Oxygen on Ni(110): Surface phases and related absolute coverages

313

Surface Science 175 (1986) 313-324 North-Holland, Amsterdam

OXYGEN ON Ni(ll0): SURFACE PHASES AND RELATED COVERAGES P.R. NORTON

*, P.E. BINDNER

and T.E. JACKMAN

ABSOLUTE

**

Chemistry and Materials Division, Atomic Energy of Canada Limited, Research Company, Chalk River Nuclear Laboratories, Chalk River, Ontario, Canada KOJ IJO

Received 3 February 1986; accepted for publication 15 April 1986

The adsorption of oxygen on Ni(ll0) was investigated by nuclear reaction analysis (NRA), XPS, A#, temperature pro~ammed reaction spectroscopy (TPRS) and LEED. At 423 K, (3 x I), (2 X 1) and (3 X 1) phases are formed in sequence with increasing 0, exposure. The coverage in the (2 X 1) phase was determined by NRA, the coverages in the other phases being determined via this calibration by XPS, TPRS and A$. Contrary to previous reports, the maximum in the intensity of half-order beams from the (2 x 1) phase is associated with a coverage of (5.6 -f.0.5) x lOI 0 atoms cme2 or 0.49 + 0.05 monolayers, and not 0.25 monolayers. The two (3 X 1) phases are associated with B = 0.33 kO.03 and 0.64+0.06 monolayers respectively. Oxygen adsorbed at 295 K is not at thermodynamic equilibrium. Annealing to T > 400 K causes significant decreases in A+ and the formation of the (2 x 1) phase for B > 0.3.

1. Introduction Oxygen chemisorption on Ni(ll0) has been the subject of many recent investigations (for example, see ref. [l] and references therein). There seems to be general (but not unanimous, see ref. [2]) agreement about the sequence of LEED patterns that form as a function of exposure ((3 x l), (2 X l), (3 X 1)). There is less agreement about the coverages at which these pattern are said to appear [l]. The coverages (6) expected from the LEED patterns assuming simpte unit cells would be 0.33, 0.5 and 0.67 monolayers (B = 1 = 1.14 x lOi 0 atoms cm-2). Th ere are only two measurements in the literature which depend upon absolute calibration methods or upon the use of well defined standards [2,3]. The X-ray emission measurements of Mitchell et al. [2] yielded a coverage of 0.25 at the maximum intensity of the (2 x 1) phase, while the RBS data of Smeenk et al. [3] gave 0 = 0.28 t 0.03 and 0.36 k 0.04 for two * Present address: Canada. ** Present address: Ottawa, Ontario,

Department Microstructural Canada.

of Chemistry, Sciences

University Laboratory,

of Western National

0039-6028/86/$03.50 0 Elsevier Science Publishers (North-Holland Physics Publishing Division)

Ontario,

Research

B.V.

London, Council

Ontario,

of Canada,

different channeling/blocking directions. The apparent discrepancy between expectation and actual coverage of the (2 X 1) phase was removed when it was found that the (2 x 1) phase is reconstructed and that it is the Ni atom periodicity that gives rise to the doubling of the unit mesh [3]. There is still a discrepancy between the two “absolute” measurements however and the sequence of LEED patterns is hard to interpret if the (2 X 1) phase corresponds to 0.25 monolayers. In this paper we report measurements of the coverage in the (2 X 1) phase by nuclear reaction analysis (NRA) which is established as a reliable and accurate absolute method [4,5]. The coverages in the other phases were determined by XPS and temperature programmed reaction spectroscopy (TPRS) using the (2 X 1) phase for calibration. The measurements were correlated via precise work function (A+) data with measurements of the intensity of the fractional order LEED beams to ensure that the coverage of a particular surface phase was maximized. It should thus be possible for workers at other laboratories to achieve identical experimental conditions through use of A+ or LEED intensity measurements. Exposure is thereby eliminated as an experimental variable.

2. Experimental The experiments were carried out in two UHV chambers. The first system has facilities for XPS, UPS, A+, LEED and line-of-sight thermal desorption spectroscopy (TDS). This apparatus will be fully described elsewhere [6]. Briefly the system is designed to allow quasi-simultaneous acquisition of TDS, A+ and LEED data. The data acquisition and control system will scan up to I2 masses and 4 analog inputs in sequence with cycle times of < 1 s. In the present experiments cycle times of -C 0.3 s were used. The LEED diffraction beams were monitored with a lens/aperture/photomultiplier assembly that was mounted on a programmable X-Y stage. By suitable choice of aperture either integrated intensities or beam profiles could be monitored. The sample temperature programmer permitted linear heating or cooling rates of 0.01 to 30 K s-i over the range 80-1200 K. During LEED observations the heating supply was chopped at 24 Hz antiphase to a 60 V retard pulse on the LEED retard grid. This enabled LEED measurements to be carried out during heating or cooling. The second, 2-level chamber, is equipped for Rutherford backscattering (RBS) and nuclear reaction analysis (NRA) as well as more conventional surface science techniques (LEED, Auger, Kelvin probe for A+ measurements, etc.). A simpler, single level version of this apparatus has been fully described

[71. The nickel (110) samples were cut from the same boule used in earlier experiments [8]. One of the samples was in fact used in both the previous

P.R. Norton et al. / Oxygen on Ni(l IO): surface phases and coverages

315

experiments on oxygen interaction [S] and the recent D*~i(llO) work [9,10]. The samples were oriented within lo of the (110) plane. The cleaning procedure involved Ar+ ion sputtering (3-5 keV) for many hours at high temperatures (> 1000 K) and oxidation and reduction. The carbon and oxygen backgrounds in the bulk of the crystal (top l/2 pm) were measured using a beam of deuterons of incident energy 972 keV and the ‘*C(d, p) 13C and i60(d, p)170 reactions. Values of - 1 x lOI C atoms and 2 x 1014 0 atoms cmm2 were routinely obtained, i.e. bulk concentrations of - 50 to 100 ppm (atomic). No sulphur segregation occurred during annealing and no other impurities were detected by XPS or Auger spectroscopy. The final cleaning treatment involved sputtering at 295 K followed by a brief oxidation/reduction cycle. Final cleanliness was judged by the magnitude of A# upon Dr saturation at 170 K. This has been shown to be a very sensitive measure of surface purity [ll]. Values of 500 mV were routinely obtained and the sequence of phase transitions during adsorption or desorption [lO,ll] could be accurately reproduced. Both samples gave identical results. The oxygen coverage was determined from the yields of the 160(d, p) I70 reaction which was calibrated in-situ against a 25 nm thick Ta,O, standard. The standard can be moved to exactly the sample position, thus eliminating all is attainable 141, with the geometrical corrections. Accuracy of - 2-3s precision of a given experiment being determined by the counting statistics. It was found that the work function maximum at 295 K was reproducible from run to run and crystal to crystal (560 _t 10 mV) so this was used as the transferable coverage calibration point. Relative coverages were then determined by measurement of 0 1s peak areas (XPS) or by temperature programmed reaction spectroscopy (TPRS). In the latter method the adsorbed oxygen overlayer was reacted with Dz while A# and the partial pressure of D,O were simultaneously monitored. The rate of reaction depends upon initial coverage and the instantaneous temperature during a temperature ramp. If the initial cover is > 0.4 monolayers, the temperature at which rapid reaction occurs overlaps that at which dissolution into the bulk becomes significant (T> 750 K). Thus the integrated areas of the PD2O versus time trace give a reliable measure of the initial coverage only for 0 I 8 I 0.4. In this coverage range XPS and TPRS data were in excellent agreement.

3. Results

Because the data are correlated via A$ measurements, a brief description of the work function behaviour is necessary as our data differ in some respects

af 3 /

I

1.0

EXPOSURE IL1 , 10 20

1 30

I

2.0

40

so

I

I

3.0

EXPOSURE @s-s x 10'1 Fig. 1. Variation of work function (A+) of Ni(llO) surface with 0, exposure: (a} 295 K, (h) 295 and 423 K. Arrow indicates decrease in Arp when adiayer produced at 295 K is annealed to 423 I(.

from earlier work 1121. The key observation in the present work is that A+ is temperature dependent in the range investigated (273 -C T-c 500 K). This is illustrated in fig. lb which shows A$ versus 0, exposure at 295 and 423 K for exposures up to -2.Z~~O-~Pas(-2‘6LwhereL=~angmnir~l x IO-” Torr s). The first maximum in A+ (A+,,, ) at 295 K is 560 -t 10 mV. At 423 K the maximum is 310 f 10 mV. Annealing the 295 K surface to 423 K causes the work function to decrease to the value it would have achieved at that exposure at 423 K. XPS measurements showed that no desorption or

P.R. Norms et al. _/ Oxygen an Ni(I IO): surface phases and coverages

317

dissolution of oxygen ocwrred during heating from 295 to 423 K. Fig. la also shows the decrease in A# and a second maximum at higher exposures at T = 295 K, a region not further discussed in this paper. The behaviour is in good agreement with earlier work [12] except that our saturation value of A# ( > 6 x Me3 Pas) in the presence of 0, is nearer the clean surface value. This will be discussed in a forthcoming paper. As described above, the point at A&,,,, at 295 K was thereafter used as the intercalibration point.

The clean sample was first exposed to Uz at - 2 x fO_’ Pa at 295 K until A@= 560 rt 10 mV (A+m,,) and then repositioned for oxygen analysis using the %(d, p) I70 reaction. The oxygen coverage thus determined was 0.43 + 0.04 monolayers or (4.9 + OS) x 1014 0 atoms cm-‘. The same coverage was achieved by oxygen exposure at 423 K, to A+ = 310 + 10 mV. The other point which should be essentially exposure independent is the coverage after exposures of - lo-* to 10-t Pas (100-1000 L). Although continued oxide growth does occur 121, it is very slow at 295 K and this “practical” saturation value should be reproducible from laboratory to laboratory. The oxygen coverage was found to be (3.4 rt 0.1) X 1Ol5 0 atoms cm-‘. Note that this value is within experimental error equal to the number of nickel atoms in three nickel (110) planes.

3.2.2. Carbon monoxide admrption Measurements were also made of the saturation CO coverages at T < 170 K. Under these conditions an excellent (2 X 1) p2mg LEED pattern was formed. The coverage was measured using both the ‘*C(d, p)13C and 160(d, p)“O reactions to be (1.05 k 0.05) X 1Or5 CO molecules cme2, or 8 = 0.92 + 0.04 monolayers. The ideal coverage would be 1.0 for this structure but, presumably, steric hindrance prevents this coverage being reached. We have found saturation coverages of - 1 x lot5 CO molecules cme2 on a variety of surfaces and have attributed this (51 to the packing density at which ne~~bo~ng molecules are separated by distances approac~ng the sum of their Van der VVaalsradii, 3.3. A+ versus coverage relationship

Fig. 2 shows the A+ versus B relationship for Oz adsorption at 423 K deduced by TPRS and XPS. The calibration point is for A&,,,, = 310 f 10 mV which carresponds to B = 0.43 + 0.04. The relationship is very non-linear and as discussed below this probably reflects the reconstruction into the (2 x 1)

P.R. Norton et al. / Oxygen on Ni(liO):

surface phasesand coverages

8 (MONOLAYERS) Fig. 2. Variation

of work function

A+ with oxygen coverage

on Ni(ll0);

T = 423 K

phase. Fig. 2 can be used to establish the coverage scales in the following figures up to B 2 0.45. The higher coverages shown in figs. 3 and 5 were determined by XPS measurements through the relative areas of the 0 Is peaks. 3.4. LEED

~bser~a~iQ~s

3.4.1. T= 423 K The oxygen-induced overlayer LEED beams are very broad at 295 K, indicating considerable disorder, while at 423 K and higher temperatures they are much sharper. The sequence of LEED patterns at 423 K is that reported by most other authors, i.e. (3 x l), (2 x l), (3 X 1). LEED data (beam energy = 47 eV) which were obtained as a function of 0, exposure by rastering over the appropriate diffraction beams, are shown in fig. 3. Fig. 3b shows the integrated intensities of the fractional order beam from the first (3 x 1) (circles), (2 x 1) (solid circles) and second (3 x 1) phases (plusses) respectively. The triangles on the (2 x 1) intensity versus exposure curve are data where the separation of intensities of the (2 X 1) and second (3 X 1) phase is difficult to effect. Fig. 3a shows measured FWHM of the beams. The changes in widths are those that might be expected from the changes in the long rate order. XPS and TPRS data indicate that the coverages associated with the maxima in the (3 x l), (2 x 1) and (3 x 1) LEED intensities are 0.33 rt 0.03, 0.49 _t 0.04 and 0.64 + 0.06 monolayers respectively.

P.R. Norton et al, / Oxygen on Ni(l IO): surface phases and coverages

319

EXPOSURE IL)

5i

4-

al

0, ON NilllO), 1 = 423K

?= = Ld

3

4

2

z 3 IL

1 0.

,

0.33 0150

8 o (3 x 1) LOW COVERAGE l (2 x 11 -I- (3 x 1) HIGH COVERAGE

10

12

14

16

EXPOSURE (Pa-s x 104)

Fig. 3, Variation of (a) widths and (b) integrated intensities of LEED beams from various oxygen-induced surface phases on Ni(ll0); adsorption temperature is 423 K: (0) third-order beam from low coverage (3 x 1) phase, (0) half-order beam from (2 X 1) phase, (+) third-order beam from high coverage

(3 X 1) phase.

3.42. Annealing of295 K oxygen overlayer to 423 K Annealing of the surface produced by oxygen adsorption at 295 K, to 423 K decreases A+ from 560 + 10 mV to 310 k 10 mV. This process is accompanied by the intensification and sharpening of the (2 X 1) overlayer beams. This is illustrated in fig. 4 in which A+ and integrated LEED intensity data (half-order beam of (2 x 1) phase) were acquired during annealing (1 K s-’ heating rate) of a surface produced by adsorption to A&,,, at 295 K. The integrated beam intensities were corrected for the Debye-Waller effect by subsequent measurements on the annealed surface as a function of T. It is clear that the decrease in A+ and increase in integrated intensity are related. Also shown in fig. 4a are the FWHM of the beams showing increased ordering during the annealing process. Since it is believed that the (2 X 1) surface is reconstructed [1,3] it is clear that the large change in dipole moment at constant coverage of 0.43 + 0.04 monolayers is associated with the reconstruction of the nickel surface. If the

320

P.R. Norfon et al. / Oxygen on Ni(l IO): surface phases and coverages I

I

I

I

8-

02 ON Ni(ll0)

l,_

a) FWHM OF HALF-OROER BEAM DURING ANNEAL FROM 295K, HEATING RATE 1 K s-’ I

I

I

I

b) A#+, AND INTEGRATED INTENSITY (01 OF HALF-ORDER BEAM DURING ANNEAL

I ’

T(K)

Fig. 4. Variation of (a) full width half maximum of half-order

(0) and (b) A+ (0) and Integrated intensity (0) beam of (2 X 1) phase during annealing of adlayer produced by 0, adsorption at 295 K to A# = 560 mV (B = 0.43 t_ 0.04 monolayers). Heating rate is 1 K s- ‘.

missing-row model for the (2 x 1) reconstruction is correct [1,3] then the nickel atoms still reside on bulk-like lattice sites. Single alignment measurements by Rutherford backscattering (RBS) would therefore not yield new information and hence were not attempted at the time of these measurements. It is clear that the surface produced by adsorption at 295 K is metastable, i.e. it is not at thermodynamic equilibrium. It is therefore critical to work at temperatures at which the equilibrium structures are formed, i.e. T > 400 to 420 K.

3.4.3. Coverage calibration at maximum of (2 x 1) phase The coverage at the ma~mum intensity of the (2 X 1) phase was determined in two ways. In the first, oxygen was adsorbed to A&,,,, = 560 f 10 mV at 295 K, the surface annealed to 423 K and then exposed to more 0, at 423 K while the half-order beams were monitored. These data showed that the half-order beams intenGfied with increased exposure at 423 K. In fact, an extra (4-6) X lo-’ Pas (- 0.3 to 0.4 L) were required to reach the intensity maximum

P.R. Norton et al. f Oxygenan Ni(f 10): surface phases and coverages

321

O2 ON Ni(llO1,T = 423K INTENSITY OF HALF-ORDER VERSUS COVERAGE

BEAM

t 0

0.1

0.2

0.3

0.4

0.5

t9 ~MONOLAYERS) Fig. 5. intensity

of half-order

beam of (2

X 1)

phase during K.

oxygen

adsorption

on Ni(ll0)

at 423

This immediately establishes that 0 at 1,, is > 0.43 f 0.04 monolayers. In addition, measurements were made in which A+ (and hence, 6’) and a half-order beam intensity were monitored as a function of exposure at 423 K. These data are illustrated in fig. 5. The coverage calibration above 6 = 0.4 was established by XPS and the 01s peak area ratios from both the above methods showed the maximum in the (2 X 1) pattern occurred at 8 = 0.49 rf: 0.04 monolayers.

(I,,,).

3.5. Evidence for minimum coveruge for (2 X 1) reconstruction The behaviour of A+ upon annealing a layer produced by adsorption at 295 K provides a signature for the reconstruction. Fig. 6 shows the change in A+ upon annealing to 423 K plotted against the value of A+ after adsorption at 295 K. The change in A+ increases rapidly at higher initial coverages. The initial coverages can be deduced from the final A+ values at 423 K through the data in fig. 2. These are shown at the top of fig. 6. Additional data (not illustrated) indicate that the Acp versus 8 scale at 295 K is approximately linear to B - 0.4. In addition, near A#,,,, A+ decreases slowly with time in vacuum even at 295 K indicating that the reconstruction is occurring at room temperature. This was confirmed by LEED observations. The onset of the reconstruction signalled by decreasing A+ on annealing is not as abrupt as that indicated

P.R. Norton et al. / Oxygen an Ni(ilO):

322

,Y

300

s!

FUNCTION CHANGE ON ANNEALING

295K 0-OVERLAYER

2 u r ;

WORK

surface phasesand couerages

TO 423K

200

z z 0 a" 100 z % a ki g

0

100

200

300

400

500

600

Ag, 295K (in”) Fig. 6. Work function (A+) change on annealing of 295 K 0 overlayer to 423 K. ‘The initial 0 coverage and corresponding A+ value at 295 K are shown on the abscissa.

by the LEED data in fig. 5. Perhaps A+ is a more sensitive probe of the local order and can detect the reconstruction at an earlier stage than LEED observations which require long range order. Fig. 6 indicates that coverages as low as - 0.1 monolayers are sufficient to permit some reconstruction at 423 K.

4. Summ~

and discussion

At T> 423 K the sequence of LEED patterns observed on exposure of a Ni(ll0) surface to 0, is (3 X 11, (2 x l), (3 X 1).By a combination of methods we have established that the coverages associated with the maxima in the fractional order beams from these phases are 0.33 k 0.03, 0.49 F 0.04 and 0.64 rf: 0.06 monolayers respectively (Fig. 3b). The absolute magnitudes rely upon a measurement of i? at A+m,, by nuclear reaction analysis and the relative coverage scale was then established by XPS and TPRS. These values are higher than those determined previously. We would expect the measurement of Mitchell et al. [2] by X-ray emission to be reliable since it is based upon a calibration of oxide standards. These authors found a value of 0.25 monolayers at the (2 x 1) maximum although it is not clear how the maximum intensity was determined. We do not understand the origin of this

P.R. Norton et al. / Oxygen on Ni(l10):

surface phases and cooerages

323

discrepancy, but offer two additional pieces of evidence that our present values are reasonable. First, our coverage versus exposure plots yield an initial sticking coefficient (S,) of > 0.8. S,, is believed to be 1 (ref. [l] and references therein). If the (2 X 1) maximum occurs at 0.25 monolayers S, would be - 0.4 to 0.5, a lower value than currently accepted. The value averaged over the first 0.4 monolayers was found to be 0.4 in the present work. Second, we have measured the saturation CO coverage at low temperature (< 170 K) by NRA. If the present oxygen coverage measurements are in error by a factor of 2, then so must the CO coverage be in error. This would yield a coverage for the (2 x 1) p2mg of 0.46 monolayers which is so low a value as to be untenable. Mitchell et al. [2] also measured the oxygen coverage as a function of exposure and after IO-* to 10-i Pas (100 to 1000 L) exposure found values of - 2.4 X lOi 0 atoms cm-‘, also smaller than found in the present work. A coverage of 3.4 X lOI 0 atoms cmm2 was only achieved after - lo2 to lo3 Pas (lo6 to lo7 L). Given the possible strong dependence of oxidation rate upon temperature, defect structure, etc., it is not possible to discuss this difference further, except to caution against using our “practical” saturation value for calibration purposes until the discrepancy is resolved. The point we wish to make is that the difference between our data and those of Mitchell et al. [2] is not removed by a single scaling factor that might indicate a calibration error. The present A+ data should be comparable with those of Benndorf et al. 1127. There are, however, a considerable number of internal inconsistencies in their paper with regard to the value of A&,,,, at 300 K, quoted values varying from - 400 to 500 mV. Also they did not use the hydrogen-adsorption test for surface purity and perfection. The qualitative behaviour found by the two groups is very similar as shown by fig. 1, but the final value of A+ near “practical” saturation (- 10e2 to 10-l L) was temperature dependent in our work, remaining higher at lower T probably because of lower mobility preventing complete oxidation. In fact, even at lower coverages we observe greater temperature dependence of A+ than did Benndorf et al. The variability of the values of A&,,,, in ref. [2] could be attributed to differing degrees of ordering of the (2 X 1) phase. If the coverage measurements of Smeenk et al. [3] are correct, then the question arises as to the extent of the reconstruction of the surface into the (2 X 1) phase in their measurements. Clearly it is essential to have an independent monitor of the intensity of the LEED beams so that measurements are made under the same surface conditions. References [l] T. Engel and K.H. Rieder, Surface Sci. 148 (1984) 321. [2] D.F. Mitchell, P.B. Sewell and M. Cohen, Surface Sci. 69 (1977) 310.

324 f3j f4] (51 [6] I?] [S] [QJ

P.& Norton et at. / Oxygen on Ni~l~~~: surfnce phases and coverages

R.G. Smeenk, R.N. Tromp and F.W. Saris, Surface Sci. IO? (1981) 429 J.A. Davies and P.R. Norton, Nucl. Instr. Methods 168 (lQ80) 611. P.R. Norton, J.A. Davies and T.E. Jackman, Surface Sci. 122 (1982) L593. P.R. Norton and P.E. Bindner, to be published. G. Sitter, J.A. Davies, T.E. Jackman and P.R. Norton, Rev. Sci. Instr. 53 (1982) 797. P.R. Norton, R.L. Tapping and J.W. Goodale? Surface Sci. 65 (1977) 13. T.E. Jackman, J.A. Davies, P.R. Norton. W.N. Unertl and K. Griffiths, Surface Sci. 141 (1984) L313, HO] K. Griffiths, P.R. Norton, J.A. Davies, W.N. Unertl and T.E. Jackman, Surface Sci. 152,053 (1985) 374. fll] V. Per&a, K. Christmann and G. Ertf, Surface Sci. 136 (1984) 307. [12] C. Benndorf, E. Egert, C. Nijbl, H. Seidei and F. Thieme, Surface Sci. 92 (1980) 636.