on the combination of a low energy hydrogen atom … · referat der erste teil der aktivitäten...

ON THE COMBINATION OF A LOW ENERGY HYDROGEN ATOM BEAM

WITH A COLD MULTIPOLE ION TRAP

von der Fakultät für Naturwissenschaften der Technischen Universität Chemnitz

genehmigte Dissertation zur Erlangung des akademischen Grades

doctor rerum naturalium

(Dr. rer. nat.)

vorgelegt von M. Sc. Gheorghe Borodi

geboren am 06.10.1970 in Caianu Mic, Rumanien

eingereicht am 20. Oktober 2008

Gutachter:

Prof. Dr. Dieter Gerlich

Prof. Dr. Manfred Albrecht

Prof. Dr. Juraj Glosík

Tag der Verteidigung, 09. Dezember 2008

http://archiv.tu-chemnitz.de/pub/2009/

3

BIBLIOGRAPHISCHE BESCHREIBUNG

ON THE COMBINATION OF A LOW ENERGY HYDROGEN ATOM BEAM WITH A COLD MULTIPOLE ION TRAP

Dissertation an der Fakultät für Naturwissenschaften der

Technischen Universität Chemnitz, Institut für Physik,

von Gheorghe Borodi

Chemnitz, 20. 10. 2008

144 Seiten inkl. 2 Publikationen in englischer Sprache mit 6 Tabellen und 66 Abbildungen

Referat Der erste Teil der Aktivitäten dieser Arbeit bestand in der Entwicklung einer modernen Ionenspeicher Apparatur zur Untersuchung chemischer Prozesse mit atomarem Wasserstoff. Die Integration eines differentiell gepumpten Radikalenstrahls in eine vorhandene temperaturvariable 22-Pol Speicherapparatur erforderte größere Änderungen an dieser. Da astrophysikalische Fragestellungen im Vordergrund standen, führt die Einleitung zunächst in das Gebiet der Astrophysik und -chemie ein. Die Grundlagen der Ionenspeicherung in temperaturvariablen Hf-Speichern sind ausführlich in der Literatur dokumentiert. Daher ist die Beschreibung der Apparatur (Kapitel 2) relativ kurz gehalten. Viel Mühe wurde in die Entwicklung einer intensiven und stabilen Quelle für Wasserstoffatome aufgewandt, deren kinetische Energie variiert werden kann. Das Kapitel 3 beschreibt dieses Modul in vielen Details, wobei der Einsatz von magnetischen Hexapolen zum Führen der Atome und die chemische Behandlung der Oberflächen zur Reduzierung der H-H Rekombination einen wesentlichen Platz einnimmt.

Durch die außergewöhnliche Empfindlichkeit der Speichertechnik kann das neue Instrument zur Untersuchung von vielen Reaktionen eingesetzt werden, die von astrochemischer und fundamentaler Bedeutung sind. Die Ergebnisse dieser Arbeit sind im Kapitel 4 zusammengestellt, einige Reprints und Entwürfe von Publikationen findet man im Anhang. Die Reaktionen von CO2

+ mit Wasserstoffatomen und -molekülen erwiesen sich als sehr geeignet, um in situ H and H2 Dichten über den gesamten Temperaturbereich der Apparatur zu bestimmen (10 K - 300 K). Zum ersten mal wurden Reaktionen von H- and D-Atomen mit den Kohlenwasserstoffionen CH+, CH2

+,

and CH4+ bei Temperaturen des interstellaren Raums untersucht. Ein sehr interessantes,

noch nicht ganz verstandenes Stoßsystem ist die Wechselwirkung von protoniertem Methan mit H-Atomen. Im Ausblick der Arbeit werden einige Ideen aufgezeigt, wie man das Instrument verbessern kann, und es werden einige Reaktionen erwähnt, die man als nächste untersuchen könnte.

BIBLIOGRAPHISCHE BESCHREIBUNG

4

Diese Dissertation ist einen Beitrag zum Projekt 5 der Forschergruppe Laboratory Astrophysics: Structure, Dynamics and Properties of Molecules and Grains in Space, die von der DFG im Zeitraum von 2000 bis 2006 unterstützt wurde.

Schlagwörter Ionen-Atom Reaktionen, Astrochemie, Hf-Multipol Ionenfalle, niederenergetischer H-Atomstrahl, H-H-Rekombination, Ratenkoeffizienten bei tiefen Temperaturen, Dehydrierung , Deuterierung, interstellare Moleküle: CH+, CD+, CH3

+, CH2D+, CH4+,

CH3D+, CH5+, CH4D+, CO2H+

Abstract The first part of the activities of this thesis was to develop a sophisticated ion storage apparatus dedicated to study chemical processes with atomic hydrogen. The integration of a differentially pumped radical beam source into an existing temperature variable 22-pole trapping machine has required major modifications. Since astrophysical questions have been in the center of our interest, the introduction first gives a short overview of astrophysics and -chemistry. The basics of ion trapping in temperature variable rf traps is well-documented in the literature; therefore, the description of the basic instrument (Chapter 2) is kept rather short. Much effort has been put into the development of an intense and stable source for hydrogen atoms the kinetic energy of which can be changed. Chapter 3 describes this module in detail with emphasis on the integration of magnetic hexapoles for guiding the atoms and special treatments of the surfaces for re-ducing H-H recombination.

Due to the unique sensitivity of the rf ion trapping technique, this instrument allows one to study a variety of reactions of astrochemical and fundamental interest. The results of this work are summarized in Chapter 4, some reprints and drafts are reproduced in the appendix. Reactions of CO2

+ with hydrogen atoms and molecules have been established as calibration standard for in situ determination of H and H2 densities over the full tem-perature range of the apparatus (10 K - 300 K). For the first time, reactions of H- and D-atoms with the ionic hydrocarbons CH+, CH2

+, and CH4

+ have been studied at tempera-

tures of interstellar space. A very interesting, not yet fully understood collision system is the interaction of protonated methane with H. The outlook presents some ideas, how to improve the new instrument and a few reaction systems are mentioned which may be studied next.

This thesis is a contribution to the project 5 of the research unit Laboratory Astrophys-ics: Structure, Dynamics and Properties of Molecules and Grains in Space which has been supported by the DFG from 2000 to 2006.

Key words Ion-atom reactions, astrochemistry, rf-multipole ion trap, low energy hydrogen atom beam, H-H recombination, low temperature rate coefficient, dehydrogenation, deutera-tion, interstellar molecules: CH+, CD+, CH3

+, CH2D+, CH4+, CH3D+, CH5

+, CH4D+, CO2H+

5

CONTENTS 1 INTRODUCTION 7

1.1 Physical conditions in space 7 1.2 Chemistry of the interstellar medium 7 1.3 Laboratory approaches 11 1.4 Overview 13

2 EXPERIMENTAL 14 2.1 The AB–22PT apparatus 14 2.2 Vacuum system and gas inlet 16 2.3 Description of the 22-pole ion trap machine 18 2.4 Determination of rate coefficients 21

3 COLD ATOMIC HYDROGEN BEAM SOURCE 25 3.1 Theoretical considerations 25

3.1.1 Theory of effusive gas sources 25 3.1.2 Mechanisms of H production and loss 27 3.1.3 Guiding and focusing H atoms 31

3.2 Technical description 33 3.2.1 The radio frequency dissociator 33 3.2.2 The nozzle cooling system 34 3.2.3 The hexapole magnets 36

3.3 Test measurements 38 3.3.1 Degree of dissociation measurements 39 3.3.2 Number density at trap location 45 3.3.3 Velocity distributions of atoms 53

4 REACTIONS 55 4.1 Temperature dependence of reactions of CO2

+ 55 4.2 Reactions of CH+ with H and D 56 4.3 CH4

+ + H, CH4+ + D 61

4.4 CH5+ + H, CH5

+ + D 65

5 SUMMARY AND OUTLOOK 69 APPENDIX 71

A On the combination of a low energy hydrogen atom beam with a 71 cold multipole ion trap

B Reactions of CO2+ with H, H2 and deuterated analogues 109

C Interactions of ions with hydrogen atoms 119 D Nomenclature 129

6

REFERENCES 131

LIST OF PUBLICATIONS AND CONFERENCE CONTRIBUTIONS 137 SELBSTSTÄNDIGKEITSERKLÄRUNG 139 CURRICULUM VITAE 141 ACKNOWLEDGEMENTS 143

7

1 INTRODUCTION

1.1 Physical conditions in space Most of the mater in the universe is assembled in large agglomerates of stars known as galaxies; however, the stars occupy only a small fraction of the space. The space be-tween stars is found to be not devoid of material but to contain so-called interstellar medium (ISM), which consist of gas (99 %) and sub-µm sized grain particles (1 %) with an average number density of 1 H atom cm-3 [kai02]. Using various methods, astrono-mers have been able to deduce the cosmic abundances of the elements. By far dominant is still hydrogen; helium has about 7 % of the hydrogen abundance by number. The bio-genic elements carbon, nitrogen and oxygen have fractional abundances with respect to hydrogen of 4 × 10-4, 9.3 × 10-5, and 7.4 × 10-4, respectively [her05]. Other elements like neon, silicone, magnesium and sulfur are less copious furnishing only 0.02 % of the hydrogen abundance. The average elemental abundances do not apply to all astronomi-cal objects, for example, stars in certain stages of evolution can be carbon rich (contain-ing more carbon than oxygen).

Early models [kee77] classified the ISM into three phases: the Cold Neutral Medium (CNM) often referred to as clouds; the Warm Ionized Medium (WIM) which is consid-ered the boundary layers of the CNM; and the Hot Ionized Medium (HIM), which is sometimes referred to as the intercloud medium or the coronal gas. These phases are thought to be in approximate pressure equilibrium with one another [cox05]. The CNM itself appears to contain a variety of cloud types, spanning a wide range of physical and chemical conditions. The densest clouds that are most protected from UV radiation from stars are referred to as dense clouds, dark cloud or molecular clouds. They are character-ized by typical number densities of 102 – 104 cm-3, large visible extinction, and their kinetic temperatures are typically on the order of 10 – 50 K [sno06]. The molecules range in size from 2 –13 atoms and are mainly organic in nature. Molecular hydrogen is the dominant species, having a concentration roughly 104 times than of the second most abundant molecule – CO. The most tenuous clouds, fully exposed to starlight are usu-ally called diffuse clouds. The concentration of diffuse matter is typically 10 – 100 at-oms cm-3 and average kinetic temperature in the range 50 – 100 K [her05]. As in all clouds, the gas is mainly hydrogen in one form and another, betraying the fact that the material in cloud comes from previous generation of stars, which are also mainly hy-drogen in content.

1.2 Chemistry of the interstellar medium The chemical complexity of the ISM has been explained for many years by gas-phase processes driven by cosmic-ray ionization. The atoms, molecules, the dust particles and the radiation field interact with each other in an extremely complex manner. Ion-molecule reactions have dominated for long time the reaction models developed for describing molecular clouds since many of them have no barriers [her73], [bla77]. At the low density in the interstellar gas, three-body processes are unimportant, so that only two-body reactions need to be considered. A very important mechanism by which mo-lecular bonds can be formed at the low temperatures and densities of the ISM is radia-tive association [ger92c]. It has recently been realized that neutral-neutral reactions in-volving radicals or atoms can have large enough rate coefficients at low temperatures that they may be important [her01], [kai02], [smi04]. Very relevant for understanding the evolution of interstellar matter is the process of isotopic fractionation occurring pre-dominantly at the low temperatures of interstellar clouds [her03], [mil03].

1 INTRODUCTION

8

One reaction of central importance which looks simple at a first glance, is the formation of the hydrogen molecule from two hydrogen atoms. In the early universe, H2 is the product of the fast associative detachment reaction with atomic hydrogen H- + H → H2 + e-. H- ion is formed by the slow radiative association reaction of H with a free electron and that this one can be easily photodetached. There have been other mechanisms pro-posed but in most circumstances the H- pathway dominates [glo03]. Unfortunately there is considerable variation in the rate coefficients of this fundamental reaction. For exam-ple, published values for the associative detachment reaction differ by nearly an order of magnitude introducing significant uncertainties into the H2 formation rate. The cosmo-logical implications have been discussed recently [glo06b], [glo08a]. The only way to remove these uncertainties will be to obtain more accurate rate coefficients for the rele-vant associative detachment processes at cosmologically relevant collision energies supplemented by further theoretical calculations.

In interstellar clouds, it is generally assumed that, hydrogen molecules are produced on the grain surfaces. A large body of theoretical work exists [pir00] stretching back to the pioneering work of Gould & Salpeter [gou63] and Hollenbach, Werner, & Salpeter [hol71]. So far no gas phase processes are known, e.g. ternary or radiative association or chemical reactions, which could explain the observed H2 abundances. Some quantum mechanical calculation have suggested that molecular hydrogen may be formed by reac-tions between hydrogen atoms and positively charged PAHs, resulting in a simultaneous dehydrogenation of the aromatic ions. Since regeneration of the original cation by addi-tion of atomic hydrogen to the dehydrogenated positive ion is thermodynamically al-lowed, the process can be cycled [che94], [her99]. In general catalytic cycles such as XY+ + H → XYH+ followed by XYH+ + H → XY+ + H2 are possible schemes. Potential candidates for XY+ are C+, CH3

+, C2H2+, or C3H2

+. Some of the relevant steps have been studied in this work.

A special very interesting subfield of interstellar chemistry is the formation and destruc-tion of hydrocarbons. The gas-phase chemical networks describing the most important formation routes of carbon-bearing molecules in local ISM have been discussed exten-sively in the literature [smi92], [hol97], [dis99], [her01], [her05]. Because hydrogen is more abundant than any other element, reactions with H and H2 dominate the networks if they are exothermic. In dense clouds the ion-chemistry is initiated by the action of cosmic rays on the major constituents producing H+, H2

+, and He+. The produced H2+

ion reacts fast with molecular hydrogen to form the stable H3+ ion. This ion plays a piv-

otal role in the subsequent ion-molecule chemistry through proton transfer. The reaction of H3

+ with C to form CH+ initiates the hydrocarbon chemistry through the chain of re-actions CH+ + H2 → CH2

+→ CH3+. Because the reaction of CH3

+ with H2 to form CH4+

is endothermic only slow radiative association reaction leading to CH5+ can occur

[ger92c]. Once protonated methane is synthesized, it is depleted by dissociative recom-bination leading mainly to the production of methyl radical rather than methane and by reaction with abundant CO molecules leading to methane. As indicated in Fig. 1.1 some depletion of CH5

+ occurs via the H atom transfer reaction CH5+ + H → CH4

+ + H. This reaction channel is more important in diffuse clouds where the abundance of H and H2 are equal. Motivated by the important of this reaction for interstellar chemistry and fun-damental research it has been studied in the present work at collision energies relevant for dark interstellar clouds.

Once simple hydrocarbons such as methane are formed, the formation of more complex hydrocarbons occurs via three types of reactions [her01]: (i) carbon insertion reactions (e.g. C+ + CH4 → C2H3

+ + H, C2H2+ + H2); (ii) condensations reactions (e.g. CH3

+ + CH4

1 INTRODUCTION

9

→ C2H5++H2); and (iii) radiative association reactions (e.g. C+ + Cn → Cn+1

++hυ). In general, carbon insertion with C+ is thought to be the dominant route. Because the larger ions CnHm

+ do not react rapidly with H2, low-temperature gas-phase chemistry produces strongly unsatured hydrocarbons, in agreement with observations of dark clouds. Recent laboratory work revealed that many CnHm

+ ions that are unreactive in a molecular hy-drogen environment became reactive in an H-atom environment. For example, the syn-thesis of benzene in interstellar clouds is initiated by the slow radiative association reac-tion C4H2

+ + H → C4H3+ + hυ [ewa99].

Fig. 1.1. Initial steps in the gas-phase carbon chemistry in diffuse and dark clouds [dis99]. Of specific importance for this work is the dehydrogenation of ions (see for ex-ample CH5

+) in collisions with H which, simultaneously leads to H2 formation.

Due to the intensity of the stellar UV radiation in diffuse clouds a substantial fraction of the matter is in the atomic form. Thus the major constituents are H atoms and H2 mole-cules, with C, N, and O atoms as minor constituents. In the outer part of the diffuse clouds where the UV is more intense, the hydrocarbon chemistry begins with photoioni-zation of C atoms by stellar UV, producing C+ ions and electrons. Formation of CH+ through the reaction C+ + H2 → CH+ + H does not proceed at low temperature because the reaction is endothermic by about 0.4 eV. In particular, C+ undergoes a slow radiative association reaction with H2: C+ + H2 → CH2

+ + hυ which has been measured to occur with a rather small rate coefficient of ~ 10-15 cm3s-1 [ger92c]. Deep inside diffuse clouds the UV radiation field is reduced and a fraction of carbon remains in its neutral form. Therefore, similarly to dark clouds, an alternative route for hydrocarbon synthesis in diffuse clouds is the reaction H3

+ + C → CH+ + H2.

1 INTRODUCTION

10

At high temperatures of ~ 100 - 4000 K such as encountered in photon-dominated re-gions (PDR) or shocked regions, gas-phase reactions with H and H2 become significant [dis99]. PDR chemistry differs from normal ion-molecule chemistry in a number of ways. Obviously, because of the high FUV flux, photoreactions are very important, as are reaction with atomic hydrogen. Likewise, vibrationally excited H2 is abundant and can play a decisive role in the PDR chemistry. If the gas get very warm (≥ 500 K), the activation barrier of reactions of atoms and radicals with H2 can be easily overcome and this type of reactions can dominate. Electron recombination and charge exchange reac-tions are important for the ionization balance. Finally, the FUV flux keeps atomic O very abundant throughout the PDR and hence burning reactions are effective. The most important reactions in the chemistry of carbon compounds are schematically shown in Fig. 1.2.

Fig. 1.2. Reactions of central importance for the carbon-bearing molecules included in the chemical network of dense photodissociation regions (PDRs) (adapted from [hol97]).

The PDR surface layer consists largely of neutral or cationic atoms created by photodis-sociation and ionization reactions. Hydrocarbon chemistry in PDR begins with photoionization of atomic carbon followed by the endothermic abstraction reaction C+ + H2 → CH+ + H and C+ + H2

* → CH+ + H. The CH+ ions rapidly formed by these reac-tions are removed by reactions with H2, electrons, and the reverse reaction CH+ + H → C+ + H2 which has been measured also in this work (see Sec. 4.2). Likewise, a small fraction of the neutral C reacts with H2

* to form CH which is destroyed by the reverse reaction with H. Photoreactions are important for C, OH, and CO but not for the small hydrocarbon radicals and cations.

In addition to H, which is the most abundant atom in the Universe, there is a density of D atoms, which is smaller by a factor of about 10-5 compared with the density of H at-oms in regions where the gas is primarily atomic. Despite the small abundances of D atoms about 30 deuterated molecules have been detected to date in the interstellar me-dium including doubly and even triply deuterated species [lis07]. Surprisingly, the measured abundance ratios of the singly deuterated molecules to their undeuterated counter parts tend to be factor of up to 104 greater than expected from the D/H elemen-tal ratio of about 1.5 ×10-5. Owing to the studies of low-temperature ion-molecule reac-tions it is now known that this enrichment is due to the phenomenon of ‘isotope frac-tionation’. In molecular clouds most of the deuterium is contained in HD. Therefore, it

1 INTRODUCTION

11

is assumed that the deuterium fractionation is initiated by H-D exchange reactions be-tween three primary molecular ions H3

+, CH3+, and C2H2

+ with HD as neutral reactant [ger02b], [mil03]. In general H-D exchange reactions are exothermic due to differences in zero-point vibrational energies. Reactions in subsequent collisions of these ions lead to the deuteration of other molecules like HCO+.

However, in dense clouds a small amount of atomic deuterium of order 10 % can be maintained through the dissociative recombination of DCO+ + e- → D + CO. Therefore, Dalgarno & Leep have extended the theory of deuterium fractionation to include in the list of reactions the atom reactions, like for example H3

+ + D → H2D+ + H and HCO+ + D →DCO+ + H [dal84]. Deuteration in collision with D atoms is enhanced in certain regions of dense interstellar clouds, e.g. a low mass pre-protostellar core, since all heavy elements such as C, N, and O are incorporated in ice mantles frozen on dust grain sur-faces [wal04]. D atoms play a fundamental role in the chemistry of the early universe and diffuse clouds. For example, the formation of HD in diffuse interstellar clouds oc-curs more efficiently in the gas phase via the sequence of reactions H+ + D → H +D+ followed by D+ + H2 → H+ +HD, which can lead to enhancements of 100 over surface formation [wat73] and can explain the observed HD/H2 abundance ratio in diffuse clouds.

Recently, in our laboratory systematic studies on possible deuteration reaction of small hydrocarbons CHn

+, (n = 3..5) and of C3Hn+, (n = 1..3) in collision with HD have been

carried out [asv04b], [sav05b], [sav05c]. The present work is an extension of these stud-ies to include H-D exchange reactions of small hydrocarbons CHn

+, (n = 1..5) in colli-sion with D and D2. For such investigations, an atomic beam-22 pole trap apparatus (AB-22PT) has been constructed, as described in the next chapter.

1.3 Laboratory approaches One of the motivations for investigating ion-atom or ion-molecule reactions at low-temperatures and low-density was and still is the need of accurate data for predicting the chemical evolution of interstellar clouds. Up to date gas-phase chemical models contain ~ 4500 reactions and photodestruction processes among 400 or so species (e.g., the UM-IST database for astrochemistry [woo07]). Unfortunately experimental data are avail-able on just one-third of all reactions. Moreover, the majority of experiments have been performed at room temperatures, some down to the temperature of liquid nitrogen, 80 K. This means that for a large number of reactions only crude theoretical estimates are utilized in the simulations. The assumption that rate coefficients of exothermic reac-tions are temperature independent or the extrapolation of results measured at high tem-perature to 10 K often leads to large uncertainties. Typical examples are the hydrogen abstraction reactions CH4

+ + H2 → CH5+ + H or C3

+ + H2 → C3H+ + H which have been investigated in our laboratory between 15 and 300 K [asv04a], [sav05b]. Although exo-thermic, these reactions are rather slow at room temperature but become faster with de-creasing temperature.

In order to study such processes over a wide range of temperatures and densities, spe-cific experimental techniques with increasingly powerful capabilities were developed in the last decades to characterize in detail the interstellar medium described above. In the following a brief overview over three types of experimental techniques involved in the study of ion-atom reaction is given. Most of the ion-atom reactions have been studied using the flowing afterglow technique (FA), pioneered by Ferguson and colleagues [fer69] and derived techniques such as the selected-ion flow tube (SIFT) technique. In-

1 INTRODUCTION

12

vestigation of reactions at lower temperatures became possible with the development of the variable-temperature selected-ion flow tube (VT-SIFT) instrument [smi79], [ada85]. In order to reach energies from thermal to ~ 2eV the flow-drift tube (FDT) technique was developed [far73].

There have been several other attempts to study reactions at very low temperatures. One of the solutions is to use supersonic expansions. The best known example is the CRESU (Cinétique de Réaction en Ecoulements Supersoniques Uniformes) apparatus developed by Rowe and coworkers [row84]. In this apparatus ions are injected into the core of a several centimeters thick supersonic flow of helium containing a small amount of the neutral molecule as reactant. Kinetic data at temperatures as low as 7 K have been ob-tained [smi00]. An extension to temperatures below 3 K has been made possible by the development of the free jet flow reactor by Smith [smi98]. Unfortunately, both methods have not yet been used for laboratory study of ion-hydrogen atom reactions.

Since the 1960’s, various molecular beam techniques have been used to the study of chemical reactions. Such experiments yield a great deal more information on the micro-scopic details of the molecular collision dynamics than do measurements of a thermally averaged rate coefficient for the same process. The first method capable of reaching meV collision energy was the merged-beam technique, pioneered by Trujillo et al. [tru66], and further developed by Gentry et al. [gen75]. A more sophisticated merged beam apparatus is the arrangement described by Gerlich [ger92], [ger08a]. A recent example is the merged-beam apparatus at the Columbia Astrophysics Laboratory aimed at studying the associative detachment reaction H- + H → H2 + e- [sav08].

Fig. 1.3. Construction details of a typical arrangement of a temperature variable 22-pole ion trap combined with a pulsed supersonic beam. For details see [ger06a].

Various ion traps have been used since the 1960’s for studying ion-molecule reactions. Using suitable electric or magnetic fields, ions can be confined for long times, and with the use of a buffer gas thermal conditions can be attained. The most common technique

1 INTRODUCTION

13

used in ion chemistry is based on the ion cyclotron resonance (ICR) cell (for a review see [mar98]) and the Penning ion trap [lui85]. Confinement of ions is based on the su-perposition of a static electric and magnetic field. As will be discussed below, there has been only one attempt to combine an ion cyclotron resonance cell with a hydrogen dis-charge source [kar79]. Very successful and versatile solutions utilize rf multi-electrode arrangements, such as stacks of ring electrodes or linear multi-poles traps, combined with cryogenic cooling.

The first trapping machine developed in the late 1980’s by Gerlich and colleagues is a linear combination of an ion source, a first mass filter for ion preparation, the ion trap, a quadrupole mass spectrometer for analyzing the trap content and an ion detector [ger92c], [ger95]. A slightly different approach was used to combine a trap and a graph-ite sublimation source for providing a neutral Cn (n = 1-3) target [sav05a]. An important modification, extending the features of trapping machines, has been realized in this the-sis, the integration of a beam of hydrogen atoms.

1.4 Overview In this work an rf trapping machine has been reconstructed thoroughly for studying re-actions between stored ions and hydrogen atoms under well-defined conditions. The basics of the innovative instrument, its features and detailed tests are described in Chap-ter 2. The central innovation, the temperature variable H-atom source, is documented in great detail in Chapter 3. Chapter 4 summarizes various experimental results obtained in this thesis, including reactions of CO2

+ and small ionic hydrocarbons CHn+ (n = 1, 4, 5)

with H, H2 and deuterated analogues.

Some preliminary results of this work have been reported as invited article on a confer-ence [luc05] and is reproduced in Appendix C. The manuscript describing the potential of the machine, a few limitations and possible improvements can be found in the Ap-pendix A. The paper discussing the results for CO2

+, which can be found in Appendix B, has been accepted for publication.

15

2 EXPERIMENTAL

2.1 The AB – 22 PT apparatus A major part of the experimental work of this thesis was to develop a temperature vari-able H-atom source, to integrate it into an existing 22-pole ion trap apparatus, and to characterize it features and limits. In the following, this instrument which is briefly called AB-22PT apparatus (Atomic Beam 22-Pole Trap) will be described. The basic operational principles are illustrated in Fig. 2.1.

Fig. 2.1. Combination of a 22PT machine with an H atom beam. Trapping experiments are usually performed in a pulsed mode. A bunch of ions is generated in an ion source and mass filtered in a quadrupole mass filter (not shown here). Using a electrostatic ion bender, the primary ions are injected into the ion trap. There they are cooled to the tem-perature of trap walls via inelastic collisions with cold buffer gas. The ions also interact permanently with the continuous H/H2 (D/D2) beam entering the trap from the left. Alter-natively, neutral target gas can be directly leaked into the trap. After a given storage time, the content of the trap is extracted, mass analyzed in the second mass filter and finally de-tected and counted.

In the early ion trapping machines, the various modules usually have been arranged co-axially. Typical modules include an ion source, a mass filter for ion preparation, an ion trap in the center, a quadrupole mass spectrometer for analyzing the trap content, and an ion detector [ger95], [asv04a], [sav05b]. In order to integrate the H-atom beam, it was necessary to modify the injection of the ion beam from linear to orthogonal using a dc quadrupole bender (see Fig. 2.1). The H-atom beam source consists of a standard rf driven plasma source, a temperature variable accommodator which allows to change the velocity distribution of the H atoms and one or two hexapole magnets for guiding the atoms and for improving the hydrogen background using differential pumping. The at-oms and the not dissociated molecular background, effusing from the cold accommoda-tor, are formed into a beam by a skimmer. The various apertures placed after the hex-apole magnet and the entrance of the trap were chosen such that the neutrals traverse the 22-pole trap in axial direction without hitting any surface.

Many details of the complete AB-22PT apparatus are shown in Fig. 2.2. The left half is dominated by the H-atom beam source and the differential pumping chamber. The modules for ion preparation (storage ion source and QP mass filter), ion trapping (22PT) and detection (QP mass spectrometer and scintillation detector), are standard for a typi-cal ion trapping instrument. Since they often have been described in the literature [ger93b], [ger95], [asv04a], [sav05b], [ger06a], [ger08a] only a few remarks are made

2 EXPERIMENTAL

16

in the following. Details concerning the H atom beam, its integration into the trapping machine, and various experimental tests will be discussed in Chapter 3.

Fig. 2.2. The Atomic Beam 22-Pole Trap Apparatus (AB-22PT)

2.2 Vacuum system and gas inlet A very good vacuum is a crucial point in an ion-trapping experiment since, in addition to reactions with the reactant gas, the ions also interact with traces of residual gas at long storage times. In order to suppress such parasitic reactions, the partial pressure of impurities should be below 10-11 mbar. The vacuum system of the AB-22 PT apparatus is shown in Fig. 2.3. Basically, it consists of several standard chambers with ConFlat flanges which are separated by differential walls. They are pumped by means of a vari-ety of turbo-molecular pumps. In order to avoid back-diffusion, also the forevacuum system is rather complex.

The atomic beam source chamber is pumped by the biggest turbo-molecular pump (Pfeiffer, TPU 2200). For H2, it has a nominal pumping speed of 2800 l/s, which is re-duced by 15 % due to the splinter shield. It is backed with a turbo-molecular pump (Pfeiffer, TPU 240) in series with a rotary pump (Pfeiffer, DUO016B). A skimmer with a diameter of 2 mm connects the source chamber with the hexapole chamber. These two chambers can be separated by closing a gate valve mounted directly after the skimmer (not shown in the figure). Two magnetically suspended turbo pumps (Leybold, T340M, 370 l/s for H2) evacuate the hexapole chamber and the central chamber housing the 22PT. The ion source chamber is pumped with one turbo-molecular pump (Pfeiffer, TMU 200MP, 105 l/s for H2). The turbo pumps are backed with a series of drag (Pfeif-fer, TPD011, 10 l/s) and diaphragm (Pfeiffer, MVP015, 15 l/s) pumps in order to get sufficient compression for hydrogen and to avoid oil pumps.

2 EXPERIMENTAL

17

To monitor the pressures, all chambers are equipped with ion gauges. For absolute pres-sure and flow measurements they are calibrated using a spinning rotor gauge (MKS SRG2), directly connected to the chambers. In order to measure accurately the number density of reactant gas inside the 22-pole trap, a spinning rotor gauge is connected with the inside of the trap using a flexible stainless steal tube. Without the neutral beam and with all gas inlets closed, the system reaches pressures of about 10-9, 10-8 and 10-10 mbar in the source, hexapole, and 22PT chamber, respectively. With the H atom beam in op-eration (at a hydrogen throughput of 5 × 10-3 mbar l/s) the corresponding pressures are 10-6, 10-7 and 10-8 mbar.

Fig. 2.3.Vacuum system of the AB-22PT apparatus.

The gas inlet system for the H atom source is shown in Fig. 2.4. It consists of two lines, one for the H2 or D2 and another one for the water vapor. At the gas bottles which are typically filled with gas up to 200 bar, pressure reducers are mounted to get a constant pressure of about 1.5 bar. For flow measurements the gas is filled into containers with calibrated volumes. Each container is equipped with an ultra high purity pressure trans-ducer (Swagelok, PTU). The gas flow into the discharge tube is controlled using needle valves. The pressure at the entrance of the discharge tube is measured with a capacitive gauge (MKS, 722A02MCD2FA). The whole gas inlet system is bakeable.

In addition to the hydrogen, a constant flow of a small amount of water is leaked into the discharge tube for reducing recombination of atoms on the walls. The role of water vapor for obtaining high dissociation degrees has been already mentioned in the litera-ture [szc00]. Our water coating procedure is described in Sec. 3.3. In order to create a reproducible flow of water vapor, the water reservoir is dipped into an ice bath while, for avoiding condensation, the gas inlet tubes are heated to a constant temperature of 60 °C.

2 EXPERIMENTAL

18

Fig. 2.4. Gas inlet system of the H atom source.

The system which supplies the neutral gas to the 22-pole trap and the ion source is pre-sented in Fig. 2.5. It allows us to introduce ultra pure gases into the machine or to pre-pare well-characterized mixtures. The neutral gas is leaked into the 22PT either con-tinuously using two sapphire sealed leak valves (Varian Model 951) or in a pulse mode via a fast piezoelectric valve, developed in our group. The ion source is supplied with gas via two lines. One of the containers is used for the precursor of the ion of interest, the other one usually has been filled with CO2 since CO2

+ ions is routinely used for calibrating the H atom density.

Fig. 2.5. Gas inlet system of the 22-pole trap apparatus

2.3 Description of the 22-pole ion trap machine Ion source For creating ions a storage ion source has been used which is itself a combination of an ion trap with an electron source [ger92]. The double-H-shaped storage volume is de-fined by a stack of 8 rectangular molybdenum plates separated by ruby balls. The plates are alternatively connected to the two phases of an rf generator. To prevent ions from leaking out of the trap, the top and the bottom of the storage volume are closed by an adequate dc bias voltage applied to the endplates, each containing a small slit for the electron beam. Ionization of the neutral precursor gas is achieved through electron

2 EXPERIMENTAL

19

bombardment from a rhenium filament (0.3 mm) that is situated above one of the side channels. Applying a voltage between the storage source and the filament, the electrons are accelerated into the source volume. To obtain maximum ion yield a repeller plate and a slit electrode focus the electron beam into the center of the storage volume. After generation, the ions are trapped in an environment where they can relax by collisions to the temperature of the source (typically 350 - 450 K). They also can react with other neutral molecules. This allows for example to chemically quench unwanted ions or to synthesize species which cannot be obtained directly by electron impact. Variation of the trapping parameters, precursor gas mixture and pressure permits a good control over the generated ion composition. At both ends of the source (in axial direction), electrodes with apertures are mounted. For producing a constant flow of ions, a negative voltage is used. Using a rather small voltage (less than -1V) the field penetrating into the exit re-gion is rather weak and the ions just diffuse toward the exit region. This leads to a nar-row energy distribution. The trap also can be kept closed by applying a positive bias voltage to the exit electrode. Superimposing an adequate pulse of a few µs or ms, the stored ions are gently extracted from the source. The storage property of the source al-lows low pressures to be used. Therefore the pressure of the precursor gas, leaked into the main chamber, can be very small. In summary, advantages of this source include high ionization efficiency, chemical ionization, and especially thermalization of all de-grees of freedom of the ions.

The quadrupole mass filters RF quadrupoles are the best characterized devices using oscillating voltages. They have become the working horse of mass spectrometry in many areas of science. Their use as mass filter or ion guide are well described in many books and papers [daw76], [ger92], [mar89]. In the AB-22PT apparatus two quadrupole mass filters are used. The first quadrupole is operated either in the mass-selective mode or in the low-mass band-pass mode depending on the application. The second quadrupole is operated exclusively in the mass-selective mode. The first quadrupole was driven by a home-built rf power sup-ply and the second one by a commercial 1.637 MHz rf power supply (BALZERS rf-generator QMH510 with control unit QHS511). The first mass filter is 245 mm long and consist of 4 rods with half-moon like shape (effective d = 10 mm) circumscribing an inner circle of a radius r0 = 4.35 mm. The second one is 260 mm long and consist of 4 rods with a diameter of d = 18 mm, circumscribing an inner circle of a radius r0 = 7.84 mm.

For operation, the potential +( 0U - 0V cos tΩ ) is applied to two opposite rods, the other pair gets -( 0U - 0V cos tΩ ), where 0U is the DC-voltage, 0V the amplitude of the rf- voltage, fπ2=Ω is the angular frequency and t is the time. The principles of operation are best explained with the help of the stability parameters a and q defined in the lit-erature. Here we use the nomenclature of [ger92].

20

20

28

rmqUaΩ

= , 20

20

24

rmqVqΩ

= (2.1)

Depending on the values of these dimensionless parameters, the motion of an ion in such a field is stable or unstable. A small fraction of the stability diagram is shown in Fig. 2.6. If the pair (a2, q2) is situated within the triangle-shaped region, the trajectories are stable and the ions will be transferred safely through the quadrupole. Outside of this region, trajectories are unstable. The amplitude of the oscillation increases exponen-

2 EXPERIMENTAL

20

tially, resulting finally in the loss of the ions by leaving the structure or hitting the elec-trodes.

Fig. 2.6. Stability diagram in the (a2, q2) plane and different operational modes of a quad-rupole mass spectrometer.

In practice, two different regions in the stability diagram are used: the so called “mass-selective” one which is closed to the tip of the triangle (a2, q2) = (0.237…, 0.706…), and “low band pass” one which is defined by the condition q2 < 0.3. When the quadru-pole is operated in the mass-selective mode, only a narrow mass range around sm is transmitted. The width (mass resolution) depends on how close to the tip the operating line (see Fig. 2.6) cuts through the triangle. All masses outside the stability region are unstable. For selecting another mass, 0U and oV must be changed in accordance with equations (2.1) such that this mass is shifted into the stability region. This mode of op-eration is well suited for mass analysis, but not for the preparation of a low kinetic en-ergy ion beam. Operating the quadrupole in the low-mass band pass mode is based on the fact that only ions with masses below a certain limit, the so called low-pass cutoff mass,

0

20

2

20

UrqVmc Ω

= , (2.2)

are transmitted through the quadrupole, while ions with masses above this value are removed by the filter [ger92]. This mode is well suited for preparing an ion beam with a narrow energy distribution because it operates in the region where the kinetic energy is an adiabatic constant of the motion.

The 22-pole ion trap The most important module of this machine is the temperature variable 22-pole ion trap shown schematically in Fig. 1.3 and in the center of Fig. 2.2. It consists of two sets of 11 stainless steel rods (diameter d = 1 mm, inscribed radius r0 = 0.5 cm). As described in Fig. 2.1 ions are usually generated externally and filled into the trap. Radial confine-ment of the ion cloud is achieved by applying a suitable rf voltage alternatively to the two sets of rods. In the present experiments (e.g. for CO2

+ ions used for testing the hy-

2 EXPERIMENTAL

21

drogen beam) the trap is operated with an rf voltage with an amplitude oV = 50 V and a frequency f = 15.86 MHz. Using the standard definitions and formulas [ger92], these operating parameters lead to an effective potential and a stability parameter of V*(r = r0) = 0.668 eV and η(r = r0) = 0.097, respectively. For the interaction with the atomic beam it is important to note that the stored ions explore a large region in radial direc-tion. For a transverse energy of 10 meV the turning radius is 0.8r0, i.e., the ions explore a cylinder with a diameter of 8 mm. In the axial direction the ions are confined by elec-trostatic voltages (<+1 V) applied to gates electrodes indicated schematically in Fig. 2.1.

For obtaining a cold environment the trapping electrodes are surrounded by a copper box which is mounted onto the cold head of a closed-cycle He refrigerator (Leybold RGD 210 with compressor RW2/3). A second heat shield is mounted onto the first stage (temperature 35 K) enables one to reach temperatures as low as T22PT = 10 K. The trans-lational and internal degrees of freedom of the ions are coupled to the cold environment by inelastic collisions with helium or hydrogen buffer gas. In most applications we util-ize a buffer gas pulse which is intense enough that the translational and internal degrees of freedom of the ions are cooled within a few milliseconds [ger03a], [ger08a], [ger08b]. The trap temperature is measured by a calibrated silicon diode (LakeShore Cryotronics, model DT-470) mounted directly onto the copper wall. In order to operate the trap at variable temperatures between 10 K and 300 K, a resistive heater is mounted between the cold head and the trap.

Daly detector Ions extracted from the trap are mass selected and then detected with high efficiency using a Daly-type detector [dal60]. The main parts (see Fig. 2.2) are the stainless steel stamp, the so-called converter, a scintillator, coated with a thin aluminum layer and a photomultiplier. The converter is set to a high negative voltage (-30 kV), attracting posi-tive ions. These impinge with high velocities upon the metal surface and kick out sev-eral secondary electrons. These negative particles are accelerated in the potential gradi-ent between the converter and the grounded scintillator situated at the opposite side. In the scintillator the fast electrons are converted into photons which are registered by the photomultiplier located outside the vacuum chamber. The output pulses of the pho-tomultiplier are accumulated in a 100 MHz counter.

2.4 Determination of rate coefficients The ion cloud injected into the trap, undergoes changes due to various processes includ-ing relaxation, wanted and unwanted reactions and loss from the trap (to low trapping field or a reaction forming a non detected mass). In all cases the temporal changes of the ion composition are measured for different storage times, pressures etc. From such re-sults reaction rate coefficients are determined by simulating all processes as exact as possible with an adequate system of coupled rate equations. Typical measuring se-quences and iterations, performed with the AB-22PT apparatus, are described in detail in the manuscripts in Appendix A and B.

As an example Fig. 2.7 shows the results for CO2+ ions injected into the trap at T22PT =

15 K. Using the effusive gas inlet, deuterium is leaked into the trap leading to a station-ary number density of [D2] = 1.3 × 1011 cm-3. Plotted are the average numbers of the indicated ions as a function of storage time. Since data are accumulated over many it-erations, we usually plot the normalized number of ions injected into the trap, Ni called briefly the number of ions per filling. Inspection of the figure reveals that the deuterium density is high enough to convert half of the CO2

+ primary ions into DCO2+ products

2 EXPERIMENTAL

22

already after 5.9 ms. As the first mass filter was operated in the rf only mode, also some 50 HCO2

+ ions were transferred to the trap, in addition to the CO2+ primary ions. How-

ever, these ions do not perturb the investigations since they react very slowly with D2 at the low temperature of the trap. From the mono-exponential decay and the density of the neutrals, one obtains directly the rate coefficient for attenuating CO2

+. If several competing processes play a role, the temporal evolution of all ions is used to extract reaction rates. Here one obtains the same result, the rate coefficient derived from the CO2

+ decay and that from the DCO2+ formation is k = 1.3 × 10-9 cm3 s−1. The absolute

error for the rate coefficients are estimated to be 20 % and are mainly due to uncertain-ties in the determination of the effective D2 number density. Note that this is in part due to the nonlinear dependence of the density along the axis of the trap.

Fig. 2.7. Reactions of CO2

+ ions with D2 molecules, leaked directly into the trap at a number density of [D2] = 1.3 × 1011 cm-3. For cooling the ions to T22PT = 15 K, He gas has been added with a number density of 4.5 × 1013 cm-3. The number of primary and product ions per filling, Ni, is plotted as a function of the storage time t. At each iteration a mix-ture of both CO2

+ and HCO2+ ions are filled into the trap. While the first ion reacts with

deuterium to DCO2+, the second one remains almost unchanged.

If the primary ions are converted very slowly into products, a linear fit to the increase of product ions is sufficient. This is illustrated in Fig. 2.8 which shows a typical result measured for the reaction CH5

+ + H → CH4+ + H2. This reaction has been studied both

as a function of the ion temperature and the kinetic energy of the H atoms (see Sec. 4.4.). For the measurements shown in Fig. 2.8 the temperature of the 22PT and the ac-commodator has been set to 100 K and 12 K, respectively. As explained above, the plot-ted number of primary and product ions is normalized to number of ions per filling. Each time about 1000 CH5

+ ions have been injected into the trap and exposed to a mix-ture of H atoms and H2 molecules. Also small amounts of CH4

+ ions (~ 0.2 counts per filling) were initially present in the 22PT. The calibration of the H and H2 densities for this measurement is based on reactions with CO2

+ (see Fig. 2.9).

In the special case shown in Fig. 2.8, however, the reverse reaction CH4+ + H2 → CH5

+ + H has to be accounted for due to the large H2 background in the 22PT. This is easy to do since the rate coefficient has been measured separately in this work using a molecu-lar beam of H2 (see Sec. 4), and also previously in our laboratory [asv04a]. The solid

2 EXPERIMENTAL

23

lines in Fig. 2.8 represent solution of a set of coupled rate equations describing both the formation and destruction of the corresponding ions. Initial conditions are the number of various ions at short storage times. The coupled differential equations are usually solved numerically with a set of assumed rate coefficients which are varied systematically until the agreement with the experimental data is satisfactory. For the measurements shown in Fig. 2.8 this procedure leads to k = 1.6 × 10-11 cm3 s−1 for the production of CH4

+.

Fig. 2.8. Reaction of CH5

+ ions with a mixture of H / H2 have been recorded at TACC = 12 K and T22PT = 100 K. Plotted is the number of primary and product ions, Ni, per filling as a function of the storage time t. For cooling the ions to T22PT = 100 K, He gas has been added with a number density of 4.3 × 1013 cm-3. The solid lines are solutions of an adequate rate equation system leading to the rate coefficient k = 1.6 × 10-11 cm3 s−1 for the production of CH4

+.

For determining absolute rate coefficients, one has to determine the effective number density of neutral reactants in the trap. For those experiments which are performed with gas leaked into the trap, the target pressure is determined using a spinning rotor gauge type MKS SRG2 connected via a short tube to the interior of the trap. This gauge which operates at room temperature has a specified accuracy of 5 %. Since it is very sensitive to vibrations, it can be used only for pressures above 10-7

mbar. Therefore we use for routine measurements an ion gauge which is calibrated with the MKS SRG2. For cali-bration, the gas of interest is introduced into the 22-pole ion trap and both pressures (the spinning rotor gauge pressure PSRG, and the ion gauge pressure PIG22PT are recorded at 300 K. Plotting PSRG versus PIG22PT and fitting the data with the linear function PSRG = C⋅PIG22PT leads to the calibration factor C. The calibration factors for the gases used in this work, H2, D2, and He, are CH2 = 80, CD2 = 105, and CHe = 285. Note that they are only correct for the geometry and pumps of the present setup and for operating the ion gauge controller (AML) with a sensitivity of 19 mbar-1 and an emission current of 0.1 mA. Knowing the calibration factor C, the number density n of the neutral reactant is calculated from the indicated pressure PIG22PT (mbar) and the trap temperature T22PT (K) using the practical relation [ger92].

3

22

2217102.4 −×⋅= cmT

PIGCnPT

PT (2.3)

2 EXPERIMENTAL

24

The overall accuracy for the determination of the absolute number density inside the trap has been estimated to be in the range of 10 % to 15 %.

The situation is quite different if the H atom beam is used instead of leaking the gas directly into the trap. In contrast to the homogeneous density in radial direction, the beam diameter is smaller than the diameter of the ion cloud. In this case one can charac-terize the flux and shape of the neutral beam or one uses a known chemical reaction for calibration purposes. As described in detail in Appendix B, we use CO2

+ ions similar to Tosi et al. [tos84] and Scott et al. [sco97a]. The special feature is that CO2

+ produces predominantly HCO+ in collisions with H while reactions with molecular hydrogen lead to HCO2

+. For a complete characterization of the cold atomic beam we have extended the previous results towards lower temperatures [bor08c]. The set of reactions used for the calibration of number density of the H / H2 beam are presented in table 1 of Appen-dix B.

A typical set of data used for calibrating the H and H2 density, is shown in Fig. 2.9. In this case, the accommodator temperature was fixed at TACC = 12 K and the trap tempera-ture at T22PT = 100 K. With the discharge ON, an atom density of [H] = 3 × 108 cm-3 has been obtained in the trap while the density of molecules, [H2] = 6 × 108 cm-3, is two times larger. A slightly reduced molecular density, [H2] = 5 × 108 cm-3 has been meas-ured if the discharge is switched OFF. As discussed in detail in the next Chapter this seeming contradiction is explained with the confinement of H atoms in the hexapole and the additional production of molecular hydrogen via H-H recombination.

Fig. 2.9. Reaction of CO2

+ ions with H2 beam (discharge OFF, left panel) and with H / H2 mixture (discharge ON, right panel) have been recorded at TACC = 12 K and T22PT = 100 K. The average number of ions, Ni, trapped and formed for each iteration, is plotted as a function of the storage time t. Note that there is some minor background on mass 29 caused by some background gas. From the production rate of HCO+ and HCO2

+ and known rate coefficients, the number densities with discharge OFF [H2]OFF = 5 × 108 cm-3 and discharge ON ([H] = 3 × 108 cm-3, [H2]ON = 6 × 108 cm-3)) has been deter-mined.

25

3 COLD ATOMIC HYDROGEN BEAM SOURCE

Although a wide body of literature exists about the formation of molecular beams, one has to be aware that the required conditions for the production of an intense hydrogen atom beam, the aim of this thesis, differs significantly from the generally studied cases. The necessity to keep the H-H recombination rate low, requires beam forming systems operating at low densities (< 1017 cm-3) and large nozzle diameters. Conditions for the beam formation cannot be chosen a priori, they are a compromise imposed by other parameters, e.g. the acceptance volume of our trap. Depending on the gas density in the beam generating vessel and the nozzle geometry, different modes of beam formation are possible. In this section some gas-dynamical aspects of beam formation are summarized in order to interpret the experimental results.

3.1 Theoretical considerations 3.1.1 Theory of effusive gas sources Molecular flow regime prevails if the density in the orifice of the gas source is low enough such that the Knudsen number 1≥kn ( Dkn /λ= , with λ the mean free path and D the diameter of the source orifice). In this molecular flow mode there is no inter-action between the particles during and after the effusion. The differential beam inten-sity )(θI per unit solid angle element ωd at an angle θ (measured with respect to the forward direction) is given by a cosine distribution [pau88]:

dvdvvfAnI ωθθ )cos()()( 00= , (3.1)

where 0n is the gas density in the source, 0A the cross sectional area of the orifice and )(vf is the Maxwell-Boltzmann velocity distribution function, which is equal to

dvvv

vvdvvf

ww ⎥⎥⎦

⎤

⎢⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛−⎟⎟

⎠

⎞⎜⎜⎝

⎛= −

2

3

22/3 exp)( π . (3.2)

The values ( ) 2/1/2 mTkv Bw = is the most probable velocity of the particles at the source temperature T . The total flow rate Φ0 through the orifice can be obtained by integrating over all velocities and over the solid angle 2π and one obtains

sparticlesvAn /41

000 ⟩⟨=Φ . (3.3)

Here ( ) 2/1/8 mTkv B ⋅=⟩⟨ π is the average velocity of the particles in the source with a temperature T . The peak intensity )0( =θI of the beam is given by

=)0(I Φ0 / π particles / s⋅sr. (3.4)

The main disadvantages of an effusion source are the low maximum intensity restricted by the low density n0 due to the molecular flow condition 1≥kn and the poor directivity. The intensity in the forward direction is proportional to the total rate of flow through the orifice. The poor directivity of the beam formed by a simple orifice can be slightly im-proved if it is replaced by a long channel, generally with a cylindrical cross section. Restricting the flow in the entire cylindrical channel to the molecular flow regime re-sults in a loss of the beam intensity. A large intensity gain can be obtained if the mo-lecular flow condition is partly released. The atomic beam that results when the gas,


26

leaving the accommodator, expands into vacuum can be described by a model of Giordmaine and Wang [gio60] who successfully applied kinetic theory to the problem of molecular beam formation by long cylindrical channels which connects a source vol-ume to an ideal vacuum. This model assumed the density )(zn to fall linearly with the distance z along the channel )/()( 0 Lznzn = , where 0n is the density in the source, z is measured from the low pressure end, and L is the total length of the channel. The peak intensity of the beam is given by two contributions: (i) a first contribution is due to the particles which pass through the entire tube without any collision and (ii) a second con-tribution arises from particles which are scattered into the axial direction by intermo-lecular collisions in the tube. This model by Giordmaine and Wang has two specific modes which depend on the ratio of the particle mean free path λ at the gas density 0n in the source to the total length L of the tube. For a sufficiently low gas pressure in the source, only the process (i) contributes to the peak intensity )0(I of the beam which is equal to: π4/)0( 00 ⟩⟨= vAnI particles / s⋅sr. We can relate the peak intensity )0(I to the total flow rate tΦ through the tube by using the Clausing formula [cla30]

⟩⟨=Φ vnKAt 0041 , (3.5)

where the geometrical factor LDK 3/4= is the Clausing factor with D and L the di-ameter and length of the tube

KI t π/)0( Φ= particles /s⋅sr. (3.6)

Thus the peak intensity is directly proportional to the total flow rate through the tube. In this case, however, the gas flow in forward direction, compared to the total rate of flow, is superior to the one obtained by a simple orifice. At relatively high source pressures the particles have a negligible probability to pass directly through the tube without any collisions. In this case, the general relation for the peak intensity is reduced to

2/12/1

0 065.0 tdvI Φ⎟⎠⎞

⎜⎝⎛ ⟩⟨

=σ

, (3.7)

where ( ) 12

−= nλσ is the collision cross section of the particles [sin89]. Equation (3.7)

shows that the peak intensity is independent of the length of the channel. As a character-istic property of this specific mode of operation we observe that the peak intensity is now related to the total flow rate tΦ by a square root dependence and by a factor de-pending on the average velocity of the particles, respectively on the equilibrium tem-perature T of the gas in the source. For a given total flow rate, the theory implies a de-crease of the peak intensity as the temperature is lowered. This phenomenon can be more dramatic if one considers also the variation of the collision cross section with the temperature. From the above discussion, one concludes that, at a low enough density in the source, the peak intensity )0(I varies proportionally to tΦ and at sufficiently high

density proportionally to 2/1tΦ .

As the gas density in the source becomes sufficiently high such that the mean free path λ is small compared to the exit diameter, the gas moves in the viscous flow regime and an adiabatic expansion occurs downstream from the orifice. This type of beam forma-tion is very interesting for the production of high intensity beams and especially for monochromatic beams [her87b].


27

3.1.2 Mechanisms of H production and loss The degree of dissociation in a hydrogen discharge is determined by the production and loss mechanisms of atoms. These depend on macroscopic parameters such as the power transferred to the plasma, the gas conditions and the properties of the discharge vessel. Typically, the discharge is sustained by applying an electromagnetic high frequency field, which is coupled into a cylindrical dielectric discharge tube with the rf field appli-cator located outside. If the applied electrical field strength is sufficiently high, the gas in the discharge tube breaks down and a steady-state glow discharge is formed. Such a “cold” plasma is characterized by a low degree of ionization: ξ = ne / nn = ni / nn = 10-5 – 10-3, where ne, ni and nn denote the densities of the free electrons, the ions and the neu-trals, respectively. The low degree of ionization, together with the fact that the electron mobility is higher than the ion mobility, leads to a substantially higher temperature of the electrons compared to the neutral and ion gas. Typically, the temperature of neutrals and ions range between 500 K and 2000 K corresponding, in the case of hydrogen, to energies of 0.06 - 0.35 eV, while the electrons have mean energies of 2 – 10 eV. As a result, the discharge properties are dominated by the electron kinetics. The free elec-trons gain energy from the electromagnetic field and dissipate it via inelastic and elastic collisions with the neutral gas components. Table 3.1. H atom production mechanisms [ged93a]

In hydrogen plasma, H atoms can be produced by various reactions, e.g., collisions of molecular hydrogen with (i) electrons, (ii) ions and (iii) electronically-excited metasta-ble neutrals. Reactions involving electron-ion collisions give extremely low yields of H atoms due to the low ion density in the plasma (<0.01 %). The dominant electron impact processes in a hydrogen discharge are listed in Table 3.1. H atoms in the ground state are mainly created (90 %) by electron impact dissociation of H2 (R1) via the excitation of the first molecular triplet b 3Σu

+ state. The H atoms, resulting from dissociation of the b 3Σu

+ state, have kinetic energies between 2 and 4.5 eV. Wall collisions involving these energetic atoms are expected to have a higher probability of recombination than atoms with thermal energies. Wise and Wood have shown that the recombination coefficient has a very strong dependence on wall temperature [wis67]. The pressure required to thermalize these fast atoms has been estimated to be ~ 0.1 mbar [ged93a]. Excitation of H2 to the bound a 3Σg

+ state which subsequently decays to the repulsive b 3Σu + state by

radiative emission contributes with less than 10 % to the production of H atoms.


28

Also dissociative ionization (R2) or dissociative attachment of electrons to the ground state of H2 (R3) are not efficient for creating H atoms as can be seen from the rate coef-ficients in Table 3.1. Although H atom production via the ion molecule reaction H2

+ + H2 → H3

+ + H is fast, the small rate coefficient for H2+ formation by electron impact

(R4) impedes that this process makes a significant contribution. Atom production from H2(v = 1) by dissociative attachment or by direct dissociation via excitation of the re-pulsive b 3Σu is significantly less than that arising from reaction (R1). This dissociation process becomes more efficient for H2 in high vibrational states. H atoms will be formed in the gas phase reaction H(n = 2) +H2 → H(n = 1) +H2(b 3Σu

+ ) which is reso-nant in energy. The b 3Σu + molecular state is unbound and leads to the formation of H atoms with 2 – 4.5 eV kinetic energy as in reaction (R1). The efficiency of H atom pro-duction by the excitation transfer reaction will depend on the number density of atoms in the n = 2 excited state. The short radiative lifetime of the 2p state ensures that it de-cays before making a gas phase collision. The radiative lifetime of the metastable 2s state is well known to be critically dependent on electric field strength. This radiative lifetime was estimated to be 4 ×10-8 s suggesting a low density of H(n = 2) atoms. How-ever, a high atom density in the plasma can give rise to the imprisonment of the Lyman alpha resonance radiation which dramatically increases the mean lifetime of the 2p state and, consequently, increases its number density.

Hydrogen atoms are lost by surface or volume recombination of two atoms. For forming H2 in the gas phase, a third collision partner is needed to conserve energy and momen-tum. The H-H recombination process can be written as:

H + H + M → H2 + M + Erec, (3.11)

where Erec=4.478 eV is the binding energy of H2 and M is either another hydrogen atom or a molecule. Erec is transferred to the two products. The rate of loss by three-body re-combination is [ged93a]

333250 )83.1(1044.1 −+×= cmpDDkR mv (3.12)

where mk = 8.3×10-33 cm6s-1 is the three-body recombination coefficient with H2, D the dissociation fraction and p is the source pressure. Volume recombination losses are not relevant at pressures below 1 mbar [smi43], [ged93a].

In conventional H beam sources, surface recombination is the major loss mechanism of atomic hydrogen. For surface recombination two kinds of mechanisms have been pro-posed: (i) the Eley-Rideal mechanism involving the collision of a gaseous atom with an adsorbed atom on the surface and (ii) the Langmuir-Hinshelwood mechanism involving the collision of two adatoms through surface diffusion [wis67]. Based on the first mechanism, Gelb and Kim calculated the recombination coefficient of H atoms on glass surfaces [gel71]. The good agreement of their model with experimental results for the recombination of H on Pyrex or Quartz [woo62], [wis67] shows only a negligible con-tribution by the Hinshelwood mechanism and justifies the assumption of the predomi-nance of the Rideal mechanisms. In the following a short summary of this model is given. The rate of recombination jR per unit surface area is

jjjj ZMR θ2= (3.13)

where jM is the number of localized sites of type j per unit area, jθ is the coverage of a j type of site, and jZ is the reactive flux of gaseous atoms per adatom. The index j


29

indicates that a surface may have sites of different types, with correspondingly different values of the adsorption energy. The reactive flux jZ is given by

)/exp(00 TkEgZZ Bajjj −= σ , where (Eaj > 0) (3.14)

with the total collision rate 0Z per unit area )/(00 hTknZ Bλ= and the thermal wave length 2/1

0 )2( −= Tmkh Bπλ . The j0σ denotes the reactive cross section of the recombi-nation reaction, ajE is the activation energy, n the gas density, m the mass of the atom and the degeneration factor g is 1/4. The coverage jθ can be calculated by the Lang-muir’s adsorption isotherm

)1/( 30

30 jjj QnQn λλθ += (3.15)

where jQ is the partition function for an atom localized in a j type of site. The parti-tion function jQ may be approximated by that of a three-dimensional harmonic oscilla-tor with an appropriate truncation.

3

300

))/exp(1())3/exp(1)(/exp(

TkhTkETkE

QBj

BjBjj υ−−

−−= (3.16)

where jE0 is the heat of adsorption and jυ is the mean frequency of the harmonic oscil-lator. The recombination coefficient jγ is defined by the ratio of the recombination rate to the total collision rate,

0/ ZRjj =γ . (3.17)

In the low-temperature regime where 1≅jθ , jγ takes the form

)/exp(2 0 TkEgM Bajjjj −≅ σγ (3.18)

which is a monotonic increasing function of temperature and is independent in n . Thus, jR is first order in n for this region. On the other hand, in the high-temperature region

where 1<<jθ and 1/ <<TkE Baj , on obtains

3

3003

00 ))/exp(1())3/exp(1)(/exp(

2Tkh

TkETkEngM

Bj

BjBjjj υ

λσγ−−

−−= (3.19)

which is a monotonically decreasing function of temperature and is linear in n . Then jR is quadratic in n . From the functional behavior of jγ in the two extreme regions of

temperature has been concluded that jγ has one maximum at an intermediate tempera-ture. There is a special situation where n and ajj EE /0 are so large that the temperature region where 1≅jθ overlaps with the region where 1)/exp( ≅− TkE Baj . Then, about the transition temperature jT the recombination coefficient jγ takes a plateau value given by

jjj gM 02 σγ ≅ (3.20)


30

which is independent of temperature and density. This type of special situations actually occurs in the cases of hydrogen recombination on certain metal surfaces [wis67]. When more than one type of adsorption sites is active, the contribution may be added inde-pendently, so that the total recombination coefficient is given by

∑=j

jγγ (3.21)

A situation in which two kinds of binding sites are available is found in hydrogen re-combination on glass surface. Measurements of the recombination rate of hydrogen at-oms on glass surfaces have been made over a wide temperature range (77-1123 K) by several authors [woo62]. According to de Boer and Van Steenis the glass surface shows to types of sites [boe72], (i) one with a strong binding energy (E01 = 44 - 48 kcal/mol) and (ii) another one with a weak binding energy (E01 = 2 - 5 kcal/mol) on top of the first layer. The second one contributes only at low temperatures. If the activation energy for the recombination process is estimated by the Hirschfelder relation [hir41]

jaj EE 005.0= (3.22)

then 1aE and 2aE also differ considerably. Thus, each layer can be considered as inde-pendent and the rate of recombination γ may be calculated by the sum of γ1 and γ2.

The left panel of Fig. 3.1 shows experimental data and calculations for Pyrex and quartz taken from the literature. The best fit of the theoretical expression with the given value of n (1014 cm-3) is obtained for E01 = 42 kcal/mol and E02 = 2.1 kcal/mol. For these val-ues theoretical maxima of γ occurs at T1 = 883 K and T2 = 48.5 K and a minimum at Tm = 111 K. Based on the above model, Singy et al. have calculated the recombination co-efficient of H atoms on copper [sin90]. The copper surface also exhibits two types of sites with bindings energies E01 = 66 kcal/mol respectively E02 = 160 K [wis67]. The right panel of Fig. 3.1 shows the hydrogen recombination coefficient on a copper sur-face. Because of the high binding energy of the first site, the surface is already saturated at room temperature leading to almost flat behavior of γ. This behavior is in good agreement with the experimental results reported by Wise and Wood [wis67].

Fig. 3.1. Hydrogen recombination coefficient on Pyrex (left panel) and copper (right panel) [sin90]


31

3.1.3 Guiding and focusing H atoms The principle of separating atoms in a Stern-Gerlach magnet is fully described in the literature [hae67], [bre31]. The most fundamental example is the separation of hydrogen atoms which is based on the interaction of the gradient of the magnetic field with the magnetic momentum of the atom. Hydrogen atoms in the ground state have an electron spin S = 1/2 with a z-component ms = ±1/2 and a proton spin I = 1/2 with mI = ±1/2. Electron and proton spin couples to the total spin F = S + I = 0 or F = 1 with mF = -1, 0, +1, respectively. The energy difference between the states with F = 0 and F = 1 is called hyperfine splitting energy hfsE , correspond to a transition frequency of 1420.4 MHz.

Fig. 3.2. Breit-Rabi diagram for a hydrogen atom in a magnetic field. The magnetic field is plotted in units of the critical field cB = 50.7mT. The energy is measured in units of the reso-nant frequency ν0 = 1420.4 MHz . For details see text.

If an external magnetic field B is applied, the interaction with the magnetic momentum of the atom leads to an energy splitting of the F = 1 state according to the Zeeman ef-fect. The energy splitting E due to an external magnetic field is given by the Breit-Rabi formula [bre31]

2

1

1241

2)1(

24 ⎟⎟⎠

⎞⎜⎜⎝

⎛+

++−++

+−= +

cc

FhfsFFBI B

BBB

ImE

BmgIEhfsE μ (3.23)

where cB is called critical field and is defined as

mTgg

EhfsBBIS

c 7.50)(

=−

=μ

(3.24)

with Sg = 2.002, Ig = 0.00304 and Bμ = 9.274078 ×10-24 J/T are the g-factors of the electron and the proton and the Bohr magneton, respectively. For high magnetic fields, the interaction with the magnetic moment of the atom is dominated by the interaction with the magnetic moment of the electron as a result of the decoupling of S and I and the about 1837 times higher magnetic momentum of the electron. Fig. 3.2 shows the energy-level diagram of a hydrogen atom in a magnetic field. It can be seen from the


32

plot that in the inhomogeneous magnetic field of a hexapole magnet the energy E of the atom becomes a function of its position. The force the atom experience is given by

gradEF −= . (3.25)

This force can be used to spatially separate atoms according to their electron spin. In a hexapole magnet B has a radial symmetry,

2

00/)(/ ⎟⎟

⎠

⎞⎜⎜⎝

⎛=

rrBrB , (3.26)

where 0r is the radial distance of the pole tips in respect to the symmetry axis of the magnet and 0B is the magnetic field on the pole tip. Passing through the hexapole mag-nets, the atoms experience a force

rrBrBgradF BB 2

0

02/)(/ μμ == (3.27)

For a hexapole magnet with parallel poles pieces ( 0r and 0B constant), it is possible to find an analytical solution for the differential equation of motion which is given by

rrB

dtrdm B ⎟⎟

⎠

⎞⎜⎜⎝

⎛±= 2

0

02

2

2μ (3.28)

Considering an atom entering the magnet at 0=z at a radial distance 1)0( rzr == , in respect to the symmetry axis z of the magnet, with a small angle α and a velocity v , one can use the approximation vzt /~ in Eq. (3.28) which leads to the following equa-tion of motion for the atoms through the hexapole magnet

rrmvB

dzrd B

20

20

2

2 2μ±= (3.29)

From Eq. (3.29) the trajectory of a ms = +1/2 atom can be calculated as

)sin()cos()( 1 zzrzr ⋅⎟⎠⎞

⎜⎝⎛+⋅= λλαλ with 2

02

02 2rmvBBμλ = (3.30)

According to the geometrical optics, a hexapole magnet can be considered for a fixed velocity of the atoms as a thick optical lens. In order to emphasize the relationship exist-ing between the characteristics of the beam transportation through the magnet and the focal properties of an optical lens, it is convenient to introduce a linear transfer matrix which acts on the individual trajectories of atoms. Using the Eq. (3.30), the transfer ma-trix hexM , corresponding to the beam transportation through the hexapole magnet for atoms with ms = +1/2 is given by

⎥⎦

⎤⎢⎣

⎡=⎥

⎦

⎤⎢⎣

⎡−

=2221

1211

)cos()sin(/)sin()cos(

mmmm

LLLL

MSs

SShex λλλ

λλλ (3.31)

with SL the lengths of the magnet. The coordinates of the individual trajectories at the exit of the magnet are given as a linear combination of the input coordinates for 0=z


33

⎥⎦

⎤⎢⎣

⎡=⎥

⎦

⎤⎢⎣

⎡)0()0(

)()(

'' rr

MLrLr

hexS

s (3.32)

where )(' zr is the slope of the trajectory as a function of the distance z on the axis. There is a direct relationship between the elements of the transfer matrix and the focal properties of a thick lens. In the case of the hexapole magnet this relationship is ex-pressed by

01

211

0 /1))sin((

ff

mLf S

−=

=−= −λλ (3.33)

with 0f and 1f , the focal lengths of the object and the image respectively. From the transfer matrix, one deduces furthermore the relation between the distances of the object ( 0L ) and image ( 1L ) planes

)cos()sin(

/)sin()cos(

0

01

SS

SS

LLLLLLLλλλλλλ

+⋅⋅−+⋅

−= . (3.34)

Here 0L and 1L are given with respect to the entrance and the exit of the hexapole mag-net, respectively. The above equations show the velocity dependence of the trajectories of atoms passing through a hexapole magnet. Since the velocity distribution of an atomic beam has a finite width one has to consider besides the most probable velocity of the beam the effective velocity distribution in order to establish the focusing properties of the hexapole magnet.

3.2 Technical description 3.2.1 The radio frequency dissociator A discharge in a hydrogen gas is a convenient method for producing an intense beam of hydrogen atoms. Both radiofrequency and microwave exited discharge sources can, in favorable circumstances, be operated over long periods of time with high efficiency. In the present work a commercially available Slevin-type rf plasma source is used, because of its very high dissociation fraction (near 95 % in the discharge tube) and the long time stability [sle81]. Dissociation of hydrogen molecules take place in a water cooled Pyrex tube, with an inner diameter of 19 mm and length of 330 mm. A 2 mm diameter capil-lary at each end of the tube effectively confines the discharge, while at the output end, the beam emerged through a capillary 20 mm long and 2 mm in diameter. The discharge is excited by feeding about 20 W of rf power into a coaxial cavity resonant at 27.12 MHz. The cavity consists of a copper cylinder, surrounding the discharge tube, and a 17-turn helix of constant pitch. Details of the cavity design and dimensions are described in [alp59].

The gas flow through the discharge tube is controlled by a needle valve allowing an accurate adjustment of the throughput. A parallel pipe, connected to the main feed sys-tem just at the entrance of the discharge tube, is used for addition of water vapor to the hydrogen gas. The source pressure is measured at the entrance of the discharge tube using a capacitance manometer (see Fig. 2.4). The specified purity of the gases used is 99.999 % for H2 and 99.7 % for D2. The stability of the discharge and the degree of dis-sociation dependent critically upon both the steady flow of bubble-free cooling water and the cleanliness of the Pyrex tube. Therefore, the cooling water is turned on several


34

hours before the source is used. A variety of methods have been proposed for the clean-ing of the discharge tube [don92]. The following method has been found the most con-venient: the discharge tube has been cleaned in sequence with boiling ethanol, boiling methanol and filled with 20 % Phosphoric acid for 24 hours. Between each step it has been flashed with deionized water. Once a clean surface is obtained, it is important to keep it clean under operating conditions. In this respect the source pressure was main-tained positive with respect to the chamber pressure at all times, in order to reduce the back diffusion of any contaminants. Also, a liquid nitrogen trap was used in the backing line when the system was being roughed from atmospheric pressure and the gas inlet was maintained at a temperature of 60 °C during operation.

3.2.2 The nozzle cooling system In the last two decades substantial effort has been made to cool the beam of H atoms from room temperature down to ~ 5 K in order to obtain high number densities. The accommodation of H atoms to the temperatures of 8.5 K has been observed in a Pyrex tube (inner diameter 4 mm, 20 mm long, coated with Teflon) [wal82].

Fig. 3.3. Details of our nozzle cooling system: (1) discharge tube, (2) precooler hold at 100 K, (2b) temperature gap, (3) copper accommodator, (4) copper bridge, (5) heater, (6) Si diode and (7) 2 mm skimmer.

For slowing down the H atoms emerging from the hot discharge to velocities which are representative for the low temperatures prevailing in the ISM, cryogenic cooling has been used. The accommodation is performed by the collisions of the atoms on the cold surface of the accommodator. In order to reduce the temperature gradients in the ac-commodator which can occur due to recombination, leading to an additional thermal load, copper accommodators have generally been used. As indicated in Fig. 3.3, the hydrogen atoms pass first through the Pyrex nozzle followed by a channel in a copper accommodator, the temperature of which can be set at values between TACC = 10 - 300 K. The total length of the channel in the accommodator is 22 mm. Initial experi-ments have been performed without precooler and a smaller channel diameter, 1.2 mm instead of 2 mm. The exit capillary of the discharge tube was inserted 6 mm in the cop-per accommodator. In order to test the efficiency of our cooling system, absolute beam density measurements have been carried out (see Sec. 3.3). The tests have proven that the gas flow accommodates efficiently to the accommodator temperature for the smaller


35

diameter of 1.2 mm while for the larger diameter of 2 mm thermalization was not com-plete. Further improvements of the cooling system have been made during the PhD. work of C. Mogo. As shown in Fig. 3.3, a precooler surrounding the exit capillary hold at 100 K and a temperature gap have been installed. After installing the precooler veloc-ity distribution measurements have shown that sufficient cooling is achieved with the diameter of 2 mm. More details concerning the accommodator characterization can be found in Sec. 3.3.

In order to reach accommodator temperatures as low as 10 K, a two-stage closed cycle He refrigerator cold head (Leybold, RGD210) with a He compressor (Leybold, RW2) has been used. This system provides a cooling power of 2 W at 10 K. A copper bridge with a total length of 12.7 cm connects the cold surface of the cold head (not shown in Fig. 3.3) with the accommodator. Depending on the temperature of the cold head, ther-mal expansion changes the position of the nozzle. The bellows and a suitable mechanics allow for readjustment under visual control with a telescope.

In order to transfer efficiently the cooling power from the cold head to the accommoda-tor a given cross section of the bridge components is needed. It can be estimated as

)()(

csACCb

radrecbb TT

PPLA−+

=κ

(3.35)

where recP is the heating power due to recombination of H atoms on the accommodator,

radP is the heating power due to radiation from the environment, bL = 12.7 cm is the total lengths of the cooling bridge, bκ = 8 W/(cm K) is the specific heat conductivity of the cooper bridge at 100 K and ACCT and CST are the temperatures of the accommodator and the surface of the cold head. At the gas throughputs the source is operated at (10-3 – 10-2 mbar ls-1), the heating power released during the thermalization of the gas to the accommodator temperature is negligible. Also the heating power carried through the discharge tube was found to be negligible. In order to estimate the heating power due to recombination, the following assumptions have been made: (i) the degree of dissocia-tion is α = 80 % at the end of the discharge tube, (ii) the gas throughput is ACCQ = 5 × 10-3 mbar l/s, and (iii) the surface recombination on the accommodator is 100 %. Under these extreme assumptions one gets the heating power due to recombination,

127.02 == recACCrec EQP α W. (3.36)

The recombination energy, Erec = 4.478 eV, is equal to the binding energy of a hydrogen molecule. The power caused by radiation is a function of the surface area of the cooling bridge and of the inner wall of the vacuum chamber surrounding the dissociator as well as of the respective emission coefficients. It has been estimated to be radP = 0.18 W. Assuming a temperature of the cold surface of CST = 10 K at a heating power of 0.3 W, a cross section area of about bA = 2 cm2 is required to cool down the accommodator to a temperature as low as ACCT = 10.3 K.

The time for cooling down the accommodator has been estimated as well. The mass of the cooling bridge and accommodator has been determined to be m ≈ 570 g. Hence the time to cool the accommodator from room temperature to 10 K is

hoursP

mctt Cudowncool 3)10300(

0 ≈⟩⟨−

+=− , (3.37)


36

where 0t = 0.5 hours is the cooling down time for the cold head without the copper bridge, cCu = 0.38 J/ (g K) is the specific heat capacity of copper and ⟩⟨P ≈ 7 W is the average cooling power provided by the cold head during the cooling process. There is a good agreement between estimated and measured times. The temperature of the ac-commodator is measured with a precision of ±0.25 K by a calibrated silicon diode (LakeShore Cryotronics, model DT-470) mounted directly onto the accommodator sur-face. Further, a heating source is fixed onto the cooling bridge (thermocoax with R = 3.6 Ω and PMax. = 20 W) which provides stable accommodator temperatures in the range 10 - 350 K. For accommodator temperatures around 100 K, the preparation and mainte-nance of a cryogenic ice layer is crucial for the efficient and stable operation of the cooling system [koc99a]. After a given time the properties of the ice layer degrade and the water coating procedure has to be repeated. Therefore, the accommodator is warmed up to room temperature to get rid of the ice layer absorbed on it. Using a heating power of 12 W the warming up time is about 50 min.

3.2.3 The hexapole magnets Hexapole magnets are commonly used in atomic beam sources because of their har-monic focusing properties. During the passage through a hexapole magnet, atoms with a proper orientation of the electron spin (spin up) are attracted towards the center line leading to oscillatory trajectories, while atoms with spin down are repelled towards the magnetic poles and lost from the beam. Hexapole magnets for atomic beam sources come in various forms: as conventional electromagnets [sin90], permanent magnets [szc00], or superconducting magnets [isa98], [sta05]. To date, best performance results for atomic beam sources were obtained with permanent magnets.

Fig. 3.4. Front view of the first hexapole magnet, consisting of 6 sectors. The arrows in the sectors indicate the orientation of the magnetization.

Two permanent hexapole magnets for guiding and focusing the H / D atoms are located in the second and third vacuum chamber (see Fig. 2.2). The geometrical dimensions of our two hexapole magnets, skimmer and apertures in the set up with ionizer or 22PT are listed in Table 3.2. The first hexapole magnet was fabricated by the Electra Trieste and the second one by the Vakuumschmelze GmbH. The magnets are made from rare earth cobalt materials with highly remanent magnetization. After assembly the magnets were encapsulated in a vacuum-tight stainless steal can of a cylindrical shape. This canning is necessary to prevent degradation of the magnet material by exposure to hydrogen. The


37

first magnet has a length of 150 mm, inner diameter of 10 mm, outer diameter of 50 mm and a hexapole coefficient of 55.240 Tm-2. It is formed from 6 sectors with a straight bore, as shown in Fig. 3.4. The second magnet is formed from 24 sectors with straight bore and has an inner diameter of 14 mm, outer diameter of 36 mm, a length of 105 mm and hexapole coefficient of 30.880 Tm-2. The arrows in the sectors indicate the orienta-tion of the magnetization, which is rotated by 120o from sector to sector for the first hexapole magnet and 60o for the second hexapole magnet. This configuration results in a hexapolar magnetic field in the bore.

Fig. 3.5. Calculated trajectories of D atoms with mj = +1/2 for the three velocity groups observed by the TOF analysis (590 ± 80) m / s (upper figure), (902 ± 195) m / s (middle figure), and (1410 ± 590) m / s (lower figure). The plots give the radial distance of the at-oms from the symmetry axis of the magnets at different positions along the axis. The no-menclature used is explained in Appendix E.

The magnetic field in the bore of the magnet can be calculated from the radius of the bore, r1, the outher radius of the magnet, r2, and the remanent field of the magnet mate-rial remB , using formulae derived by Halbach [hal85]. For a hexapole magnet consisting of M wedge-shaped magnets one finds:


38

M

MMrr

rrBrB rem /3

)/3sin()/(cos15.1)( 32

2

21

2

1 πππ⎥

⎦

⎤⎢⎣

⎡−⎟⎟

⎠

⎞⎜⎜⎝

⎛= (3.38)

For the same magnet material and geometry, the field of a magnet consisting of 24 seg-ments is a factor 2.3 larger than a 6 segment magnet.

For obtaining high number densities of atoms and low background in the 22-pole trap the position of the hexapole magnets, skimmer and apertures have been optimized. As can be seen in Fig. 3.5 the usable phase space at the exit of the first hexapole magnet is limited by the aperture ME1 introduced for differential pumping, and at the exit of the second hexapole magnet by the aperture ME2 introduced to collimate the atoms through the 22-pole trap. For a detailed understanding of these limitations many trajectory of atoms with mj = +1/2 passing through the present two-magnet system have been calcu-lated using the model described in Sec. 3.1.3 and the experimentally determined veloc-ity distributions [bor08a], [mog09]. For both hydrogen and deuterium, time of flight measurements (TOF) revealed that the hexapole magnet system favor transmission of three velocity groups with energies of 3.6 meV, 8.2 meV, and 21 meV. A few selected trajectories illustrate the situation in Fig. 3.5. They have been calculated for D atoms starting on the beam axis with an angle of 0.8° and velocities of (590 ± 80) m/s (upper figure), (902 ± 195) m/s (middle figure) and (1410 ± 590) m/s (lower figure). Inspection of Fig. 3.5 shows that atoms with a kinetic energy of 8.2 meV are well-collimated through the 22-pole trap while atoms with a lower kinetic energy of 3.6 meV or a higher kinetic energy of 21 meV enter the 22-pole trap at larger angles.

3.3 Test measurements

Fig. 3.6. Experimental set-up for the characterization of the H atom beam

In order to determine the number density of H atoms at the location of the 22-pole trap and for characterizing the H atom source concerning recombination, stability, velocity distribution, etc. the trap has been replaced by a commercial ion source (see Fig. 3.6). Both the neutral beam and background gas are ionized by the electron bombardment ion source, mass analyzed in the quadrupole mass spectrometer and detected via the Daly detector. Using a mechanical shutter placed between the second hexapole and the ion-izer, it is possible to separate the direct beam from other contributions. The sensitivity


39

of the detector has been calibrated by leaking in effusively up to 10-5 mbar of H2 / D2 into the chamber and by comparing the ion signal with the pressure, determined with a spinning rotor gauge. Typically, emission currents of 0.1 - 5 mA have been used. For an electron energy of 50 eV the sensitivity for D2 is 0.2 mbar-1. Electron induced fragmen-tation is slightly reduced (from ~5 % to ~ 1 %) if the electron energy is lowered to 20 eV resulting in a sensitivity of 0.01 mbar-1. The error of the sensitivities is smaller than 5 %; however, an additional systematic error is due to the fact, that the directed beam is detected more efficiently than gas with random motion. Proper tuning of the ion source may lead to a rather small difference. The results have not been corrected for that leading to an overall uncertainty of ~ 30 %.

A residual gas analyzer was mounted onto the source chamber for optimizing the water coating procedure by monitoring the amount of water desorbed from the cold accom-modator during the warming up cycle.

3.3.1 Degree of dissociation measurements In order to avoid confusion it must be stated that there are several locations for deter-mining the dissociation degree. In the discharge tube, values close to 100 % can be achieved, the value at the exit of the source is usually slightly smaller. At the location of the ions trap, or in the ion source, it is better to talk about the H / H2 ratio which in-cludes also the molecular background originating from other sources.

With the arrangement shown in Fig. 3.6, the H / H2 ratio has been determined. For get-ting the same transmission properties for hydrogen atoms and molecules, the hexapole magnets have been taken out. To avoid mass discrimination effects, the quadrupole mass spectrometer was tuned to the molecular masses of H2

+ or D2+, and the ion signal

IOFF_beam (proportional to the beam flux with discharge OFF) and ION_beam (proportional to the beam flux with discharge ON) has been measured. The degree of dissociation is defined as [cul93]

beamOFF

beamON

II

_

_1−=α (3.39)

Fig. 3.7. Typical mass spectra for deuterium beam with open (left panel) and close (right panel) shutter measured for switched OFF and running discharge.

Four typical mass spectra are compared in Fig. 3.7. They have been recorded for a deu-terium beam with a throughput of 2.25 × 10-3 mbar l s-1, an rf power of 20 W, and with


40

an optimal ice layer at TACC = 100 K. The left side shows the ion signal measured for discharge OFF and with open shutter (proportional to the total flux of D2 molecules) and closed shutter (proportional to the flux of D2 background). Switching the discharge ON (right panel), the D2 flux is converted into a D / D2 mixture. Fragmentation of D2 leads to a decrease of D2

+ ion signal while the newly formed D atoms lead to the production of D+ ions. Using formula (3.39) a degree of dissociation α = 93 % has been determined at the location of the ion source. Under optimal operating conditions of the H atom source, we estimate the relative uncertainty of α to be 2 % for α = 30 – 95 % and 5 % for α = 2 – 10 % With the method outlined here, the degree of dissociation for the Slevin-type source and the source with the accommodator has been measured as a func-tion of gas throughput, rf power and different surface coating of the accommodator.

Slevin-type source The first set of measurements has been carried out for the Slevin-type source. The de-gree of dissociation has been measured for commercial grade deuterium without addi-tives to the plasma discharge. The measurements were carried out using a simple set up, one differential pumping system, with the source placed in axial direction and the dis-tance between the exit capillary to ionizer of 8 cm. Fig. 3.8 shows the dependence of the dissociation degree on the source pressure. The deuterium discharge has been operated with an rf power of 15 W. The discharge tube used in this measurement has an exit cap-illary with a diameter of 2 mm and lengths of 45 mm. The highest degree of dissociation α = 80 % can be seen to occur for a source pressure of around 0.175 mbar. This type of source has been used and characterized in several laboratories. As was already dis-cussed in Sec. 3.1, H-H recombination on the Pyrex tube is the dominant loss to the degree of dissociation. Depending on the recipes used for the cleaning of the discharge tube, gas purity and vacuum conditions, large variations of the degree of dissociation have been observed. For example, using commercial grade hydrogen a degree of disso-ciation of 65 % has been measured while for purified hydrogen values around 90 % could be observed at otherwise same conditions [don92]. An extremely high degree of dissociation, 95 %, has been reported in [sle81] using purified hydrogen. Somewhat lower values between 80 and 85 % have been reported in [cha88].

Fig. 3.8. Dependence of the degree of dissociation α on source pressure, measured for commercial grade deuterium and using an rf power of 15 W. Note that the Pyrex nozzle with an exit capillary of 2 mm in diameter and 45 mm length was maintained at room temperature and no additives were mixed to the deuterium gas.


41

Investigation of recombination losses on the accommodator Fig. 3.9 illustrates the dependence of the degree of dissociation α on the source pressure for both hydrogen and deuterium operating the discharge with an rf power of 20 W. The accommodator temperature was 100 K and its surface was coated with an optimal layer of ice. The degree of dissociation for hydrogen and deuterium are αH = 92 % and αD = 90 %, respectively. This demonstrates a high dissociation efficiency for the Slevin-type source and low recombination losses on the accommodator surface. From the absolute increase in dissociation degree caused by the water admixture, it can be concluded that the water reduces not only wall recombination rate in the accommodator but also may have an effect on the interior surfaces of the discharge tube. This behavior has been ob-served in an earlier experiment for a microwave source running with 5 % D2O in hydro-gen [mil74]. More recent studies [dal98], [koc99a], [szc00] have shown that a small amount of oxygen (0.1-1 flux %) or water (0.5 – 1 %) admixture to the hydrogen or deuterium result in an increase dissociation efficiency at gas throughputs up to 3 mbar ls-1.

Fig. 3.9. Degree of dissociation for hydrogen and deuterium as a function of source pres-sure at TACC = 100 K and rf input power of 20 W. In order to reduce H-H recombination on the copper accommodator its surface was coated with ice.

The dependency of the dissociation degree on accommodator temperature is shown in Fig. 3.10, both for hydrogen and deuterium and using an optimal ice layer. The source was running at optimum pressure and rf power for both gases. In this case the dominant effect is H-H recombination on the accommodator surface. The results plotted in Fig. 3.10 demonstrate that large effects occur if the temperature of the accommodator is varied. In the temperature range TACC = 100 – 130 K, the dissociation degree in hydro-gen and deuterium reaches a flat top of about 95 % indicating a very slow recombina-tion rate. At higher temperatures, TACC = 130 – 230 K, a steep decrease is observed for both gases with a faster recombination rate for hydrogen than for deuterium. For TACC > 230 K, a degree of dissociation ~ 3 % has been observed indicating very fast recombina-tion on the copper accommodator for both gases. The observed decrease in α can be explained by desorption of the ice layer formed on the accommodator when the tem-perature is increased (see Fig. 3.14). For TACC < 100 K, a steep decrease in α has been observed for both gases with a lower recombination rate for hydrogen. Around 20 K, a


42

minimum in α has been found and, approaching 10 K, the degree of dissociation in-creases to values around αH = 10 % and αD = 5 %. This increase is attributed to the ab-sorption of hydrogen molecules on the accommodator surface at temperatures below 12 K, resulting in a good recombination inhibiting surface [wal82], [her87b]. The observed behavior of the degree of dissociation is also in good agreement with recently published results [koc99a].

Fig. 3.10. Dependence of the degree of dissociation α for hydrogen (filled circles) and deuterium (filled and open triangles) on the temperature of a copper accommodator. For each measurement the water coating has been made at an accommodator temperature of 100 K. Note that the measurements for hydrogen were performed in one day and those for deuterium in three consecutive days.

Fig. 3.11. Dissociation degree α of deuterium as a function of accommodator temperature at different conditions. The data plotted as solid circles have been measured for an opti-mal coating. The open circles are data obtained after the water was removed from the ac-commodator. The triangles represent data for a discharge tube cleaned with distilled water and pumped down for two days. The squares are data measured for a clean discharge tube after two weeks of pumping down.

The influence of various methods to prepare the surface of the accommodator is illus-trated in Fig. 3.11. The data, plotted with solid circles, have been obtained with an op-timum ice layer while the open circles data have been measured two hours later with the ice layer removed by heating the accommodator at 50 °C. The triangles are data meas-


43

ured for a discharge tube rinsed in distilled water at the last step of the cleaning proce-dure and pumped down for two days and the filled squares are data for the same dis-charge tube after two weeks of running time. It is obvious from Fig. 3.11 that the quality of a discharge source dependents critically on the preparation of the surface. For exam-ple, at TACC = 100 K the dissociation fraction obtained with a not prepared surface was 33 % while optimal conditions lead to 95 %. The decrease in α during the pumping down time of the discharge tube rinsed in distilled water suggests that de degree of dis-sociation depends on the amount of water in the vacuum system. From these observa-tions one can conclude that the ice layer formed onto the cold accommodator surface determines the wall recombination properties in the accommodator.

Fig. 3.12. Decay of the dissociation degree α for deuterium gas after several hours of op-eration. The differences are due to different coating times which are defined by the open-ing time of the leak valve to the water reservoir, at 100 K accommodator temperature, source pressure 0.1 mbar and rf input power of 20 W. Details are explained in the text.

Important for trapping experiments is the stable operation of the beam source over many hours. Fig. 3.12 shows the dependence of the dissociation degree of deuterium for a source running for hours under optimal operating conditions (PBT = 0.1 mbar, P = 20 W). The accommodator temperature was set to 100 K and different ice layers were formed on to accommodator surface depending on the opening time to the water reser-voir. Comparison of the 3 curves reveals that the stability of the degree of dissociation depends on the thickness of the ice layer formed on the accommodator. Also the mor-phology is expected to play a role. Exposing the accommodator for 6 min to water vapor at a pressure of 0.066 mbar, we have obtained a dissociation fraction of 95 % at the start of the measurement with a loss rate of only 1 % per hour. Using only 4 min. the meas-urements start with 90 % and decreases to 57 % over a period of 12 hours. Increasing the exposure time to 8 min. a thick ice layer of inadequate structure is formed on the accommodator resulting in a drastic decrease in the dissociation fraction after 8 hours of running time. Such behavior also has been reported in [koc99a] for a discharge operated with an admixture of about 2.5 flux % oxygen. Fig. 3.14 shows the partial pressure of water for this measurement, detected by the residual gas analyzer during the warming up phase of the accommodator (thermal programmed desorption).

Water coating procedure As was originally observed by Singy et al. and also confirmed in our experiments, if no special precautions are used, the behavior of the recombination coefficient of hydrogen


44

atoms on the accommodator surface can be attributed to the formation of a stable ice layer [sin90]. Therefore, the preparation and maintenance of a cryogenic ice layer is crucial for the efficient and stable operation of the cooling system. Since our source operates better at gas throughput in the range of 10-3 mbar ls-1 the water coating proce-dure developed for polarized atomic beam sources, which usually operates of gas throughput in the range of several mbar ls-1, cannot be simply applied [koc99a], [szc00].

Fig. 3.13. Water coating procedures at TACC = 100 K. Plotted is the pressure at the en-trance of the discharge tube PBT and the pressure in the H atom source chamber PIGHAS as a function of time.

In order to implement this method for our source it was necessary to form the ice layer with discharge OFF and a large amount of water vapors were added to the H2 or D2 gas. Flow measurements were performed for H2 and D2 at different accommodator tempera-tures (see next section). Also the H2O flow was measured at TACC = 300 K. It is gener-ally assumed that a suitable ice layer can be formed on the accommodator surface when its temperature is about 100 K. Fig. 3.13 shows our optimized method for the formation of specific ice layers at TACC = 100 K. Plotted is the pressure at the entrance of the dis-charge tube PBT and the pressure in the H atom source chamber PIGHAS as a function of time. Whit discharge switched OFF, the throughput of D2 is fixed at QACC_D2 = 3.6 × 10-

3 mbar ls-1. The resulting pressures are PBT_D2 = 0.25 mbar and PIGHAS_D2 = 5 × 10-7 mbar. When the pressure is stabilized, the valve to the water reservoir is opened for 6 minutes and water vapor is added to the D2 gas at a throughput of QACC_H2O = 4.2 × 10-

3 mbar ls-1. The partial pressure at the entrance of the discharge tube was PBT_H2O = 0.066 mbar. Following the traces in Fig. 3.13 reveals that, as soon as the valve is opened, some water escapes into the vacuum chamber for a short time. After 6 min the valve is closed. Subsequently, the D2 flow is increased to QACC_D2 = 7.2 × 10-3 mbar ls-1 for 30 min. The corresponding pressures are PBT_D2 = 0.46 mbar and PIGHAS_D2 = 1×10-6

mbar. Finally, the operational pressure is fixed at PBT_D2 = 0.12 mbar and the discharge is turned ON.

Using this procedure, a high degree of dissociation has been obtained for both hydrogen and deuterium. In order to maintain a high dissociation degree the surface coating has to be regenerated after one day of running time. After warming up the accommodator to TACC = 300 K, the described water coating procedure is repeated. Fig. 3.14 shows the pressure in the discharge tube and in the H atom source chamber during the warming up cycle of the accommodator after water has been deposited for 6 min. In order to analyze the discharge products frozen onto the accommodator during continuous operation, the


45

composition of the beam desorbed from the accommodator surface has been determined using the residual gas analyzer, mounted on the source chamber. No partial pressure signals apart from water could be observed. As can be seen from the time dependence the first amount of water is desorbed at TACC = 150 K. A maximum is observed at TACC = 240 K. Various measurements of the water partial pressure desorbed from accommo-dator in combination with long time measurements for the degree of dissociation has been used to optimize the water coating procedure.

Fig. 3.14. Pressure in the discharge tube (left scale) and H atom source chamber (right scale) during the warming up of the accommodator. After one day of running, the ac-commodator must be heated up for regenerating the ice layer. During the measurement shown, no D2 was running through the source.

3.3.2 Number density at trap location In this subsection three methods are used for the determination of number density to the 22-pole trap position. Gas flow measurements, ionizer with QPMS and chemical prob-ing in the trap itself.

Determination of number density using kinetic theory of gases The flux calibration for the H atom source is based on the measurement of the pressure decay in a calibrated volume CALV (see Fig. 2.4). As was already discussed in Sec. 2.2 the pressure in the calibrated volume is measured using an ultra high purity pressure transducer (Swagelok, PTU) specified to have a relative accuracy of 0.036 %. The source pressure is measured at the entrance of the discharge tube by a capacitive gauge (MKS, 722A02MCD2FA). It is safe to assume that the mass flux through the leak valve and the exit of the accommodator is conserved. As a result the throughput ACCQ out of the accommodator is given by the time evolution of the pressure CALP as

CALCAL

ACC Vdt

tdPQ )(= . (3.40)

At gas throughputs in the range of 10-3 mbar ls-1 through the accommodator and dis-charge tube, molecular flow regime is assumed. In this flow regime the pressure de-creases exponential with time and the throughput ACCQ is just calculated from the time derivative of PCAL

CALCALACC VtPQ )(1τ

= , (3.41)


46

where τ is the time constant for the pressure decay in the calibrated volume. For each pressure value in the calibrated volume, the corresponding throughput can be deter-mined after the time constant τ has been obtained from an exponential fit to the time evolution of CALP . For the calibration of the molecular flow ACCQ out of the accommoda-tor, the calibrated volume with CALV = 0.129 l is filled to about 250 mbar. A typical flux measurement is shown in Fig. 3.15 recorded at an accommodator temperature TACC = 12.5 K.

Fig. 3.15. Calibration of hydrogen flux for the H atom source at an accommodator tem-perature of TACC = 12.5 K. Using Eq. (3.40) the gas flow )(tQACC is calculated. Interest-ing for calibration is the throughput QACC as a function of the pressure PBT at the entrance of the discharge tube.

The pressure in the calibrated volume and in the discharge tube have been recorded by the data acquisition system every 9 s for ~ 90 min. The time dependence of the through-put is determined by numerical differentiation, using Eq. (3.40). The dependence of the gas throughput on the pressure in the calibrated volume indicates the pressure depend-ence of the throughput through the leak valve. This method allows us to adjust the throughput over a wide range with an uncertainty of less then 0.5 %.

Measurements were carried out for both hydrogen and deuterium and for different ac-commodator temperatures. The left side of Fig. 3.16 shows the dependence of the hy-drogen throughput on the source pressure for various accommodator temperatures. The lines represent linear fits, QACC (TACC) = CACC+DT (TACC) PBT. The right side shows the conductance of the accommodator and the discharge tube as a function of TACC

-1/2. From the linear fit of the data one obtains

12/12/121_ ))(109.01006.1())((

2

−−−−+ ×+×= lsTKlsTC ACCACCHDTACC . (3.42)

Following the same procedure but for deuterium one obtains

12/12/121_ ))(097.01057.0())((

2

−−−−+ ×+×= lsTKlsTC ACCACCDDTACC (3.43)


47

Fig. 3.16. Left panel: Dependence of the hydrogen throughput on source pressure for the indicated accommodator temperatures. The lines represents linear fits, QACC = CACC+DT PBT. (Right) Temperature dependence of the accommodator and discharge tube conductance CACC+DT.

The atom density at the accommodator exit can be estimated once the gas throughput and the degree of dissociation are known. The atomic flow rate out of the accommoda-tor is given as

)(21069.2)( 1191 −− ××××= lsmbarQatomssF ACCatoms α (3.44)

where ACCQ is the molecular gas throughput, α is the degree of dissociation and the factor 2.69 × 1019 converts the flow rate into atoms s-1. Assuming the accommodator to be a point-like source, the intensity distribution function can be describe as

2cos

rFI atoms π

θ= (3.45)

where θ is the polar angle with respect to the beam axis, and r is the radius of the ac-commodator. The number density of atoms just outside the accommodator exit is given as

⟩⟨

=vInACC (3.46)

where aACCB mTkv ⋅=⟩⟨ π/8 is the average velocity of the particles in the accommoda-tor. The density field at the center line of the effusive stream for a circular orifice has been derived by Howard [how61]

⎟⎟⎠

⎞⎜⎜⎝

⎛

+−=

221

2)(

rzznzn ACC (3.47)

where ACCn is the number density of atoms just outside the accommodator, r is the ra-dius of the accommodator and z is the position along the AB-22PT axis. Fig. 3.17 shows the estimated density of D atoms at the exit of the accommodator as a function of source pressure and accommodator temperature. Inspection of these curves reveals that the atom density increases towards its maximum by increasing the source pressure and at


48

higher pressures the density drops down. D-atom densities as high as 1.0 ×1014 cm-3 have been achieved when the source was running under optimal conditions: source pres-sure of about PBT = 0.2 mbar, rf power of P = 20 W and the accommodator coated with an optimal ice layer at TACC = 100 K. At the lowest accommodator temperature TACC = 12 K, atom density around [D] = 2 × 1013 cm-3 has been obtained at a source pressure PBT = 0.1 mbar. The measurements performed at TACC = 300 K shows the atom density at the exit capillary (2 mm in diameter, 45 mm lengths) of the Slevin-type source. As can be seen in Fig. 3.17 the maximum atom density [D] = 5.7 × 1013 cm-3 is obtained for a source pressure of PBT = 0.2 mbar.

Fig. 3.17. Number density of D atoms at the accommodator exit as a function of source pressure for indicated accommodator temperatures.

In order to determine the atom density at the location of the 22-pole trap which is situ-ated at a distance of 46 cm from the accommodator exit, the relation (3.47) has been used. Densities around 1.2 × 108 cm-3 have been obtained using a 2 mm in diameter ac-commodator hold at TACC = 100 K. These estimates are in excellent agreement with measurements performed with the universal detector and chemical probing using reac-tions in the 22PT (see appendix B). Larger densities are expected when the distance between the source and the 22PT is shorter. Experimental test performed for a distance of 8 cm indicated an atom density of 2.2 × 109 cm-3. However, not only the atom density increases but also the molecular background which is coming from the source chamber.

Determination of number density using an universal detector The number density of H atoms and the H2 background at the location of the trap has been determined by replacing the 22PT with the calibrated universal detector which is based on electron bombardment ionization followed by mass selection and ion counting (see Fig. 3.6). Using a mechanical shutter placed between the second hexapole and the ionizer, it is possible to separate the direct beam from background contributions. The density of H2 background with discharge switched OFF, [H2]OFF_bg, is determined from the ion signal of H2

+ recorded with shutter close while the density of H2 beam, [H2]OFF_beam, is determined by the differences in the ion signal of H2

+ recorded with shutter open and closed. Switching the discharge ON the H2 flux is converted into an H / H2 mixture. The fragmentation of H2 leads to a decrease in the ion signal of H2

+ which results in a reduced number density of H2 beam, [H2]ON_beam, while the H atoms


49

formed lead to an increase of the H+ ion signal (see Fig. 3.7). Using the experimental set up with the hexapole magnets removed, conservation of the mass flux through the leak valve, the accommodator, and the ionizer is guaranteed. This means that one obtains for each molecule in the beam two atoms. In order to convert the flux into density in the ionizer one has to account for the different velocities of H and H2. This requires division by 2 , assuming that, in the accommodator, atoms and molecules are fully thermalized to the same temperature. Accounting for these two factors, the mean number density of H atoms is simply obtained from

[H] = 2 ([H2]OFF_beam - [H2]ON_beam) . (3.48)

When the hexapole magnets are in place, the conservation of the mass flux doesn’t hold and the atom density has to be determined directly from the ion signal of H+. The cali-bration factor which converts the H+ ion signal in number densities has been determined in a setup without hexapole magnets.

Fig. 3.18 shows the number density of D atoms (left panel) and D2 background (right panel) as a function of the accommodator temperature. The rf source was running at the optimum pressure and rf power (PBT = 0.1 mbar, P = 20 W). The accommodator surface was coated as described above. In the left panel, the densities of D atoms with and with-out hexapole magnets are compared. Without the guiding field, the atom density has a maximum value around 108 cm3 in a temperature range between 50 and 100 K and de-creases above 100 K and below 50 K due to surface recombination. A minimum has been observed in atom density at an accommodator temperature around 20 K and a re-increase when the temperature approaches 10 K. More details concerning H-H recombi-nation are discussed in the previous section. With hexapole magnets in place the atomic signal is enhanced by a factor of 15±5 at 50 K. Atom densities as large as 2 × 109 cm-3 have been obtained for the temperature range TACC = 50 – 100 K. At the lowest tempera-ture, TACC = 12 K, the D atom density drops to 1 × 108 cm-3.

Fig. 3.18. (Left panel) The density of D atoms has been measured as a function of the ac-commodator temperature TACC. The rf source was running under optimal conditions (PBT = 0.1 mbar, P = 20 W) and the accommodator surface coated with an optimal ice layer. The decrease in density above 100 K and below 50 K is due to surface recombination. The re-increase below 15 K is most probably caused by condensation of D2 molecules on the accommodator surface. (Right panel) Total density of D2 measured with discharge OFF and ON as a function of accommodator temperature.


50

The right panel of Fig. 3.18 shows the total density of D2 (background and not dissoci-ated molecules in the beam) measured with discharge OFF and ON for different ac-commodator temperatures. With the discharge switched OFF, the density of D2 in-creases from 7 × 107 cm-3 at TACC = 215 K to 2.1 × 108 cm-3 at an accommodator tem-perature of TACC = 13 K. Switching the discharge ON the flux of D2 is dissociated into a D / D2 mixture. At TACC = 100 K most of the molecules in the beam are dissociated lead-ing to a reduced D2 background. For TACC = 300 K or TACC = 12 K the dissociation frac-tion is only few percent and the D2 background with discharge ON approaches the value with discharge OFF. [D] / [D2] ratio around 50 has been measured at an accommodator temperature of TACC = 100 K and a reduce value around 0.5 at the lowest accommodator temperature of TACC = 12 K.

Determination of number density using chemical reactions As already discussed in Sec. 2.4, number densities of atomic and molecular hydrogen can be determined by injecting CO2

+ ions into the trap and using known rate coeffi-cients. In collision with H atoms CO2

+ produces predominantly HCO+ while in collision with molecular hydrogen produces HCO2

+. The rate coefficients for these reactions have been measured using the AB-22PT as a function of ion energy and the kinetic energy of the neutral beam target [bor08c]. The density of H / H2 in the trap depends on a variety of parameters, like for example, the operational pressure and rf power of the source, H-H recombination on the accommodator, transmission properties of the hexapole mag-nets for a given geometry and the efficiency of the vacuum system and the 22-pole trap cold head. Therefore careful characterization for the neutral beam target has been made with and without hexapole magnets in place and for different parameters of the neutral beam and different trap temperatures.

First experimental tests were carried out for the characterization of the D / D2 beam ef-fusing from a cold accommodator hold at TACC = 55 K in a configuration without hex-apole magnets. Geometrical dimensions of the accommodator, skimmer and different apertures are given in Table 3.2. Temperature dependence rate coefficients for the reac-tions of CO2

+ with H and D atoms have been measured using this setup (see paper in the Appendix B).

Fig. 3.19. (Left panel) D2 density in the trap measured with discharge OFF as a function of baratron pressure PBT. This measurements have been recorded at TACC = 55 K and T22PT = 55 K. In order to determine the beam and background density the shutter is open (solid symbols) and closed (open symbols). (Right panel) Dependence of D2 density in the beam (filled circles)) and molecular background (filled squares) on 22PT temperature when the source is running at a fix pressure PBT = 0.2 mbar and the accommodator temperature kept at TACC = 55 K.


51

The left panel of Fig. 3.19 shows the total (shutter open) and background (shutter close) density of D2 in the trap measured with discharge OFF as a function of baratron pres-sure PBT. For these measurements the trap temperature was adjusted to the accommoda-tor temperature T22PT = 55 K. From a linear fit of experimental data one obtains the background density [D2]OFF_bg and total density [D2]OFF_tot in the trap as a function of baratron pressure PBT. The density of D2 beam is obtain from the difference [D2]OFF_beam = [D2]OFF_tot - [D2]OFF_bg. On the right panel in Fig. 3.19 is plotted the beam (filled cir-cles) and background (filled squares) density of D2 molecules in the trap for different trap temperatures T22PT. The source pressure was fixed at PBT = 0.2 mbar and the dis-charge switched OFF. As can be seen in Fig. 3.19 the density of D2 molecules in the beam has a constant value around [D2]OFF_beam = 1.2 × 108 cm-3 and no influences of the 22PT temperature was observed within the measured range. This suggests that the neu-tral beam traverse the linear multipole ion trap in axial direction without hitting any of the surfaces surrounding the interaction region. Inspection of Fig. 3.19 shows that the density of the D2 background increases by cooling down the 22PT. The data points cor-responding to the D2 background were fitted with the function [D2]OFF_bg = 8.2 × 107 × sqrt(300K / T22PT) cm-3.

Fig. 3.20 shows typical number densities of H (left panel) and H2 (right panel) measured with the hexapole magnets in place for different accommodator and trap temperatures. In order to obtain large [H] / [H2] ratios inside the 22PT different beam geometries have been tested (see Table 3.2). Also, the effect of water coating has been investigated at an accommodator and trap temperature of 100 K. Inspection of Fig. 3.20 shows that atom densities as high as 5.2 × 109 cm-3 have been achieved for a 2 mm in diameter accom-modator hold at TACC = 55 K the surface of which was not coated. Reducing the diame-ter of the accommodator to 1.2 mm smaller densities around 2 × 108 cm-3 have been measured for TACC = 12 K and a somehow larger value around 5 × 108 cm-3 for TACC = 100 K. However, the situation is improved using the water coating and the actual den-sity is enhanced by a factor 4.5 as is expected from the increase in the dissociation frac-tion (see Sec. 3.3.1). As can be seen in Fig. 3.20 the atom density is influenced by the 22PT temperature. This suggests that a fraction of atoms hit the interior of the 22PT and recombine forming H2 background. A maximum atom density is observed for trap tem-peratures around 100 K and a minimum value at temperatures around 20 K.

Fig. 3.20. Number density of H atoms (left panel) and H2 background (right panel) in the 22PT as a function of trap temperature T22PT, and accommodator temperatures TACC.


52

This behavior could be explained by the low H-H recombination on the 22PT walls (made from copper) at temperatures around 100 K and a higher recombination probabil-ity at 20 K. The density of the H2 background, measured as a function of trap and ac-commodator temperature with the discharge switched ON, is plotted in the right panel of Fig. 3.20. Setting the trap temperature to 300 K, an H2 background number density of about 5 × 108 cm-3 has been measured for TACC = 100 K and a larger value of 2.5 × 109

cm-3 for TACC = 55 K. The increase in H2 background results from different operational pressure of the rf source and different beam geometries (see Table 3.2). Cooling the trap from 300 K to 20 K the density of H2 background increases by almost a factor of 4 for each accommodator temperature as is expected from the temperature dependence [H2]bg_T = [H2]bg_300K × sqrt(300K / T22PT) cm-3. Operating the trap at 10 K reduces sig-nificantly the H2 background due to additional cryopumping. For example, using the 1.2 mm in diameter accommodator hold at TACC = 100 K the H2 background drops from 2 × 109 cm-3 measures at T22PT = 20 K to a value around 5 × 107 cm-3 measured at the lowest trap temperature of T22PT = 10 K. The net result is a reduction of 37 in the H2 back-ground for a period of time up to 30 minutes. After two hour of operation the 22PT sur-face is saturated and the H2 background increase. However, by warming up the 22PT for short time helps to remove the H2 adsorbed on the surface and cryopumping becomes again efficient. With the 2 mm in diameter accommodator hold at TACC = 55 K a large background is observed in the 22PT and cryopumping is efficient for only 10 min.

Table 3.2. Geometric dimensions of the accommodator, skimmer, hexapole magnets, and dif-ferent apertures. For nomenclature see appendix D.


53

3.3.3 Velocity distribution of atoms In order to design a suitable accommodator and test its efficiency, beam density meas-urements have been carried out using the calibrated universal detector. Measuring the density of molecules in the beam as a function of gas throughput and accommodator temperature allows us to probe the translational temperature of molecules in the beam. The basis for the following is that the beam density is inversely proportional to the av-erage velocity.

Fig. 3.21. Influence of the thermalization of D2 molecules in the accommodator. Left panel: Density of D2 molecules as a function of source pressure PBT for various accom-modator temperatures (discharge OFF). The lines are linear fits, [D2]OFF_beam = BT PBT + A. The plot in the right panel indicates that the fitting parameter BT, (relative number den-sity) increases proportional to 1/sqrt(TACC).

A collection of various data is presented in Fig. 3.21. The measurements have been per-formed with deuterium and the 1.2 mm copper accommodator (inner diameter 1.2 mm, lengths 22 mm). The discharge has been OFF. The left panel shows the D2 density, [D2]OFF_beam, determined with the ionizer, as a function of the pressure PBT. Several ac-commodator temperatures between 15 K and 300 K have been used. The slope of these dependences, BT, is plotted in the right panel as a function 1/sqrt(TACC). As can be seen the relative density BT increases linearly with 1/sqrt(TACC). This allows the conclusion that the molecules really get thermalized to the accommodator temperature, with a mi-nor uncertainty at 15 K. It can be assumed that the same holds for the H atoms. Some additional measurements were performed using the 2 mm accommodator (diameter 2 mm, lengths also 22 mm) It has been found that the number of collisions inside the accommodator were not sufficient for a complete accommodation of the molecules to the accommodator temperature. Note that these measurements have been performed without a precooler.

For a complete characterization of the H atom beam, velocity distributions measure-ments for a 2 mm copper accommodator have been recorded during the PhD work of C. Mogo, using the universal detector in combination with a chopper wheel. A typical time of flight distribution of D2 molecules measured with discharge OFF is shown in Fig. 3.22. The accommodator temperature has been set to TACC = 11 K and the source was running at a throughput of QACC_D2 = 2.5 × 10-2 mbar ls-1. The comparison of the data with a Maxwell–Boltzmann distribution, calculated for an effusive beam with


54

Tfit =11.2 K, reveals that there are enough wall collisions in the accommodator to ther-malize the molecules.

Fig. 3.22. Measured and simulated time of flight distribution for D2 molecules effusing from a cold accommodator at TACC = 11 K. In this experiment the 2 mm accommodator has bee used (inner diameter 2 mm, length 22 mm). The solid line is an M-B distribution calculated for Tfit = 11.2 K.

For the atomic beam, the velocity distributions are structured as can be seen in Fig. 3.23. These features can be explained by the focusing and guiding properties of the two hexapole magnets and the boundary conditions imposed by the present beam ge-ometry (see Sec. 3.2). Comparison with the calculated thermal distribution (dotted line) reveals that the magnets favor mainly the transmission of atoms with kinetic energies of 8.2 meV and 3.6 meV. For more details see [bor08a], [mog09].

Fig. 3.23. Normalized time of flight distribution of H atoms measured at TACC = 50 K us-ing an ionizer in combination with a chopper wheel. The transmission features of the two hexapoles favor mainly two groups the kinetic energies of which are indicated by the two Gaussians. Without the hexapole magnets one obtains the thermal distribution indicated by the dotted line. The maximal transmission of H atoms is obtained for a kinetic energy of 8.2 meV.

4 REACTIONS

Already during the development and the various tests and improvements of the instru-ment, a variety of reaction systems has been studied, utilizing the various features of the combination of a temperature variable ion trap with the atomic hydrogen beam. Motiva-tions for selecting specific systems range from testing and calibrating the machine, over open questions concerning the reactivity of fundamental systems such as CH+ + H, to exploring challenging unknowns such as the proton affinity of methane. Some of the results already have been accepted for publication, some drafts have been submitted and some material is still in the make. This chapter summarizes the state of scientific results achieved during the experimental phase of this thesis. In some cases we refer to the manuscripts which can be found in the Appendix.

4.1 Temperature dependence of reactions of CO2+

Collisions of CO2+ with atomic and molecular hydrogen have been established in the

literature as standard for calibrating the densities of these two targets [tos84], [sab93]. As described in [bor08c], the rate coefficients k(T) for the following reactions have been measured,

CO2+ + H → HCO+ + O , (4.1)

CO2+ + H2 → HCO2

+ + H , (4.2)

CO2+ + D → DCO+ + O , (4.3)

CO2+ + D2 → DCO2

+ + D . (4.4)

For obtaining reliable thermal rate coefficients for these reactions, we used, instead of the beam, neutral reactants which were leaked directly into the trap with number densi-ties up to 1011 cm-3. Typical measurements are shown in Sec. 2.4 and are discussed in detail in the publication [bor08c] which is reproduced in Appendix B. The thermal rate coefficients measured in this way, are now used routinely to determine the effective number density of the hydrogen atoms and molecules in the trap. Blocking the molecu-lar beam with the shutter, also the net molecular background can be determined. For establishing this calibration standard, the hexapole magnets were taken out. This allows us to assume conservation of hydrogen flux in the beam, independent on whether the discharge is switched OFF or ON and to convert the hydrogen flux into densities in the trap [bor08c].

The AB-22PT machine has the flexibility to select separately the ion temperature T22PT (10 to 300 K) and the velocity distribution of H atoms TACC (12 to 300 K). The resulting rate coefficients for reactions (4.1) – (4.4) are shown in Fig. 4.1 as a function of the temperature T22PT. The errors of the absolute values for reactions with molecular target are estimated to be 20 % and those with atomic reactants 40 %. Inspection of Fig. 4.1 reveals that the rate coefficients for molecular target increase significantly with decreas-ing temperature,

k4.2= 9.5 × 10-10 cm3s-1 (T/300K)-0.15 , (4.5)

k4.4= 4.9 × 10-10 cm3s-1 (T/300K)-0.30 . (4.6)

The fact that, at low temperatures, one finally reaches the Langevin limit, marked with short lines in Fig. 4.1, may be taken as an indication that, in the collision complex, time is required for dissociating the hydrogen molecule and for integrating the atom into the

55

4 REACTIONS

protonated carbon dioxide. Reaction (4.3) has been investigated at TACC = 55 K and T22PT = 55 - 300 K while reaction (4.1) at different combinations of TACC and T22PT. No significant temperature dependence has been observed between 50 and 300 K. For D, an average value of k 4. 3= 2.2 × 10-10 cm3s-1 has been measured while for H we report two values k 4.1 = 4.5 and 4.7 × 10-10 cm3s-1. At room temperature the obtained rate coeffi-cients are in reasonable overall agreement with results from previous measurements performed with different flow systems [feh71], [tos84], [sco97a]. This is surprising due to the overall difficulties in determining the actual H atom concentrations in the various instruments. Reactions with H and D atoms also have been investigated with the ICR technique [kar79]. However, the results from this instrument deviate significantly. Some explanations are given in [bor08c].

Fig. 4.1. Temperature dependence of rate coefficients for reactions of CO2

+ with H, H2, D, and D2

(reactions (4.1) - (4.4)). With molecular target, the reactivity increases with fal-ling temperature, reaching finally the Langevin limit kL. The rate coefficients are fitted with the function k(T) = α (T22PT / 300 K)β. For the reactions with H or D atoms, no change with temperature could be observed.

4.2 Reactions of CH+ with H and D From a fundamental point of view, the C+ + H2 collision system is one of the model systems for experimental and theoretical studies of the kinetics, dynamics, and energy requirements of an endothermic ion-molecular reaction. Consequently, it has been the subject of numerous experimental and theoretical studies [che84]. The reverse reaction

CH+ + H → C+ + H2 + 0.398 eV (4.7)

represents an important destruction mechanism of CH+, the abundance of which is poorly understood in diffuse interstellar clouds. Simple models underestimate the ob-served abundances and, therefore, it is assumed that shock waves, turbulences or UV radiation must play a role. Until 1984 it was accepted that, at 100 K, the exothermic reaction (4.7) is slow, k = 2 × 10-12 cm3s-1 and even slower at lower temperatures whereas phase space theory predicts that the rate coefficient can approach the Langevin limit of 2 × 10-9 cm3s-1 at low temperatures [che84]. At and above room temperature,

56

4 REACTION

this was confirmed by experimental results obtained with the SIFDT apparatus [fed84], [fed85]. The observed large rate coefficient and the negative temperature dependence indicates that the potential energy surface has no barrier or only a small one.

With the AB-22PT apparatus, the rate coefficient for reaction (4.7) has been measured as a function of the ion temperature and the kinetic energy of the H atoms. As an exam-ple, two sets of data are plotted in Fig. 4.2. The two temperatures, TACC and T22PT, have been set both to 50 K. Plotted is the number of ions per filling, Ni, averaged over many iterations. The number density of H / H2 with discharge OFF and ON has been cali-brated as described in Sec. 2.4 and is given in the figure caption.

Fig. 4.2. The reactions of CH+ ions with H2 (left panel, [H2] = 6 × 108 cm-3) and a mixture of atomic and molecular hydrogen ([H] = 4 × 108 cm-3, [H2] = 1.7 × 109 cm-3) have been recorded at TACC = 50 K. For cooling the ions to T22PT = 50 K, He gas has been added with a number density of 1.5 × 1013 cm-3. Plotted is the number of primary and product ions per filling, Ni, as a function of the storage time t. The product ions CH2

+ will undergo H-atom abstraction with the H2 gas leading finally to CH3

+. The lines are solutions to an adequate rate equation system leading to the rate coefficients for reactions (4.7) – (4.9).

In the left panel, the discharge has been switched OFF and the injected CH+ ions react only with the beam of H2. The first product is CH2

+ (triangle) which is formed via the fast hydrogen abstraction reaction

CH+ + H2 → CH2+ + H . (4.8)

This products undergo a second reaction with hydrogen, leading to CH3+ via

CH2+ + H2 → CH3

+ + H . (4.9)

This is the final product under the conditions of our experiment, since the next step would be formation of CH5

+ via the slow radiative association process [ger92c]

CH3+ + H2 → CH5

+ + hυ . (4.10)

Due to the very small rate coefficient and the low number density of hydrogen no CH5+

ions can appear in the range plotted in Fig. 4.2.

Switching the discharge ON (right panel) the H2 flux is converted into a H / H2 mixture. As can be seen on the right side of Fig. 4.2, the H atoms lead to the production of C+ ions via reaction (4.7). Inspection of the data reveals that the decay rate of the primary

57

4 REACTIONS

CH+ ion increases by a factor of 3.5 if the discharge is switched ON (τON = 0.39 s, τOFF = 1.38 s). This is only to a small extend due to the large amount of H atoms pene-trating the ion cloud. Comparison with the left panel, however, indicates, that also more CH2

+ products are formed: this means that also more molecules come into the trap in-stead of less. The reason is explained in detail in [bor08a]. Briefly, the hexapole mag-nets guide a large flux of atoms towards the trap some of which recombine at surfaces. Especially efficient for increasing the H2 background is H-H recombination inside of the trap.

The solid lines in Fig. 4.2 are the solutions of a rate equation system describing the chemical interaction of all trapped ions with H and H2. The rate coefficients resulting for reactions (4.7) – (4.9), are listed in Table 4.1. The measurements have been per-formed at a few selected trap and accommodator temperatures, more work needs to be done.

The experimental and theoretical results are summarized in Fig. 4.3. The non-thermal results of the SIFDT technique have been converted using the approximation KEcm = 3/2 kBT, the validity of which has been shown for atomic ions in drift fields. However, the contribution of rotational energy of CH can differ from translational energy and this may lead to differences in the temperature dependences. Our low temperature data which are close to Langevin limit show that there is no barrier that would significantly hinder the reaction. The theoretical values of phase space theory describe well the be-havior of the reaction [che84], [hal07]. Recent RIOSA-NIP calculations [sto05] which explains all reactive channels including hydrogen atom exchange, underestimate the measured value by almost a factor 10. This indicates that more theoretical work is needed for understanding low temperature processes from first principles.

B

+

Fig. 4.3. Temperature dependence of the rate coefficient for reaction CH+ + H → C+ +H2. The results from this work cover the temperature range from 50 to 80 K. At higher ener-gies results from a SIFDT experiment [fed85] are included. The results of phase space theory [hal07] are slightly larger while calculations using the RIOSA-NIP method are one order of magnitude to small [sto05].

As mentioned in the introduction, isotope scrambling at low temperatures is a very sen-sitive probe for reaction dynamics and molecular structure. Therefore, the deuterated

58

4 REACTION

analogue of reaction (4.7) has been investigated also with deuterium as target gas. Two different product channels are possible in collision of CH+ with atoms,

CH+ + D → C+ + HD , (4.11)

→ CD+ + H . (4.12)

Since the D2 background is always present in the 22PT also reactions of CH+ with D2 need to be taken in account. Three different product channels are possible,

CH+ + D2 → CHD+ +D , (4.13) → CD+ + HD , (4.14) → CD2

+ + H . (4.15)

Fig. 4.4 shows a typical measurement for CH+ reacting with a pure beam of D2 (dis-charge OFF, left panel) and a mixture of D / D2 (discharge ON, right panel). The tem-peratures have been set to TACC = 36 K and T22PT = 80 K. Plotted is the number of pri-mary and product ions, Ni, as a function of storage time. It should be noted that the products CHD+ and CD2

+ were recorded only at 90 ms storage time, with discharge OFF, based on a mass spectrum. In order to explain the measured ion abundances, a coupled rate equation model has been developed. The processes included in this model are shown schematically in Fig. 4.5. With discharge OFF, initially injected CH+ ions react with D2 which is present in the trap with a number density of [D2] = 1.6 × 109 cm-3 via reactions (4.13) – (4.15).

Fig. 4.4. Reactions of CH+ with a beam of D2 (left panel, [D2] = 1.6 × 109 cm-3) and a mixture of D and D2 (right panel, [D] = 8 × 108 cm-3 and [D2] = 3 × 109 cm-3) have been recorded at TACC = 36 K and T22PT = 80 K. The solid lines represent solutions of a system of coupled differential equations describing primary and secondary reactions. The result-ing rate coefficients are summarized in Table 4.1.

The CD+ products react further with D2 by a chain of D atom abstraction leading to CD2

+ and finally to CD3+. Also, the products CHD+ react with D2 by H-D exchange

leading to CD2+. The CD2

+ products undergo further reactions with D2 and form finally CD3

+ via D atom abstraction. Switching the discharge ON, both D and D2 are present in the trap with a number density of [D] = 8 × 108 cm-3 and [D2] = 3 × 109 cm-3.

59

4 REACTIONS

Fig. 4.5. Schematic illustration of the reactions used to model the chemical evolution of CH+ ion in a mixture of D and D2. The product CD2H+ has not been detected.

As indicated in Fig.4.5 by the red arrows the presence of D in the 22PT opens up sev-eral new reactions channels. In a collision with D, CH+ either reacts to C+ via D atom transfer or to CD+ via H-D exchange. The CD+ products undergo further reaction with D and form C+ via D atom transfer. This contribution to the production of the atomic car-bon ion is less efficient than the direct path via reaction (4.11). The CHD+ products re-act also with D to form CD2

+ via H-D exchange.

The solid lines in Fig. 4.4 are the solutions of a rate equation system describing the chemical interaction of all trapped ions with D and D2. In the simulations the rate coef-ficients k6, k7, and k10 have been assumed to be equal to the analogue reactions with hy-drogen. The resulting rate coefficients are listed in Table 4.1.

Table 4.1. Measured reaction rate coefficients for indicated reactions and comparison with published values

a T is either T22PT or the temperature of the respective instrument.

60

4 REACTION

4.3 CH4+ + H, CH4

+ + D An interesting reaction system which proceeds via the intermediate collision complex [CH4

+⋅⋅H] is

CH4+ + H → CH3

+ + H2 . (4.16)

Despite an exothermicity of 2.7 eV, the hydrogen abstraction reaction (4.16) has not been observed at 300 K in an early ICR experiment [kar79]. Note that the detection limit of the ICR experiment was 10-11 cm3s-1. This experimental finding has stimulated us to investigate this reaction in the AB-22PT machine. Fig. 4.6 shows a typical ex-perimental result, the number of various trapped ions, Ni, as a function of storage time t. The primary CH4

+ ions have been relaxed to the ambient temperature T22PT = 80 K via collisions with He buffer gas ([He] = 3.7 × 1012 cm-3) and exposed to a mixture of H / H2 delivered by the cold accommodator TACC = 100 K. Although the H-atom density is lower than the H2 background (see figure caption), formation of CH3

+ via reaction (4.16) prevails. In addition a few CH5

+ are formed via the hydrogen abstraction reaction

CH4+ + H2 → CH5

+ + H . (4.17)

The solid lines in Fig. 4.6 represent the solutions of a rate equation system describing the chemical interaction of all trapped ions with H and H2, leading to the rate coeffi-cients 5 × 10-10 cm3s-1 for the reaction (4.16) and 1.1 × 10-10 cm3s-1 for the reaction (4.17). These values are in good agreement with previous studies [asv04a]. Additional measurements have been performed at an accommodator and trap temperature of 50 K.


+ with hydrogen atoms ([H] = 8 × 108 cm-3) and molecules ([H2] = 1.5 × 109 cm-3) effusing from a cold accommodator TACC = 100 K. After injection the primary ions are relaxed to the ambient temperature T22PT = 80 K using an intense pulse of He. Plotted is the average number of primary and product ions trapped per filling, Ni, as a function of storage time t. The lines are the solutions of a model calculation, resulting in a set of rate coefficients (see Table 4.2).

If one assumes that the rate coefficient obtained at 300 K by the ICR technique is cor-rect; reaction (4.17) would show a very steep temperature dependence. Such changes at low temperatures may be explained by the formation of a long lived [CH4

+⋅⋅H] complex in combination with a bottle neck hindering the transition towards the product channel.

61

4 REACTIONS

However, the ICR data are too unreliable to make such a conclusion and it is better to perform additional measurements over the full temperature range.

In addition to hydrogen also deuterium has been used as target gas. Three different product channels are possible in collision of CH4

+ with D

CH4+ + D → CH3

+ + HD , (4.18) → CH3D+ + H , (4.19) → CH2D+ + H2 . (4.20)

Experimental investigations revealed that collision of CH4+ with D2 lead to only one

product [asv04a]

CH4+ + D2 → CH4D+ + D . (4.21)

It is obvious from Fig. 4.7 that a rich chemistry can be initiated if trapped CH4+ ions are

exposed to a mixture of D and D2. A detailed explanation of the complex chemical in-terplay is not easy since, in addition to D atom abstraction and D atom transfer, also H-D exchange reactions are possible. Since there are several isotopic combinations having the same mass, one has to find methods to separate the various channels. Switching the discharge OFF or ON allow us to measure separately reactions of primary ions with a pure D2 beam or with a beam of both D and D2. Moreover, measurements performed at short times (see left panel in Fig. 4.8) allow us to minimize the role of secondary reac-tions. In this time window (0.1 s) accurate rate coefficients for the initial reactions have been determined. Performing measurements for long storage time (1 s, right panel) helps us to extract information about secondary reactions. Fig. 4.8 shows a typical set of data, recorded with [D2] = 2 × 109 cm-3 at TACC = 50 K.

Fig. 4.7. Schematic illustration of the reactions used to model the chemical evolution of CH4

+ ions stored in a mixture of D and D2

The primary CH4+ ions have been relaxed to the ambient temperature T22PT = 50 K via

collisions with an intense pulse of He buffer gas. Plotted is the number of ions per fill-ing, Ni, averaged over many iterations. In this experiment about 1200 CH4

+ ions are injected each time into the trap. The primary CH4

+ ions react with D2 via D atom ab-straction and form CH4D+. The CH4D+ products undergo further reactions with D2 via

62

4 REACTION

H-D exchange leading to CH3D2+ and in further steps finally to CD5

+. In addition, some tens CH5

+ ions are formed in collision of primary CH4+ with H2 background gas present

in the vacuum chamber. The CH5+ products react slowly with D2 and only small

amounts of CH4D+ and CH3D2+ are formed in the time window shown in Fig. 4.8.

Fig. 4.8. Reaction of CH4+ with a beam of D2 ([D2] = 2 × 109 cm-3 at TACC = 50 K) re-

corded for two time windows (left panel 0.1 s, right panel 1 s). By collision with an in-tense pulse of He buffer gas, the injected CH4

+ and CH3+ primary ions are relaxed to the

ambient temperature T22PT = 50 K. Both CH4+ and CH3

+ ions react with D2 producing CH4D+ respectively CH2D+, CHD2

+. There is also some loss due to reaction of CH4+ with

H2 background gas leading to CH5+ products. The solid lines are solutions of a system of

coupled differential equations describing primary and secondary reactions. The rate coef-ficients are summarized in Table 4.2.


+ with a mixture of D / D2 ([D] = 2.9 × 109 cm-3, [D2] = 3.1 × 109 cm-3 at TACC = 50 K) recorded for two time windows (left panel 0.1 s, right panel 1 s). By collision with an intense pulse of He buffer gas, the injected CH4

+ primary ions are relaxed to the ambient temperature T22PT = 50 K. As illustrated in Fig. 4.7, a rich chemistry is started with CH4

+ + D / D2 (for details see text). The solid lines are solutions of a system of coupled differential equations describing primary and secondary reactions and the rate coefficients are summarized in Table 4.2.

63

4 REACTIONS

It is obvious from Fig. 4.8 that secondary reactions become important only for long storage time. In this experiment, some small amounts of CH3

+ were injected also into the trap. As can be seen on the right panel in Fig. 4.8, the number of CH3

+ ions decay exponentially in collision with D2 while CH2D+ and CHD2

+ products are formed [smi82b]. Despite the fact that the products CH2D+ and CHD2

+ are indistinguishable from CH4

+ respectively CH5+ since they have the same mass, the rate coefficient for

reaction CH3++ D2 can be derived from the decay of primary CH3

+ ions.



Switching the discharge ON, the primary CH4+ ions react with both D and D2 which are

present in the trap with a number density of [D] = 2.9 × 109 cm-3 and [D2] = 3.1 × 109 cm-3. As illustrated in Fig. 4.7 in addition to CH4D+ products formed in the first collisions of CH4

+ with D2, collisions of CH4+ with D form (i) CH3

+ via D atom transfer, (ii) CH3D+ via H-D exchange and (iii) CH2D+ via H-D exchange followed by H abstraction. The primary CH3

+ products react with both D and D2 via H-D exchange to form CH2D+, CHD2

+ and in further steps CD3+. D atom transfer or D atom abstraction in

collision of CH3+ with D or D2 do not occur at low temperatures since both processes

are endothermic. In analogy CH3D+ products react with both D and D2 via H-D ex-change to form CH2D2

+ and in further steps CD4+. In contrast to CH3

+ equivalents the CH4

+ equivalents react with D and D2 via D atom transfer and D atom abstraction, re-sulting in a complex chemistry. However, measurements performed in the first millisec-

64

4 REACTION

onds (see left panel in Fig. 4.9) allow us to derive the rate coefficients for the primary reactions (4.18) and (4.19). Due to the mentioned mass overlap problem it is hard to determine an accurate rate coefficient for reaction (4.20). However, measurements per-formed for long storage time (see right panel in Fig. 4.9) allow us to derived an upper limit for the rate coefficient (4.20) by observing its contribution to the mass 17 and 18. Also the rate coefficient for CH3

++ D reaction can be determined accurately since the rate coefficient for CH3

++ D2 reaction is known from previous measurements with dis-charge OFF. The solid lines in Fig. 4.8 represent the solutions of a rate equation system describing the chemical interaction of primary and product ions with D, D2 and H2. Re-sulting rate coefficients are summarized in the Table 4.2.

4.4 CH5+ + H, CH5

+ + D The thermodynamical data of the reaction

CH5+ + H ↔ CH4

+ + H2 (4.22)

are quite controversial. The astrochemical database UMIST [teu00] and the NIST Chemistry WebBook [lin03] report endothermicities of 19 kJ/mol and 14.0 kJ/mol, re-spectively. Reaction (4.22) has been investigated in forward and backward direction by SIFDT technique [fed85] in the regime from thermal to 120 meV center of mass kinetic energy, KEcm. From a Van’t Hoff plot of the experimental data an endoergicity ΔH = -5 kJ/mol, and an entropy change ΔS0297 = -31 Jmol-1K-1 has been obtained. This means that reaction (4.22) is endoergic but exoentropic. Due to this endothermicity the rate coefficient should decrease steeply from ~ 10-10 cm3s-1 at 20 meV (150 K) to ~ 10-12 cm3s-1 at 8 meV (60 K) as indicates the thin dotted line in Fig. 4.10. The rate coefficient in backward direction, has been measured from 300 K down to 15 K, using a temperature-variable 22-pole trap [asv04a], see full squares in Fig. 4.10.

Fig. 4.10. Temperature dependence of measured rate coefficients for the reactions CH5

+ + H → CH4

+ + H2 and its backward reactions [luc05]. The high temperature data have been obtained with a SIFDT apparatus while low temperature data have been measured in two different 22-pole traps.

65

4 REACTIONS

In the present work, rate coefficients k(TACC, T22PT) have been measured for the forward direction varying the accommodator temperatures TACC between 10.8 K and 100 K. Also the trap temperatures T22PT has been set to values between 10 K and 300 K. As can be seen in Fig. 4.10 (open circles, TACC= 92 K), the rate coefficient increase slightly from 2.3 × 10-11 cm3s-1 at T22PT = 10 K to 2.8 × 10-11 cm3s-1 at T22PT = 300 K. A slightly lower rate coefficient is obtained for TACC = 10.8 K (filled circles). At the lowest accommoda-tor and trap temperature ~10 K a rate coefficient around 10-11 cm3s-1 have been meas-ured. However, as discussed in [bor08a], the hydrogen beam has a mean kinetic energy of 3.6 meV under these conditions. The small variation in the forward rate coefficient by changing the trap temperature T22PT, which define the internal energy of CH5

+ ions, can be taken as an argument that a direct process is involved and the H atom impact determines predominantly the reactivity.

Finally, we present experimental results for collisions of protonated methane, CH5+,

with a mixture of D / D2.

CH5+ + D → CH4

+ + HD (4.23) → CH4D+ + H (4.24) → CH3D+ + H2 (4.25)

CH5+ + D2 → CH4D+ + HD (4.26)

→ CH3D2+ + H2 (4.27)

Fig. 4.11. Reaction of CH5+ with D2 beam (left panel, [D2] = 1.4 × 109 cm-3) and D / D2

mixture (right panel, [D] = 2.9 × 109 cm-3 and [D2] = 3.1 × 109 cm-3) recorded at TACC = 50 K and T22PT = 50 K. Plotted is the average number of CH5

+ and product CH4+,

CH4D+ and CH3D+ ions as a function of storage time t. In the time window presented the number of injected primary ions (~ 3000 per filling) is almost unchanged, despite the rather large number density of the target gas. The solid lines are solutions of a system of coupled differential equations describing the reaction system; the resulting rate coeffi-cients can be found in Table 4.3.

Fig. 4.11 shows a typical measurement, the reaction of CH5+ with D2 beam (left panel)

and a mixture of D / D2 (right panel) recorded at TACC = 50 K and T22PT = 50 K. Plotted is the average number of primary CH5

+ and product CH4+, CH4D+ and CH3D2

+ ions as a function of storage time. The product CH3D2

+ has been recorded only at a fixed storage time of 90 ms by scanning a mass spectrum. The number density of D / D2 with dis-

66

4 REACTION

charge OFF and ON has been calibrated using the standard method and is given in the figure caption. The chemical evolution of CH5

+ in a mixture of D and D2 is illustrated in Fig. 4.12.

With discharge OFF initially injected CH5+ ions react with D2 via reactions (4.26) –

(4.27). The primary products CH4D+ and CH3D2+ react further with D2 by a chain of H-

D exchange reactions leading to CD5+. As can be seen on the left side of Fig. 4.11 some

small amount of CH4+ initially injected to the trap react with D2 via reaction (4.21) and

form CH4D+. However, this contribution is unimportant compare to reaction (4.26).

Switching the discharge ON, both D and D2 are present in the trap and a rich chemistry is initiated (see Fig. 4.12). In the first collision with D, CH5

+ ions form (i) CH4+ via D

atom transfer, (ii) CH4D+ via H-D exchange, and finally (iii) CH3D+ via H-D exchange followed by H abstraction. As in the case of D2, the CH4D+ products undergo several H-D exchange reactions with D producing CH3D2

+, CH2D3+, and finally CD5

+. Similarly CH3D+ products, which are overlapping with the primary CH5

+ ions, react with D and D2 via a chain of H-D exchange reaction leading to CD4

+. Finally, CH4+ products react

with D2 via D atom abstraction reaction (4.21) and with D via D atom transfer reaction (4.18), and H-D exchange reaction (4.19).

The solid lines in Fig. 4.10 represent the solutions of a rate equation system describing the chemical interaction of all trapped ions with D and D2. In order to explain the meas-ured ion abundances also information from CH4

+ + D / D2 reactions have been used (see Fig. 4.12). In addition, it was assumed that H-D exchange in collisions of CH4D+ and CH3D+ with D and D2 proceeds fast with a rate coefficient of 10-9 cm3s-1. This assump-tion is based on the experimental observation that H-D exchange in CH4D+ + D2 is fast, resulting in a rate coefficient of 10-9 cm3s-1 (see left side of Fig. 4.11). Resulting rate coefficients for the reactions (4.23) – (4.27) are summarized in Table 4.3.

Fig. 4.12. Schematic illustration of the reactions used to model the chemical evolution of CH5

+ ions stored in a mixture of D and D2

67

4 REACTIONS



68

5. SUMMARY AND OUTLOOK

In this work a trapping machine has been constructed dedicated to study low energy reactions between stored ions and radicals, especially hydrogen atoms. An effusive beam of H-atoms which is skimmed, double differentially pumped, focused and guided by two hexapole magnets is combined with a temperature variable ion trapping appara-tus. The neutral beam traverse the linear 22-pole ion trap in axial direction avoiding collisions with the cold surfaces in the trap. In the center of our activities was the devel-opment of an intense hydrogen atom source the kinetic energy of which can be changed using an accommodator, connected to a cold head. In order to characterize the H atom source, various diagnostic tools has been set up. First investigations were conducted towards the study of the experimental conditions by which H atoms can be cooled to low temperatures and the efficiency of this process with respect to recombination. We have found a procedure to obtain very high dissociation degrees (more than 90 % at 100 K) and to keep them constant over many hours. The method is based on coating the walls of the precooler and the accommodator with specific layers of water. The H atom beam has been characterized in detail using time of flight analysis. The focusing force and the geometric boundary conditions of the magnets used favor transmission of three kinetic energies, 3.6, 8.2, and 21.0 meV. The effective H-atom number density has been determined in the interaction region either using a calibrated ionizer or, directly in the trap, via chemical probing with suitable ions, especially CO2

+. Number densities in the range of 109 cm-3 have been reached with a rather low H2 / H ratio in the trap. With the high sensitivity of the rf ion trapping technique, this allows us to determine rate coeffi-cients as small as 10-13 cm3s-1 for ion - hydrogen reactions.

During the various stages of development, the instrument has been used for studying reactions between ions of astrochemical importance and cold hydrogen atoms. The reac-tions of CO2

+ with hydrogen atoms and molecules already have been mentioned as a calibration standard for in situ determining H-atom densities. It has been found that the rate coefficients for reactions with atoms do not depend on the temperature of the CO2

+ ions nor on the velocity of the atoms and they are rather small. For collisions with molecules, the reactivity increases significantly with falling temperature, reaching the Langevin values at 15 K.

For the first time, reactions of small ionic hydrocarbons CHn+ (n = 1, 4, 5) with H and D

atoms have been studied at temperatures representative for dark interstellar clouds. The reactions of the investigated ionic species are important for models of interstellar clouds and also for understanding the reaction dynamics at low and very low energies. One of the important points of the research program was the reaction of CH+ + H since this re-action is an important destruction mechanism of CH+ in diffuse clouds. It has been found that at low temperatures, the rate coefficient is fast as predicted by statistical theories and decreases when the collision energy increases. The temperature depend-ence is in accordance with the trend predicted by the high temperature SIFDT data and recent calculations based on phase space theory. Surprising results were measured for collision of CH+ with D. The same rate coefficient has been obtained for both D-atom transfer and H-D exchange indicating the importance of the atom exchange channel.

The methane cation, CH4+ shows also some interesting features in collision with H at-

oms. Reaction rate coefficients, measured at T22PT = 50 K and TACC = 50 K are a factor of 60 larger than values frequently used in astrophysical models. Using D instead of H in reactions with CH4

+ it has been found that the rate coefficient for H-D exchange is 2.5 × 10-11 cm3s-1.

69

SUMMARY AND OUTLOOK

One important results of these studies is the investigation of the CH5+ + H collision

since the thermodynamical data of this reaction are quite controversial. Rate coefficients around 10-11 cm3s-1 have been measured at the coldest conditions T22PT = 10 K and TACC = 12 K which indicates an endoergicity smaller than 5 kJ mol-1 as derived in ref. [fed85]. Another question was whether protonated methane, CH5

+, is subject to H-D exchange in collisions with D atoms at low temperatures. It turns out that the rate coef-ficient measured at T22PT = 50 K and TACC = 50 K is small 1.0 × 10-12 cm3s-1. However, H-D exchange in collision of CH5

+ with D is 106 times faster than in collision with HD [asv04b].

Outlook There are several technical improvements in progress. Especially important is the reduc-tion of the molecular background in the trap. With this aim a beam catcher has been installed at the end of the instrument and a cryopumping tube has been mounted be-tween the beam source and the trap (see Fig. 6.8 of [ger08b]). For further lowering the kinetic energy of the H atoms below 1 meV, weaker hexapole magnets should to be used and the accommodator has to be operated at the limit of H-condensation. A new cold head, reaching 3.6 K already has been already connected to the accommodator. A big step forward in experiments between ions and H atoms would be to combine a suit-able rf trap for ions with a magnetic trap for the atoms. There have been several suc-cessful activities to load H atoms into a magnetic trap with a cryogenic filling technique exploiting interatomic collisions and heat exchange with liquid helium covered surfaces. Densities up to 1014 cm-3 and temperatures below the mK range have been achieved [kil98].

However, even without these improvements, this versatile instrument opens up already many possibilities to study systems which are of interest for low temperature plasmas or which are of central importance for testing fundamental theories. To the first group be-long radiative association reactions with H atoms which may be important for catalytic formation of molecular hydrogen in interstellar space. A critical test case for the appara-tus is the 3.6 meV endothermic electron transfer from D to H+. The threshold behavior is well characterized by theory [esr00]. The reverse process leads to a fast loss of deu-terons; however, so far no measurements are available.

The formation of molecular hydrogen via the associative electron detachment reaction H- + H → H2 + e- is of great importance for the chemistry of the early universe but not known with the required accuracy. Storing H- and monitoring the loss of ions due to collisions with H is feasible with the trapping technique. The detection of the velocity distribution of the ejected electrons is planned. Very challenging are studies of Hn

+ col-lision systems with various H-D ratios. Using laser probing of the ions [scl06], the ob-servation of ortho to para transitions, e.g. o-H3

+ + H → p-H3+ + H, is in principle possi-

ble. Easier, from the point of view of the experiment, is the exothermic H - D exchange in H3

+ + D. The potential energy surface has been recently published [moy04]; however, the accuracy in the vicinity of the H3D+ transition state is not sufficiently characterized.

70

71

Appendix A

On the combination of a low energy hydrogen

atom beam with a cold multipole ion trap G. Borodi, A. Luca, C. Mogo, M. Smith*, D. Gerlich

Department of Physics, Technische Universität Chemnitz, 09107 Chemnitz, Germany

*Department of Chemistry, The University of Arizona, P.O. Box 210041, 1306 E. University Blvd. Tucson, Arizona 85721-0041

Submitted to Rev. of Scientific Instruments

1

On the combination of a low energy hydrogen atom beam with a cold multipole ion trap

G. Borodi, A. Luca, C. Mogo, M. Smith*, D. Gerlich

Department of Physics, Technische Universität Chemnitz, 09107 Chemnitz, Germany

*Department of Chemistry, The University of Arizona, P.O. Box 210041, 1306 E. University Blvd. Tucson, Arizona 85721-0041

Abstract An experiment has been developed for studying reactions between stored ions and radi-cals or condensable molecular species under well-defined conditions. A skimmed and doubly differentially pumped neutral beam source has been added to a temperature vari-able (T22PT = 10 K - 300 K) ion trapping apparatus such that the neutrals traverse the linear 22-pole ion trap (22PT) in the axial direction, avoiding collisions with the cold surfaces in the trap. Special emphasis has been given to the integration of an intense source for hydrogen atoms the velocity distribution of which can be changed using a temperature variable accommodator (TACC=10 K - 300 K). The effusive H (or D) atom beam which also can be guided and selected using one or two hexapole magnets, has been characterized in detail. Complete thermalization to TACC has been confirmed using time of flight analysis. The focusing force and the geometric boundary conditions of the magnetic fields favor transmission of three groups of kinetic energies, centered at 3.6, 8.2, and 21.0 meV. The kinematic conditions for collision between such a beam travers-ing the trapped and stored ion cloud are discussed. The effective H-atom number den-sity is determined in the interaction region either using a calibrated ionizer or, directly in the trap, via chemical probing with CO2

+ ions. Values of some 109 cm-3 have been reached. Due to the tremendous dynamic range and the high sensitivity of the rf ion trapping technique, this allows us to determine rate coefficients smaller than 10-13 cm3s-1 for ion - H-atom reactions. The potential and present limits of the machine are illus-trated with the interaction of CH+ and CH5

+ ions with hydrogen atoms and molecules. Under most conditions, the density of the molecular background is still higher than that of atoms. It needs to be further reduced for studying reactions with ions like H2

+ which otherwise would get lost via fast reactions with H2. In the outlook some ongoing im-provements are mentioned including cryopumping and preparing atoms with sub-meV energies.

Keywords Ion-atom reactions, rf-multipole ion trap, effusive hydrogen atom source, cold H atoms, low temperature rate coefficient, reaction kinetics

2

1 Introduction Hydrogen in the universe Hydrogen is the most abundant baryonic species in the universe. It plays, both in its atomic and molecular form, a central role in many fields of natural science. Hydrogen plasmas can be found in laboratory discharges, flames, atmospheres, and astrophysical environments. Although pure hydrogen plasmas seem to be rather simple, the interac-tions of its components (H, H+, H-, H2, H2

+, H3+, electrons, and photons) are usually

rather complex since many processes compete with each other. In order to model such a system, many relevant parameters are needed. The knowledge, covering the energy range from thermal up to keV, has been reviewed recently [jan03]. This energy range is suitable for modeling the plasma physics of the solar corona and solar wind or technical applications, e.g. for describing the boundary region of a magnetically confined fusion plasma.

For questions raised in astrochemistry, the interactions of various forms of hydrogen, especially H and D atoms, need to be understood at low temperatures, down to a few K. At such cold conditions, other processes such as small barriers, isotope enrichment, nu-clear spin restrictions, or cluster formation play a role [ger06b]. Recent publications emphasize that more experimental information on low temperature hydrogen chemistry is needed for tracing the chemical and physical processes occurring in the cosmic dark age of our universe [glo08a] or in dense interstellar clouds [wal04]. During the period before the first stars formed, the temperature dropped below 100 K and the density fell below 1 hydrogen atom per cm3. At those times, matter consisted just of hydrogen, he-lium and traces of lithium gas. Collisions with H atoms have played a central role. The first molecule in the universe was probably HeH+ created via radiative association of He with H+ [puy07]. Forming H2 molecules, the dominant coolant in primordial clouds, needed the detour via associative electron detachment, H- + H → H2 + e-. Published values for this process differ by nearly an order of magnitude introducing significant uncertainties into the predictions [glo06a].

After heavier atoms have been produced in stars, chemistry became much richer. None-theless, no efficient gas phase process for H2 formation has been found and for decades it is assumed that hydrogen molecules are formed on the surface of interstellar grains [glo03]. One of the aims of the new instrument was to search for special ions X+ which can play the role of a H-H catalyst at low temperatures. In general, the first step in such a process must be radiative association of X+ and H, followed by the hydrogen abstrac-tion XH+ + H → X+ + H2. The net result is formation of a hydrogen molecule from the two atoms. Potential candidates for X+ are C+, CH3

+, C2H2+, or C3H2

+, and certainly many larger ions.

In addition to hydrogen, there is a small fraction of deuterium which plays a special role in low temperature chemistry. Despite a deuterium abundance of only 10-5, many deuterated molecules have been detected in the interstellar medium including doubly and even triply deuterated species. This phenomenon, called ‘isotope fractionation’ is due to a complex interplay between zero point energies, symmetry selection rules, and barriers [ger06b]. Many details of D-H exchange reactions must be known in order to correlate observed abundances of deuterated molecules with the cosmic D/H isotopic abundance ratio. An important experimentally unexplored class of reactions is isotope enrichment in collisions of ions or neutrals with D-atoms.

3

Experimental studies involving H atoms H-atoms are used in many experiments in fundamental physics and chemistry leading to the development of a variety of sophisticated beam sources and instruments. Typical examples include the hydrogen maser which uses a beam of atoms, a separating magnet and a resonator [gol60], high resolution scattering experiments using crossed beams [toe79], attempts to observe Bose-Einstein condensation of a trapped, dilute gas of atomic hydrogen [fri98], or resonance enhanced two-photon spectroscopy of magneti-cally confined H-atoms [hij00]. Polarized hydrogen and deuterium atoms are standard targets in storage ring experiments. In recent years several sophisticated instruments have been constructed and used to study H2 formation on macroscopic surfaces or inter-stellar grain analogues [pir00], [per05], [isl07]. Very important are other chemical proc-esses induced via H bombardment of surfaces, e.g. during the growth of diamond layers. Experimental studies of reactions between solid methanol and D-atoms at 10 K have been reported recently [nag07].

There have been several approaches to study gas phase reactions between ions and at-oms. The literature until 1992 dealing with this subject has been reviewed by Sablier and Rolando [sab93]. The laboratory studies done since 1993 for reactions of positive and negative ions with H, N, or O atoms have been summarized very recently in a com-prehensive review by Snow and Bierbaum [sno08]. Low energy ion beam experiments with H atoms as target are rare. One is briefly mentioned in [gen79]; however, no fur-ther publication appeared. An ambitious new experimental approach is the merged an-ion-neutral beams arrangement which is under construction specifically for measuring detailed cross sections for the reaction H- + H → H2 + e- [bru08]. This fundamental reac-tion has been studied so far only once at thermal conditions, many decades ago [scm67].

Most results for ion – H atom reactions have been obtained with swarm techniques op-erating at room temperature. Some of the earliest investigations of such processes were performed in the Boulder Aeronomy Laboratory using initially a flowing afterglow ar-rangement and later a selected ion flow tube (SIFT) [feh71], [feh73], [ada76]. The com-bination of an ion drift tube and an hydrogen atom source has provided information at higher energies [fed84], while a variable temperature SIFT apparatus extended the tem-perature range down to 120 K [ada85]. There has been one attempt to combine an ion cyclotron resonance cell with a hydrogen discharge source [kar79]; however, many complications due to H - H recombination, metastable atoms and UV photons have made it extremely hard to get reliable results. Reactions of CmHn

+ and some other posi-tive ions with H and H2 are summarized in [sco97] and [sco97a]. Studies with negative ions, e.g., the interaction of Cn

− or CnH− anions with H, have been reported recently from the Boulder ion group [bar01].

A problem with most of the previous experiments is that they do not work at very low temperatures. There have been several activities to create radicals in free expansions or in a laval flow and to study their reactions in these cold environments [can08]. The pre-sent strategy to reach low collision energies is based on confining cold ions via suitable effective potentials created by fast oscillatory electric fields. As summarized recently in [ger08a] this method has removed many difficulties encountered by other techniques. The instrument which will be described in detail in what follows, is a combination of an rf ion trap with a cold effusive beam of neutrals.

This paper is organized as follows. In the experimental section we give first a short overview over the whole apparatus and a brief summary of the low temperature ion trapping technique. After an overview of the vacuum system, the hydrogen atom source

4

is described in detail, followed by a description of tests of its combination with the trap and some hints to the kinematics of this trap - beam arrangement. The selected reactions presented in a separate section, have been selected mainly for illustrating the features of the instrument. The outlook proposes further improvements of the apparatus and gives hints to challenging experiments which are under way.

2 Experimental In the first rf ion trapping machines, the various modules for ion preparation and detec-tion have been arranged coaxially with the ion trap in the center [ger95]. In order to integrate a neutral beam and to provide efficient differential pumping, it was necessary to make major modifications. There have been several previous attempts to combine an ion trap and a beam of neutrals. One of the very early successful realizations was the study of spin-dependent collision processes between stored ions and a beam of polarized atoms [deh69]. In our group a first setup has been assembled by E. Haufler [hau96] who studied, with a clever modification of an ion trapping machine, the electron transfer from a pulsed supersonic beam of H2 molecules or rare gas atoms to cold stored ions. In a slightly different geometry, the association of H2O molecules with trapped CH3

+ ions has been observed at low temperatures [luc02]. This experiment demonstrated the pos-sibility to utilize condensable gases together with low temperature ion traps. Other ex-amples of beam-trap arrangements are the addition of an atomic nickel beam [scl03] or a carbon beam source [sav05a] to an rf ring electrode trap. The first exposure of a laser-cooled ion crystal, confined in a linear quadrupole trap to a slow molecular beam has been reported very recently [wil08].

2.1 AB-22PT apparatus In order to get an impression of the new instrument, Fig 1 shows many details of the AB-22PT (Atomic Beam 22-Pole Trap) apparatus. The left part is dominated by the H-atom beam source and the chambers, enabling differential pumping. The modules for ion preparation (storage ion source and QP mass filter), ion trapping (22PT) and detec-tion (QP mass spectrometer and scintillation detector), are standard for a typical ion trapping instrument. Since the rf trapping technique and machines with multi-electrode traps have been described thoroughly in the literature [ger92], [ger93b], [ger95]; [asv04a] [sav05b] [ger06a], [ger08a], [ger08b] only the most important facts are sum-marized in the following. The H atom beam, its integration into the trapping machine, and various experimental tests will be discussed in great detail in a separate section.

Primary ions are continuously generated by electron bombardment of neutral gas in a labyrinth type rf storage ion source [ger92]. Advantages of this special ion source in-clude high collection efficiency of ions, chemical ionization, and many thermalizing collisions of the ions already in the source. The double-H-shaped storage volume (see Fig. 1) is defined by a stack of 8 rectangular molybdenum plates separated by ruby balls. The plates are alternatively connected to the two phases of an rf generator. To prevent ions from leaking out of the trap, the top and the bottom of the storage volume are closed by an adequate dc bias voltage applied to the endplates, each containing a small slit for the electron beam. Ionization of the neutral precursor gas is achieved through electron bombardment from a rhenium filament (0.3 mm) that is situated above one of the side channels.

After generation, the ions are trapped in an environment where they can relax by colli-sions to the temperature of the source (typically 350 - 450 K). They also can react with other neutral molecules. This allows for example to chemically quench unwanted ions

5

or to synthesize species which cannot be obtained directly by electron impact. Variation of the trapping parameters, precursor gas mixture and pressure permits a good control over the generated ion composition. In most applications the trap is kept closed by ap-plying a positive bias voltage to the exit electrode. Superimposing an adequate pulse of a few µs or ms, the stored ions are gently extracted from the source. The storage prop-erty of the source allows low pressures to be used. Therefore the pressure of the precur-sor gas, leaked into the main chamber, can be very small.

The ions are transferred to the electrostatic quadrupole bender using an rf-quadrupole. It is 245 mm long and consists of 4 half-moon shaped rods with an effective rod diameter of d = 10 mm circumscribing an inner circle of a radius r0 = 4.35 mm. Depending on the needs this one is operated as an ion guide (high frequency, no DC difference, high transmission of low energy ions), in the low mass pass mode (high frequency and DC difference) or as a standard mass spectrometer (high mass resolution but impairment of the energy distribution). For details see [ger92].

After the bender, the ions enter the 22PT, the most important module of this instrument. [ger92c], [ger93b]. The radial rf field is created by 22 stainless steel rods with a diame-ter d = 1 mm, equally spaced on an inscribed radius r0 = 0.5 cm. Using operating condi-tions in the adiabatic regime (see [ger92]), this creates a wide nearly field-free effective trapping environment. As emphasized in [ger08a], the influence of the micro motion driven by the oscillating confining field - effects referred to as radio frequency heating - is negligible if the trap is operated at high enough frequencies. Operating conditions must be chosen according to the mass of the ions. For the CO2

+ ions used for testing the hydrogen beam, it is operated with an rf amplitude of V0 = 50 V and a frequency of f = 15.86 MHz. Using the standard definitions and formulas [ger92], these operating parameters lead to an effective potential and a stability parameter of V*(r=r0) = 0.668 eV and η(r=r0) =0.097, respectively. For the interaction with the atomic beam it is important to note that the stored ions explore a wide region in the ra-dial direction, the turning radius being determined by the transverse kinetic energy. For 10 meV the turning radius is 0.8r0, i.e., the ion cloud has a diameter of 8 mm. In the axial direction, two electrodes are used to close the trap with dc voltages of less than +100 mV, which is sufficient to confine the ions. The volume of the ion cloud is typi-cally 1.5 cm3.

For obtaining a cold, localized environment, the trap is surrounded by a copper box which is mounted onto the cold head of a closed-cycle He refrigerator (Leybold RGD 210 with compressor RW2/3). Via thin sapphire sheets, the trapping electrodes are also connected to the cold head. A second heat shield which is connected to the first stage of the cryogenerator (see Fig. 1) reduces the load from the 300 K black body radiation to below 0.2 W. This allows one to reach temperatures as low as T22PT = 10 K. With an-other cold head (Sumitomo SRDK-101E), 3.6 K has been reached. The trap temperature is measured using a calibrated silicon diode (LakeShore Cryotronics, model DT-470) mounted directly onto the copper wall. In order to operate the trap at variable tempera-tures between 10 K and 300 K, a resistive heater is mounted between the cold head and the trap. The translational and internal degrees of freedom of the ions are coupled to the cold environment by inelastic collisions with helium or hydrogen buffer gas. In most applications we utilize a pulse of buffer gas which is intense enough that the transla-tional and internal degrees of freedom of the ions are cooled within a few milliseconds [ger03a], [ger08a], [ger08b].

With the exit electrode gate momentarily lowered to a small negative voltage, the ions leave the trap and are injected into a QP mass spectrometer via a lens system. This

6

quadrupole is 260 mm long and consists of 4 rods with a diameter of d = 18 mm, cir-cumscribing an inner circle with r0 = 7.84 mm. Using a commercial rf power supply at a frequency of 1.637 MHz (BALZERS rf-generator QMH510 with control unit QHS511), it is operated in the mass-selective mode [daw76]. The main parts of the detector (see Fig. 1) are the stainless steel converter ("door knob"), a scintillator, coated with a thin aluminum layer and a photomultiplier [dal60]. The converter is set to a high negative voltage (-30 kV), attracting positive ions. These impinge with high velocities upon the metal surface and kick out several secondary electrons. These negative particles are accelerated in the potential gradient between the converter and the grounded scintillator situated at the opposite side. In the scintillator the fast electrons are converted into pho-tons which are registered by the photomultiplier located outside the vacuum chamber. The output pulses of the photomultiplier are accumulated in a 100 MHz counter or in a multi channel scaler.

The basis of trapping experiments with "destructive" detection schemes is (i) to fill the trap repetitively with the same amount of primary ions, (ii) to store them for various times and (iii) to extract all ions, primaries and products for mass analysis. Independent on whether one is interested in metastable lifetimes of ions, in recording ion spectra or in determining rate coefficients for reactions, this sequence is repeated many times for all ion masses of interest. Several parameters are changed, the most important being the storage time. Others include the number density of reactant, coolant or chemical probing gas, the temperature of the trap, laser wavelength or intensity and, especially in the re-sults reported here, various parameters of the neutral target beam. In order to avoid space charge heating usually only ~ 1000 primary ions are filled into the 22PT. Depend-ing on the conversion rates (relaxation, reaction, or decay rates), they are trapped for time periods varying between ms and min. In general there is no loss of ions and also no creation of additional ones. Therefore the number of ions per filling, remains constant, i.e., ∑Ni = N0. For better statistics, the measurements are averaged over many iterations, but the Ni(t) are normalized to one iteration for better comparison.

The results of such a procedure are illustrated in Fig. 2 showing the interaction of CH+ ion with hydrogen molecules and atoms. The data on the left side has been recorded with the discharge switched OFF. It is obvious from the plot on the right side where the discharge is turned ON that additional products are formed. In this example, the trap has been filled each time with 120 ± 11 primary ions. The ion creation and filling procedure is usually so stable that the fluctuations are purely statistical. If only H2 is present (left side), the fast hydrogen abstraction reaction CH++H2 → CH2

+ + H is followed by CH2

++H2 → CH3++H. Under the conditions of our experiment the final product is CH3

+. Further hydrogenation of this ion only can occur via radiative association of CH3

+ with H2. For this process a time constant of more than 7 hours can be estimated from the hy-drogen number density in the trap (6 × 108 cm-3) and the very low rate coefficient (5 × 10-14 cm3s-1 at T22PT = 50 K) [ger92c]. If the discharge is switched ON (right panel), a significant fraction of the hydrogen molecules dissociate to H atoms. It is obvious that they convert the primary CH+ ions into C+ ions via the 0.398 eV exothermic reaction CH+ + H → C+ + H2. The number densities of H and H2 for the two situations ON and OFF are reported in the figure caption.

Comparison of the two sets of data leads to the surprising observation that the decay rate of the CH+ ions increases by a factor of 3.5 if the discharge is switched ON. Expo-nential fits to the CH+ data lead to decay time constants of τOFF = 1.38 s and τON = 0.39 s. As will be explained in more detail below, this is not due to the reactions

7

of CH+ with H atoms but due to an increase in the overall hydrogen density. As can be seen from the production rate of CH2

+ products, there are also more H2 molecules if the discharge is running. The reason is the efficient guiding of atoms in the hexapole mag-nets and creation of H2 molecules via H - H recombination inside the box surrounding the trap.

For evaluating such experimental data, the temporal evolution of the number of stored ions, Ni, is modeled by solving numerically a system of differential equation simulating the interactions in the trap. Sometimes the situation is very simple and only two chan-nels are of importance, in other cases, too many reaction rates play a role and additional independent measurements are needed for fitting the data [sav05b]. In general the trapped ion cloud is not only modified by the reaction of interest, but relaxation proc-esses, reactions with background gas and loss or gain from the trap can play a role. Tests have to be performed to ensure that the trapping field is sufficient. Loss also can be caused by reactions leading to non detected masses. In most cases excellent fits can be obtained as illustrated in Fig. 2.

2.2 Vacuum system An excellent vacuum is required for ion-trapping experiments in order to avoid reac-tions with traces of residual gas during long storage times. As can be seen from Fig. 1, the vacuum system of the AB-22PT apparatus is built from bakeable, UHV quality components, separated by differential walls. The left part of Fig. 1 shows the double differentially pumped neutral beam source. The H atom source chamber, containing the discharge tube and a cold head, is pumped by a turbo-molecular pump (Pfeiffer, TPU 2200). For H2, it has a nominal pumping speed of 2800 l/s which is reduced by 15 % due to the splinter shield. It is backed with a turbo-molecular pump (Pfeiffer, TPU 240) in series with a rotary pump (Pfeiffer, DUO016B). A skimmer with a diameter of 2 mm connects the source chamber with the hexapole chamber. These two chambers can be separated by closing a gate valve mounted directly after the skimmer (not shown in the figure). Two magnetically suspended turbo pumps (Leybold, T340M, 370 l/s for H2) evacuate the hexapole chamber and the central chamber housing the 22PT. The ion source chamber is pumped with one turbo-molecular pump (Pfeiffer, TMU 200MP, 105 l/s for H2). The turbo pumps are backed with a series of drag (Pfeiffer, TPD011, 10 l/s) and diaphragm (Pfeiffer, MVP015, 15 l/s) pumps in order to get sufficient compression for hydrogen and to avoid oil pumps.

To monitor the pressures in the ion trap regions also at 10-10 mbar and below, the cham-bers are equipped with Bayard-Alpert ionization gauges. For absolute pressure and flow measurements they are calibrated using a spinning rotor gauge (MKS SRG2), connected directly to the chambers. The number density of the reactant gas inside the 22-pole trap is determined with an SRG2 as described below.

A gas inlet system allows us to introduce pure gases and well-characterized mixtures into various locations of the machine. The ion source is supplied with gas via two sap-phire sealed leak valves. One is used for the precursor gas of the ion of interest, the other one is used for preparing CO2

+ ions which are routinely used for calibrating the H atom density. For controlling the flow entering the 22PT, either of two sapphire sealed leak valves (Varian Model 951) are used or, for pulsed applications, fast piezoelectric valves developed in our group.

The hydrogen atom source uses a constant flow of ultra pure hydrogen gas (H2 99.9999 vol %, Messer Griesheim or D2 99.7 wt %, Messer Griesheim). Important for obtaining and maintaining high dissociation degrees is a second gas line leading to the hydrogen

8

atom source. It is used for leaking small amounts of water into the discharge tube. In order to create a reproducible flow, the water reservoir is held at 0° C and to avoid con-densation, the gas inlet tubes are heated to a constant temperature of 60 °C.

2.3 H atom beam Discharge tube Special emphasis has been given to the development of a well-defined, effusive beam source for hydrogen atoms, mounted onto a cold head and the integration of two hex-apole guiding magnets (see Fig. 1). There are many publications describing the creation of H-atoms by dissociation of hydrogen molecules using radio-frequency or microwave discharges. In some applications thermal dissociation, e.g. on hot filaments, is also prac-ticable [sca91]. Depending on the boundary conditions of the experiment some of which are mentioned in the Introduction, quite different solutions have been found. In the pre-sent work, we allow for a rather large acceptance volume, defined by the ion trap; how-ever, restrictions are imposed by the need to transmit across a broad velocity range of the H atoms and to keep the H2 background as low as possible. The solution described in detail in what follows, makes use of a well-established rf driven plasma source the design of which is described in [sle81]. It has been chosen because of very high disso-ciation yields and, a very important point for long time trapping experiments, because of long time stability.

The discharge tube is constructed from Pyrex glass, with an inner diameter of 19 mm and a length of 330 mm. On both ends it is terminated with capillaries (see Fig. 1). Their inner diameter and length has been varied between 2 - 2.5 mm and 3 - 5 cm, respec-tively. The discharge is excited by feeding rf power into a coaxial cavity resonant at 27.12 MHz. The cavity consists of a copper cylinder, surrounding the discharge tube, and a 17-turn helix of constant pitch. Details of the cavity design and dimensions are described in [alp59]. With a power of 15 – 20 W the highest dissociation degree is achieved at hydrogen pressures of about ~ 0.2 mbar. The tube and exit capillary are cooled by water circulation for avoiding overheating (see Fig. 1) and for decreasing recombination losses on the walls [wis67].

The literature contains many hints, recipes and tricks for delivering dissociation degrees up to 95% [sle81], [koc99a], [szc00]. It certainly depends on parameters such as the power transferred to the plasma, the properties of the discharge vessel or the location of the coupling coil close to the exit of the tube. Much more important, however, are the schemes for reducing H-H recombination on the walls. An important first step is the procedure for passivating the glass surfaces of the discharge tube towards recombina-tion. In our case we found most effective a procedure whereby cleaning was performed by successive ethanol and methanol washes each followed by deionized water rinse. Finally the tube has been exposed to 20 % phosphoric acid in water for 24 hours. After mounting the tube into the apparatus, a continuous flow of hydrogen was maintained.

The dissociation degree one can achieve with a specific source is a complex interplay of production and loss mechanisms of H atoms. In our setup (see Fig. 1 and [sle81]) the discharge is sustained by the electromagnetic high frequency field coupled into the dis-charge tube with an rf field applicator located outside [alp59]. The result is a “cold” plasma, characterized by a low degree of ionization (ξ < 10-4). The temperatures of neu-trals and ions range between 500 K and 2000 K while the electrons have kinetic ener-gies between 2 and 10 eV. As discussed in [ged93a], H atoms are mainly created by electron impact dissociation of H2 via the repulsive b 3Σu

+ state leading to kinetic ener-gies between 2 and 4.5 eV. Since wall collisions with such energies have a high recom-

9

bination probability, one needs a hydrogen density of at least 0.1 mbar for slowing down the atoms [ged93a]. Hydrogen atoms can get lost by ternary gas phase processes or by surface collisions. The first process becomes relevant only at pressures above 1 mbar [ged93a]. The second process, H-H recombination on surfaces, is usually de-scribed by the recombination coefficient γ. Important for understanding our H atom source is the temperature dependence of γ. Experimental [wis67] and theoretical [gel71] results, reported for Pyrex and quartz, show in very good agreement that γ has two max-ima (γ ~ 0.1) around 900 K and 50 K and a pronounced minimum (γ < 10-4) at 110 K. For a copper surface a minimum (γ < 10-4) has been found slightly below 40 K [sin90].

A process which is interesting in itself, is the reduction of H-H recombination on sur-faces by coating them with H2O or, at low temperatures, with H2. This effect has been mentioned in the literature [sin90], [szc00]. It also has been shown [koc99a] that per-manent addition of traces of oxygen to the hydrogen can decrease recombination losses. In our application the water layer is formed in advance, operating the precooler and the accommodator (see below) at 100 K. The result is a dissociation degree above 90% and stable operation over hours. The procedure and various tests are described in [bor08].

Hexapole magnets Two permanent hexapole magnets are used for guiding and selecting the hydrogen at-oms. Their function is based on the interaction of the inhomogeneous magnetic field with the magnetic momentum of the atom. In good approximation, the same force acts on H and D since the dominant contribution is due to the spin of the electron. The con-tributions from the nuclear spin can be ignored in our application. Passing through an hexapolar magnetic field, the atom experiences, depending on the orientation of the spin, an attractive or repulsive force

rrBF B 2

0

02μ±= , (1)

where µB is Bohr´s magneton, 0r is the distance of the pole tips from the symmetry axis, and B0 is the magnetic field at the pole tip. This force is proportional to r and independ-ent on the angle. Due to this and due to the rotational symmetry, it is straight forward to solve the equation of motion describing the trajectories of hydrogen atoms in a hexapo-lar magnetic field.

Magnetic hexapolar fields can be created by conventional electromagnets [sin90], per-manent magnets [szc00] or using superconducting electromagnets [tos06]. Here, perma-nent magnets have been installed. The first one (Electra Trieste, length 150 mm, inner diameter 2r0= 1 cm, outer diameter 5 cm) is formed from 6 sectors with a straight bore. The hexapole coefficient is specified with B0/r0

2 = 5.52 T cm-2 corresponding to a mag-netic field of more than 1 T at the tip. The second magnet (Vakuumschmelze GmbH, length 10.5 cm, inner diameter 2r0 = 1.4 cm, outer diameter 3.6 cm) is assembled from 24 sectors with a straight bore. The hexapole coefficient is 3.09 T cm-2. For obtaining high number densities of atoms and low background in the 22-pole trap the position of the hexapole magnets, skimmer and apertures have been optimized.

For a detailed analysis of the transmission of the two magnets and the geometrical boundary conditions indicated in Fig. 1, the transition regions from weak magnetic fields into the hexapoles have to be taken into account. However, for a qualitative un-derstanding of the observed transmission maxima which will be presented in Sect. 2.4, a simple estimate is sufficient. It is based on the harmonic frequencies derived from the two hexapole coefficients. In the first magnet, a D atom oscillates in the radial direction

10

with 2.80 kHz, in the second one with 2.10 kHz (oscillation periods are 357 and 476 µs, respectively). The frequencies for H are √2 larger. Since the motion in the axial direc-tion is not coupled with the transverse one, trajectories inside the magnets are sine func-tions. The number of nodes depends on the axial velocity.

As will be discussed below in Sec. 2.4, time of flight measurements revealed that the magnetic transfer system favors transmission of three groups having kinetic energies of 3.6 meV, 8.2 meV, and 21 meV, both for hydrogen and deuterium. In Fig. 3 one typical trajectory illustrates schematically the imaging properties. In this example D atoms move with such a kinetic energy that one gets 1/2 (0.45) of an oscillation period in the first magnet and 1/4 (0.24) in the second one. A more precise calculation, accounting for the length of the magnets and the velocity of D atoms with 8.2 meV leads to the number in brackets. The minor differences can be explained with corrections due to the transition regions.

Accommodator For slowing down the H atoms emerging from the hot discharge to low temperatures, cryogenic cooling has been used. Similar approaches are reported in the literature. For example a cold Teflon coated Pyrex tube (4 mm inner diameter, 20 mm length) has thermalized a flow of H atoms down to 8.5 K [wal82]. In the present experiment a cop-per accommodator is used. As can be seen in Fig. 3, the hydrogen atoms first pass through a glass tube which is water cooled on one end and surrounded by a copper block at the other end the temperature of which is fixed at 100 K. The final temperature (variable between TACC = 10 K and 300 K) is determined by the accommodator, a cop-per block with a 2 mm diameter channel with a total length of 22 mm. The efficiency of the system has been tested measuring density and velocity distributions (see Sec. 2.4). Initial experiments have been performed without the precooler. In this case an accom-modator with an inner diameter of 1.2 mm was required for full thermalization. After installing the precooler, sufficient wall collisions could be achieved with a 2 mm chan-nel.

In order to reach accommodator temperatures as low as 10 K, a two-stage closed cycle He refrigerator cold head (Leybold, RGD210) with a He compressor (Leybold, RW2) has been used. This system provides a cooling power of 2 W at 10 K. As indicated schematically in Fig. 1, copper bridges, the heat conductance of which have been opti-mized [bor08], connect the cold head with the precooler and the accommodator. The temperature of the accommodator is monitored with a calibrated silicon diode (Lake-Shore Cryotronics, model DT-470) mounted directly onto the accommodator surface. A heater is fixed onto the cooling bridge (thermocoax with R = 3.6 Ω, maximum heating power 20 W) which provides stable accommodator temperatures between 10 and 320 K. This heater is also important for regeneration of the ice layer (see below). Using 12 W, the accommodator is warmed up to room temperature in 50 min.

Very important for cooling the H atoms is the two step arrangement allowing for a sud-den change of the temperature from the glass tube exit to the accommodator. This has already been mentioned in the literature [her87b]. In our design, the precooler holds the exit capillary at a constant temperature of 100 K where the H-H recombination coeffi-cient has a minimum. For getting down to 10 K, the critical temperature region, where recombination is especially efficient (50 K, see above), is kept as small as possible. The construction uses a special Teflon part which also acts as a gasket (see Fig. 3).

A detailed drawing of all the elements, confining the atomic beam is given in Fig. 4. The overall geometry has been optimized in such a way that a high flux of atoms can

11

penetrate the trap without hitting any surface, while differential pumping reduces the background of molecular hydrogen to a minimum. One of the problems with the ar-rangement is that the hexapole magnets, being closed tubes, also transfer background. An additional differential cryopumping stage will reduce this effect. The design which is also used for creating ultracold effusive beams is described in [ger08b] (see Fig. 6.8).

Precise adjustment of all apertures, the magnets etc. onto the axis of the apparatus is controlled using an alignment telescope from the detector side. Using an UHV window, the location of the accommodator and of the ion trap is monitored continuously. Ther-mal expansion of the cold head is corrected for by changing the length of the bellows using suitable adjustment screws.

2.4 Test measurements Flow measurements Two important parameters for characterizing the atomic beam source are the total gas flow and the pressure in the discharge tube. The flux Q, leaked through the needle valve, is determined directly by recording the pressure decrease in a calibrated volume (VCAL = 129 cm3)

Q = VCAL (PCAL(t0) - PCAL(t1)) / (t1 - t0) (2)

Due to the high stability of the manometer monitoring PCAL (PTU, Swagelok, relative accuracy of 0.036 %) it takes only a few minutes to get a reliable value from Eq. 2. In addition, we always monitor the pressure PBT at the entrance of the discharge tube using an absolute pressure transducer (MKS Baratron Type 722A). Typical standard experi-mental conditions are Q = 3 × 10-3 mbar l/s and PBT = 0.2 mbar for H2 and at TACC = 300 K. This means that the overall conductance of the discharge tube with accommodator is C = Q / PBT = 0.015 ls-1. Inspection of the details of the geometry of the discharge tube (see Fig. 1) reveals that this conductance is mainly determined by the gas line, the en-trance and exit glass tubes (typically d = 2 mm and l = 5 cm) and the accommodator. In these narrow tubes the operating conditions are just at the upper limit of the molecular flow regime. The pressure inside the discharge tube is about 0.7 × PBT. Note that a criti-cal point is the special Teflon adapter used for sealing the joint between the exit tube and the accommodator (see Fig. 3). A detailed analysis of the conductances of the sys-tem, accounting also for the contribution of viscous flow (~ 25%) and the temperature dependence of CACC can be found in [bor08].

Electron bombardment ion source For performing a variety of tests on the H atom source and for determining the number density of H atoms at the location of the 22PT, the ion trap has been removed and re-placed by a commercial electron bombardment ion source. The ions produced from the traversing beam are injected into the quadrupole mass spectrometer and detected via the Daly detector. Using a mechanical shutter placed between the second hexapole and the ionizer, the atomic and molecular beam can be separated from the background gas.

The overall sensitivity of the arrangement has been calibrated by leaking up to 10-4 mbar of H2 and D2 gas into the chamber and by comparing the ion count rate with the pres-sure determined with the spinning rotor gauge. Emission currents between 0.1 and 5 mA have been used. For an electron energy of 50 eV, the sensitivity for D2 has been deter-mined to be ε = 0.2 mbar-1 (I+ = ε p I-). Electron induced fragmentation of hydrogen molecules is slightly reduced if the electron energy is lowered to 20 eV (from ~5 % to ~1 %); however, this lowers the sensitivity to 0.01 mbar-1. For effusive gas, the error of the sensitivities is smaller than 5 %. In the case of the molecular beam the detection

12

efficiency may be slightly higher due to the directed motion. There may be also minor deviations due to differences in the ionization or probing volume. Both effects which compensate each other, have been assumed to be negligible and the same sensitivity has been used in both cases. Errors are estimated to be less than 30 %.

One simple application of this arrangement is the determination of the dissociation de-gree of the discharge source, α. This measurement must be performed without hexapole magnets for getting the same transmission for atoms and molecules. To avoid mass dis-crimination, the parameter α is just determined from the two count rates of the molecu-lar H2

+ or D2+

ions, ION and IOFF, determined with discharge ON and OFF. The degree of dissociation is defined in accordance with [cul93] as

OFFON II /1−=α (3)

The relative uncertainty of α is 2 % for α > 30 % and increases to 5 % for smaller dis-sociation degrees. With this simple experimental tool, systematic studies have been per-formed. Very high dissociation degrees up to 95 % have been observed, depending on geometrical arrangements, the treatment of the glass tube and as a function of parame-ters such as gas throughput, rf power and different surface coatings of the accommoda-tor. Note that the parameter α characterizes the beam at the location of the ion source. For analyzing reactions in the ion trap, it is better to talk about the H / H2 ratio which includes also the molecular background. As will be seen below, this value is often much smaller, depending on the temperature of the trap.

Absolute densities Most important for studying chemical reactions between ions and the atoms are the ab-solute number densities of the hydrogen atoms one can achieve in the region of the trap. Fig. 5 shows the temperature dependence of the number density of deuterium, nD, measured with the calibrated ionizer (squares) and using chemical probing with CO2

+

(filled circles). The second method is described in detail in [bor08c] and will be briefly mentioned below. The discharge tube has been prepared with the H2O coating proce-dure described above. The minimum of nD at 20 K can be explained with efficient H-H recombination on a H2O surface in this temperature range. Going further down in tem-perature adsorption of hydrogen molecules on the accommodator surface reduces this effect. Similar observations have been reported in the literature [wal82], [her87b]. The decrease of nD above 120 K is due to evaporative loss of the passivating water surface from the accommodator leading also to recombination loss. More details can be found in [bor08].

It is not straightforward to determine the transfer efficiency of the two hexapole mag-nets just by comparing absolute number densities. Since we work with permanent mag-nets they have to be removed from the vacuum system complicating quantitative com-parisons. In addition, as already discussed above, the imaging features depend on the velocity of the atoms, i.e. on the temperature of the accommodator. Many comparisons, discussed in [bor08], indicate an overall amplification factor of one order of magnitude. The maximum effective number densities reached with the geometry shown in Fig. 1, was 2 × 109 cm-3, removing the magnets leads typically to 108 cm-3.

One disadvantage of the magnets is that their installation also leads to an increase of the molecular hydrogen background. This is due to two effects, the reduction of the effi-ciency of differential pumping by the long tube like structures, and the significant pro-duction of molecular background via atom - atom recombination on the walls. Many atoms hit the inner surface of the magnets and produce there additional molecules. This

13

is not so problematic in the differentially pumped first chamber. Most critical are atoms which hit a surface inside the trapping box. Inspection of the geometry (Fig. 4) leads to the suspicion that there are atoms which make it just through the second magnet and the aperture ME2, but not through SA. Such atoms first increase the number density of at-oms in the trap; however, they finally form molecules. Note that they contribute espe-cially to the background in the trap due to the low pumping speed.

The consequences have already been mentioned in combination with the results shown in Fig. 2. Instead of getting a lower flow of molecules when the discharge is switched on, the reactions shown in the right panel indicate an increase of H2 almost by a factor of 3. An efficient way to improve the H / H2 ratio inside the trap is to operate the copper box at very low temperature. One reason is the reduction of molecules due to condensa-tion on the walls (cryopumping), and the other one, the reduction of H-H recombination inside the trap due to the hydrogen coating. For example, an [H]/[H2] ratio of 3.7 has been obtained at 10 K. Operating at 20 K, the H2 molecular background increases dra-matically (factor 10 - 50) while the H-atom density drops some 30 % since, at this tem-perature, H-H recombination on copper is very efficient ([H]/[H2] = 0.06). More work needs to be done to avoid surface collisions inside the trap or to make use of surfaces passivated with an H2 layer.

For those experiments which are performed with gas leaked directly into the trap, the number density is determined in situ using a spinning rotor gauge (MKS SRG2, accu-racy 5 %,). Both the gas line and the tube to the manometer are connected separately to the center region of the cold housing surrounding the trap. The SRG itself is at room temperature. Since we operate in a pressure regime where the influence of thermo tran-spiration can be neglected [mil63], the number density n can be calculated from the in-dicated pressure PSRG and the measured temperature T22PT using

3

22PT

SRG17 cmK/

mbar/10182.4 −×=T

Pn . (4)

Since the SRG is rather sensitive to vibrations, the ion gauge on the 22PT chamber is calibrated for routine operations, PSRG = C⋅PIG22PT. Note that the calibration factor C includes the conductance of the entrance and exit from the trap and the pumping speed on the 22PT chamber. It also depends on the gas, on the installed ion gauge and on the settings of the ion gauge controller. Typical values are CH2 = 80, CD2 = 105, and CHe = 285. For avoiding errors, the ion gauge controller (AML) is always set to the fixed sen-sitivity (nitrogen, 19 mbar-1) and to an emission current of 0.1 mA. The overall accuracy for the determination of the absolute number density inside the trap has been estimated to be in the range of 10 % to 15 %. This estimate also includes the fact that the number density is not constant in the trap but drops towards the entrance and exit electrode.

Chemical probing with CO2+

The discussions above clearly indicate that one needs reliable methods to determine the density of hydrogen atoms and molecules in situ. As elaborated in detail [bor08c] a method based on CO2

+ ion reactions with H and H2 has been developed. In collisions with H atoms, this ion produces predominantly HCO+ while collisions with molecular hydrogen lead to HCO2

+. As described in [bor08c], many calibration measurements have been performed without the hexapole magnets. This allows us to assume conserva-tion of hydrogen flux, meaning that the number of H atoms, independent on whether they are single or bound in a molecule, remains unchanged if the discharge is switch on.

14

In order to convert the flux into density in the trap one has to account for the different velocities of the atom and the molecule. Since the accommodator thermalizes them to the same temperature, this requires division by 2 . Accounting for the fact that each molecule produces two atoms, one gets the number density of H atoms simply from the number density of molecules without and with the discharge, using

[H] = 2 ([H2]OFF - [H2]ON) . (5)

During the calibration measurements [bor08c], the rate coefficients k(T) have been ex-tended towards lower temperatures. The rate coefficients for reactions with atoms do not depend on the temperature of the CO2

+ ions nor on the velocity of the atoms. A mean value of k(T) = 4.6 × 10-10 cm3s-1 is recommended for H as target, and 2.2 × 10-10 cm3s-1 for D. For collisions with molecules, the reactivity increases with falling temperature, reaching the Langevin values at 15 K. These experimental results have been parameterized as k = α (T / 300 K)β with α = 9.5 × 10-10 cm3s-1 and β = - 0.15 for H2 and α = 4.9 × 10-10 cm3s-1 and β = - 0.30 for D2.

These rate coefficients are now used routinely as calibration standard for determining the densities of hydrogen and deuterium atoms and molecules in the region where the ions are trapped. This method has the advantage that it is determined automatically for the overlap integral of the spatial distributions of the ions and the neutrals. As discussed above, the CO2

+ ions usually explore a cylinder with a diameter of 7 to 8 mm. Changing the mass requires either readjusting the effective potential or one has to use a correction factor. The neutral beam can be assumed to have a homogeneous density in the region of overlap which is indicated schematically in Fig. 4. Typical beam diameters are 5 mm. As discussed above, an additional uncertainty in the determination of the reactant den-sity is due to those H atoms which hit a surface inside the trap. It is assumed that they thermalize to the temperature of the walls and that the probing reaction accounts prop-erly for their contribution.

TOF distributions of cold D2 and of D atoms The velocity distribution of atoms and molecules emerging from the accommodator has been determined using time of flight measurements. A chopper wheel with 2 slits (2mm wide) interrupts the beam on a 37 mm radius. With a chopping frequency of 200 Hz, one gets a time resolution of ~20 μs. The neutrals are detected using the ionizer men-tioned above. The flight path is (374 ± 2) mm, the uncertainty is due to the lengths of the ionization volume. The data shown in Fig. 6. have been recorded at TACC =11 K for D2 molecules (discharge switched off). Comparison with a Maxwell–Boltzmann distri-bution, calculated for an effusive beam with TMB =11.2 K, reveals that there are enough wall collisions in the channel (1.2 mm diameter and 22 mm length) to thermalize the molecules. The small peak at a TOF of 0.3 ms is caused by the 300 K effusive beam emerging from the skimmer hole.

For the atomic beam, the velocity distribution is structured as can be seen from Fig. 7. As already explained above this is due to the harmonic guiding field of the magnets. For the boundary conditions imposed by the present geometry, H-atoms with the specific energies indicated in the figure are transmitted efficiently. Best transmission is obtained for H atoms with a kinetic energy of (8.2 ± 3.3) meV and for TACC = 50 K. As illustrated in Fig. 3, this leads to one half wave in the first magnet and to a quarter wave in the sec-ond one, resulting in a parallel beam. Cooling the accommodator to 12 K leads to a beam in which the 3.6 meV part dominates with an half width smaller than 1 meV. Comparison with the Maxwell–Boltzmann distributions (dotted lines) reveals that, at

15

this temperature, most atoms are not guided in the magnets but removed from the ther-mal beam. For generating a slower beam, one has to use weaker magnets.

2.5 Kinematics, reaction temperature In order to assess the capabilities and limitations of the present arrangement for study-ing reactive collisions at temperatures of a few K or, it is better to say at total energies of a few meV, a detailed analysis of kinematic averaging is required. Some special as-pects, e.g. how to reach very low relative velocities or how to cool efficiently the inter-nal energy of molecular ions, have been discussed recently in [ger08a], [ger08b] and [ger09]. Averaging over quantities like velocity or angular spread of reactants is a quite general problem in gas phase experiments and many investigations have been devoted to such problems [cha71], [lev74], [ger89], [ger92]; however, the present experiment needs some special attention.

The situation in an ion trap is simple and well-defined if the stored ions and the neutral reactants are in thermal equilibrium with the surrounding walls, at a common tempera-ture T22PT. Adding non-reactive buffer gas for cooling does not change the situation as long as the densities are low enough to make ternary processes very slow. In these cases one extracts reaction rate coefficients from the measured data Ni(t) (see for example Fig. 2) by fitting the parameters of an appropriate rate equation system. In the case of a simple two-channel process, the decay of the number of primary ions is described by

1211 NnkdtdNN −==& (6)

where n2 is the number density of the neutrals and k = k(T22PT) is the thermal rate coef-ficient (by definition). N1(t) decreases exponentially with the time constant τ = (kn2)-1 while the number of product ions, N2(t), augments complementarily. For slow reactions or low number densities (i.e., t « τ), the linear approximation N2 = N1(0) t/τ. is often sufficient.

The situation is more complicated if one performs experiments with a beam-beam, beam-gas cell or beam ion-cloud arrangement. For deriving the correct relations be-tween measured quantities and rate coefficients (or cross sections), one usually starts with the ideal case of two well-collimated monochromatic beams with velocities v1 and v2, interacting in a scattering volume dτ . The number of products (or collisions) per unit time is given by [ger89]

τσ dnnggNd 212 )(=& (7)

Here g = |g|= |v1 - v2| is the relative velocity, n1 and n2 are the projectile and target den-sities, respectively. σ(g) is the elementary (or intrinsic) integral cross section. The cor-responding elementary rate coefficient is obviously given by

)(ggk σ= (8)

In contrast to the ideal case, the conditions of a realistic experiment are less well de-fined. Therefore, the actually obtained product rate, 2Nd & , is an average over the veloc-ity distributions of the reactants. Instead of Eq. 7 one obtains the six-dimensional inte-gral

( ) ( ) τdnnddNd ∫ ∫=1 2

2211212 ,,v v

vrvrvv& (9)

16

Assuming that the velocity distributions are independent of the spatial coordinate r, the density functions ni(r,vi) can be factored using normalized probability functions fi(vi) (i = 1,2)

( ) ( ) ( )iiii fnn vrvr =1, (10)

With f1(v1) and f2(v2) we define f(g), the distribution of the relative velocity

( ) ( ) ( )∫∫=*)(

221121

21 vv

vvvv ffdddggf (11)

The asterisk indicates symbolically that the integration must be restricted to that sub-space (v1, v2) where |v1 - v2| ∈ [g, g+dg]. f(g)dg denotes the probability that the relative velocity lies in the interval [g, g+dg]. With the mean relative velocity, given by

( )∫∞

=⟩⟨0

dggfgg (12)

we define the effective cross section and the effective rate coefficient as

( ) ( ) ( )∫∞

><=><

0dggfg

gggeff σσ , (13)

( ) ( )∫∞

=0

dggfggkeff σ . (14)

With this, we obtain a result that is very similar to Eq. (7),

( ) ( ) ( ) τσ dnnggNd eff rr 212 ⟩⟨⟩⟨=& . (15)

For evaluating specific experimental results, the velocity and spatial distributions func-tions must be known. For the present experiment, we assume here for simplicity that the beam of hydrogen atoms is mono-energetic and that the ions are thermalized to the tem-perature of the trap. For this situation, Eq. 11 can be evaluated analytically, resulting in the generalized Maxwell-Boltzmann distribution [cha71], [ger92].

( ) ( ) ( ) ( ) .2

exp2

exp2 21

2

221

2

2

1

2/122 ⎥

⎦

⎤⎢⎣

⎡⎟⎟⎠

⎞⎜⎜⎝

⎛+−−⎟⎟

⎠

⎞⎜⎜⎝

⎛−−×= vg

kTmvg

kTm

vgkTmgf π (16)

In order to be consistent with the literature, we have used in this formula v1 for the labo-ratory velocity of the H atoms and T2 and m2 for the ions, which play in this treatment the role of the target. An approximation for the half width (FWHM) of this distribution is

2121/1.11 kTEmmE =Δ , (17)

where m1 and E1 is that mass and the laboratory energy of the H atoms, respectively. The following example illustrates the importance of this result. For the H + CH5

+ sys-tem at T2 = 10 K and E1= 3.6 meV, one obtains ΔE = 1.4 meV while the half width of the H atom peak (see Fig. 7) is only 0.75 meV. At low H atom velocities, the general-ized Maxwell-Boltzmann distribution approaches a normal Maxwellian fM(g;TC) with a reduced temperature TC = m1/(m1+m2)T2. This has the interesting consequence that, in the case of H + CH5

+, one can reach sub-K collision temperatures with 10 K ions, pro-vided the H-atom beam is slow enough.

17

Inspection of Fig. 7 reveals that our H-atom beam is not yet ideal and the assumption of a monoenergetic beam cannot be made. Therefore, further analysis requires to integrate numerically the generalized Maxwell-Boltzmann distribution over the velocity distribu-tion of the H atom beam. Also an analytical approximation, describing the hydrogen atoms with the three Gaussians shown in Fig. 7 lead to rather good results. For the CH+ + H system, presented in the following chapter, the resulting corrections are negligible since the elementary rate coefficients depend only weekly on the relative velocity. In contrast, a detailed treatment is required for the CH5

+ + H collision system. This is the subject of a separate paper [ger09a].

3 Typical applications During the development of the instrument, a variety of reaction systems have been stud-ied, demonstrating the various features of the ion trap - neutral beam combination. The high sensitivity and the wide dynamic range of the trapping technique allows us to work with very low number densities or to measure very small rate coefficients. A criterion for estimating the limits of the instrument is that 1 product ion must be formed from 10.000 primary ions after an interaction time of 100 s. This means that, for a fast reac-tion, i.e., k = 10-9 cm3s-1, a density of 103 cm-3 would be sufficient if there are no com-peting or interfering processes. Using the typical effective H-atom number density of 109 cm-3, achieved with the present setup, rate coefficients of k = 10-15 cm3s-1 are within our reach. This is very important because it opens up the possibility to observe radiative attachment of H atoms to small polyatomic ions. Note that up to 106 ions already have been stored, however, energy broadening by space charge becomes a problem. Also storage times over 15 min have been used in special cases.

Several systems have been selected for testing the machine. For example the reactions of CO2

+ with hydrogen atoms and molecules already has been mentioned as a calibra-tion standard for in situ determining H-atom densities [bor08c]. Other systems were chosen to answer open questions such as the unknown reactivity of CH+ + H at low temperatures. A general observation is that, in the case of certain hydrocarbon ions, col-lisions with H2 often lead to hydrogenation while reactions with H atoms cause dehy-drogenation, i.e. formation of H2. The near thermoneutral system CH5

+ + H ↔ CH4

+ + H2 is used as an example for such an H atom transfer. Low temperature rate co-efficients will reduce significantly the uncertainty in the proton affinity of methane; however, as will be shown below, the molecular background complicates the observa-tion of H-abstraction via the H target.

CH+ + H The CHn

+ (n = 0-5) ions and their collisions with H atoms play an important role in the carbon chemistry in diffuse interstellar molecular clouds. A still unsolved problem is the observed abundance of CH+. One of the key reactions, besides its formation, is the de-struction of this ion via the exothermic hydrogen abstraction process

CH+ + H → C+ + H2 + 0.398 eV . (18)

The endothermic reverse process has been studied in great detail, especially in the threshold region using the guided ion beam technique [erv84]. Thermal rate coefficients are extremely small at low temperatures as can be seen from a detailed evaluation of measured cross sections, k = exp(-4575 K / T ) × 10-9cm3s-1, reported in [ger87]. Also radiative association, C+ + H2 → CH2

+ + hν is very slow as discussed in [ger92c]. With other words, C+ formation via reaction (18) can be studied without being perturbed by the molecular hydrogen background. Therefore the exothermic reaction (18) is an ideal

18

test case for the new instrument. An additional interesting experimental aspect is that almost exclusively the population of the rotational states changes with T22PT. At 5 K the molecule is predominantly in its rotational ground state while, at 100 K, states up to j = 7 are populated. Due to the mass ratio, the collision energy is primarily determined by the H atoms, i.e. via TACC.

First results already have been reported in [luc05]. The details of the measuring proce-dure have been discussed above in combination with the two sets of data plotted in Fig. 2. Also the number densities of H and H2 with discharge OFF and ON which have been calibrated with CO2

+, are given in the caption of this figure. For TACC = T22PT = 50 K a rate coefficient of 1.3 × 10-9 cm3s-1 has been determined. Raising the accommo-dator temperature to 100 K and the trap temperature to 80 K, the rate coefficient drops to 8.7 × 10-10 cm3s-1. Inspection of Fig. 7 reveals that the higher temperature of the ac-commodator leads to a significant increase of atoms having a kinetic energy around 8.2 and 21 meV.

These two measured values are compared with other experimental and theoretical re-sults in Fig. 8. The non-thermal results of the SIFDT technique have been converted using the simple approximation KEcm = 3/2 kBT. This may be questioned since the influ-ence of translational and rotational energy may be quite different in the present reaction. The same holds for our data, which are just plotted at T = T22PT. It is obvious that we can obtain more detailed information, perhaps even state specific cross sections, by varying in a more systematic way the energy of the H-atom beam and the rotational temperature of CH+. Corresponding work is in progress.

Reaction (18) has been studied recently by quasiclassical trajectory (QCT) calculations and phase space theory (PST), employing a global single-valued potential energy sur-face [hal07]. The results shown in Fig. 8 are in reasonable agreement with the experi-mental values; however there are still many open questions which are due to the use of classical or statistical methods. So far, the only full quantum mechanical treatment of reaction (18), a RIOSA-NIP calculation [sto05], underestimates the measured value by almost a factor 10. This indicates that more experimental and theoretical work is needed for understanding this fundamental triatomic hydrocarbon collision system in detail and from first principles. An important aspect is to correctly describe the competition be-tween hydrogen abstraction, hydrogen exchange and inelastic collisions which can be distinguished experimentally using the combinations CH+ + D or CD+ + H.

CH5+ + H

Another interesting project which has been started with the new instrument is to explore the fluxional CH5

+ ion via collisions with cold H- or D-atoms [ger05a]. Taking the "es-tablished" values for the endothermicity of the reaction

CH5+ + H → CH4

+ + H2 (19)

no CH4+ products should be observed at temperatures below 50 K; however, experimen-

tal studies performed with the current apparatus, resulted in rate coefficients which are in contradiction to this [luc05]. The energetics of this system have been thoroughly dis-cussed recently based on potential energy surfaces calculated with high level ab initio methods [wan05]. One of the conclusions of these calculations was that reaction (19) is endothermic with 12.7 ± 5.2 kJ/mol at 0 K. Problematic with such predictions is the fact that both the methane cation [wor07] and the protonated methane [coy04] are extremely floppy molecules in which many isoenergetic structures exist due to the nearly free scrambling motion of hydrogen atoms. Also the CH6

+ collision complex belongs to this class of fluxional molecules.

19

Above room temperature, reaction (19) has been investigated in the forward and back-ward direction using the SIFDT technique [fed85]. From a Van’t Hoff plot of the ex-perimental data an endoergicity of 5 kJ/mol, and an entropy change -31 Jmol-1K-1 has been obtained. The experimental finding that formation of CH5

+ in CH4+ + H2 collisions

is exothermic but endoentropic has been corroborated in our laboratory by measuring this process at an extended temperature range from 300 K down to 15 K [asv04a]. The obtained rate coefficients are plotted as full squares in Fig. 9. This figure also shows the present state of our rate coefficients for reaction (19), measured as a function of T22PT [ger09a]. One set of data (open circles) has been obtained with TACC = 92 K, the other one (filled circles) with TACC = 12 K. Note that T22PT determines the internal excitation of the ions while variation of the accommodator temperature leads to the H-atom veloc-ity distributions shown in Fig. 7. The small variations indicate that reaction (19) is al-most thermoneutral. It also allows us to conclude that internal energy of the CH5

+ ions does not play a significant role.

In order to evaluate these observations quantitatively, an analytical function, describing the dependence of the elementary rate coefficients (see above) both on the kinetic en-ergy of the H-atoms and on the internal excitation of CH5

+, has been used for calculat-ing effective rate coefficients. Accounting for the thermal energy of the ions and the velocity distributions shown in Fig. 7 the solid lines following the data points in Fig. 9 have been obtained. The model and the results are described in detail elsewhere [ger09].

An obvious conclusion for the experiment is that we need colder H atoms in order to really probe the threshold onset of reaction (19). As an example, the dashed line in the lower right corner of Fig. 9 shows an effective rate coefficient, calculated with our ana-lytical rate coefficients but assuming an H-atom beam with a kinetic energy of 1 meV.

Deuteration with D Several other reactions of hydrogen atoms with small hydrocarbon ions CHn

+ have been studied with the present technique [bor08]. A wide range of applications of the AB-22PT machine is to follow the sequential deuteration of hydrogen containing ions in collisions with D atoms. These exchange reactions which are always exothermic be-cause of gain in zero point energy, are sensitive probes for reaction dynamics and mo-lecular structure, especially at low temperatures. For example, it is not yet understood why collisions of CH+ with D result in the same amount of C+ products via hydrogen abstraction and CD+ ions via H-D exchange. Such studies are complicated by the D2 background; however, detailed modeling of the kinetics occurring in the trap allows one to extract a lot of interesting rate coefficients.

4 Conclusions and outlook An apparatus has been developed with the aim to study low energy reactions between stored ions and radicals. It combines a linear rf multipole ion trap with an effusive beam source. Alternatively a supersonic beam can be used, leading to a higher velocity of the neutrals but to lower internal temperatures. Another application of the present setup is to use a cold effusive beam in the pulsed mode. Combining this with a fast shutter allows one to select the velocity of the neutrals entering the trap. This can be used to cool ions to temperatures below 1 K and to study ultracold collisions.

Special emphasis has been put onto the integration of a stable and intense hydrogen atom source with an accommodator, connected to a cold head. This allows us to control the velocity distribution of the atomic or molecular beam while the ions can be thermal-ized separately. So far, the instrument has been used for establishing the reactions of

20

CO2+ with hydrogen atoms and molecules as a calibration standard over the full tem-

perature range. First reaction studies have concentrated on the small hydrocarbon ions CHn

+ (n = 1, 4, 5). The possibility to separately vary the temperature and the velocity of the colliding partners offers an important tool for investigation of reaction dynamics. Detailed results will be presented in forthcoming papers.

There are several technical improvements in progress or possible. Especially important is the reduction of the molecular background in the trap. With this aim a beam catcher has been installed at the end of the instrument and a special cryopumping tube has been mounted between the beam source and the trap (see Fig. 6.8 of [ger08b]). For further lowering the kinetic energy of the H atoms below 1 meV, weaker hexapole magnets need to be used and the accommodator has to be operated at the limit of H-condensation. For this a new cold head, reaching 3.6 K already has been installed. A big step forward in experiments between ions and H atoms would be to combine a suitable rf trap for ions with a magnetic trap for the atoms. There have been several successful activities to load H atoms into a magnetic trap with a cryogenic filling technique ex-ploiting interatomic collisions and heat exchange with liquid helium covered surfaces. Densities up to 1014 cm-3 and temperatures below the mK range have been achieved [kil98]. A typical application is ultrahigh resolution two-photon spectroscopy.

However, without these improvements, this versatile instrument already opens up many possibilities to study systems which are of interest for low temperature plasmas or which are of central importance for testing fundamental theories. To the first group be-long radiative association reactions with H atoms which may be important for catalytic formation of molecular hydrogen in interstellar space. A critical test case for the appara-tus is the 3.6 meV endothermic electron transfer from D to H+. The threshold behavior is well characterized by theory [esr00]. The reverse process leads to a fast loss of deu-terons; however, so far no measurements are available.

The formation of molecular hydrogen via the associative electron detachment reaction H- + H → H2 + e- is of central importance for the chemistry of the early universe but not known with the required accuracy. With the current setup, storing H- and monitoring the loss of ions due to collisions with H is straight forward while the detection of the veloc-ity distribution of the ejected electrons is still a dream. Very challenging are studies of Hn

+ collision systems with various H-D ratios. Using laser probing of the ions [scl06], the observation of ortho - para transitions, e.g. o-H3

+ + H → p-H3+ + H, is in principle

possible. Easier, from the point of view of the experiment, is the exothermic H - D ex-change in H3

+ + D. The potential energy surface has been recently published [moy04]; however, for predicting the low temperature behavior of this reaction, the energies in the vicinity of the H3D+ transition state need to be characterized with very high accu-racy, a challenge for quantum chemistry.

Acknowledgments Financial support of the Deutsche Forschungsgemeinschaft (DFG) via the Forschergruppe FOR 388 "Laboratory Astrophysics" is gratefully acknowledged.

21

References [ada76] N. G. Adams, D. Smith, The selected ion flow tube: a technique for studying ion-neutral

reactions. Int. J. Mass. Spectrom. Ion Phys. 21 (1976) 349.

[ada85] N.G. Adams, D. Smith, Laboratory studies of the reactions of HCO+ (and DCO+) and N2H+ (and N2D+) with D (and H) atoms: Interstellar implications. Ap. J. 294 (1985) L63.

[alp59] W. W. Macalpine, R. O. Schildknecht, Coaxial resonators with helical inner conductor. proceedings of the IRE (1959) 2099.

[asv04a] O. Asvany, I. Savic, S. Schlemmer, D. Gerlich, Variable temperature ion trap studies of CH4+

+ H2, HD and D2: negative temperature dependence and significant isotope effect. Chem. Phys 298 (2004) 97.

[bar01] C. Barckholtz, T. P. Snow, V. M. Bierbaum, Reactions of Cn- and CnH- with atomic and

molecular hydrogen. Ap. J. 547 (2001) L171.

[bor08] G. Borodi, On the combination of a low energy hydrogen atom beam with a cold multipole ion trap. PhD Thesis TU Chemnitz, (2008).

[bor08c] G. Borodi, A. Luca, D. Gerlich, Reactions of CO2+ with H, H2 and deuterated analogues.

submitted to Int. J. Mass. Spec. (2008) .

[bru08] H. Bruhns, H. Kreckel, M. Lestinsky, W. Mitthumsiri, M. Schnell, B. Seredyuk, D.W. Savin, M.E. Bannister, C.C. Havener, A. Dorn, O. Heber, M.L. Rappaport, A.M. Covington, A novel merged-beams apparatus for studying anion-neutral reactions. Bull. Am. Phys. Soc. 53 (2008) 149.

[can08] A. Canosa, F. Goulay, I.R. Sims, B.R. Rowe, Gas phase reactive collisions at very low temperature: recent experimental advances and perspectives. in: Low temperatures and cold molecules, World Scientific Publishing, ed. by Ian W M Smith, ISBN 978-1-84816-209-9(2008) 295.

[cha71] P. J. Chantry, Doppler broadening in beam experiments. J. Chem. Phys. 55 (1971) 2746.

[coy04] A. B. McCoy, B. J. Braams, A. Brown, X. Huang, Z. Jin and J. M. Bowman, Ab Initio Diffusion Monte Carlo Calculations of the Quantum Behavior of CH5+ in Full Dimensionality. J. Phys. Chem. A 108 (2004) 4991.

[cul93] R. W. McCullogh, J. Geddes, A. Donnelly, M.Liehr, M. P. Hughes and H. B. Gilbody, A new microwave discharge source for reactive atom beams. Meas. Sci. Technol. 4 (1993) 79.

[dal60] N. R. Daly, Scintillation type mass spectrometer ion detector. Rev. Sci. Instr. 31 (1960) 264.

[daw76] P. H. Dawson, Quadrupol Mass Spectrometry, Elsevier Scientific Publishing, Amsterdam. book(1976) .

[deh69] H.G. Dehmelt, Radiofrequency Spectroscopy of Stored Ions II: Spectroscopy. Adv. At. Mol. Phys. 5 (1969) 109.

[erv84] K. M. Ervin, P. B. Armentrout, Threshold behavior of endothermic reactions: C+(2P) + H2 → CH+ + H. J. Chem. Phys. 80 (1984) 2978.

[esr00] B. D. Esry, H. R. Sadeghpour, E. Wells and, I. Ben-Itzhak, Charge exchange in slow H+ + D(1s) collisions. J. Phys. B 33 (2000) 5329.

[fed84] W. Federer, H. Villinger, F. Howorka, W. Lindinger, P. Tosi, D. Bassi and E. Ferguson, Reaction of O+, CO+, and CH+ Ions with Atomic Hydrogen. Phys. Rev. Lett. 52 (1984) 2084.

[fed85] W. Federer, H. Villinger, P. Tosi, D. Bassi, E. Ferguson and W. Lindinger, Laboratory studies of ion reactions with atomic hydrogen. G.H.F. Diercksen et. al. (eds), Molecular Astrophysics, D. Reidel Publishing Company(1985) 649.

[feh71] F. C. Fehsenfeld, E. E. Ferguson, Rate constants for the reactions of CO2+ + H and H+ + CO2.

J. Geophys. Res. 76 (1971) 8453.

[feh73] F. C. Fehsenfeld, C. J. Howard, E. E. Ferguson, Thermal energy reactions of negative ions with H atoms in the gas phase. J. Chem. Phys. 58 (1973) 5841.

22

[fri98] D.G. Fried, T.C. Killian, L. Willmann, D. Landhuis, S.C. Moss, D. Kleppner, T.J. Greytak, Bose-Einstein condensation of atomic hydrogen. Phys. Rev. Lett. 81 (1998) 3811.

[ged93a] J. Geddes, R. W. McCullough, A. Donnelly, H. B. Gilbody, Dissociation of hydrogen in high - frequency discharges. Plasma Sources Sci. Technol. 2 (1993) 93.

[gel71] A. Gelb, S. K. Kim, Theory of atomic recombination on surfaces. J. Chem. Phys. 55 (1971) 4935.

[gen79] W.R. Gentry, Molecular beam techniques: applications to the studies of ion-molecule colli-sions. in: Gas Phase Ion Chemistry, Vol.2, M.T. Bowers, Ed., Academic Press, New York (1979) 221.

[ger03a] D. Gerlich, Molecular Ions and Nanoparticles in RF and AC traps. Hyperfine Interactions 146- 147 (2003) 293.

[ger05a] D. Gerlich, Probing the structure of CH5+ ions and deuterated variants via collisions. Phys.

Chem. Chem. Phys 7 (2005) 1583.

[ger06a] D. Gerlich, M. Smith, Laboratory astrochemistry: studying molecules under inter- and circumstellar conditions. Physica Scripta 73 (2006) C25.

[ger06b] D. Gerlich, F. Windisch, P. Hlavenka, R. Plašil, J. Glosik, Dynamical constraints and nuclear spin caused restrictions in HmDn

+ collision systems and deuterated variants. Phil. Trans. R. Soc. Lond. A 364 (2006) 3007.

[ger08a] D. Gerlich, The study of cold collisions using ion guides and traps. in: Low temperatures and cold molecules, World Scientific Publishing, ed. by Ian W M Smith, ISBN 978-1-84816-209-9 (2008) 121.

[ger08b] D. Gerlich, The production and study of ultra-cold molecular ions. in: Low temperatures and cold molecules, World Scientific Publishing, ed. by Ian W M Smith, ISBN 978-1-84816-209-9 (2008) 295.

[ger09a] D. Gerlich, G. Borodi, Buffer gas cooling of poliatomic ions in rf-multi-electrode traps, Faraday Discussion, 142, (2009)

[ger87] D. Gerlich, R. Disch, S. Scherbarth, C++H2(j) → CH++H: The effect of reagent rotation on the integral cross section in the threshold region. J. Chem. Phys. 87 (1987) 350.

[ger89] D. Gerlich, Kinematic averaging effects in thermal and low energy ion-molecule collisions: influence on product ion kinetic energy distributions. J. Chem. Phys. 90 (1989) 127.

[ger92] D. Gerlich, Inhomogeneous Electrical Radio Frequency Fields: A Versatile Tool for the Study of Processes with Slow Ions. Adv. in Chem. Phys. LXXXII (1992) 1.

[ger92c] D. Gerlich, S. Horning, Experimental investigations of radiative association processes as related to interstellar chemistry. Chem. Rev. 92 (1992) 1509.

[ger93b] D. Gerlich, Experimental Investigations of Ion Molecule Reactions Relevant to Interstellar Chemistry. J. Chem. Soc. Farady Trans. 89 (1993) 2199.

[ger95] D. Gerlich, Ion-Neutral Collisions in a 22-pole trap at very low energies. Physica Scripta T59 (1995) 256.

[glo03] S. C. O. Glover, Comparing gas-phase and grain-catalyzed H2 formation. Ap. J. 584 (2003) 331.

[glo06] J. Glosik, P. Hlavenka, R. Plašil, F. Windisch, D. Gerlich, Action spectroscopy of H3+ using

overtone excitation. Phil. Trans. R. Soc. Lond. A 364 (2006) 2931.

[glo06a] S. Glover, D. W. Savin, H3+ cooling in primordial gas. Phil. Trans. R. Soc. A 364 (2006) 3107.

[glo08a] S. C. O. Glover, T. Abel, Uncertainties in H2 and HD chemistry and cooling and their role in early structure formation. Mon. Not. R. Astron. Soc. 388 (2008) 1627.

[gol60] H.M. Goldenberg, D. Kleppner, N. F. Ramsey, Atomic hydrogen maser. Phys. Rev. Lett. 5 (1960) 361.

23

[hal07] P. Halvick, T. Stoecklin, P. Larrégaray, L. Bonnet, Cross sections and low temperature rate coefficients for the H + CH+ reaction: a quasiclassical trajectory study. Phys. Chem. Chem. Phys. 9 (2007) 582.

[hau96] E. Haufler, Niederenergetische Elektronen- und Protonen-Transferprozesse, Experimente zur Stoßdynamik in Hochfrequenz-Ionenführungsfeldern. PhD Thesis TU Chemnitz (1996) .

[her87b] A. Hershcovitch, A. Kponou, T. O. Niinikoski, Cold high - intensity atomic hydrogen beam source. Rev. Sci Instr. 58 (1987) 547.

[hij00] T.W. Hijmans, P.W.H. Pinske, A.P. Mosk, M. Weidemüller, M.W. Reynolds, J.T.M. Walraven and C. Zimmermann, Resonance enhanced two-photon spectroscopy of magnetically trapped atomic hydrogen. Hyperfine Interactions 127 (2000) 175.

[isl07] F. Islam, E. R. Latimer, S. D. Price, The formation of vibrationally excited HD from atomic recombination on cold graphite surfaces. J. Chem. Phys. 127 (2007) 064701-1.

[jan03] R. K. Janev, D. Reiter, U. Samm, Collision Processes in Low-Temperature Hydrogen Plasmas. Berichte des Forschungszentrums Jülich 4105 (2003) .

[kar79] Z. Karpas, V. Anicich, W. T. Huntress, Jr., An ion cyclotron resonance study of reactions of ions with hydrogen atoms. J. Chem. Phys. 70 (1979) 2877.

[kil98] T. C. Killian, D. G. Fried, D. Landhuis, S. Moss, L. Willmann, T.J. Greytak, D. Kleppner, Two-photon spectroscopy of trapped atomic hydrogen. American Physical Society, DAMOP Meeting, abstract #HP.94 (1998) .

[koc99a] N. Koch, E. Steffens, High intensity source for cold atomic hydrogen and deuterium beams. Rev. Sci Instr. 70 (1999)1.

[lev74] R. D. Levine, R. B. Bernstein, Molecular Reaction Dynamics. in "Molecular Reaction Dynamics", Oxford University Press, Oxford (1974).

[luc02] A. Luca, D. Voulot, D. Gerlich, Low temperature reactions between stored ions and condensable gases: formation of protonated methanol via radiative association. WDS'02 Proceedings of Contributed Papers, Part II, Safrankova (ed), Matfyzpress (2002) 294.

[luc05] A. Luca, G. Borodi, D. Gerlich, Interactions of ions with Hydrogen atoms. Progress report in XXIV ICPEAC 2005, Rosario, Argentina, July 20-26, 2005, Edited by F.D.Colavecchia, P.D.Fainstein, J.Fiol, M.A.P.Lima, J.E.Miraglia, E.C.Montenegro, and R.D.Rivarola XXIV (2005) 20.

[mil63] G.A. Miller, On the calculation of thermal transpiration. J. Phys. Chem. 67 (1963) 1359.

[moy04] G.E. Moyano, D. Pearson, M. A. Collins, Interpolated potential energy surface and dynamics for atom exchange between H and H3+, and D and H3+. J. Chem. Phys. 121 (2004) 12396.

[nag07] A. Nagaoka, N. Watanabe, A. Kouchi, Effective rate constants for the surface reaction between solid methanol and deuterium atoms at 10 K. J. Phys. Chem. A 111 (2007) 3016.

[per05] H. B. Perets, O. Biham, G. Manicó, V. Pirronello, J. Roser, S. Swords and G. Vidali, Molecular hydrogen formation on ice under interstellar conditions. Ap. J. 627 (2005) 850.

[pir00] V. Pirronello, O. Biham, G. Manicó, J. E. Roser and G. Vidali, Laboratory studies of molecular hydrogen formation on surfaces of astrophysical interest. in: "Molecular Hydrogen in Space", F. Combes and G. Pineaudes Forets (eds.) Cambrigde University Press, Standford (2000) 71.

[puy07] D. Puy, M. Signore, Primordial chemistry from molecules to secondary cosmic microwave background anisotropies. New Astronomy Reviews 51 (2007) 411.

[sab93] M. Sablier, C. Rolando, Gas - phase ion - atom reactions. Mass Spectrometry Reviews 12 (1993) 285.

[sav05a] I. Savić, I. Cermak, D. Gerlich, Reactions of Cn (n=1-3) with ions stored in a temperature-variable radio frequency trap. Int. J. Mass. Spec. 240 (2005) 139.

[sav05b] I. Savić, D. Gerlich, Temperature variable ion trap studies of C3Hn+ with H2 and HD. Phys.

Chem. Chem. Phys 7 (2005) 1026.

24

[sca91] L. Schäfer, C.-P- Klages, U. Meier, K. Kohse-Höinghaus, Atomic hydrogen concentration profiles at filaments used for chemical vapor deposition of diamond. Appl. Phys. Lett. 58 (1991) 571.

[scl03] S. Schlemmer, A. Luca, D. Gerlich, Reaction of trapped ions with metal atoms: O+ + Ni and NiN2

+ + Ni. Int. J. Mass. Spec. 223-224 (2003) 291.

[scl06] S. Schlemmer, O. Asvany, E. Hugo, D. Gerlich, Deuterium Fractionation and Ion-Molecule Reaction at low temperatures. in: Proceedings IAU Symp. No 231, D.C. Lis, G.A. Blake & E. Herbst, eds. (2005) 125.

[scm67] A.L. Schmeltekopf, F.C. Fehsenfeld, E.E. Ferguson, Laboratory measurement of the rate constant for H- + H → H2 + e. ApJ 118 (1967) L155.

[sco97] G. B. I. Scott, D. A. Fairley, C. G. Freeman, M. J. McEwan, N. G. Adams and L. M. Babcock, CmHn

+ reactions with H and H2: an experimental study. J. Phys. Chem. A 101 (1997) 4973.

[sco97a] G. B. I. Scott, D. A. Fairley, C. G. Freeman, M. J. McEwan, P. Spanel and D. Smith, Gas phase reactions of some positive ions with atomic and molecular hydrogen at 300 K. J. Chem. Phys. 106 (1997) 3982.

[sin90] D. Singy, P. A. Schmelzbach, W. Grüebler, W. Z. Zhang, Study of the surface recombination in the production of a dense polarized hydrogen atomic beam at low temperature. Nuclear Instr. and Methods in Phys. Research B 47 (1990) 167.

[sle81] J. Slevin, W. Stirling, Radio frequency atomic hydrogen beam source. Rev. Sci Instr. 52 (1981) 1780.

[sno08] T. P. Snow, V. M. Bierbaum, Ion Chemistry in the Interstellar Medium. Ann. Rev. Anal. Chem. 1 (2008) 229.

[sor94] A. Sorgenfrei, Ion-Molekülreaktionen kleiner Kohlenwasserstoffe in einem gekühlten Ionenspeicher. PhD Thesis, Univ. Freiburg(1994) .

[sto05] T. Stoecklin, Ph. Halvick, Low temperature quantum rate coefficient of the H + CH+ reaction. Phys. Chem. Chem. Phys 7 (2005) 2446.

[szc00] D. Szczerba, L. D. van Buuren, J. F. J. van den Brand, H. J. Bulten, M. Ferro-Luzzi, S. Klous, H. Kolster, J. Lang, F. Mul, H. R. Poolman and M. C. Simani, A polarized hydrogen/deuterium atomic beam source for internal target experiments. Nuclear Instr. and Methods in Phys. Research A 455 (2000) 769.

[toe79] J. P. Toennies, W. Welz, G. Wolf, Molecular beam scattering studies of orbiting resonances and the determination of van der Waals potentials for H–Ne, Ar, Kr, and Xe and for H2–Ar, Kr, and Xe. J. Chem. Phys. 71 (1979) 614.

[tos06] T. Tosaka, K. Koyanagi, M. Ono, T. Kuriyama, I. Watanabe and K. Tsuchiya, Design study of superconducting sextupole magnet using HTS coated conductor for neutron-focusing device. Physica C (2006) 445.

[wal04] C. M. Walmsley, D. R. Flower, G. Pineau des Forets, Complete depletion in prestellar cores. Astron. & Astroph. 418 (2004) 1035.

[wal82] J. T. M. Walraven, I. F. Silvera, Helium - temperature beam source of atomic hydrogen. Rev. Sci. Instr. 53 (1982) 1167.

[wan05] B. Wang, Hua Hou, Computational Study of the Ion-Molecule Reactions Involving Fluxional Cations: CH4

+ +H2 to CH5+ + H and Isotope Effect. J. Chem. Phys. 109 (2005) 8537.

[wil08] S. Willitsch, M.T. Bell, A.D. Gingell, S.R. Procter, T.P. Softley, Cold reactive collisions between laser-cooled ions and velocity-selected neutral molecules. Phys. Rev. Lett. 100 (2008) 043203-1.

[wis67] H. Wise, B. J. Wood, Reactive collisions between gas and surface atoms. Adv. At. Mol. Phys. 3 (1967) 291.

[wor07] H J. Wörner, X. Qian, F. Merkt, Jahn-Teller effect in tetrahedral symmetry: Large-amplitude tunneling motion and rovibronic structure of CH4

+ and CD4+. J. Chem. Phys. 126 (2007)

144305-1.

25

Figure captions Fig. 1: Schematic view of the AB-22PT apparatus, a combination of an atomic beam source with a temperature variable 22-pole ion trap. The central element of the instru-ment is the rf trap which is mounted onto a cold head. Primary ions are produced in a storage ion source, mass selected, and injected into the trap via an electrostatic quadru-pole bender. He buffer gas is used to cool stored ions to the temperature of the sur-rounding walls (T22PT). Primary and product ions are extracted, mass analyzed in the QP mass spectrometer and detected using a scintillation detector [dal60]. The left part of the apparatus is dominated by the H-atom source. Note that, in reality, the cold head and the discharge tube are at a right angle. Hydrogen molecules are dissociated in a water cooled rf discharge tube. The atom flow passes trough a glass tube which is water cooled at the entrance and connected to the precooler (100 K) at the exit. The accom-modator (copper) cools the atoms to the final temperature which can be set at any value between TACC = 10 and 300 K. A skimmer and a second aperture limit the beam travers-ing the trap. The function of the hexapole magnets M1 and M2 is explained in the text and below in Fig. 3.

Fig. 2: Reactions of CH+ ions with H2 and H at T22PT = TACC = 50 K. Plotted is the number of primary and product ions per filling, Ni, as a function of the storage time t. The lines are solutions of a set of differential equations accounting for the competing processes. With discharge OFF (left panel, [H2] = 6 × 108 cm-3) hydrogen abstraction leads in a first step to CH2

+ and in a second one to CH3+. Switching the discharge ON

(right panel) leads to atomic hydrogen with [H] = 4 × 108 cm-3 and to additional prod-ucts, especially to C+. The additional increase of the molecular density ([H2] = 1.7 × 109 cm-3) which can be seen from the faster decay and more CH2

+ and CH3+ prod-

ucts, is explained in the text.

Fig. 3: Temperature variable H-atom source with the two hexapole magnets. Hydro-gen gas is dissociated in a water cooled rf discharge tube. After precooling the gas to 100 K in the exit glass tube, an accommodator determines the final temperature of the ensemble. A skimmer selects a beam of H atoms which passes through two hexapole magnets. The harmonic guiding field and the geometry favor the transmission of spe-cific energies. With atoms, having a kinetic energy of 8.2 meV, one obtains 1/2 wave in the first and 1/4 in the second magnet, as illustrates schematically by the trajectories.

Fig. 4: Optimized geometry of the H atom beam. The indicated apertures and dis-tances (all in mm, axial dimensions are not at scale) are needed for estimating the beam flux, the resulting number density of H-atoms in the 22PT, the gas conductance through the differential walls, and the hydrogen background pressures in the various chambers.

Fig. 5: Density of D atoms, nD, measured as a function of the accommodator tempera-ture, TACC. The measurements have been performed with an electron bombardment ion source which has been installed in place of the 22PT. The two data points shown as filled circles with error bars, have been determined in situ via chemical probing (see text). The agreement corroborates the reliability of the independent methods. The de-crease of nD above 100 K and below 50 K is due to surface recombination. The re-increase below 15 K is caused by the reduced recombination of adsorbing D2 molecules on the surface.

Fig. 6: Time of flight distribution of a cold effusive beam of D2 (TACC = 11.0 K). The agreement with a Maxwell-Boltzmann distribution (solid line, TMB = 11.2 K) indicates that the effusing molecules are thermalized to the temperature of the accommodator.

26

Fig. 7: Relative time of flight distribution of H atoms measured at three different tem-peratures of the accommodator. Without the hexapole magnets, thermal distributions result in TOF distributions indicated by the thin dotted lines. The transmission features of the two hexapoles favour mainly three groups of hydrogen atom with kinetic energies which are indicated in the upper panel. Maximal transmission is obtained for H atoms with (8.2 ± 3.3) meV. The central trajectory for this case is plotted in Fig. 3.

Fig. 8: Temperature dependence of the rate coefficient for the hydrogen abstraction reaction CH+ + H → C+ + H2. The two AB-22PT results at T22PT 50 K and 80 K have been reported in [luc05]. At higher energies experimental results from a SIFDT experi-ment [fed85] are included. A detailed discussion of theoretical aspects of the system can be found in [hal07], including results from phase space theory (PST) and quasi classical trajectories (QCT). Quantum mechanical calculations, using the RIOSA-NIP method, significantly underestimate hydrogen abstraction [sto05].

Fig. 9: Temperature dependence of rate coefficients for the CH6+ collision system. The

surprising behavior of the exothermic reaction CH4+ + H2 → CH5

+ + H has been dis-cussed in [asv04a]. In those measurements the ions as well the neutrals have been at the temperature of the 22PT. For the CH5

+ + H collision system, two sets of data are shown for TACC = 12 and 92 K. The lines are the result of a detailed analysis of the data, using an analytical ansatz for the cross section and accounting for the experimental boundary conditions [ger09a]. The dashed line is the prediction from the model for an H atom beam with a kinetic energy of 1 meV.

G. Borodi et al., AB–22PT Fig. 1


1 10 100 1000

10-10

10-9

k / c

m3 s-1

T / K

SIFDT

PST

AB-22 PT

RIOSA-NIP

QCT

109

Appendix B

Reactions of CO2+ with H, H2 and deuterated analogues

G. Borodi, A. Luca, D. Gerlich Department of Physics, Technische Universität Chemnitz, 09107 Chemnitz, Germany

Int. J. Mass Spectrom., in print DOI: 10.1016/j.ijms.2008.09.004

M

R

GD

a

ARRAA

KIARCL

1

1

ctngpodsamartoinn

1d

ARTICLE IN PRESSG ModelASPEC-13860; No. of Pages 8

International Journal of Mass Spectrometry xxx (2008) xxx–xxx

Contents lists available at ScienceDirect

International Journal of Mass Spectrometry

journa l homepage: www.e lsev ier .com/ locate / i jms

eactions of CO2+ with H, H2 and deuterated analogues

. Borodi, A. Luca, D. Gerlich ∗

epartment of Physics, Technische Universität Chemnitz, 09107 Chemnitz, Germany

r t i c l e i n f o

rticle history:eceived 26 July 2008eceived in revised form 30 August 2008ccepted 5 September 2008vailable online xxx

eywords:on–atom reactions

a b s t r a c t

Combining a temperature variable 22-pole ion trap with a cold effusive beam of neutrals, rate coefficientsk(T) have been measured for reactions of CO2

+ ions with H, H2 and deuterated analogues. The neutral beamwhich is cooled in an accommodator to TACC, penetrates the trapped ion cloud with a well-characterizedvelocity distribution. The temperature of the ions, T22PT, has been set to values between 15 and 300 K.Thermalization is accelerated by using helium buffer gas. For reference, some experiments have beenperformed with thermal target gas. For this purpose hydrogen is leaked directly into the box surroundingthe trap. While collisions of CO2

+ with H2 lead exclusively to the protonated product HCO2+, collisions

+ +
strochemistryf-multipole ion trapold H atomsow temperature rate coefficient
with H atoms form mainly HCO . The electron transfer channel H + CO2 could not be detected (<20%).Equivalent studies have been performed for deuterium. The rate coefficients for reactions with atoms arerather small. Within our relative errors of less than 15%, they do not depend on the temperature of theCO2

+ ions nor on the velocity of the atoms (k(T) lays between 4.5 and 4.7 × 10−10 cm3 s−1 with H as target,and 2.2 × 10−10 cm3 s−1 with D). For collisions with molecules, the reactivity increases significantly withfalling temperature, reaching the Langevin values at 15 K. These results are reported as k = ˛ (T/300 K)ˇ

1 and

rts

1

trothibeprt

with ˛ = 9.5 × 10−10 cm3 s−

. Introduction

.1. Instruments for ion chemistry

The last decades have seen the development of many sophisti-ated instruments, specific tools and clever strategies for studyinghe interaction between ions and molecules in great detail. Tech-iques include crossed and merged beams, swarm methods, ionuides and traps. Many skillful methods have been invented forreparing the reactants. A recent example is the preparationf ultracold molecules via stark deceleration [1]. For extractingetailed information on the products, emerging from the colli-ion region, the instruments are often equipped with sophisticatednalytical tools; for example, ion imaging detectors allow one toap product velocity distributions. A central motivation of such

ctivities is to understand the dynamics of elastic, inelastic andeactive collisions from first principles. For this aim, the best iso measure state-to-state differential cross-sections as a function

Please cite this article in press as: G. Borodi, et al., Int. J. Mass Spectrom. (2

f a variety of parameters. Another goal of studying ion chem-stry is to understand (or model) thermal plasmas and chargedon-equilibrium environments such as boundary layers of mag-etically confined fusion plasmas or interstellar and circumstellar

∗ Corresponding author.E-mail address: [email protected] (D. Gerlich).

mvitmAwd

387-3806/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.ijms.2008.09.004

ˇ = −0.15 for H2 and ˛ = 4.9 × 10−10 cm3 s−1 and ˇ = −0.30 for D2.© 2008 Elsevier B.V. All rights reserved.

egions. The present needs for astrochemistry – in most cases theemperature dependence of thermal rate coefficients – have beenummarized recently [2].

.2. Cold collisions, low temperatures

A nice overview of experimental techniques used in beam scat-ering studies of elementary ion–molecule reactions has been givenecently by Herman [3] who has strongly influenced the fieldsf gas phase ion chemistry and mass spectrometry over morehan four decades. The “classical” ion beams were restricted toigher energies, typically above 1 eV, while, in the thermal regime,

on–molecule reactions were studied for long time predominantlyy swarm and magnetic trapping techniques. The reason is thatlectrostatic fields are not well-suited for handling slow chargedarticles, especially if space charges or field distortions play aole. One of the approaches to reach low collision energies washe merged beam technique where the two reactant beams which

ove in the same direction, can be operated with high laboratoryelocities [4]. A rather general solution is based on confining slowons via suitable effective potentials created by fast oscillatory elec-

008), doi:10.1016/j.ijms.2008.09.004

ric fields. As summarized recently in [5] this method has removedany difficulties encountered by standard electrostatic techniques.typical application is the so-called guided ion beam method, inhich a suitable electric radio frequency (rf) field is used to con-uct ions through a scattering cell, filled with the neutral reaction

dx.doi.org/10.1016/j.ijms.2008.09.004

http://www.sciencedirect.com/science/journal/13873806

mailto:[email protected]

dx.doi.org/10.1016/j.ijms.2008.09.004

ING ModelM

2 l of Ma

psctia

1

nrgisbcetdprbe[

atRndatbadmsrbS

1

wtrahihtitil

m

C

a

C

C

C

bt[hs

tad

C

C

b

ipnrw

2

2

bsrtctiamttabs

tsihtsfiiwtedA

ARTICLEASPEC-13860; No. of Pages 8

G. Borodi et al. / International Journa

artner. The results presented in this paper, also have been mea-ured with a beam-cell arrangement; however, the roles have beenhanged: the neutral reactants are formed into a beam while thearget are a few hundred ions, confined in an rf trap. In the follow-ng, the instrument will be described only briefly since the detailsre reported elsewhere [6].

.3. Reactions with H and D atoms

One of our motivations to combine an ion trap with a beam ofeutrals was to investigate low energy reactions between ions andadicals, especially hydrogen atoms. H-atom beams are used as tar-ets in many experiments in fundamental physics and chemistry,ncluding collisions with multiply charged ions, photoionizationtudies, or high resolution scattering experiments using crossedeams [7]. Collisions with H atoms play a central role in astro-hemistry, especially in the chemistry of the early universe [8]. Forxample, a fundamental process is the weakly endothermic elec-ron transfer from D to H+, an important step in the formation ofeuterated molecules in interstellar clouds. While this process isrobably well enough characterized by theory [9], more complexeactions such as the H–D exchange in H3

+ + D certainly needs toe studied in more detail theoretically and in a low temperaturexperiment [10,11]. Some first measurements have been reported4]; however not with internally cold ions as discussed in [5].

In ion chemistry, studies with atoms are rather scarce especiallyt energies of interest for interstellar chemistry. The older litera-ure dealing with this subject has been reviewed by Sablier andolando [12]. Most ion–H atom experiments utilize swarm tech-iques operated at room temperature. The combination of an ionrift tube and a hydrogen atom source has provided informationt higher energies [13], while a variable temperature SIFT appara-us extended the temperature range down to 120 K [14]. There haseen an attempt to combine an ion cyclotron resonance cell withhydrogen discharge source [15]; however, many complications

ue to H–H recombination, metastable atoms and UV photons haveade it extremely hard to get reliable quantitative results from this

pecial arrangement. The laboratory studies done since 1993 foreactions of positive and negative ions with H, N, or O atoms haveeen summarized very recently in a nice comprehensive review bynow and Bierbaum [16].

.4. CO2+ reacting with H or H2

One of the first ions the reactivity of which was investigatedith H atoms at thermal energies, was CO2

+ [17]. Using a driftube, these measurements have been extended later into the energyange between 60 and 140 meV by Tosi et al. [18]. For calibrating thetomic hydrogen density, these authors and later Scott et al. [19]ave made use of the special feature that CO2

+ produces predom-nantly HCO+ in collisions with H while reactions with molecularydrogen lead to HCO2

+. This was one of our motivations to selecthis well-characterized system for calibrating and testing our newnstrument. In addition, because of the importance of these reac-ions in connection with planetary and cometary atmospheres ornterstellar clouds, we have extended the previous results towardsower temperatures.

At low collision energies, reactions of CO2+ with hydrogen

olecules lead just to one product,

+ +


O2 + H2 → HCO2 + H, (1)

while for collision with H atoms, three channels are energeticallyllowed,

O2+ + H → HCO+ + O, (2)

oFnte

PRESSss Spectrometry xxx (2008) xxx–xxx

O2+ + H → HOC+ + O, (2a)

O2+ + H → H+ + CO2. (2b)

It must be mentioned that, so far, we have not yet distinguishedetween the two isomers formed via reactions (2) and (2a) althoughhe relevant technique is available in instruments using ion traps20]. In addition no products have been found from reaction (2b);owever, as discussed below, we did not make use of the full sen-itivity of the trap.

In contrast to the well-characterized reactions with H and H2,he analogue processes with deuterium have been studied only inn ICR experiment [15]. In the present work we have performedetailed measurements for

O2+ + D2 → DCO2

+ + D, (3)

O2+ + D → DCO+ + O. (4)

Also in reaction (4) no attempt has been made to distinguishetween isomers or to detect possible traces from charge transfer.

In the following we briefly describe the experimental setup,ncluding the hydrogen atom source and the various measuringrocedures. Some remarks concerning kinematic averaging in theeutral beam-stored ion cloud arrangement are made. Variousesults are summarized and discussed which have been obtainedith different beam arrangements.

. Experimental

.1. The AB-22PT

The experimental studies have been performed in the atomiceam 22-pole ion trapping apparatus (AB-22PT) which is shownchematically in Fig. 1. A first description and some preliminaryesults have been reported previously [21]. Since a detailed charac-erization of this new instrument is published in [6], together with aollection of typical applications, here only the most important fea-ures are summarized briefly. Concerning the basics of the rf basedon trapping technique we refer to the relevant literature [23,24]. Inddition, there are several more recent applications of cryogenic rfulti-electrode traps which just have been reviewed [5,25]. Impor-

ant for the present study is that the internal degrees of freedom ofhe CO2

+ ions can be cooled to temperatures between 15 and 300 Knd that the linear multipole is well-suited for running a moleculaream (or also a laser beam) through the trap without hitting anyurface.

As indicated in Fig. 1 the instrument combines a standard ionrapping machine [26,2] with a doubly differentially pumped beamource the temperature of which can be varied. Primary ions arenjected into the 22PT via an electrostatic quadrupole bender. Theyave been prepared in a well-established combination of a elec-ron impact storage ion source with a quadrupole mass selector (nothown in the figure, for details see [23]). In the trap, the ions are con-ned in a wide near field-free environment. The typical ion volume

s 1.5 cm3. Radial confinement is achieved by an effective potentialhich is created by applying two opposite phases of an rf generator

o the two sets of 11 electrodes. The electrostatic entrance and exitlectrodes are used to control the injection and extraction proce-ure. Voltages of less than 100 mV are sufficient to close the trap.nalysis and detection of the ion cloud is carried out with a sec-

008), doi:10.1016/j.ijms.2008.09.004

nd quadrupole mass spectrometer and a Daly type detector [27].or studying chemical reactions, it is sufficient to store a rather lowumber of ions, typically N0 = 1000 primary ions are injected eachime. Depending on conversion rates (relaxation, reaction, decay,tc.), they are trapped for time periods varying between ms and

dx.doi.org/10.1016/j.ijms.2008.09.004

ARTICLE IN PRESSG ModelMASPEC-13860; No. of Pages 8

G. Borodi et al. / International Journal of Mass Spectrometry xxx (2008) xxx–xxx 3

Fig. 1. Schematic diagram of the combination of an Atomic Beam (left part) with a 22-pole ion trap (AB-22PT apparatus) used for studying the interaction of a slow effusiveH-atom/H2 molecular beam with stored cold CO2

+ ions. A short description and first results have been published previously [21]. A detailed documentation of this versatileinstrument can be found elsewhere [6]. Before the primary ions enter the electrostatic quadrupole bender, they have been thermalized in a storage ion source and selectedin a first quadrupole mass spectrometer (not shown, for details see [22] and [25]). The H-atoms produced in an rf discharge, are first precooled to 100 K and then thermalizedto the temperature of the accommodator, TACC. For determining accurate relative rate coefficients of reactions with H and H2, a series of measurements has been performedw ) for br etic hed eld fob

ms

wetaegiwiidf

2

2ssothddc

whtIdtaa

dfthd

AtpTeeFtrfdcil[

thptypical time of flight distribution, recorded at TACC = 50 K, is shownin Fig. 2. Without magnets one obtains a smooth thermal distribu-tion indicated by the dotted line, whereas specific velocity groupsdominate if the two hexapoles are integrated into the machine. The

Fig. 2. Normalized time of flight distribution of H atoms measured at TACC = 50 K with

ithout the hexapole focusing magnets. This leads to straight trajectories (solid linesegion between the marked ion cloud and the neutrals (solid line) With the magnotted lines indicate schematically the imaging properties of the harmonic guiding fieam, allowing to account for reactions with background gas.

in. In general there is no loss of ions or creation of additional ones,o the total number of ions per filling remains constant,

∑Ni = N0.

The ion trapping region is surrounded by a copper box (see Fig. 1)hich is thermally connected to the cold head of a close cycle refrig-

rator system. Depending on the cryocooler head used, nominalemperatures as low as 3.6 K have been reached. The translationalnd internal degrees of freedom of the ions are coupled to the coldnvironment efficiently by inelastic collisions with helium bufferas. This either can be leaked into the inner box continuously ornjected as a short and intense gas pulse. The pulse is synchronized

ith the filling of the trap with ions. In general complete thermal-zation to the temperature of the surrounding walls can be achievedn a few ms. As emphasized in [5], the influence of the micro motionriven by the oscillating confining field – effects referred to as radiorequency heating – is negligible if the trap is operated properly.

.2. The atomic or molecular beam source

The component dominating the otherwise well-characterized2PT machine, is the doubly differential pumped neutral beamource. The overall geometry, defined by the accommodator, thekimmer, several apertures and the entrance and exit electrodef the trap, has been constructed in such a way that the neutralsraverse the linear multipole ion trap in axial direction withoutitting any of the surfaces surrounding the interaction region. Foristinguishing between the direct beam and atoms or moleculesiffusing indirectly into the trap, a shutter is placed in the chamberontaining the 22-pole trap (see Fig. 1).

The vacuum system consists of three differential pumping stagesith nominal effective pumping speeds of 2400, 350 and 300 l/s (for

ydrogen). Important details such as the conductance of the aper-ures in the walls separating the chambers, etc., are given in [6].n order to be able to perform measurements with target numberensities smaller than 108 cm−3 an excellent vacuum is manda-ory. Nonetheless, especially at elevated temperatures, we have toccount for parasitic reactions of stored ions with impurities suchs N2, CO2, H2O.

Special emphasis has been given to the development of a well-


efined, temperature variable hydrogen atom source, profitingrom the variety of relevant publications. In most of our applica-ions, we used a standard rf driven plasma source for dissociatingydrogen [22]. For slowing down the atoms emerging from the hotischarge to low velocities, cryogenic cooling has been used [28].

aofaaa

oth atoms and molecules, to conservation of flux, and to a well defined overlappingxapoles in place, atoms are guided and focused, depending on their velocity. Ther atoms with a kinetic energy of 8.2 meV. The shutter is used for blocking the direct

s indicated in Fig. 1, the hydrogen atoms pass first through a glassube surrounded by a precooler followed by a channel in a cop-er block the temperature of which can be set to values betweenACC = 10 and 300 K. The arrangement has been tested under a vari-ty of conditions, using either chemical probing in the ion trap or anlectron ionization source installed at the location of the 22PT (seeig. 1). There are many descriptions and specific hints in the litera-ure for achieving high dissociation degrees and for reducing H–Hecombination on the walls of the discharge tube and the tubes, seeor example [29,30]. We have found a procedure to obtain very highissociation degrees (more than 90% at 100 K) and to keep themonstant over many hours. The method which is based on coat-ng the walls of the precooler and the accommodator with specificayers of water, is described in more detail in the thesis of Borodi31].

The efficiency of the accommodator and the transmission fea-ures of the two hexapole magnets shown schematically in Fig. 1,ave been tested by measuring velocity distributions using a chop-er wheel and the already mentioned universal detector [6]. A

008), doi:10.1016/j.ijms.2008.09.004

n ionizer at the location of the 22-pole. Without the hexapole magnets (see Fig. 1)ne obtains the thermal distribution indicated by the dotted line. The transmissioneatures of the two hexapoles favor mainly two groups the kinetic energies of whichre indicated by the two Gaussians. As illustrated by the dashed trajectory in Fig. 1nd discussed in detail in [6], the maximal transmission of H atoms is obtained forkinetic energy of 8.2 meV.

dx.doi.org/10.1016/j.ijms.2008.09.004

ING ModelM

4 l of Ma

igtstWaswaf

2

e2sgokttetteftlFmraipv

2

cbaacr

t(strhroodasw

T

acMtp

T

mstbndcrwtic

[Itiiwtf[E

3

itftstaawcnHmc(dt

TPow



nteraction of their fields with the magnetic moment of the hydro-en atoms leads to a harmonic guiding potential and, therefore,o focusing and imaging properties. In the present case the fieldtrength of the two different magnets and the geometry favor theransmission of atoms, having a kinetic energy of (8.2 ± 3.3) meV.

ith such a speed, the oscillation in transverse direction leads tobout 1/2 sinus wave in the first magnet and to 1/4 wave in theecond one, which is weaker and shorter. As illustrated in Fig. 1ith the dashed trajectories the result is a nearly parallel beam of

toms passing through the trap. More details and modifications foravoring slower atoms are discussed in [6].

.3. Measuring procedure

For studying the title reaction, CO2+ ions have been produced by

lectron bombardment of carbon dioxide gas (Messer-Griesheim.7 purity) in a storage ion source [23]. In such a source, the longtorage time and the high number density of the neutral precursoras (>10−5 mbar) already leads to a significant pre-thermalizationf the ions. Furthermore, the initial excitation of the ions has beenept rather low by setting the electron energy (∼15 eV) just abovehe ionization energy of carbon dioxide (13.777 eV). After extrac-ion of primary ions from the storage source using the pulsed exitlectrode, the ions are mass selected in a quadrupole, deflected inhe static quadrupole by 90◦ (included in Fig. 1) and injected intohe trap. There they are cooled to the ambient temperature andxposed to the neutral reactant gas for a period which is variedrom milliseconds to seconds or minutes. After this storage time,he trap content is extracted by opening the exit electrode, ana-yzed in a second quadrupole mass filter and the ions are counted.or determining reaction rate coefficients, the sequence ion for-ation, injection, relaxation, reaction, extraction, and analysis is

epeated many times for all ions of interest. Several parametersre changed the most important being the reaction time. Othersnclude the number density of the reactant or coolant gas, the tem-erature of the trap, and, especially in the results reported here,arious parameters of the neutral beam.

.4. Averaging over velocity and spatial distributions

In order to understand the potential and the limitations of theombination of a thermal ion cloud interacting with an effusiveeam, one needs some basic knowledge in kinematic and spatialveraging. Some relevant hints can be found in [23]. Other specialspects, e.g., how to reach very low relative velocities or how toool molecular ions to very low temperatures, have been discussedecently in [5] and [25].

So far, in most low temperature trapping experiments, the neu-ral reactants have been leaked into the box surrounding the trapsee Fig. 1), usually also together with the cooling buffer gas. Underuch conditions, one can reach a thermal equilibrium at a commonemperature T22PT given by the surrounding walls. The measuredesults are thermal rate coefficients k(T22PT). If there is some ioneating, e.g., via a parasitic low frequency components from thef generator or other electric noise, the translational distributionf the trapped ions is no longer thermal. In such a situation it isften sufficient to approximate the motion of the ions which may beetermined, for example, with laser methods, by a Maxwellian withmean temperature, Tion. As can be shown analytically [32], under


uch conditions one measures a thermal rate coefficients k(Tcoll)ith a collision temperature

coll = (mtarget Tion + mion Ttarget)(mion + mtarget)

. (5)

snti3


Another situation is a monoenergetic ion beam passing throughscattering cell containing a thermal target gas. This spe-

ial case can be treated analytically, resulting in a generalizedaxwell–Boltzmann distribution [33,23]. At low beam velocities,

his function approaches a normal Maxwellian with a reduced tem-erature

coll ={

mion

mion + mtarget

}Ttarget. (6)

From a kinematic point of view, we have in the present experi-ent, combining a cold effusive or supersonic neutral beam with a

tored ion cloud, the same situation. In reality, however, inspec-ion of Fig. 2 reveals that the assumption of a monoenergeticeam cannot be made and a detailed analysis requires to integrateumerically (or analytically) the generalized Maxwell–Boltzmannistribution over the velocity distribution of the beam. For the rateoefficients presented in this paper, the resulting corrections areather small since the elementary rate coefficients depend onlyeekly or not at all on the relative velocity. In contrast, a detailed

reatment is required for the CH5+ + H collision system as discussed

n [34]. In the following we just report the measured rate coeffi-ients as k(T22PT, TACC).

Another important aspect which is discussed in more detail in6] is the averaging over the spatial distributions of the reactants.n a first approximation one can assume that the densities of neu-rals and ions are homogeneous in the region of overlap which isndicated schematically in Fig. 1. The neutral beam diameter is typ-cally 5 mm while, in the 22PT, the ions usually explore a cylinder

ith a diameter of 7–8 mm. For example, in the case of CO2+ ions,

he 22PT has been operated with an rf amplitude V0 = 50 V and arequency f = 16 MHz. Using the standard definitions and formulas23], one obtains a diameter of 2rm = 7.2 mm for a transverse energym = 1 meV (8.1 mm for Em = 10 meV).

. Results

The AB-22PT apparatus has been used in various ways for study-ng the interaction of hydrogen atoms and molecules with coldrapped CO2

+ ions. For obtaining reliable thermal rate coefficientsor reactions (1) and (3), molecular hydrogen was leaked into therap directly with number densities up to 1011 cm−3. The so mea-ured thermal rate coefficients have been used to determine in therap the effective number density of the effusive molecular beamnd together with the shutter, the molecular background. Withtomic hydrogen, a series of measurements has been performedith the hexapole magnets taken out. This allows us to assume

onservation of hydrogen flux, i.e., the number of H-atoms/s doesot change if the discharge is switched ON. For this analysis, eachis counted independent on whether it is single or bound in aolecule. Other parameters which have been selected in various

ombinations are the two temperatures TACC (12–300 K) and T22PT15–300 K) and the position of the shutter which can block theirect beam. The number density of He buffer gas was always seto a few 1013 cm−3.

For reaction (1), a typical time dependence is shown in Fig. 3 at22PT = 15 K using the gas inlet leaking the H2 directly into the trap.lotted is the number of primary and product ions, Ni, as a functionf storage time. As already discussed above, the trap is usually filledith some thousand primary ions. For better statistics, the mea-

008), doi:10.1016/j.ijms.2008.09.004

urements are averaged over many iterations; however, the Ni(t) areormalized to one iteration (ions per filling). Inspection shows thathe number density of 1.4 × 1010 cm−3 is high enough for convert-ng half of the primary CO2

+ ions into HCO2+ products already after

0 ms. For evaluating such experimental data, the time dependence

dx.doi.org/10.1016/j.ijms.2008.09.004

ARTICLE IN PRESSG ModelMASPEC-13860; No. of Pages 8

G. Borodi et al. / International Journal of Mass Spectrometry xxx (2008) xxx–xxx 5

Table 1Measured reaction rate coefficients and comparison with published values.

Reaction ka ˛b ˇb T (K)c TACC (K) Remarks

CO2+ + H2 → HCO2

+ + H 15.3 Langevin9.5 −0.15 15–300 this work

8.7 300 [19] SIFT9.0 60–140 meV [18] DT(5.8)d Non-thermal [15] ICR

CO2+ + D2 → DCO2

+ + D 10.9 Langevin4.9 −0.30 15–300 This work

(4.1) Non-thermal [15] ICR

CO2+ + H → HCO+ + O 19.4 Langevin

4.7 125 300 This worke

4.5 300 55 This work4.7 300 [19] SIFT2.9 60–140 meV [18] DT(1.1) Non-thermal [15] ICR6 300 [17] FA

CO2+ + D → DCO+ + O 13.8 Langevin

2.2 55–300 55 This work(0.84) Non-thermal [15] ICR

a The unit of k is 10−10 cm3 s−1. The errors of the absolute values for reactions with molecular target, k1 and k3, are estimated to be 20%, those with atomic reactants, k2 andk4, are 40%. Relative errors for the temperature dependence are less than 15%.

b The temperature dependence is approximated with k(T) = ˛(T22PT/300 K)ˇ , the unit of ˛ is 10−10 cm3 s−1. These measurements were performed with hydrogen, directlyl

he dritails s

angem

octotci∼iew

Faptt4o(oi

Ds

ossr

eaked into the 22-pole trap (see Fig. 1).c T is either T22PT or the temperature of the respective instrument. In the case of td The number in brackets are ICR data [15] which should be take with care. For dee This measurement was performed using a single differential pumping stage arr

f the number of stored ions, Ni, is simulated by solving numeri-ally a system of differential equation describing the changes in therap. Here the situation is very simple since only two channels aref importance. No other masses have been found within 3 decades,wo of which are shown in the figure. The resulting rate coeffi-ient is given in Table 1. A similar result for deuterium is shownn Fig. 4. Note that the number density, [D ] = 1.1 × 1011 cm−3, is


28 times higher leading to a faster conversion of the primary

ons. Very important hints from such measurements are the mono-xponential decay of the primary ions over 2 orders of magnitudeithout any indication of a curvature and the convergence of the

ig. 3. Reactions of CO2+ ions with H2 molecules, leaked directly into the trap at

number density of [H2] = 1.4 × 1010 cm−3. Plotted is the number of primary androduct ions per filling, Ni , as a function of the storage time t. At each iteration,ypically 700 CO2

+ ions (full squares) are filled into the trap. For cooling the ionso T22PT = 15 K within a few ms, He gas has been added with a number density of.8 × 1013 cm−3. In the present example, data are evaluated after 10 ms. The numberf the primary ions decreases exponentially. As can be seen from the simulationsolid lines) and the constant sum, �, the HCO2

+ products grow at the same rate. Nother ions have been seen within the plotted range. Rate coefficients are summarizedn Table 1.

pOssaram

FsmTr

ft tube, the mean kinetic energy is given.ee text.ent.

CO2+ products towards the constant sum. This allows the conclu-

ion that, within a few % accuracy, no other ions are involved.The situation is quite different if the neutral beam is used instead

f leaking the gas directly into the trap. For the measurementshown in Fig. 5 both the accommodator and the 22PT have beenet to the same temperature, 55 K. The hexapole magnets have beenemoved from the instrument in order to get the same transmissionroperties for hydrogen atoms and molecules. On the left (dischargeFF), Fig. 5 shows results for reactions of CO2

+ ions with D2. Onlyome ten DCO2

+ products are formed per filling in the time windowhown. Closing the shutter (open symbols) blocks the beam and

008), doi:10.1016/j.ijms.2008.09.004

llows to determine the density of the D2 background gas using theate coefficients which has been determined at the same temper-ture of the walls of the trap, T22PT = 55 K (see Table 1). From sucheasurements we obtain a mean number density of the molecu-

ig. 4. Same as Fig. 3, here for deuterium. The significant higher number den-ity ([D2] = 1.1 × 1011 cm−3, T22PT = 15 K, [He] = 4 × 1013 cm−3) allows us to follow theono-exponentially decay over two orders of magnitude (time constant � = 7.5 ms).

he solid lines are solutions of a simple rate equation system accounting just foreaction (3). The resulting rate coefficient is included in Table 1.

dx.doi.org/10.1016/j.ijms.2008.09.004

ARTICLE ING ModelMASPEC-13860; No. of Pages 8

6 G. Borodi et al. / International Journal of Ma

Fig. 5. Experimental results obtained without hexapole magnets. The reactions ofCO2

+ ions with the pure D2 beam (discharge off, left panel) and with a D/D2 mixture(discharge on, right panel) have been recorded at TACC = 55 K and T22PT = 55 K. Dueto the lower target density (some 108 cm−3) the number of product ions is muchssoc

las

cFuimvtatditf

[

[a

aFadvrtceutuhtnrot

fletvtmstuod

imttOooloictatatio

badalfaster with H2 than with H, due to the larger rate coefficient. Wealso have performed measurements with the discharge tube placedat a much shorter distance of 8 cm to the trap by removing the dif-ferential pumping stage with the hexapole magnet. The numberdensity achieved in this way was rather large, [H] = 4.5 × 109 cm−3;

Fig. 6. Experimental results obtained with the hexapole magnets installed (see+ 9 −3

maller in comparison to Figs. 3 and 4. For subtracting the background, the shutterhown in Fig. 1 is opened (solid symbols) and closed (open symbols). The evaluationf data sets which have been taken without the magnets, is simplified due to fluxonservation. The resulting rate coefficients are summarized in Table 1.

ar beam in the trap, [D2]OFF = 1.3 × 108 cm−3. As already mentionedbove this mean value accounts for the fact, that the trapped ionspend about 60% of their time outside of the beam (see Fig. 1).

Switching the discharge ON (right panel) the deuterium fluxonsists of a D/D2 mixture. As can be seen on the right side ofig. 5, the fragmentation of D2 leads to a decrease of DCO2

+ prod-cts while the D atoms formed by dissociation produce now DCO+

ons via reaction (4). Note that there is some minor background onass 30, most probably caused by some background gas. Conser-

ation of mass flux through the leak valve, the accommodator, andhe trap means that one obtains for each molecule in the beam twotoms. In order to convert the flux into density in the trap one haso account for the different velocities of D and D2. This requiresivision by

√2, assuming that the accommodator fully thermal-

zes atoms and molecules to the same temperature. Accounting forhese two factors, the mean number density of D atoms is obtainedrom

D] =√

2 ([D2]OFF − [D2]ON) . (7)

Typical atom number densities measured in this way areD] = 108 cm−3 corresponding to an effective dissociation degree ofbout 60% in the 22PT.

Such measurements have been performed at different trap andccommodator temperatures, for both hydrogen and deuterium.rom a large set of data, absolute rate coefficients for reactions (2)nd (4) have been determined. No significant temperature depen-ence has been observed between 50 and 300 K. For D, an averagealue of k4 = 2.2 × 10−10 cm3 s−1 is included in Table 1 while foreaction (2) we report two values, k2 = 4.5 and 4.7 × 10−10 cm3 s−1

o emphasize that these values have been measured at differentombinations of TACC and T22PT. For all subsequent measurements,specially with the hexapole magnets installed, these values aresed as calibration standards. The errors of k2 and k4 are estimatedo be 40% since the uncertainties of four rates enter into the eval-ation. Another uncertainty is the charge transfer channel (2b);owever, Scott et al. [19] reported that its contribution is smaller


han 5% of the total rate coefficient. In our measurements, we haveot yet searched for small amounts of ions on mass 1 or 2 whichequires some effort with the quadrupole mass spectrometer. Theverall charge balance of the detected ions excludes, that the chargeransfer rate coefficient k2b exceeds 20% of k2.

Fatci[


Installing the two hexapole magnets leads to an increase of theux of atoms. Inspection of time of flight distributions, see forxample Fig. 2, allows us to estimate that, for the given geome-ry, an amplification factor of up to 10 can be achieved for specificelocity groups. Total averaged effective number densities of upo 3.5 × 109 cm−3 have been achieved. One disadvantage of the

agnets is that also the molecular hydrogen background increasesince the efficiency of differential pumping is reduced by the longube like structures. Moreover, one has to account for the molec-lar background which is created by atom–atom recombinationn walls. As a consequence flux conservation cannot be used foretermining the effective number density of the atoms in the trap.

The situation is illustrated with a surprising experimental resultn Fig. 6. The overall higher flux caused by guiding the atoms in the

agnetic field, leads to more reactions and, already within 100 ms,o an obvious decay of the primary ions. Remarkable is the fact thathe decay rate increases by a factor of 3 if the discharge is switchedN (�ON = 0.33 s, �OFF = 1.05 s). This is partly due to the large amountf H atoms being transferred to the trap; however, also the densityf molecular hydrogen in the trap increases instead of becomingess due to dissociation. This can be seen from the larger numberf HCO2

+ products formed by reaction (1) with discharge ON. Thisncrease of the H2 number density in the trap is most probablyaused by the fact that the loss of molecules due to fragmenta-ion is counter-compensated by recombination. Apparently a lot oftoms collide with the surfaces surrounding the beam path. Withhe geometry of the beam, emerging from the magnets, and thepertures, also wall collisions in the trap cannot be ruled out. Due tohe low conductance, hydrogen molecules formed inside the cool-ng shield (see Fig. 1) lead to a significant increase of the backgroundf molecules.

Cryopumping can improve significantly the situation. So far, theest H/H2 ratio in the 22PT has been achieved by operating the trapt 10 K and the accommodator at 12 K. At these conditions, an atomensity of [H] = 2.1 × 108 cm−3 has been obtained in the trap with3 times lower molecular density, [H2] = 7 × 107 cm−3. Nonethe-

ess, even under such conditions, the primary CO2+ ions still react

008), doi:10.1016/j.ijms.2008.09.004

ig. 1). The CO2 ions react with an intense beam of D-atoms ([D] = 3.4 × 10 cm )t TACC = 100 K and T22PT = 100 K. A surprising observation which is discussed inhe text, is that the decline of primary ions is three times slower with the dis-harge switched OFF. Remarkable is also that the total number density of moleculesn the trap is higher with discharge ON, [D2]ON = 3.3 × 109 cm−3 than with OFF,D2]OFF = 1.4 × 109 cm−3.

dx.doi.org/10.1016/j.ijms.2008.09.004

ING ModelM

l of Ma

hb

4

mtsubtiof

rmsHH[ccatItUtar

hktaTcas

FDeatfi

k2fitpHsorsgtwbsTirtfTbc

cvpwtbmsWH[



owever, at the expenses of an almost 4 times larger molecularackground, [H2] = 1.7 × 1010 cm−3.

. Discussions

For the first time, reactions between CO2+ ions and hydrogen in

olecular and atomic form have been measured down to the lowemperatures prevailing in dense interstellar clouds. The results areummarized in Fig. 7 and in Table 1. The errors of the absolute val-es for reactions with molecular target, k1 and k3, are estimated toe 20%, those with atomic reactants, k2 and k4, are 40%. Althoughhe overall arrangement radical beam-trap is well-characterized,t cannot be completely excluded that some problems have beenverlooked, especially caused by collisions of radicals with sur-aces.

At room temperature the obtained rate coefficients are ineasonable overall agreement with results from previous measure-ents performed with different flow systems [17–19,35]. This is

urprising due to the overall difficulties in determining the actualatom concentrations in the various instruments. Reactions withand D atoms also have been investigated with the ICR technique

15]. However, the results from this instrument deviate signifi-antly. Not only the overall reactivity is off by a factor ∼5 in thease of H atom reactions and ∼3 in the case of D atom reactions butlso the branching ratio is completely different, favoring the chargeransfer channel (2b). Inspection of the schematic diagram of theCR based machine (Fig. 1 in Ref. [15]) allows one to suspect thathe erroneous measurements may be caused by metastable atoms,V photons or secondary reactions involving surfaces. For example,

he assumption must be questioned that all H atoms which strikesurface in the ICR cell, are lost and re-enter the gas phase as a

ecombined molecules.For the molecular target, a significant temperature dependence

as been observed. The data have been fitted using the function= ˛(T/300 K)ˇ, a simple parameterization commonly used in reac-

ion networks describing interstellar chemistry [36]. The fact that,


t low temperatures, one reaches finally the Langevin limit (seeable 1 and Fig. 7) may be taken as an indication that, in the collisionomplex, time is required for dissociating the hydrogen moleculend for integrating the atom into the protonated carbon dioxide. Atatistical calculation may corroborate this hypothesis.

ig. 7. Temperature dependence of rate coefficients for reactions of CO2+ with H, H2,

, and D2 (reactions (1)–(4)). The indicated overall error of the rate coefficients isxplained in the text. For the reaction with hydrogen atoms, no change with temper-ture could be seen. With molecular reactants, the reactivity increases with fallingemperature, reaching finally the Langevin limit kL. The rate coefficients and thetting function are given in Table 1.

5

tbstttowcer

iCoiassstiwt

PRESSss Spectrometry xxx (2008) xxx–xxx 7

For CO2+ + H the Langevin rate coefficient is

L = 1.9 × 10−9 cm3s−1 while reaction (2) occurs with less than5% of the collision rate, also at low temperatures. The rate coef-cient with deuterium atoms, reaction (4), is even smaller. Fromhose collisions, which are attracted according to the Langevinrescription, only 16% react. For a system like CO2

+ colliding withdetailed dynamical calculations on precise potential energy

urfaces are feasible today. Without such information, one cannly speculate. It has been stated in [19] that there seems to be aather general rule that ion–H atom reactions have rate coefficientsubstantially less than the Langevin capture rate. One explanationiven by Scott et al. is the influence of spin statistics. The two reac-ants which are doublets, interact attractively on a singlet surfacehile triplet surfaces may be repulsive. Other restrictions may be

ased on conservation of spin: reaction (2) requires the decay of ainglet HCO2

+ intermediate into a singlet HCO+ and triplet O atom.his is spin-forbidden. Another simple guess for H–AB + collisionsn general may be that there are some steric restrictions, i.e., aeaction is rather direct and occurs only if the atom approacheshe molecule from the right direction. This argument also holdsor a wide temperature range due to the atom–ion mass ratio.he significant difference in the rate coefficients for H and D maye due to differences in the mobility of the two atoms or may beaused by tunneling contributions.

Another unexplained experimental observation is that theharge transfer channel (2b) is negligibly small as concluded fromarious authors and also from the charge balance in our trap-ing experiment. However, instead of entering speculations, weant to conclude the discussions by stating that a much bet-

er k2/k2b ratio can be determined with the AB-22PT instrumenty optimizing the system for low masses. Related measure-ents have been delayed until several improvements of the

etup are made, especially the integration of cryogenic pumping.ith the improved sensitivity, also the two isomeric structures

OC+ and HCO+ can be distinguished using chemical probing20].

. Conclusions

Our studies of the title reaction have been motivated initially byhe need to use reactions (2) and (4) for determination of the num-er density of hydrogen atoms directly in the trap. Such calibrationtandards are necessary for a beam-trap arrangement since, in con-rast to the usual operation where the 22PT housing is filled witharget gas with in a well-defined velocity and density distribution,he diameter of the skimmed beam is smaller than the diameterf the ion cloud in order to avoid (or reduce) collisions with thealls. For this purpose it is advantageous that the relevant rate

oefficients have been found to be independent on the collisionnergy, which is determined by the motion of the H atoms and theotational population of the ions.

There is a variety of other measurements which have been stud-ed with the AB-22PT machine, including CH+ + H, CH4

+ + H andH5

+ + H [21]. Detailed publications are in preparation. A wide fieldf interesting research is the deuteration of hydrogenated ionsn collisions with D-atoms. A system of basic interest mentionedlready in the introduction is the H–D exchange in H3

+ + D colli-ions or other isotopic analogues. Most probably it is rather slow

008), doi:10.1016/j.ijms.2008.09.004

ince it is hindered by small barriers [5]. The early merged beamtudies [4] almost reached low enough collision energies; however,hose experiments have been performed with internally hot H3

+

ons. The combination of the low temperature multipole ion trapith a beam of very slow hydrogen atoms will reveal the details of

he H4+ potential energy surface and the low energy dynamics.

dx.doi.org/10.1016/j.ijms.2008.09.004

ING ModelM

8 l of Ma

A

htasFa

R

[[

[[

[[[[[

[

[

[

[[[

[

[[[[[

[[



cknowledgments

This work is dedicated to Zdenek Herman on the occasion ofis 75th birthday and in recognition of his outstanding contribu-ions to the field of ion–molecule reactions. We thank M. Smithnd S. Schlemmer for many contributions to this work. Financialupport of the Deutsche Forschungsgemeinschaft (DFG) via theorschergruppe FOR 388 “Laboratory Astrophysics” is gratefullycknowledged.

eferences

[1] S.Y.T. van de Meerakker, N. Vanhaecke, G. Meijer, Ann. Rev. Phys. Chem. 57(2006) 159.

[2] D. Gerlich, M. Smith, Phys. Scripta 73 (2006) C25.[3] Z. Herman, Int. J. Mass. Spectrom. 212 (2001) 413.[4] W.R. Gentry, in: M.T. Bowers (Ed.), Gas Phase Ion Chemistry, vol. 2, Academic

Press, New York, 1979, p. 221.[5] D. Gerlich, in: W. Ian, M. Smith (Eds.), Low temperatures and cold molecules,

World Scientific Publishing, 2008, ISBN 978-1-84816-209-9, p. 121.[6] G. Borodi, A. Luca, C. Mogo, M. Smith, D. Gerlich, Rev. Sci. Instr., submitted for

publication.[7] J.P. Toennies, W. Welz, G. Wolf, J. Chem. Phys. 71 (1979) 614.[8] D. Puy, M. Signore, New Astron. Rev. 51 (2007) 411.[9] D.W. Savin, Ap. J. 566 (2002) 599.


10] G.E. Moyano, D. Pearson, M.A. Collins, J. Chem. Phys. 121 (2004) 12396.11] D. Gerlich, F. Windisch, P. Hlavenka, R. Plasil, J. Glosik, Phil. Trans. R. Soc. Lond.

A 364 (2006) 3007.12] M. Sablier, C. Rolando, Mass Spectrom. Rev. 12 (1993) 285.13] W. Federer, H. Villinger, F. Howorka, W. Lindinger, P. Tosi, D. Bassi, E. Ferguson,

Phys. Rev. Lett. 52 (1984) 2084.

[[[

[


14] N.G. Adams, D. Smith, Ap. J. 294 (1985) L63.15] Z. Karpas, V. Anicich, W.T. Huntress Jr., J. Chem. Phys. 70 (1979) 2877.16] T.P. Snow, V.M. Bierbaum, Ann. Rev. Anal. Chem. 1 (2008) 229.17] F.C. Fehsenfeld, E.E. Ferguson, J. Geophys. Res. 76 (1971) 8453.18] P. Tosi, S. Iannotta, D. Bassi, H. Villinger, W. Dobler, W. Lindinger, J. Chem. Phys.

80 (1984) 1905.19] G.B.I. Scott, D.A. Fairley, C.G. Freeman, M.J. McEwan, P. Spanel, D. Smith, J. Chem.

Phys. 106 (1997) 3982.20] M.A. Smith, S. Schlemmer, J. von Richthofen, D. Gerlich, ApJL 578 (2002)

87.21] A. Luca, G. Borodi, D. Gerlich, in: F.D. Colavecchia, P.D. Fainstein, J. Fiol, M.A.P.

Lima, J.E. Miraglia, E.C. Montenegro, R.D. Rivarola (Eds.), Progress report in XXIVICPEAC 2005, Rosario, Argentina, July 20–26, 2005, p. 20.

22] J. Slevin, W. Stirling, Rev. Sci. Instr. 52 (1981) 1780.23] D. Gerlich, Adv. Chem. Phys. LXXXII (1992) 1.24] D. Gerlich, in: P.B. Armentrout (Ed.), Low Temperatures and Cold Molecules,

Elsevier Ltd, 2003, ISBN 978-1-84816-209-9, p. 182.25] D. Gerlich, in: I.W.M. Smith (Ed.), Low Temperatures and Cold Molecules, World

Scientific Publishing, 2008, ISBN 978-1-84816-209-9, p. 295.26] D. Gerlich, Phys. Scripta T59 (1995) 256.27] N.R. Daly, Rev. Sci. Instr. 31 (1960) 264.28] J.T.M. Walraven, I.F. Silvera, Rev. Sci. Instr. 53 (1982) 1167.29] N. Koch, E. Steffens, Rev. Sci. Instr. 70 (1999) 1.30] D. Szczerba, L.D. van Buuren, J.F.J. van den Brand, H.J. Bulten, M. Ferro-Luzzi,

S. Klous, H. Kolster, J. Lang, F. Mul, H.R. Poolman, M.C. Simani, Nuclear Instr.Methods Phys. Res. A 455 (2000) 769.

31] G. Borodi, Ph.D. Thesis TU Chemnitz, 2008.32] D. Gerlich, J. Chem. Phys. 90 (1989) 127.

008), doi:10.1016/j.ijms.2008.09.004

33] P.J. Chantry, J. Chem. Phys. 55 (1971) 2746.34] G. Borodi, A. Luca, C. Mogo, D. Gerlich, in preparation.35] W. Federer, W. Dobler, H. Ramler, W. Lindinger, 3rd International Swarm Sem-

inar, Innsbruck, Austria, 1983, p. 191.36] Y.H. Le Teuff, T.J. Millar, A.J. Markwick, Astron. Astroph. Suppl. Ser. 146 (2000)

157.

dx.doi.org/10.1016/j.ijms.2008.09.004

119

Appendix C

Interaction of ions with hydrogen atoms

A. Luca, G. Borodi, D. Gerlich

Faculty of Natural Science, University of Technology, 09107 Chemnitz, Germany

1

INTERACTIONS OF IONS WITH HYDROGEN ATOMS

ALFONZ LUCA, GHEORGHE BORODI, DIETER GERLICH Institut für Physik, Technische Universität Chemnitz,

Chemnitz, 09107, Germany

This progress report presents recent advances in developing a versatile technique for in-vestigation of collisions of ions with open shell neutral intermediates. Combination of a 22-pole ion trap with a beam of H atoms allows accurate determination of rate coeffi-cients at temperatures between 10 K and 300 K. New experimental results on hydrogen abstraction in collisions of CH+, CH4

+ and CH5+ ions with H atoms are reported at tem-

peratures between 10 K and 100 K. In the case of CH+ and CH4+, large rate coefficients of

1.3 × 10-9 cm3s-1 and 6.0 × 10-10 cm3s-1 have been obtained at 50 K. CH+ reacts with D at-oms with a total rate coefficient of 2.4 × 10-9 cm3s-1 the branching ratio being 50 % for hydrogen abstraction and 50 % for atom exchange. For collisions of CH5

+ with H atoms rate coefficients of 9 × 10-12, 1.3 × 10-11, and 2.3 × 10-11 cm3s-1 have been determined at trap and nozzle temperatures of 10, 50, and 100 K, respectively. This indicates that this reaction is almost thermoneutral in contrast to thermodynamical data reported in the lit-erature.

1. Introduction

Hydrogen, in the atomic or molecular form, is the most abundant species in interstellar clouds and therefore its interaction with other molecules, molecular ions and particle surfaces has to be carefully taken in account in order to under-stand chemical evolutions and explain observed abundances. The lack of spe-cific information on its reaction dynamics lets plenty of related questions unan-swered. For example it is still not yet known how the hydrogen molecule itself or more complex molecules like methanol are formed in space. Do processes in the gas phase or on particle surfaces play the dominant role? In many cases valuable information can be derived from observed abundances of molecules and structures of deuterated species. For this laboratory experiments are asked for to be performed at conditions relevant for interstellar space.

There are plenty of experimental results on reactions with molecular hydro-gen. Unfortunately, radicals such as hydrogen atoms are not as simple to handle and, therefore, only few experiments with H or D atoms have been performed in the gas phase. A general review of ion-atom reactions has been published.1 Except early ICR studies2 most experiments are based on the flow tube tech-

Progress report in XXIV ICPEAC 2005, Rosario, Argentina, July 20-26, 2005, Edited by F.D.Colavecchia, P.D.Fainstein, J.Fiol, M.A.P.Lima, J.E.Miraglia, E.C.Montenegro, and R.D.Rivarola

2

nique. Using the selected ion flow drift tube approach (SIFDT) reactions can be studied at superthermal energies.3 At temperatures down to 120 K ion – hy-drogen atom reactions have been investigated by a variable temperature modifi-cation of the selected ion flow tube.4 However, no general low temperature reaction studies of ions with H atoms have been reported below 100 K.

In this progress report a general methodology is presented which combines a wide field free trapping technique for confining and thermalizing mass se-lected ions and an atomic hydrogen beam as target. It allows the investigation of reactions of H atoms with ions at temperatures lower than 10 K and can be ex-tended to other radical atoms and condensable neutrals.5,6 Results are reported for three benchmark ion-atom reaction systems.

2. Experimental

Fig. 1 shows the novel experimental setup, the Atomic Beam 22-Pole Trap Ap-paratus (AB-22PT) that has been developed to study H-atom reactions with molecular ions. It is a combination of an effusive source of H / D atoms with a standard trapping apparatus consisting of an ion preparation unit, the central 22PT and an ion detection system. A thorough description of the used rf devices can be found in Refs. 7 and 8. Discharge tube

Accommodator12-300 K

HexapoleMagnet 22PT

CH+

10 - 300 K

H / H2

Mass Filter & Detection

Figure 1. Schematic diagram of the Atomic Beam 22-Pole Trap Apparatus (AB-22PT) combined with a low temperature H atom source.

Primary ions are prepared in a standard storage ion source. CO2+ ions are

generated directly by electron bombardment from carbon dioxide, both CH+ and CH4

+ from methane, while CH5+ is produced via the subsequent reaction CH4

+ + CH4 → CH5

+ + CH3. The ions are extracted from the storage ion source, mass filtered in an rf quadrupole, deflected by 90° in a electrostatic quadrupole bender, and injected into the 22PT. There they are stored for time between ms and some seconds. The translational and internal energy distributions of injected

3

ions accommodate to the cold 22PT environment, Tion = 10 ÷ 300 K, via radia-tion and via collisions with buffer gas. In order to accelerate the thermalization process, Helium is introduced in a pulsed mode during injection of the ions. For ~10 ms, densities of some 1015 cm-3 are achieved. In addition, Helium is let into the 22PT continuously. Typically the He densities, being between 1012 and 1013 cm-3, are several orders of magnitudes higher than the density of particles from the beam source. After a certain reaction time, the ions are extracted from the 22PT, mass analyzed, and detected using single ion counting technique.

For dissociation of molecular hydrogen resp. deuterium a standard rf driven plasma source9 is used. The generated hydrogen atoms pass through an tempera-ture variable accommodator with 1.2 mm inner diameter and 22 mm length resulting in a translational temperature TH. The effusive beam is skimmed and differentially pumped twice. Two hexapole magnets are used for guiding the H / D atoms (weak field seeking). The number density of H and H2 and deuterated analogues in the interaction region has been determined using a calibrated uni-versal detector based on ionization via electron bombardment at the 22PT posi-tion. Comparison of the densities measured with and without hexapole magnets reveals that the magnetic guiding field increases the density by more than a factor 10 at beam temperatures between TH = 35 K and 90 K. The maximal en-hancement, a factor of 25, is reached at 60 K.

Real in situ determination of atomic and molecular densities is achieved via the reaction of CO2

+ with H / D and H2 / D2. At 300 K the rate coefficients for these reactions are known from previous studies. At low temperatures they have been measured carefully with the present AB-22PT. Density determination has been performed before and after each set of measurements with primary hydro-carbon ions. Therefore long time drifts of the discharge source can be excluded or corrected for. At TH = Tion = 100 K, effective H and H2 densities of 109 cm-3 have been obtained typically. At temperatures around 22 K the H - H recombi-nation on the accommodator surface is high and atom density is rather low, 107 cm-3. At the lowest temperature, TH = 12 K, hydrogen condensate on the surface and the atom density increases to 2 × 108 cm-3. If the 22PT is at lowest temperature of 10 K the H2 background, 5 × 107 cm-3, is rather low due to con-densation. Using the described configuration of the AB-22PT, rate coefficients lower than 10-13 cm3s-1 can be determined. In order to further reduce the back-ground of molecules, beam catchers based on effective cryo-pumping can be introduced.

4

3. Results and Discussion

3.1. CH+ + H, CH+ + D

From a fundamental point of view, the C+ + H2 collision system is one of the model systems for experimental and theoretical studies of the kinetics, dy-namics, and energy requirements of endothermic ion-molecular reactions. Con-sequently, it has been the subject of numerous experimental and theoretical studies, see e.g. Ref. 10 and references therein. The reverse reaction

CH+ + H → C+ + H2 + (0.398 ± 0.003) eV (1)

represents an important destruction mechanism of CH+ the formation of which is poorly understood in diffuse interstellar molecular clouds. Simple models underestimate the observed abundances and, therefore, it is assumed that shock waves, turbulences or UV radiation must play a role. Until 1984 it was accepted that, at 100 K, the exothermic reaction (1) is slow, k = 2 × 10-12 cm3s-1 and even slower at lower temperatures whereas phase space theory predicts that the rate coefficient can approach the Langevin limit of 2 × 10-9 cm3s-1 at low temperatures.11 This was confirmed by experimental results obtained with the SIFDT apparatus.3, 12 The observed large rate coefficients and a negative tem-perature dependence indicates that the potential energy surface has no barrier or only a small one.

With the AB-22PT technique the rate coefficient for reaction (1) has been determined at 50 K and 100 K. For TH = Tion = 50 K a rate coefficient of 1.3 × 10-9 cm3s-1 has been obtained. At an accommodator temperature of 100 K and a trap temperature of 80 K, a value of 8.7 × 10-10 cm3s-1 has been measured. Note that in this case the internal temperature of the CH+ ion was somewhat smaller than the collisional temperature. The experimental and theoretical re-sults are summarized in Fig. 2. The non-thermal results of SIFDT technique have been converted using the approximation KEcm = 3/2 kT, the validity of which has been shown for atomic ions in drift fields. However, the contribution of rotational energy of CH+ can differ from translational energy and this may lead to differences in the “temperature” dependences. Our low temperature data which are close to Langevin limit show that there is no barrier that would sig-nificantly hinder the reaction. The theoretical values of phase space theory de-scribe well the behavior of the reaction.11 Recent RIOSA-NIP calculations13 which account for all reactive channels including hydrogen atom exchange, show the correct temperature trend; however, they underestimate the measured

5

value by almost a factor 10. This indicates that more work needs to be done in order to understand low temperature processes from first principles.

10 100 1000

10-10

10-9

Phase-Space Theory

RIOSA-NIP Theory

SIFDT

CH+ + H → C+ + H2

k / c

m3 s-1

T / K

AB-22PT

Figure 2. Experimental and theoretical rate coefficients for the indicated hydrogen abstraction reaction.

The H - H exchange in CH+ + H collisions is thermoneutral and it can be expected, e.g. from phase space theory, that the rate coefficient is much smaller than C+ production which is exothermic.11 Using the isotopically labeled system CH+ + D both channels, hydrogen abstraction and atom exchange

CH+ + D → C+ + HD + 0.434 eV (1a) → CD+ + H + 0.046 eV (1b)

can be distinguished. Also this system has been investigated in the trap at TD = Tion = 80 K. Surprisingly the same rate coefficient (1.2 ± 0.2) × 10-9 cm3s-1, has been obtained for both channels indicating the importance of the atom exchange channel. It is obvious that more detailed theoretical studies are needed in order to understand the dynamics of this basic reaction system.

3.2. CH4+ + H

An interesting reaction system which proceeds via the intermediate collision complex CH5

+ is

CH4+ + H → CH3

+ + H2. (2)

6

With the ICR technique,2 no hydrogen abstraction reaction has been observed with H and D atoms although this process is exothermic by 2.7 eV and forma-tion of CH5

+ collision complex is as well exothermic by 4.6 eV. Note that the detection limit of the ICR experiment was 10-11 cm3s-1.

Our investigations show that reaction (2) is highly reactive at low tempera-tures. The values 6.0 × 10-10 cm3s-1 and 5.1 × 10-10 cm3s-1 have been obtained at TH = Tion = 50 K and at TH = 100 K and Tion = 80 K, respectively. Based on the Langevin limit, 2 × 10-9 cm3s-1, this means that every forth collision leads to reaction. Assuming that the rate coefficient obtained at 300 K by the ICR tech-nique is correct, the reaction has very strong negative temperature dependence. This may be explained by the formation of a long lived CH5

+* complex in com-bination with a bottle neck hindering the transition towards the product channel. In order to make final conclusions on energetic and dynamics, additional meas-urements over the full accessible temperature range are planned.

10 100 1000

10-11

10-10

10-9

AB-22PT

22PT

CH4+ + H2 → CH5

+ + H

CH5+ + H → CH4

+ + H2

k / c

m3 s-1

T / K

TH = 100 KSIFDT

Figure 3. Temperature dependent rate coefficients for forward (squares) and backward (circles) reaction CH4

+ + H2 ↔ CH5+ + H.

3.3. CH5+ + H

An interesting reaction system including two fluxional CH4+ and CH5

+ ions is

CH4+ + H2 ↔ CH5

+ + H (3)

It has been investigated in both directions by SIFDT technique12 in the regime from thermal (300 K) to 0.12 eV center of mass kinetic energy, KEcm. An exoer-gicity, ΔH = 5 kJ/mol, and an entropy change, ΔS0

297 = 31 Jmol-1K-1 has been

7

derived for forward direction. From these results it has been concluded that reaction (3) is exoergic but endoentropic. Therefore the rate coefficient in backward direction, kb, should decrease steeply from 10-10 cm3s-1 at 20 meV (150 K) to 10-12 cm3s-1 at 8 meV (60 K) as indicated by the dotted line in Fig. 3. The rate coefficient in forward direction, kf, has been recently measured from 300 K down to 15 K using a 22PT,14 see full squares in Fig. 3.

In the present work, reaction (3) has been investigated in backward direc-tion at 10, 50, and 100 K. The temperature of the accommodator has been set at the 22PT value except for the lowest temperature where only 12 K have been achieved. The rate coefficient, kb, increases slightly from 9 × 10-12 cm3s-1 at 10 K to 2.3 × 10-11 cm3s-1 at 100 K, see solid circles in Fig. 3. Approaching 300 K, kb should increase significantly according SIFDT results. Our result at TH = 100 K and Tion = 300 K, see open circles in Fig. 3, shows that kb does not reach SIFDT value and therefore the internal temperature (Tion) of CH5

+ does not influence reactivity significantly. The small variation of kb with internal temperature of CH5

+ ions can be taken as an argument that a direct process is involved and the H atom impact determines predominantly the reactivity. It should be noted that this statement is based on both our and SIFDT results the difference between which may be also caused be experimental uncertainties. Our result at TH = Tion = 100 K agrees with prediction done in Ref. 12 within the error of measurement whereas kb does not drop significantly at lower temperatures as predicted. This indicates that the reaction is almost thermoneutral.

4. Conclusion

The presented results demonstrate that the combination of an rf ion trap with a neutral beam is a versatile tool for revealing information on reactions between ions and radical atoms at low temperatures. Using H or D atoms, the fundamen-tal collision systems C+ + H2 or HD, CH3

+ + H2 and CH4+ + H2 have been stud-

ied in detail in the reverse direction. A comparison of the experimental findings with theoretical results or expectations indicate that our understanding of these systems is still quite limited. Some of the observations can be explained with statistical models; however it is obvious that one has to account for nuclear spin restrictions which are known to be important in H and D atoms containing sys-tems at low temperatures. For the simple tri-atomic CH2

+ and CHD+ collision system the measurements indicate that a statistical theory is not satisfactory. In order to account correctly for direct collisions, quantum mechanical calculations beyond the so far used approximations13 are needed. Reactions which proceed via the CH5

+ intermediate may be described by statistical methods since already

8

the ground state of protonated methane is very fluxional and all H atoms are equivalent. This, however, does not explain the observed increase of reactivity by more than a factor 50 going from room temperature to 50 K. Predictions for the reactions occurring on the CH6

+ potential energy surface are handicapped by the fact, that already the asymptotic energies are uncertain. Besides that it is desirable to calculate correctly the actual phase spaces of the two competing channels CH5

+ + H and CH4+ + H2 at the temperatures of the present work. In

contrast to the conclusions made in Ref. 12, it is expected that a low tempera-tures the formation of CH5

+ should be “exoentropic”.

Acknowledgments

The development of this instrument involved contributions of several people. Especially we would like to mention the fruitful cooperation with Mark Smith. Financial support of the Deutsche Forschungsgemeinschaft (DFG) via the For-schergruppe FOR 388 "Laboratory Astrophysics" is gratefully acknowledged.

References

1. M. Sablier, C. Rolando, Mass Spec. Rev. 12 , 285 (1993). 2. Z. Karpas, V. Anicich, W. T. Jr. Huntress, J. Chem. Phys. 70, 2877 (1979). 3. W. Federer, H. Villinger, F. Howorka, W. Lindinger, P. Tosi, D. Bassi, E.

Ferguson, Phys. Rev. Lett. 52, 2084 (1984). 4. N. G. Adams, D. Smith, Astrophys. J. 294, L63 (1985). 5. A. Luca, D. Voulot, D. Gerlich, WDS'02 Proceedings of Contributed Pa-

pers, Part II, Safrankova (ed), Matfyzpress, 294 (2002). 6. S. Schlemmer, A. Luca, D. Gerlich, Int. J. Mass Spectrom. 223-224, 291

(2003). 7. D. Gerlich, Adv. Chem. Phys. 82, 1 (1992). 8. D. Gerlich, Physica Scripta T59, 256 (1995). 9. J. Slevin, W. Stirling, Rev. Sci. Instr. 52, 1780 (1981).

10. M. P. Hierl, R. A. Morris, A. A. Viggiano, J. Chem. Phys. 106, 10145 (1997).

11. W. J. Chesnavich, V. E. Akin, D. A. Webb, Ap. J. 287, 676 (1984). 12. W. Federer, H. Villinger, P., Tosi, D. Bassi, E. Ferguson, W. Lindinger; G.

H. F. Diercksen et. al. (eds), Molecular Astrophysics, D. Reidel Publishing Company, 649 (1985).

13. T. Stoecklin, Ph. Halvick, Phys. Chem. Chem. Phys. 7, 2446 (2005). 14. O. Asvany, I. Savic, S. Schlemmer, D. Gerlich, Chem. Phys. 298, 97

(2004).

129

Appendix D

131

REFERENCES [ada85] N.G. Adams, D. Smith, Laboratory studies of the reactions of HCO+ (and DCO+) and N2H+ (and

N2D+) with D (and H) atoms: Interstellar implications. Ap. J. 294 (1985) L63.

[alp59] W. W. Macalpine, R. O. Schildknecht, Coaxial resonators with helical inner conductor. proceedings of the IRE (1959) 2099.

[and65] J. B. Anderson, J. B. Fenn, Velocity distribution in molecular beams from nozzle sources. Phys. of Fluids. 8 (1965) 780.

[asv04a] O. Asvany, I. Savic, S. Schlemmer, D. Gerlich, Variable temperature ion trap studies of CH4+ +

H2, HD and D2: negative temperature dependence and significant isotope effect. Chem. Phys. 298 (2004) 97.

[asv04b] O. Asvany, S. Schlemmer, D. Gerlich. Deuteration of CHn+ (n=3-5) in Collisions with HD

Measured in a Low Temperature Ion Trap. Ap. J. 617 (2004) 685-692.

[bla77] J. H. Black, A. Dalgarno, Models of interstellar clouds. I. The Zeta Ophiuchi cloud. Ap. J. Suppl. Ser. 34 (1977) 405.

[boe58] J. H. De Boer, Bull. Soc. Chim. Belges 67 (1958) 284.

[bor08a] G. Borodi, A. Luca, C. Mogo, M. Smith, D. Gerlich, On the combination of a low energy hydrogen atom beam with a cold multipole ion trap, submitted to Rev. Sci. Instr. (2008).

[bor08c] G. Borodi, A. Luca, D. Gerlich, Reactions of CO2+ with H, H2 and deuterated analogues. Int. J.

Mass. Spec. in print DOI: 10.1016/j.ijms.2008.09.004

[bre31] G. Breit, I. I. Rabi, Measurement of nuclear spin. Phys. Rev. 38 (1931) 2082.

[cha88] N. Chan, D.M. Crowe, M. S. Lubell, F. C. Tang, A. Vasilakis, F. J. Mulligan and J. Slevin, A polarized atomic hydrogen beam. Z. Phys.D - Atoms, Molecules and Clusters 10 (1988) 393.

[che84] W. J. Chesnavich, V. E. Akin, D. A. Webb, Reevaluation of the rate constants for the reactions C+ + H2 → CH+ + H and CH+ + H → C+ + H2 at temperatures characteristic of ambient and shocked diffuse interstellar clouds. Ap. J. 287 (1984) 676.

[che94] P. Cassam - Chenai, F. Pauzat, Y. Ellinger, Is Stripping of Polycyclic Aromatic Hydrocarbons a Route to Molecular Hydrogen?. Molecules and Grains in Space, 50th International Meeting of Physical Chemistry, Proceedings of the Conference held in Mont Sainte-Odile, France, 312 (1994) 543.

[cla30] P. Clausing, Über die Strahlformung bei der Molekularströmung. Physica 9 (1930) 471.

[cox05] D. P. Cox, The three-phase interstellar medium revisited. Annu. Rev. Astron. Astrophys. 43 (2005) 337.

[cul93] R. W. McCullogh, J. Geddes, A. Donnelly, M.Liehr, M. P. Hughes and H. B. Gilbody, A new microwave discharge source for reactive atom beams. Meas. Sci. Technol. 4 (1993) 79.

[dal60] N. R. Daly, Scintillation type mass spectrometer ion detector., Rev. Sci. Instr. 31 (1960) 264.

[dal84] A. Dalgarno, S. Lepp, Deuterium fractionation mechanisms in interstellar clouds. Ap.J. 287 (1984) L47.

[dal98] P. F. Dalpiaz, W. Kubischta, F. Masoli, A. Reali, P. Slavich, Comparison of deuterium dissociation with a surface-wave discharge dissociator and SiO2 sputtered nozzle. Nuclear Instr. and Methods in Phys. Research A 402 (1998) 171.

[daw76] P. H. Dawson, Quadrupol Mass Spectrometry. Elsevier Scientific Publishing, Amsterdam. book (1976).

[dis99] E.v. Dishoeck, M.R. Hogerheijde, Models and Observations of the Chemistry Near Young Stellar Objects. in: Lada, C. J. and Kylafis, N.D. (eds): The Origin of Stars and Planetary Sys-tems, Kluwer, Dordrecht (1999) 97-140.

[don92] A. Donnelly, M. P. Hughes, J. Geddes, H. B. Gilbody, A microwave discharge atom beam source of high intensity. Meas. Sci. Technol. 3 (1992) 528.

132

[esr00] B. D. Esry, H. R. Sadeghpour, E. Wells and, I. Ben-Itzhak, Charge exchange in slow H+ + D(1s) collisions. J. Phys. B 33 (2000) 5329.

[ewa99] M. J. McEwan, G. B. I. Scott, N. G. Adams, L. M. Babcock, R. Terzieva and E. Herbst, New H and H2 reactions with small hydrocarbon ions and their roles in benzene synthesis in dense interstellar clouds. Ap. J. 513 (1999) 287.

[far73] M. McFarland, D. L. Albritton, F. C. Fehsenfeld, E. E. Ferguson and A. L. Schmeltekopf, Flow-drift technique for ion mobility and ion-molecule reaction rate constant measurements. I. Apparatus and mobility measurements. J. Chem. Phys. 59 (1973) 6610.

[fed84] W. Federer, H. Villinger, F. Howorka, W. Lindinger, P. Tosi, D. Bassi and E. Ferguson, Reaction of O+, CO+, and CH+ Ions with Atomic Hydrogen. Phys. Rev. Lett. 52 (1984) 2084.

[fed85] W. Federer, H. Villinger, P. Tosi, D. Bassi, E. Ferguson and W. Lindinger, Laboratory studies of ion reactions with atomic hydrogen. G.H.F. Diercksen et. al. (eds), Molecular Astrophysics, D. Reidel Publishing Company (1985) 649.

[feh71] F. C. Fehsenfeld, E. E. Ferguson, Rate constants for the reactions of CO2+ + H and H+ + CO2. J.

Geophys. Res. 76 (1971) 8453.

[feh73] F. C. Fehsenfeld, C. J. Howard, E. E. Ferguson, Thermal energy reactions of negative ions with H atoms in the gas phase. J. Chem. Phys. 58 (1973) 5841.

[fer69] E. E. Ferguson, F. C. Fehsenfeld, A. L. Schmeltekopf, Flowing afterglow measurements of ion-neutral reactions. Adv. At. Mol. Phys. 5 (1969) 1.

[ged93a] J. Geddes, R. W. McCullough, A. Donnelly, H. B. Gilbody, Dissociation of hydrogen in high - frequency discharges. Plasma Sources Sci. Technol. 2 (1993) 93.

[gel71] A. Gelb, S. K. Kim, Theory of atomic recombination on surfaces. J. Chem. Phys. 55 (1971) 4935.

[gen75] W. R. Gentry, D. J. McClure, C. H. Douglas, Merged beams at Minnesota (molecular beams). Rev. Sci Instr. 46 (1975) 367.

[ger02b] D. Gerlich, and S. Schlemmer, Deuterium Fractionation in Gas-Phase Reactions Measured in the Laboratory. Planetary and Space Science 50 (2002) 1287-1297.

[ger06a] D. Gerlich, M. Smith, Laboratory astrochemistry: studying molecules under inter- and circumstellar conditions. Physica Scripta 73 (2006) C25.

[ger08a] D. Gerlich, The study of cold collisions using ion guides and traps. in: Low temperatures and cold molecules, World Scientific Publishing, ed. by Ian W M Smith, ISBN 978-1-84816-209-9 (2008) 121.

[ger08b] D. Gerlich, The production and study of ultra-cold molecular ions. in: Low temperatures and cold molecules, World Scientific Publishing, ed. by Ian W M Smith, ISBN 978-1-84816-209-9 (2008) 295.

[ger09a] D. Gerlich, G. Borodi, Buffer gas cooling of poliatomic ions in rf-multi-electrode traps, Faraday Discussion, 142, (2009)

[ger92] D. Gerlich, Inhomogeneous Electrical Radio Frequency Fields: A Versatile Tool for the Study of Processes with Slow Ions. Adv. in Chem. Phys. LXXXII (1992) 1.

[ger92c] D. Gerlich, S. Horning, Experimental investigations of radiative association processes as related to interstellar chemistry. Chem. Rev. 92 (1992) 1509.

[ger95] D. Gerlich, Ion-Neutral Collisions in a 22-Pole Trap at Very Low Energies. Physica Scripta, T59 (1995) 256-263.

[gio60] J. A. Giordmaine, T. C. Wang, Molecular beam formation by long parallel tubes. J. Appl. Phys. 31 (1960) 463.

[glo03] S. C. O. Glover, Comparing gas-phase and grain-catalyzed H2 formation. Ap. J. 584 (2003) 331.

[glo06b] S. Glover, D. W. Savin, A. K. Jappsen, Cosmological implications of the uncertainity in H- destruction rate coefficients. Ap. J. 640 (2006) 553.

133

[glo08a] S. C. O. Glover, T. Abel, Uncertainties in H2 and HD chemistry and cooling and their role in early structure formation. Mon. Not. R. Astron. Soc. 388 (2008) 1627.

[gou63] R. J. Gould, E. E. Salpeter, The interstellar abundance of the hydrogen molecule. I. Basis processes. Ap. J. 138 (1963) 393.

[hae67] W. Haeberli, Sources of polarized ions. Ann. Rev. Nucl. Sci. 17 (1967) 373.

[hal07] P. Halvick, T. Stoecklin, P. Larrégaray, L. Bonnet, Cross sections and low temperature rate coefficients for the H + CH+ reaction: a quasiclassical trajectory study. Phys. Chem. Chem. Phys. 9 (2007) 582.

[hal85] K. Halbach, Application of permanent magnets in accelerators and electron storage rings (invited). J. Appl. Phys. 57 (1985) 3605.

[her01] E. Herbst, The chemistry of interstellar space. Chem. Soc. Rev., 30 (2001) 168.

[her03] E. Herbst, Isotopic fractionation by ion-molecule reactions. Space Science Rev., 106 (2003) 293.

[her05] E. Herbst, Molecular ions in interstellar reaction network. J. of Physics 4 (2005) 17.

[her73] E. Herbst, W. Klemperer, The Formation and Depletion of Molecules in Dense interstellar Clouds. Ap. J. 185 (1973) 505.

[her87b] A. Hershcovitch, A. Kponou, T. O. Niinikoski, Cold high - intensity atomic hydrogen beam source. Rev. Sci Instr. 58 (1987) 547.

[her99] E. Herbst, V. Le Page, Do H atoms stick to PAH cations in the interstellar medium?. Astron. & Astroph. 344 (1999) 310.

[hir41] J. O. Hirschfelder, Semi-empirical calculations of activation energies. J. Chem. Phys. 9 (1941) 645.

[hol71] D. Hollenbach, E. E. Salpeter, Surface Recombination of Hydrogen Molecules. Ap. J. 163 (1971) 155.

[hol97] D. Hollenbach, A.G.G.M. Tielens, Dense Photodissociation Regions (PDRs). An. Rev. Astron. Astroph. 35 (1997) 179.

[how61] W. M. Howard, Density field for rarefied flow through an orifice. Phys. Fluids 4 (1961) 521.

[isa98] L. G. Isaeva, B. A. Lazarenko, S. I. Mishnev, D. M. Nikolenko, S. G. Popov, I. A. Rachek, Yu. V. Shestakov, D. K. Toporkov, D. K. Vesnovsky, S. A. Zevakov, High field superconducting sextupole magnets. Nuclear Instr. and Methods in Phys. Research A 411 (1998) 201.

[kai02] R. I. Kaiser, Experimental Investigation on the Formation of Carbon-Bearing Molecules in the interstellar Medium via Neutral-Neutral Reactions. Chem. Rev. 102 (2002) 1309.

[kar79] Z. Karpas, V. Anicich, W. T. Huntress, Jr., An ion cyclotron resonance study of reactions of ions with hydrogen atoms. J. Chem. Phys. 70 (1979) 2877.

[kee77] C. F. McKee, J. P. Ostriker, A theory of the interstellar medium: three components regulated by supernova explosions in an inhomogeneous substrate. Ap. J. 218 (1977) 148.

[kil98] T. C. Killian, D. G. Fried, D. Landhuis, S. Moss, L. Willmann, T.J. Greytak, D. Kleppner, Two-photon spectroscopy of trapped atomic hydrogen. American Physical Society, DAMOP Meeting, abstract #HP.94 (1998).

[koc99a] N. Koch, E. Steffens, High intensity source for cold atomic hydrogen and deuterium beams. Rev. Sci Instr. 70 (1999) 1.

[lin03] P.J. Linstrom, W.G. Mallard, NIST Chemistry WebBook, NIST Standard Reference Database Number 69. March 2003, National Institute of Standards and Technology, Gaithersburg MD, 20899 (http://webbook.nist.gov) (2003).

[lis07] D. C. Lis, M. Gerin, E. Roueff, T. G. Phillip, D. R. Poelman, Interstellar deuteroammonia. Tracing physical conditions in dense, cold interstellar medium. Astrophys. Space Sci. doi10.1007/978-1-4020-6935-2_15

134

[luc05] A. Luca, G. Borodi, D. Gerlich, Interactions of ions with Hydrogen atoms. Progress report in XXIV ICPEAC 2005, Rosario, Argentina, July 20-26, 2005, Edited by F.D.Colavecchia, P.D.Fainstein, J.Fiol, M.A.P.Lima, J.E.Miraglia, E.C.Montenegro, and R.D.Rivarola XXIV (2005) 20.

[lui85] J. A. Luine, G. H. Dunn, Ion-molecule reaction probabilities near 10 K. Ap. J. 299 (1985) L67.

[mar89] R. E. March, R. J. Hughes, Quadrupole Storage Mass Spectrometry. John Wiley, New York (1989).

[mar98] A.G. Marshall, C.L. Hendrickson, G.S. Jackson, Fourier transform ion cyclotron resonance mass spectrometry: A Primer. Mass. Spec. Rev 17 (1998) 1.

[mil03] T. J. Millar, Deuterium fractionation in interstellar clouds. Space Science Rev., 106 (2003) 73.

[mil74] T. M. Miller, Atomic beam velocity distributions with a cooled discharge source. J. App. Phys. 45 (1974) 1713.

[mog09] C. Mogo, Proton affinities of molecules and thermodinamical equilibrium in a variable tempera-ture 22-pole ion traps. PhD-thesis. TU Chemnitz. (2009) in preparation

[moy04] G.E. Moyano, D. Pearson, M. A. Collins, Interpolated potential energy surface and dynamics for atom exchange between H and H3

+, and D and H3+. J. Chem. Phys. 121 (2004) 12396.

[mun63] M. S. B. Munson, F. H. Field, J. L. Franklin, Reactions of gaseous ions XIII. The System Meth-

ane- Hydrogen. J. Am. Chem. Soc. 85 (1963) 3584.

[pau88] H. Pauly, Other Low - Energy Beam Sources. in G. Scoles, Atomic and Molecular Beam Methods 1 (1988) 83.

[pir00] V. Pirronello, O. Biham, G. Manicó, J. E. Roser and G. Vidali, Laboratory studies of molecular hydrogen formation on surfaces of astrophysical interest. in: "Molecular Hydrogen in Space", F. Combes and G. Pineau des Forets (eds.) Cambrigde University Press, Standford (2000) 71.

[row84] B. R. Rowe, G. Dupeyrat, J. B. Marquette, P. Gaucherel, Study of the Reactions N2+ + 2N2 →

N4+ + N2 and O2

+ + 2O2 → O4+ + O2 from 20 to 160 K by the CRESU Technique. J. Chem.

Phys. 80 (1984), 4915-4921.

[sab93] M. Sablier, C. Rolando, Gas - phase ion - atom reactions. Mass Spectrometry Reviews 12 (1993) 285.

[sav05a] I. Savić, I. Cermak, D. Gerlich, Reactions of Cn (n=1-3) with ions stored in a temperature-variable radio frequency trap. Int. J. Mass. Spec. 240 (2005) 139.

[sav05b] I. Savić, D. Gerlich, Temperature variable ion trap studies of C3Hn+ with H2 and HD. Phys.

Chem. Chem. Phys 7 (2005) 1026.

[sav05c] I. Savić, S. Schlemmer, D. Gerlich, Low-temperature laboratory measurements of forming deuterated C3H3

+. Ap. J. 621 (2005) 1163-1170

[sav08] D. W. Savin, Progress towards laboratory measurements of H2 formation for early universe chemistry. Atomic and Molecular Physics of the Early Universe (2008).

[scl06] S. Schlemmer, O. Asvany, E. Hugo and D. Gerlich, Deuterium Fractionation and Ion-Molecule Reactions at Low Temperatures. Proceedings of the International Astronomical Union, Sympo-sium, Published online by Cambridge University Press doi: 10.1017/S1743921306007113 S231 (2006) 125-134

[sco97a] G. B. I. Scott, D. A. Fairley, C. G. Freeman, M. J. McEwan, P. Spanel and D. Smith, Gas phase reactions of some positive ions with atomic and molecular hydrogen at 300 K. J. Chem. Phys. 106 (1997) 3982.

[sin89] D. Singy, P. A. Schmelzbach, W. Grüebler, W. Z. Zhang, Production of intense polarized hydrogen atomic beams by cooling the atoms to low temperature. Nuclear Instr. and Methods in Phys. Research A 278 (1989) 349.

135

[sin90] D. Singy, P. A. Schmelzbach, W. Grüebler, W. Z. Zhang, Study of the surface recombination in the production of a dense polarized hydrogen atomic beam at low temperature. Nuclear Instr. and Methods in Phys. Research B 47 (1990) 167.

[sle81] J. Slevin, W. Stirling, Radio frequency atomic hydrogen beam source. Rev. Sci Instr. 52 (1981) 1780.

[smi00] I. W. M. Smith, B. R. Rowe, Reaction Kinetics at Very Low Temperatures: Laboratory Studies and Interstellar Chemistry. Accounts of Chemical Research 33 (2000) 261-268.

[smi04] I. W. M. Smith, E. Herbst, Q. Chang, Rapid neutral-neutral reactions at low temperature: a new network and first results for TMC-1. Mon. Not. R. Astron. Soc. 350 (2004) 323.

[smi98] M. A. Smith, Low Temperature Rate Studies of Ions and Radicals in Supersonic Flows. Interna-tional Reviews in Phys. Chem. 17 (1998), 35-63.

[smi43] W. V. Smith, The surface recombination of H atoms and OH radicals. J. Chem. Phys. 11 (1943) 110.

[smi82a] D. Smith, N.G. Adams, E. Alge, Isotope exchange and collisional association in the reaction of CH3

+ and its deuterated analogs with H2, HD, and D2. J. Chem. Phys. 77 (1982) 1261.

[smi79] D. Smith, N.G. Adams, Recent advances in flow tubes: Measurement of ion-molecule rate coefficients and product distribution. in: Gas Phase Ion Chemistry, Vol.1, M.T. Bowers, Ed., Academic Press, New York (1979) 1.

[smi92] D. Smith, The ion chemistry of interstellar clouds. Chem. Rev. 92 (1992) 1473.

[sno06] T. P. Snow, B. J. McCall, Diffuse atomic and molecular clouds. Annu. Rev. Astro. Astrophys. 44 (2006) 367.

[sta05] M. Statera, M. Stancari, M. Capiluppi, G. Ciullo, M. Contalbrigo, P. F. Dalpiaz, F. Giordano, P. Lenisa, and M. Wang, A high intensity superconducting atomic beam source. Ieee Transactions on applied superconductivity 15 (2005) 1164.

[sto05] T. Stoecklin, Ph. Halvick, Low temperature quantum rate coefficient of the H + CH+ reaction. Phys. Chem. Chem. Phys 7 (2005) 2446.

[szc00] D. Szczerba, L. D. van Buuren, J. F. J. van den Brand, H. J. Bulten, M. Ferro-Luzzi, S. Klous, H. Kolster, J. Lang, F. Mul, H. R. Poolman and M. C. Simani, A polarized hydrogen/deuterium atomic beam source for internal target experiments. Nuclear Instr. and Methods in Phys. Research A 455 (2000) 769.

[teu00] Y. H. Le Teuff, T. J. Millar, A. J. Markwick, The UMIST database for astrochemistry 1999. Astron. & Astroph. Suppl. Ser. 146 (2000) 157.

[tos84] P. Tosi, S. Iannotta, D. Bassi, H. Villinger, W. Dobler and W. Lindinger, The reaction of CO2+

with atomic hydrogen. J. Chem. Phys. 80 (1984) 1905.

[tru66] S. M. Trujillo, R. H. Neynaber, E. W. Rothe, Merging beams, a different approach to collision cross section measurements. Rev. Sci. Instr. 37 (1966) 1655.

[wal04] C. M. Walmsley, D. R. Flower, G. Pineau des Forets, Complete depletion in prestellar cores. Astron. & Astroph. 418 (2004) 1035.

[wal82] J. T. M. Walraven, I. F. Silvera, Helium - temperature beam source of atomic hydrogen. Rev. Sci. Instr. 53 (1982) 1167.

[wat73] W. D. Watson, Formation of the HD molecule in the interstellar medium. Ap. J. 182 (1973) L73.

[wis67] H. Wise, B. J. Wood, Reactive collisions between gas and surface atoms. Adv. At. Mol. Phys. 3 (1967) 291.

[woo07] J. Woodall, M. Agúndez, A. J. Markwick-Kemper, T. J. Millar, The UMIST database for astrochemistry 2006. Astron. & Astroph. 466 (2007) 1197.

[woo62] B. J. Wood, H. Wise, The kinetics of hydrogen atom recombination on Pyrex glass and fused Quartz. Technical rept (1962) 1049.

137

LIST OF PUBLICATIONS AND CONFERENCE CONTRIBUTION

List of Publications M. Vasilescu, G. Borodi, S. Simon, MAS NMR and SEM study of local structure changes induced by heat treatement in La2B4Al2O12. Journal of Optoelectronics and Advanced Materials Vol. 5, No. 1, March 2003, p. 153 – 156

A. Luca, G. Borodi, D. Gerlich, Interactions of ions with Hydrogen atoms. Progress report in XXIV ICPEAC 2005, Rosario, Argentina, July 20-26, 2005, Edited by F.D.Colavecchia, P.D.Fainstein, J.Fiol, M.A.P.Lima, J.E.Miraglia, E.C.Montenegro, and R.D.Rivarola

G. Borodi, A. Luca, D. Gerlich, Reactions of CO2+ with H, H2 and deuterated ana-

logues, Int. J. Mass Spectrom. 280, 218–225, (2009)

D. Gerlich, G. Borodi, Buffer gas cooling of poliatomic ions in rf-multi-electrode traps, Faraday Discussion, 142, invited contribution (2009)

Conference Contributions A. Luca, G. Borodi, D. Gerlich, Reactions of stored ions with H-atoms Poster, TMR Astrophysical Chemistry Final Network Meeting, Perugia April 2002, Italy

A. Luca, G. Borodi, D. Gerlich, Reactions of stored Ions with H-Atoms and Molecules Poster, FGLA symposium, Großbothen September 2002, Germany

G. Borodi, A. Luca, D. Gerlich, Reaction between ions and H-atoms, Talk, DPG-Fruhjahrstagung, Hannover March 2003, Germany

G. Borodi, A. Luca, R. Plasil, D. Gerlich, Combination of Supersonic Molecular Beam and Low Temperature 22–Pole Ion Trap. Association of CH3

+ + CO, Poster, DPG-Fruhjahrstagung, Munchen March 2004, Germany

G. Borodi, A. Luca, C. Mogo, D. Gerlich, Combination of a cold rf-ion trap with a slow H atom beam for investigating ion-atom collisions, Poster, Symposium organized by the FOR 388, Schloss Hotel Pillnitz, Dresden June 2005, Germany

A. Luca, G. Borodi, C. Mogo, D. Gerlich, Reactioons of small hydrocarbon ions CHn+

with H and D atoms at low temperatures, Poster, Symposium organized by the FOR 388, Schloss Hotel Pillnitz, Dresden June 2005, Germany

139

SELBSTSTÄNDIGKEITSERKLÄRUNG

Ich erkläre, dass ich die vorliegende Arbeit selbstständig und nur unter Verwendung der angegebenen Literatur und Hilfsmittel angefertigt habe. Ich erkläre, nicht bereits früher oder gleichzeitig bei anderen Hochschulen oder an dieser Universität ein Promotionsverfahren beantragt zu haben. Falls diese Erklärung nicht zutrifft, füge ich eine Stellungnahme diesem Antrag bei. Ich erkläre, obige Angaben wahrheitsgemäß gemacht zu haben und erkenne die Promotionsordnung der Fakultät für Naturwissenschaften der Technischen Universität Chemnitz vom 10. Oktober 2001 an.

Gheorghe Borodi

141

CURRICULUM VITAE

Personal Name: Borodi Gheorghe

Birth date and place: 06.10.1970, Caianu Mic, Romania

Citizenship Romanian

Status Married, three children

Studies School: 1985-1989 High school: “Liviu Rebreanu”, Bistrita, Roma-

nia, Mathematics - Physics University:

1990-1996

M. SC. Degree in Atomic and Molecular Physics obtained at “Babes-Bolyai” University, Cluj-Napoca, Romania

Master thesis title: “Determination of U traces in fossils and minerals”

B. SC. Degree in Physics obtained at “Babes-Bolyai” University, Cluj-Napoca, Romania

Diploma thesis title: “Determination of geologi-cal ages using Rd-Sr method”

PhD studies: 2001 – 2005 Ion trap studies in the group of Prof. D. Gerlich, Faculty of Natural Sciences, Institute of Physics, TU Chemnitz, Germany

2007 – 2008 Scientific employ in the group of Prof. D. Ger-lich, Faculty of Natural Sciences, Institute of Physics, TU Chemnitz, Germany

143

ACKNOWLEDGEMENTS

At this point I would like to express my gratitude to all the people who have contributed to the results of this dissertation.

First of all I would like to thank Prof. Dr. Dieter Gerlich for the opportunity to study in Germany, for the interesting research project and for the excellent su-pervision

Thanks to Dr. Alfonz Luca for his introduction into the field of trapping tech-niques and his help and supervision during the research work

Thanks to Prof. Dr. Stephan Schlemmer and Prof. Dr. Mark Smith for many con-tributions to this work

Thanks to engineers Adelheid Steinrücken, Doreen Kunte, Angelika Hiller and Gunter Vales for the excellent technical support.

Thanks to the Prague team namely Prof Dr. Juraj Glosik and Dr. Radek Plasil for the interesting discussions and especially for the support with the data acquisi-tion system

Thanks to the administrative personal, namely Ivonne Schubert and specially Annet Kurasch who with her infinite patience helped me to solve so many prob-lems.

I also would like to thank the present and some of the former group members, namely Dr. Oskar Asvany, Silko Barth, Dr. Ivo Čermák, Silvio Decker, Dr. Hans – Jürgen Deyerl, Mirko Kämpf, Thomas Mehner, Cesar Mogo, Dr. Christophe Nicolas, Dr. Igor Savić, Dr. Didier Voulot, for the nice atmosphere and their help.

At last but not least I am especially thankful to my wive who encouraged and supported me during all this period in a way that I will never be able to forget.

Financial support by the Deutsche Forschungsgemeinschaft (DFG) is gratefully ac-knowledged, especially via the Forschergruppe FOR 388 "Laboratory Astrophysics".

on the combination of a low energy hydrogen atom … · referat der erste teil der aktivitäten...

Documents