angle-integrated sticking probabilities: n2, h2 and d2 on w{001}

3
Vacuum/volume 31/numbers 10-12/pages 507 to 509/1981 0042-207X/81/100507-03502 00/0 Printed ,n Great Br,tam @ 1981 Pergamon PressLtd Angle-integrated sticking probabilities: N 2, and D 2 on W{O01} Heather Bickley, James S Arlow, M A Morris and David A King, The Donnan Laboratorms, The Untverstty, Ltverpool L69 3BX, UK H2 A new techmque has been used to measure/nctdence-angle-averaged sttcktng probab/httes, (s>, for N 2, H u and D u on W{O01} A comparison with normal-mctdence values, s±, obtained by molecular beam techntques provtdes a measure of the angular dependence of s For N 2 on W{O01}, the initial (zero-coverage hm/t) sttckmg probabthty (S>o ts 0 63+0 03, comparing closely wtth So ±= 0 58±0 01 reported by King and Wells, indicating a relattvely small vartatton m s with incidence angle After saturatton and anneahng at 1000 K to form an underlayer, mtrogen Is readsorbed at 340 K with (S>o=O 1 For both H u and D 2 at 340 K we hnd (S>o to be 0 72±0 02, whtch ts slgmhcantly htgher than the value reported for So ~, 0 60±0 02 The stgnthcance of the results for adsorption mechamsms ts dtscussed 1. Introduction Relatively few accurate, absolute methods are available for measuring sticking probabfllUes at single crystal surfaces I For high sticking probabilities (0 1 ~<s~< 1) the reflection-detector techniques are the most accurate, where the reflection coefficient is measured by sampling the back-scattered flux for a given molecular beam incidence angle Studies of the dependence ors on incidence angle by these methods are restricted experimentally to the range from normal incidence to ~60 ° from normal, and generally very small variations have been noted 2 4, with the exception of H2 and D 2 on W~ll0) 4, where So is reported to decrease by ~60°~, over this range, and Hz/D 2 on stepped Pt{lll~j planes s, where the reactivity was highest when the incident beam was directed at the steps A comparison of angle- averaged sticking probabilities, (s), with normal incidence values, s~, from molecular beam experiments should provide a sensitive measure of the incidence-angle dependence of s and to this end we have applied a recently developed 6 absolute method for measuring ~ to the adsorption of N2, H 2 and D2 on W~0011 Accurate absolute values of s ~ are available 3 v 8 for each of these systems, allowing a direct comparison For H2 and D 2 on W I001 I, there is some controversy in the literature concerning the initial sticking probability v'8, the coverage dependence of s 8 l o, and the existence of an isotope effect 7 8 10 tx ~4 We have attempted to resolve these disagreements in the present study 2. Experimental The principles and operation of the ultrahigh vacuum chamber used for measuring absolute sticking probabilities have been described elsewhere 6 Gas is introduced to the adsorption cell via a capillary, and a steady state pressure is maintained In the cell by continuous pumping The crystal is suspended in front of a tube connected to an ionization gauge (IG-1) and possessing an orifice directed at the centre of the crystal The orifice diameter is 1 mm, and the (accurately measured) orifice-to-crystal distance IS 2 26 mm A second 'dummy' tube in the adoption cell is directed at a glass disc, I e an inert surface, but is otherwise identical to the first, connected to a second ionization gauge, IG-2 If the sticking probability on the crystal is non-zero, the reflected flux from the crystal Into IG-1 is reduced compared to that into IG-2 The sticking probability is obtained by comparing pressures P1 and P2 measured on IG-1 and IG-2, respectively If the fraction of randomly distributed molecules in the adsorption cell which would have entered IG-I but are blocked from doing so by the crystal IS f, then (s) is given by (s~ = (1/~ (P2-PO/P2 whence (s) is independent of the gauge cahbratlon factor An accurate evaluation of f is necessary Steckelmacher 12 has derived accurate formulae forffor various experimental configur- ations In the present work we have used an elhptlcally shaped crystal, with major axis 9 mm and minor axis 5 5 mm, for this geometry,j is accurately given by our original expression 6, and is evaluated as 0 69 The W crystal was cut and polished to within 1° of the 1100~ plane, and was cleaned m sztu by temperature cycling between 400 and 1700 K in an oxygen pressure of 10-6 torr for several days (until no CO desorptlon peak was observed) and subsequently annealed m tacuo at 2500 K Temperatures were measured using a W/25°o Re-W thermocouple spotwelded to the side of the crystal, independently calibrated using an optical pyrometer 3. Results and discussion 3.1. N 2 on W[001}. Six Independent runs were performed to evaluate the variation of (s) with surface coverage, with the 507

Upload: heather-bickley

Post on 15-Jun-2016

217 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Angle-integrated sticking probabilities: N2, H2 and D2 on W{001}

Vacuum/volume 31/numbers 10-12/pages 507 to 509/1981 0042-207X/81/100507-03502 00/0 Printed ,n Great Br,tam @ 1981 Pergamon Press Ltd

A n g l e - i n t e g r a t e d st ick ing probabi l i t ies: N 2, and D 2 on W{O01} Heather Bickley, James S Ar low, M A Morr is and David A King, The Donnan Laboratorms, The Untverstty, Ltverpool L69 3BX, UK

H2

A new techmque has been used to measure/nctdence-angle-averaged sttcktng probab/httes, (s>, for N 2, H u and D u on W{O01} A comparison with normal-mctdence values, s ±, obtained by molecular beam techntques provtdes a measure of the angular dependence of s For N 2 on W{O01}, the initial (zero-coverage hm/t) sttckmg probabthty (S>o ts 0 63+0 03, comparing closely wtth So ± = 0 5 8 ± 0 01 reported by King and Wells, indicating a relattvely small vartatton m s with incidence angle After saturatton and anneahng at 1000 K to form an underlayer, mtrogen Is readsorbed at 340 K with (S>o=O 1 For both H u and D 2 at 340 K we hnd (S>o to be 0 72±0 02, whtch ts slgmhcantly htgher than the value reported for So ~, 0 60±0 02 The stgnthcance of the results for adsorption mechamsms ts dtscussed

1. Introduction

Relatively few accurate, absolute methods are available for measur ing sticking probabfllUes at single crystal surfaces I For high sticking probabi l i t ies (0 1 ~<s~< 1) the reflection-detector techniques are the most accurate, where the reflection coefficient is measured by sampling the back-scat tered flux for a given molecular beam incidence angle Studies of the dependence ors on incidence angle by these methods are restricted experimentally to the range from normal incidence to ~ 6 0 ° f rom normal , and generally very small var ia t ions have been noted 2 4, with the exception of H2 and D 2 on W ~ l l 0 ) 4, where So is reported to decrease by ~60°~, over this range, and H z / D 2 on stepped Pt{ll l~j planes s, where the reactivity was highest when the incident beam was directed at the steps A compar i son of angle- averaged st icking probabi l i t ies , ( s ) , with n o r m a l incidence values, s~, f rom molecular beam experiments should provide a sensitive measure of the incidence-angle dependence of s and to this end we have applied a recently developed 6 absolute method for measuring ~ to the adsorp t ion of N2, H 2 and D2 on W~0011 Accurate absolute values of s ~ are available 3 v 8 for each of these systems, allowing a direct compar i son For H2 and D 2 on W I001 I, there is some controversy in the l i terature concerning the initial sticking probabi l i ty v'8, the coverage dependence of s 8 l o, and the existence of an isotope effect 7 8 10 tx ~4 We have

a t tempted to resolve these disagreements in the present study

2. Experimental

The principles and opera t ion of the u l t rahigh vacuum chamber used for measur ing absolute sticking probabi l i t ies have been described elsewhere 6 Gas is in t roduced to the adsorp t ion cell via a capillary, and a s teady state pressure is main ta ined In the cell by cont inuous pumping The crystal is suspended in front of a tube

connected to an ionizat ion gauge (IG-1) and possessing an orifice directed at the centre of the crystal The orifice diameter is 1 mm, and the (accurately measured) orifice-to-crystal distance IS 2 26 m m A second ' dummy ' tube in the adopt ion cell is directed at a glass disc, I e an inert surface, but is otherwise identical to the first, connected to a second ionizat ion gauge, IG-2 If the sticking probabi l i ty on the crystal is non-zero, the reflected flux from the crystal Into IG-1 is reduced compared to tha t into IG-2 The sticking probabi l i ty is ob ta ined by compar ing pressures P1 and P2 measured on IG-1 and IG-2, respectively If the fraction of r andomly dis tr ibuted molecules in the adsorp t ion cell which would have entered IG-I but are blocked from doing so by the crystal IS f, then ( s ) is given by

( s ~ = (1 /~ ( P 2 - P O / P 2

whence ( s ) is independent of the gauge cahbra t lon factor An accurate evaluat ion of f is necessary Steckelmacher 12 has

derived accurate formulae fo r f fo r var ious experimental configur- at ions In the present work we have used an elhptlcally shaped crystal, with major axis 9 m m and minor axis 5 5 mm, for this geome t ry , j is accurately given by our original expression 6, and is evaluated as 0 69

The W crystal was cut and polished to within 1 ° of the 1100~ plane, and was cleaned m sztu by tempera tu re cycling between 400 and 1700 K in an oxygen pressure of 10-6 torr for several days (until no CO desorpt lon peak was observed) and subsequently annealed m tacuo at 2500 K Tempera tures were measured using a W/25°o Re-W thermocouple spotwelded to the side of the crystal, independent ly cal ibrated using an optical pyrometer

3. Results and discussion

3.1. N 2 on W[001}. Six Independent runs were performed to evaluate the var ia t ion of ( s ) with surface coverage, with the

507

Page 2: Angle-integrated sticking probabilities: N2, H2 and D2 on W{001}

Heather Btckley et al Angle-integrated st,ckmg probabd,t,es N 2, H= and D u on W{001}

a • • i i

0 6 ?&- --o~ o~ . . •

<S>, S J" 0 5 ~\ 04 %

03

02 o

01 ,

0 ~ ~ 0 2 4 6 8

Coveroge / otoms cn52 x 1014

Figure l. Variation of <s) with coverage for N 2 on W ~0011 Each symbol represents a separate run Crystal temperature 340 K The full line is taken from the data for s ± against coverage of King and Wells 3. at 340 K

0 8 '

O6

O2

01 , J , ~ ,mw~x

0 5 10 15 20

Coveroge / otoms cr52 x 1014

Figure 2. Varmtlon of (6) with coverage of H 2 and D z on W{001] Each symbol represents a separate run Filled circles D z Crystal temperature 340 K Also shown are the data for ~ against coverage from King and Thomas s (KT) and MadeyV(M)

crystal at ~ 340 K The results are presented in Figure 1, together with the results for s ± obta ined by King and Wells 3 using a col l imated molecular beam technique Since coverages in the present work are dependent on the gauge cahbra t lon factor, which was not evaluated, the coverages have been normal ized to the value obta ined by King and Wells 3 at s ± = 0 03, viz 5 8 X 10 I4

a tom cm -2, thus assuming that ( s ) ~ s ~ near ' sa tura t ion ' Extrapola t ing to zero coverage gives an initial sticking probabi l i ty (6)o in the range 0 63 4- 0 03, which is only marginal ly higher than the value at no rmal inodence , So ± = 0 59_+ 0 013 The differences in the observed dependence of ( s ) and s ± on coverage are also relatively minor , and are p robab ly bet ter a t t r ibuted to experi- mental inaccuracies ra ther than to any real effect

Recently it has been repor ted ~3 that after adsorb ing 5 to 6 x 1 0 1 4 N a toms cm -2 on W[0011 at room temperature , anneal ing to ~ 1 2 0 0 K caused most of the adsorba te to be absorbed into the bu lk For example, the half-order L E E D beams whlch are character is t ic of the overlayer s t ructure are removed by this procedure However, a subsequent brief exposure to ni t rogen at r oom tempera ture was sufficient to restore the half-order beams to their original intensity This effect was studied in the course of the present work Ni t rogen adsorp t ion at 340 K was cont inued until ( s ) had fallen to ~ 0 02, the surface was annealed at 1200 K for I0 s and the crystal allowed to cool to 340 K before commencing r eadso rpuon The sticking p robab lh ty during the initial stages of readsorp t ion was found to be 0 1 4- 0 03 Thus, we est imate tha t the res tora t ion of half-order beam intensity during readsorp t ion repor ted in the L E E D study of Grlffiths and King a3 occurred with a total up take of ~ 6 5 × 1014 a toms cm 2 This confirms the observa t ion I3 tha t the desorpt ton spect rum after readsorp t lon showed an increase m peak area corresponding to a coverage of 6 5 × 1014 atoms cm 2, and rules out the possibility tha t the relatively small desorptxon increment resulted from permanen t absorp t ion into the bulk during t empera tu re pro- g r amnung Res tora t ion of half-order intensities thus seems to imply the (seemingly unhkely) process of the re turn of N a toms from the bulk to the overlayer dur ing r eadso rpuon at room tempera tu re

3.2. H2 and D2on W{001}. Results obta ined from the var ia t ion of ( s ) with coverage for bo th H 2 and De are shown in Figure 2, and are compared with the s ± da ta repor ted by Madey 7 and King and T h o m a s s for this system In each case, coverages have been

normal ized to the now generally accepted 7,s 14 sa tu ra t ion coverage for this system of 20 x l0 ~4 a toms cm -2

The only notable difference between the results for ( s ) and s ~ is in the initial sticking p robabd l ty Thus, all three sets of da ta show a virtually l inear fall in sticking probabi l i ty with coverage, a l though the curve of King and T h o m a s s does show a sharper fall at low coverages Taking the mean of the two da ta sets for So I as 0 55 4- 0 05, the value for ( s )0 , 0 72 4- 0 02, is significantly higher, suggesting an appreciable incidence angle dependence ofs for this system, i e (S)o/So ~ = 1 3 4-0 1 The virtually linear dependence of s on coverage confirms the results of Madey ~ and King and Thomas s, and contras ts with the da ta of T a m m and Schmidt l°, who reported tha t s is initially almost independent of coverage (Jaeger and Menzel 9 have also recently inferred that s is initially independent of coverage from ESD data) In all three sets of data, including the present, where a l inear dependence is found the crystal was cooled to the adsorpt ion tempera ture before dosing with hydrogen whereas T a m m and Schmldt ~° cooled the crystal in a hydrogen ambient , suggesting that their result may be an artefact produced by crystal cooling in the early stages of adsorpt ion

Results for D 2 adsorpt ion are also shown in Figure 2 Within experimental error, there is no difference between the H 2 and D 2

data, and we conclude that s(D2)/s(H2)= 1 0 0 4 - 0 0 4 , in agree- ment with King and Thomas s Madey ~ reported a value of 1 12 4- 0 12, while Schmidt and coworkers 4' lo, ~x report values in the range 14_+005 There is no obvious reason for this experimental discrepancy

3.3. General discussion. In considering the results, it is useful to derive a relat ionship between the incidence-angle-averaged stick- ing probabil i ty, ( s ) , and the differential value s o for a given incidence angle 0 (measured from the surface normal) The parameter ( s ) refers to the average value for a fully randomized Maxwel l -Bol tzman gas, the incident flux of gas molecules I ° can therefore be represented by a cosine law expression

I ° = I cos 0

where I is the flux at no rmal incidence Hence the angle integrated sticking probabi l i ty IS given by

(~/2 (*n/2 Fx/2 ( s ) = | s°l c o s 0 d 0 / | I c o s 0 d 0 = s°cosOdO (1)

do do Jo

5 0 8

Page 3: Angle-integrated sticking probabilities: N2, H2 and D2 on W{001}

Heather Btckley et al Angle-)ntegrated stsckmg probabdltles N2, H=

In order to relate ( s ) to s °, a funct ional dependence ofs ° on 0 must be assumed For example, a reasonable funct ion (based on energy accommoda t ion considerat ions) may be

s o = a{ 1 - e x p ( - - b/cosO)} (2)

where a~< 1 Exper imenta l values of ( s ) and s ± thus allow an evaluat ion of a and b f rom equat ions (1) and (2)

For N 2 o n W{001}, we observe a small dependence o f s on 0 (So±/(S)o = 1 09_+ 0 04) F r o m equat ions (1) and (2), the da ta can be fitted with a = 0 8 and b = 1 3, which applies a l imiting value for s as O~n/2 of 0 8 The funct ional form of equa t ion (2) with these values of a and b is shown in Figure 3, together with the da ta of

s 1 0 ~H2" 0 8 ~ ~ N z

/iNz

0 6

o % 3'o do 9o Incidence Angle / °

Figure 3. Full lines calculated variation of s with incidence angle from equation (2) for H2 and N 2 on W{ 001 ), consistent with the (s) and s ~ data for these systems Dashed lines experimental data of Stembruchel and Schmldt 4

Stelnbruchel and Schmldt 4, showing good agreement As dis- cussed in detail elsewhere 1' 3.1 s, 16 adsorp t ion into the dissoclat- lvely adsorbed fl state occurs t h r ough initial t rapping of the incident molecule into a weakly b o u n d precursor state, and subsequent transfer to the chemlsorbed state The t rapping probabi l i ty ~t at no rmal incidence was est imated as 0 60 from the low tempera tu re l imiting value ofs 3, ~ was found to decrease with increasing gas t empera tu re 3 The present results indicate an Increase in ct wi th incidence angle, the increase is smaller than would be ant ic ipated if only the normal componen t of incident molecule t rans la t iona l energy must be accommoda ted We note tha t a cosine law desorptlon dis t r ibut ion implies tha t s is independent of 0, as discussed recently by Comsa 1 ~, the angular d is t r ibut ion of N 2 backscat tered from steel, a l u m m l u m and glass is close to a cosine law dis t r ibut ion 1 s The present results would imply a slight f lat tening of the desorpt lon d is t r ibut ion compared with cos 0, and are discussed in relat ion to the desorpt lon flux dis t r ibut ion observed for N 2 f rom W{310} In Ref 19

The da ta for H 2 on W{001 } show a larger angular dependence of s, compar i son with the s ± da ta of Madey 7 gives So'/(S)o = 1 4, and with the da ta of King and T h o m a s s gives a rat io of 1 2 F rom equat ions (1) and (2) above, a ra t io of 1 2 is ob ta ined with a = 1

and D= on W{001 }

and b = 0 92, which Implies tha t s tends to unity as 0 ~ g / 2 The funct ional dependence is shown in Figure 3, together with the da ta of Stelnbruchel and Schmldt ¢ for H 2 on W{001 } In this case there appears to be an irreconcilable discrepancy with their da ta A higher So±/(S)o rat io implies a different funct ional dependence of s o f rom that suggested by equa t ion (2) The contras t between N 2

and H 2 may reflect different adsorp t ion mechanisms for these two gases Based on an analysis of available experimental data, King 1 has suggested tha t H E adsorp t ion on Wl0011 is domina ted by direct transfer at an incident site to the &ssoclatively chemlsorbed fl state, wi thout t rapping in a precursor state In the absence of the ' scrambling ' influence of the precursor state, it would therefore appear tha t direct ional effects are impor t an t We would again predict, by the principal of microscopic reversibility, a ' squashed ' cosine law desorp t lon dis t r ibut ion for this system

The absence of an isotope effect in the sticking probabi l i ty of H 2 and D E on W{001 } is ano ther impor t an t indicator of adsorp t ion mechanism If sticking were domina ted by p h o n o n transfer to the lattice, a s t rong mass dependence for the incident particle might be expected, since the sticking probabi l i ty would be a sensitive funct ion of the mass rat io ~ On the other hand, if sticking IS domina ted by electron-hole pairing, as has been recently suggested20 22, the ISOtope effect would be small, and domina ted by the different mean velocities of H 2 and D : The observed increase in s with incidence angle for this system is also consistent with electron-hole pmrlng, since the impact t ime is increased at grazing incidence

References

i D A King, Sohd St Mat Sct, 7, 167 (1968) 2 D A King and M G Wells, SurfScz, 29, 454 (1972) a D A King and M G Wells, Proc R Soc, Lond, A339, 245 (1974) 4 C S Stelnbruchel and L D Schmldt, Phys Ret B10, 4209 (1974), Phys Rel Lett, 32, 11 (1974) 5 R G Gale, M Salmeron and G A Somorjal, Phys Ret Lett, 38, 1027 (1977) 6 M A Morns and D A King, Vacuum, 30, 23 (1980) 7 T E Madey, SurfScl, 36, 281 (1973) 8 D A King and G Thomas, SurfScl, 92, 201 (1980) 9 R Jaeger and D Menzel, SurJace Scl, 100, 581 (1980) 10 p W Tamm and L D Schmldt, J Chem Phys, 51, 5352 (1969) 11 p W Tamm and L D Schmldt, J Chem Phys, 55, 4253 (1971) 12 W Steckemacher, Vacuum, 30, 261 (1980) 13 K Grlffiths and D A King, Proc 4th Int ConJ Sohd Surfaces, Suppl Le Vide, 201,237 (1980) 14 L C Feldman, R J Sllverman and ] Stensgaard, SurfScl, 87, 410 0979) is K C Janlda, J E Hurst, C A Becker, P J Cowln, L Wharton and D J Auerbacher, SurfSc4 93, 270 (1980) 16 S P Slngh-Boparal, M Bowker and D A King, SurJSct, 53, 55 (1975) 17 F C Hurlbut, J Appl Phys, 28, 844 (1957) l a g Comsa, Proc 7th Int Vac Congr and 3rd Int Conf Sohd SurJaces, Vienna, 1977, p 1317 19 R C Cosser, S R Bare, S M Franos and D A King, Vacuum, 31,503 (1981) 20 G P BrlVlO and T B Grlmley, SurfSct, 89, 226 (1979) 21 j K Norskov and B I Lundqvlst, SurfSct, 89, 251 (1979) 22 R Brako and D M Newns, Solid State Commun, 33, 713 (1980), D M Newns, Vacuum, 31,685 (1981)

509