J. Phys. Chem. 1991, 95, 1755-1759
a small photoeffect because an overlap is only given with the conduction band. The much higher current under illumination of O2compared to MV2+ and C02can again be explained by the nature of the surface states. Dioxygep is assumed to cause the surface states, and reduced oxygen species are found on the surface.I4 No charge transfer seems to be necessary from the ZnPc surface to solvated species in solution, but reduced oxygen species from the film surface could diffuse into the electrolyte and are replaced by dioxygen molecules from solution. The surface states are populated by electrons only under illumination so that the described mechanism only works under illumination of the surface even though O2 is adsorbed also in the dark.
1755
To conclude, it has been shown that zinc phthalocyanine, especially when embedded in poly(viny1idene fluoride), leads to an efficient photoreduction of dioxygen. A high ratio of the current under illumination to that in the dark is obtained in a short response time. A high sensitivity toward dioxygen and visible light is observed. The electrodes exhibit excellent mechanical and electrochemical stability. Acknowledgment. We gratefully acknowledge financial support for D.S. by the Deutscher Akademischer Austauschdienst (DAAD) for a research visit to Japan. We are thankful to R. Memming (Hannover, FRG) for intensive and fruitful discussions.
Temperature Rises Produced by a Molecular Beam Striking a Platinum Surface. 2 T. Taot and E. F. Greene* Department of Chemistry, Brown University, Providence, Rhode Island 0291 2 (Received: August 10, 1990)
Measurements of the temperature rises produced when molecular beams strike Pt surfaces are interpreted to give information on the probability of the transfer of energy as a result of collisions. Previous work introducing the method is supplemented by showing the effect of roughening the surface by the deposition of Pt black, of varying the temperature of the nozzle from which the beam emerges, and of adding internal energy to the molecules of the beam by irradiating them with an infrared laser. Experiments with binary gas mixtures provide a method for determining relative accommodation coefficients for the two components without any need for external calibration.
Introduction In an earlier paper,' here denoted I, we reported measurements of temperature rises AT produced by molecular beams issuing from a nozzle a t room temperature and then striking a Pt surface initially also a t room temperature. Interpretation of the results gave values for a coefficient y (closely related to CY,the classical energy accommodation coefficient) that measures the fraction of the scattered molecules whose energy is determined by the temperature of the surface. Because this method of learning about energy transfer between gases and solids differs in several respects from other experiments previously reported by others (see the review by Goodman and Wachman2 and other references in I), we describe here further experiments that characterize it more fully. In particular we show the effect on the AT when ( I ) the nozzle source of the beam is heated above room temperature, (2) the polycrystalline Pt surface struck by the beam is roughened by electrochemical deposition of Pt black, and (3) some of the molecules in a beam of SF6are excited by the absorption of infrared radiation from a laser. Apparatus and Procedure In most respects the'apparatus and procedure are the same as those described in I. Briefly stated, a manipulator in a highvacuum system permits translation of a small Pt surface in three Cartesian directions relative to the exit hole of the nozzle from which the molecular beam emerges. After the beam is turned on, Tj, the temperature of the Pt (measured by thermocouple TCl), rises and in about 3 min becomes steady. This rise, AT = T j Tb,where Tb is the temperature of the background (measured on thermocouple TC3), increases as n, the beam flux reaching the Pt surface, is increased, e.g., by increasing P,, the pressure in the nozzle, or by decreasing x, the distance from the nozzle along the axis of the beam. Asymptotic values AT, = lim (n m) AT obtained by short linear extrapolation of A T 1 versus either x 2 or P i ' gives the value of the rise, AT, or ATp, respectively, when losses due to radiation and heat conduction from the Pt are negligible. The main differences from I are that there is a heater
-
'Present address: Teknor Apex, Attleboro, MA 02703.
0022-3654191 12095- 1755$02.50/0
for the nozzle so that its temperature T,, need no longer be kept at Tb,the ambient one, a second Pt surface coated with Pt black is mounted 3.5 mm nearer the nozzle and offset from the polycrystalline one so that AT can be recorded by using either surface during a given run, and there is provision for using other nozzles. One of these permits vibrational excitation of the molecules of SF6by their absorption of radiation from a CO, laser, while the other is simply a round hole in a thin Ni foil. The Pyrex nozzle from I is heated by closely wound turns of resistance wire, and T, is measured on thermocouple TC2 (chromel-alumel) that is inserted into the nozzle from the rear so the molecules that are to form the beam flow over the junction just before they emerge into the vacuum chamber. The tip of the nozzle passes through a hole in a radiation shield that reduces the heating of TCI by radiation from heated nozzles. The Pt black surface is prepared by first making TCI as in I by spot welding together two (2.0 X 2.2 mm2) polycrystalline Pt foils 25-pm thick to enclose the junction between one alumel and two chrome1 wires that support TCl in its frame. Next, additional Pt is deposited electrochemically from a solution of 0.10 g of PtClz (Alfa Products) in 2 mL of concentrated HCI (12 M) that is diluted to 10 mL with deionized water. The normal rapid dissolution of the fine alumel wire (25 pm) in the acid can be slowed sufficiently by applying the potential before placing T C l in the solution and keeping the time of electrolysis short. In 3 min a current density of 1.3 A cm-, at 3.6 V gives good blackening over about 90% of the surface. (Better coverage would be desirable but is difficult to achieve.) The blackened Pt is then immediately rinsed in deionized water and dried in air. The coating typically increases the mass of T C l from 0.060 to 0.080 g. To produce an increase in the vibrational energy of the molecules of an SF6beam, we use a C 0 2laser (wavelength 10.6 pm) (1) Greene, E. F.; Tao, T.; Thantu, N. J . Phys. Chem. 1989,93, 6778. Note: In 1 on p 6782,column 2, y i should be X i in the expression for Emnd and a square bracket, [, is missing before c in the equation defining the rate coefficient c'; also there are four errors in eq 15 that are corrected in eq 2 of this paper. (2)Goodman, F. 0.; Wachman, H. Y.Dynamics of Gas-Surface Scattering; Academic: New York, 1976;Chapter 10.
0 1991 American Chemical Society
1156 The Journal of Physical Chemistry, Vol. 95, No. 4, 1991 40
Tao and Greene
m
-
30 . h
Y
v
a
a
I-.
r; Y
20 -
v l-
a
10.
b
Figure 2. Electron micrographs of Pt (a) and Pt black (b) surfaces (width of each picture 29 pm).
0
100
300
200
400
500
600
x2 (mm2>
Figure 1. Effect of heating nozzle: (a) Temperature increment on thermocouple TCI (Pt surface) vs distance x from nozzle with ( A T , upper three curves) and without ( A T r ,lower three curves) an Ar beam; the lines serve merely to connect the points; (b) reciprocal of temperature increment AT due to A t gas alone vs square of distance x; the lines are linear fits to t\e points. PAr= 40 kPa; T,, 300 K (O), 356 K (0),and 397 K (0). Insert in (b): lim (x 0) AT, vs nozzle temperature T,,; the lines are linear fits to the points.
-
that enters the vacuum chamber through a ZnSe window and intersects the molecular beam at a right angle to its axis. To allow the laser beam to excite SF, molecules either before they leave the tip of the nozzle or downstream where the flow expands into the vacuum, the nozzle is fitted with an infrared-transmitting window of ZnSe. The assembly, which is similar to the one described by Lester et al.,3 consists of a V-shaped groove (approximately 0.25 mm wide by 0.1 mm deep) in a stainless steel block covered by the plane ZnSe window. The groove and window form a capillary 3.8 mm long with a triangular cross section through which the gas flows to form the beam. The third nozzle consists simply of a hole of 0.05" diameter in a 0.025-mm-thick Ni foil (an electron microscope aperture, Ealing Co.) hard soldered to a 6-mm stainless steel tube. Results Heated Nozzle. As the Pyrex nozzle is heated above room temperature, the flow through it, in moles per unit time at constant P,, decreases approximately as TO.'.It is linear in P, at constant T, within the uncertainty of our measurements (although the extrapolated line intersects the P, axis not at zero but near 50 kPa). The temperature rises seen by TC1 increase with T, at constant P, as Figure l a shows for an Ar beam. The radiation shield, through which only the tip of the nozzle projects, reduces AT,, the temperature rise on TCl due only to (3) Lester, M. 1.; Casson, L. M.; Spector, G. B.; Flynn, G. W.; Bemstein,
R. 9. J . C h ~ mPhys. . 1984.80,
1490.
radiation from the heated nozzle (molecular beam off), to less than 10%of the total rise. Nevertheless, it is not negligible. The quantity of interest is AT, = AT - AT,, the rise due to the molecular impacts on TCl . The reciprocal plots in Figure l b (see eq 11 of I) permit extrapolation of AT, to x = 0 to yield values of AT,, the temperature rise expected when losses due to heat conduction and radiation from TCl are negligible compared to losses to molecules leaving the surface. The insert in Figure 1b gives the variation of AT, with T, and permits an estimate that ATx would be zero for T, = 246 f 10 K. Pt Black. As expected, the Pt black surface is a more efficient radiator than the polycrystalline Pt one is. We measure its rate of energy loss due to both conduction and radiation by warming it with a beam from a small He-Ne laser and observing its rate of cooling after the beam is turned off. A plot of In (q - Tb) versus time is linear with a Slbpe (decay constant) c' equal to 0.031 1 f 0.0003 s-l, a value much larger than 0.0143 f 0.0002 s-I, that for the polycrystalline Pt surface.' One may expect that the rougher a surface, the more collisions incoming molecules are likely to make with it, and thus the more closely they should come to achieving equilibrium with it before they return to the gas phase. The electron micrographs reproduced in Figure 2 permit comparison of the Pt black and Pt surfaces. The former has an irregular, rough surface with features having a linear dimension of 1 pm or less while the latter is relatively smooth. Figure 3 shows reciprocal plots, A T 1 vs x2, for beams of Ar and He and 10% by volume of Ar in He incident on the Pt black surface. The intercepts give AT, = 45 f 4,67 f 6, and 11 1 f 11 K, respectively, values that can be compared with the values (see I) on Pt, AT, = 50 f 4, 62 f 5, and 180 f 18 K, respectively. As is expected, because AT, depends on the net energy transferred per molecule accommodating on TCl rather than on the number of molecules, the values for the pure gases do not differ significantly on the two surfaces, but the mixture gives an appreciably lower value on the Pt black. Laser Excitation of SF,. The ability to increase the internal energy of a molecule without producing much change in its translational energy makes possible study of the efficiency of energy transfer from the internal degrees of freedom to the surface. The C02laser (Adkin, Model MIRL-50) we use is CW with a power of lo2W cmW2,i.e., about 10 times lower than that Lester et aL3 used to heat a beam of SF,. The laser is tuned to give a
The Journal of Physical Chemistry, Vol. 95, No. 4, 1991 1757
Temperature Rises of Pt Surface in Molecular Beam
TABLE I: Temperature Rises (K) Caused by Laser and SF, Beams on Pt and pt Black Surfacesa Pt ( x = 16.0 mm) Pt black (x run AT, ATOL A TBL 6T AT, ATOL 1
4.95 4.83 5.33 5.0
2 3 av
1.10 0.94 2.43 1.5
9.55 8.46 12.11 10.3
3.50 2.69 5.01 3.8 f 1
4.35 4.30 4.19 4.5
2.02 2.02 4.42, 2.8
= 12.5 mm) A TBL
ST
10.50 9.08 15.09 11.6
4.13 2.16 5.88 4.3 1
a Nozzle: pressure 40 kPa, temperature 300 K. Laser spot centered on exit plane of nozzle. x distance to nozzle exit; subscripts i, j : i = B, 0 SF, beam on, off; j = L, 0 laser beam on, off; 6T ATBL - ATBo - AToL.
A
/
‘h
c
a
v
0.2
nn 0
100
200
300
400
500
x2 (mm2)
Figure 3. Reciprocal of temperature increment ATvs square of distance x from nozzle for beams of Ar (0),He ( 0 ) ,and 10%Ar in He (0) striking Pt black surface: P, = 40 kPa, T, = 298 K.
single bright spot (1 -mm diameter) from the lowest order (TEMm) mode. The frequency is not stabilized, but the intensity does not drift significantly during the few minutes needed for a run. Moving an external mirror permits placement of the laser spot up- or downstream so it intersects the molecular beam either inside or outside the transparent capillary. As may be expected, a nonabsorbing gas such as He produces the same temperature rise at the detector whether the laser is on or not, although the laser does heat the nozzle if the beam touches it. This heating raises the temperature of the nozzle no more than 3 O C in the 20 min required for a run. For example, when the laser spot is half on the exit of the capillary and half outside in the vacuum and the Pt surface is 12.5 mm from the nozzle tip, we find AT equal to 4.0,6.2, and 2.2 K (each f0.05 K) when the He and laser beams are on-off (BO), on-on (BL), and off-on (OL) respectively. We note that for the on-on case the 6.2 K rise includes contributions from the He striking the surface of TCl, from the laser light scattered to T C l , and from any possible increase in the kinetic energy of the He atoms because of heating of the nozzle by the laser, while for the off-on case the 2.2 K rise includes only the contribution from the scattered laser light. Thus, we conclude that in these experiments with nonabsorbing gases the contributions to the temperature rise seen by T C l caused by scattered light from the laser and from impacts of molecules in the beam are independent and that the heating of the nozzle by the laser has a negligible effect. Extensive study of the complex spectrum of SF6 has led to assignment of its transitions in its u3 mode that are in and near resonance with lines of C 0 2 in its P branch near 10.6 pme4 Because the P( 16) line of C 0 2 has the biggest absorption coefficient at room t e m p e r a t ~ r e it, ~is the one used here in exciting the SF6 molecules. Table 1 gives the temperature rises measured with both the Pt (x = 16.0 mm) and Pt black ( x = 12.5 mm) surfaces when the laser spot is centered on the very end of the capillary. On both surfaces there is a clear temperature increase 6T = ATBL- AT, - AToL beyond that coming from the heating (4)McDowell, R.S.;Galbraith, H. W.; Cantrell, C. D.; Nereson, N. G.; Hinkley, E. D. J . Mol. Spectrosc. 1977, 68, 288.
of the nozzle by the radiation (ATOL) and by the unirradiated gas molecules (ATBo). Although the direct heating of the nozzle by the laser beam has a negligible effect on AT as shown with the He beam, with an SF6 beam the internal energy increase resulting from the absorption of the laser radiation may be partly transferred to translational energy by collisions in the nozzle. Lester et al,,3 who focused their 1-mm spot 1 mm from the exit of the nozzle, estimated that in their experiment about 75% of the energy absorbed remained as internal energy (after ca. 2000 collisions/SF6 molecule) when the beam arrived at the detector. Because in our experiment the laser is focused somewhat nearer to the end of the nozzle, there should be fewer collisions to provide internal to translational energy transfer before collisions become infrequent as the gas expands on leaving the nozzle. We assume that more than 75% of the energy absorbed from the radiation remains as internal energy when the beam reaches TC1. No detectable temperature rise is found when the laser spot is moved downstream just enough to miss the nozzle and intersect the gas beam only after it is expanding into the vacuum. This is understandable because outside the nozzle the radiation can at most saturate only the few transitions in resonance with the laser, whereas inside the capillary, where the pressure is much higher, the more frequent collisions repopulate the absorbing state(s) and the lines are broadened so that more transitions are in re~onance.~Thus more molecules are excited, and the lines are further from saturation because the excited molecules can transfer their energy to other molecules. This can increase the absorption greatly.
Discussion Our measurements of N, the total flow through the Pyrex nozzle as it varies with changes in T , and P,, can be compared with N = P , , L ~ * C D { ( ~ ’ / R T , W ) [ ~+/ (1)][q+1)/(v‘-1)}1/2 ’)”
(1)
This is an expression for the flow from a free-jet source approximated as being isentropic and quasi-one-dimensional. See., for example, eq 2.10 in the excellent review by Miller.5 We have modified his equation so the flow is in mol s-I and a factor CD, the discharge coefficient, is included to make allowance for the presence of the boundary layer. (See the useful and thorough article by Weaver and Frankl.6) A* is the area of the exit of the source, R the gas constant, W the molecular weight of the gas, and y’ the ratio of specific heats. Our measured flows conform to eq 1 by being linear in P, but show a somewhat larger decrease of the flow with increasing temperature if CD is taken to be a constant. The differences in the extrapolated temperature rises AT, for mixtures of the monatomic gases Ar and He found for the Pt and Pt black surfaces can be understood semiquantitatively with the help of AT = Tj - Tb = t[(mlnlrl + m2n272)/(mlXl + m2X2ml71 + nz7z)l x ( 5 / 4 ) T n - Tbl/(l 4- (CgC i- CXiUil;’/ZAj)/k~(niyi nz’Y2)1 which is eq 15 of I with four misprints corrected, and model A (5) Miller,D. R. In Atomic and Molecular Beam Methods; Sales, G., Ed.; Oxford: New York, 1988; Vol. 1, p 14. ( 6 ) Weaver, B. D.; Frankl, D. R. Reu. Sci. Instrum. 1987, 58, 2115.
1758 The Journal of Physical Chemistry, Vol. 95, No. 4, 1991
of 1. The latter is based on the following assumptions: the absence of slip in the velocity, i.e., the Ar and He atoms reach the same speed as a result of collisions during the expansion; the thermal energy of the gas in the nozzle is converted so completely into the energy of directed flow that Til, the temperature that characterizes the distribution of speeds parallel to the flow, becomes negligible [Le., ' 1 2 ( m l X+ , m2Xz)u2= 5/zk~T,,],where u is the flow speed); the composition of the mixture remains uniform as the gas expands. In eq 2 m is the atomic mass, n the number of molecules in the beam arriving at TCI in unit time and per unit area, y the fraction of the molecules affected by the temperature of TCI, X the mole fraction, t the emissivity of T C l , D the Stefan-Boltzmann constant, Xi the thermal conductivity, ai the cross-sectional area of the fine thermocouple wires, li the length of the wires, and A,, the area of TCI. We consider the limit ATn obtained from eq 2 for large fluxes (which is equal empirically to the limits ATp for large nozzle pressure and AT, for x = 0): ATn = ATx = ![mlXiyi/m2X~y, + 11 X ( 5 / 4 ) T n - rbj!/(miXl/m2x2 + I)(xI?'I/?'I+ X2) (3)
We see that a measurement of ATx together with the assumptions used in our model determines yl/yz. For 10%by volume of Ar in He with Tb = T,, = 300 K our value of ATx, 180 f 18 K, on Pt gives 7Hc/yAr = 0.59 f 0.05, while on Pt black our value ATx = 1 11 f 11 K gives yHe/yAr = 0.81 f 0.05. The notable increase of AT, for the mixture compared to either pure gas (see I) follows because for Ar and He atoms moving at the same speed the heavier Ar has more, and the lighter H e less, kinetic energy than the average and because, with yHeless than yAr,TCI responds more to the Ar than the He. On the Pt black surface both Ar and He are expected to make more collisions with TCI, so that both y's increase. However, if yAris already near unity, for example, if we assume 0.9, on Pt, it can increase only slightly. Then with a change from Pt to Pt black, yArmay increase from 0.9 to 1 while the corresponding increase of yHe,derived from the measured ratios given above, would be from 0.53 f 0.04 to 0.81 f 0.05. This method of comparing the heating caused by one component in a binary mixture with the cooling produced by the other should provide a useful kind of internal calibration for the measurement of relative values of y for the components. Here it is useful to mention the recent work of Fernandez de la Mora and R~sell-Llompart,~ who used small thermocouples to study the focusing of heavy molecules such as CBr, and W(CO)6 seeded into nozzle beams of H2 and He. Their thermocouples, like ours, were much more sensitive to the heavier than to the lighter molecules in the beam. The result was that the temperature rises recorded could be used to measure the angular distributions of the heavier molecules. We note that some of the restrictions imposed by model A could be relaxed. The mixture need not be assumed to stay uniform during the expansion, if a mass spectrometer capable of scanning the whole angular range of the beam were available; slip in the velocity need not be assumed to be absent, if the speeds of the He and Ar could be measured (at present our velocity selector is not able to reach the higher speeds, ca. 2 km s-l, required); the average energy of the beam, i/zm(ul12),could be evaluated. There is a noticeable difference between the predictions of model A and our measurements not only for mixtures but also for the pure gases. As we stated in I, at T,, = 300 K, AT, is 50 f 4 K for Ar and 63 f 6 K for He, while model A gives 75 K for both pure gases. This difference is also present at higher nozzle temperatures. The results in Figure 1b show that for Ar the extrapolated temperature rises measured relative to those predicted by model A (ATx,cxp/ATx,modelA) are 0.67, 0.72, and 0.67, each f0.05, for T,, equal to 300, 356, and 397 K respectively, while for He at 300 K the ratio is 0.84. A possible reason for these differences may be that in the experiments some of the kinetic energy of the flow is degraded
Tao and Greene to heat and conducted away to the nozzle. This would reduce the fractionfof the energy remaining in the molecules of the beam. If we take the energy originally available to be S/zkBT,,for a monatomic gas expanding enough to cool T,, (see I) nearly to zero and 2kBTj [=E = 2kB(T,, AT,)] as a measure of the energy left in the molecules when they arrive at TCI, f becomes 0.93 for Ar (ATx = 50 K) and 0.97 for He (ATx = 62 K) with T,, = 300 K, i.e., 7 and 3%, respectively, of the energy available in the nozzle chamber is lost during the expansion. More extensive and accurate measurements should show whether this explanation of the difference is realistic. The linearity of the reciprocal plots (Figures 1 and 3 of this paper and Figures 6-9 of I) suggests that there are two pieces of information to be obtained from them. As we show above, the intercept on the ordinate gives AT,,, a measure of the energy of the molecules of the beam. The slope also depends on AT,, but in addition varies with y. Another way of finding y values that was not stated explicitly in I is to differentiate eq 11 of I to get d(AT')/dn-' = [m(U1l2) /4kg - Tb]-'(xX;1ailiWi/2Aj i-
+
i
tPt'Jc)/ykB = (ATn)-'C/ykB (4)
where C is a constant. We change the variable for A T i from n-l to PLi or x2 by making the assumption that the flux n (x, y = z = 0) of molecules striking T C l , when it is set on the axis of the beam, is proportional to N/x2 or to P,,/x2. This requires also that the angular distribution of the beam around its axis be independent of the gas, something that Weaver and Frank16 show is true for Ar and He. Thus we get relative values of y from YHc/YAI
= ([d(ATi)/dX21Ar/ [d(ATi)/dX2)1He) I(ATx)Ar/(ATx)H,l(NAr/NHc)
nAj[(l/2)ym(ul12) + ~vEint(Tn)I+ nvAjyvEint,L = 2nAj(ykBTj [CXiail;i/2nAj + tptuc/n](Tj - Tb) +
+
1
yvEint(T,)/21 (6)
This is eq 9 of I with some additions needed because the gas is not monatomic. ,!Tint( Ti) is the internal energy at equilibrium at Tiwhen the laser is off, ,Tint,'the average increment to the internal energy in a molecule that absorbs a quantum from the laser, n, the number of beam molecules excited by the laser arriving at TC1 per unit area in unit time, and y and yv the fractions of the beam molecules accommodating their translational and internal energy respectively on TC1. We next subtract 2nAjyTb from both sides of eq 6, take Ein,(Tj)- Einc(Tn) equal to cv,int(Tj - T,,), and put T,, equal to Tb and AT equal to T j - Tb to get I/zm(Ui12)- 2kBTb + (nvyv/ny)[Eint,LI = [2k+ ~ c'mjcp/nAjy + (yv/y)cv,intlAT (7)
where c'is the decay constant for TCI measured as it cools after being heated by a small He-Ne laser, and mj and cp are respectively the mass and specific heat of TCl (see 1).
(7) Fernandez de la Mora, J.; Rosell-Llompart, J . J . Chem. Phys. 1989, 91, 2603.
(5)
Using our results from Figures 7 and 8 of I together with our measured value of NHe/NAr(=2.3 f 0.2), we find yHe/yAr to be 0.54 f 0.05 on Pt. (Unfortunately the corresponding measurement for the Pt black surface seems to be very imprecise, and changes in the apparatus make redoing it temporarily impractical.) Although the Pt black surface does appear to provide nearly complete accommodation for Ar atoms in a beam, a still more efficient surface would be desirable. One interesting possibility is to use one described by Nindi and Stulik,* who bombarded a silver surface contaminated by microparticles of diamond to get a surface covered with closely packed silver cones having bases of ca. 1-pm diameter. We can make estimates of both the average energy absorbed by each SFs molecule and the efficiency of the transfer of this internal energy to the surface. To do this we note that when becomes constant at steady state, equating the energy arriving at and leaving TCl gives
(8) Nindi, M.; Stulik, D. Vacuum 1988, 38, 1071.
J . Phys. Chem. 1991, 95, 1759-1768 When the SF6beam is on and the laser off, n, is zero and AT becomes ATBO. When the laser is on, its incremental effect is measured by 6T = ATBL - ATOL - AT,,, so we get 6T/ATBO = (nvy,/ny)Ei,,.L/[I/zm(vl12)- 2kBTb1 (8) where values of 6T and ATBo are given in Table I. To evaluate n/ny from eq 8, we take l/zm(v112) = 2.5kB(T,- T I I=) 6 kJ mol-', = 11.3 kJ mol-', and T b = 300 K. We also assume that a negligible fraction of the energy absorbed by the SF6 molecules from the laser beam is transferred to the kinetic energy of the molecules during the expansion. If we now take y and yvto be unity on the Pt black surface, the result is nv/n = 0.09, i.e., 1.O kJ mol-' or 0.09 quantum absorbed/molecule of SF6in the beam. This very approximate value is about one-fortieth of the excitation achieved by Lester et aL3 for their SF6 beam from a nozzle at 290 K with a laser IO times more powerful and excitation somewhat farther upstream from the exit of the nozzle. We can also find relative values for yvon the Pt and Pt black surfaces: Yv.Pt
=
Yv.Pt black(6TC'X2/Aj)Pt/(6Tc~2/Aj)Pt
black
(9)
from which with values for 6T and c'given above and A, = 0.040 f 0.004 cm2 (0.044 f 0.004 cm2) for the Pt (Pt black) surface we get ~ ~ , black ~ ~ = 0.7 / fy 0.3. ~ Because , ~ the value on the Pt black surface is likely to be close to unity, this suggests that vibrational energy transfer from SF6 to the Pt surface is also efficient. This kind of experiment, if suitably refined, and others such as ones with a variable temperature bolometer9 may provide (9) Faubel, M.; Schlemmer, S . J . Phys. E: Sci. Instrum. 1988, 21, 75.
1759
useful measurements of the total accommodation of energy on surfaces for comparison with studies of changes in the populations of individual quantum states. The latter, although much more detailed, may be difficult to integrate over the experimental distributions of final states, something that is needed for a determination of the survival probability of vibrational energy after an excited molecule collides with a surface.1° Conclusion The experiments reported here on temperature rises produced when beams strike surfaces show that (1) for monatomic gases 2kBT may be a useful measure of the total (kinetic and local thermal) energy in a beam for comparison, for instance, with the total available energy, the enthalpy of the gas in the nozzle, (2) the probability of energy transfer to the surface increases significantly, as expected, when the surface is roughened, and (3) energy transfer from the internal degrees of freedom of a molecule can also be measured, and, for the vibration of SF,, is nearly as efficient as is the transfer of translational energy.
Acknowledgment. We are grateful to Prof. G. J. Diebold for loaning us the laser, to Drs. M. A. Pickering and J. T. Keeley of the CVD Division of Morton Thiokol, Inc. for providing the ZnSe windows, to the Department of Energy, Division of Chemical Science, for its support under Grant DE-FG02-85ER13441, and to the National Science Foundation through the Materials Research Laboratory of Brown University for electron micrographs. Registry No. Pt, 7440-06-4; SF6, 2551-62-4. (10) Houston, P. L.; Merrill, R. P. Chem. Rev. 1988, 88, 657.
Photoinduced Pathways to Dissociation and Desorption of Dioxygen on Ag( 110) and Pt( 111) S. R. Hatch,
X.-Y.Zhu,J. M. White, and Alan Campion*
Department of Chemistry, University of Texas, Austin, Texas 78712 (Received: August 13, 1990)
The rates for photodissociation and photodesorption of dioxygen on Ag( 110) under UV irradiation have been measured with TPD and HREELS. O2adsorbs molecularly with the 0-0axis parallel to the surface and aligned along the Ag[ lTO] azimuth. This alignment permits polarization measurements to be made at normal incidence. Photolysis rates are found to be independent of the azimuthal orientation of the electric field of the light. Dependences on angle of incidence are consistent with the angle dependence of metal absorption predicted by Fresnel's equations. Taken together, these.observationsrule out direct photoexcitation and suggest substrate excitation followed by dissociative electron capture as the photodissociation mechanism. Similar energy thresholds (2.8 eV) suggest that both channels are activated by a common process. A reaction model is proposed whereby substrate charge transfer induces dissociation. Dissociating adatoms, in competition with dioxygen for available binding sites, force some desorption of the latter. Analysis using this model reproduces the dependences on initial O2coverage and extent of irradiation for both channels and allows calculation of the photodissociation cross section. In contrast, angle-dependent photolysis rates and wavelength dependences for O2on Pt( 11 1) suggest that direct intraadsorbate excitation contributes significantly to photodesorption and may participate in photodissociation.
I. Introduction Recent studies have established that photochemical reactions of molecular adsorbates on metal surfaces readily occur, despite the possibility of rapid quenching of excited electronic states by nonradiative energy transfer to the surface.'" Both photoinduced (1) Costello, S . A,; Roop, B.; Liu, Z.-M.; White, J. M. J . Phys. Chem. 1988, 92, 1019. (2) Domen, K . ; Chuang, T. J. J . Chem. Phys. 1989, 90, 3318. (3) Pimental, G . L.; Grassian, V. H. J . Chem. Phys. 1988, 88, 4478. (4) Nuzzo, R . G.; Dubois, L. H. J . Am. Chem. SOC.1986, 108, 2881. (5) Avouris, P.; Walkup, R. E. Annu. Reu. Phys. Chem. 1989, 40, 173. (6) Ho, W. In Desorption Induced by Electronic Transitions, DIET IV; Springer Series in Surface Science, Springer: Berlin, in press.
0022-3654191/2095-1759$02.50/0
dissociation and desorption of molecules adsorbed on metals under UV irradiation have been observed. While the detailed mechanisms responsible for surface photochemistry remain to be elucidated, two limiting cases have emerged for metallic substrates: direct photodissociation following absorption of a photon by a bond or dissociative electron within the adsorbatesubstrate capture by the adsorbate following the photoexcitation of a substrate electron.I*l2 Substrate-mediated photochemistry may (7) Liu, Z.-M.; Costello, S . A,; Roop, B.; Coon, S. R.; Akhter, S.;White,
J. M.J . Phys. Chem. 1989, 93, 7681.
(8) White, J. M. Presented at the 9th International Summer Institute in Surface Science at Milwaukee, WI, 1989. (9) Hanley, L.; Guo, X.; Yates, Jr., J. T. J . Chem. Phys. 1989, 91, 7720.
0 1991 American Chemical Society