Examining Nanoenvironments in Solids on theScale of a Single, Isolated Impurity Molecule

W. E. Moerner*

Optical spectroscopy of single impurity molecules in solids can be used as an exquisitelysensitive probe of the structure and dynamics of the specific local environment around thesingle molecule (the "nanoenvironment"). Recently observed effects such as spectraldiffusion, perturbations by external fields, changes in molecular photophysics, shifts invibrational modes, optical modification of the absorption spectrum, dynamics due to amor-phous system physics, and magnetic resonance of a single molecular spin attest to thevitality of and growing interest in this new field, which may lead to optical storage on thesingle-molecule level.

The power of high-resolution laser spec-troscopy has recently been extended intothe fascinating domain of individual impu-rity molecules in solids, where the singlemolecule acts as a probe of the detailedlocal environment of unprecedented sensi-tivity (1-5). In ongoing experiments atseveral laboratories around the world, ex-actly one molecule hidden deep within asolid sample is probed at a time by tunablelaser radiation, which represents detectionand spectroscopy of 1.66 x 10-24 moles ofmaterial, or 1.66 yoctomoles. Single mole-cule studies completely remove the normalensemble averaging that occurs when alarge number of molecules are probed at thesame time. Thus, the usual assumption thatall molecules contributing to the ensembleaverage are identical can now be directlyexamined, and strong tests of truly micro-scopic theories can be completed.

Single molecule spectroscopy (SMS) insolids is related to, but distinct from, thewell-established field of sepectroscopy ofsingle electrons, atoms, or ions confined inelectromagnetic traps or on surfaces (6-8).The vacuum environment and confiningfields of an electromagnetic trap are quitedifferent from the environment of a singlemolecule in a solid where the lattice con-strains the molecule, preventing rotation inmost cases, and where the molecule isbathed in the phonon vibrations of thesolid available at a given temperature. Thenear-field probing of atoms on surfaces withscanning tunneling microscopy or atomicforce microscopy (9) usually requires astrong bond between the molecule of inter-

The author is at the IBM Research Division, AlmadenResearch Center, 650 Harry Road, K13/801, San Jose,CA 95120-6099, USA, and Laboratory of PhysicalChemistry, Swiss Federal Institute of Technology, Uni-versitatstrasse 22, CH-8092, Zurich, Switzerland.*Present address: IBM Research Division, AlmadenResearch Center, 650 Harry Road, K13/801, San Jose,CA 95120-6099, USA.

46

est and the underlying surface as well astolerance of the perturbing forces from thetunneling electrons or the tip. In contrast,SMS usually operates noninvasively in theoptical far field with a corresponding loss inspatial resolution but with no loss of spec-tral resolution. In recent work, single-mol-ecule imaging has been achieved with near-field optical techniques with 100-nm reso-lution at room temperature (10, 1 1). How-ever, the spectral resolution that will beavailable under such conditions is three tofour orders of magnitude poorer than theresults of high-resolution SMS describedhere.

This article presents an overview ofsome of the recent advances produced bythe spectroscopy of single impurity centersin solids, with emphasis on individual, iso-lated molecular impurities. One of severalreviews may be consulted for more informa-tion (4, 5). These SMS studies are signifi-cant because (i) new physical effects havebeen observed in the single molecule re-gime and (ii) it is now possible in a singlenanoenvironment to probe directly theconnection between specific microscopictheories (12) of local structure, dynamics,and host-guest interactions and the statisti-cal mechanical averages that are measuredin conventional experiments.

How Single Molecule Spectroscopyin Solids Is Accomplished

One may ask: How is it possible to useoptical radiation to isolate a single impuritymolecule hidden deep inside a host matrix?There are several requirements necessary toensure that the fluorescence emission fromthe molecule dominates over backgroundsignals (13, 14). First, only a small volumeof sample on the order of a few hundredcubic micrometers is probed by the use of athin sample and a small (micrometer-sized)laser spot produced by a lens, an optical

SCIENCE * VOL. 265 * 1 JULY 1994

fiber, or an aperture. In addition, the im-purity concentration in the transparenthost is low, in the range of 10-7 to 10-imol/mol. These two actions alone are in-sufficient to guarantee that only one impu-rity molecule in the probed volume is inresonance with the laser at a time. Addi-tional selectivity on the order of a factor of104 or so is provided by the careful selectionof guest and host and the well-known prop-erties of inhomogeneously broadened ab-sorption lines in solids (15-18), summa-rized below.

For the usual case of detection by fluo-rescence excitation, the quantum yield forphoton emission per absorption eventshould be high. A further requirement onthe guest species stems from the photophys-ics of optical absorption in solids. Theprobability that a single molecule will ab-sorb a photon from the incident beam issimply (*pk/A, where Urpk is the peak absorp-tion cross section and A is the area of thelaser beam. It has been recognized since thefirst SMS experiments that to produce ab-

A

BLaserbeam\Sample

I0Vk 1 0

0 0 000 ~ sm~0

0 0 0

MoleculeLine center Wings of line

Fig. 1. (A) Schematic showing an inhomoge-neous line at low temperatures and the princi-ple of SMS in solids. The entire line is formed asa superposition of Lorentzian profiles of theindividual absorbers, with a distribution of cen-ter resonance frequencies caused by randomstrains and imperfections. In the inset, severaldopant molecules are sketched with differentnanoenvironments. (B) Controlling the numberof impurity molecules in resonance in theprobed volume by changing the laser wave-length. The laser linewidth is negligible on thescale shown.

...... """' ', -- `.: "' `........... 1,11,11-1,1,1,1,11, "' ---` .."I""""""""",l.....-..--..0.....'.. ----

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

sorption events efficiently with the incidentphotons rather than unwanted scatteringsignals, Cpk should be as large as possible(1, 14, 19-21). Because apk depends linear-ly upon the oscillator strength and inverselyupon the optical homogeneous linewidth,AVH, strong absorptions and narrow linesgive the largest Upk values. Strong absorp-tions are conveniently provided by electric-dipole-allowed, singlet-singlet transitionsin aromatic molecules where the oscillatorstrength is often near unity. The narrowestoptical linewidths in solids occur for zero-phonon, zero-vibration electronic transi-tions at low temperatures, because suchtransitions can only broaden by phononscattering. For rigid molecules like penta-cene or perylene in a solid below 10 K, thelinewidth of the optical transition near 500THz is on the order of 10 MHz, and 0pkreaches values near 10-11 cm2, thousandsof times the area of a single molecule.

There is an additional crucial effect im-portant for SMS resulting from the extremenarrowness of zero-phonon transitions: in-homogeneous broadening (15, 16) (see Fig.1, top). This effect occurs when AvH be-comes narrower than the (mostly static)distribution of resonance frequencies in thesolid caused by dislocations, point defects,or random internal strain and electric fieldsand field gradients (inset). Inhomogeneousbroadening is a universal feature of high-resolution optical spectroscopy of defects insolids (17, 18), and of other spectroscopiesin which zero-phonon lines are probed. (Adynamic inhomogeneous broadening alsooccurs for Doppler-broadened lines of gas-es.) Inhomogeneous broadening facilitatesthe selection of a single impurity, becausethe different guest molecules in the probevolume take on slightly different resonancefrequencies according to the local fields atthe location of the impurity. One thensimply uses the tunability of a narrow-bandlaser to select different single-impurity cen-ters. Naturally, this selection must be done

A

B

in a region of the optical spectrum wherethe number of molecules per homogeneouswidth is on the order of or less than one.This may be accomplished by (i) tuning thelaser wavelength out into the wings of theinhomogeneous line (Fig. 1, bottom right),(ii) using a sample with a very low dopinglevel, or (iii) producing a large inhomoge-neous linewidth; all three of these methodshave been used.A further requirement on the absorption

properties of the probe molecule is the ab-sence of strong bottlenecks in the opticalpumping cycle. In organic molecules, inter-system crossing (ISC) from the singlet statesinto the triplet states represents a bottle-neck, because photon emission usually ceasesfor a relatively long time equal to the tripletlifetime when ISC occurs. This effect re-sults in premature saturation of the emis-sion rate from the molecule and reductionof upk compared to the case with no bottle-neck (22). Thus, molecules with weak ISCyield and short triplet lifetime are prefera-ble, such as rigid, planar aromatic dyes.A final requirement for SMS is the

selection of a guest-host couple that allowsfor the photostability of the impurity andweak spectral hole-buming. By this spectralhole-burning, we include any light-inducedchange in the resonance frequency of themolecule caused either by photochemistryof the molecule or by a photophysicalchange in the nearby environment (23).For example, most detection schemes withoverall photon collection efficiency of 1 to

A j15000 1

.a0

a,

C

0.1% require that the quantum efficiency forhole-burning, q, be less than 10 6 to 10-7.This requirement is necessary to providesufficient time-averaging of the single mole-cule line shape before it changes appreciablyor shifts to another spectral position.

All ofthese concepts may be used to derivespecific expressions for the signal-to-noise ra-tio (SNR) for SMS (4, 24), and values on theorder of 20 can be expected with 1-s integra-tion time and spot diameters of a few mi-crometers for a system like pentacene impuri-ties in ap-terphenyl crystal. To date, SMS hasbeen achieved in four host-guest combina-tions with the three impurity moleculesshown in Fig. 2: pentacene in p-terphenylcrystals (1), perylene in poly(ethylene) (25),terrylene in poly(ethylene) (26), and quiterecently, terrylene in the Shpol'skii matrixhexadecane (27). The Shpol'skii matrices inparticular should provide a large new class ofmaterials that allow SMS.

Examples of SingleMolecule Spectra

Fluorescence excitation spectra for a 10-gm-thick sublimed crystal of pentacene inp-terphenyl at 1.5 K are shown in Fig. 3 [forspecific details of the experimental appara-tus, see (13)1. In this method, the opticalabsorption is measured by recording thetotal fluorescence emission shifted to longwavelengths as the laser frequency isscanned over the resonance. The 18-GHzspectrum in Fig. 3A (obtained by the scan-

Fig. 3. Fluorescence excitationspectra for pentacene in p-ter-phenyl at 1.5 K. (A) Broad scan ofthe inhomogeneously broadenedline; all the spikes are repeatablefeatures. (B) Expansion of a2-GHz region [area circled in (A)]showing several single mole-cules. The laser detuning is refer-enced to the line center at592.321 nm. (C) Low-power scanof a single molecule at 592.407nm showing the lifetime-limited

10t.0 width of 7.8 MHz and a Lorentzian---11\10.

fit; cps, counts per second.

c

Fig. 2. Structures of molecules used so far forSMS: (A) pentacene, (B) perylene, and (C)terrylene.

450

50 ~=-50 0 50

MHZ

SCIENCE * VOL. 265 * 1 JULY 1994 47

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

ning of a 3-MHz linewidth dye laser overthe entire inhomogeneous line) contains20,000 points; to show all the fine structureusually requires several meters of linearspace. The structures appearing to be spikesare not noise; all features shown are staticand repeatable. Near the center, the "spec-tral roughness" is a fundamental effectcalled statistical fine structure (SFS) (28,29) which arises directly from statisticalvariations in the spectral density of reso-nance frequencies. It is immediately obvi-ous that the inhomogeneous line is far fromGaussian in shape and that there are tailsextending out many standard deviationsboth to the red and to the blue. Figure 3Bshows an expanded region in the wing ofthe line. Each of the narrow peaks is theabsorption line of a single molecule. Thedifferent molecules have slightly differentpeak heights because the laser transverseintensity profile is bell-shaped and the mol-ecules are not all located at the center ofthe focal spot. Although these spectra seemnarrow, they are in fact slightly power-broadened by the probing laser.

Upon the close examination of an indi-vidual single-molecule peak at lower inten-sity (Fig. 3C), the quantum-limited line-width of 7.8 ± 0.2 MHz can be observed(30). Here, the quantum limit applies be-cause the optical linewidth has reached theminimum value allowed by the lifetime ofthe optical excited state. This value is inexcellent agreement with previous photonecho measurements of AVH using large en-sembles of pentacene molecules (31, 32).In summary, it is clear that well-isolatedsingle molecule spectra like those in Fig. 3allow many spectroscopic studies of thelocal environment to be performed, becausesuch narrow lines are much more sensitiveto local perturbations than are broad spec-tral features.

Historical Background andSummary of Results

The foundations for SMS were laid in 1987when the SFS signal scaling as the squareroot of the number of absorbers was firstobserved (28, 29) for pentacene in p-ter-phenyl. The first single molecule spectrawere also recorded in this model systemwith a sophisticated zero-scattering-back-ground absorption technique, FM spectros-copy, and either Stark or ultrasonic strainmodulation of the absorption line (1, 2).The SNR in these experiments of about 5:1was limited by the shot noise of the laser. In1990, Orrit and Bernard demonstrated thatfluorescence excitation produces superiorsignals if the emission is collected efficientlyand if the scattering sources are minimized(3). Subsequent experiments have used flu-orescence excitation exclusively.

For example, with SMS methods it hasbeen possible to observe the quantum-lim-ited linewidth of a single molecule (30),temperature-dependent dephasing (13), op-tical saturation (13), the effects of appliedelectric fields in crystals (33) and polymers(26), and shifts caused by external hydro-static pressure (34), all of which representapplications of the tools of the spectrosco-pist to the narrow spectral line of a singlemolecule in a solid at low temperatures.With a proper time-delayed photon count-ing apparatus, the fluorescence emissionlifetime of a single molecule can be directlymeasured (35). More surprisingly, severalunexpected effects have also been observed,including the spectral diffusion of a singlemolecule due to dynamics of the nearbyhost (36), reversible optical modificationthat could lead to optical storage on a singlemolecule scale (24, 25), and even magneticresonance of a single molecular spin (37-39). The correlation properties of the emit-ted photons have been used to good advan-tage to measure the photophysics of singlemolecules (40), resonance frequency fluc-tuations due to host degrees of freedom(41), and a quantum-optical effect, photonantibunching (42). It has also been possibleto obtain fairly detailed information aboutthe nanoenvironment of a single moleculeby the measurement of the vibrationalmode properties in the electronic groundstate (43-45).

Imaging Single Molecules in Solids

As a survey of some specific SMS studies,Fig. 4 shows a three-dimensional "image" ofsingle molecules in a solid first obtained forpentacene in p-terphenyl some years ago(13). The z axis of the image is the usualfluorescence signal, the horizontal axis isthe laser frequency detuning (300-MHzrange), and the axis going into the page isone transverse spatial dimension (40-pmrange). The spatial scan is produced byslowly translating the laser focal spot acrossthe face of the crystal and obtaining spectraat each position. There are three large,

clear, single molecule peaks localized inboth frequency and position at the center,upper left, and upper right. The spatialresolution of this image is clearly limited bythe 5-pum-diameter laser spot; however, inthe frequency dimension, the features arefully resolved.

The frequency widths of the peaks inFig. 4 are slightly larger than the lifetime-limited width (Fig. 3C) because of the useof higher probing intensity. The oddlyshaped peak between the two strong mole-cules at the center and upper right resultsfrom a molecule that is spectrally diffusingbecause of changes in its local nanoenvi-ronment during the measurement (see thenext section).

In recent work at ETH-Znrich, four-dimensional images have been obtained forpentacene in p-terphenyl with two spatialand one frequency dimension as well as thesignal dimension (46). This "fluorescencemicroscopy" experiment used an image in-tensifier-video camera combination specif-ically configured as a photon counter ineach pixel. Single molecules resemble star-like objects that appear and disappear as thelaser frequency is scanned. Future work inthis area may increase the spatial resolutionby near-field optical techniques (10). Atlow temperatures, the narrowness of thespectral features compared to room temper-ature should provide unprecedented detailof the local environment of the impuritymolecule.

Direct Observations of SpectralDiffusion and Host Dynamics

When a new regime is first opened for study,new physical effects can often be observed.In the course of the early SMS of pentacenein p-terphenyl, an unexpected phenomenonappeared: resonance frequency shifts of indi-vidual pentacene molecules in a crystal at1.5 K (36), called spectral diffusion by anal-ogy to amorphous systems (47). Here, spec-tral diffusion means changes in the center(resonance) frequency of a defect due toconfigurational changes in the nearby host

Fig. 4. Three-dimen-sional "images" of sin-gle molecules. The mea-sured fluorescence sig-nal (z axis) is displayedover a range of 300 MHzin excitation frequencynear 592.544 nm (hori-zontal axis) and 40 plmin spatial position (axisinto the page). (Ren-dered with IBM Data Ex-plorer, San Jose, Cali-fornia.)

SCIENCE * VOL. 265 * 1 JULY 1994

gnawicclcnc;QR.... 2^----------^-------------

48

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

that affect the frequency of the electronictransition through guest-host coupling. Ex-perimentally, two distinct classes of singlepentacene molecules were identified: class Imolecules, which have center frequenciesthat were stable in time like the three largemolecules in Fig. 4, and class II molecules,which showed spontaneous, discontinuousjumps in resonance frequency of 20 to 60MHz on a 1- to 420-s time scale, which areresponsible for the distorted single-moleculepeak in Fig. 4.

To illustrate the behavior of the type IImolecules, Fig. 5A shows a sequence ofexcitation spectra of a single molecule tak-en as fast as allowed by the available SNR.The laser was scanned once every 2.5 s with0.25 s between scans, and the hopping ofthis molecule from one resonance frequencyto another is clearly evident. By acquiringhundreds to thousands of such spectra, atrajectory or trend of the resonance fre-quency can be obtained (Fig. 5B, for thesame molecule as Fig. 5A). For this mole-cule, the optical transition energy appearsto have a preferred set of values and per-forms spectral jumps between these valuesthat are discontinuous on the 2.5-s timescale of the measurement. The behavior ofanother molecule is shown in Fig. 5C at 1.5K and in Fig. 5D at 4.0 K. This moleculewanders in frequency space with manysmaller jumps, and both the rate and rangeof spectral diffusion increase with tempera-ture, suggesting a phonon-driven process.The occurrence of class II molecules was

Fig. 5. Examples of single mole-cule spectral diffusion for penta-cene in p-terphenyl at 1.5 K. (A) Aseries of fluorescence excitationspectra each 2.5 s long spacedby 2.75 s showing discontinuousshifts in resonance frequency,with zero detuning at 592.546 nm.(B) Trend or trajectory of the res-onance frequency over a longtime scale for the molecule in (A).(C) Resonance frequency trendfor a different molecule at 592.582nm at 1.5 K and (D) at 4.0 K.

quite common in the wings of the inhomo-geneous line, but only class I moleculeswere observed close to the center, suggest-ing a connection to disorder in the crystalstructure. In addition, the jumping rate didnot depend on the probing laser power. Thespectral diffusion appeared to be a sponta-neous process rather than a light-induced,spectral hole-burning effect (30).

Because the optical absorption is highlypolarized (48) and the peak signal from themolecule does not decrease when the spec-tral jumps occur, it is unlikely that themolecule is changing orientation in thelattice. Because the resonance frequency ofa single molecule in a solid is extremelysensitive to the local strain, the conclusionis that the spectral jumps are due to internaldynamics of some configurational degrees offreedom in the surrounding lattice. Thesituation is analogous to that for amorphoussystems, which are postulated to contain amultiplicity of local configurations that canbe modeled by a collection of double-wellpotentials [the two-level system, or TLS,model (49)1. The dynamics result fromphonon-assisted tunneling or thermally ac-tivated barrier-crossing in these potentialwells. One possible source for the tunnelingstates (13) could be discrete torsional libra-tions of the central phenyl ring of thenearby p-terphenyl molecules about themolecular axis. The p-terphenyl moleculesin a domain wall between two twins or nearlattice defects may have lowered barriers tosuch central-ring tunneling motions. Fur-

*1Aa

200 B0

-200400

W 0.0 100.0 200.0 300.0 400.0 50sa

t 400 [7-7

-200

o0.0 500.0 1000.0 1500.0 200C.

Time (s)

ther study is necessary to identify conclu-sively the specific molecular motions re-sponsible for the effect. These direct obser-vations of the dynamics of a nanoenviron-ment of a single molecule have sparked newtheoretical studies of the underlying micro-scopic mechanism (12, 50).

Spectral diffusion similar to that shownin Fig. 5 was also observed in SMS ofperylene in poly(ethylene) (24, 25) and ofterrylene in poly(ethylene) (26, 51). Asopposed to the crystalline system, spectraldiffusion is expected here (47) and is ob-served in a variety of time regimes. On thefast (sub-second) time scale, spectral shiftson the order of 10 to 100 MHz producesingle-molecule line shapes that fluctuatefrom measurement to measurement. Onlonger time scales, jumping behavior can beobserved that is similar to that shown inFig. 5B. Detailed measurements of terrylenein poly(ethylene) (51) also suggest that athigher laser powers, more spectral diffusionis observed that may be regarded as a type oftransient spectral hole-burning. Orrit andcolleagues (41, 52) have used photon cor-relation techniques to study such processes.

These measurements of spectral diffusionin crystals and polymers illustrate the powerof SMS in detecting changes in the na-noenvironment of a single molecule. Thespectral changes can be followed in realtime for each impurity molecule; no ensem-ble averages over "equivalent" centers needbe made that would obscure the effects.

Optical Modification of `

Single Molecule Spectra:Applications to Optical Storage

A spectral hole usually refers to a dip in theinhomogeneously broadened absorption spec-trum of many molecules when light irradia-tion alters the absorption frequencies of aportion of the resonant absorbers (23). In thesingle molecule regime, however, only theisolated absorption line of one center is pre-sent. When photochemical or photophysicalchanges occur as a result of optical excitation,the absorption line simply appears to vanish as

0. the resonance frequency of the single centershifts away from the laser frequency. Becauselaser-induced resonance frequency shifts formthe foundation for the spectral hole-burningprocess, I will continue to use the term "hole-burning" even for the single molecule case.

At first glance, the spontaneous spectraldiffusion of the last section seems similar to

D.0 single molecule hole-burning, and in somecases it may be hard to distinguish the two(51), because when a single moleculechanges resonance frequency, one may notknow which mechanism produced thechange. Because hole-burning is driven by

o.o the light and does not spontaneously occur

as the spectral diffusion of the last section,

SCIENCE * VOL. 265 * 1 JULY 1994

..

Am.

49

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

one must observe a dependence on the laserintensity to verify that hole-burning is oc-curring. The exact microscopic mechanismfor light-induced resonance frequency shiftsneeds further study and may be related to thegeneration of molecular internal vibrationalmodes during fluorescence emission or tononradiative decay of the excited state.

This effect was first convincingly ob-served for perylene in poly(ethylene) (24,25), where a clear increase in hole-burningrate could be observed at higher laser powerbecause the single molecules showed revers-ible hole-burning. Here, reversibility meansthe spontaneous return of the resonancefrequency to its original value after hole-burning, which allowed many measure-ments of the stochastic buming-time kinet-ic distribution for the same molecule (25).Irreversible hole-buming was also observed,and Fig. 6 illustrates this effect for a spectralregion containing four different molecules.Starting from the bottom, the first 10 tracesshow that the molecule near + 2 GHz isspectrally diffusing while the other threemolecules are relatively stable with a smallamount of spectral diffusion. After trace 10,the laser was deliberately brought into res-onance with the molecule at 0.5 GHz,which was observed to stop fluorescing dis-continuously after _28 s, indicating the hole-

2

EaC1L.

C59c

U

CD

2500

2000

1500

1000

500

-2 -1 0 1 2Laser frequency (GHz)

Fig. 6. Persistent spectral hole-burning of asingle perylene in polyethylene arbitrarily cho-sen out of four molecules in a 4.6-GHz scanrange (T = 1.5 K, 0 GHz detuning at 448.019nm). The burning event for the molecule at 0.5GHz (arrow) was executed after the tenth tracecounting from the bottom or in terms of time 29min after the first trace shown [Adapted from(24)].50

burning. Trace 11 was then recorded, whichshows that the molecule shifted its resonancefrequency outside of the laser scan range. Theremaining traces show that the molecule didnot return to its original frequency for at least15 min. The exact location of the new reso-nance frequency was unknown, but by anal-ogy with previous nonphotochemical hole-burning studies on large ensembles (53, 54),the shift could have been as large as 100cm-'. In recent work on the new system ofterrylene in hexadecane (55), well-definedtwo-state switching behavior was observedthat could be partially controlled by the laserirradiation.

In principle, single molecule hole-burn-ing is a controllable process that couldallow the modification of the transitionfrequency of any arbitrarily chosen mole-cule in the polymer host. This ability tochange leads naturally to the possibility ofoptical storage at the single molecule level(Fig. 7). For highest density, near-fieldoptical excitation with a 50-nm-diameterpulled fiber tip should be used, a processthat has demonstrated densities of 45 giga-bits per square inch with magneto-opticrecording (56). One can imagine a thinlayer of the material with a sufficientlybroad inhomogeneous line so that singleimpurity molecules are isolated and spreadover a large range of frequency space. Theresonance frequencies constitute the ad-dresses of all the bits to be written in asingle focal volume. A binary sequence ofl's and 0's can be produced by altering orignoring each single molecule absorption.

Single molecule optical storage, whilehighly speculative at present, provides sev-eral advantages and disadvantages. Amongthe advantages are (i) the extreme arealdensity (up to perhaps 10`3 bits per squareinch for a 50 nm by 50 nm storage volumeand a doping level of lo-3, giving 100 bits

Near-field fiber tip

Recording medium

,a ads E -aI

Fig. 7. Illustration of possible scheme for ultra-high density optical storage with the use oflight-induced spectral modification of singlemolecules and near-field optics. In each cross-hatched volume, many bits are recorded withthe use of single molecules at different posi-tions in the inhomogeneous line.

per spot), (ii) the increased signal fromeach molecule resulting from the use ofsub-diffraction-limited beams, and (iii) theease of access of different bits by the simpletuning of the laser. Some disadvantages arethe required low temperatures, stochastical-ly variable burning times, high doping lev-el, and variable frequency locations of theindividual bits from laser spot to laser spot.The last problem can be overcome in prin-ciple by "pre-formatting" the resonancefrequencies or perhaps even by careful de-sign of the inhomogeneous distribution. Inspite of these problems, optical modifica-tion of single molecule spectra not onlyprovides a unique window into the photo-physics and low-temperature dynamics ofthe amorphous state, but it also allows suchnovel optical storage schemes to be studiedand developed.

Correlation of Photons Emittedby a Single Molecule

The stream of photons emitted by a singlemolecule contains information about thesystem encoded in the arrival times of theindividual photons. Figure 8 schematicallyshows the time domain behavior of thephoton stream for a single molecule with adark triplet state, here taken to be penta-cene. While cycling through the singletstates of S0O->SO, photons are emitted untilintersystem crossing occurs. Because thetriplet yield is 0.5% (57), 200 photons areemitted on average before a dark periodthat has an average length equal to thetriplet lifetime, rT. This effect causes"bunching" of the emitted photons, asshown in the upper half of Fig. 8. Thesubsequent decay in the autocorrelation ofthe emitted photons for pentacene in p-ter-phenyl was reported by Orrit and Bemard(3), and this phenomenon has been used tomeasure distributions in the triplet yieldand `TT from molecule to molecule (40)resulting from distortions by the local na-noenvironment. Such correlation measure-ments can extract information on muchshorter time scales (down to microseconds)

200 Photons

/ime

X.1Fig. 8. Schematic of the photon emissionstream from a single molecule showing bunch-ing on the scale of the triplet lifetime (upperhalf) and antibunching on the scale of theinverse of the Rabi frequency (lower half).

SCIENCE * VOL. 265 * 1 JULY 1994

W119w~ r, 1 1 1

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

than those of the frequency scans describedabove; however, the dynamical process un-der study must be "stationary," that is, itmust not change during the many secondsneeded to record a valid autocorrelation.For terrylene in poly(ethylene), which hascomplex dynamics driven by TLSs in thepolymer, the amplitude fluctuations in thesingle-molecule fluorescence signal result-ing from resonance frequency shifts some-times cause a characteristic falloff in theautocorrelation that yields informationabout the TLS-phonon coupling (41, 52).

By contrast, in the nanosecond timeregime (lower half of Fig. 8), the emittedphotons from a single quantum system areexpected to show antibunching, whichmeans that the photons "space themselvesout in time;" that is, the probability for twophotons to arrive at the detector at thesame time is small. This is a quantum-mechanical effect, which was first observedfor sodium atoms in a low-density beam(58). For a single molecule, antibunching iseasy to understand: After photon emission,the molecule is definitely in the groundstate and cannot emit a second photonimmediately. A time on the order of theinverse of the Rabi frequency, X-1, mustelapse before the probability of emission ofa second photon is appreciable. Antibunch-ing for a single molecule was observed atIBM for pentacene in p-terphenyl (42),demonstrating that quantum optics experi-ments can be performed in solids and with asingle molecule. Of course, if more than onemolecule is emitting, the antibunching ef-fect as well as the bunching effect quicklydisappears because the molecules emit inde-pendently. Thus, high-contrast anti-bunch-ing is further proof that the spectral featuresare indeed those of single molecules.

Fig. 9. Examples of dispersed flu-orescence spectra showing vi-brational mode frequencies andintensities for two molecules ofterrylene in poly(ethylene). (A)Molecule probably located in acrystalline region (laser wave-length = 577.19 nm). (B) Mole-cule probably located in an amor-phous region (laser wavelength =

572.46 nm).p

c

Vibrational Modes ofSingle Molecules

Until recently, all SMS studies used thetotal fluorescence excitation technique inwhich all long-wavelength-shifted photonspassing through a long-pass filter are record-ed. With the use of a grating spectrographand a charge-coupled device array detector,vibrationally resolved emission spectra fromsingle molecules both in crystals (43) and inpolymers (44, 45) have been obtained.Such experiments may also be regarded as

resonance Raman studies, because the laseris in resonance with the 0-0 electronictransition, and even-parity vibrationalmodes of the ground state are detected bythe measurement of the shift between thelaser wavelength and the wavelength of theemission peak. The ability to examine vi-bronic or vibrational features of individualabsorbers can generate much more detailabout the identity of the absorber and thenature of the interactions with the nanoen-

vironment which produce shifts or intensitychanges in the vibrational spectrum.

Typical dispersed fluorescence spectrafor two different single molecules of ter-rylene in poly(ethylene) at 1.6 K are shownin Fig. 9. In addition to small (acm-)shifts and intensity changes of variousmodes from molecule to molecule, twodifferent classes of spectra were observed(44), as shown in the upper and lower partsof the figure. After considering various pos-

sibilities, the investigators concluded thatthe upper type of spectrum resulted from a

terrylene molecule on the surface of or

inside the crystalline lamellae of the poly-mer, while the other spectrum resulted froma single terrylene located in an amorphousregion. Such results demonstrate the wealth

300 500 1200 1400

Vibrational wave number (cm-1)

SCIENCE * VOL. 265 * 1 JULY 1994

of spectroscopic detail that can be obtainedfrom individual molecules to probe trulylocal aspects of the structure of amorphoussolids. For example, the lowest frequencymode in Fig. 9 is a long axis ring expansionof the molecule; the shift to lower energy inthe amorphous site can be understood asresulting from the reduced local density(greater free volume) compared to the crys-talline site. Such measurements will stimu-late new theoretical calculations of thespectral changes that result from specificlocal distortions for comparison with thesingle molecule spectra.

Magnetic Resonance of aSingle Molecular Spin

Historically, the standard methods of elec-tron paramagnetic resonance and nuclearmagnetic resonance have been limited insensitivity to about 108 electron spins andabout 1015 nuclear spins, respectively, duechiefly to the weak interaction between theindividual spins and the radio-frequencyelectromagnetic fields used for detection.Using a combination of SMS and opticallydetected magnetic resonance (ODMR),two research groups have independentlyobserved the magnetic resonance transitionof a single molecular spin (37, 38) using thepentacene in p-terphenyl model system. Inessence, ODMR allows higher sensitivitybecause the weak spin transition is effec-tively coupled to a much stronger opticaltransition with oscillator strength near uni-ty. The method involves selecting a singlemolecule with the laser as described aboveand monitoring the intensity of opticalemission (here, fluorescence from the firstexcited singlet state) as the frequency of amicrowave signal is scanned over the fre-quency range of the triplet spin sublevelsplittings (59). Because the emission rate isdependent on the overall lifetime of thetriplet (bottleneck) state, the emission rateis affected when the microwave frequency isresonant with transitions among the tripletspin sublevels.

Figure 10 shows examples of the 1480-MHz magnetic resonance transition amongthe T. to T, triplet-spin sublevels at 1.5 Kfor a single molecule of pentacene in p-ter-phenyl, where the signal plotted is thechange in the fluorescence emission rate asa function of the applied microwave fre-quency (37). Traces A and B show the lineshapes at large N when many pentacenemolecules are pumped by the laser, where01 and 02 refer to the specific substitution-al sites for pentacene in the host. Traces Cto G show the line shapes for four differentsingle molecules. As is the case for large N,the line shape of the ODMR transition fora single spin is broadened by hyper-fineinteractions with nearby proton nuclear

51

.,._ K IM ---'El M111 0

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

spins. This broadening occurs because manydifferent configurations of the protons in themolecule are sampled on the time scale ofthe measurement of the triplet state transi-tion. In contrast, in the large N experimentan ensemble average is measured rather thana time average.

These observations open the way for avariety of new studies of magnetic interac-tions in solids at the level of a singlemolecular spin, such as the use of externalmagnetic fields to reduce proton spin flipsand hence the hyper-fine broadening. Al-ready single spin coherence has been ob-served (39), because no matter what timethe molecule enters the triplet state, thestart of the microwave irradiation providesa zero-time reference so that the signalsfrom subsequent entries into the triplet canadd coherently. In future work, the proper-ties of amorphous organic materials may bestudied in greater detail, as the selection ofa single molecular spin removes all orienta-tional anisotropy as well as all inhomoge-neous broadening. Imaging on the spatialscale of a single molecule is possible with asufficiently large magnetic field gradient.The power of magnetic resonance in gener-al in the study of fine and hyper-fine inter-actions, local structure, and molecularbonding can now be enhanced with thesefirst demonstrations of useful sensitivity inthe single spin regime.

A Verylarge N, 01B Very large N. 020

-50C0

So-

IL

0c

0

ID

CD'acc

aI

-300

-25c

1475 1480 1485 1490 1495 1500

Microwave frequency (MHz)Fig. 10. Single-spin magnetic resonance forpentacene in p-terphenyl at 1.5 K. The fluores-cence detected magnetic resonance (FDMR)signal is shown as a function of the microwavefrequency applied to the sample. (A and B) TheTx to T. transition for a large number of mole-cules for sites 01 and 02- (C to G) A selection ofsingle-molecule magnetic resonance lineshapes [adapted from (37)].

52

Outlook

The attainment ofSMS in solids opens up anew frontier of single-absorber experimentsin which the measured properties of theabsorbing center are not averaged overmany "equivalent" absorbers. The signifi-cance of such experiments is fourfold. First,the properties of a single absorber are mea-sured without ensemble averaging, whichmeans that tests of specific theoretical mod-els are much stronger. Second, the sensitiv-ity to specific properties of the nanoenvi-ronment such as the local phonon modesand the true local fields is extremely high.This sensitivity means, for example, thatthe identity of the mysterious two-levelsystems in amorphous materials may finallybe determined. Third, it provides a windowinto the spectral hole-burning process on amolecule by molecule basis. Thus, the ex-act local coupling through which opticalpumping of a single molecule gives rise tochanges in the nanoenvironment that shiftthe resonance frequency may be studied.Fourth, this regime is essentially unexplored,which means that surprises and unexpectedphysical effects can occur (such as the obser-vation of spectral diffusion in a crystal).

While as a general technique SMS is notapplicable to all molecular impurities, it canbe applied to the large number of absorbingmolecules (and perhaps ions) in solids thathave zero-phonon transitions, reasonable ab-sorption strength, and efficient fluorescence.The detectability of the resulting single-centersignal, which ultimately depends on the spe-cific sample and weak or absent spectral hole-burning, must be evaluated in each case.Single-molecule excitation spectra should beobservable at higher and higher temperaturesif the concentration and background are bothreduced sufficiently. One method for makingthis reduction may be to use near-field exci-tation (10, 11) to reduce the scattering vol-ume and increase the single center signal.

Other future experiments may be con-templated. Detailed study of the spectraldiffusion process in crystals and polymerswill help eventually to identify the actualmicroscopic nature of the two-level sys-tems. Nonlinear spectroscopy to measurethe ac Stark effect for a single isolatedmolecule may also be performed. With theproper choice of lifetimes, quantum jumpsand other processes observed for single ionsin vacuum electromagnetic traps would beexpected to become directly observable.The door is also open to true photochemi-cal experiments on single absorbers at lowtemperatures and even the possibility ofoptical storage with the use of single mole-cules. Future efforts to increase the numberof probe-host couples that allow single-molecule spectra will lead to an even largerarray of possibilities.

One novel experiment that is now pos-sible would be to use the emission from asingle molecule as a light source of sub-nanometer dimensions for near-field opticalmicroscopy (60). Of course, this experi-ment would involve the technical difficultyof placing the single molecule at the end ofa pulled fiber tip. Another possibility wouldbe to perform cavity quantum electrody-namics studies with a single molecule in asolid placed in a high-Q (quality factor, orthe center frequency of the cavity/the line-width of the cavity resonance) cavity. In allcases, improvements in SNR would be ex-pected to open up a new level of physicaldetail and possibly new applications for thismethod. Because this field is relatively new,the possibilities are only limited at presentby the imagination and the persistence ofthe experimenter and the continuing scien-tific interest in the properties of singlequantum systems in solids.

REFERENCES AND NOTES

1. W. E. Moerner and L. Kador, Phys. Rev. Lett. 62,2535 (1989).

2. L. Kador, D. E. Horne, W. E. Moerner, J. Phys.Chem. 94,1237 (1990).

3. M. Orrit and J. Bernard, Phys. Rev. Lett. 65, 2716(1 990).

4. W. E. Moerner and Th. Basch6, Angew. Chem.105, 537 (1993); Angew. Chem. tnt. Ed. EngI. 32,457 (1993).

5. M. Orrit, J. Bernard, R. Personov, J. Phys. Chem.97,10256 (1993).

6. W. M. Itano, J. C. Bergquist, D. J. Wineland,Science 237, 612 (1987), and references therein.

7. F. Diedrich, J. Krause, G. Rempe, M. 0. Scully, H.Walther, IEEE J. Quantum Electron. 24, 1314(1988), and references therein.

8. H. Dehmelt, W. Paul, and N. F. Ramsey, Rev.Mod. Phys. 62, 525 (1990).

9. G. Binnig and H. Rohrer, ibid. 59, 615 (1987).10. E. Betzig and R. J. Chichester, Science 262, 1422

(1993).11. W. P. Ambrose, P. M. Goodwin, J. C. Martin, R. A.

Keller, Phys. Rev. Lett. 72, 160 (1994).12. P. D. Reilly and J. L. Skinner, ibid. 71, 4257

(1993).13. W. P. Ambrose, Th. Basch6, W. E. Moerner, J.

Chem. Phys. 95, 7150 (1991).14. W. E. Moerner, J. Lumin., in press.15. A. M. Stoneham, Rev. Mod. Phys. 41, 82 (1969).16. K. K. Rebane, Impurity Spectra of Solids (Plenum,

New York, 1970), p. 99.17. D. A. Wiersma, Adv. Chem. Phys. 47, 421 (1981).18. See Laser Spectroscopy of Solids, vol. 49 of

Springer Topics in Applied Physics, W. M. Yenand P. M. Seizer, Eds. (Springer, Berlin, 1981).

19. W. E. Moerner and L. Kador, Anal. Chem. 61,1217A (1989).

20. W. E. Moerner, NewJ. Chem. 15,199 (1991).21. K. K. Rebane and 1. Rebane, J. Lumin. 56, 39

(1993).22. H. de Vries and D. A. Wiersma, J. Chem. Phys. 72,

1851 (1980).23. See Persistent Spectral Hole Burning: Science

and Applications, vol. 44 of Springer Topics inCurrent Physics, W. E. Moerner, Ed. (Springer,Berlin, 1988).

24. Th. Basch6, W. P. Ambrose, W. E. Moerner, J.Opt. Soc. Am. B 9, 829 (1992).

25. Th. Basch6 and W. E. Moerner, Nature 355, 335(1992).

26. M. Orrit, J. Bernard, A. Zumbusch, R. I. Personov,Chem. Phys. Lett. 196, 595 (1992).

27. T. Plakhotnik, W. E. Moerner, T. Irngartinger, U. P.Wild, Chimia 48, 31 (1994).

SCIENCE * VOL. 265 * 1 JULY 1994

C ,. Molecule

)o'. -10

)o .. ..Molecule III *. e"..Or,.\* f..0.........¶. *00

to D ..... Molecule lil

to , ~~~~Molecule lV

/ o*o. 0 ., * 0*S**.".

1o -G Molecule IV(! . .(high resolution)

-1 01 --- 'W -.111 WIMM.

-400

-40C

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

..SDa

28. W. E. Moerner and T. P. Carter, Phys. Rev. Lett.59, 2705 (1987).

29. T. P. Carter, M. Manavi, W. E. Moerner, J. Chem.Phys. 89, 1768 (1988).

30. W. E. Moerner and W. P. Ambrose, Phys. Rev.Lett. 66, 1376 (1991).

31. H. de Vries and D. A. Wiersma, J. Chem. Phys. 69,897 (1978).

32. F. G. Patterson, H. W. H. Lee, W. L. Wilson, M. D.Fayer, Chem. Phys. 84, 51 (1984).

33. U. P. Wild, F. Guttler, M. Pirotta, A. Renn, Chem.Phys. Lett. 193 451 (1992).

34. M. Croci, H.-J. Muschenborn, F. Guttler, A. Renn,U. P. Wild, ibid. 212, 71 (1993).

35. M. Pirotta et al., ibid. 208, 379 (1993).36. W. P. Ambrose and W. E. Moerner, Nature 349,

225 (1991).37. J. Kohler etal., ibid. 363, 242 (1993).38. J. Wrachtrup, C. von Borczyskowski, J. Bernard,

M. Orrit, R. Brown, ibid., p. 244.39. J. Wrachtrup, C. von Borczyskowski, J. Bernard, M.

Orrit, R. Brown, Phys. Rev. Lett. 71, 3565 (1993).40. J. Bernard, L. Fleury, H. Talon, M. Orrit, J. Chem.

Phys. 98, 850 (1993).

41. A. Zumbusch, L. Fleury, R. Brown, J. Bernard, M.Orrit, Phys. Rev. Lett. 70, 3584 (1993).

42. Th. Basch6, W. E. Moerner, M. Orrit, H. Talon,ibid. 69, 1516 (1992).

43. P. Tchenio, A. B. Myers, W. E. Moerner, J. Phys.Chem. 97, 2491 (1993).

44. _ , Chem. Phys. Lett. 213, 325 (1993).45. A. B. Myers, P. Tchenio, W. E. Moerner, J. Lumin.,

in press.46. F. Guttler, T. Irngartinger, T. Plakhotnik, A. Renn,

U. P. Wild, Chem. Phys. Lett. 217, 393 (1994).47. J. Friedrich and D. Haarer, in Optical Spectrosco-

py of Glasses, 1. Zschokke, Ed. (Reidel, Dor-drecht, the Netherlands, 1986), p. 149.

48. F. Guttler et al., J. Lumin. 56, 29 (1993).49. See Amorphous Solids: Low-Temperature Proper-

ties, vol. 24 of Springer Topics in Current Physics,W. A. Phillips, Ed. (Springer, Berlin, 1981).

50. G. Zumofen and J. Klafter, Chem. Phys. Lett. 219,303 (1994).

51. P. Tch6nio, A. B. Myers, W. E. Moerner, J. Lumin.56, 1 (1993).

52. L. Fleury, A. Zumbusch, M. Orrit, R. Brown, J.Bernard, ibid., p. 15.

53.

54.

55.

56.57.

58.

59.

60.

61.

J. M. Hayes, R. P. Stout, G. J. Small, J. Chem.Phys. 74, 4266 (1981).R. Jankowiak and G. J. Small, Science 237, 618(1987).W. E. Moerner, T. Plakhotnik, T. Irngartinger, M.Croci, V. Palm, U. P. Wild, J. Phys. Chem., inpress.E. Betzig et al., Appl. Phys. Lett. 61, 142 (1992).H. de Vries and D. A. Wiersma, J. Chem. Phys. 70,5807 (1979).H. J. Kimble, M. Dagenais, L. Mandel, Phys. Rev.Lett. 39, 691 (1977).Effectively, the triplet state of the molecule is formedby two unpaired electrons with spin parallel to eachother leading to a net electronic spin of 1. In zeroexternal magnetic field, the triplet state splits intothree spin sublevels because of the anisotropy ofthe molecular electronic wave function and the mag-netic structure of the nearby environment.K. Lieberman, S. Harush, A. Lewis, R. Kopelman,Science 247, 59 (1990).thank W. P. Ambrose, Th. Basch6, L. Kador, A.

B. Myers, P. Tchenio, and U. P. Wild for fruitfulcollaborations.

~~~...- ^. ... .......... .........sn| \ \ _ Sss'S sy xy? sysfayx ~sk ? t. ss.s.f.x. .n..

Stimulation of GAL4 DerivativeBinding to Nucleosomal DNA by

the Yeast SWI/SNF ComplexJacques Cate, Janet Quinn, Jerry L. Workman,

Craig L. Peterson*

The SWI/SNE protein complex is required for the enhancement of transcription by manytranscriptional activators in yeast. Here it is shown that the purified SWI/SNE complex iscomposed of 10 subunits and includes the SW11, SW12/SNF2, SW13, SNF5, and SNF6gene products. The complex exhibited DNA-stimulated adenosine triphosphatase(ATPase) activity, but lacked helicase activity. The SWI/SNE complex caused a 10- to30-fold stimulation in the binding of GAL4 derivatives to nucleosomal DNA in a reactionthat required adenosine triphosphate (ATP) hydrolysis but was activation domain-inde-pendent. Stimulation of GAL4 binding by the complex was abolished by a mutant SW12subunit, and was increased by the presence of a histone-binding protein, nucleoplasmin.A direct ATP-dependent interaction between the SWI/SNE complex and nucleosomal DNAwas detected. These observations suggest that a primary role of the SWI/SNE complexis to promote activator binding to nucleosomal DNA.

The yeast SWI1, SWI2/SNF2, SWI3,SNF5, and SNF6 gene products are requiredfor the induced expression of a large set ofgenes (1). Furthermore, SWI/SNF productsare required for the enhancement of tran-scription by several gene-specific activatorproteins in yeast, such as GAL4 (2), Dro-sophila ftz (2), mammalian glucocorticoidJ. C6tb and J. L. Workman are in the Department ofBiochemistry and Molecular Biology and the Centerfor Gene Regulation, The Pennsylvania State Univer-sity, University Park, PA 16802, USA. J. Quinn and C.L. Peterson are in the Program in Molecular Medicineand the Department of Biochemistry and MolecularBiology, University of Massachusetts Medical Center,Worcester, MA 01605, USA.*To whom correspondence should be addressed.

and estrogen receptors (3), and LexA-GAL4and LexA-Bicoid fusion proteins (4). TheSWI/SNF gene products function as compo-

nents of a large multi-subunit protein com-plex of approximately 2 x 106 daltons (5,6). One activity of this complex is to asso-ciate with the mammalian glucocorticoidreceptor (3). These observations suggestedthat homologs of the SWI/SNF genes wouldbe present in Drosophila and mammals (3).Candidate homologs of the SW12 gene havebeen identified in Drosophila (7), mouse (8),and human (9, 10). Protein chimeras be-tween either BRG1 (one of two putativehuman homologs) or brahma (brm, the pu-tative Drosophila homolog) and SWI2 arefunctional in yeast, which suggests thatthese relatives are functional homologs (9,I 1). Gel filtration data suggest that theBRG1 protein may also be a subunit of alarge protein complex (9).A current hypothesis of how the SWI/

SNF complex facilitates activator functionsuggests that the complex antagonizes chro-matin-mediated transcriptional repression.The relation between the SWI/SNF com-plex and chromatin structure was suggestedbecause mutations in genes that encodechromosomal proteins alleviate the pheno-types of swi and snf mutants. Mutations thatinactivate the SIN] gene, which encodes aputative nonhistone chromatin compo-nent, or in the SIN2 gene, which encodeshistone H3, alleviate the defects in growthand in transcription caused by mutations in

Table 1. Purification of the SWI/SNF complex was followed by protein immunoblots probing for theSWI2-HA-6HIS fusion protein. Similar levels of purification were obtained in at least four preparations.

Volume Concentration pTrotaln Specific(ml) (mg/mi) (mg) (units*)

WOE 130 21.5 2800 1Ni2+ eluate 43 1 43 65Mono 0 2 0.35 0.7 3,965Superose 1.5 0.02 0.03 91,195

*One unit is equivalent to the amount of SW12-HA-6HIS fusion protein in 100 >g of whole cell extract as measuredby immunoblots. The overall yield is estimated to be approximately 25 percent.

SCIENCE * VOL. 265 * 1 JULY 1994 53

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

Examining Nanoenvironments in Solids on the Scale of a Single, Isolated Impurity MoleculeW. E. Moerner

DOI: 10.1126/science.265.5168.46 (5168), 46-53.265Science 

ARTICLE TOOLS http://science.sciencemag.org/content/265/5168/46

REFERENCES

http://science.sciencemag.org/content/265/5168/46#BIBLThis article cites 52 articles, 4 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

1994 by the American Association for the Advancement of Science

on Septem

ber 28, 2020

http://science.sciencemag.org/

Dow

nloaded from