Scanning Electron Microscopy Scanning Electron Microscopy

Volume 1986 Number 2 Article 2

6-30-1986

High Resolution Scanning Auger Microscopy of Mineral Surfaces High Resolution Scanning Auger Microscopy of Mineral Surfaces

M. F. Hochella Jr. Stanford University

A. M. Turner Perkin-Elmer Corporation

D. W. Harris Perkin-Elmer Corporation

Follow this and additional works at: https://digitalcommons.usu.edu/electron

Part of the Biology Commons

Recommended Citation Recommended Citation Hochella, M. F. Jr.; Turner, A. M.; and Harris, D. W. (1986) "High Resolution Scanning Auger Microscopy of Mineral Surfaces," Scanning Electron Microscopy: Vol. 1986 : No. 2 , Article 2. Available at: https://digitalcommons.usu.edu/electron/vol1986/iss2/2

This Article is brought to you for free and open access by the Western Dairy Center at DigitalCommons@USU. It has been accepted for inclusion in Scanning Electron Microscopy by an authorized administrator of DigitalCommons@USU. For more information, please contact digitalcommons@usu.edu.

SCANNING ELECTRON MICROSCOPY /1986/II (Pages 337-349) SEM Inc., AMF O'Hare (Chicago), IL 60666-0507 USA

0586-5581/86$1.00+05

HIGH RESOLUTION SCANNING AUGER MICROSCOPY OF MINERAL SURFACES

M. F. Hochella, Jr.*, A. M. Turner** and D. W. Harris**

*Department of Geology, Stanford University, Stanford, CA 94305 USA **Physical Electronics Laboratories, Perkin-Elmer Corporation

1161-C San Antonio Rd., Mountain View, CA 94043 USA

(Received for publication March 04, 1986: revised paper received June 30, 1986)

Abstract

There are a number of cases where scanning Auger microscopy can be used to determine the near-surface composition of minerals with ex­tremely high lateral resolution. This involves collecting Auger spectra with reasonable signal to noise ratios without encountering significant beam induced charging or surface degradation, even if the beam is impinging on a grain less than lµm in diameter. We typically use a 3 keV beam with less than 10 nA beam current on a sam­ple surface that is tilted (to increase backscat­tered and secondary electron emission efficiency) and relatively flat. To further minimize surface degradation, vacuum levels are kept high and the beam is rastered or defocused whenever possible. The Auger spectra of minerals can be used to study modification of surfaces due to geochemical influ­ences or to identify submicron grains if the near­surface composition is representative of the bulk composition. Also, high lateral resolution step scans can be performed across sharp interfaces be­tween two grains, allowing for short-range studies of solid-solid interactions in rocks at grain boundaries. We also report on preliminary attempts to chemically quantify Auger peak inten­sities for silicate minerals. Measurements of peak-to-peak heights for oxygen and silicon lines for eight silicate minerals of well-known composi­tion indicate that Auger sensitivity factors can vary significantly with 0/Si ratio.

KEY WORDS: scanning Auger microscopy, high reso­lution scanning Auger microscopy, step scans, mineral surfaces, submicron grains, pyrite, lizar­dite, garnet.

*Address for correspondence: Michael F. Hochella, Jr., Dept. of Geology, Stan­ford University, Stanford,CA 94305 USA

Phone No. (415) 723-1070

337

Introduction

Chemical microanalysis has played an inte­gral role in the advancement of the major geolo­gic subfield of petrology, the study of the ori­gin and evolution of rocks. This is because a large part of petrology not only depends on the minerals in a rock and their distribution and re­lationship to one another (which can commonly be determined optically with a petrographic micro­scope), but also on the specific chemical compo­sition of the constituent phases. Even the chemi­cal variation within a single mineral can be cru­cial in unravelling the history of a rock. The ability to perform chemical microanalysis of min­erals in rocks was not possible until well after the initial development of the first electron probe microanalyzer (EPMA) by Castaing and Guini­er in the late 1940's (see Castaing, 1960). Es­pecially over the last decade, EPMA has evolved into a microanalytic technique which is highly quantitative and relatively easy and convenient to use. Scanning electron microscopy (SEM) and scanning transmission electron microscopy (STEM), when used in conjunction with energy dispersive x-ray spectroscopy (EDX or EDS), provide addi­tional chemical microanalysis tools for geologic research. SEM/EDS is relatively easy to use and provides semi-quantitative to quantitative analy­sis of mineral grains down to several microns in size. STEM/EDS, while requiring more advanced operator training and more difficult sample pre­paration procedures, allows for semi-quantitative analysis with lateral resolution in the neighbor­hood of O.Olµm.

Auger electron spectroscopy and scanning Au­ger microscopy (AES/SAM) are relatively new in the chemical microanalysis field. These tech­niques, first developed for general laboratory use in the late 1960's (see Palmberg et al., 1969; Wells and Bremer, 1969; and MacDonald, 1970), are typically used as surface sensitive analytic probes for conducting materials measur­ing compositions to a depth of 5-50~. However, when a finely focused primary electron beam is utilized, the analyzed volume can be several or­ders of magnitude smaller than with EPMA or SEM/ EDS. Therefore, the technique can be used as a near-surface analytic tool with exceptionally high lateral resolution, or if the near-surface composition of a sample is representative of its

M. F. Hochella, Jr., A. M. Turner and D. W. Harris

bulk composition, it can be used as a high spatial resolution microprobe.

Venables et al. (1976) were among the first to use SAM as an ultra-high resolution microprobe on conducting materials, and the literature is extensive on the uses of high spatial resolution SAM on conductors and semi-conductors (see, e.g., Holm, 1982; Olson et al., 1983; and Harris, 1984, for reviews). The growth in the use of AES/SAM on insulating surfaces has not been nearly so rapid. This is primarily due to charging prob­lems, and also because of the surface degrada­tional effects of the electron beam which can easily affect Auger spectra due to the surface sensitive nature of the technique. Nevertheless, in many cases these problems can be minimized or overcome, and there are several important low to moderate spatial resolution AES/SAM studies on mineral surfaces. Poppa and Elliot (1971) were among the first to use AES on minerals in a study of the (001) surface of muscovite which was beina evaluated as a vapor deposition substrate. Elec: tron beam degradational effects on quartz surfaces have been studied by Carriere et al. (1973) and Carriere and Lang (1977). The first use of AES on minerals in geologic studies concerned the surface characterization of lunar dust (Morrison et al., 1970; Connell et al., 1971; Gold et al., 1974, 1975; Grant et al., 1974; and Baron et al., 1977). More recent studies include those of Per­ry et al. (1982a,b; 1983a,b; 1984a,b) who have investigated metal ion interaction on and weather­ing of feldspar and sulfide surfaces, Mucci et al. (1985) and Mucci and Morse (1985), who have used AES to study adsorbed layers on carbonates after reaction with seawater, and Bisdom et al. (1985) who have used SAM and SEM/EDS to study soil sam­ples. Remand et al. (1981, 1982, 1983, 1985) have used moderately high spatial resolution SAM and other techniques to investigate how surface preparation and air exposure can affect optical measurements (color and reflectance) on sulfide minerals in polished ore cross sections. Also, White and Yee (1985) have used AES, among other techniques, to study the surfaces of ferrous iron-containing minerals in contact with various aqueous solutions, and Mackinnon and Mogk (1985) and Mogk et al. (1985) have used SAM to study the near-surface composition of particulates collec­ted from the stratosphere.

The use of high spatial resolution SAM as a microprobe for rocks has only been recently ex­plored (Hochella, 1985; Hochella et al., 1985, 1986). In these studies, it has been demonstrated that SAM can be used to identify submicron grains in rocks while eliminating charging and minimizing electron stimulated surface degradational effects. The same techniques can be used to perform surfJce studies with very high lateral resolution on insu­lators. In this paper, we will review sample handling of rock specimens prior to Auger analysis and the instrumental conditions used to control charging and surface degradation. We will then briefly discuss important factors involving the chemical quantification of Auger intensities for silicates. Finally, we will present several pre­liminary experimental results which demonstrate the ability of high spatial resolution SAM to give important insight into certain geologic problems.

338

Experimental Techniques

The scanning Auger microscope used in this study was a Perkin-Elmer Phi 600. The 600 de­sign is based on a vertically mounted electromag­netically focused electron beam column which is coaxial with a single pass cylindrical mirror electron energy analyzer. The electron optics are sufficient to produce a beam as small as 0.035µm in diameter at 10 keV and 0.05 nA beam current. Auger spectra presented in this paper were collected with a 3 keV primary beam with currents between approximately 1 and 20 nA pro­ducing a beam diameter at or below 0.3µm in di­ameter. The SEM micrographs were taken in the Phi 600 SAM of the same uncoated specimens used for surface analysis. The beam currents used for SEM work were well below 1 nA, resulting in beam diameters down to 0.05µm. Vacuum during our ex­periments was maintained in the 10-0 -10- 10 torr range depending on the outgassing of individual samples.

Sample preparation of rock specimens for SAM analysis can be relatively simple if the sample is free of gross organic contamination. We have found that it is appropriate to introduce dry rock chips (which have only been handled with clean tools) directly into the ultra-high vacuum environment of the instrument for observation. Considerable outgassing may occur if the rock contains clay and zeolite minerals. If the sam­ple needs to be degreased, cleaning in even di­lute detergent solutions should be done with ex­treme caution. For example, we have determined that sodium and phosphorus can be sorbed in small quantities onto the surfaces of calcite grains in limestone after the sample has been rinsed in a dilute alkali phosphate detergent. Boiling in acetone and then ethanol often serves to ade­quately degrease unclean rock specimens, leaving behind only a thin contaminant layer of residual solvent (Stephenson and Binkowski, 1976, and Ho­chella et al., 1986). However, modification of the mineral surface chemistry itself is always a possibility when any cleaning procedure is used.

Modern Auger instruments can typically ac­commodate samples which are up to a cubic centi­meter in volume. These samples are most easily affixed to the flat tops of sample holders with screws or clips. Small rock fragments or grains can be mounted by pressing them into indium foil attached to a sample holder. Fractured, sawed, or polished surfaces can be analyzed, although sawing can modify surface composition due to pos­sible contamination from the blade and contamina­tion or leaching by the cutting lubricant, even if water is used. It is well known that polishing typically modifies surface composition to some degree, and although polished surfaces are occa­sionally important to analyze (see below), the surface chemistry will, at best, only be roughly representative of the bulk.

Unless a mineral surface has been exposed by fracturing under UHV conditions, it will com­monly show excess carbon and oxygen in the Auger spectrum compared to any bulk measurement. We occasionally also see very small amounts of ni­trogen, and we presume that hydrogen is also present as molecular or dissociated H20 and/or

High resolution SAM of mineral surfaces

hydrocarbons. This contaminant layer can be easily removed by a few seconds of ion sputter­ing. Hochella et al. (1986) have demonstrated on feldspar surfaces that very small amounts of sputtering will not interfere with semi-quanti­tative chemical analysis due to differential sputtering effects (Pantano et al., 1975, and Pantano, 1981).

Argon ion sputtering was performed with a differentially pumped ion gun (analytic chamber pressure in the 10-s torr range) at between 1 and 4 keV accelerating potential. The angle of the ion beam to the sample normal was between 60 and 70°. Calibrated sputtering rates in our sys­tem under these conditions at 4 keV are 100~/min­ute for Si0 2.

Sample charging is always the most prevalent problem when performing Auger analyses of mineral surfaces. We have been relatively successful at reducing charging problems using the following guidelines: 1) beam voltage typically 3 keV; 2) beam currents below 10 nA, and sometimes near 1 nA, resulting in_beam current densities less than a few amps/cm2; 3) the angle between the sample surface normal and the primary beam be­tween 40 and 60° to increase backscattered and secondary electron emission efficiency; and 4) analyzing relatively smooth sample surfaces whenever possible. Also, a surface that is not charging under certain beam conditions may charge badly under the same conditions after sputtering. Exposing the sputtered surface to air for a few seconds, and thereby reestablishing a thin con­taminant surface layer, often alleviates the charging problem.

Electron beam damage of the mineral surface, including electron stimulated desorption and ad­sorption (see, e.g., Madey and Yates, 1971; Car­riere and Lang, 1977; and Pantano and Madey, 1981), and electric field and thermally induced ion migration or breakdown (Dawson et al, 1978; Gossink et al., 1980; Yau et al., 1981; Clark et al., 1979; Storp, 1985), have been minimized by following the suggestions of Levenson (1982) and references therein. These precautions include maintaining the highest possible vacuum during analysis, using the lowest beam currents and ac­celerating voltages necessary to obtain adequate signal to noise ratios, and rastering and/or de­focusing the beam whenever possible.

Chemical Quantification of Auger Line Intensities for Minerals

Palmberg (1973) was among the first to review all of the major fundamental principles of quanti­tative AES. Reviews and discussion of the advancements that have been made in this field can be found in, e.g., Powell (1980), Prutton (1982), Seah (1983), and references therein.

Unfortunately, chemical quantification using AES continues to lag behind the present state of EPMA quantitative analysis. Part of the reason is an incomplete understanding of the variations in electron escape depths (see Powell, 1984, and references therein) and backscattering (see, e.g., Ichimura et al., 1983) as a function of the sample being probed. Other less fundamental, but

339

nevertheless important, aspects of quantitative AES remain partially unresolved. These include the effects of surface roughness (deBernardez et al., 1984; Wehbi and Roques-Carmes, 1985), the effects of electron beam induced surface damage (reviewed by Levenson, 1982), and anisotropic Au­ger emission due to crystallographic orientation (Armitage et al., 1980, and references therein; Le Gressus et al., 1983).

Semi-quantitative to quantitative chemical analysis of near-surface regions with AES is still performed most readily with standards that are similar to the unknown (see, e.g., Seah, 1983, and references therein). Elemental sensitivity factors (Davis et al., 1976; Mroczkowski and Lichtman, 1983; and Payling, 1985) can typically provide no better than semi-quantitative results, but they are very easy and convenient to use.

Semi-quantitative and quantitative analysis of mineral surfaces have rarely been performed with AES. We are currently developing a set of sensitivity factors for the silicate minerals, by far the most abundant mineral group in the Earth's crust. To this point, we have carefully measured Auger peak-to-peak heights from eight well-known silicate microprobe standards obtained from the C. M. Taylor Corporation. The standards were de­greased, then fractured in air and stored in etha­nol until insertion into the instrument. The fractured surfaces were analyzed after very light sputtering to remove the thin contaminant layer due to air and ethanol exposure. Hochella et al. (1986) have shown that this treatment does not significantly affect the near-surface composition of alkali feldspars, and presumably similar sili­cates used in this study. Sensitivity factors de­rived from our quartz standard at 3 keV (using the 0 KL2,3L2,3 and Si L3M2,3M2,3 lines) are nearly identical to those given for Si02 in Davis et al. (1976). Using these quartz sensitivity factors (S0 = 0.50, S . = 0.36) to estimate the atomic 0/Si ratio fo? 7all of the other silicate standards from their Auger spectra, we have obtained the de­viation of the apparent atomic 0/Si ratio from the known or actual 0/Si ratio (Fig. 1). The devia­tion from the ideal line gets progressively great­er as 0/Si increases from that of quartz. This implies that the Auger emission current for oxygen and/or silicon is changing and/or that there is a systematic change in the Auger peak shapes from standard to standard. The forms of both the Si and O differentiated Auger peaks show a remarkable consistency over this range of standards suggest­ing that the deviation from the ideal line is due to more than peak shape changes. The Auger emis­sion current is dependent on a number of factors, but because the instrument settings and beam con­ditions were not varied, and because we are deal­ing with elemental ratios, we only need to con-· sider factors specific to individual constitu­ents, namely the effects of surface topography (see Holloway, 1978), escape depth, ionization cross section for the initial vacancy, or the probability that the excited atom will decay through the Auger transition of interest. The surface topography of the analysis area for each standard was smooth and flat. Unfortunately, it is not yet clear whether the escape depth curves

M. F. Hochella, Jr., A. M. Turner and D. W. Harris

5

en 0 4 «i ::,

u3 ell

2

2

1/

Qt

3 4 5 6 7 apparent 0/Si

Fig. 1. Apparent oxygen to silicon ratios, which have been calculated using the sensitivity fac­tors derived from the O KL2,3L2,3 and Si L3M2,3M2,3 lines for pure quartz, for eight silicate mineral standards plotted against the actual oxygen to silicon ratios. The 1/1 line (solid) is drawn to show the deviation from ide­ality. The standards, obtained from the C. M. Taylor Corporation, are quartz (Qt), orthoclase (Or), albite (Ab), wollastonite (Wo), diopside (Di), forsterite (Fo), sphene (Sp), and kyanite (Ky). will vary significantly over this range of sili­cate composition. At least a part of the varia­tion may come from changes in the ionization cross section and the probability of the Auger decay process. If the energy of the primary beam is constant, these factors will depend on the chemical environment of the atom of interest. Although silicon is coordinated to 4 oxygens for all of these silicate standards, the average num­ber of silicon atoms surrounding each oxygen varies directly with the 0/Si ratio. In sili­cates, neighboring silicon atoms have the largest effect on the electronic structure of the oxygen because the Si-0 bond is the strongest in these materials. A complete study of the quantitative analysis of silicates via the standards approach is forthcoming (Hochella, Turner, Harris, and Taylor, in preparation).

Applications

The following three examples illustrate the types of geochemical problem solving for which high spatial resolution SAM is well suited. In each case, information is gained that might be difficult or impossible to obtain using more con­ventional microanalytic techniques such as EPMA, SEM/EDS, or STEM/EDS. High Lateral Resolution Step Scans

In this example, SAM is used to measure cat­ion diffusion profiles between two garnet crys­tals with different compositions which have been joined and subjected to high pressure and tem­perature. These diffusion experiments have been conducted by El phi ck et al. (1985), Loomis et al. (1985), and Ganguly (per. comm.) (see also Loomis, 1978, 1983). Garnet is one of the most important metamorphic minerals, and the

340

compositional zoning within a garnet can be used to help interpret the history of the rock in which it has grown. It is important to under­stand chemical diffusion within garnets because it is a critical factor in controlling composi­tional zonation. In the laboratory, diffusion gradients are usually measured with EPMA line or step scanning across the garnet couple interface. Accurate diffusion modeling for garnet requires a precise measurement of experimental diffusion distances. However, within the laboratory time scale, the diffusion paths of cations in garnet crystals under realistic temperature and pressure conditions are very short.

In this study, we compared EPMA step scans with SAM step scans for Mn and Fe across the in­terface of a spessartine (Feo.2 Mn2.a Al2Si3012l -almandine (Fe2. 1Mgo.9Al2Si3012l garnet couple which has been subjected to 30 Kbar and 1200°C for 2.5 hours followed by 800°C for 9.5 hours (see Fig. 2). The expected diffusion lengths for this experiment are predicted to be less than O.lµm (Elphick et al., 1985). The EPMA step scans, taken from Elphick et al. (1985), were per­formed across the polished interface with a step size of 0.5µm at an accelerating voltage of 15 keV and a beam current of 50 nA. The EPMA scan shows the apparent interface to be about 2.5-3.0 µm wide. In an attempt to improve the analytic resolution, step scans were taken at 10 keV and 10 to 20 nA, but these measurements did not re­sult in a noticeable narrowing of the apparent profile. The SAM step scans were performed across the same interface of the same sample (af­ter the carbon coating used for microprobe analy­sis was removed) using a step size of 0.1 µm and beam conditions of 5 keV and 10 nA. The apparent width of the interface is reduced to approximately lµm.

It is not surprising that, in this case, SAM shows a narrower apparent interface than EPMA by a factor of 2 to 3. Although the electron beam in the EPMA may be more finely focused, the x-ray excitation volume will still be on the order of 1-2 µmin diameter in these materials for the ac­celerating voltages used. However, although not as severe, AES/SAM also suffers from broadening phenomena even though the escape depth for Auger electrons is no more than a few tens of Angstroms. The broadening factors which affect AES have been discussed by, for example, Janssen and Venables (1978), El Gomati and Prutton (1978), El Gomati et al. (1979), and Cazaux ( 1983). The most se­vere compromising effect to the spatial resolution of AES is the Auger electron emission due to back­scattered electrons. El Gomati and Prutton (1978) and El Gomati et al. (1979) have modeled this backscattering with Monte Carlo methods and sug­gest that the chemical resolution (for semi-quan­titative analysis) for AES/SAM should be roughly twice the diameter of the beam outline on the sur­face. If this is the case, and if we assume that the true interface for the garnet pair studied is O.lµm or less in width, we would predict the ap­parent interface width given by SAM to be approxi­mately lµm given the diameter of the primary beam (0.15-0.20µm). Lower accelerating voltage and a finer primary beam with SAM should result in even narrower apparent interfaces on this and other

High resolution SAM of mineral surfaces

1.0 - --I

Mn I I .... 0.8 I Fe I C

I ,--Q) I

(.) I I I... I I Q) I I a. 0.6 I I

I (.) I I

.E \ I I

0 ,,

.... Cll 0.4 I\ Q)

I I \

.2:: I \ ~ I I

I ai 0.2 I \ I... I I

Fe I \ I \

-------- I I Mn I

0.0 '--2 1 0 1 2

distance from interface (µ.m)

Fig. 2. Two EPMA step scans (solid curves) and two SAM step scans (dashed curves) for Mn and Fe across the same interface of a spessartine-alman­dine garnet couple. The EPMA step scans are from Elphick et al. (1985). The SAM step scans are scaled to the EPMA step scans.

samples (Hochella, Ganguly, Turner and Harris, in preparation). High Lateral Resolution Surface Analysis

We demonstrate here the ability of SAM to determine the near-surface composition of very fine individual mineral grains in rocks. These analyses are now possible for grains in the sub­micron size range (Hochel la et al., 1986). In the example presented here, the near-surface com­position of a small sulfide grain in a gold ore sample from the Carlin Mine, Nevada, is observed for reasons given below. The rock containing the gold at Carlin is a carbonaceous and argillaceous limestone which has been partially silicified (Hausen and Kerr, 1968; Wells and Mullens, 1973). Although these and related unoxidized gold ores form an important gold reserve, the gold is high­ly disseminated and has rarely been directly ob­served. The nature of the gold and its distribu­tion in the host rock is important in understand­ing the formation of these types of deposits (Radtke, 1985; Bakken and Einaudi, 1986) and de­veloping the most efficient and highest yielding method of extraction. Recently there has been ex­perimental evidence that gold in aqueous solution can be readily deposited on sulfide surfaces by an adsorption/reduction process (Jean and Bancroft, 1985). This process may be very important in na­ture, even when the gold concentration in solution is very low. Therefore, we have used SAM to probe the surface of a number of pyrite (FeS2 ) and sphalerite (ZnS) grains in this rock.

Small rock fragments from lightly crushed specimens were placed in distilled water in an ultrasonic bath to help disperse the clays. The fragments were then pressed into indium foil and inserted into the instrument. One of the pyrite grains studied is shown in Fig. 3. An Auger spectrum, collected over a small area on this

341

Fig. 3. SEM photomicrograph of a specimen from the Carlin Mine gold-bearing carbonaceous lime­stone. An Auger spectrum was obtained from a face of the euhedral lOµm pyrite cube (indicated by the arrow) and is shown in Fig. 4. The euhe­dral elongated prism to the lower left of the py­rite cube is a quartz crystal.

w -0 ..... w z -0

K

s

200

0

400 600 800 1000

electron energy (eV)

Fig. 4. Differentiated Auger spectrum obtained from the face of a lOµm (on edge) pyrite crystal shown in Fig. 3. The oxygen in the spectrum is from tiny secondary silicate grains on the sur­face of the pyrite and from a thin oxidation layer covering the pyrite crystal.

grain, is shown in Fig. 4. The spectrum indi­cates the presence of 0, Si, S, K, Ca, and Fe. The Si, K, Ca and a portion of the O is from sub­micron grains of secondary silicates on the sur­face of the pyrite cube (see Hochella et al., 1986). The remainder of the O is from a thin Fe-oxide/sulfate surface layer which is common to metal sulfides exposed to air or oxidizing solu­tions (see Perry et al., 1983b, 1984a). The ma­jor gold Auger line, which would be present at 69 eV, is not observed here or on any other spectra taken on the several pyrite or sphalerite grains

M. F. Hochella, Jr., A. M. Turner and D. W. Harris

Fig. 5. SEM photomicrograph of a specimen from an experimental run product of a hydrothermal flow through experiment (Potter et al., 1985; Po­nader and Liou, 1985; Ponader, 1986). The large grain is a plagioclase feldspar, a starting phase in the experiment, which was fractured after the experimental run exposing a fresh surface. The top face, which was exposed during the run, shows a 1-2µm thick crust of new mineral growth.

Fig. 6. SEM photomicrograph of micron to submi­cron serpentine crystals (var. lizardite) which are encrusting a ferro-magnesian mineral, proba­bly olivine. This specimen is from the same run products referred to in Fig. 5. An arrow points to the crystal from which an Auger spectrum was obtained (see Fig. 7).

studied. Further experiments will be required to confirm that gold is not likely to be found on the present pyrite surfaces in this rock. If so, the next step will be to look for gold parti­cles or layering within sulfide grains using ion sputtering to obtain chemical profiles with depth. Identification of Submicron Grains

SAM can be used to identify mineral grains

34'2

w "O ...... .---. w ...... z "O

Si

C

0

400

Si Mg

800 1200 1600 2000

electron energy (eV)

Fig. 7. Differentiated Auger spectrum obtained from the single crystal marked in Fig. 6. This spectrum, along with the photomicrograph in Fig. 6, identifies the mineral as lizardite.

Fig. 8. Submicron silicon grains on ZnSe and ThF4 thin films over a silicon substrate. A step scan for Si is superimposed on the photomicro­graph showing the extremely high lateral resolu­tion of SAM on this electrically insulated sur­face. See text for details.

in the micron to submicron size range in rocks assuming, of course, that the near-surface compo­sition of the grain is representative of its bulk composition. A good example of where it is par­ticularly important to identify grains that are typically very small is in rock/water interaction experiments. In these experiments, crushed rock and aqueous solutions are allowed to react at high temperature and pressure to simulate hydro­thermal conditions. Full interpretation of these experiments not only depends on the change in wa­ter chemistry with time, but also on the identi­fication of newly formed phases. The latter is extremely helpful in understanding elemental re­distribution and reaction paths. The example that follows is from the experiments of Ponader and Liou (1985) and Ponader (1986) who used a

High resolution SAM of mineral surfaces

special hydrothermal fluid flow through system (Potter et al., 1985) to simulate rodingite for­mation. The starting minerals in their experi­mental runs were plagioclase feldspar (Ca1-xNaxAl2-xSi2+x0s), olivine ((Mg,Fe)2Si04), enstatite (MgSi03 ), and diopside (CaMgSi20G). A photomicrograph of one of the feldspar grains after a run (Fig. 5) shows an example of the thin crust of new mineral growth, approximately 1-2µm thick, which can form on the surfaces of starting phases. Fig. 6 shows a high magnifica­tion photomicrograph of micron to submicron crys­tals, which appear to be only 0.2-0.4µm thick, encrusting one of the ferro-magnesian minerals, probably olivine. SAM analysis of one of these submicron crystals (Fig. 7) shows Mg, Si, and 0 (and contaminant C) in the proportions charac­teristic of the mineral serpentine. The mor­phology of these crystals further indicates that they are the lizardite variety of serpentine. Firm identification of this sort plays an impor­tant role in understanding the relevance of these experiments to natural systems.

It should be noted that SAM is capable of lateral resolution better than a few thousand Angstroms on insulators (or electrically isolated conductors) if the material of interest is within a few microns of a supporting substrate which is conducting. An example of this is shown in Fig. 8 where submicron silicon particles are laying on a ZnSe thin film over an insulating ThF4 thin film. These films are supported by a silicon substrate. A step scan for Si is super­imposed on Fig. 8, showing the intensity of the Si L3 M2, 3M2, 3 signal while scanning over a 0.2µm wide section of a silicon particle. The beam diameter for the conditions used (20 keV, beam current in the pA range) was approximately 500~, and the Si signal rises and falls sharply at the boundaries of the particle. The charging problem has been alleviated due to the conductive path immediately beneath the thin insulating films; however, beam damage can be substantial at these high beam energies and careful monitoring for damage is required.

Summary and Conclusions

It is shown that high lateral resolution scanning Auger microscopy can be used to chemi­cally analyze the surfaces of electrically insu­lating minerals. The techniques demonstrated in this paper can be used to perform surface analy­sis on individual oxide, silicate, or sulfide grains in rocks, or for the identification of mineral grains with sizes down to below one mi­cron. The latter assumes that the composition of the near-surface of the grain is representative of its bulk composition. This will not be the case if the sample near-surface has undergone significant oxidation (or reduction), leaching, or chemical exchange, or if any sort of sorption has occurred.

Generally, beam and sample conditions can be found which will minimize charging and sample de­gradation problems, while still allowing for the collection of acceptable Auger spectra. An ac­celerating voltage of 3 keV and beam currents

343

below 10 nA with tilted and relatively flat sur­faces for analysis work best for high resolution SAM on insulators. The tilt angle we typically use is between 40 and 60° (angle between the sam­ple surface normal and the primary beam). This increases backscattered and secondary electron emission efficiency and allows the use of higher beam current densities without encounterino charging. Still, to help prevent the onset of sample degradation, the beam is rastered or de­focused whenever possible, and vacuum levels are maintained as high as possible. Even with these precautions, we have noticed that carbonate min­erals can quickly break down under electron beam exposure, probably due to both heating and elec­tron stimulated desorption. However, most of the oxide, non-hydrous silicate, and sulfide minerals that we have analyzed to this point have shown no signs of significant electron beam induced surface degradation after reasonable amounts of data collection time.

In this study, we have also begun to take the steps necessary to perform reliable semi­quantitative chemical analysis with SAM on sili­cate mineral surfaces. Plots like that shown in Fig. 1 (apparent 0/Si vs. actual 0/Si) may be of practical value in determining certain atomic ra­tios for minerals. However, these plots should not be directly compared with data from other Au­ger instruments unless the electron energy analy­zer used has approximately the same transmission characteristics as that for the Phi 600, and un­less the same beam conditions and sample form and preparation are used.

The applications of high lateral resolution SAM to geologic materials and problem solving are numerous and diverse. Examples of three major types of SAM experimentation on minerals and rocks are given in this paper. High resolution step scans across interfaces will help us under­stand grain boundary chemistries and will supple­ment the information obtained from sputter depth profiling. The surface compositions of individu­al grains in rocks, including very fine material such as clays, can now be studied for potentially important geochemical information. Submicron grains in natural rocks or experimental run pro­ducts can often be identified with SAM if their near-surface compositions are representative of their bulk compositions. More consistent identi­fication will come with further advancements in the chemical quantification of Auger line inten­~ities for minerals.

Scanning Auger microscopy is becoming an im­portant addition to the microanalysis tools which are currently available and most useful for geo­logic materials characterization, namely EPMA, SEM/EDS, and STEM/EDS. The major disadvantages of SAM compared to these techniques are the beam induced charging and surface degradation problems to which SAM is very sensitive. The major advan­tage of SAM is its combination of a very high la­teral resolution and extremely high depth resolu­tion, sensitivity for the lighter elements (ex­cept Hand He), and ease of sample preparation.

M. F. Hochella, Jr., A. M. Turner and D. W. Harris

Acknowledgements

MFH is sincerely grateful for the generous support of the Perkin-Elmer Corporation and the National Science Foundation through Stanford's Center for Materials Research. We are also in­debted to Prof. J. Ganguly of the University of Arizona, Dr. C. M. Taylor of the C. M. Taylor Corporation, and B. M. Bakken and H. B. Ponader of Stanford University for their consultation and the contribution of samples and standards. The manuscript was improved by thoughtful reviews from Profs. P. H. Holloway, D. W. Mogk, and C. G. Pantano and Ors. P. M. Fenn and G. Remand.

References

Armitage AF, Woodruff DP, Johnson PD. (1980). Crystallographic incident beam effects in quanti­tative Auger electron spectroscopy. Surface Sci­ence 100, L483-L490.

Bakken BM, Einaudi MT. (1986). Spatial and tempo­ral relations between wall-rock alterations and gold mineralization, main pit, Carlin Gold Mine, Nevada. Proceedings of the Symposium Gold '86, J. Macdonald (ed.), Toronto, Canada (in press).

Baron RL, Bilson E, Gold T, Colton RJ, Hapke B, Steggert MA. (1977). The surface composition of lunar soil grains: A comparison of the results of Auger and x-ray photoelectron (ESCA) spectros­copy. Earth and Planetary Science Letters 37, 263-272. Bisdom EBA, Henstra S, Kooistra MJ, van Ooij WJ, Visser TH. (1985). Combined high-resolution scanning Auger microscopy and energy dispersive x-ray analysis of soil samples. Spectrochimica Acta 40B, 879-884.

Carriere B, Lanq B. (1977). A study of the charg­ing and dissociation of Si0 2 surfaces by AES. Surface Science 64, 209-223. Carriere B, Deville J-P, Goldsztaub S. (1973). Auger electron spectroscopy of insulating silicon compounds. Vacuum, 22, 485-487. Castaing R. (1960). Electron probe microanalysis. Advances in Electronics and Electron Physics 13, 317-386. Cazaux J. (1983). Mathematical and physical con­siderations on the spatial resolution in scanning Auger electron microscopy. Surface Science 125, 335-354. Clark DE, Pantano CG, Hench LL. (1979). Corro­sion of Glass. Books for Industry and the Glass Industry, New York, 15-19. Connell GL, Schneidmiller RF, Kraatz P, Gupta YP. (1971). Auger electron spectroscopy of lunar ma­terials. Proceedings of the Second Lunar Science Conference, Geochimica et Cosmochimica Acta, Per­gamon Press, New York, Supplement 3, 2, 2083-2092.

Davis LE, MacDonald NC, Palmberg PW, Riach GE, Weber RE. (1976). Handbook of Auger Electron Spectroscopy, 2nd Edition. Eden Prairie, Minne­sota, Physical Electronics Div., 190 pp.

344

Dawson PT, Heavens OS, Pollard AM. (1978). Glass surface analysis by Auger electron spectroscopy. Journal of Physics C: Solid State Physics 11, 2183-2193. deBernardez LS, Ferron J, Goldberg EC, Buitrago RH. (1984). The effect of surface roughness on XPS and AES. Surface Science 139, 541-548.

El Gomati MM, Prutton M. (1978). Monte Carlo cal­culations of the spatial resolution in a scanning Auger electron microscope. Surface Science 72, 485-494. El Gomati MM, Janssen AP, Prutton M, Venables JA. (1979). The interpretation of the spatial resolu­tion of the scanning Auger electron microscope. Surface Science 85, 309-316.

Elphick SC, Ganguly J, Loomis TP. (1985). Experi­mental determination of cation diffusivities in aluminosilicate garnets: I. Experimental methods and interdiffusion data. Contributions to Minera­logy and Petrology 90, 36-44.

Gold T, Bilson E, Baron RL. (1974). Observation of iron-rich coating on lunar grains and a rela­tion to low albedo. Proceedings of the Fifth Lu­nar Conference, Geochimica et Cosmochimica Acta, Pergamon Press, New York, Supplement 5, 3, 2413-2422. Gold T, Bilson E, Baron RL. (1975). Auger analy­sis of the lunar soil: Study of processes which change the surface chemistry and albedo. Proceed­ings of the Sixth Lunar Science Conference, Geo­chimica et Cosmochimica Acta, Pergamon Press, New York, Supplement 6, 3, 3285-3303.

Gossink RG, van Doveren H, Verhoeven JAT. (1980). Decrease of the alkali signal during Auger analy­sis of glasses. Journal of Non-Crystalline Solids 37, 111-124. Grant RW, Housley RM, Szalkowski FJ, Marcus HL. (1974). Auger electron spectroscopy of lunar sam­ples. Proceedings of the Fifth Lunar Conference, Geochimica et Cosmochimica Acta, Pergamon Press, New York, Supplement 5, 3, 2423-2439.

Harris OW. (1984). Surfaces, science ... fact, not fiction. Solid State Technology 27, 278-285.

Hausen OM, Kerr PF. (1968). Fine gold occurrence at Carlin, Nevada, in: Ore Deposits of the United States, 1933-1967, J.D. Ridge (ed), Am. Inst. Min­ing Metall. Petroleum Eng., New York, 908-940. Hochella Jr. MF. (1985). Development of scanning Auger microscopy as a high resolution microprobe for geologic materials. EOS Transactions, Ameri­can Geophysical Union, 66, 407-408.

Hochella Jr. MF, Harris OW, Turner AM. (1985). Scanning Auger microscopy: Development and appli­cations for geologic materials. Geological Soci­ety of America Abstracts with Programs 17, 611. Hochella Jr. MF, Harris OW, Turner AM. (1986). Scanning Auger microscopy as a high resolution microprobe for geologic materials. American Min­eraloqist 71, (in press). Holloway PH. (1978). Quantitative Auger electron spectroscopy - Problems and prospects. Scanning Electron Microsc. 1978; I: 361-374.

High resolution SAM of mineral surfaces

Holm R. (1982). Some applications of Auger micro­analysis in the chemical industry. Scanning Elec­tron Microsc. 1982; III: 1043-1052.

Ichimura S, Shimizu R, Langeron JP. (1983). Back­scattering correction for quantitative Auger analysis. III. A simple functional representa­tion of electron backscattering factors. Sur­face Science 124, L49-L54.

Janssen AP, Venables JA. (1978). The effect of backscattered electrons on the resolution of scanning Auger microscopy. Surface Science 77, 351-364.

Jean GE, Bancroft GM. (1985). An XPS and SEM study of gold deposition at low temperatures on sulphide mineral surfaces: Concentration of gold by adsorption/reduction. Geochimica et Cosmo­chimica Acta 49, 979-987. Le Gressus C, Duraud JP, Massignon D, Lee-Deacon 0. (1983). Electron channeling effect on secon­dary electron image contrast. Scanning Electron Microsc. 1983; II: 537-542.

Levenson LL. (1982). Electron beam-solid interac­tions: Implications for high spatial resolution Auger electron spectroscopy. Scanning Electron Microsc. 1982; III: 925-936.

Loomis TP. (1978). Multicomponent diffusion in garnet: II. Comparison of models with natural da­ta. American Journal of Science 278, 1119-1137.

Loomis TP. (1983). Compositional zoning of crystals: A record of growth and reaction history. Advances in Physical Geochemistry 3, 1-60. Loomis TP, Ganguly J, Elphick S. (1985). Experi­mental determination of cation diffusivities in aluminosilicate garnets: II. Multi-component simulation and tracer diffusion coefficients. Contributions to Mineralogy and Petrology 90, 45-51. MacDonald NC. (1970). Auger electron spectroscopy in scanning electron microscopy: Potential mea­surements. Applied Physics Letters 16, 76-80.

Mackinnon IDR, Mogk OW. (1985). Surface sulfur measurements on stratospheric particles. Geo­physical Research Letters, 12, 93-96. Madey TE, Yates JT. (1971). Electron stimulated desorption as a tool for studies of chemisorp­tion: A review. Journal of Vacuum Science and Technology 8, 525-555.

Mogk OW, Mackinnon !DR, Rietmeijer FJM. (1985). Auger spectroscopy of stratospheric particles: The influence of aerosols on interplanetary dust. Journal of Geophysical Research, Proceedings of the Sixteenth Lun~~ and Planetary Science Confer­ence, Part I, 30-31.

Morrison GH, Gerard JT, Kashuba AT, Rothenberg EV, Potter NM, Miller GB. (1970). Elemental abun­dances of lunar soil and rocks. Proceedings of the Apollo 11 Lunar Science Conference, Geochimi­ca et Cosmochimica Acta, Pergamon Press, New York, Supplement 1, 2, 1383-1392.

Mroczkowski S, Lichtman D. (1983). Calculated Au­ger sensitivity factors compared to experimental handbook values. Surface Science 131, 159-166.

345

Mucci A, Morse JW. ( 1985). Auger spectroscopy de­termination of the surface-most adsorbed layer composition on aragonite, calcite, dolomite, and magnesite in synthetic seawater. American Journal of Science 285, 306-317.

Mucci A, Morse JW, Kaminsky MS. (1985). Auger spectroscopy analysis of magnesian calcite over­growths precipitated from seawater and solutions of similar composition. American Journal of Sci­ence 285, 289-305.

Olson RR, Palmberg PW, Hovland CT, Brady TE. (1983). Applications of AES in microelectronics, in: Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, D. Briggs and M.P. Seah (eds), John Wiley and Sons, New York, 217-246.

Palmberg PW. (1973). Quantitative analysis of solid surfaces by Auger electron spectroscopy. Analytical Chemistry 45, 549A-556A.

Palmberg PW, Bohn GK, Tracy JC. (1969). High sen­sitivity Auger electron spectrometer. Applied Physics Letters 15, 254-255.

Pantano CG. (1981). Surface and in-depth analysis of glass and ceramics. Ceramic Bulletin 60, 1154-1167.

Pantano CG, Madey TE. (1981). Electron beam dam­age in Auger electron spectroscopy. Applications of Surface Science 7, 115-141.

Pantano CG, Dove DB, Onoda Jr. GY (1975). Glass surface analysis by Auger electron spectroscopy. Journal of Non-crystalline Solids 19, 41-53.

Payling R. (1985). Modified elemental sensitivity factors for Auger electron spectroscopy. Journal of Electron Spectroscopy and Related Phenomena, 36, 99-104.

Perry DL, Tsao L, Gaugler KA, Taylor JA. (1982a). Studies of the bonding of iron on feldspar sur­faces using x-ray photoelectron spectroscopy, scanning electron microscopy, and scanning Auger microscopy. Book of Abstracts, 184th American Chemical Society National Meeting, Division of Geochemistry, Abstract 27. Perry DL, Tsao L, Gaugler KA. (1982b). Studies of the bonding of cobalt on feldspar surfaces using x-ray photoelectron spectroscopy (XPS), scanning electron microscopy (SEM), and scanning Auger mi­croscopy (SAM). Abstracts of the 1982 Annual Meeting of the Materials Research Society, Bos­ton, MA, Symposium D.

Perry DL, Tsao L, Gaugler KA. (1983a). Surface study of HF- and HF/H2 SQ4-treated feldspar using Auger electron spectroscopy. Geochimica et Cosmo­chimica Acta 47, 1289-1291.

Perry DL, Tsao L, Taylor JA. (1983b). Surface studies of galena (PbS) using x-ray photoelectron spectroscopy and scanning Auger microscopy. Book of Abstracts, 185th American Chemical Society Na­tional Meeting, Division of Geochemistry, Ab­stract 82.

Perry DL, Tsao L, Taylor JA. (1984a). Surface studies of the interaction of copper ions with metal sulfide minerals. The Electrochemical Soci­ety, Extended Abstracts 84-1.

M. F. Hochella, Jr., A. M. Turner and D. W. Harris

Perry DL, Tsao L, Taylor JA. (1984b). The galena/ dichromate solution interaction and the nature of the resulting chromium (III) species. Inorganica Chimica Acta 85, L57-L60.

Ponader HB. (1986). Serpentinization of lherzo­lite: flow through experiments at 300°C, 300 bars. EOS Transactions, American Geophysical Union 67, (in press).

Ponader HB, Liou JG. (1985). A flow through appa­ratus for the study of rodingitization. EOS Transactions, American Geophysical Union 66, 423.

Poppa H, Elliot AG. (1971). The surface composi­tion of mica substrates. Surface Science 24, 149-163. Potter JM, Pohl DC, Guillemette RN, Ponader HB, Liou JG. ( 1985). A sys tern for fl ow through experi­mental studies at hydrothermal conditions. Neues Jahrbuch fuer Mineralogie, Monatshefte 1985, 329-335. Powell CJ. (1980). Quantification of surface analysis techniques. Applied Surface Science 4, 492-511. Powell CJ. (1984). Inelastic mean free paths and attenuation lengths of low-energy electrons in solids. Scanning Electron Microsc. 1984; IV: 1649-1664. Prutton M. (1982). How quantitative is analysis in the scanning Auger electron microscope? Scan­ning Electron Microsc. 1982; I: 83-91. Radtke AS. (1985). Geology of the Carlin Deposit, Nevada. United States Geological Survey Profes­sional Paper no. 1267, 124p. Remond G, Hollov1ay PH, Gressus C. (1981). Elec­tron spectroscopy and microscopy for studying surface changes of mechanically prepared pyrite and quartz. Scanning Electron Microsc. 1981; I: 483-492. Remond G, Holloway PH, Hovland CT, Olson RR. (1982). Bulk and surface silver diffusion related to tarnishing of sulfides. Scanning Electron Mi­crosc. 1982; III: 995-1011. Remond G, Picot P, Giraud R, Holloway PH, Ruzkow­ski P. (1983). Contribution of electron spectros­copies to x-ray spectroscopy applied to the geo­sciences. Scanning Electron Microsc. 1983; IV: 1683-1706. Remond G, Holloway PH, Kosakevitch A, Ruzakowski P, Packwood RH, Taylor JA. (1985). X-ray spectrom­etry, electron spectroscopies and ootical micro­reflectometry applied to the study 6f ZnS tar­nishing in polished sulfide ore specimens. Scan­ning Electron Microsc. 1985; IV: 1305-1326.

Seah MP. (1983). A review of quantitative Auger electron spectroscopy. Scanning Electron Microsc. 1983; I I: 521-536. Stephenson DA, Binkowski NJ. (1976). X-ray photo­electron spectroscopy of silica in theory and ex­periment. Journal of Non-Crystalline Solids 22, 399-421.

346

Storp S. (1985). Radiation damage during surface analysis. Spectrochimica Acta 40B, 745-756.

Venables JA, Janssen AP, Harland CJ, Joyce BA. (1976). Scanning Auger electron microscopy at 30 nm resolution. Philosophical Magazine 34, 495-500.

Wehbi D, Roques-Carmes C. (1985). Surface rough­ness contribution to the Auger electron emission. Scanning Electron Microsc. 1985; I: 171-177.

Wells JD, Mullens TE. (1973). Gold-bearing arse­nian pyrite determined by microprobe analysis, Cortez and Carlin gold mines, Nevada. Economic Geology 68, 187-201.

Wells OC, Bremer CG. (1969). Improved energy analyser for the scanning electron microscope. Journal of Scientific Instrumentation (Journal of Physics E) 2, 1120-1121.

White AF, Yee A. (1985). Aqueous oxidation-reduc­tion kinetics associated with coupled electron­cation trunsfer from iron-containing silicates at 25°C. Geochimica et Cosmochimica Acta, 49, 1263-1275.

Yau YW, Pease RFW, Iranmanesh AA, Polasko KJ. (1981). Generation and applications of finely fo­cused beams of low energy electrons. Journal of Vacuum Science and Technology 19, 1048-1052.

Discussion with Reviewers

D. W. Mogk: In addition to the applications of SAM discussed in this paper (high lateral resolu­tion step scans, surface analysis, and identifi­cation of submicron grains) other applications include 1) characterization of surface reactions (such as adsorption, or catalysis in geologic systems); 2) demonstration of inhomogeneities be­tween bulk and surface chemistries; 3) depth pro­filing of surface chemistries; and 4) two dimen­sional mapping of components on surfaces. Don't these additional techniques demonstrate the fur­ther potential of SAM in petrologic investiga­tions beyond developing a higher resolution mi­croprobe? Authors: Absolutely, although the points that you mention are beyond the scope of this paper. Many of these points have already been addressed, at least in a preliminary sense, in papers refer­enced in the Introduction of this article.

The reason that this paper emphasizes the high resolution microprobe aspect of SAM analysis is because it is this feature, along with its in­herent ~urface sensitive nature, that sets it apart from most other techniques. If x-ray photo­electron spectroscopy had lateral analytic reso­lution comparable to SAM, it would generally be the method of choice, especially for insulators. Therefore, it seems appropriate to push the lat­eral analytic resolution of SAM for minerals as far as possible while being careful to avoid sur­face charging or degradation due to higher beam current densities.

D. W. Mogk: The really exciting potential of SAM is the possibility of examining grain bounda­ries and microcracks. It is in this setting that crystals interact with their geologic

High resolution SAM of mineral surfaces

environment. For the petrologist, the challenge will be to ask appropriate questions and to de­sign experiments that can be directly addressed by SAM. Will we not see an explosion of new stud­ies concerninq the kinetics of dissolution/ precipitation of minerals, and other surface re­actions such as catalysis and adsorption applied to geologic systems? Authors: Yes, we believe that the time is right for a much better physical understanding of the reactions that occur especially at the mineral/ aqueous solution interface, including adsorption, exchange, precipitation, and dissolution. These processes, of course, control factors which are of fundamental importance in many geologic sys­tems, including weathering rates, crystal growth rates, and solution compositions. Although a num­ber of techniques will have to be employed in these studies, including, for example, XPS, FTIR, and EXAFS, SAM should play a major role, espe­cially where small individual grains must be examined.

D. W. Mogk: Ion millinq can be used to systema­tically strip away layers at surfaces. Can the relative chronology of surface reactions be de­termined by depth profiling of the surface chemi­cal "stratigraphy"? Authors: This type of study is certainly possi­ble using SAM, although in many cases SIMS or small area XPS may be the method of choice. SIMS is particularly good at profiling a light element in a heavier matrix, and XPS avoids the charging problems that SIMS and SAM can experience when profiling in insulators. However, before embark­ing on a depth profiling study with any of these techniques, two potential problems should be con­sidered. First, differential sputtering can occur when the sputtering rates of the matrix constitu­ents are appreciably different. This effect can quickly and significantly modify the near-surface composition. Second, surface roughening can oc­cur, especially when sputtering through more than a few microns. This roughening greatly reduces the effective depth resolution.

D. W. Mogk: One further advantage of using SAM is its ability to analyze for light elements (ex­cept Hand He). This will provide many opportuni­ties for petrologists to open new lines of inves­tigation beyond the standard procedures using the electron microprobe or EDS systems. Authors: SAM's ability to easily detect oxygen is extremely useful in geologic research. For example, because the compositions for minerals derived by SAM are presently no better than semi­quantitative, knowing the approximate Si/0 ratio for a silicate may help you to identify it with­out the need for quantitative analysis. Also, distinguishing between oxides and metals or sul­fates and sulfides becomes trivial. Unfortunate­ly, although there has been much interest in the carbon content and its distribution in mantle­derived minerals and rocks, it may not be possi­ble for Auger spectroscopy to routinely dis­tinguish between this carbon and adventitious carbon from air exposure. UHV fracture will have to be employed. Finally, it is obviously some­times important to analyze for the light metals in

347

minerals that are very difficult to detect by tech­niques that depend on x-ray fluorescence.

P. H. Holloway: While you mention semi-quantita­tive to quantitative analysis for SAM using stand­ards similar to the unknown, elemental standards and correction factors have been developed for AES similar to the ZAF corrections used for the electron microprobe (see, e.g., Holloway, 1978). Please comment on their usefulness. Authors: We have not attempted to use any other quantification scheme other than the "exact stand­ards" method as it is described in your paper (Holloway, 1978). At this relatively early stage of the development of quantitative analysis of silicates with Auger spectroscopy, we feel most comfortable using this approach because of its simplicity. On the other hand, making extensive matrix corrections for minerals which typically have relatively complex chemistries would be far more uncertain. Also, we have access to excellent silicate standards and we are confident that their near-surface compositions, as we have prepared them, are representative of the bulk. Eventually, we would like to combine the use of these stand­ards with theoretical matrix correction factors.

P. H. Holloway: While the interface width data in Fig. 2 are impressive, does the SAM offer an advantage if the initial width of the undiffused interface measured by the EPMA had been decon­voluted from the width after diffusion? Authors: There is one main advantage. When the crystals in the diffusion couple are joined (at high pressure and temperature), a small amount of diffusion takes place. The amount of diffusion during this process can only be estimated, and therefore cannot be exactly accounted for in a deconvolution scheme. A direct high resolution measurement would leave no doubt as to the short diffusion distances. Nevertheless, deconvolution schemes will be very useful in this situation, and one is currently being developed (J. Ganguly, per. comm.).

P. H. Holloway: In this same light, does the dif­fusion width measured by SAM (Fig. 2) represent a true width or does it still contain instrumental broadening? Authors: The diffusion width measured by SAM is a convolution of the actual diffusion width and the width of the analysis area at any point along the step scan (due to the beam width, analytic broad­ening phenomena like Auger electron emission due to backscattered electrons, and beam instability).

P. H. Holloway: For the pyrite grain surfaces analyzed (see Fig. 3), did you look for the Au Auger line at 2024 eV? The 69 eV Au peak is very surface sensitive and Remand et al. have shown that chalcopyrite tends to form an outer iron ox­ide layer which screens the underlying Cu sulfide layer. Such an oxide on pyrite may prevent the Au 69 eV Auger electron from reaching the analyzer without inelastic scattering. Authors: Yes, we have taken narrow scans over the 2024 eV region, but have still failed to see a Au Auger signal. The advantage of using the 2024 eV Au Auger line is that it has no neighboring Auger

M. F. Hochella, Jr., A. M. Turner and D. W. Harris

lines from other elements prominent in these samples, and, of course, these electrons come from a greater average depth in the sample. How­ever, the 69 eV line is still perhaps the most important Au line to monitor for these samples (it is over 20 times more intense than the 2024 eV line at 3 keV beam energy) as long as you sput­ter to make sure that a very thin surface layer is not masking the signal as you point out. In these measurements, after an initial spectrum of a sulfide grain was collected, the surface oxide layer was sputtered away and the spectrum recol­lected. It is possible that we will see Au with more extensive sputtering down into the sulfide.

C. G. Pantano: The preferential sputtering ef­fect does become a problem in silicates where high atomic number cations (e.g. Ca, Ti, Zr, Sn) are present in the structure. Did you examine the dependence of the relative Auger signals upon sputtering for the standards used in Figure 1? Authors: Yes, vie looked very carefully at one sample as a test case for possible differential sputtering effects under the sputtering condi­tions used to remove the surface contamination layer from all of the samples listed in the fig­ure caption of Fig. 1. The test was performed on an alkali feldspar (Na,K)AlSi 308 ). One sample surface was exposed by fracturinq under UHV con­ditions. Another surface was exposed to air, stored in ethanol, and then inserted into the instrument and sputtered (4 keV Ar ions for 10 seconds). Carefully measured peak-to-peak heights from both samples for all constituent elements, including K, were identical within experimental error. At least for elements as heavy as K, this amount of sputtering does not seem to be a prob­lem for semi-quantitative analysis. We plan to test heavier elements and different sputtering conditions for our silicate standards in future studies.

G. Remand: What kind of accuracy might be ex­pected when using sensitivity factors for "quantitative" analysis when large topographic variations or crystallographic orientation changes are present? Authors: The accuracy of an analysis will not generally suffer due to topographic variations as long as the analysis does not require absolute intensities. This is because surface roughness typically affects the Auger intensities of all elements uniformly, so the use of relative in­tensities is preferred to negate its consequen­ces. However, there is at least one case where all elements will not be uniformly affected. This is when EP (the energy of the primary beam) is allowed to approach E

11 (the minimum beam energy

required to ionize core level U for the UVW Auger transition) for some element in the sample. In this case, surface roughness will reduce the Auger intensity for this element relative to the other elements (see Holloway, 1978). Obviously, this situation should be avoided. To avoid the surface roughness problem when measuring absolute intensities, we suggest looking at cleavage sur­faces (in the case of minerals) or conchoidally fractured surfaces (in the case of glasses or minerals with poor cleavage) which are typically

348

exceptionally flat. An added benefit of looking at flat rather than rough surfaces is that sample charging tends to be much more easily controlled. This can be rationalized by realizing that the beam is impinging on the surface of a rough sam­ple at many angles, many not appropriate for charge neutralization for the beam conditions used (see Hochella et al., 1986).

The quality of AES quantification can be seri­ously compromised by crystallographic orientation effects. This is not so much due to anisotropic Auger electron emission, as the large acceptance angle of cylindrical mirror electron energy anal­yzers commonly used for Auger analysis tends to average its effects. However, this is not the case for electron diffraction (channelling) which de­pends solely on the relative orientation of the primary electron beam and the sample and not on the type and/or position of the analyzer. It has been demonstrated that diffraction effects can re­sult in significant variations (>20%) in relative Auger intensities (see, e.g., Bishop et al. (1984) Crystalline effects in Auger electron spectroscopy, Surface and Interface Analysis, 6, 116-128). The only direct observations that we-have made of the effects of crystallographic orientation on SAM analyses of mineral surfaces are on the (001) and (010) faces of an albite single crystal with the primary electron beam at approximately 50° to the sample normal in both cases. We observed no appre­ciable relative intensity changes for any of the constituent elements.

G. Remand: Could the accuracy of quantitative measurements be improved by measuring the area of the Auger peak rather than using the peak-to­peak height for the intensity? Authors: Yes, but we have not as yet attempted this.Complications in measuring Auger peak areas include modeling backgrounds and knowing where to cut off the Auger peak from energy loss features. As our Auger quantification work for Earth mate­rials is refined further, we will begin using peak areas. At this stage of our work, we feel that using peak-to-peak measurements are acceptable as long as there is no peak shape change between the standards and the unknowns. If there is, it may be advisable to simply measure the negative excursion (high KE) side of the differentiated peak. The positive excursion (low KE) side of the Auger peak is more likely to vary from sample to sample be­cause its shape is characteristic of energy loss features (see, e.g., Seah (1979) Quantitative Auger electron spectroscopy: Via the energy spec­trum or the differential?, Surface and Interface Analysis,_!_, 86-90).

P. M. Fenn: How stable are clays in the electron beam? Authors: From our limited experience thus far with sheet silicates, we believe that clays will be stable except for loosely bound and/or volatile interlayer species. This is suggested by our anal­ysis of the mineral apophyllite, a sheet silicate with the formula Ca4K(Si4010)2F·8H20. If Fig. l of this paper is used to correct for the ap­parent 0/Si ratio as determined by Auger peak-to­peak heights and sensitivity factors derived from Si0 2, the amount of oxygen in the analysis

High resolution SAM of mineral surfaces

suggests that the interlayer H70 has been lost. This is not surprising, of course, considering the heating effects of the primary beam.

P. M. Fenn: Have you tried analyzing for Li in minerals with AES? Authors: Yes, but not extensively. In Hochella et al. (1986), we report on the analysis of Li in the mineral spodumene which has a Li content of approximately 10 atomic percent. We observed a definite Li Auger signal at approximately 37 eV. Unfortunately, this peak overlaps a weak Al Auger line, making any sort of quantification more dif­ficult in this case. Other elements which are relatively common in minerals and which have po­tential peak overlap problems with Li are Mg, Ti, Cr, Mn, and Fe.

349