3808
J. Phys. Chem. 1987, 91, 3808-3814
Adsorption and Decompositlon of Trimethylphosphlne on Pt( 111) G. E. Mitchell, M. A. Henderson, and J. M. White* Chemistry Department, University of Texas at Austin, Austin, Texas 78712 (Received: November 25, 1986; In Final Form: March 19, 1987)
The chemistry of trimethylphosphine (P(CH,),) on a Pt( 1 1 1 ) surface was studied with temperature programmed desorption, temperature programmed secondary ion mass spectrometry, Auger electron spectroscopy, and high-resolution electron energy loss spectroscopy. P(CH3), is molecularly chemisorbed at 100 K and bonds through the phosphorus atom for coverages less than 0.25 monolayer (ML). At 100 K, the sticking coefficient is large and independent of the coverage. Upon heating, a maximum of 0.25 ML of P(CH3)3decomposes to produce 0.18 ML CH4(g), 0.78 ML H2(g), 0.25 ML adsorbed phosphorus, 0.58 ML adsorbed carbon, and 0.003 ML C2H4(g). Overall activation energies and preexponential factors for methane formation (23.2 f 0.5 kcal/mol and (7 2) X IO9 s-I) and ethylene formation (22 f 4 kcal and 2 X lO9*I SKI)were calculated from TPD data. The formation of CH4(g) and C2H4(g)is discussed and compared to the formation of these products from decomposition of CH31, CH20,and CH2CO on Pt( 111). A reaction mechanism which explains interesting aspects of the experimental results is proposed.
*
1. Introduction
Analogies between the chemistry of coordination compounds and adsorbates at metal surfaces can be helpful in understanding complex heterogeneous reaction mechanisms.l To make these comparisons, we need a data base over a wide enough variety of surface reactions in order to recognize where similarities and differences are likely to occur. Substituted phosphines are very common ligands in coordination compounds and understanding the bonding of these molecules at metal surfaces can contribute to the required data base. This work is part of a larger investigation of the surface chemistry of substituted phosphines (PR3 where R = halogens, CH,, OCH,, or hydrogen) on metal surfaces. There have been several studies of PF3*v3and PH,@ adsorption on clean metal surfaces, but to our knowledge, there has been no previous study directly related to the chemistry of trimethylphosphine on well-characterized metal surfaces. It has been used as a displacement reagent in several recent studies concerned with the chemistry of hydrocarbons adsorbed on transition metals.1@'2 As part of one of the latter studies, Tsai et a L f l reported the formation of methane at 483 K in TPD after room temperature P(CH3)3adsorption on P t ( l l 1 ) . Recently we have investigated the surface chemistry of ketene (CH2C0)'3,'4 formaldehyde (H2CO),I5and methyl iodide CH3II6on a P t ( l l 1 ) surface. In TPD, these adsorbates evolve methane at 260-380, 245-255, and 270-300 K, respectively. The large range in temperature for evolution of the same product (CH,) from different sources on the same surface is interesting and warrants further study. (1) Muetterties, E. L.; Rhodin, T. N.; Band, E.; Brucker, C. F.; Pretzer, W. R. Chem. Rev. 1979, 79, 91. (2) Nitschke, F.; Ertl, G.; Kuppers, J. J . Chem. Phys. 1981, 74, 5911. (3) Garfunkel, E. L.; Maj, J. J.; Frost, J. C.; Farlas, M. H.; Somorjai, G. A. J . Phys. Chem. 1983, 87, 3629. (4) Mitchell, G. E.; Henderson, M. A.; White, J. M., submitted to Surf. Sci. (5) Hegde, R. I.; Tobin, J.; White, J. M. J . Vac. Sci. Techno/. 1985, A3, 339. (6) Hegde, R. I.; White, J. M. J . Phys. Chem. 1986, 90, 2159. (7) Kiskinova, M.; Goodman, D. W. Surf.Sci. 1981, 108, 64. (8) Hegde, R. 1.; White, J. M. Surf. Sci. 1985, 157, 17. (9) Greenlief, C. M.; Hegde, R. I.; White, J. M. J . Phys. Chem. 1985, 89, 5681. (IO) Friend, C. M.; Muetterties, E. L. J . Am. Chem. SOC.1981, 103,773. ( I 1 ) Tsai, M.-C.; Muetterties, E. L. J . Am. Chem. SOC.1982, 104, 2534. (12) Tsai, M.-C.; Friend, C. M.; Muetterties, E. L. J . Am. Chem. SOC. 1982, 104, 2539.
(13) Mitchell, G. E.; Radloff, P. L.; Greenlief, C. M.; Henderson, M. A.; White, J. M. Surf.Sci. 1987, 183, 403. (14) Radloff, P. L., Mitchell, G. E.; Greenlief, C. M.; White, J. M.: Mims, C. A. Surf.Sci. 1987, 183, 377 (15) Henderson, M. A.; Mitchell, G. E.; White, J . M. Surf. Sci,, in press (formaldehyde work).
0022-3654/87/2091-3808$01.50/0
The work reported here is a detailed study of the adsorption and thermal reaction of trimethylphosphine on P t ( l l 1 ) using high-resolution electron energies loss spectroscopy (HREELS), temperature programmed desorption (TPD), secondary ion spectroscopy (SIMS), and Auger electron spectroscopy (AES).
2. Experimental Section The experiments were performed in an ion pumped UHV Torr. Located radially chamber with a base pressure of 2 X around the chamber on a single level are a quadrupole mass spectrometer used for (SIMS) and TPD, a single-pass cylindrical mirror analyzer for (AES), and a high-resolution electron energy loss spectrometer. Details of the experimental apparatus and the sample-cleaning procedure are presented elsewhere.I3 Trimethylphosphine was obtained by thermal decomposition of its silver iodide complex ( [P(CH3),]A11),. The complex was obtained as a powder (Aldrich Chemical Co., Inc.) and ca. 5 g was placed in a glass sample vial and evacuated. Heating to approximately 375 K with a hot air gun provided sufficient pressure of P(CH3)3for dosing, leaving the AgI as a solid. The purity of P(CH3)3was confirmed by mass spectrometry and careful monitoring with AES failed to detect any Ag or I on the sample. P(CH3)3was dosed through a directional doser which was a long open-ended '/,-in. stainless steel tube.17 The enhancement factor (defined as the ratio of the isotropic pressure which would give an equivalent flux at the sample divided by the actual ion gauge reading during a dose (2 X lo-" Torr)) was (1.3 X 2) X IO4. This factor was estimated by assuming a sticking coefficient of unity. Note that this calculation ignores ion gauge sensitivity factors which, if included, would increase the enhancement factor. The standard flux used here was selected so the ion gauge pressure increased by 2 X lo-" Torr. Temperature programmed desorption (TPD) was performed with a heating rate of 12 K/s. The input of the mass spectrometer has a central stop plate (integral to the ion energy filter) which prevents line-of-sight access of desorbing molecules to the ionizer during TPD. Desorption of molecular P(CH,), was therefore not reliably measured since it readily adsorbs or undergoes displacement or decomposition reactions at the chamber walls. SIMS measurements were performed with a primary Ar' ion current of 2.8 f 0.1 nA and beam energy of 500 eV. Temperature programmed SIMS (TPSIMS) used a heating rate of 4.2 K/s. HREEL spectra were taken by averaging many scans (20-65). This data was collected in the pulse counting mode (one point per (16) Henderson, M. A,; Mitchell, G. E.; White, J. M. Sur$ Sci., in press (methyl iodide work). (17) Mitchell, G . E.; Ph.D. Dissertation, University of Texas at Austin, 1986.
0 1987 American Chemical Society
The Journal of Physical Chemistry, Vol. 91, No. 14, 1987
Trimethylphosphine on Pt( 11 1)
3809
TPD Product Y i e l d s
P (CH313 / P t ( 1 1 1 ) TPD
2.0
'-
1.5 i2 1
1.0
W .->
/ I
A
llethane Xi0
0
Ethylene Xi00
n a +ffl
0.5
CI
.+ C
3
480 n L a
0.0
I
\
I
I
I
5
10
15
Exposure- / sec Figure 2. TPD product yields vs. P(CH,)3 exposure. Exposure is given in seconds at standard flux (see text). Methane yields are expanded by 10 and ethylene yields are expanded by 100.
d
m C PI .?I
cn
L
aJ
u a, E 0
fL\
L
CI U 0)
520
ffl
I
a
H2
m/e = 2
480 A
ffl Lo
m
z
0
200
400
675
600
800
1000
Temperature / K
Figure 1. TPD spectra for H, ( m / e 2 ) , CH4 ( m / e 16), and CzH4( m / e 27) after (a) 1-, (b) 2-, (c) 3-, (d) 5-, (e) lo-, (f) 1800-s exposures of P(CH3)3at 100 K. The exposures are at the standard flux used throughout (see text). The heating rate was 12 K/s.
meV) and in the specular direction. Spectra were smoothed by least-squares fitting to a third-order polynomial using an 11-point moving window.
3. Results 3.1. TPD. Figure 1 is a set of TPD spectra recorded as a function of increasing exposure to P(CH3),. At the lowest exposure H2is the only desorbing product, but after larger exposures CH,, C2H4, and P(CH3), are also desorbed. Hydrogen desorption from P(CH,),/Pt( 11 1) occurs in three peaks (366, 436, and 521 K) at the lowest exposures (Figure 1, bottom). The relative areas of these three peaks are equal (3~20%) for the lowest two exposures used, but at higher exposures they are not resolved. This result suggests a stepwise demethylation of P(CH,),, with subsequent decomposition of the adsorbed methyl fragments at low exposure. Both the low- and high-temperature peaks areas increase with increasing exposure and a peak at about 675 K builds in at high exposure. The middle peak (436 K) appears to be saturated at even the lowest exposure but is obscured by the low-temperature peak after a 3-s or larger dose. The temperature at the maximum of the high- (520 K) temperature peak is constant but the low-temperature peak moves to higher temperature with increasing exposure indicative of a kinetic order less than unity. Hydrogen desorption from H(a)/clean Pt( 1 11)
follows second-order kinetics ( Tpdecreasing function of coverage) initial with Tp = 395 K at 0.07 M L (1 M L 1.5 X l O I 5 coverage of H(a).Is Since the lowest temperature H2peak position increases with increasing coverage, and the other peaks are above the temperature for desorption limited H2, all hydrogen desorption from P(CH,),/Pt( 11 1) appears to be limited by reactions which deposit H atoms on the surface. The integrated total desorption yields for H2, CH,, and C2H4 as a function of exposure are given in Figure 2. With the standard flux, the H2desorption increases linearly for exposures between 0 and 3 s and is saturated (0.78 ML of H2 = 1.56 ML of H(a)) for any dose larger than 5 s. The hydrogen desorption yields were calibrated as follows. On Pt( 11 1) 0.25 ML of H2S decomposes to yield 0.25 ML of H2(g).19 We compared the yield of hydrogen from H2S with that from a saturation dose of H 2 at 100 K, a dose easily repeated throughout the course of experiments. On this basis a saturation H2 exposure yields 1 ML of H(a). The yield of H2 from P(CH,), was calculated on the basis of this calibration. Tsai and Muetterties reported H2 desorption maxima at 493 and 533 K1' in reasonable agreement with our 520 K peak. The temperature of their first peak, however, is about 100 K higher than ours which may indicate differences in chemistry due to their elevated (-300 K) adsorption. Most of the methane desorbs in a single peak with Tp = 480 K but there is a trace amount desorbing at ca. 325 K Figure 1, middle). No desorption of CH4 is detected at the two smallest exposures (Figures 1 and 2) indicating that formation of C H 4 requires an initial coverage of P(CH3), greater than 0.07 ML (determined from H2TPD area). The desorption yield of methane saturates (0.18 ML) at slightly higher exposures than either H 2 or C2H4 (Figure 2). The constant position and the shape of the 480 K CH4 peak suggests a first-order desorption process. Ethylene desorbs in a single peak centered at 507 K (Figure 1, top). (For the largest exposure (Figure I f ) there is some interference at m / e 27 from multilayer P(CH3), desorption which is evident as a low-temperature tail.) As for CH,, C2H4 desorption also requires a minimum P(CH3), exposure (Figure 2) but saturates by 5-s exposure (0.16 ML of P(CH,),). The temperature of the desorption peak is, like CH4 desorption, independent of initial coverage (the pattern expected for first-order kinetics) but the yield of C2H4is much smaller than CH4 (Figure 2 ) . The amount of ethylene desorbing relative to the saturation methane yield was estimated by assuming that the mass spectrometer transmission is the same for m / e 16 and m / e 27." Published cracking patterns2' and ion gauge sensitivity factors22were used. (18) Zhou, X.-L.; White, J . M. Surf. Sci., in press.
(19) Koestner, R. J.; Salmeron, M.; Kollin, E. B.; Gland, J. L. Surf. Sci. 1986, 172, 668.
3810 The Journal of Physical Chemistry, Vol. 91, No. 14, 1987
Mitchell et al.
tSIMS o f P M e 3 / P t (111) M o n o l a y e r (xi01 (heated t o 1 4 0 ~ ) PMezCH2+
c 01 ,
c
u
1.0
0.0
1 1'
0
'
A
Plpt
0
C/Pt x3
I
,
I
5
10
15
Ir
PMe3+
)eeM;'
1
1 J 20
Exposure / sec Figure 3. P(120 eV)/Pt(238 eV) and C(272 eV)/Pt(238 eV) AES intensity ratios as a function of exposure. AES was performed after heating t o >900 K.
At saturation the yield of C2H4 is only 2% of the CH, yield. The TPD data were analyzed to determine kinetic parameters for CH4 and C2H4 formation from P(CH3), decomposition. Pumping speeds for methane and ethylene in our vacuum chamber were not measured accurately but an upper limit of 0.25 s was determined for the pumping time constant ( 7 ) for methane, and the pumping speed for ethylene is expected to be similar. The product r p ( p = heating ramp rate) is therefore
U
.d
u)
C 01 U
z1
c
L
H
m r
'0
0 U
01
v1
__350 0
150
300 450 Temperature / K
600
750
Figure 5. TPSIMS of CH3+ ( m / e 15), P(CH3)*+( m / e 61), and HP(CH,),+( m / e 77) for multilayer P(CH3), adsorbed at 100 K.
the sputtering event. Ions at m / e 13, 14, and 15 (CH+, CH2+, and CH3+) are detected from the monolayer but are less intense in the multilayer spectrum. Information on the thermal evolution of the surface adlayer can be deduced by following the SIMS signals while heating the sample. Several secondary ions were followed in a TPSIMS experiment (Figure 5) and some important trends can be delineated: (1) m / e 77 (HPMe,') decreases precipitously at about 150 K as do other ions ( m / e 61,62, and 76 which are not shown). This is consistent with a higher yield of these ions from multilayer P(CH,), compared to chemisorbed P(CH3),. (2) Between about 160 K and roughly 350 K the HPMe,' ( m / e 77) yield is approximately constant; it begins decreasing near 350 K and levels off again near 480 K. The m / e 61 (PMe2+) ion behaves like m / e 77 except the temperatures at which the yield begins decreasing and then levels off are higher (385 and 510 K, respectively). (3) The CH3+( m / e 15) intensity increases slightly as the multilayer desorbs, stays approximately constant until about 390 K, increases in intensity until it maximizes near 480 K, then decreases to about zero near 530 K . We associate the CH3+ ion signal with the formation of CH4 and attribute it to methyl groups attached directly to the surface. The onset (390 K) and the maximum in this signal (480 K) coincide with the onset and maximum in the CH4 TPD, respectively, and both the TPD and SIMS signals decrease to zero at about the same temperature. The decrease in the HP(CHJ,+ signal can be associated with dissociation of P(CH3),. The onset of hydrogen desorption coincides approximately with the downward deflection of the HP(CH,),+ signal (350 K). The gas-phase electron beam excited mass spectrum of P(CH& gives strong cracking at m / e 61 (P(CH,),'). Thus we anticipate a relatively strong m / e 61 signal in SIMS when molecular P(CH3), is present. This is consistent with the spectra in Figure 4. P(CH,),(a) should also have a sizeable SIMS signal a t m / e 61. Our interpretation of the behavior of m / e 77 and 61 is that the yield of P(CH&+ stays relatively constant as P(CH,),(a) decomposes to P(CH,)2(a) (=350 K) and begins to decrease strongly only when P(CH,), begins decomposing ( E 385 K).
-500
500
1500 2500 Energy Loss /
4500
3500
Figure 6. HREEL spectra for (a) 1-, (b) 5-, (c) IO-, (d) 15-, and (e) 1800-s exposures of P(CH,), at t h e standard flux and at 100 K. TABLE I: Vibrational Assignments for P(CH,),/Pt(III) and Comparison to Assignments for a Pt" Complex and P(CH3),
P(CH,),/Pt(W transmultilayer monolayer PtCI,[P(CH,)3]," P(CH,)t 2940 2890 1410 1310
1410 1290
940
955
710 685 380 -210
2980 vwsh 2960 m 2900 m 1430 sh 1416 m 1300 vw 1285 s 950 sbr 864 sh 856 m 146 ms 670 s 340 373 284 222
2968 a , 2955 e' 2894 a l 1430 e 1417 a l 1312 a l 1293 e 960 a l 947 e 708 e 653 a , 263 e, rc 305 a,,r
QReference27. bReference 26. c r = Raman of liquid phase. 3.4. HREELS. Vibrational spectra for various exposures of P(CH,),/Pt( 11 1) are plotted in Figure 6. The results are consistent with associative adsorption at all exposures at 100 K and multilayer coverage at high exposures. The observed losses for the chemisorbed layer (Figure 6b) and the multilayer (Figure 6e) are listed in Table I. The bands were assigned by comparison to the vibrational frequencies for P(CH3)32s.26 and for transPtC12[P(CH3)3]2,27 also listed in Table I. The presence of the Pt-P stretching mode (u(PtP) = 350-380 cm-') indicates chemisorbed P(CH,), and the absence of thick multilayers of P(CH3),. At lower exposures of P(CH3), (Figure 6, a and b), this band is present but it decreases for a 10-s exposure ( 2 5 ) Halmann, M. Spectrochim. Acta 1960, 16, 407. (26) Park, J. D.; Hendra, P. J. Spectrochim. Acta, Part A 1968, 24A, 208 1.
The Journal of Physical Chemistry, Vol, 91, No. 14, 1987
3812
Mitchell et al.
P(CH313 / P t (111)
BOOK
ul U
lJl
.d
U
.,
c
3
f 3
n L a
n L a
x300
\
570K
x io0
\
*
z
U
CI
.rl
.,
Lo
Ln
C a,
C a,
U
U
H C
H C
400K x300
-500
500
1500 2500 E n e r g y Loss / cm-l
3500
4500
Figure 7. HREEL spectra for a large multilayer of P(CH3)3adsorbed at 100 K and heated to various temperatures. Spectra were taken after cooling to 100 K.
and is completely obscured (or screened) for the 1800-s dose. The v(PtP) mode is at 350 cm-' for the lowest exposures of P(CH3)3 used (1 s at the standard flux) (Figure 6a) and increases to 380 cm-l with higher coverage. For the 1-s dose v(PtP) is the only loss with significant intensity. Its presence is also diagnostic of undissociated P(CH3), as it is absent for phosphorus adsorbed on Pt( 11 1) after decomposition of either P(CH3)3(see below) or PH3.4 Though the relative intensities vary, there is little change in the positions of most of the intramolecular P(CH3), vibrations as a function of coverage. One exception is the band close to 700 cm-I (P-C stretching mode) whose frequency increases by about 30 cm-I in the multilayer spectra (Figure 6e) compared to the monolayer (Figure 6, a and b). We assign this band to the asymmetric P-C stretching mode in the multilayer and to the symmetric stretch for the chemisorbed layer. If both of these bands were present simultaneously, the HREELS (resolution ~ 8 0 cm-I) would not be able to resolve them (see Table I). In organometallic complexes, the P-C stretching modes increase 20-40 cm-' when coordinated to a metal compared to the free and a similar increase in frequency should occur for chemisorbed P(CH3)3compared to the multilayer. Assigning both the 710-cm-I loss for the multilayer and the 685-cm-l loss for small coverages to the same mode, va or us, would give the wrong shift upon coordination. We have thus assigned the 710-cm-' band for the multilayer to the asymmetric P-C stretching mode, and the 685-cm-' band found for low coverages to the symmetric P-C stretching mode. By analogy to the bonding of P(CH3), in coordination compound^,*^-*^ adsorbed P(CH,), is most likely bonded through the P atom, at an on top site, with C,, skeletal symmetry. The behavior of the P-C stretching modes are consistent with C30 symmetry for the chemisorbed layer since, in this geometry, dipole scattering from the P-C asymmetric stretch will be forbidden. (27) Park, J. D.; Hendra, P. J. Spectrochim. Acta, Part A 1969, ZSA, 909. (28) Park, J . D.; Hendra, P. J. Spectrochim. Acta Part A 1969, 25A, 227. (29) Goodfellow, R. J.; Evans, J. G.;Goggin, P. J.; Duddell, D. A. J . Chem. SOC.A 1968, 1604.
-500
500
1500
2500
3500
4500
Energy L o s s / cm-l Figure 8. HREEL spectra for a large multilayer of P(CH3)3adsorbed at 100 K and heated to various temperatures. Spectra were taken after cooling to 100 K.
The spectrum for multilayer P(CH,), (1800-s dose, Figure 6e) compares very well with the infrared and Raman spectra of gas and liquid phase P(CH,)3.25,26Two losses detected in the multilayer spectra which are not included in Table I are the shoulders near 1630 and 3150 cm-I. We assign these to combination bands (u(PC) + p(CH3)) and (u(CH,) + b(PC)), respectively. Figures 7 and 8 are HREEL spectra for multilayer P(CH,), adsorbed at 100 K and heated to successively higher temperatures. The annealing temperatures were chosen to coincide with features in the TPSIMS and TPD spectra. The 140 K anneal desorbs multilayer P(CH3),, 300 K coincides with the onset of hydrogen desorption, 400 K eliminates the low-temperature hydrogen peak, 450 K coincides with the onset of the 520 K hydrogen desorption peak, and 570 K leaves only the broad high-temperature hydrogen peak to be desorbed. That heating to 140 K was sufficient to desorb the multilayer P(CH,), is confirmed by the appearance of u(PtP) at 380 cm-' (Figure 7). This spectrum is very similar to that for a 5-s dose at 100 K (Figure 6b). There is a large increase in the specular elastic peak intensity (16 times more intense at 140 K compared to 100 K). We take this as evidence for increased surface order after desorption of the multilayer. Consistent with TPD and TPSIM spectra, there is little change in HREELS for heating between 140 and 300 K. For the 400 and 450 K anneals, the HREEL spectra exhibit all the losses detected in the 300 K spectrum with minor changes in frequency, but with large changes in intensities. In particular, the relative intensities of v(PtP) and p(CH3) both decrease in relation to u(PC) and the CH, stretching and deformation modes (Figure 8). TPD and TPSIMS suggest that heating to 400 K should produce P(CH3)2(a)and heating to 450 K should produce a significant coverage of PCH,(a). It is difficult, however, to discriminate losses due to either of the latter species in the presence of each other and in the presence of P(CH3),. This can be seen by comparison of the vibrational spectra of methyl-, dimethyl-, and trimethylphosphine for those modes which involve motions of the methyl groups (Table 11). On the basis of the data in Table 11, and the resolution of the HREELS, the HREEL spectra of
The Journal of Physical Chemistry, Vol. 91, No. 14, 1987 3813
Trimethylphosphine on P t ( l l 1 ) TABLE II: Infrared or Raman Active Methyl Group Vibrational Frequencies for Trimethylphosphine, Dimethylphosphine, and Methylphosphine
mode description u(CH,)
6,(CH,) 6,(CH3) P(CHJ
v,(PCJ U,(PC,)
WC)
P(CH,),(g)' 2968 a , 2955 e 2894 a l 1430 e 1417 a , I312 a , 1293 e 960 a, 947 e
708 e 653 a, 263 e, rd 305 a,, r
HP(CH,),(S)~ 2966 al, b,, a2, b2 2907 a,, b, 2902 a2, b2 1421 bl 1411 a , 1290 a, 1272 b2 987 bl 948 a , 824 bl 726 b2 707 bl 663 a, 271 a,, bl
H,PCH,(s)' 2981 e 2918 a, 1422 e 1276 a,
lolo\ 983
e
676 a,
"Reference 26. bReference 33. 'References 25 and 34. d r = Raman. The symmetry types are indicated for P(CH3), in C,,, HP(CH3)2 in C,, and H2PCH3in the C,, point group. P(CH3)3(a) and P(CH3),(a) should be very similar in both the number and the frequencies of losses. HREELS of PCH,(a) should differ from P(CH3)3(a) and P(CH3)2(a); if not in the number of bands at least in their relative intensities. In the limit of C3, symmetry of PCH,(a), there would only be four dipoleallowed losses for this ligand. At lower symmetry (such as caused by a tilting of the P-C bond) other modes would become dipole allowed but their intensities would be determined partly by the degree of perturbation from the C3, geometry. The frequencies for most variations of PCH3(a), as for P(CH3)2(a), will fall in the ranges found for vibrations of P(CH3)3(a). The decreased intensity of the loss near 370 cm-I after 400 K or higher anneals (Figure 8) indicates the decomposition of P(CH3)3(a)and the continued presence of the -700 cm-I loss as well as those due to vibrations of methyl groups are evidence for P(CH3)2(a)or PCH3(a). By analogy to dimethylphosphine coordination c ~ m p o u n d s , P(CH3),(a) ~~~' would bridgebond between two adjacent Pt atoms. PCH3(a) could bond in a double or triple bridge site. Since, in general, metal adsorbate stretching frequencies decrease with increased coordination number,'2 the Pt-P stretches of the demethylated species P(CH3)2(a), PCH3(a), and P(a) probably occur at lower frequency than that for P(CH3)3(a) and are unresolved from the tail of the elastic peak. When the sample was heated to 800 K the broad tail on the elastic peak is the only sign of significant inelastic loss intensity. At this temperature all the hydrogen has desorbed leaving only phosphorus and carbon. Decomposition of P(CH3)3most likely deposits CH3 fragments on the surface, but for kinetic reasons we expect the coverage to be low. Some of the CH3(a) will react to form CH,(g) and some will decompose to CH2(a) and CH(a). There may be small coverages of CH3(a), CH2(a), or CH(a) present on the surface for the 400-450 K HREELS (Figure 8), but no bands unambiguously assignable to these species are found. 1. Discussion
At loo HREELS are consistent with predominantly associative adsorption of trimethylphosphine at all exposures. Although we were unable to reliably measure P(CH3)3desorption in TPD' there is ample evidence 'IMS and HREELS) which indicates P(CH3.)3 adsorbs at loo with approximately the Same sticking coefficient for coverages below and above a single layer* Sublimation Of 'lid P(CH3)3 OCCUrS between -'40-155 K, leaving Only P(CH3)3' TPSIMS and HREEL (30) Carty, A. J. In Caralytic Aspects of Meral Phosphine Complexes, Alyea, E. C., Meek, D.W. Eds.; American Chemical Society: Washington, DC 1969; Adv. Chem. Ser. No. 196, p 163. (31) Hayter, R. G. Inorg. Chem. 1963, 2, 1031. (32) Steininger, H.; Lehwald, S.;Ibach, S.Surf. Sci. 1982, 123, 264.
spectra suggest that the chemisorbed P(CH3), is stable up to about 350 K. Decomposition yielding HZ,CH4, P(a), C(a), and C2H4 occurs at higher tempertures. The maximum amount of P(CH3), that decomposes is 0.25 ML. I t produces 0.78 M L of H2, 0.18 ML of CH4, 0.58 ML of adsorbed carbon, 0.25 ML of adsorbed phosphorus (some of which diffuses into the bulk), and 0.003 ML of C2H4. The HREELS evidence is consistent with C3, symmetry for the chemisorbed P(CH3)3. However, the frequencies of most vibrations which would be dipole forbidden for C3, symmetry (especially e type vibrations of the methyl groups) lie close to allowed modes and would not be resolved if they were present. The best evidence for C3, symmetry is provided by the dominance of the symmetric P-C stretching mode over the asymmetric P-C mode for the chemisorbed layer. The changes in the HREELS after the 400 and 450 K anneals are consistent with the decomposition of P(CH3)3and the presence of the demethylated species. The best evidence for the formation of P(CH3)2(a) or PCH3(a) is the presence of a P-C stretching mode near 700 cm-l and the decreased intensity of the Pt-P stretching mode (found near 380 cm-I for P(CH3)3(a)). A mechanism for decomposition of P(CH3)3and formation of products which is consistent with our observations is given by reactions 1-9, where unless otherwise labeled, the species are indicates an equilibrium. adsorbed. The double headed arrow ( s ) P(CH3)3
-
P(CH3)2 + CH3
(1)
+ CH3
(2)
P(CH3)2 --* P(CH3)
+ CH3 +H
(4)
CH2 + C H
+H
(5)
CH
+H
(6)
CH4(g)
(7)
P(CH3) CH3
9
CH2 C(a)
CH3 + H 2CHz 2H
P
-+
9
--
-+
(3)
C&,(g)
(8)
H2(g)
(9)
The equivalent amounts of H2which desorb in the three peaks at low initial coverage suggest that reactions 1-3 occur in stepwise fashion under these conditions. Support for stepwise demethylation is provided by the behavior of the PMe2+and HPMe3+ TPSIMS signals but at the higher coverages used for the TPSIMS, the steps are not so clearly separated. We infer that the activation barrier for demethylation becomes progressively higher in each step, which is consistent with simple notions regarding these processes. For instance, if we assume (1) both the phosphorus atom and the carbon atom must interact with the surface to allow P-C bond cleavage, and (2) the equilibrium Pt-P-C bond angle increases with decreasing x for P(CH3),(a), then the difficulty of P-C bond cleavage should increase after successive demethylations. According to the scheme presented, there are three parameters which could limit the rate of methane formation. The rate constant for reaction 7 is one of these, but near 480 K this rate constant must be large based on the ease of forming methane (270 K) from CH31/Pt( 111).16 This leaves the availability of H(a) or CH3(a) as potentially rate limiting. In our mechanism these are in fact strongly coupled by reaction 4. From results for ketene, formaldehyde, and methyl iodide,l>16 reactions 4 and 5 should be facile at which methane is formed from pat the (CH3),/Pt(11 l ) , but the rapid desorption of hydrogen (reaction 9) at these temperatures will tend to keep H(a) low, The correspondence in the peak temperature for methane TPD (Figure 1) and TPSIMS of CH3+ (Figure 5) is consistent with the coverage of CH3(a) limiting the rate of methane formation. If we interpret the increase in the CHI+ SIMS vield at 390 K as resulting from adsorbed methyl fragments, thdn, though we know no Gay of quantifying the CH3+signal in terms of CH3(a) coverage, the that p-c bond latter is clearly not vanishingly small. This breaking (reactions 1-3) cannot be solely responsible for the kinetic parameters calculated from the TPD spectra (section 3.1). The
J . Phys. Chem. 1987, 91, 3814-3820
3814
preexponential factor determined from the CH4 TPD spectra (7 X lo9 S-I) is small compared to that expected for a unimolecular reaction ( l o 3 s-l) and is consistent with the rate of methane formation being determined by competition among the various pathways and being characterized by parameters for more than one elementary step. The kinetic parameters calculated in section 3.1 should not be taken as representative of a single elementary reaction step but are dependent on the overall mechanism. The rate of reaction 8 is limited by the coverage of CH2(a) which is expected to peak at a slightly higher temperature than CH3(a) (by eq 4). In agreement with this, C2H4 desorption peaks a t 507 K, 27 K higher than Tpfor CHI. Our reaction scheme shows competition for H(a) between reactions 7 and 9. This competition could explain the dip in H2 desorption which occurs concomitant to CH4 desorption. When the coverage of adsorbed methyl becomes high enough or the reverse of reaction 4 is fast enough, then H(a) reacts preferentially to form CHI. Although not explicitly included in the above scheme, available sites for decomposition must also be an important parameter in many of the individual steps. In particular, reactions 1-6 might require uncovered metal sites to form new Pt-C or Pt-H bonds. The increase in Tp,with increasing initial P(CH3), coverages, for the nominally 400 K H2 TPD peak (Figure 1, bottom) may be attributable to decreasing availability of decomposition sites. At very low initial exposures, no methane is formed by TPD. Under these conditions the equilibria ((4), (5), and perhaps (6)) favor decomposition, effectively inhibiting the formation of CHI and C2H4. Alternative mechanisms for methane formation would involve direct transfer of H atoms to C H 3 units from an adjacent P(CH,),(a) fragment or even simultaneous loss of both C H 3 and H from the same P(CH,),. The existence of coordination complexes such as (Ph,P)2PtP(mesityl)=CPh235,36 would suggest the (33) McKean, D. C. McQuillan, G. P. J . Mol. Struct. 1980, 63, 173. (34) Lannon, J. A,; Nixon, E. R . Spectrochim. Acta, Parr A 1967, 23A, 2713.
feasibility of these alternative mechanisms, since fragments of the latter type (i.e. containing P = CH2 units) would be formed as intermediates. The coadsorption of deuterium atoms with P(CH3), is of little help in determining the source of the hydrogen atoms as D(a) recombination desorption occurs below the methane formation temperature in TPD. The relatively large yield of CH3D detected by Tsai et al. (and also by the present authors) in D(a) + P(CH,), coadsorption experiments is probably due to isotrope exchange with H atoms in P(CH,),(a) during TPD. Although we cannot rule out the direct H atom transfer mechanism, we believe the mechanism presented in steps 1 through 9 adequately accounts for our experimental observations.
5. Summary Trimethylphosphine adsorption at 100 K is associative, with a sticking coefficient which is large and independent of coverage onto the clean or P(CH3), covered surface. Solid P(CH3), sublimes between 140 and 160 K. Chemisorbed P(CH3)3 is bonded to the platinum surface through the phosphorus atom probably in C,, symmetry. P(CH3), decomposes by demethylation to produce P(CH,)2(a) and PCH3(a) as intermediates. A maximum of 0.25 ML of P(CH3), dissociates upon heating to yield 0.78 ML of H,(g), 0.58 ML of C(a), 0.25 ML of adsorbed phosphorus, 0.18 M L of CH,(g), and 0.003 M L of C,H4(g). A reaction mechanism, consistent with our results and which explains interesting aspects of the decomposition of P(CH,), and formation of products, is presented. Acknowledgment. This work was supported in part by the U S . Army Research Office. (35) Van der Knapp, T. A.; Bickelhaupt, F.; Kraaykamp, J. G.; Van Koten, G.; Bernards, J. P. C.; Edzes, H. T.; Veerman, W. S.; deBoer, E.; Baerends, E. J. Organometallics 1984, 3, 1804. (36) Cowley, A. H.; Norman, N. C.; Quashie, S. J . Am. Chem. SOC.1984, 106, 5007.
Microstructure from X-ray Scattering: The Disordered Open Connected Model of Microemulslons Thomas N. Zemb,t* S. T. Hyde,t Paul-Joel Derian,t Ian S. Barnes,t and Barry W. Ninham*+ Department of Applied Mathematics, Research School of Physical Sciences, Australian National University, Canberra, ACT 2601, Australia, and Department de Physico-Chimie, DESICP, CEN Saclay, 91 191 Gif-sur-Yvette, France (Received: December 10, 1986; In Final Form: February 18, 1987)
We describe the first structural model of microemulsions consistent with the peak position in small-angle X-ray scattering (SAXS) and conductivity variation observed in three-component microemulsions. The analysis is carried out for a set of disordered connected cylinders. A general method for analyzing scattering from random media is developed from the variation of the product ZD* (surfactant surface area and characteristic distance) with component volume fractions.
Introduction
Interpretation of the features of the scattered intensity of concentrated microemulsions presents so serious a problem that the utility of scattering as a probe of microstructure has been in doubt. This is because the scattered intensity reflects only the pair-correlation function of the scattering structure, and it is well-known that very different structures in space can be described by identical pair-correlation functions.' 'Australian National University. DESICP.
*
0022-3654/87/2091-3814$01.50/0
Hence, in the absence of additional physical information which allows discrimination between a multiplicity of possible models, the assignment of microstructure from scattering data remains ambiguous or even arbitrary. For instance, any scattering intensity measurement can be explained in terms of a microstructure made up from either interacting monodisperse ellipsoids or polydisperse spheres, provided one admits a free choice of the mass distribution function and the interaction potential. Although in the case of monodisperse interacting spheres a general solution to the problem (1) Welberry, T. R. Rep. f r o g . Phys. 1985, 48, 1543-1 593
0 1987 American Chemical Society
