- Email: [email protected]
Deuterium uptake in titanium thin films: t ssc00621 the effect of oxide, and the metal (Ti and Fe) overlayers
Surface Science 160 (1985) 235-252 North-Holland, Amsterdam
235
D E U T E R I U M U P T A K E IN T I T A N I U M T H I N FILMS: t ssc00621THE E F F E C T O F O X I D E , A N D M E T A L (Ti A N D Fe) OVERLAYERS Michael C. B U R R E L L * a n d Neal R. A R M S T R O N G **
Department of Chemistry, University of Arizona, Tucson, Arizona 85721, USA Received 3 December 1984; accepted for publication 5 April 1985
Titanium thin films prepared in UHV were reacted with deuterium (PD2 < 1 X 10-5 Torr) to various loadings, as determined by microgravimetry using a quartz crystal microbalance. The kinetics of deuterium absorption favor a mechanism in which an a-phase surface deuteride forms on the film during the early stages of the reaction, resulting in a constant rate of deuterium uptake during most of the reaction. Surface characterization by AES and ELS, however, demonstrated spectral changes which were dependent on the bulk film stoichiometry. Electron-beam decomposition of the surface deuteride during AES analysis is postulated to explain this result. Oxidation of the titanium film surface caused a decrease in the deuterium absorption rate, completely inhibiting the reaction when oxides of thickness 20 ,~ or greater were formed. Fresh titanium layers on top of the oxide renewed the ability of the Ti film to take up D2 at the previous rate. Iron adlayers were found to accelerate the D2 absorption rate of Ti films, or to likewise reactivate oxidized Ti surfaces.
1. Introduction D u r i n g the past decade, metal hydrides have received considerable a t t e n t i o n in various technological applications. Metals which react rapidly with hydrogen to form solid hydrides have been used as getter p u m p s for h y d r o g e n isotopes in prototype fusion reactors. Metal hydrides have also been used as n e u t r o n energy m o d e r a t o r s [1,2]. Recently there has been increased a t t e n t i o n directed towards metal hydrides as media for h y d r o g e n storage a n d transportation for ultimate use as a fuel [3], or for other energy conversion schemes [4]. T h e most p r o m i s i n g materials for storage applications are the intermetallic alloys FeTi, Mg2Ni, L a N i s , Z r M n 2 , TiMnl.5, a n d related systems [1-3]. The rates of hydrogen a d s o r p t i o n a n d desorption in metal hydride systems are extremely sensitive to the c o m p o s i t i o n of the surface, particularly the presence of c o n t a m i n a n t s or oxide layers. Most studies have c o n c e n t r a t e d on * Presently at General Electric Company, Corporate Research and Development, Schenectady, New York, USA. ** To whom correspondence should be addressed.
0 0 3 9 - 6 0 2 8 / 8 5 / $ 0 3 . 3 0 © Elsevier Science Publishers B.V. ( N o r t h - H o l l a n d Physics Publishing Division)
236
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
the thermodynamic properties of these materials in bulk form, and only a few studies have appeared in which the surface composition of hydrides have been characterized under controlled conditions. One reason for this is the high equilibrium hydrogen pressures often necessary for hydride formation, which precludes the use of conventional vacuum surface spectroscopies. The hydride of titanium, Till 2 is readily formed and thermodynamically stable in ultrahigh-vacuum environments and lends itself to studies of surface composition of the hydride. Malinowski [5,6] has investigated the effect of contaminants on the deuterium pumping speed of titanium films, and demonstrated that Auger electron spectroscopy (AES) (in the derivative lineshape mode) can be useful for identifying the hydride composition as well as surface contaminants. Others have shown that the hydrides of titanium (TiHx; x ~<2) are characterized by surface electronic properties distinct from the pure metal using Auger electron, photoelectron, and electron energy-loss spectroscopies (ELS) [7,8]. The hydrogenation kinetics of the alloy FeTi are also dependent on the initial cleanliness of the surface [9-11]. The presence of a readily formed oxide layer prohibits hydrogen absorption and must be removed by an "activation" procedure. As for most hydride forming materials, this involves repeated heating of the FeTi alloy to temperatures in excess of 400°C in vacuum or under a hydrogen atmosphere, which apparently dissolves the surface oxide. Formation of FeTi--H requires an equilibrium hydrogen pressure of several atmospheres and surface studies of this material in vacuum environments is not possible. Recently, however, evidence has been given [11] that heating FeTi in a vacuum environment may result in the segregation of pure titanium ( - 100 monolayers) to the alloy surface at 800°C. Thus, the hydrogen adsorption reactions occurring at titanium film surfaces may be representative of similar reactions occurring at activated FeTi alloy and other alloy surfaces under actual hydriding conditions. It is also of interest in this study to determine the effect of adding small adlayers of Fe to the atomically clean or partially oxidized Ti surface, as a way of approximating the surface chemistries of the activated FeTi alloy. Oxide layers which readily form on Till 2 and other reactive metal hydride surfaces exposed to air, O 2, or H20 control the rate and extent of subsequent adsorption or release of hydrogen. Using only quartz crystal microgravimetry (QCM), Kasemo and Tornqvist [12] have shown that the oxide formed on titanium film surfaces drastically slows the hydrogen adsorption reaction, but that evaporation of small amounts of fresh Ti onto the oxide surface restored the hydrogen absorption capability of the entire material. The oxide prevented hydrogen dissociation on the surface, but apparently did not act as a diffusional barrier. Malinowski has observed that the oxide formed on TiD~ surfaces inhibited the release of deuterium by thermal decomposition; films with oxidized surfaces had to be heated to higher temperatures before the
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
237
material decomposed [7]. The electrochemical behavior of TiH~ films, in particular the anodic potential at which hydrogen is released, was also found by us in earlier studies to depend on the extend of the oxide layer on the surface [13]. These results indicated that an understanding of the reactivities of titanium hydride surfaces required first a knowledge of the composition and thickness of the oxide layer that in many instances isolates the material from its surroundings. Other studies in our laboratory [14,15] have demonstrated that it is possible to form an oxide layer at room temperature and low pressures (10 -8 Tort < Po2 < 1 atm) whose thickness and composition can be determined in a quantitative fashion using AES and microgravimetry (QCM), and ranges in thickness from
10-40 h. In the studies described in this paper, titanium deuteride films (TiDx; x = 0.5 to 2) were prepared and their surfaces characterized by AES and ELS. Using data treatment procedures developed in this laboratory for N(E) Auger spectra [16], we show that the lineshapes of the Ti LMM series (particularly the LMV transition) are even more sensitive to the presence of the hydride in the lattice. A very distinctive lineshape change (chemical shift) can be observed when hydrogen is introduced due to its contribution to the valence band density of states [17]. We have also found, however, that electron-beam damage effects are an important consideration in determination of hydriding levels by AES and these are discussed as well. The quartz crystal microbalance (QCM) was used to monitor the total deuterium uptake, and hence the absolute deuterium content of the entire thin film was known accurately and could be varied systematically. These films were subsequently reacted with oxygen, and the rate and extent of surface oxidation were determined from the mass uptake data, and the oxide characterized using AES. Rates of hydrogen uptake with and without oxide, with and without Ti and Fe overlayers were examined. The effect of deuterium loading (from zero concentration up to TiD2) on the surface oxidation reaction is also discussed.
2. Experimental All measurements were performed in a Physical Electronics 540A thin film analyzer with a base pressure of 10-~0 Torr. Auger spectra were collected in the direct current mode using a previously described system [18], using a 5 kV, 0.5/~A electron beam. The quartz crystal microbalance (QCM) has also been described in previous publications [14,15,19]. It consisted of a 1 / 4 inch diameter AT-cut crystal with a 10 MHz fundamental resonance frequency. Gold electrodes were deposited onto either side of the crystal, and were contacted by wires leading to an external gate oscillator circuit, whose output was measured with a digital counter. The QCM mount was rotatable within the
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
238
vacuum chamber, such that it could alternatively face the AES optics and the Ti evaporation source. Titanium films were deposited onto the exposed face of the QCM from a wire filament (Alfa), under UHV conditions. Changes in the QCM frequency, after sufficient cooling following evaporation, yielded the film thickness. Subsequent exposure to gaseous deuterium or oxygen (Matheson) yielded further frequency shifts, which were measured as a function of time, and related to adsorbed mass by the Sauerbrey relation [20]:
A m / A F = 4.4 X 10 -9 g / H z - c m 2.
(1)
Auger spectra and ELS were measured before and after the gas dosing experiments described previously [14]. AES data were treated to remove inelastic electron scattering contributions to the background by a previously described deconvolution procedure using the ELS spectra obtained in the region of the Auger spectrum [16].
3. Results and discussion
3.1. Preparation of TiDx films Fig. 1 shows the QCM response of a typical thick (at least 1000 A), freshly evaporated clean titanium film, exposed to deuterium at a pressure of 1 × 10-5 Torr. Within 30 min, the film became saturated and the rate of D 2 uptake
TI g00
I
+
02
1000"
~3'
N
cq I
/
13
6130
EXPOSURE
12(3(3 TIHE,
18~[]
2~OB
~eo
Fig. 1. Deuterium uptake (frequency change in the QCM) versus time for a clean titanium film. The region up to point 1 corresponds to the conversion up to TiD0. 3. The region up to point 2 corresponded to conversion up to TiD1. s. The limiting TiD 2 composition was reached at point 3.
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
H-H
239
~H+ H
///////
//~/I
/ H y.dride////~)I M eta I a
/
1
b Fig. 2. Models for hydride formation. (a) Surface hydride model, (b) Precipitate model.
leveled off. The D / T i ratio was calculated from the frequency shifts due to titanium deposition and subsequent deuterium uptake at saturation, and the atomic weights of Ti and D:
ND NTi
Afo/2 AfTi/47 .
(2)
For the film shown in fig. 1, the frequency shifts were AfT i = 10000 Hz (from the Ti deposition stage) and Alp = 870 Hz (at point 3, fig. 1, from the deuteriding stage), indicating formation of TiD z throughout the film. The adsorption rate decreased monotonically from the clean surface value as the film was converted to an overall D / T i ratio ~< 0.3 (point 1 in fig. 1) consistent with previous studies by Kasemo and Tornqvist [12]. The rate was then approximately constant with an apparent sticking coefficient of - 0 . 0 4 (at hydrogen pressures of 10 -5 Torr) until the D / T i ratio approached 1.8 (point 2 in fig. 1), after which the rate decreased until the limiting composition TiD 2 was attained (point 3). If the system was evacuated during the adsorption regime where the rate was constant (between points 1 and 2), a mass loss was observed corresponding to desorption of a small fraction ( < 5%) of the hydrogen content of the film. The desorption rate was insensitive to the total hydrogen content in the film for overall D / T i ratios from 0.5 to 1.8. A desorption of H z in this manner has been shown to fit a simple associative/desorption rate law [12]. In previous studies by Kasemo, the adsorption/desorption rates and the total amount of desorbed H2, did, however, increase with the hydrogen pressure used in the preceding adsorption cycle. Two models may be proposed for the conversion of a titanium film to the
240
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
hydride (or deuteride), as shown in figs. 2a and 2b. The absorption/desorption kinetics favor a surface hydride model, (fig. 2a), in which the reaction rate is controlled by diffusion of dissociated hydrogen through a surface hydride layer. The observed desorption of H 2 (or D2) from the Ti film, when the gas exposure is halted, may be due to super-saturation of the surface hydride layer. An alternative mechanism in which the hydrogen concentration changes uniformly throughout the sample (and in the surface region), during the hydriding process (fig. 2b) is inconsistent with our observations and has been previously discounted by others [6,7,12]. First, no desorption was observed if the hydriding reaction was interrupted prior to the constant-rate adsorption region. Presumably the initial D 2 adsorbed (50-100 monolayers) results in the formation and coalescence of the surface hydride. Secondly the adsorption/desorption kinetics did not change as the total D / T i ratio in the film increased from 0.5 to 1.8, indicating that the surface is not changing significantly during this stage of reaction. (If the composition was changing uniformly throughout the film, the rate of D 2 uptake would change with exposure). The T i / H phase diagram when extrapolated to room temperature, exhibits a solution c~-phase for H / T i ratios up to ca. 0.001, and a 7-phase for H / T i ratios of 1.5-2 [1,2]. Films with intermediate hydrogen concentrations therefore contain a mixture of both a- and 7-phases. In the surface hydride model of metal hydride formation (fig. 2a), the a 7 phase transition occurs more rapidly on the surface than in the bulk, and a 7 hydride layer forms on the surface. Subsequent hydriding must occur through this layer. One would expect to observe only 7 hydride of the surface for films hydrided to any intermediate concentration Till x (x = 0.1 to 2). To test this hypothesis, atomically clean titanium films were prepared and reacted to various deuterium loadings. Surface characterization by AES and ELS were performed in situ immediately following this loading.
3.2. AES results The d ( E N ( E ) ) / d E Auger spectra were first recorded for pure titanium, TiD0.v, TiDa. 3 and TiD 2 thin films. As noted by Malinowski and others, changes in the peak fine structure distinguish metal from TiD 2 and intermediate hydride compositions [7,8,13,17], and can be interpreted in terms of valence band redistribution as deuterium is added to the lattice [13]. These changes in peak shape can be summarized as follows: (1) A shoulder appears on the low-energy side of the M2,3VV peak (27 eV), which eventually becomes a doublet as the bulk composition approaches TiD 2. (2) The L2,3M2,3M2, 3 peak (380 eV) becomes broader, and the doublet structure in this peak disappears. (3) The L2.3M2,3V peak (410 eV) decreases in magnitude (with respect to the LMM peak), and a prominent shoulder appears on the low-energy side. (4) A shoulder also appears on the low-energy side of the L2,3W
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
Ti D2 A L M V
.2
l
N(E)
241
L
l
(b)
NE}
(a) 3~113
45~ 401~ 35~ ELECTRON K I N E T I C ENERGY, eV
501~
400
eV
420
Fig. 3. Ti(LMM) Auger spectra for (a) clean Ti and (b) full saturated, TiD 2. The original EN(E) spectra were treated by a deconvolution procedure outlined in ref. [16]. Spectral features (I) and (II) explained in text.
Fig. 4. Ti(LMV) spectra of TiD2. Spectrum (a) has been treated to remove the secondary cascade from electron sources of higher energy. Spectrum (b) represents the result of an FFT deconvolution of (a) with an energy-loss function obtained with the electron beam at 400 eV. Spectral features (I) and (II) explained in text.
peak (460 eV). The new spectral features appear gradually as the overall composition approaches TiD 2. This observation suggests (in contrast to the surface hydride model) that the surface composition within the sampling depth of the AES experiment is related to the bulk film composition and does not approach stoichiometric TiD 2 until the entire film has been converted. Lamartine, Haas, and Solomon [17] and previous work from this laboratory [13], documented studies of titanium hydride and deuteride surfaces of various compositions using AES, XPS, and ELS. Those studies differed from this study because the samples were not prepared in the same vacuum chamber where analysis occurred. The hydrides were ion-sputtered prior to analysis to remove the surface oxide. The AES results, however, were essentially identical to those observed here. The Ti (2p) binding energy in the XPS results shifted towards higher energies as the hydrogen or deuterium content increased. The bulk plasmon energy in the ELS shifted to larger energies, again related to the overall bulk composition of the specimen. We have extended these results to AES data collected in the EN(E) mode. Following correction for the C M A transmission efficiency a sequential back-
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
242
ground subtraction correction for the secondary electron background and the inelastic loss contribution was applied to the data [16], yielding the spectra shown in fig. 3. The Ti (LMV) peak in the TiD 2 clearly shows the additional low-energy shoulder associated with hydrogen uptake. A higher-resolution LMV peak in TiD 2 was obtained from the original data by deconvoluting an instrument-specimen response measured at E 0 = 400 eV, using a fast Fourier transform (FFT) technique developed in this laboratory [21]. The initial spectrum of the LMV peak is shown in fig. 4a and, the resultant deconvolved spectrum is shown in fig. 4b. This FFT procedure removes the sample and instrumental contributions to the peak broadening and helps resolve the shoulder on the low-kinetic-energy side. The high-kinetic-energy component (I) in the LMV spectrum figs. 3b and 4b, in a simple one-electron model, represents ejection of an Auger electron originating from the Ti 3d valence level (following L-shell ionization). The lower-energy component (II) arises from a valence Auger electron ejected from a level of H(ls) character. The observed energy difference of 5 eV corresponds to the measured difference of the two valence band maxima observed in UPS measurements of Till 2 [17,22,23]. 3.3. E L S results
Changes in the electron energy-loss spectra (ELS) also occurred between pure and fully deuterided titanium films, as shown in fig. 5. The data shown were obtained using primary beam energies of 100, 300, 500, and 1000 eV (curves a through d, respectively). The use of low-energy electron beams emphasizes features due to surface processes, while higher-energy electron beams give spectra which emphasize bulk processes. The feature noted in the
TiD 2
N(E
c
1 30
Energy
20
10
4 0
3 2
1
13,0 2,oI ol ko
Los s,eV
Fig. 5. Energy-loss spectra for Ti and TiD 2 films obtained at electron beam energies of 100, 300, 500 and 1000 eV (curves a through d respectively). All data are plotted as energy loss from the primary energy ( F W H M of electron beam ca. 800-900 mV). Spectra features 1-4 and 1'-4' are detailed in the text.
M.C Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
243
ELS of clean titanium (feature 1, right side of figs. 5 and 6) at about 7 eV is most likely an interband transition; its intensity was rapidly quenched by even small doses of D 2. The features 2, 3 and 5 at 11-12, 16-17, and 32 eV respectively arise from surface and bulk plasmon excitations, and M2, 3 ionization, respectively. In the ELS from TiD 2 (left side of fig. 5) several changes are noted. A new peak appears at 4-5 eV (feature 1% a and b) which we attribute to excitation of an electron from the valence band maxima at - 5 eV (H (ls) - like band) to the conduction band. This peak is at least partially of bulk origin, because it is still present (although not totally resolved) even when incident beams of 500 eV were used. The surface and bulk plasmons were shifted to higher energy-loss values of 12-13 eV (feature 2') and 19 eV (feature 3'), respectively. The M2, 3 (3p) ionization edge shifted from 31.3 to 32.4 eV (feature 4'), as determined from ELS data of higher resolution than shown here. This is consistent with partial charge transfer from Ti to D observed in XPS Ti (2p) core level chemical shifts [13,17]. The changes in the ELS measurement (from Ti- to TiD2-1ike spectra), like the changes in the AES spectra, occurred gradually and monotonically as the deuterium content of the film was increased. 3.4. Hydriding mechanisms Two possible explanations may be offered to explain why a surface hydride layer of constant composition is not apparent on the surface of titanium films deuterided in situ to intermediate compositions, TiD x (x < 2): (1) The electron beam used to generate the AES and ELS data may induce decomposition of the deuteride due to localized heating or electron-stimulated desorption [21]. The deuterium content in the region of the beam may be depleted by these processes, but due to the high diffusivity of deuterium atoms, is partially replenished by deuterium from the bulk [6]. The measured Auger spectra may reflect the steady-state deuteride composition under electron-beam irradiation, which would be expected to vary with the bulk composition. Although beam currents of 0.5 /~A or less were used to avoid altering the surface (at 0.5 ~A, 5 kV beam, 0.1 mm spot size - power d e n s i t y = 25 W / c m 2), the damage threshold for these materials may be much lower than expected for other materials (such as metal oxides [24]). The heating caused by the electron beam may be enough to facilitate D 2 desorption. (2) The alternate explanation as to why the AES data depend only on the bulk composition is that the surface composition is the same as the overall bulk composition. This situation is less likely in view of the analysis of D 2 adsorption kinetics, which suggest a constant surface composition. The presence of a surface 7-hydride on Ti surfaces exposed to hydrogen has also been proposed to explain valence band structure in UPS spectra [22,23]. Fukuda, Honda, and Rabalais [22] observed the temperature dependence of
244
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
the Ti (3d) valence band profile and suggested a T - a phase transition occurring at the surface as the Specimen temperature was raised to 240°C. Another peak in the UPS data at ca. 5 eV below the Fermi level attributed to H (ls) levels was present even at low exposures, and did not increase further at larger exposures. These results were explained by the presence of a surface hydride layer which forms during the early stages of reaction, and is not perturbed by the UPS measurement. Based on these UPS measurements, and the kinetic analysis of D 2 uptake we still favor the model which proposes a T-phase surface hydride. 3.5. Oxidation of TiD x films The Q C M was used to monitor the rate of oxygen adsorption by several TiD x films prepared as described above, and the oxide formed was examined by AES. Fig. 6 compares the QCM response observed when clean Ti, TiD0. 7, and TiD 2 films were exposed to oxygen at a pressure of 2 x 10 v Torr. A frequency shift of 8 Hz corresponds to the adsorption of one equivalent monolayer (i.e. not corrected for surface roughness) of oxygen, and the data of fig. 6 represent multi-layer oxide formation. The adsorption rates for other deuterided films of various compositions were, within the experimental error, between the rates of the TiD 2 and TiD0. 7 curves shown. All of the deuterided films exhibited only a slightly lower oxygen adsorption rate (at 2 X 10 7 Torr) than the clean metal (by about 10-30%) [19], indicating that the high reactivity of the metal surface is affected only slightly by the presence of deuteriuim in
0 2 Uptake
N I
TiD07
c <1 I
EXPOSURE
TIHE,
sec
Fig. 6. QCM frequency change (mass uptake) versus time for clean Ti, TiD0. 7 and TiD 2 films exposed to 2 × 10 7 Torr O 2.
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films I
i
245
I
/~i~M) O(I
301~
B75 ELECTRON
450 KINETIC
525 ENERGY, ~V
61~0
Fig. 7. Background corrected AES spectra of TiD 2 films following uptake of (a) 380 n g / c m 2 and (b) 1300 n g / c m 2 of oxygen.
the metal lattice. The spread in the adsorption rate data is due primarily to the irreproducibility of films with identical surface roughness, i.e. the surface area varied slightly between specimens. Some decrease in the oxygen adsorption rate between TiD E and Ti is expected based solely on the volume expansion which occurs upon deuteriding - the density decreases from 4.5 g / c m 3 to 3.8 g / c m 3. There are less Ti atoms per unit area on the surface of TiD E (for oxygen adsorption) than on the pure metal surface. Based on these data, it is not possible to distinguish whether the small change in the oxygen adsorption rate for the deuterides is caused by subtle electronic changes, changes in surface roughness, or a decrease in the number of active adsorption sites. Examination of the oxidized TiD x surfaces at various stages of reaction by AES revealed the formation of an oxide which approached TiO 2 in composition, as revealed by the AES O(KLL) and Ti(LMM) peak areas. The oxide was substoichiometric at relatively low exposures (e.g., 2 × 10 - 7 Torr 02 for 15 minutes), and showed the characteristic Ti 3+ d -~ d transition in the ELS data [25]. The AES and ELS spectra obtained were virtually identical in peak shape and position to those obtained from oxidized titanium surfaces [14,15,19]. For example, fig. 7 shows the background corrected AES spectra of two TiD 2 film surfaces following adsorption of 380 n g / c m z and 1300 n g / c m z oxygen, respectively (8 n g / c m 2 ~ 1 monolayer). The measured peak area ratios ( I o / I T i ) were 0.34 and 0.43. This is less than the corresponding ratio measured from a pure TiO s specimen (Io/ITi = 0.59) because the oxide depth is still thinner than the AES sampling depth. At higher oxygen pressures ( > 10 -4 Torr), both clean and deuterided Ti films formed a TiO E layer at least as thick as the sampling depth of the AES data ( > 25-30 ,~).
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
246
1350
g
I
-80
Atm
jo ,--poo-,
o__.<__
-40
~f, H z 90C
45C
Jo
~
Oxide
0 Fig. 8. QCM frequency change (mass uptake) of a TiD 2 film exposed to 0 2 partial pressures of (1) 2 x 1 0 -5 Ton', (2) 5 × 1 0 -5 Ton., (3) 7 x 1 0 -5 Torr, (4) 9 × 1 0 5 Ton., (5) 0.2 Torr, (6) atmospheric pressure. The oxide thickness has been computed by correcting the mass uptake for the surface roughness of the film, R -- 3.0 [15].
In a previous publication [15], the thickness of the oxide formed on clean titanium film surfaces was found to vary markedly with the oxygen pressure. This dependence was interpreted in terms of an oxide growth mechanism in which oxygen anions migrated through the existing surface oxide under the driving force of an electrostatic field gradient set up across the oxide by electron transfer from the metal to form charged oxygen species on the surface. The oxide thickness increased with oxygen pressure and tended to grow under conditions of constant field across the layer. The oxidation of the deuterided films likewise showed a dependence on the oxygen pressure. Fig. 8 shows the Q C M uptake curve for a TiD 2 film exposed to oxygen at pressures ranging from 10 -5 Torr to atmospheric pressure. The limiting oxide thickness increased linearly with oxygen pressure between 10-5 and 10 -4 Torr, as previously observed for pure titanium films. The magnitude of the oxide thickness (indicated by the Q C M frequency shifts) was also similar to that observed for Ti films exposed to the same pressure steps, as illustrated by the dashed lines in fig. 8. However, at pressures above 10 - 4 Torr, the thickness of the oxide formed on the TiD 2 surface was much greater than that observed on the pure Ti films. The limiting oxide thickness at atmospheric pressure was over twice as large for TiD 2 films than it was for the Ti films, about 100 A compared to 40 A (when roughness factors were accounted for). Apparently the oxide formed at low pressures on the TiD 2 surface is not as effective in preventing further oxidation of the surface at higher pressures. The oxide formed on the titanium films has been hypothesized to consist of a glassy-type structure with channels of atomic dimensions through which oxygen anions could migrate to propagate the oxide thickness [15,19]. A similar type of oxide network probably forms on the TiD 2 surface. However, the network structure and dimensions are expected to differ significantly from the oxide formed on the clean titanium. The oxide network which forms on the
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
247
TiD2 surface probably contains channels of larger atomic dimensions through which oxygen anions could migrate more easily, and this would lower the resistance of partially oxidized TiD 2 to further oxidation at higher oxygen pressures, as compared to pure titanium. It is not clear whether the deuterium remains in the surface oxide or if it diffuses inward or out of the material during oxidation. (The removal of deuterium from the surface region during the oxidation reaction, is not expected to be a rate-limiting step in the overall reaction due to the high diffusivity of deuterium.) No D 2 or D20 evolution was detectable by the R G A (sensitivity - 1 0 - u Torr) during oxidation of 10-8 Torr. It is conceivable that if a glassy oxide network is formed, deuterium could remain bound either to O or Ti in the oxide. The possible formation of a hydroxide or hydrated oxide (TiOx • n D 2 0 ) during early oxidation cannot be ruled out, although the ultimate oxide stoichiometry was TiO v (Hydrated forms would exhibit higher O / T i ratios.) It is known that hydrogen is somewhat soluble in TiO 2, thus some deuterium is likely present in the oxide even if not directly bound. 3. 6. Effect of surface oxidation on deuterium absorption by TiDx
The rate of deuterium absorption by a titanium film subject to several sequential surface modifications was measured. These modifications included oxidation of the partially deuterided film surface and deposition of monolayer and multilayer iron films onto the titanium film surface. A schematic representation of some of the film surfaces that were examined with respect to their deuterium absorption properties is shown in fig. 9. The Q C M mass uptake curves obtained for deuterium loading of these surfaces (labeled a through f in
a[ T'O0. J
b TiOx Ti DO.9
O
_~,Fe (-19A) c Ti~x----Ti DO.9
dlFe(82~,) TiO X
t
,Ti (-10~)
e~_F~Oxide TiOx
TiD1.5
TiOx TiDI.5
I
Fig. 9. Schematic representation of various surfaces exposed t o time was observed. Details of each experiment listed in text.
D2
for which mass uptake versus
248
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
450
300
b
ef
Fig. 102 QCM frequency shift (mass uptake of D2) versus time for the same surfaces shown in fig. 9. The t = 0 axis has been displaced to the right for curves c-f for the sake of clarity. fig. 9) are shown in fig. 10. Titanium films, 2500 ,~ thick, freshly evaporated onto the Q C M , were exposed to D 2 at a pressure of 1 × 10-5 Torr until a total frequency shift of 975 Hz was observed, corresponding to an overall film composition of TiD0. 9. The linear absorption rate was 1.7 n g / c m 2 s (figs. 9a and 10a). At this point, the system was returned to v a c u u m and oxygen (ca. 10-8 Torr) was admitted until as additional frequency shift (due to oxidation) of 25 Hz (ca. 1 monolayer of oxygen, accounting for surface roughness) was observed, before again returning the c h a m b e r to vacuum. Subsequently, deuterium uptake at rates at 1 × 10 .5 Torr I32 were observed and found to be only slightly less (1.3 n g / c m 3 - s, not shown) than the clean surface value. Thus, small a m o u n t s of surface oxidation had a small effect on the ability of the film to adsorb deuterium. It is k n o w n that the first two monolayers of oxygen adsorbed on Ti surfaces are incorporated below the surface plane [26,27]. Apparently, this leaves the Ti atoms remaining on the surface and the necessary hydrogen dissociation reaction m a y still occur. The surface oxide was then allowed to grow to a greater depth by exposure to 0 2 at 3 × 10 .5 Torr until a 100 Hz additional frequency shift was noted in the Q C M indicating an oxide depth of - 10 A. The surface was returned to the deuterium environment and the rate of deuterium uptake at a D 2 pressure of 1 × 10 5 Torr was again measured and found to be reduced to 0.4 n g / c m 2 s, less than one-third of the clean surface value. A mass uptake curve is shown in fig. 10b, which can be c o m p a r e d to the clean surface curve (fig. 10a). A n o t h e r Ti film (not shown) which had been intentionally oxidized to a greater extent at 8 × 10 .5 Torr 02 (about 15 A oxide was formed, when corrected for surface roughness) prior to any D 2 exposure was found to be completely passivated toward the deuterium adsorption reaction. A small a m o u n t of fresh Ti was next evaporated onto the oxide surface, producing an additional frequency shift of 250 Hz, or about 8 A of titanium. At these small thicknesses, the metal film consists of islands and did not completely cover the oxide surface [15]. W h e n deuterium was admitted at 1 × 10 .5 Torr, the film
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
249
was reactivated and adsorbed D 2 at a slow linear rate of 0.03 n g / c m 2. s. The film continued to adsorb D 2 well past the amount which would have been required to convert the Ti overlayer to TiD 2, confirming that the oxide layer does not act as a complete diffusional barrier to deuterium and the underlying film continues to react if actives sites for D 2 dissociation are present on the surface. It is clear that these active dissociation sites are still present when the oxide coverage is less than 3-5 monolayers, are completely removed when the oxide is grown to ca. 15 A or greater thickness and can be partially restored with monolayer coverages of fresh metal. These may be the same range of conditions necessary to achieve in the activation/passivation cycle of some of the hydriding transition metals.
3. 7. Effect of iron adlayers A small amount of iron was evaporated onto the surface of the QCM, onto which had been previously deposited a 2000 A thick titanium film. The frequency shift due to the iron was 358 Hz, or 10-15 A (from the bulk density of Fe). Such low coverages result in the formation of Fe islands (as confirmed by AES) rather than a continuous layer. The persistence of the low-energy Ti MVV peak (at 27 eV) in the derivative Auger spectrum of fig. 11a confirms that some Ti is still at or very near the surface. This clean film was next exposed to D 2 at a pressure of I x 10 -5 Torr, and adsorbed D 2 at a rate of 14 n g / c m 2. s - over eight times the rate observed for clean titanium films. The rate was approximately linear until the entire Ti layer had been converted to TiD 2. This result shows that iron may be an excellent catalyst for hydrogen absorption by the underlying titanium film, by providing D 2 dissociation sites of higher activity than Ti. The Auger spectrum of the surface was recorded again after the deuterium exposure and is shown in fig. 1lb. No changes were observed in the Fe MVV line at 40 eV or the L M M lines (575-725 eV), indicating that an iron hydride was probably not formed. No solid iron hydrides have been reported, and the solubility of H 2 in Fe is quite low (estimated to be about 10 -9 cm 3 H z / g Fe from ref. [1] p. 84). As pointed out in ref. [1], however, the apparent solubility of hydrogen in Fe depends on the condition of the surface and is probably higher than this value. When examined at higher resolution, the Ti MVV peak (partially overlapping the Fe MVV peak in fig. l l b ) showed the characteristic doublet observed in the TiD x spectra previously observed [17]. The Ti L M M lines are not entirely characteristic of TiD 2. They are significantly different, however, from those observed on the F e / T i film before deuterium adsorption and are characteristic of a substoichiometric deuteride. These spectral differences may be indicative of some degree of alloy formation at the F e / T i interface in addition to titanium hydride formation, which prevent the titanium layers very near the surface from forming stoichiometric TiD2, even though the overall
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
250
f/
_before
[)2
(a) Ti MVV j, MVV
Ti LMM Fe LMM
(b) I
f
after D2
doublet
. . . O. . . . . . 2. 0 0
4 6 6 - . eV. . 6 0.0
8()O
Fig. 11. Derivative Auger spectra of clean Ti films onto which 10-15 A of Fe had been evaporated, (a) before exposure to D 2 and (b) after extensive deuteriding to T i D 2.
film stoichiometry is T i D 2. The deposition of Fe onto pure or partially deuterided titanium films which had been intentionally oxidized was found to reactivate the surface for further deuterium absorption. For example, a 2500 ,~ Ti film which had been converted to TiD0. 9 was intentionally oxidized ( - 20 A oxide), reducing the rate of D 2 absorption at 1 × 10 -5 Torr from 1.7 to 0.4 n g / c m 2. s (see figs. 10a and 10b). A small a m o u n t of Fe was then evaporated onto the surface (Aft, = 342 Hz, - 19 ,~ Fe) and the deuterium uptake rate was again measured; the mass increase remained linear in time, and doubled to 0.84 n g / c m 2 - s (figs. 9c and 10c). W h e n a second layer of Fe was then evaporated onto such surfaces, this time covering completely the TiO x layer (Afw = 1460 Hz, - 82 A Fe, fig. 9d), the deuterium adsorption rate at 1 × 10 -5 Torr remained linear and was 1.9 n g / c m 2. s. This is slightly above the clean titanium surface value. The total frequency shifts noted during the deuteriding of the film with the Fe adlayers indicated that the Ti film beneath had been converted from TiD0. 9 to TIDE5 (and would have continued to react if the D 2 exposure had not been interrupted). The Fe surface is at least as reactive as Ti towards deuterium adsorption even though a stable iron hydride phase is not formed. Deuterium
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
251
atoms, formed on the Fe surface by dissociative chemisorption, are apparently able to traverse the Fe layer despite their low solubility. This process may involve rapid diffusion along grain boundaries, if some degree of crystallinity exists in the film, to the F e / T i interface, where exothermic dissolution in the Ti occurs. Because of the small thicknesses of these Fe layers, some of the underlying Ti film may still be uncovered, such that D atoms formed on the surface can enter the bulk film through these "openings". The determination of the deuterium solubility in evaporated Fe films, and the effect of the Fe layer thickness on deuterium uptake by underlying Ti films, are currently under investigation. Oxidation of the Fe layers at 3 x 10-5 Torr 02 produced an oxide layer of about 15 A, as deduced by the QCM frequency shift of 20l Hz (assuming R = 3.0). When these films were exposed to deuterium at 1 × 10-5 Torr, the mass gain again remained linear with time, but was reduced to 0.25 n g / c m : • s (see figs. 9e and 10e). Thus surface oxidation has the effect of reducing the D 2 absorption rate by Ti films with Fe adlayers in a similar fashion to oxidation of pure Ti films. When small amounts of Ti (ca. 10 ,~) were evaporated onto the oxidized Fe layers (fig. 9f), no change in the deuterium uptake rate was observed. Thus, it appears that the rate of deuterium absorption is limited by the transport of D atoms through the oxidized Fe layer, and not by the availability of dissociation sites. This is in contrast to the effect of the oxide on D 2 uptake on Ti surfaces. There the oxide was not a diffusional barrier, but prevented dissociative chemisorption.
4. Conclusions The characterization and reactions of titanium film surfaces exposed to deuterium can be summarized as follows. (1) Clean titanium films react readily with D 2 at low partial pressure (10-7-10 -5 Torr) to form TiD 2 throughout the film at saturation. The kinetics of D z absorption suggest that a surface hydride layer forms during the early stages of reaction. AES and ELS spectra of the film surfaces at intermediate compositions (TiDx, x < 2), however, are sensitive to the overall bulk film composition. This condition likely results from disruption of the hydride surface layer under irradiation by the electron probe beam (even at low power densities). A steady-state deuterium composition within the sampled region is established which is proportional to the bulk deuterium content. (2) Titanium deuteride film surfaces are more reactive to oxidation in high partial pressures of 02 than pure Ti surfaces. The total amount of oxygen adsorbed at atmospheric pressure was - 5700 n g / c m 2 for the TiD 2 films, compared to 2200 n g / c m 2 for clean titanium films. (3) Oxidation of the titanium film surface slows the deuterium absorption rate
252
M.C_ Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
appreciably, requiring a b o u t 15 ,h, of oxide to completely passivate the surface. Small a m o u n t s of Ti (or Fe) evaporated o n t o the oxide surface provided sites for further D 2 dissociation. The u n d e r l y i n g oxide film did not act as a complete diffusional barrier for D atoms, such that the bulk film could c o n t i n u e to a b s o r b D 2. (4) I r o n adlayers of b o t h s u b m o n o l a y e r a n d multilayer thicknesses were f o u n d to increase the D 2 a b s o r p t i o n rate b y the u n d e r l y i n g t i t a n i u m film. Because F e does n o t form a hydride c o m p o u n d , its role seems to be purely catalytic in the d e u t e r i u m a b s o r p t i o n reaction. These results also have some i m p l i c a t i o n s in describing the h y d r i d i n g reactions at FeTi surfaces. The formation of a clean t i t a n i u m hydride surface layer is n o t a necessary prerequisite for b u l k H 2 absorption, Fe a d a t o m s can also act as a dissociation center. Segregation of the elemental constituents at the FeTi surface d u r i n g a c t u a l h y d r i d i n g cycles requires further investigation to describe the roles of the Fe a n d Ti atoms at those surfaces.
References [1] W.M. Mueller, J.P. Blackledge and G.C. Libowitz, Eds., Metal Hydrides (Academic Press, New York, 1968). [2] K.M. Mackay, Hydrogen Compounds of the Metallic Elements (Spon, London, 1966). [3] J.J. Reilly and G.D. Sandrock, Hydrogen Storage in Metals, in: Hydrogen in Metals II, Topics in Applied Physics 29, Eds. G. Alefeld and J. ViSlkl(Springer, Berlin, 1978) p. 201. [4[ C.G. Libowitz, Metal Hydrides for Thermal Energy Storage, in: Proc. 9th Intersoc. Energy Conversion Conf., 1974, p. 322. [5] M.E. Malinowski, J. Nucl. Mater. 63 (1976) 386. [6] M.E. Malinowski, J. Vacuum Sci. Technol. 15 (1978) 39. [7] M.E. Malinowski, J. Less-Common Metals 83 (1983) 1. [8] T.N. Wittberg and P.S. Wong, J. Electron Spectrosc. Related Phenomena 31 (1983) 81. [9] G.D. Sandrock and P.D. Goodell, J. Less-Common Metals 73 (1980) 161. [10] D. Khatamian, N.S. Kazama, F.D. Manchester, G.C. Weatherly and C.B. Alcock, J. LessCommon Metals 91 (1983) 267. [11] M. Polak, M. Hefetz, M.H. Mintz and M.P. Dariel, Surface Sci. 126 (1983) 739. [12] B. Kasemo and E. Tornqvist, Appl. Surface Sci. 3 (1979) 307. [13] R.K. Quinn and N.R. Armstrong, J. Electrochem. Soc. 125 (1978) 1790. [14] M.C. Burrell and N.R. Armstrong, J. Vacuum Sci. Technol. A1 (1983) 1831. [15] M.C. BurreU and N.R. Armstrong, Langmuir, submitted. [16] M.C. Burrell and N.R. Armstrong, Appl. Surface Sci. 17 (1983) 53. [17] B.C. Lamartine, T.W. Haas and J.S. Solomon, Appl. Surface Sci. 4 (1980) 537. [18] M.C. Burrell, R.S. Kaller and N.R. Armstrong, Anal. Chem. 59 (1982) 2511. [19] M.C. Burrell, PhD Dissertation, University of Arizona (1984). [20] G. Sauerbrey, Z. Physik 155 (1959) 206. [21] K.W. Nebesny and N.R. Armstrong, unpublished results; K.W. Nebesny, PhD Dissertation, University of Arizona (1984). [22] Y. Fukuda, F. Honda and J.W. Rabalais, Surface Sci. 91 (1980) 165. [23] D.E. Eastman, Solid State Commun. 10 (1972) 1933. [24] C.G. Pantano and T.E. Madey, Appl. Surface Sci. 7 (1981) 115. [25] V.E. Henrich, H.J. Zeiger and G. Dresselhaus, Natl. Bur. Sttl. Special Publication 455 (US Govt. Printing Office, Washington, DC, 1976) p. 133-138. [26] J.B. Brignolas, M. Bujar and J. Bardollo, Surface Sci. 108 (1981) L453. [27] M.J. PeUin, V. Anatono and D.M. Gruen, in: Proc. 43rd Annual Conf. on the Physics of Electronics, Santa Fe, 1983, Abstract A9.
235
D E U T E R I U M U P T A K E IN T I T A N I U M T H I N FILMS: t ssc00621THE E F F E C T O F O X I D E , A N D M E T A L (Ti A N D Fe) OVERLAYERS Michael C. B U R R E L L * a n d Neal R. A R M S T R O N G **
Department of Chemistry, University of Arizona, Tucson, Arizona 85721, USA Received 3 December 1984; accepted for publication 5 April 1985
Titanium thin films prepared in UHV were reacted with deuterium (PD2 < 1 X 10-5 Torr) to various loadings, as determined by microgravimetry using a quartz crystal microbalance. The kinetics of deuterium absorption favor a mechanism in which an a-phase surface deuteride forms on the film during the early stages of the reaction, resulting in a constant rate of deuterium uptake during most of the reaction. Surface characterization by AES and ELS, however, demonstrated spectral changes which were dependent on the bulk film stoichiometry. Electron-beam decomposition of the surface deuteride during AES analysis is postulated to explain this result. Oxidation of the titanium film surface caused a decrease in the deuterium absorption rate, completely inhibiting the reaction when oxides of thickness 20 ,~ or greater were formed. Fresh titanium layers on top of the oxide renewed the ability of the Ti film to take up D2 at the previous rate. Iron adlayers were found to accelerate the D2 absorption rate of Ti films, or to likewise reactivate oxidized Ti surfaces.
1. Introduction D u r i n g the past decade, metal hydrides have received considerable a t t e n t i o n in various technological applications. Metals which react rapidly with hydrogen to form solid hydrides have been used as getter p u m p s for h y d r o g e n isotopes in prototype fusion reactors. Metal hydrides have also been used as n e u t r o n energy m o d e r a t o r s [1,2]. Recently there has been increased a t t e n t i o n directed towards metal hydrides as media for h y d r o g e n storage a n d transportation for ultimate use as a fuel [3], or for other energy conversion schemes [4]. T h e most p r o m i s i n g materials for storage applications are the intermetallic alloys FeTi, Mg2Ni, L a N i s , Z r M n 2 , TiMnl.5, a n d related systems [1-3]. The rates of hydrogen a d s o r p t i o n a n d desorption in metal hydride systems are extremely sensitive to the c o m p o s i t i o n of the surface, particularly the presence of c o n t a m i n a n t s or oxide layers. Most studies have c o n c e n t r a t e d on * Presently at General Electric Company, Corporate Research and Development, Schenectady, New York, USA. ** To whom correspondence should be addressed.
0 0 3 9 - 6 0 2 8 / 8 5 / $ 0 3 . 3 0 © Elsevier Science Publishers B.V. ( N o r t h - H o l l a n d Physics Publishing Division)
236
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
the thermodynamic properties of these materials in bulk form, and only a few studies have appeared in which the surface composition of hydrides have been characterized under controlled conditions. One reason for this is the high equilibrium hydrogen pressures often necessary for hydride formation, which precludes the use of conventional vacuum surface spectroscopies. The hydride of titanium, Till 2 is readily formed and thermodynamically stable in ultrahigh-vacuum environments and lends itself to studies of surface composition of the hydride. Malinowski [5,6] has investigated the effect of contaminants on the deuterium pumping speed of titanium films, and demonstrated that Auger electron spectroscopy (AES) (in the derivative lineshape mode) can be useful for identifying the hydride composition as well as surface contaminants. Others have shown that the hydrides of titanium (TiHx; x ~<2) are characterized by surface electronic properties distinct from the pure metal using Auger electron, photoelectron, and electron energy-loss spectroscopies (ELS) [7,8]. The hydrogenation kinetics of the alloy FeTi are also dependent on the initial cleanliness of the surface [9-11]. The presence of a readily formed oxide layer prohibits hydrogen absorption and must be removed by an "activation" procedure. As for most hydride forming materials, this involves repeated heating of the FeTi alloy to temperatures in excess of 400°C in vacuum or under a hydrogen atmosphere, which apparently dissolves the surface oxide. Formation of FeTi--H requires an equilibrium hydrogen pressure of several atmospheres and surface studies of this material in vacuum environments is not possible. Recently, however, evidence has been given [11] that heating FeTi in a vacuum environment may result in the segregation of pure titanium ( - 100 monolayers) to the alloy surface at 800°C. Thus, the hydrogen adsorption reactions occurring at titanium film surfaces may be representative of similar reactions occurring at activated FeTi alloy and other alloy surfaces under actual hydriding conditions. It is also of interest in this study to determine the effect of adding small adlayers of Fe to the atomically clean or partially oxidized Ti surface, as a way of approximating the surface chemistries of the activated FeTi alloy. Oxide layers which readily form on Till 2 and other reactive metal hydride surfaces exposed to air, O 2, or H20 control the rate and extent of subsequent adsorption or release of hydrogen. Using only quartz crystal microgravimetry (QCM), Kasemo and Tornqvist [12] have shown that the oxide formed on titanium film surfaces drastically slows the hydrogen adsorption reaction, but that evaporation of small amounts of fresh Ti onto the oxide surface restored the hydrogen absorption capability of the entire material. The oxide prevented hydrogen dissociation on the surface, but apparently did not act as a diffusional barrier. Malinowski has observed that the oxide formed on TiD~ surfaces inhibited the release of deuterium by thermal decomposition; films with oxidized surfaces had to be heated to higher temperatures before the
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
237
material decomposed [7]. The electrochemical behavior of TiH~ films, in particular the anodic potential at which hydrogen is released, was also found by us in earlier studies to depend on the extend of the oxide layer on the surface [13]. These results indicated that an understanding of the reactivities of titanium hydride surfaces required first a knowledge of the composition and thickness of the oxide layer that in many instances isolates the material from its surroundings. Other studies in our laboratory [14,15] have demonstrated that it is possible to form an oxide layer at room temperature and low pressures (10 -8 Tort < Po2 < 1 atm) whose thickness and composition can be determined in a quantitative fashion using AES and microgravimetry (QCM), and ranges in thickness from
10-40 h. In the studies described in this paper, titanium deuteride films (TiDx; x = 0.5 to 2) were prepared and their surfaces characterized by AES and ELS. Using data treatment procedures developed in this laboratory for N(E) Auger spectra [16], we show that the lineshapes of the Ti LMM series (particularly the LMV transition) are even more sensitive to the presence of the hydride in the lattice. A very distinctive lineshape change (chemical shift) can be observed when hydrogen is introduced due to its contribution to the valence band density of states [17]. We have also found, however, that electron-beam damage effects are an important consideration in determination of hydriding levels by AES and these are discussed as well. The quartz crystal microbalance (QCM) was used to monitor the total deuterium uptake, and hence the absolute deuterium content of the entire thin film was known accurately and could be varied systematically. These films were subsequently reacted with oxygen, and the rate and extent of surface oxidation were determined from the mass uptake data, and the oxide characterized using AES. Rates of hydrogen uptake with and without oxide, with and without Ti and Fe overlayers were examined. The effect of deuterium loading (from zero concentration up to TiD2) on the surface oxidation reaction is also discussed.
2. Experimental All measurements were performed in a Physical Electronics 540A thin film analyzer with a base pressure of 10-~0 Torr. Auger spectra were collected in the direct current mode using a previously described system [18], using a 5 kV, 0.5/~A electron beam. The quartz crystal microbalance (QCM) has also been described in previous publications [14,15,19]. It consisted of a 1 / 4 inch diameter AT-cut crystal with a 10 MHz fundamental resonance frequency. Gold electrodes were deposited onto either side of the crystal, and were contacted by wires leading to an external gate oscillator circuit, whose output was measured with a digital counter. The QCM mount was rotatable within the
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
238
vacuum chamber, such that it could alternatively face the AES optics and the Ti evaporation source. Titanium films were deposited onto the exposed face of the QCM from a wire filament (Alfa), under UHV conditions. Changes in the QCM frequency, after sufficient cooling following evaporation, yielded the film thickness. Subsequent exposure to gaseous deuterium or oxygen (Matheson) yielded further frequency shifts, which were measured as a function of time, and related to adsorbed mass by the Sauerbrey relation [20]:
A m / A F = 4.4 X 10 -9 g / H z - c m 2.
(1)
Auger spectra and ELS were measured before and after the gas dosing experiments described previously [14]. AES data were treated to remove inelastic electron scattering contributions to the background by a previously described deconvolution procedure using the ELS spectra obtained in the region of the Auger spectrum [16].
3. Results and discussion
3.1. Preparation of TiDx films Fig. 1 shows the QCM response of a typical thick (at least 1000 A), freshly evaporated clean titanium film, exposed to deuterium at a pressure of 1 × 10-5 Torr. Within 30 min, the film became saturated and the rate of D 2 uptake
TI g00
I
+
02
1000"
~3'
N
cq I
/
13
6130
EXPOSURE
12(3(3 TIHE,
18~[]
2~OB
~eo
Fig. 1. Deuterium uptake (frequency change in the QCM) versus time for a clean titanium film. The region up to point 1 corresponds to the conversion up to TiD0. 3. The region up to point 2 corresponded to conversion up to TiD1. s. The limiting TiD 2 composition was reached at point 3.
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
H-H
239
~H+ H
///////
//~/I
/ H y.dride////~)I M eta I a
/
1
b Fig. 2. Models for hydride formation. (a) Surface hydride model, (b) Precipitate model.
leveled off. The D / T i ratio was calculated from the frequency shifts due to titanium deposition and subsequent deuterium uptake at saturation, and the atomic weights of Ti and D:
ND NTi
Afo/2 AfTi/47 .
(2)
For the film shown in fig. 1, the frequency shifts were AfT i = 10000 Hz (from the Ti deposition stage) and Alp = 870 Hz (at point 3, fig. 1, from the deuteriding stage), indicating formation of TiD z throughout the film. The adsorption rate decreased monotonically from the clean surface value as the film was converted to an overall D / T i ratio ~< 0.3 (point 1 in fig. 1) consistent with previous studies by Kasemo and Tornqvist [12]. The rate was then approximately constant with an apparent sticking coefficient of - 0 . 0 4 (at hydrogen pressures of 10 -5 Torr) until the D / T i ratio approached 1.8 (point 2 in fig. 1), after which the rate decreased until the limiting composition TiD 2 was attained (point 3). If the system was evacuated during the adsorption regime where the rate was constant (between points 1 and 2), a mass loss was observed corresponding to desorption of a small fraction ( < 5%) of the hydrogen content of the film. The desorption rate was insensitive to the total hydrogen content in the film for overall D / T i ratios from 0.5 to 1.8. A desorption of H z in this manner has been shown to fit a simple associative/desorption rate law [12]. In previous studies by Kasemo, the adsorption/desorption rates and the total amount of desorbed H2, did, however, increase with the hydrogen pressure used in the preceding adsorption cycle. Two models may be proposed for the conversion of a titanium film to the
240
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
hydride (or deuteride), as shown in figs. 2a and 2b. The absorption/desorption kinetics favor a surface hydride model, (fig. 2a), in which the reaction rate is controlled by diffusion of dissociated hydrogen through a surface hydride layer. The observed desorption of H 2 (or D2) from the Ti film, when the gas exposure is halted, may be due to super-saturation of the surface hydride layer. An alternative mechanism in which the hydrogen concentration changes uniformly throughout the sample (and in the surface region), during the hydriding process (fig. 2b) is inconsistent with our observations and has been previously discounted by others [6,7,12]. First, no desorption was observed if the hydriding reaction was interrupted prior to the constant-rate adsorption region. Presumably the initial D 2 adsorbed (50-100 monolayers) results in the formation and coalescence of the surface hydride. Secondly the adsorption/desorption kinetics did not change as the total D / T i ratio in the film increased from 0.5 to 1.8, indicating that the surface is not changing significantly during this stage of reaction. (If the composition was changing uniformly throughout the film, the rate of D 2 uptake would change with exposure). The T i / H phase diagram when extrapolated to room temperature, exhibits a solution c~-phase for H / T i ratios up to ca. 0.001, and a 7-phase for H / T i ratios of 1.5-2 [1,2]. Films with intermediate hydrogen concentrations therefore contain a mixture of both a- and 7-phases. In the surface hydride model of metal hydride formation (fig. 2a), the a 7 phase transition occurs more rapidly on the surface than in the bulk, and a 7 hydride layer forms on the surface. Subsequent hydriding must occur through this layer. One would expect to observe only 7 hydride of the surface for films hydrided to any intermediate concentration Till x (x = 0.1 to 2). To test this hypothesis, atomically clean titanium films were prepared and reacted to various deuterium loadings. Surface characterization by AES and ELS were performed in situ immediately following this loading.
3.2. AES results The d ( E N ( E ) ) / d E Auger spectra were first recorded for pure titanium, TiD0.v, TiDa. 3 and TiD 2 thin films. As noted by Malinowski and others, changes in the peak fine structure distinguish metal from TiD 2 and intermediate hydride compositions [7,8,13,17], and can be interpreted in terms of valence band redistribution as deuterium is added to the lattice [13]. These changes in peak shape can be summarized as follows: (1) A shoulder appears on the low-energy side of the M2,3VV peak (27 eV), which eventually becomes a doublet as the bulk composition approaches TiD 2. (2) The L2,3M2,3M2, 3 peak (380 eV) becomes broader, and the doublet structure in this peak disappears. (3) The L2.3M2,3V peak (410 eV) decreases in magnitude (with respect to the LMM peak), and a prominent shoulder appears on the low-energy side. (4) A shoulder also appears on the low-energy side of the L2,3W
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
Ti D2 A L M V
.2
l
N(E)
241
L
l
(b)
NE}
(a) 3~113
45~ 401~ 35~ ELECTRON K I N E T I C ENERGY, eV
501~
400
eV
420
Fig. 3. Ti(LMM) Auger spectra for (a) clean Ti and (b) full saturated, TiD 2. The original EN(E) spectra were treated by a deconvolution procedure outlined in ref. [16]. Spectral features (I) and (II) explained in text.
Fig. 4. Ti(LMV) spectra of TiD2. Spectrum (a) has been treated to remove the secondary cascade from electron sources of higher energy. Spectrum (b) represents the result of an FFT deconvolution of (a) with an energy-loss function obtained with the electron beam at 400 eV. Spectral features (I) and (II) explained in text.
peak (460 eV). The new spectral features appear gradually as the overall composition approaches TiD 2. This observation suggests (in contrast to the surface hydride model) that the surface composition within the sampling depth of the AES experiment is related to the bulk film composition and does not approach stoichiometric TiD 2 until the entire film has been converted. Lamartine, Haas, and Solomon [17] and previous work from this laboratory [13], documented studies of titanium hydride and deuteride surfaces of various compositions using AES, XPS, and ELS. Those studies differed from this study because the samples were not prepared in the same vacuum chamber where analysis occurred. The hydrides were ion-sputtered prior to analysis to remove the surface oxide. The AES results, however, were essentially identical to those observed here. The Ti (2p) binding energy in the XPS results shifted towards higher energies as the hydrogen or deuterium content increased. The bulk plasmon energy in the ELS shifted to larger energies, again related to the overall bulk composition of the specimen. We have extended these results to AES data collected in the EN(E) mode. Following correction for the C M A transmission efficiency a sequential back-
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
242
ground subtraction correction for the secondary electron background and the inelastic loss contribution was applied to the data [16], yielding the spectra shown in fig. 3. The Ti (LMV) peak in the TiD 2 clearly shows the additional low-energy shoulder associated with hydrogen uptake. A higher-resolution LMV peak in TiD 2 was obtained from the original data by deconvoluting an instrument-specimen response measured at E 0 = 400 eV, using a fast Fourier transform (FFT) technique developed in this laboratory [21]. The initial spectrum of the LMV peak is shown in fig. 4a and, the resultant deconvolved spectrum is shown in fig. 4b. This FFT procedure removes the sample and instrumental contributions to the peak broadening and helps resolve the shoulder on the low-kinetic-energy side. The high-kinetic-energy component (I) in the LMV spectrum figs. 3b and 4b, in a simple one-electron model, represents ejection of an Auger electron originating from the Ti 3d valence level (following L-shell ionization). The lower-energy component (II) arises from a valence Auger electron ejected from a level of H(ls) character. The observed energy difference of 5 eV corresponds to the measured difference of the two valence band maxima observed in UPS measurements of Till 2 [17,22,23]. 3.3. E L S results
Changes in the electron energy-loss spectra (ELS) also occurred between pure and fully deuterided titanium films, as shown in fig. 5. The data shown were obtained using primary beam energies of 100, 300, 500, and 1000 eV (curves a through d, respectively). The use of low-energy electron beams emphasizes features due to surface processes, while higher-energy electron beams give spectra which emphasize bulk processes. The feature noted in the
TiD 2
N(E
c
1 30
Energy
20
10
4 0
3 2
1
13,0 2,oI ol ko
Los s,eV
Fig. 5. Energy-loss spectra for Ti and TiD 2 films obtained at electron beam energies of 100, 300, 500 and 1000 eV (curves a through d respectively). All data are plotted as energy loss from the primary energy ( F W H M of electron beam ca. 800-900 mV). Spectra features 1-4 and 1'-4' are detailed in the text.
M.C Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
243
ELS of clean titanium (feature 1, right side of figs. 5 and 6) at about 7 eV is most likely an interband transition; its intensity was rapidly quenched by even small doses of D 2. The features 2, 3 and 5 at 11-12, 16-17, and 32 eV respectively arise from surface and bulk plasmon excitations, and M2, 3 ionization, respectively. In the ELS from TiD 2 (left side of fig. 5) several changes are noted. A new peak appears at 4-5 eV (feature 1% a and b) which we attribute to excitation of an electron from the valence band maxima at - 5 eV (H (ls) - like band) to the conduction band. This peak is at least partially of bulk origin, because it is still present (although not totally resolved) even when incident beams of 500 eV were used. The surface and bulk plasmons were shifted to higher energy-loss values of 12-13 eV (feature 2') and 19 eV (feature 3'), respectively. The M2, 3 (3p) ionization edge shifted from 31.3 to 32.4 eV (feature 4'), as determined from ELS data of higher resolution than shown here. This is consistent with partial charge transfer from Ti to D observed in XPS Ti (2p) core level chemical shifts [13,17]. The changes in the ELS measurement (from Ti- to TiD2-1ike spectra), like the changes in the AES spectra, occurred gradually and monotonically as the deuterium content of the film was increased. 3.4. Hydriding mechanisms Two possible explanations may be offered to explain why a surface hydride layer of constant composition is not apparent on the surface of titanium films deuterided in situ to intermediate compositions, TiD x (x < 2): (1) The electron beam used to generate the AES and ELS data may induce decomposition of the deuteride due to localized heating or electron-stimulated desorption [21]. The deuterium content in the region of the beam may be depleted by these processes, but due to the high diffusivity of deuterium atoms, is partially replenished by deuterium from the bulk [6]. The measured Auger spectra may reflect the steady-state deuteride composition under electron-beam irradiation, which would be expected to vary with the bulk composition. Although beam currents of 0.5 /~A or less were used to avoid altering the surface (at 0.5 ~A, 5 kV beam, 0.1 mm spot size - power d e n s i t y = 25 W / c m 2), the damage threshold for these materials may be much lower than expected for other materials (such as metal oxides [24]). The heating caused by the electron beam may be enough to facilitate D 2 desorption. (2) The alternate explanation as to why the AES data depend only on the bulk composition is that the surface composition is the same as the overall bulk composition. This situation is less likely in view of the analysis of D 2 adsorption kinetics, which suggest a constant surface composition. The presence of a surface 7-hydride on Ti surfaces exposed to hydrogen has also been proposed to explain valence band structure in UPS spectra [22,23]. Fukuda, Honda, and Rabalais [22] observed the temperature dependence of
244
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
the Ti (3d) valence band profile and suggested a T - a phase transition occurring at the surface as the Specimen temperature was raised to 240°C. Another peak in the UPS data at ca. 5 eV below the Fermi level attributed to H (ls) levels was present even at low exposures, and did not increase further at larger exposures. These results were explained by the presence of a surface hydride layer which forms during the early stages of reaction, and is not perturbed by the UPS measurement. Based on these UPS measurements, and the kinetic analysis of D 2 uptake we still favor the model which proposes a T-phase surface hydride. 3.5. Oxidation of TiD x films The Q C M was used to monitor the rate of oxygen adsorption by several TiD x films prepared as described above, and the oxide formed was examined by AES. Fig. 6 compares the QCM response observed when clean Ti, TiD0. 7, and TiD 2 films were exposed to oxygen at a pressure of 2 x 10 v Torr. A frequency shift of 8 Hz corresponds to the adsorption of one equivalent monolayer (i.e. not corrected for surface roughness) of oxygen, and the data of fig. 6 represent multi-layer oxide formation. The adsorption rates for other deuterided films of various compositions were, within the experimental error, between the rates of the TiD 2 and TiD0. 7 curves shown. All of the deuterided films exhibited only a slightly lower oxygen adsorption rate (at 2 X 10 7 Torr) than the clean metal (by about 10-30%) [19], indicating that the high reactivity of the metal surface is affected only slightly by the presence of deuteriuim in
0 2 Uptake
N I
TiD07
c <1 I
EXPOSURE
TIHE,
sec
Fig. 6. QCM frequency change (mass uptake) versus time for clean Ti, TiD0. 7 and TiD 2 films exposed to 2 × 10 7 Torr O 2.
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films I
i
245
I
/~i~M) O(I
301~
B75 ELECTRON
450 KINETIC
525 ENERGY, ~V
61~0
Fig. 7. Background corrected AES spectra of TiD 2 films following uptake of (a) 380 n g / c m 2 and (b) 1300 n g / c m 2 of oxygen.
the metal lattice. The spread in the adsorption rate data is due primarily to the irreproducibility of films with identical surface roughness, i.e. the surface area varied slightly between specimens. Some decrease in the oxygen adsorption rate between TiD E and Ti is expected based solely on the volume expansion which occurs upon deuteriding - the density decreases from 4.5 g / c m 3 to 3.8 g / c m 3. There are less Ti atoms per unit area on the surface of TiD E (for oxygen adsorption) than on the pure metal surface. Based on these data, it is not possible to distinguish whether the small change in the oxygen adsorption rate for the deuterides is caused by subtle electronic changes, changes in surface roughness, or a decrease in the number of active adsorption sites. Examination of the oxidized TiD x surfaces at various stages of reaction by AES revealed the formation of an oxide which approached TiO 2 in composition, as revealed by the AES O(KLL) and Ti(LMM) peak areas. The oxide was substoichiometric at relatively low exposures (e.g., 2 × 10 - 7 Torr 02 for 15 minutes), and showed the characteristic Ti 3+ d -~ d transition in the ELS data [25]. The AES and ELS spectra obtained were virtually identical in peak shape and position to those obtained from oxidized titanium surfaces [14,15,19]. For example, fig. 7 shows the background corrected AES spectra of two TiD 2 film surfaces following adsorption of 380 n g / c m z and 1300 n g / c m z oxygen, respectively (8 n g / c m 2 ~ 1 monolayer). The measured peak area ratios ( I o / I T i ) were 0.34 and 0.43. This is less than the corresponding ratio measured from a pure TiO s specimen (Io/ITi = 0.59) because the oxide depth is still thinner than the AES sampling depth. At higher oxygen pressures ( > 10 -4 Torr), both clean and deuterided Ti films formed a TiO E layer at least as thick as the sampling depth of the AES data ( > 25-30 ,~).
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
246
1350
g
I
-80
Atm
jo ,--poo-,
o__.<__
-40
~f, H z 90C
45C
Jo
~
Oxide
0 Fig. 8. QCM frequency change (mass uptake) of a TiD 2 film exposed to 0 2 partial pressures of (1) 2 x 1 0 -5 Ton', (2) 5 × 1 0 -5 Ton., (3) 7 x 1 0 -5 Torr, (4) 9 × 1 0 5 Ton., (5) 0.2 Torr, (6) atmospheric pressure. The oxide thickness has been computed by correcting the mass uptake for the surface roughness of the film, R -- 3.0 [15].
In a previous publication [15], the thickness of the oxide formed on clean titanium film surfaces was found to vary markedly with the oxygen pressure. This dependence was interpreted in terms of an oxide growth mechanism in which oxygen anions migrated through the existing surface oxide under the driving force of an electrostatic field gradient set up across the oxide by electron transfer from the metal to form charged oxygen species on the surface. The oxide thickness increased with oxygen pressure and tended to grow under conditions of constant field across the layer. The oxidation of the deuterided films likewise showed a dependence on the oxygen pressure. Fig. 8 shows the Q C M uptake curve for a TiD 2 film exposed to oxygen at pressures ranging from 10 -5 Torr to atmospheric pressure. The limiting oxide thickness increased linearly with oxygen pressure between 10-5 and 10 -4 Torr, as previously observed for pure titanium films. The magnitude of the oxide thickness (indicated by the Q C M frequency shifts) was also similar to that observed for Ti films exposed to the same pressure steps, as illustrated by the dashed lines in fig. 8. However, at pressures above 10 - 4 Torr, the thickness of the oxide formed on the TiD 2 surface was much greater than that observed on the pure Ti films. The limiting oxide thickness at atmospheric pressure was over twice as large for TiD 2 films than it was for the Ti films, about 100 A compared to 40 A (when roughness factors were accounted for). Apparently the oxide formed at low pressures on the TiD 2 surface is not as effective in preventing further oxidation of the surface at higher pressures. The oxide formed on the titanium films has been hypothesized to consist of a glassy-type structure with channels of atomic dimensions through which oxygen anions could migrate to propagate the oxide thickness [15,19]. A similar type of oxide network probably forms on the TiD 2 surface. However, the network structure and dimensions are expected to differ significantly from the oxide formed on the clean titanium. The oxide network which forms on the
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
247
TiD2 surface probably contains channels of larger atomic dimensions through which oxygen anions could migrate more easily, and this would lower the resistance of partially oxidized TiD 2 to further oxidation at higher oxygen pressures, as compared to pure titanium. It is not clear whether the deuterium remains in the surface oxide or if it diffuses inward or out of the material during oxidation. (The removal of deuterium from the surface region during the oxidation reaction, is not expected to be a rate-limiting step in the overall reaction due to the high diffusivity of deuterium.) No D 2 or D20 evolution was detectable by the R G A (sensitivity - 1 0 - u Torr) during oxidation of 10-8 Torr. It is conceivable that if a glassy oxide network is formed, deuterium could remain bound either to O or Ti in the oxide. The possible formation of a hydroxide or hydrated oxide (TiOx • n D 2 0 ) during early oxidation cannot be ruled out, although the ultimate oxide stoichiometry was TiO v (Hydrated forms would exhibit higher O / T i ratios.) It is known that hydrogen is somewhat soluble in TiO 2, thus some deuterium is likely present in the oxide even if not directly bound. 3. 6. Effect of surface oxidation on deuterium absorption by TiDx
The rate of deuterium absorption by a titanium film subject to several sequential surface modifications was measured. These modifications included oxidation of the partially deuterided film surface and deposition of monolayer and multilayer iron films onto the titanium film surface. A schematic representation of some of the film surfaces that were examined with respect to their deuterium absorption properties is shown in fig. 9. The Q C M mass uptake curves obtained for deuterium loading of these surfaces (labeled a through f in
a[ T'O0. J
b TiOx Ti DO.9
O
_~,Fe (-19A) c Ti~x----Ti DO.9
dlFe(82~,) TiO X
t
,Ti (-10~)
e~_F~Oxide TiOx
TiD1.5
TiOx TiDI.5
I
Fig. 9. Schematic representation of various surfaces exposed t o time was observed. Details of each experiment listed in text.
D2
for which mass uptake versus
248
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
450
300
b
ef
Fig. 102 QCM frequency shift (mass uptake of D2) versus time for the same surfaces shown in fig. 9. The t = 0 axis has been displaced to the right for curves c-f for the sake of clarity. fig. 9) are shown in fig. 10. Titanium films, 2500 ,~ thick, freshly evaporated onto the Q C M , were exposed to D 2 at a pressure of 1 × 10-5 Torr until a total frequency shift of 975 Hz was observed, corresponding to an overall film composition of TiD0. 9. The linear absorption rate was 1.7 n g / c m 2 s (figs. 9a and 10a). At this point, the system was returned to v a c u u m and oxygen (ca. 10-8 Torr) was admitted until as additional frequency shift (due to oxidation) of 25 Hz (ca. 1 monolayer of oxygen, accounting for surface roughness) was observed, before again returning the c h a m b e r to vacuum. Subsequently, deuterium uptake at rates at 1 × 10 .5 Torr I32 were observed and found to be only slightly less (1.3 n g / c m 3 - s, not shown) than the clean surface value. Thus, small a m o u n t s of surface oxidation had a small effect on the ability of the film to adsorb deuterium. It is k n o w n that the first two monolayers of oxygen adsorbed on Ti surfaces are incorporated below the surface plane [26,27]. Apparently, this leaves the Ti atoms remaining on the surface and the necessary hydrogen dissociation reaction m a y still occur. The surface oxide was then allowed to grow to a greater depth by exposure to 0 2 at 3 × 10 .5 Torr until a 100 Hz additional frequency shift was noted in the Q C M indicating an oxide depth of - 10 A. The surface was returned to the deuterium environment and the rate of deuterium uptake at a D 2 pressure of 1 × 10 5 Torr was again measured and found to be reduced to 0.4 n g / c m 2 s, less than one-third of the clean surface value. A mass uptake curve is shown in fig. 10b, which can be c o m p a r e d to the clean surface curve (fig. 10a). A n o t h e r Ti film (not shown) which had been intentionally oxidized to a greater extent at 8 × 10 .5 Torr 02 (about 15 A oxide was formed, when corrected for surface roughness) prior to any D 2 exposure was found to be completely passivated toward the deuterium adsorption reaction. A small a m o u n t of fresh Ti was next evaporated onto the oxide surface, producing an additional frequency shift of 250 Hz, or about 8 A of titanium. At these small thicknesses, the metal film consists of islands and did not completely cover the oxide surface [15]. W h e n deuterium was admitted at 1 × 10 .5 Torr, the film
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
249
was reactivated and adsorbed D 2 at a slow linear rate of 0.03 n g / c m 2. s. The film continued to adsorb D 2 well past the amount which would have been required to convert the Ti overlayer to TiD 2, confirming that the oxide layer does not act as a complete diffusional barrier to deuterium and the underlying film continues to react if actives sites for D 2 dissociation are present on the surface. It is clear that these active dissociation sites are still present when the oxide coverage is less than 3-5 monolayers, are completely removed when the oxide is grown to ca. 15 A or greater thickness and can be partially restored with monolayer coverages of fresh metal. These may be the same range of conditions necessary to achieve in the activation/passivation cycle of some of the hydriding transition metals.
3. 7. Effect of iron adlayers A small amount of iron was evaporated onto the surface of the QCM, onto which had been previously deposited a 2000 A thick titanium film. The frequency shift due to the iron was 358 Hz, or 10-15 A (from the bulk density of Fe). Such low coverages result in the formation of Fe islands (as confirmed by AES) rather than a continuous layer. The persistence of the low-energy Ti MVV peak (at 27 eV) in the derivative Auger spectrum of fig. 11a confirms that some Ti is still at or very near the surface. This clean film was next exposed to D 2 at a pressure of I x 10 -5 Torr, and adsorbed D 2 at a rate of 14 n g / c m 2. s - over eight times the rate observed for clean titanium films. The rate was approximately linear until the entire Ti layer had been converted to TiD 2. This result shows that iron may be an excellent catalyst for hydrogen absorption by the underlying titanium film, by providing D 2 dissociation sites of higher activity than Ti. The Auger spectrum of the surface was recorded again after the deuterium exposure and is shown in fig. 1lb. No changes were observed in the Fe MVV line at 40 eV or the L M M lines (575-725 eV), indicating that an iron hydride was probably not formed. No solid iron hydrides have been reported, and the solubility of H 2 in Fe is quite low (estimated to be about 10 -9 cm 3 H z / g Fe from ref. [1] p. 84). As pointed out in ref. [1], however, the apparent solubility of hydrogen in Fe depends on the condition of the surface and is probably higher than this value. When examined at higher resolution, the Ti MVV peak (partially overlapping the Fe MVV peak in fig. l l b ) showed the characteristic doublet observed in the TiD x spectra previously observed [17]. The Ti L M M lines are not entirely characteristic of TiD 2. They are significantly different, however, from those observed on the F e / T i film before deuterium adsorption and are characteristic of a substoichiometric deuteride. These spectral differences may be indicative of some degree of alloy formation at the F e / T i interface in addition to titanium hydride formation, which prevent the titanium layers very near the surface from forming stoichiometric TiD2, even though the overall
M. C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
250
f/
_before
[)2
(a) Ti MVV j, MVV
Ti LMM Fe LMM
(b) I
f
after D2
doublet
. . . O. . . . . . 2. 0 0
4 6 6 - . eV. . 6 0.0
8()O
Fig. 11. Derivative Auger spectra of clean Ti films onto which 10-15 A of Fe had been evaporated, (a) before exposure to D 2 and (b) after extensive deuteriding to T i D 2.
film stoichiometry is T i D 2. The deposition of Fe onto pure or partially deuterided titanium films which had been intentionally oxidized was found to reactivate the surface for further deuterium absorption. For example, a 2500 ,~ Ti film which had been converted to TiD0. 9 was intentionally oxidized ( - 20 A oxide), reducing the rate of D 2 absorption at 1 × 10 -5 Torr from 1.7 to 0.4 n g / c m 2. s (see figs. 10a and 10b). A small a m o u n t of Fe was then evaporated onto the surface (Aft, = 342 Hz, - 19 ,~ Fe) and the deuterium uptake rate was again measured; the mass increase remained linear in time, and doubled to 0.84 n g / c m 2 - s (figs. 9c and 10c). W h e n a second layer of Fe was then evaporated onto such surfaces, this time covering completely the TiO x layer (Afw = 1460 Hz, - 82 A Fe, fig. 9d), the deuterium adsorption rate at 1 × 10 -5 Torr remained linear and was 1.9 n g / c m 2. s. This is slightly above the clean titanium surface value. The total frequency shifts noted during the deuteriding of the film with the Fe adlayers indicated that the Ti film beneath had been converted from TiD0. 9 to TIDE5 (and would have continued to react if the D 2 exposure had not been interrupted). The Fe surface is at least as reactive as Ti towards deuterium adsorption even though a stable iron hydride phase is not formed. Deuterium
M.C. Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
251
atoms, formed on the Fe surface by dissociative chemisorption, are apparently able to traverse the Fe layer despite their low solubility. This process may involve rapid diffusion along grain boundaries, if some degree of crystallinity exists in the film, to the F e / T i interface, where exothermic dissolution in the Ti occurs. Because of the small thicknesses of these Fe layers, some of the underlying Ti film may still be uncovered, such that D atoms formed on the surface can enter the bulk film through these "openings". The determination of the deuterium solubility in evaporated Fe films, and the effect of the Fe layer thickness on deuterium uptake by underlying Ti films, are currently under investigation. Oxidation of the Fe layers at 3 x 10-5 Torr 02 produced an oxide layer of about 15 A, as deduced by the QCM frequency shift of 20l Hz (assuming R = 3.0). When these films were exposed to deuterium at 1 × 10-5 Torr, the mass gain again remained linear with time, but was reduced to 0.25 n g / c m : • s (see figs. 9e and 10e). Thus surface oxidation has the effect of reducing the D 2 absorption rate by Ti films with Fe adlayers in a similar fashion to oxidation of pure Ti films. When small amounts of Ti (ca. 10 ,~) were evaporated onto the oxidized Fe layers (fig. 9f), no change in the deuterium uptake rate was observed. Thus, it appears that the rate of deuterium absorption is limited by the transport of D atoms through the oxidized Fe layer, and not by the availability of dissociation sites. This is in contrast to the effect of the oxide on D 2 uptake on Ti surfaces. There the oxide was not a diffusional barrier, but prevented dissociative chemisorption.
4. Conclusions The characterization and reactions of titanium film surfaces exposed to deuterium can be summarized as follows. (1) Clean titanium films react readily with D 2 at low partial pressure (10-7-10 -5 Torr) to form TiD 2 throughout the film at saturation. The kinetics of D z absorption suggest that a surface hydride layer forms during the early stages of reaction. AES and ELS spectra of the film surfaces at intermediate compositions (TiDx, x < 2), however, are sensitive to the overall bulk film composition. This condition likely results from disruption of the hydride surface layer under irradiation by the electron probe beam (even at low power densities). A steady-state deuterium composition within the sampled region is established which is proportional to the bulk deuterium content. (2) Titanium deuteride film surfaces are more reactive to oxidation in high partial pressures of 02 than pure Ti surfaces. The total amount of oxygen adsorbed at atmospheric pressure was - 5700 n g / c m 2 for the TiD 2 films, compared to 2200 n g / c m 2 for clean titanium films. (3) Oxidation of the titanium film surface slows the deuterium absorption rate
252
M.C_ Burrell, N.R. Armstrong / Deuterium uptake in titanium thin films
appreciably, requiring a b o u t 15 ,h, of oxide to completely passivate the surface. Small a m o u n t s of Ti (or Fe) evaporated o n t o the oxide surface provided sites for further D 2 dissociation. The u n d e r l y i n g oxide film did not act as a complete diffusional barrier for D atoms, such that the bulk film could c o n t i n u e to a b s o r b D 2. (4) I r o n adlayers of b o t h s u b m o n o l a y e r a n d multilayer thicknesses were f o u n d to increase the D 2 a b s o r p t i o n rate b y the u n d e r l y i n g t i t a n i u m film. Because F e does n o t form a hydride c o m p o u n d , its role seems to be purely catalytic in the d e u t e r i u m a b s o r p t i o n reaction. These results also have some i m p l i c a t i o n s in describing the h y d r i d i n g reactions at FeTi surfaces. The formation of a clean t i t a n i u m hydride surface layer is n o t a necessary prerequisite for b u l k H 2 absorption, Fe a d a t o m s can also act as a dissociation center. Segregation of the elemental constituents at the FeTi surface d u r i n g a c t u a l h y d r i d i n g cycles requires further investigation to describe the roles of the Fe a n d Ti atoms at those surfaces.
References [1] W.M. Mueller, J.P. Blackledge and G.C. Libowitz, Eds., Metal Hydrides (Academic Press, New York, 1968). [2] K.M. Mackay, Hydrogen Compounds of the Metallic Elements (Spon, London, 1966). [3] J.J. Reilly and G.D. Sandrock, Hydrogen Storage in Metals, in: Hydrogen in Metals II, Topics in Applied Physics 29, Eds. G. Alefeld and J. ViSlkl(Springer, Berlin, 1978) p. 201. [4[ C.G. Libowitz, Metal Hydrides for Thermal Energy Storage, in: Proc. 9th Intersoc. Energy Conversion Conf., 1974, p. 322. [5] M.E. Malinowski, J. Nucl. Mater. 63 (1976) 386. [6] M.E. Malinowski, J. Vacuum Sci. Technol. 15 (1978) 39. [7] M.E. Malinowski, J. Less-Common Metals 83 (1983) 1. [8] T.N. Wittberg and P.S. Wong, J. Electron Spectrosc. Related Phenomena 31 (1983) 81. [9] G.D. Sandrock and P.D. Goodell, J. Less-Common Metals 73 (1980) 161. [10] D. Khatamian, N.S. Kazama, F.D. Manchester, G.C. Weatherly and C.B. Alcock, J. LessCommon Metals 91 (1983) 267. [11] M. Polak, M. Hefetz, M.H. Mintz and M.P. Dariel, Surface Sci. 126 (1983) 739. [12] B. Kasemo and E. Tornqvist, Appl. Surface Sci. 3 (1979) 307. [13] R.K. Quinn and N.R. Armstrong, J. Electrochem. Soc. 125 (1978) 1790. [14] M.C. Burrell and N.R. Armstrong, J. Vacuum Sci. Technol. A1 (1983) 1831. [15] M.C. BurreU and N.R. Armstrong, Langmuir, submitted. [16] M.C. Burrell and N.R. Armstrong, Appl. Surface Sci. 17 (1983) 53. [17] B.C. Lamartine, T.W. Haas and J.S. Solomon, Appl. Surface Sci. 4 (1980) 537. [18] M.C. Burrell, R.S. Kaller and N.R. Armstrong, Anal. Chem. 59 (1982) 2511. [19] M.C. Burrell, PhD Dissertation, University of Arizona (1984). [20] G. Sauerbrey, Z. Physik 155 (1959) 206. [21] K.W. Nebesny and N.R. Armstrong, unpublished results; K.W. Nebesny, PhD Dissertation, University of Arizona (1984). [22] Y. Fukuda, F. Honda and J.W. Rabalais, Surface Sci. 91 (1980) 165. [23] D.E. Eastman, Solid State Commun. 10 (1972) 1933. [24] C.G. Pantano and T.E. Madey, Appl. Surface Sci. 7 (1981) 115. [25] V.E. Henrich, H.J. Zeiger and G. Dresselhaus, Natl. Bur. Sttl. Special Publication 455 (US Govt. Printing Office, Washington, DC, 1976) p. 133-138. [26] J.B. Brignolas, M. Bujar and J. Bardollo, Surface Sci. 108 (1981) L453. [27] M.J. PeUin, V. Anatono and D.M. Gruen, in: Proc. 43rd Annual Conf. on the Physics of Electronics, Santa Fe, 1983, Abstract A9.