J. Phys. Chem. 1990, 94, 3381-3384 TABLE 11: Total Energies (hartrees) and Relative Energies (kcal/mol) of the 22y+ and 'B2 States of C3+ Obtained with the DZP Basis Set at the CISD/TZZP Optimized Geometry Employing CISD, CISD+Q, CISDTQ, and FULL CI Energies with Various Choices of Active Orbital Spaces" active space method 2ZY+ *B2 AE(2Z,+-2B2) 12 orbitals ClSD -I 13.04280 -113.05279 6.3 - 1 13.06858 - 1 13.06722 -0.9 ClSD+Q -113.071 IO -113.07037 -0.5 ClSDTQ -113.07399 -113.07142 -1.6 FULL CI 24 orbitals ClSD -113.14591 -I 3.16668 13.0 6.3 CISD+Q -113.18694 -I 3.19693 5.1 CISDTQ -113.21412 -I 3.222 20 14.5 35 orbitals ClSD -113.21383 -I 3.23690 8.1 ClSD+Q -113.26488 -I 3.27772 6.9 ClSDTQ -113.2987 -1 3.3097 15.7 45 orbitals ClSD -1 13.23245 -I 3.25745 C1SD-t-Q -113.28563 -113.30054 9.4 #See text for details.
state functions (CSF's). For the 2B2state in C2usymmetry, the number of CSF's is 3 925 903. (d) 45 orbitals. Here all of the virtual orbitals are active. Only ClSD and CISD+Q energies are reported in this active space. These results are directly comparable to the CISD/TZZP results in Table I and differ by less than 0.3 kcal/mol. The final results of this study on the relative energies of the 2Zu+ and 2B2states of C3+are shown in Table 11. Similar to the CASSCF results, all the CI relative energies in the valence space of only I2 orbitals are quite small. In fact the CISD+Q, CISDTQ, and FULL CI methods all predict the 2Zu+state to be below 2Bz in this restricted space. As the number of active orbitals increases, however, there is a systematic stabilization of the 2B2state relative to 2Zu+. This observation explains why the CASSCF 2Z,,+ - 2B2 separation is predicted to be misleadingly low. Thus, for example, the ClSDTQ relative energy increases from -0.5 to 5.1 to 6.9 kcal/mol as the number of active orbitals increases from 12 to 24 to 35. Similar trends are found for the CISD and CISD+Q relative energies. Noting that the ClSD and CISD+Q relative energies are increased by a little over 1 kcal/mol on going from 35 active orbitals to the full space of 45 orbitals, we estimate the CISDTQ 2Zu+- 2B2energy difference in the full space to be near 8 kcal/mol. Our final estimate of this barrier should probably be lowered by a similar amount, however, when we note that the CISDTQ method overestimates the stability of the 2B2state in comparison with the FULL CI results in the 12 active orbital case.
338 1
Thus, we estimate the actual difference in energy between the 2Zu+ and 2Bzstates to be around 7 kcal/mol. The insensitivity of the CISD and CISD+Q relative energies to the expansion of the basis set from DZP to TZ2P quality, and the near FULL CI quality of the CISDTQ wave functions employed, lead us to suspect that the actual barrier is probably 7 f 4 kcal/mol. In any event, we consider it very unlikely that the minimum on the C3+ potential energy surface occurs at the linear geometry. It is worth noting that the coefficient of the principal configuration in the CI expansion for the largest CISDTQ wave function is quite small for both states, 0.834 for 2Zu+and 0.867 for 2B2. The high barriers to linearity predicted by the CISD method obviously result from the neglect of higher (than double) excitations. Finally, the origin of the symmetry breaking observed for the *Z,+ and 2B2states can be gleaned from the qualitative valence structures of Figure I . For these two states, the unpaired electron resides principally on the terminal carbon atoms. Following Allen et al.I2 (and references therein) we speculate that the symmetry breaking in these states results from the desire for the two terminal carbons in either of the localized structures to have different orbital sizes. That is, the two electrons in the lone pair of one of the terminal carbons want to be in a more diffuse orbital than the orbital on the other terminal carbon atom which has only one electron in it. The resonance energy gained by delocalization in the high-symmetry solution is apparently insufficient to overcome the orbital size effect, and symmetry breaking results. While simpler resolutions to this problem may often be obtained, the full-valence CASSCF method is an obvious solution. To conclude, our finding that the 2B2state lies about 7 kcal/mol below the 2Zu+state on the C3+ potential energy surface lends convincing support to the original interpretation by FKTBZ of the Coulomb explosion results. While Vager and Kanter's analysis points out some interesting potential ambiguities in Coulomb explosion data, it should not be interpreted as implying that C3+ is a floppy linear molecule. C3+is, in fact, a strongly bent molecule with a substantial barrier to linearity. Acknowledgment. This research was supported by the U S . Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Fundamental Interactions Branch, Grant DE-FG09-87ER13811. We thank Professor Michael A. Duncan for inspiring this work. ~~
(12) Allen, W. D.; Horner, D. A.; DeKock, R. L.; Remington, R. B.; Schaefer, H. F. Chem. Phys. 1989, 133, 11.
Single Quantum Well Electrodes for Photoelectrochemistry C. A. Parsons,*.t M. W. Peterson, B. R. Thacker, J. A. Turner, and A. J. Nozik* Solar Energy Research Institute, Golden, Colorado 80401 -3393 (Received: January 24, 1990)
Photocurrent spectra and flat-band potential measurements of a single quantum well electrode at a liquid interface are reported for the first time. The quantum wells were well characterized by photoreflectance. Hot photoelectron distributions were characterized by photoluminescence.
Recently, very interesting results have been reported on the photoelectrochemical (PEC) behavior of superlattices used as photoelectrodes in PEC cells.'-* In all these experiments the superlattices consisted of from 20 to 200 quantum wells formed by alternating layers of smaller and larger band-gap semiconductors; the barrier thicknesses varied from values needed to form minibands and true superlattices (540 A)9 to values resulting in 'Current address: Bandgap Technology Corp., Broomfield, C O 80021.
0022-3654/90/2094-3381$02.50/0
isolated multiple quantum wells (2100 A). It has been reported in several studies that hot photogenerated carriers relax more (1) Nozik, A. J.; Thacker, B. R.; Turner, J. A.; Klem, J.; Morkoc, H. Appl. Phys. Lett. 1987, 50, 34. (2) Nozik, A. J.; Thacker, B. R.; Turner, J. A,; Peterson, M. W. J . Am. Chem. SOC.1988, 110,7630. (3) Nozik, A. J.; Thacker, B. R.; Olson, J. J. Nature (London)1987, 326, 450; 1985, 316, 6023. Thacker, B. R.; Turner, J. A.; Olson, J . M. J . Am. Chem. SOC.1985, 107, 7805.
0 1990 American Chemical Society
3382 The Journal of Physical Chemistry, Vol. 94, No. 9, 1990 slowly in semiconductor quantum wells compared to the bulk semiconductors.i0 Hence, hot carrier effects are expected to be enhanced in quantized semiconductor structures as compared to bulk semiconductors. Previous studies have shown that both multiple quantum well (MQW) and superlattice (SL) structures in liquid junction PEC cells exhibit structured photocurrent (PC) spectra indicative of energy level However, attempts to determine the flat-band position of these samples, and hence the energetics of electron transfer, were not successful. Also, because of band bending in the presence of an applied electric field, energy levels in adjoining wells no longer align in energySsTherefore, it becomes difficult to relate observed photocurrent at a given excitation energy to a specific well energy level. This is an essential requirement in experiments designed to show a photocurrent contribution from hot carriers.s One solution to the difficulty introduced by multiple wells is to simply work with a single quantum well (SQW) in a configuration such that the only PC observed is from the SQW. In this case each peak in the PC spectrum can be related to only one energy level in the SQW, depending upon the relative rates of electron relaxation vs electron transfer.* Using a SQW also greatly simplifies modeling the results from these experiments. A potential disadvantage is in the much reduced absorption length and associated PC signal of a single well as compared to multiple wells. We report here the first results of PEC investigations on SQW samples, including PC, capacitance-voltage, photoluminescence (PL), and photoreflectance (PR). The SQW samples were grown in an atmospheric-pressure MOCVD reactor at 725 OC on p+-GaAs substrates. The structure was a 0.35-pm GaAs buffer layer, an inner Ga,-,AI,As barrier layer varying in thickness from 170 A to 3.5 pm, a GaAs well layer of 130-320 A, and finally an outer barrier of Gal-,A1,As of 270-it thickness. The carrier concentrations of the nondeliberately doped layers were determined by Hall measurements of individual, separately grown epilayers to be -loi5 n-type for the GaAs layers and p-type for the Ga,-,AI,As layers. Because the well width is less than one-tenth the calculated space-charge width, junction effects can be ignored. The aluminum fraction x was nominally 0.3 but was determined by PL and PR for each sample. For the two types of samples (A and B) discussed here, the nominal well width was 320 and 130 A, respectively. PEC electrodes were fabricated from pieces cleaved from the samples. The electrolyte was acetonitrile with 0.1 M tetraethylammonium tetrafluoroborate; the redox couple was 2 mM each of cobaltocene and cobaltocenium. Bias voltages quoted here are vs a Ag/AgCI reference electrode. PC vs voltage, capacitance vs voltage, and PC vs excitation energy data were collected as described else where.'^^ The excitation source was the monochromatized output of a 1000-W filament lamp at normal incidence to the sample; typically this provided about 50 pW/cm2. PL and PR were performed on samples in air. PL was excited by means of a sync-pumped, cavity-dumped dye laser operating at 600 nm and providing 7-ps pulses at an 800-kHz repetition rate ~~
(4) Zuhoski, S. P.; Johnson, P. B.; Ellis, A. B.; Biefeld, R . M.; Ginley, D. S. J . Phys. Chem. 1988, 92, 3961. ( 5 ) Johnson, P. B.; Ellis, A. B.; Biefeld, R. M.; Ginley, D. S. Appl. Phys. Left. 1985, 47, 877. (6) Lemasson, P.; Van Huong, C. N.; Chu, X., Sivananthan, S.;Faurie, J. P.J . Electrochem. Soc. 1988, 135, 1282. (7) Shtrikman, H.; Mahalu, D.; Tenne, R. Presented at the 7th International Conference on Photochemical Conversion and Storage of Solar Energy, Evanston. IL, Aug 1988. ( 8 ) Nozik, A. J.; Turner, J. A,; Peterson, M. W. J. Phys. Chem. 1988, 92, 2493. (9) Peterson, M. W.; Turner, J. A.; Parsons, C. A,; Nozik, A. J.; Arent, D. J.; Van Hoof, C.; Borghs, G.; Houdre, R.; Morkoc, H . Appl. Phys. Letr. 1988, 53, 2666. (IO) Edelstein, D. C.; Tang, C. L.; Nozik, A. J. Appl. Phys. Lett. 1987, 51. 48. Wise, F. W.; Tang, C. L. Solid State Commun. 1989.69. 821. Ryan, J. F.; Taylor, R. A.; Turberfield, A. J.; Maciel, A.; Worlock, J. M.; Gossard, A. C.; Weigmann, W. Phys. Reu. Lett. 1984.53. 1841. Xu, Z. Y . ;Tang, C. L. Appl. Phys. Left.1984,44,692. Uchiki, H.; Kobayashi, T.; Sakai, H. J . Appl. Phys. 1987, 62, 1010. Tatham, M.; Taylor, R. A,; Ryan, J. F.: Wang, W. 1.: Foxon, C. T. Solid-Sfate Electron. 1988, 31, 459.
Letters TABLE I: Transition Energies (in eV) for a Type B Sample (130-A Well with a 2.5-pm Inner Barrier) Determined by PR, PC, and PL and Assignments of the Transitions transition PR peaks assignmentC P C peaks PL peaksb 1.502 HI, I 1.506 1.502 L1, 1 1.513 1.548 H3, 1 1.551 1.614 H2, 2 1.612 1.613 1.621 L2, 2 1.714 H3. 3 1.654 1.669 p;;4Q 1.737 1.713 1.747
(
1.808
1.867
H5, US
U4" ( UL4, UL5, us4
1.872
H6, 6
1.922
UL6, U6O
1.966
L H 7 . U74
U8Q { UH8, UL7, U7"
Unresolved. P L spectra were taken a t 77 K. In order to compare the PL spectra with the PC and P R spectra taken a t room temperature, 0.088 eV has been subtracted from the P L peak positions. 4 are lost in the rising absorption edge of the Ga,-,AI,As barrier. The PC spectrum shown in Figure I C was taken at an applied bias of -1.5 V vs Ag/AgCI, but similar spectra were observed over a relatively wide range of -0.8 to -1.5 V. The often observed red shift of the transitions caused by the Stark shifti4was not observed in this voltage range. We believe this is caused by the absence of p-doping in the GaAs buffer layer which produces a large potential drop across that layer; studies are in progress with photoelectrodes containing p-doped buffer layers to eliminate this complication. Capacitance-voltage data for a type B SQW sample with the thicker inner barrier were also obtained. Previous attempts to determine the flat-band potential of MQW and SL samples failed because of nonlinear Mott-Schottky plots. By use of the SQW samples, however, linear Mott-Schottky plots were achieved. With cobaltocene/cobaltocenium present in solution, the flat-band potential (measured at 5 kHz in the dark or light) was found to be +0.07 V vs Ag/AgCI with a day-to-day variation of about 10%; with decamethylferrocene/ferrocenium the flat-band potential (14) For example see: Yu, P. W.; Sanders, D. G.; Evans, K. R.; Reynolds, D. C.; Bajaj, K. K.; Stutz, C. E.; Jones, R. L. Phys. Reo. B 1988, 38, 7796. Mender, E. E.; Bastard, G.;Chang, L. L.; Esaki, L.; Morkoc, H.; Fisher, R. Phys. Reo. E 1982, 26,7101. Vina, L.; Collins, R. T.; Mendez, E. E.; Want, W. I . Phys. Reo. B 1986, 33, 5939. Vina, L.; Collins, R. T.; Mendez, E. E.; Chang, L. L.; Esaki, L. Supperlatrices Mirostruct. 1987, 3, 9.
Figure 2. Time-integrated photoluminescencespectrum of 130-A SQW (Lb = 2.5 gm) showing hot luminescence tail and structure due to emission from excited well levels. The PL spectrum of bulk GaAs is also shown for comparison. Spectra were taken at 77 K.
shifted positive to +0.7 V vs Ag/AgCl. This indicates that redox couples can cause surface charging and band-edge movement in SQW electrodes as they do with bulk semiconductor electrodes. Only a slight frequency dispersion of the capacitance data was found over the range of 1-20 kHz. The flat-band potentials of SQW electrodes with p-doped buffer layers were also measured over the frequency range 1-20 kHz, and no frequency dispersion was found.15 The carrier concentration deduced from the capacitance-voltage data was in the range (p = 1 X 10I6 ~ m - that ~ ) was found previously for our bulk epitaxial samples of Gal,A1,As by using Hall effect measurements. It thus appears that the electrical properties of the SQW-solution interface are controlled by the thick Gal,A1,As inner barrier. The development of models and related experimental data to understand in detail the effects of the various interfaces on electrical properties is in progress. PC from a SQW with a solid-state Schottky barrier has been previously reported by Polland et al. using a high-intensity picosecond dye laser.I6 Our observations here confirm their suggestion that such spectra should be observable with low-intensity excitation. The value of the overall PC quantum yield (electrons out divided by photons in) at 1.55 eV in Figure I C is 0.57%; the internal quantum yield, based on the number of photons absorbd in the well, is a more relevant figure with respect to efficiency of charge transfer at the interface. Using the appropriate value for the absorption coefficient in the we11,I’ we obtain a value of 41% for the PC internal quantum yield at 1.55 eV; if we further correct for reflection of light at the interfaces,]*the value for the internal quantum increases to 63%. The photocurrent produced by absorption in the well consists of three component^:'^ thermionic emission (ITE), defect and/or phonon-assisted, nonresonant tunneling (Iph)(hopping conduction20), and resonant tunneling (IRT). ITE and I,,, are temperature dependent, while IRT is insensitive to temperature; also, the latter component generally dominates at low temperature. Initial studies2’ of the temperature dependence (11-300 K) of the photocurrent from gold Schottky barriers made with our SQWs (15) Thacker, B. R.; Szmyd, D.; Nozik, A. J. To be submitted for publication. (16) Polland, H. J.; Horikoshi, Y.; Gobel, E. 0.;Kuhl, J.; Ploog, K. Sur/. Sci. 1986, 174, 278. (17) Masumoto, Y.; Matsuura, J.; Tarucha, S.; Okamoto, H. Phys. Rev. B 1985, 32, 4275. (18) Aspnes, D. E.; Kelso, S . M.; Logan, R. A.; Bhat, R. J . Appl. Phys. 1986, 60, 754. (19) Choi, K. K.; Levine, B. F.; Bethea, C. G.; Walker, J.; Malik, R. J. Appl. Phys. Lerr. 1987, 50, 1814. Dutta, M.; Choi, K. K.; Newman, P. G. Appl. Phys. Lert. 1989, 55, 2429. (20) Calecki, D.; Palmier, J. F.; Chomette, A. J . Phys. C 1984, 17, 5017. (21) McMahon, W.; Thacker, B. R.; Nozik, A. J. Unpublished results.
J . Phys. Chem. 1990, 94. 3384-3387
3384
indicate about an order of magnitude decrease in the photocurrent from 300 to about 200 K and little change between about 100 and 12 K. This behavior is consistent with the above picture for the photocurrent. Future work will involve quantitative analyses and comparisons of the various theoretical photocurrent components with the experimental results; the role of tunneling and/or thermionic emission from excited levels in the well will also be examined. PL can be used to estimate the temperature of the photoexcited carriers by fitting the high-energy tail of the luminescence spectra to a Boltzmann Figure 2 shows the time-integrated PL from sample B when excited with 600-nm (2.07-eV) light which is absorbed by both the well and barrier layers. The time-averaged electron temperature estimated from a fit to the data in Figure 2 neglecting band-filling effects is 827 K. Also shown in Figure 2 is a PL spectrum for a 1.7-pm epilayer of GaAs taken under the same excitation conditions as the SQW sample. The PL tail is very small and fits to an electron temperature of 80 K, very close to the lattice temperature of 77 K. Further (22) Shah, J.; Leite, R. C. C. Phys. Re&. Lett. 1969, 22. 1304. (23) Nozik, A. J.; Parsons, C. A.; Dunlavy, D. J.; Keyes, B. M.; Ahrenkiel, R. K.Solid State Commun., submitted for publication.
analysis is required before definite conclusions can be made concerning the relative hot electron temperatures of MQW vs SQW samples as a function of excitation intensity. Finally, the peaks observed in the PL spectra of Figure 2 are tabulated in Table I; reasonable agreement is seen between the levels as determined by PL and those presented earlier by PR and PC . In summary, we report the first demonstration of PC from a SQW/liquid junction. The SQW samples are well characterized by PR, PL, and PC spectroscopy. The concurrent determination of the flat-band potential for SQW electrodes will allow experiments probing the participation of hot electrons in charge-transfer processes in electrochemical environments to be performed. Such experiments are currently in progress. Acknowledgment. This work was supported by the US.Department of Energy, Office of Basic Energy Science, Division of Chemical Sciences. We thank D. J. Dunlavy, B. M. Keyes, and R. K. Ahrenkiel for the PL spectra and J. Goral for the T E M data. Registry No. GaAs, 1303-00-0; GaAIAs, 37382-15-3; acetonitrile, 60-35-5; tetraethylammoniumtetrafluoroborate, 429-06-1 ; cobaltocene, 1277-43-6; cobaltocenium-, 12241-42-8.
Electrochemical Generation of Two-Dimensional Silver Particulate Films at Monolayer Surfaces and Their Characterization on Solid Substrates Xiao Kang Zhao and Janos H. Fendler* Department of Chemistry, Syracuse University, Syracuse, New York 13244-4100 (Received: December 15, 1989)
Electrochemical reduction of silver ions at the interfaces of monolayers prepared from arachidic acid (1) and dialkyl polymerizable phosphate surfactants 2 and 3 resulted in the two-dimensional formation of uniform films which consisted of interconnecting, 250 8, diameter, roughened silver particles. No formation of silver particles could be observed in the absence of monolayers or at monolayer interfaces prepared from positively charged dioctadecyldimethylammonium bromide (4) and polymerizable dialkylammonium chloride surfactant (5). The silver particulate films were transferred on quartz, highly oriented graphite, and celluloid-coated copper grids as substrates. Absorption spectra of the silver particulate film on quartz indicated an interband transition at 321 nm and an absorption maximum at 387 nm. Scanning tunneling microscopy of the silver particulate film on graphite revealed the presence of interconnecting particles with heights of 100-200 8, and average short and long axes of 200 and 300 A, respectively. Electron diffraction rings of the silver particulate film on a grid were analyzed for a face-centered cubic polycrystalline structure with a unit lattice constant of 4.08 8,. Resistivity of the silver particulate film on quartz was R cm. determined to be 4.5 X
Introduction Electrochemical reduction of silver ions a t negatively charged monolayer surfaces is reported here to result in the two-dimensional formation of a uniform silver film consisting of roughened, 250 8, diameter, interconnected particles. Further, the crystal structure of silver particulate film remained intact upon transfer onto solid substrates which allowed the characterization of this unique system by spectroscopy, scanning tunneling microscopy, electron diffraction, and electrical measurements. Experimental Section Arachidic acid (Sigma, 1) was purified by recrystallization Preparationand purificationOf *’ 3’ 4’ and have been described.’-, Spectroscopic grade CHCI, (Aldrich), the spreading solvent, and AgNO, (Baker Analyzed Reagents) were used as 0022-3654/90/2094-3384$02.50/0
received. Water was purified by using a Millipore Milli-Q filter system provided with a 0.22-pm Millistack filter a t the outlet. Surfactants were spread in a circular, 50 mm deep, trough having a surface area of 69.4 cm2. The schematics of the experimental setup are shown in Figure 1. A 1.O mm diameter, 3 cm long silver electrode was immersed into the subphase. Electrical connection as made through a 20 pm diameter platinum electrode which was floated (subsequent to monolayer formation) on the water surface at the middle of the trough. The surface of the aqueous 1 .O X M AgNO, solution was cleaned with a water aspirator just ( I ) Reed, W.; Gutherman, L.; Tundo, P. Fendler, J . H. J . Am. Chem. SOC. 1984, 106. 1897-1907.
(2) Serrano, J.; Mucio, S.:Millan, S.;Reynoso, R.; Fucugauchi, L. A.; Reed, W.: Nome, F.; Tundo, P.; Fendler, J . H. Macromolecules 1985, 18, ,999-2005. ( 3 ) Yuan, Y . ;Tundo, P : Fendler. J . H. Macromolecules 1989. 22, 29-35.
E 1990 American Chemical Society