Electron stimulated neutral and ion emission from single crystal NaNO3

*H CJLJ __ __

Nuclear Instruments

and Methods in Physics Research B 103 (1995) 284-296

Beam Interactions with Materials&Atoms

I!!!!! ELSEVIER

Electron stimulated neutral and ion emission from single crystal NaNO, J.-J. Shin a, S.C. Langford a, J.T. Dickinson a,*, Y. Wu b a Physics Deparbnent, Washingron State Universiry. Pullman, WA 99164-2814, USA

I. y Department

of Chemistry, Rutgers, The State University of New Jersey, Piscataway,

NJ 088.55-0849,

USA

Received 8 February 1995; revised form received 3 May 1995 Abstract Quadrupole mass spectrometry of the neutral products accompanying irradiation of single crystal NaNO, with a pulsed beam of l-3 keV electrons shows intense NO and 0, emissions and smaller amounts of Na and NO,. Time-of-flight and intensity measurements as a function of pulse width and temperature are consistent with NO emission via a thermally assisted, electronic emission mechanism. The activation energy for the thermally assisted process is about 0.11 eV -similar to the excitation energies for internal vibrational modes of the NO; anion. The emission of neutral 0, involves fairly complex reactions with dissociation products and is also thermally assisted at temperatures above 380 K. Mass-selected ion emission measurements indicate copious H+ emission, where the temperature dependence of the H+ intensity is similar to that of NO. We propose that H+ emission is limited by the density of appropriate surface defect sites, and that these sites are directly related to NO; dissociation.

1. Introduction The interaction of ionizing radiation with wide band gap materials can be dramatically altered by the presence of defects which allow for excitations at sub-band gap energies. In much the same way, materials containing molecular ions or molecules often exhibit intramolecular excitations at energies well below the nominal band gap. These excitations can alter the physical arrangement of the atoms within the molecular ion (isomerization) and, in some cases, produces dissociation. The alkali nitrates are a particularly well studied class of materials where exposure to ionizing radiation dissociates the molecular ion component (Refs. [l-3] and references therein). However, a molecular level description of the dissociation mechanism is not available. Further, few experiments have been performed in vacuum, where the emission of dissociation products can be directly observed. We report mass-selected measurements of the neutral and ionic products produced by irradiating crystalline NaNO, with pulses of l-3 keV electrons at current densities of 30 to 600 p_A/cm*. The doses employed here are sufficient to slowly alter the surface composition and were

* Corresponding author. Tel. + 1 509 335 4914, fax + 1 509 335 7816, e-mail [email protected].

chosen specifically because of their dramatic effect on interactions with near-UV laser radiation (to be discussed elsewhere). The role of defects on the dissociation and emission processes is a major focus of this study. Our measurements show that surface and near-surface NO; dissociates upon electron bombardment of NaNO, to yield neutral NO by a thermally assisted electronic process, probably involving dissociative electron capture. Other neutral species (principally 0,) are created from the fragments of this primary dissociation. The time evolution of the measured NO and 0, intensities indicate that substantial emission occurs after the end of the electron pulse. Electron stimulated desorption (ESDI of H+ is also observed from NaNO,, similar to previous ESD observations of H+ from H,O or H, sorbed on various surfaces [4-61. We also report X-ray photoelectron spectroscopy (XPS) measurements of the compositional changes in the nearsurface region due to electron irradiation. An accompanying work presents a parallel study of emissions produced by pulsed UV irradiation of NaNO, at 248 nm [7], where laser-induced defects dramatically alter the laser-material interaction at quite modest exposures. A third study in preparation investigates the combined (synergistic) effects of co-focused laser and electron beams. The radiolysis of NaNO, by energetic electrons is a complex, but interesting example of electron stimulated desorption (ESDI [8]. The interaction of ionizing radiation with sodium nitrate is of substantial practical importance

0168-583X/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0168-583X(95)00610-9

J.-J. Shin et al./ Nucl. Instr. and Meth. in Phys. Res. B 103 (1995) 284-296

in the storage of high level radioactive wastes from nuclear weapons production in the United States. Sodium nitrate is a major component of this waste, and has been exposed to high fluxes of ionizing radiation for several years. The effects of ionizing radiation on this material thus have important implications as to the chemical evolution, analysis. and ultimate treatment of these wastes. The work presented here implies that secondary electrons play a critical role. In view of the importance of these low-energy electrons, a parallel study has been carried out by Knutsen and Orlando [9] which directly map out thresholds leading to degradation and desorption.

2. Experiment Sodium nitrate crystals were grown from saturated aqueous solution. After drying, the crystals were cleaved in air into 2-mm thick slices and mounted in a vacuum chamber. After pumpdown, samples were annealed in vacuum for several hours at 400 K. Most experiments were conducted at a background pressure less than lo-’ Pa. Pulsed electron irradiation was provided by rapid deflection of the beam from a Varian Model 981-2455 Auger electron gun onto the sample; typical pulse durations and repetition rates were 50 ps-200 ms and l-30 Hz, respectively. Incident electron energies varied from 1 to 3 keV with current densities ranging from 30 to 600 p.A/cm*. The total electron currents were determined by directing the beam onto a positively biased copper plate and measuring the current to the plate with an electrometer. Time resolved current measurements showed that the transit time for the electron beam in the deflection process was much less than 1 ps. Sample heating was provided by a stainless steel resistance heater; the temperature of the heater was measured with a chromel-alumel thermocouple. In all cases, Joule heating due to the pulsed electron beam is negligible. (Worst case calculations show that Joule heating is less than 5-8°C.) Neutral emission measurements were made with a Quadrupole Mass Spectrometer (QMS) mounted with its axis along the sample surface normal. Mass resolution was better than f 1 amu/e. The QMS mass filter was typically tuned to a specific mass/charge ratio and the output detected as a function of time. Neutral particles were ionized by electron impact (70 eV electrons) in the QMS ionizer. The resulting ions were drawn into the mass filter section, mass selected, and detected with a Channeltron Electron Multiplier (CEM) (lo4 gain). The output of the CEM was amplified with a fast electrometer. The time resolved signals accompanying 20-100 successive pulses were digitized and averaged with a LeCroy 9450 digital oscilloscope. We refer to the resulting signal for a particular mass as a time-of-flight (TOF) distribution, I(t), realizing that I(t) represents a convolution of the instantaneous density of ionizable atoms/molecules in the QMS ionizer

285

with the time-response of the electrometer. The measured r(t) also displays a mass-dependent delay due to the ion time-of-flight through the mass filter. We note that in continuous operation, the QMS ionizer produces a stream of low energy electrons (N 50 eV) which can significantly alter the NaNO, surface over time. The high electron beam currents employed in this work ensured that virtually all the damage during the experiment was due to the electron beam. Care was required, however, to avoid excessive exposure to low-energy electrons from the QMS ionizer between experiments. The effects of excessive damage accumulation could be almost completely reversed by a brief anneal at 500 K. To probe different emission processes which would yield different TOF curves, 1(t) was curve-fit by an equation of the form: I(t)-/bdt’R(t-t’)[‘dt’M(t’-t”)S(t”).

(1)

where R(t - t’) is the output of the electrometer at time t due to a unit input at time t’, M(t’ - t”) is the input to the electrometer at time t’ due to a unit emission at time t” (assuming a Maxwell-Boltzmann velocity distribution), and S(t”> (which we call the source function) is the emission leaving the surface as a function of time t”. The electrometer response function R(t - t’) accounts for the finite time response of the electrometer (rise time 30 ps on the more sensitive scales), which is a significant fraction of the duration of the shortest pulses (50-100 ps). It is completely determined by direct measurements of the electrometer response to short current pulses and thus introduces no adjustable parameters into the modeling procedure. The function M(t’ - t”) accounts for the time required for particles to travel from the sample to the quadrupole ionizer, where we assume a Maxwell-Boltzmann velocity distribution. Since the QMS signal reflects the particle density in the ionizer, it is convenient to transform the Maxwell-Boltzmann velocity distribution to a spatial distribution which can be integrated as a function of time over the ionizer volume, i.e.,

(2) where cr is the ionizer efficiency, m is the particle mass, k is the Boltzmann constant, and T is the effective temperature of the surface. The integral on the right is performed over the volume of the ionizer, AV, where r is the particle displacement relative to the point of emission and rX is the component of the displacement normal to the sample surface.

J.-J. Shin et al./ Nucl. burr. and Meth. in Phys. Rex B 103 (199.5) 284-296

286

An additional, mass dependent delay is introduced by the time required for ions produced in the ionizer to travel through the mass filter. This delay is readily calculated and has been experimentally verified at several masses. The ion flight time through the mass filter is accounted for by appropriately reducing 1’ in the integral over t” in Eq. (1). The source functions S(f”) for most electron-stimulated emission processes are well described by a constant emission intensity for the duration of the electron pulse [S(r) - constant]. This corresponds to one adjustable parameter in Eq. (I). In the present work, constant source functions alone could not adequately fit the NO and 0, signals; in addition, the best of these fits yielded unphysical temperature values ( < 200 K). A more satisfactory description of the TOF data is obtained by assuming an additional, “delayed” source of NO or 0, that rises gradually during the electron pulse and falls gradually afterwards. For simplicity, we describe the rise and fall of emission intensities with functions appropriate to the accumulation and decay of an intermediate state with average lifetime T, where the intermediate state is produced at a constant rate during the electron pulse: S(r”) N (1 -e-r”/7

) during the electron pulse,

(3a)

TOF)]. Using pulse counting techniques, good signal-tonoise and time resolution (1 ks per channel) were obtained for the ion signal. The output of the QMS was amplified by an Ortec Model 579 fast filter amplifier (rise time 5 ns), then discriminated and counted by standard nuclear physics instrumentation. The resulting TOF signals were used to determine the ion kinetic energies. For XPS measurements, high purity NaNO, powder was compressed into disks (about 8 mm in diameter and 0.5 mm thick) and mounted in a Kratos XSAMSOO-MCD system. Photoelectrons produced by Mg Ka X-rays were detected with a cylindrical energy analyzer operated at a pass energy of 65 eV. Electron damage was performed by rastering the sample with 500 eV electrons at a current density of about 2.0 p,A/cm’. During rastering, the system pressure rose from 1 X IO-’ Pa to about 2 X IO-’ Pa. The damage induced by the X-ray beam during data acquisition was negligible compared with changes induced by the electron beam.

3. Results 3.1. Neutral

emission

and S( l”) u e-“‘/’

afterwards.

(3b)

As discussed below, the need for a source function containing relaxation is evidence for a diffusion limited emission mechanism, possibly due to the diffusion of excitons or charged carriers to the surface. The model given by Eqs. (l)-(3) was fit to experimental data using a nonlinear least squares technique (the Marquardt algorithm [lo]). The parameters were highly over determined and yielded stable solutions with reasonable uncertainties. We note that value of the constant T is affected by the details of the velocity distribution of the desorbed particles. Since the Maxwell-Boltzmann distribution corresponding to room temperature is the “slowest” physically reasonable velocity distribuand “broadest” tion, any deviations in the velocity distribution will require more delayed emission in order to account for the TOF observations. Thus deviations from the Maxwell-Boltzmann distribution should not alter the principal conclusions of this work. Measurements of ionic species were made by grounding the filament and grid structures of the QMS ionizer, providing a line-of-sight, nominally field-free path from the sample through the mass filter to the detector. Since the radial electric fields in the mass filter have little effect on the component of ion velocity along the axis of the filter, the QMS/sample arrangement can be treated as a time-of-flight (TOF) tube whose length is equal to the distance from the sample to the detector (30 cm). [We have tested this condition many times by injecting pulses of ions of known kinetic energy into the mass filter and measuring

The principal uncharged molecular products produced by electron bombardment of crystalline NaNO, are NO (30 amu/e) and 0, (32 amu/e>. Fig. 1 shows the mass spectrum of the neutral emissions accompanying irradiation with long (200 ms) electron pulses. The signal intensities accompanying long pulses (> 1 ms) are strongly affected by the pumping speed of the system, which is somewhat mass dependent. To correct for the buildup of a gas phase background signal due to emission processes, we blocked the path of molecules emitted directly from the sample and subtracted the intensity of the blocked signal from the intensity of the non-blocked signal. As noted below, the relative intensities of the NO and 0, signals is a strong function of pulse width. Significantly, the intensity of the Na signal is quite small relative to the NO and 0, signals.

4001

NO

0'

0

10

2a 30 Mass (amu)

40

50

Fig. 1. The relative intensities of the principle neutral emissions from annealed NaNO, produced by exposure to 200-ms, 300 pA/cm* pulses of 2500-eV electrons.

287

J.-J. Shin et al. / Nucl. Instr. and Meth. in Phys. Res. B 103 (1995) 284-296

In evaluating the N and 0 signals, allowance must be made for the dissociation of nitrogen-and oxygen-containing molecules by the electron beam or in the QMS ionizer. Dissociation of N, and 0, in the QMS ionizer typically produces N and 0 signals equal to 10% of the N, and 0, intensities, respectively. Similar contributions to both the N and 0 signals are expected from NO. Thus no more than 30% of the N and 0 signals in Fig. 1 can be attributed to dissociation of larger molecules in the ionizer. We therefore conclude that measurable amounts of atomic N and 0 are emitted directly from the surface. The size of the N and 0 signals relative to the NO and 0, signals also argues strongly against dissociation in the electron beam. Assuming conservative dissociation cross sections (10 A’) and transit times across the electron beam (20 l.~s), the electron density in the near-surface region (primaries plus secondaries) would have to be at least three orders of magnitude higher than the electron beam current density to produce the observed signals. As the pulse width increases, the intensity of the 0, signal increases strongly relative to NO. The variation in NO and 0, yield (per incident electron) is plotted as a function of pulse width in Fig. 2 for pulse widths between SO and 1000 us. The NO yield (vs pulse width) is nearly constant. In contrast, the 0, yield increases dramatically as the pulse width increases, saturating at - 1 ms. Thus, although we will propose that the initia1 excitations responsible for the NO and 0, emissions are similar, the emission processes contrast markedly. Although the NO and 0, emission intensities respond differently to changes in pulse width, both emissions show

Yield ol NO and O1 Emtnioa n Phw Width 2500 eV, S30 @/cm2 20’

(a) NO

P

I :: E, 10

lb*

B

******+*+

3 : % *

0 0

SO0

moo

Pulse Width(w)

(b) 0, * I

+

+

*

t

=

+ * **

o0

500

IO00

Puke Wldtt, (w)

Fig. 2. Comparison of the neutral yield of (a) NO and (bb)0, yield produced by 530~&A/cm* pulses of 2500-eV electrons as the pulse width is varied from 50 p,s to 1 ms.

NO and O1 Intmdty

YS &earn Current 2500 ev. so-ma Pulses

80’ (a) NO Q

.

f ‘:

40.

**

l

l

*

l

% 3

+

*

l

.

I

l

Fig. 3. NO and 0, neutral intensities produced by 50-ms pulses of 2500-eV electrons as the beam current is raised from 30 to 300 PA/cm*.

a linear dependence on current density (at fixed pulse width). Fig. 3 shows the NO and 0, intensities produced by SO-ms pulses of 2500 eV electrons as the beam current density is raised from 30 to 300 PA/cm’. Similar measurements as a function of electron kinetic energy at constant current show small but measurable increases in NO and 0, emission intensities as the electron energy is raised from 1 to 3 keV. The weak energy dependence strongly suggests that secondary electrons (whose yield is only a weak function of primary electron energy for ionic crystals over this range of energies [l 11) are the major contributors to electron-induced excitations producing both NO and 0,. For low molecular weight matrices, secondaries are produced down to N 1 pm for our incident energies. Both NO and 0, emission intensities are strong functions of temperature. The emission intensities are displayed in Arrhenius form in Fig. 4. To avoid a phase transition at 560 K [12], the maximum temperature was kept below 520 K. Although larger temperature ranges are desired for activation energy measurements, the available range is sufficient for reasonable estimates. The NO plot becomes linear at temperatures slightly above room temperature, with an apparent activation energy of about 0.11 eV; an average of several runs yielded an activation energy of 0.11 + 0.04 eV. Similar measurements of the Na and NO, emission intensities as a function of temperature yielded activation energies very close to the NO activation energy. The behavior of the 0, emission is more complex, with a distinct minimum at about 380 K. The initial decrease in 0, intensity as the temperature is raised from room temperature suggests the presence of a temperature dependent

J.-J. Shin et al./Nucl.

288 Arrbfniur Ptof

or NO

nad 0, Emhsion

Instr. and Meth. in Phys. Rex B 103 (1995) 284-296 NO and 01 TOF Signah with Model Curve Fits 2500 eV. 460 pA/cm’, lOO+a Pulsar

lntewtttes

2500 eV, MO pA/cm2. 200 p.3 PI&S

I (a) NO

E.= 0.11 eV

c

Prompt soureecmly

0

100

200 300 Time@)

400

5

(b) 0,

0

Fig. 4. Arrhenius plots of the neutral (a) NO and (b) 0, by 200.ps, 300.PA/cm* pulses of 2500-eV electrons.

produced

reaction. Nevertheless, 0, emission is thermally activated at the higher temperatures with an apparent activation energy of 0.16 + 0.05 eV. Typical NO and 0, TOF signals accompanying 100~ks, 460-PA/cm2 pulses of 2500 eV electrons are shown in Fig. 5 for two different sample temperatures. In each case, the onset of electron exposure is at time t = 0. At a pulse width of 100 p,s, the NO peak intensity is about twice that of 0,. Consistent with the temperature dependence shown in Fig. 4, the NO intensity at 430 K is about twice that at competing

NO and Oz TOF Signals Accmnpanying 2500 eV, 466 ti/cm’ 20

1OOys Pulsed

(0) NO

I

/*;k 0

200

400

600

800

1006

Time Us)

Time (116)

Fig. 5. The time-of-flight curve of (a) neutral NO and (b) neutral 0, produced by 100.KS, 460-p,A/cm’ pulses of 2500-eV electrons.

Fig. 6. (a) NO and (b) 0, emissions accompanying 100~ps, 460~p,A/cm2 pulses of 2500.eV electrons. The signals have been fit to the Maxwell-Boltzmann distribution with the step source function only (constant emission during pulse-light curve) and with the delayed + step source function (dark curve): the temperature is fixed at 320 K.

320 K, while the 0, intensity at the two temperatures is about the same. Significantly, the time-behavior of the NO and 0, emission intensities (peak position and width) is not strongly affected by temperature. As discussed below, temperature changes would strongly affect the TOF of emissions which are rate limited by thermal desorption; we therefore rule out thermal desorption during the electron pulse as a rate limiting step. The NO and 0, desorption products have very low apparent kinetic energies. For instance, assuming (1) an emission rate proportional to the electron current during the electron pulse (a “square wave” source), and (2) that the velocities correspond to a thermal distribution at the sample temperature (320 K), one predicts the TOF signals given by the lighter lines in Fig. 6. (Only the first 500 ks of emission are shown to avoid the effect of accumulated volatiles. TOF measurements made after blocking the direct path from the sample to the QMS ionizer show that volatiles do not contribute to the observed signal until 600 /J,S after the beginning of the electron pulse.) The signals predicted from a square wave source function leads the observed NO and 0, signals by about 40 bs. This delay is far greater than experimental uncertainties. Unphysical temperatures (below 200 K) are required to account for these velocities on the basis of a “square wave” source, and the resulting predicted emissions still yield a poor fit to the time-resolved intensity distribution. Thus we conclude that the instantaneous rates of NO and 0, emission are not described by square wave source functions corresponding to “prompt” emission alone.

J.-J. Shin et al./ Nucl. Instr. and Meth. in Phys. Rex B 103 (1995) 2X4-296

A more satisfactory description of the TOF signals can be obtained by assuming the presence of an additional, “delayed” source of NO or 0, that rises gradually during the electron pulse and falls gradually afterwards. We treat this delayed emission component using the source function given in Eq. 2, i.e., an exponentially rising emission during the electron pulse and an exponentially decaying emission after the electron pulse. Very successful non-linear least square fits to the data were obtained by employing a velocity distribution corresponding to the measured sample temperature (typically 320 K) and using r and the intensities of the prompt and delayed components as fitting parameters. The darker lines in Fig. 6 shows the resulting fits. For the NO emission, the best fit values are r= 13.5 + 40 ps with 35 I_ 5% of the NO emission being derived from the prompt component; for the 0, emission, r = 120 f 40 p..s with 10 + 4% derived from the prompt component. Within the uncertainty of the fitting procedure, the decay time constants for NO and 0, emission are equal. An analogous analysis of the NO emission accompanying low fluence laser irradiation at 248 nm yielded a very similar decay constant (7 = 100 f 20 ps). The signals accompanying electron pulses longer than a few hundred ps are strongly affected by the pumping speed of the system. Fig. 7a shows the NO signal accompanying 200-ms pulses. The time constant for the rise and

Fig. 7. NO signal accompanying 2OO-ms,400~PA/cm’

pulses of 2500-eV electrons (a) with the beam block out (detecting NO in the background as well as NO emitted directly from the sample); (b) with the beam block in (detecting only NO in the background); and (c) the difference between (a) and (b).

289

Lading edge of NO and 02 TOF Sig~nls-Direct Emission Only 2500 eV, 400 pA/cm’, 2OO.uuPulses

O,,

Timeoui)

T1320K

Time(I-)

Fig. 8. The leading edges of (a) NO and (b) 0, difference signals (corresponding to emission directly from the sample) accompanying 2GO-ms, 400~p,A/cm* pulses of 2500-eV electrons.

decay of the signal in Fig. 7a is about 40 ms, consistent with the measured pumpout time of the vacuum system (the time constant for the decay of pressure transients). To obtain actual emission intensities as a function of time, we blocked the path of NO molecules emitted directly from the sample and subtracted the intensity of the blocked signal (Fig. 7b) from the intensity of the non-blocked signal (Fig. 7a). The resulting difference signal (Fig. 7c) is quite ‘ ‘ square’ ’ on this time scale. However, on a faster time scale, the leading edge of the NO signal rises gradually, as shown in Fig. 8a. A similar subtraction process and curve fitting procedure was applied to the 0, signal produced by a 200-ms pulse; the leading edge of the difference data is displayed in Fig. 8b. Although the background subtracted 0, signal rises more slowly than the NO signal, it eventually becomes more intense than the NO signal. Fitting the model of Eqs. (l)-(3) to the 0, data yielded r= 75 + 30 ps, with the prompt component making up 40 f 10% of the total emission intensity. Although the noise in the NO data precludes reliable estimates of the model parameters, a satisfactory fit of the NO TOF data is provided by T= 75 p..s (same as for the 0, emission) and a prompt component of 90% of the total NO emission. These and many other curve fits to similar TOF curves all require both prompt and delayed emission, with a time constant r for the delayed component of about 100 t~,s. The neutral Na emissions accompanying electron bombardment are relatively weak. As noted below, the Na Is XPS spectrum of electron bombarded NaNO, also shows no evidence for metallic Na. Apparently, most of the Na+ in irradiated NaNO, is incorporated into stable oxides and

J.-J. Shin et al. / Nucl. lnsir. and Meth. in Phys. Rex B 103 (19951284-296

290

nitrites. Nevertheless, preliminary results indicate that extensive bombardment of single crystal NaNO, (e.g., total dose = 1 C/cm’ at high duty cycles) produces a dark coloration consistent with the formation of metallic Na colloids; this coloration disappears during a 520 K anneal. Measurements of the neutral Na signal produced by rapidly heating NaNO, after electron irradiation showed a small, broad peak at - 500 K; subsequent heating cycles (without additional electron irradiation) yielded progressively smaller Na desorption peaks. This response to a 500 K anneal (progressive desorption vs a single desorption peak) is similar to the behavior of electron irradiated NaCl [ 13,141; the thermal desorption of Na during subsequent heating cycles therefore appears to involve neutral Na produced by the diffusion of defects from the bulk.

H+ Emission vs Pulse Repetition Rate l-p.9 pulses, 2500 eV. 400 pA/cm*

0 T = 20 C, Pump closet

Frequency (Hz)

Fig. 10. Total H+ emission accompanying I-ks., 400~PA/cm2 pulses of 2500-eV electrons as a function of pulse repetition rate. Measurements are made at high and low temperatures (290 K and 490 K) and at two pumping speeds (low = one pump off).

3.2. H + emission Significant Hf emissions were observed from all samples tested. At the level of sensitivity employed, the only ionic species observed during electron bombardment of “dry” NaNO, (vacuum annealed to remove occluded water) is H +. Despite the relatively intense neutral NO and 0, signals described above, the corresponding positive ion emissions are not observed at this level of sensitivity. The absence of NO+ in particular rules out gas phase ionization as a source of H+, due to the high intensity of the neutral NO signal (even at the shorter pulse widths) and its high cross section for ionization at low electron energies [ 151. The neutral H and H 2 signals were barely measurable with our impact ionization scheme. The Hf intensity as a function of beam current also displayed a linear dependence characteristic of a direct emission process, as opposed to higher-order functional relations that would arise from a gas phase ionization step: [Yield(H+) - I(to produce H or Hz) X I(to ionize)]. Much weaker NaC emissions were observed from some samples before vacuum annealing, suggesting that surface water might facilitate alkali ion desorption. A typical H+ TOF signal produced by I-p,s pulses of 2500 eV electrons from annealed NaNO, is shown in Fig. 9. The H+ signal shows a sharp peak 7 p,s after the

0

10

20 Ihe

xl

40

SO

M

Fig. 9. H+ TOF signal accompanying I-p,s, 400~PA/cm’ pulses of 2X0-eV electrons. The corresponding energy distribution is shown in the inset.

beginning of the electron pulse, and a smaller peak at about 15 ps. The kinetic energies corresponding to these peaks are about 7.5 eV and 2.1 eV, respectively. In a recent study of EDS of H+ from NaNO,, Knutsen and Orlando have also observed a 7.5 eV peak in the HC energy distribution [9]. Emission measurements as a function of temperature, pulse repetition rate, and pumping speed show that background gases do not play a significant role in H+ emission. Signals due to adsorbed background gases characteristically grow stronger with decreasing pulse repetition rate (more time for adsorption), grow stronger with decreasing temperature (higher sticking probabilities), and grow stronger with decreasing pumping speed (higher pressures of background gases). Fig. 10 shows the relative H+ emission intensities as a function of pulse repetition rate at two different temperatures and pumping speeds. Decreasing the pumping speed (by closing the valve to one of two vacuum pumps attached to the system) did not change the HC intensities, while raising the sample temperature from 300 K to 500 K increased the H+ intensities a factor of 4. Both results argue against background gases as a source of hydrogen. As the pulse repetition rate is increased from low values, the H+ intensity remains constant until a pulse repetition rate of 1 kHz and thereafter drops slowly. On time scales less than 1 ms, the emission appears to be limited by the “resupply” of a critical component (e.g., sorbed hydrogen, a defect, or free charge). The time to resupply adsorbed species from background gasses would generally be orders of magnitude higher at the pressures employed in this work. (For instance, the time to form a monolayer of H,O would be about 1000 s, assuming unity sticking probability.) An Arrhenius plot of the H+ intensity vs temperature appears in Fig. 1 la; the apparent activation energy of H + desorption from electron bombardment is about 0.11 eV, close to the activation energy of NO emission. Suspecting that the source of hydrogen involved in the HC emission was water diffusing from the bulk of the sample to the

J.-J. Shin et al. /Nucl.

291

Instr. and Meth. in Phys. Res. B 103 (19951 284-296

Arrhenius Plots of H* J?SD and H,O ISD Mensilks

XPS N 1s spectra a!+a funclion of irradiation

I

,

410

4OB Binding

I

I

406

404

Energy

time

I

402

kV)

1ooo T(C)

Fig. I I. (a) Arrhenius plot of H+ emission produced by I-p,s, 400~p,A/cm* pulses of 2500-eV electrons. (b) Arrhenius plot of water emission from NaNO, during heating (no electron bombardment).

surface, we examined the rate of Ha0 desorption vs temperature (with no electron beam). Fig. llb shows an Arrhenius plot of desorbed Ha0 from the sample (no electron beam) as a function of sample temperature; the apparent activation energy is about 0.3 eV. The contrasting activation energies of electron-stimulated HC and thermally stimulated Hz0 emission suggests that H+ emission is rrot rate limited by the diffusion of water to the surface. Conversely, the close agreement between the activation energies for H+ and NO emission suggests that the rate limiting process for HC emission is the formation of active sates by NO; decomposition. We propose that these active sites react with surface Ha0 to form a hydrogen species (yet unidentified) that can be desorbed by the electron beam. An extensive study of H+ emission from low energy electron irradiation (5-80 eV> of NaNO, will soon be reported by Knutsen and Orlando [9]. This work corroborates our observation of strong H + emission from NaNO, crystallized from aqueous solutions.

Fig. 12. Evolution of the XPS N Is spectra as a function of irradiation time.

Is peak is decreasing, indicating a net loss of nitrogen from the surface-consistent with the release of a nitrogen containing species. A Na compound remains on the surface because an increase in the Na 1s signal is observed (spectra not shown) during irradiation. The Na 1s peak position (1071.8 eV) does not change during irradiation suggesting that little metallic Na is generated. The corresponding evolution of the 0 1s spectra is shown in Fig. 13. As the electron exposure increases, the peak corresponding to fresh NO; decreases and a new feature at 530.9 eV increases in intensity. The latter 0 1s

XPS

0 1s spectra as P function of irrPdlat.Ion Ume

3.3. XPS Fig. 12 summarizes the evolution of the XPS N 1s spectra from compressed NaNO, powder as a function of irradiation time. Unexposed NaNO, yields an N 1s peak at a binding energy of 407.3 eV. Following electron irradiation, this peak decreases in intensity and a new peak appears at a binding energy of 403.7 eV. The peak position of this latter feature is identical to the N 1s peak from fresh NaNO, (spectra not shown). The total area of the N

Fig. 13. Evolution of the XPS 0 Is spectra as a function of irradiation time.

292

J.-J. Shin et al./Nucl. Total N Is NO3-

signal P( P fumttonolhadlatioa

Dose

Instr. and Meth. in Phys. Rex B 103 (1995) 284-296

time

( 10“elcm’)

Time

bnin)

Fig. 14. The total N 1s NO; signal (407.3 eV) as function of time during irradiation.

feature does not correspond to NO;; the 0 1s peak from fresh NaNO, appears at a binding energy of 532.3 eV. Aduru et al. attribute this low binding energy feature to peroxide species (Oi-) [16]. These results indicate that NaNO, decomposes to NaNO, and Na,O, under electron irradiation. NaNO, may also be induced to decompose, since the 403.7 eV peak in the N 1s region eventually decreases in intensity. The intensity of the N 1s peak corresponding to NO; is plotted in Fig. 14 as a function of electron exposure. This signal eventually decays to a low, nonzero value, indicating that a steady state is reached. The decay does not follow a simple exponential; therefore the decomposition of NaNO, does not follow first order kinetics. From the first several data points, the beam damage cross-section in the early stages of irradiation is N 2 X lo-” cm*, indicating that NaNO, is very susceptible to damage by energetic electrons. (On the NaNO, surface, the area per _t ion pair is 2.1 X 10-t’ cm2.) A damage cross section of this magnitude strongly suggests that secondary electrons play a dominant role in the damage process.

4. Discussion 4. I. Neutral NO Cunningham has shown from chemical analysis of NaNO, exposed to high doses of ionizing radiation that the principal products are NaNO, and occluded 0, [ 171. However, careful ESR measurements on KNO, irradiated with X-rays at low temperatures suggest that NO, rather than NO;, is the principal nitrogen-containing product of NO; dissociation at low doses [18], and that NO; and occluded 0, are produced by reactions among the primary decomposition products. Although the ESR evidence for radi-

olytic NO production is somewhat controversial [l], our observations of copious NO during the electron irradiation of NaNO, in vacuum indicate that NO is a major product of electronically stimulated dissociation of NO,. In a parallel study of laser-induced photodecomposition, NO is the only neutral product detected at the lowest laser fluences [7]. The high NO emission intensities, as we11 as the extreme sensitivity to electron beam induced damage indicated by XPS, require that the great majority of the NO; dissociation events involve secondary electrons. Direct excitation of surface NO, into dissociative states by primary and energetic (backscattered) secondaries would readily account for the “prompt emission” of NO and 0,. However, the long tails in the emission intensities require additional emission some tens of ps after the electron pulse, when the secondary electron energies are negligibly small. Cunningham has proposed that low energy secondary electrons induce NO, dissociation in irradiated nitrates by dissociative electron attachment [3,4]: NO;(s)

+ e-+

NO:-

* --) NO(g)

+ O;(s)

+ e-.

(4)

ESR measurements on irradiated KNO, held at 77 K for a week show gradually decreasing NO:- signals, and the growth of plausible decay products, including 0; [ 191, consistent with this mechanism. Dissociative attachment is also consistent with the general consensus that NO; dissociation in X-and -y-irradiated nitrates is due to secondary reactions involving charge trapped at electron excess or deficient centers [l]. This process would also account for the extremely high efficiency of NO; radiolysis in NO,doped alkali halides, where some process of energy Iocalization (e.g., trapping of charge carriers) must be invoked to account for the high yields [20]. We note that the charge state of the products is consistent with their gas phase electron affinities. (The gas phase electron affinities of NO and 0, are 0.026 eV and 0.45 eV respectively, favoring 0; over NO- as the ionic species in Reaction (4). The crystalline environment will change the absolute values of these binding energies but is unlikely to change their order due to the highly ionic nature of the crystal.) Because the electron kinetic energies required for dissociative electron attachment are negligible, the diffusion of charge carriers after the electron pulse would continue to produce NO from surface sites for some time after the end of the electron pulse. This would account for a most unusual aspect of the NO TOF signals: the presence of the “delayed” emission component. As noted above, the lack of a significant change in the NO and 0, TOF as the temperature is raised to 400 K rules out delayed emission due to slow thermal desorption of adsorbed decomposition products. (The reduction in the residence time (corresponding to 7 in Eq. (3)) expected at elevated temperatures should significantly reduce the “delay” in the TOF signals.) We note that NO:production by X-rays is very

J.-J. Shin et al./ Nucl. Instr. and Meth. in Phys.Res. B 103 (1995) 284-296 efficient in NaNO, at low temperatures (> 1 radical per 100 eV at 77 K) [21]. The accumulation of NO:- radicals during the electron pulse and a decay afterwards would readily account for the observed delayed emission (T = 100 ps from the curve fits of Figs. 5 and 7). The temperature dependence of the NO intensities is consistent with previous measurements of the temperature dependence of the NO; yield during low dose y-irradiation of NaNO, (NO; activation energy 0.12 f 0.01 eV) [22]. Nevertheless, many plausible, thermally activated, rate limiting steps can be ruled out in the case of the NO emission intensities. Thermally activated diffusion of NO is not likely rate limiting; the large size of the NO molecule relative to the interstitial spaces in NaNO, prohibits significant diffusion on these time scales. As noted above, the NO TOF signals indicate that thermal desorption is not rate limiting. Furthermore, the activation energy for NO emission is well below the energies required for the diffusion of interstitials and vacancies in NaNO, (typically l-2 eV for interstitials and about 0.75 eV for vacancies, respectively) [23]; thus the diffusion of interstitials and vacancies is not rate limiting. We therefore consider the role of thermal vibrations. Molecular vibrations can strongly affect electron trappmg, and would therefore affect the kinetics of NO emission by Reaction (4). Electron trapping at NO; to form NO:- involves a transition from the planar NO, geometry to the pyramidal geometry of NO;-; this change in geometry would naturally be assisted by the appropriate molecular vibrations [3]. Significantly, the activation energy for NO emission is comparable to the energies required to excite internal vibrational modes in NO;, which range from 0.09 to 0.18 eV [24,25]. This agreement strongly supports Cunningham’s NO:- * intermediate as an important route to decomposition. We therefore envision NO emission by the following process. Primary electrons and energetic secondaries excite some surface NO; ions into directly dissociative states, accounting for much of the “direct” emission. A large number of low energy secondaries are also produced, some of which diffuse back to the surface at a rate determined by the charge mobility and charge distribution as a function of depth. Electron transport via the NO; rr * state (the lowest unoccupied molecular orbital of the NO; ion), for instance, would not be thermally activated. Electrons arriving at the surface then produce NO by dissociative electron attachment at a rate proportional to the density of the relevant NO; vibrational states. Dissociative attachment involving secondaries created very near the surface may also contribute to the prompt component of emission. Assuming that charge diffusion does not depend on temperature, the time-dependence of the emission intensities will not depend strongly on temperature. However, temperature does affect the vibrational state densities, and thus strongly affects the dissociation probability and thus the NO yield.

293

4.2. Neutral 0, and other neutral emissions The 0, emission intensities display a complex dependence on pulse width and temperature, and thus must involve a different reaction sequence. We point out that the NO; production in X-and y-irradiated nitrates is attributed to a reactions of the form NO;(b)

+ NO;(b)

+ O(b),

(5)

where (b) denotes species either in the bulk or at the surface. Occluded 0,, as well as NO;, is observed in irradiated nitrates, suggesting that Reaction (5) is relevant. However, XPS 0 Is signals suggest that little NO; is produced in the near surface region. The absence of significant emissions of atomic 0 also argues against the production of 0, (g) via an atomic 0 intermediate. Finally, the strong O$- peak in the XPS 0 1s signal indicates that a large fraction of the 0; produced by NO emission is removed. Although it is difficult to account for all the relevant aspects of 0, emission with a simple model, we propose that the majority of the observed emissions result from a second electron attachment reaction which has two branches: 0, production (6a) and Oi- formation (6b): 0;

+ em--+ O,(g)

0;

+ e--t

Ot-,

+ F-center + e-,

(6a)

(6b)

where the F-center is an NO; vacancy with a bound electron. Estimates of the energy required to form Oiindicate that the Madelung energy of the NaNO, lattice (7.64 eV> is almost sufficient to stabilize a second electron on the 0; ion. (The second electron affinity of gas phase 0, is 7.70 eV.) Despite the crudeness of a purely electrostatic model of the NaNO, lattice and a gas phase approximation for the 0, ions, this suggests that the energy required to form Oz- from 0; is small. Thus, Ozformation from 0; formation appears to be energetically feasible, and that a reasonable amount of lattice relaxation about an incipient 0$-(s) ion would stabilize the second electron on the oxygen ion. At short pulse widths, significantly less 0, is emitted than NO, suggesting that O:formation is initially favored over 0, emission. Initially, O;- formation would essentially consume secondary electrons. Neglecting the small electron excess due to primary electrons, the consumption of electrons must be balanced by the accumulation of positively charged defects (e.g., Ff-centers -NO; vacancies without bound electrons). Assuming that the defect mobility is sufficiently low, electron diffusion toward the surface would build up an electrical dipole with excess negative charge at the surface itself. This dipole field would raise the electrostatic energy required to form the Oz- ion, and the surface Oz- concentration would saturate at the level where the Oi- ion is barely stable. Further reactions would then yield O,(g) alone. In the limit

294

J.-J. Shin et al./Nucl.

lnstr. and Meth. in Phys. Rex B 103 (1995) 284-296

of long pulses. most of the NO emitted from the surface produces 0,, and the emission intensities become very nearly equal, i.e., the dipole field reaches a high steady state level at which little additional O:is produced. During the relatively long period between electron pulses, the charge distribution in the subsurface region would decay to normal levels, allowing for Oz- production at the onset of the next pulse. Favored Oiproduction in the short pulse limit implies that it would also be favored in the initial stages of the longer pulses. Thus, the leading edge of the 0, signal would lag somewhat behind the leading edge of the NO signal at all pulse widths which is readily observed in the TOF data of Figs. 5 and 6 and 8. In terms of the model used to fit the TOF data of Figs. 6 and 8, there also was significantly smaller prompt emission components for 0, relative to NO. Given the present signal-to-noise and time resolution of our data, we have not pursued adding another parameter to account for this additional delay. Another process necessary to maintaining an equilibrium stoichiometry is the diffusion of oxygen-containing species from the bulk. In the absence of an anion-replacement process, the accumulation of F-centers produced by NO and 0, emission would eventually yield neutral Na in the form of intense neutral Na emissions and/or Na colloids. However, the observed neutral Na emissions are far too weak to balance the observed NO and 0, emissions, and XPS measurements rule out significant accumulation of neutral Na on the surface. Diffusion of NO; from the bulk would account for the small but significant equilibrium concentration of NO; observed by XPS after extensive electron irradiation. This process would also account for the ability of NaNO, surfaces to yield sustained and reproducible NO and 0, emissions after extensive irradiation. The apparent activation energies corresponding to Na and NO, emission are very close to that for NO, suggesting that the rate limiting steps in these emissions involve similar reactions. For instance, localized F-center accumulation may still produce small amounts of neutral Na, which can desorb thermally. Similarly, small amounts of NO, may be produced by ionization of NO; produced by Reaction (5). As noted above, the temperature dependence of NO; production is believed to be very similar to that observed for NO emission in this -and may indeed involve some of the same reaction steps (e.g., dissociative electron attachment). In contrast, the complex temperature dependence of the 0, intensities is consistent with the competing reactions outlined above. In particular, the equilibrium surface charge of ionic materials of commercial purities is expected to reverse in sign as the temperature is raised [26]. In MgO, thermally activated hole release at high temperatures can strongly affect the surface charge densities [27]. In NaNO,, hole transport to the surface would lower the electrostatic energy required for O$- production and thus lower the 0,

emission intensity as the temperature is raised. Nevertheless, at sufficiently high temperatures (> 380 Kl, the steadily increasing NO emission intensities apparently allow for enhanced 0, production despite the presence of competing reactions. We note that the formation of peroxyonitrite ions (ONOO-) (responsible for the yellow color of UV-irradiated nitrates [28]) may also provide a temperature-sensitive, alternate pathway to 0, emission at low temperatures. The production of F-centers by NO and 0, emission is potentially of great interest. Although the surface F-center concentration may be small ( < 10T3 surface sites and thus hard to detect), F-centers may play an important role in the H+ emission discussed below. F-centers produced by electron irradiation may also help account for the enhanced susceptibility of electron-irradiated NaNO, to UV laser damage at 248 nm [7]. 4.3. Ion emission Despite many studies of H+ ESD, little has been done on H+ emission from ionic solids. However, Knutsen and Orlando have recently completed a study of low energy electron EDS of ionic species from NaNO, [9], in which they observed intense Hf emission whose peak energy increases from about 5 eV to about 7.5 eV as the incident electron energy increases from 35 to 65 eV. The width of this peak also increases with incident electron energy, mostly due to broadening in the high energy side of the energy distribution. They report a weak threshold in electron energy for H+ emission at 24 eV and a strong increase in the yield per incident electron at 33 eV. They attribute these emissions to core hole excitations of the O(2s) level of a terminal OH group (near 24 eV) and of the Na(2p) level (near 33 eV). The energy distributions in the present work peak at 7.5 eV. but are significantly broader (especially on the high energy side) than those reported by Knutsen and Orlando. Although the reasons for this broadening are not clear. it is consistent with trend established in their work, where higher incident energies were associated with broader peaks. The strong temperature dependence of HC yield indicates the presence of a precursor which is formed by a thermally activated or assisted process. Since the activation energy for H+ emission agrees well with the activation energy for NO emission, NO, dissociation would seem to be involved in the formation of this precursor. Although there are many possibilities, defects involving NO, vacancies (F-centers) are likely hydrogen binding sites. F-centers interact strongly with electronegative atoms and molecules, including atomic H. In addition, F-center aggregation may produce isolated clusters of highly reactive neutral alkali atoms; this process is readily observed in irradiated alkali halides, for instance [29]. The observed coloration after extensive electron bombardment suggest that similar colloids may be produced in NaNO,.

J.-J. Shin et al. /Nucl.

Instr. and Meth. in Phys. Res. B 103 (1995) 284-296

Electron stimulated H+ desorption from metal and semiconductor surfaces is strongly enhanced by submonolayer coverages of alkali metals [4,5]. Yasue et al. report a similar enhancement of ESD H+ yields from Sic1 11) due to submonolayer Na coverages [6]. They observe HC energy distributions similar to those reported here, with three distinct components peaking at 2.9, 4.4, and 6.7 eV, respectively. The highest energy component is attributed to Coulomb repulsion following an Auger transition involving Na 2p electrons; the 4.4-eV component is attributed to a bonding-to-antibonding transition within an Na-H stmcture, followed by ionization. Such a structure on the NaNO, surface, derived by F-center aggregation and populated by reaction with diffusing water, would account for the observed H+ emission. This binding site would have to be stable to at least 480 K to be consistent with the high emission intensities at elevated temperatures in Fig. 11. We note that alkali metal aggregates in crystals such as NaCI are thermally stable up to 500 K [30].

295

understood in terms of doses at least as high as those employed in this study. Electron beam induced NO, dissociation produces very reactive species, including NO, 0,, and atomic Na, which could drive unusual chemical reactions in the complex mixtures that make up typical high level radioactive wastes. Although the electron energies employed in this work are sufficient for direct ESD of H+ from adsorbed water, this does not appear to be the principal mechanism for H+ emission in vacuum. The similarity between the thermal activation energies of Hf and neutral NO emission suggests instead that the rate limiting step is the production of associated surface defects, e.g., surface F-centers or neutral Na; subsequent interactions between water and these defects would create an Na-H structure from which H’ could be desorbed in a known fashion. The possible production of colloidal Na needs to be verified; in the context of radioactive waste containing large quantities of NaNO,, metallic Na would be of some concern due to reactions with water to produce H *.

5. Conclusions Acknowledgements NaNO, readily decomposes under l-3 keV electron bombardment in vacuum to yield NO and 0,. Small amounts of NO, are also observed as well as some H,. NO is produced by the thermally assisted, electronic dissociation of the NO; ion. A major component of the NO emission is delayed relative to the onset and termination of the electron pulse. This requires a dissociation mechanism involving long-lived electronic excitations. Dissociative electron attachment via a vibrationally excited state would account for this delayed emission, where electron transport to surface (via mobile electrons or excitons) is the rate limiting step. Direct excitation of dissociative states may play an important role in the more “prompt” NO emission. The high NO yields per incident electron require that secondaries play a dominant role. We attribute the emission of other species (including 0,) to reactions involving the products of NO; dissociation, e.g., 0;. As emphasized in the introduction, defects play an important role in many of the emission processes observed in this work. Reproducible results are obtained only because modest exposures produce a reasonably stable, equilibrium stoichiometries. Defects appear to be especially important in the emission of O,, where the emission intensities fall to very low levels in the limit of short pulses. As noted above, the defect interactions which cause and are caused by 0, emission are quite complex. Nevertheless, they must be understood in order to account for consequences of exposure to high electron doses. Recent work has shown that electron doses similar to those employed in this work strongly enhance the laser-induced emissions from NaNO, at 248 nm. Further, the effect of ionizing radiation due to the decay of radioactive components of high level radioactive wastes must surely be

The authors would like to thank Tom Orlando and Wayne Hess, Pacific Northwest Laboratory, and Ted Madey, Rutgers University, for helpful discussions. This work was supported by the Department of Energy under Contract DE-FG06-92ER142.52, the Pacific Northwest Laboratory under the Department of Energy’s Strategic Environmental Research and Development Program, and the Washington Technology Center.

References

[I] E.R. Johnson, The Radiation-Induced

Decomposition of Inorganic Molecular Ions (Gordon and Breach, New York, 1970). [2] Yu.A. Zakharov and V.A. Nevostruev, Russ. Chem. Rev. 37 (1968) 143. [3] J. Cunningham, in: Radical Ions, eds. E.T. Kaiser and L. Kevan (Interscience, New York, 1968) pp. 475-523. [4] D.L. Doering, S. Semancik and T.E. Madey, Surf. Sci. 133 (1983) 49. 151 A.-M. Lanzillotto, M.J. Dresser, M.D. Alvey and J.T. Yates, Jr., J. Chem. Phys. 89 (1988) 570. 161 T. Yasue, A. Ichimiya and S. Ohtani, Vacuum 41 (1990) 561. 171 R.L. Webb, S.C. Langford, and J.T. Dickinson, this issue, Nucl. Instr. and Meth. B 103 (1995) 297. [81 R.D. Ramsier and J.T. Yates, Surf. Sci. Rep. 12 (1991) 246. 191 K. Knutsen and T.M. Orlando, Low energy (S-80 eV) electron-stimulated desorption of H+, OH+, and Na+ from solution grown NaNO, crystals, submitted to Surf. Sci. I101 W.H. Press, B.P. Plannery, S.A. Teukolsky, and W.T. Vetterling, Numerical Recipes in Pascal: The Art of Scientific Computing (Cambridge University. Cambridge, 1989) pp. 572-580.

296

J.-J. Shin et al. / Nucl. Instr. and Meth. in Phys. Res. B 103 (1995) 284-296

[I I] P.A. Redhead. J.P. Hobson, and E.V. Komelsen, The Physical Basis for Ultrahigh Vacuum (Chapman and Hall, London, 1967) pp. 159-163. [12] Riki Kawashima and Kazuo Suzuki, J. Phys. Sec. Jpn. 52 (1983) 1857. [I31 M. Szymofiski, J. Ruthowski, A. Poradzisz, Z. Postawa, and B. Jergensen. Desorption Induced by Electronic Transitions, DIET II (Springer, Berlin, 1985) pp. 160-168. [14] G. Betz, E. Wolfrum, P. Wurz, K. Mader, B. Strehl, W. Husinsky, R.F. Haglund. Jr.. and N.H. Tolk, Desorption Induced by Electronic Transitions, DIET III (Springer, Berlin, 1988) pp. 278-283. [I 51 Earl W. McDaniel, Atomic Collisions: Electron and Photon Projectiles (Wiley, New York, 1989) p. 386. [16] S. Aduru, S. Contarini, and J.W. Rabalais, J. Phys. Chem. 90 (1986) 1683. [ 171 J. Cunningham, 1. Phys. Chem. 67 (1963) 1772. [ 181 J. Cunningham, J. Phys. Chem. 66 (1962) 779.

[19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] 1301

J. Cunningham, J. Phys. Chem. 66 (1962) 779. A. Russell Jones, J. Chem. Phys. 35 (1961) 751. J. Cunningham, J. Phys. Chem. 71 (1967) 1967. J. Cunningham. J. Phys. Chem. 65 (1961) 628. P. Cerisier and P. Gaune, J. Solid State Chem. 6 (1973) 74. D. Smith, D.W. James, and J.P. Devlin, J. Chem. Phys. 54 (1971) 4437. L. Angeloni and R. Righini, Chem. Phys. Lett. 154 (1989) 115. K.L. Kliewer and J.S. Koehler. Phys. Rev. 140 (1965) A1226. M.M. Freund, F. Freund, and F. Batllo, Phys. Rev. Lett. 63 (1989) 2096. R.C. Plumb and J.O. Edwards, J. Phys. Chem. 96 (1992) 3245. H.W. Etzel, Phys. Rev. 118 (1960) 1150. W. Hayes and A.M. Stoneham, Defects and Defect Processes in Nonmetallic Solids (Wiley, New York. 1985) pp. 305-307.