Efficient infrared dye sensitization of van der Waals surfaces of

Mar 25, 1986 - Polaroid Corporation, Waltham, Massachusetts 02254. B. A. Parkinson*. Central Research and DevelopmentDepartment, Experimental Station ...
0 downloads 0 Views 626KB Size
Langmuir 1986,2, 549-553

549

Efficient Infrared Dye Sensitization of van der Waals Surfaces of Semiconductor Electrodes? Mark Spitler* Polaroid Corporation, Waltham, Massachusetts 02254

B. A. Parkinson* Central Research and Development Department, Experimental Station E328/105, E. I. du Pont de Nemours & Co., Inc., Wilmington, Delaware 19898 Received March 25, 1986 van der Waals surfaces of the layered semiconductors n-WS2and n-WSe2can be sensitized with high quantum yields by an infrared absorbing thiapentacarbocyanine dye. A quantum yield of electrons per absorbed photon of 0.6-0.8 is as high as any previously observed on sin le-crystal semiconductor surfaces and results in sensitized photocurrent densities in excess of 40 pA/cmf. The potential, wavelength, and dye concentration dependence of the sensitized photocurrent are studied in the presence and absence of the supersensitizerhydroquinone. Adsorbed dye aggregates could be identified and selectively photooxidized. Introduction

A frequent observation made in the study of the dye sensitization of photocurrent a t semiconductor electrodes is that the quantum efficiency for conversion of absorbed photons to electrons is low, usually on the order of a few percent.' In several recent examinations of this problem, it was concluded that inefficient sensitization could be attributed to states a t the electrode surface which facilitated the return of the transferred electron from the solid to the oxidized dye layer quenching the production of photocurrent.2 In some cases these states have been attributed to hydrolyzed surface of oxide electrodes such as ZnO or Sn02. It is fairly clear from these examples that any surface layer on a semiconductor electrode could lead to an efficient quenching of the photocurrent through a back reaction. Such surface layers are absent a t electrodes made of the group group VI (group 6) dichalcogenides? and it was our belief that these materials, in this case WSez, would make an excellent substrate for the study of dye sensitization. The van der Waals (0001) surfaces of these layered semiconductors do not oxidize or interact strongly with solvents and therefore provide an abrupt interface between the electronic states of an adsorbed dye and the energy bands of the semiconductor. The saturation of the bonding on these surfaces also prevents any chemical reactions or hydrogen bonding interactions with the surface which may preturb the dye structure a t the interface. In addition single-crystal electrodes of these compounds can be easily cleaved to yield virgin surfaces, assuring a clean reproducible substrate for each experiment. Also, a t biases several hundred millivolts positive of the flatband potential, there should be minimal complications stemming from surface recombination or trapping in these electrodes. The result should be a high efficiency of current production and thus a useful model system for the study of dye-sensitized electron transfer. To study the dye sensitization of these materials an infrared dye must be employed because WSe2 has an in'In this paper the periodic group notation in parentheses is in accord with recent actions by IUPAC and ACS nomenclature committees. A and B notation is eliminated because of wide confusion. Groups IA and IIA become groups 1 and 2. The d-transition elementa comprise groups 3 through 12, and the p-block elements comprise groups 13 through 18. (Note that the former Roman number designation is preserved in the last digit of the new numbering: e.g., III+3 and 13.)

direct band gap of 1.25 eV. An infrared dye was found in the form of a thiapentacarbocyanine dye with an absorption maximum a t about 1.1eV. It is known to sensitize silver halides4 and has been shown to produce cathodic currents through a photoreduction a t p-GaAs electrodes.6 In the experiments described herein, the energy of the WSe2conduction band is sufficiently positive that this dye should sensitize through photooxidation over this band. Experimental Section The WSez used in this work was grown by chemical vapor transport as has been described earlier! The doping density was determined through a Mott-Schottkyanalysis of capacitance data and found to be about 1 X lo1' cm9. Flatband potentials for these crystals in a LiCl-saturated MeOH electrolyte were obtained through the same method and established to be -0.14 V vs. AgfAgC1.

Crystals were mounted as electrodes in a manner which permitted cleavage of the crystal to expose a fresh face before and between immersions in the electrolyte! Cleaving was accomplished in the usual manner by the use of transparenttape pressed against the crystal which was subsequently removed along with a thin layer of the crystal. The infrared dye used in this work was obtained from the Accurate Chemical Co. as the iodide salt of 3,3'-diethyl9,11:15,17-dineopentylene-2,2'-thiapentacarbocyanine. Its structure is given in Figure 1 along with spectra of this dye in methanol and benzene. The purity of the dye was established spectroscopically through the benzene spectrum where the observed extinction coefficients of the spectral maxima were found to be within 5% of literature values: Hydroquinone was sometimes present as a supersensitizer. It was reagent-grade material and used as received. To prevent the rapid decomposition of the dye in solution, both stock and working solutions of the dye were deoxygenated through the continuous bubbling of argon. In the presence of oxygen the dye oxidized within minutes in methanol and immediately in water. Spectral analysis of the orange-colored decomposition (1)Kavassalis, C.; Spitler, M. T. J. Phys. Chem. 1983,87,3166 and references therein. (2) (a) Bressel, B.; Geriacher, H.Ber. Buncenges. Phys. Chem. 1983, 87,398.(b) Arden, W.;Fromhen, P. J. Electrochem. SOC.1980,127,370. (3)Stickney,J. L.;Rosasco, S. D.; Schardt, B. C.; Solomun, T.; Hubbard, A. T.; Parkinson, B. A. Surf. Sci. 1984,136,15. (4) Brooker, L. G. S.; Vittum, P. W. J. Photog. Sci. 1957, 5 , 71. Brooker, L.G. S. In Recent Progress in the Chemistry of Natural and Synthetic Coloring Matters; Gore, T. S., et al., Eds.; Academic Press:

New York, 1962;p 573. (5)Tributach, H.; Gerischer, H. Ber. Bunsenges. Phys. Chem. 1969, 73,850. (6)Kline, G.; Kam, K.; Canfield, D.; Parkinson, B. A. Sol. Energy Mater. 1981,4,301. Kline, G.;Kam, K.; Ziegler, R.; Parkinson, B. A. Solar Energy Materials 1982,6,337.

0743-7463f 86f 2402-0549$01.50f 0 0 1986 American Chemical Society

Spitler and Parkinson

550 Langmuir, Vol. 2, No. 5, 1986 6

Dye Spectra

L

1.5 . u)

n

..

E1

Y

iC.'

600

700

BOO

900

1000

1100

L

WAVELENGTH (nml

E 10-

Dye Structure

-8

2

3 0

P

,

.

D

I

U

.

.

E

.

05-

c

F i g u r e 1. Structure is given of the infrared dye used in these experiment (below). The absorption spectra are shown for the dye in methanol and in benzene at concentrations of 1.0 x and 9.0 X lo+ M, respectively (above).

A

Wavelength (nm)

B l

'IWW

900

1000

1100

1200

Wavelength (nm)

F i g u r e 2. (A) Action spectrum of WSe2 is given after 5 min of M solution of the dye (open squares). immersion in a 2.5 X After 3 h in this solution the photocurrent response increases (squares with dots). Without removing the electrode from the cell, the dye concentration in the electrolyte is increased to 1.1 x M (dark squares). The maximum of the photocurrent response increases by about 25% and shifts to the blue by 15 nm as is made clear by the difference spectrum between the two higher spectra in the upper figure, which is shown in the lower figure. product(s) showed no absorption to the red of 500 nm. Dyesensitized photocurrents were measured with an automated photocurrent spectroscopy system which has been described previou~ly.~A lock-in technique with chopped light was used

(7) Folmer, J. C. W.; Tuttle, J. R.; Turner, J. A.; Parkinson, B. A. J. Electrochem. SOC.1985, 132, 1607.

Sensitization of Semiconductor Electrode Surfaces

Langmuir, Vol. 2, No. 5, 1986 551 1500

.

I ........ I- . ..... ........ 2

1

1000

Y

r

g J 8

I

.............

0 = 0.044

500 -

a.

I 900

1000

1100

1200

Wavelength (nm)

Figure 4. The potential bias of the electrode influences the action spectrum of aggregate forms of the dye. A bias of -0.4 V results in a maximum at 1110 (small (dark squares) whereas a bias of 0.05 V shifts this to 1080 nm. (Large squares) The dye conM. centration was 1.0 X

0 900

1000

1100

1200

Wavelength (nm)

Figure 5. A freshly cleaved WSez electrode immersed in a 1.1 X M dye solution gives action spectra with distinct and reproducible ripples on the spectral envelope. value. With addition of the reducing agent the oxidation of the surface dye is hindered until about +500-600 mV. Cyclic voltammetry was employed to determine the oxidation potential of the dye. The results were not unambiguous. At a Pt electrode in the methanolic electrolyte a reversible oxidation wave was observed a t +510 mV followed by an irreversible wave a t +650 mV, potentials which correspond to the decline in current in the photocurrent voltage curve of Figure 3. In general the electrochemical oxidation of thiacyanine dyes in methanolic solutions is an irreversible process8 so that the reversible wave a t +510 mV is not readily explainable although the cyclohexene rings on the methylene bridge may play a role in stabilizing an oxidized product. A unique characteristic of these dye-sensitized electrodes is that the action spectrum depends on the bias potential applied to the electrode. In Figure 4 is shown the photocurrent response of WSe2 at -0.4 and +0.05 V. Whereas the positively biased electrode shows a maximum a t 1080 nm, the negatively biased electrode, which has a significantly reduced response, reveals a broader maximum which is red-shifted (-1110 nm). After fresh cleavage of the WSe2 electrode and immersion in a 1.1 x M dye solution, the distinct and reproducible structure in the action spectrum of Figure 5 were observed. In contrast to adsorption from low-con(8)Piechowski, A. P. J. Electroanal. Chem. 1983, 145, 67.

900

1000

1100

1200

Wavelength (nm)

Figure 6. The quantum yield of this system at a +0.184-V bias and at the sensitization maximum was found to be 4.4% for incident radiation without correction for reflection or absorption losses in the cell arrangement. centration solutions of the dye, the dye must initially adsorb from this high-concentration solution to give rise to aggregates and the resulting modulated structure in action spectra. When the bias of the electrode in this experiment is changed to +184 mV, the current increases and these spectral features change somewhat as has been noted above (Figure 6). A quantum yield measurement a t the sensitization maximum gave a photo-to-electron yield for incident radiation of 4.4%. With an incident power of 21.2 pW/cm2, the current density in this case was 42.2 pA/cm2. In an attempt to measure the absorption spectrum of the adsorbed dye, thinly cleaved samples of the electrode were immersed in 1 X M solutions of the dye and inserted in a spectrophotometric arrangement constructed to measure a b s ~ r b a n c e . ~ However, interference fringes in the spectra caused by these thin samples precluded all but an estimate of the upper limit of this absorption of 7 f 1%. In Figure 7A is seen further evidence of the role of aggregates in the sensitization of current. When a sensitized WSe2 electrode was illuminated a t 1100 nm for an extended period of several hours, the action spectrum shows evidence of the preferential oxidation of forms of this dye absorbing a t this wavelength. The structural features of the action spectrum of this complement of surface aggregates are apparent in the difference spectrum before and after “burn off“ shown in Figure 7B. Although the experiments presented and discussed here were done with WSe2, essentially identical results were obtained with isostructural n-WS2, but in general the crystals of the latter material were of lower quality and thickness, limiting the number of possible cleaving experiments.

Discussion The absorption spectra of the dye given in Figure 1 reveal the spectral characteristics of two forms of the molecule which exist in s ~ l u t i o n .The ~ cis form of the dye, formed by isomerization about the double bonds in the methine bridge between the thiazole nuclei, absorbs at 800 nm in methanol in contrast to the trans form which absorbs a t 980 nm. The shift of the red peak to 1080 nm in benzene has been attributed to the higher index of refraction of this solvent. The absence of the cis absorption in benzene indicates that it is the polar character of the (9) Spitler, M. T.; Calvin, M. J. Chem. Phys. 1977, 67, 5193. (10)Hem, A. H. In The Theory of the Photographic Process, 4th ed.; James, T. H.,Ed.; Macmillan: New York, 1977; p 235.

Spitler and Parkinson

552 Langmuir, Vol. 2, No. 5, 1986

Extended Illumination

0’ 900

I 1000

1100

1200

Wavelength (nm)

,..“ .. . .

..

rr

a

E

..

.. 5

a

-100 900

1000

1100

1200

Wavelength (nm)

Figure 7. (A) After an electrode in a 1.0 X lo4 M dye solution is exposed to 1100-nm radiation, its action spectrum shows evidence of a “hole burning” effect where aggregates which absorb at this wavelength are preferentially oxidized. (B) Difference between the spectra taken before and after the “hole burning”. The before spectra would show a smooth peak much like the spectrum in Figure 6.

methanol which stabilizes the cis form of the dye. In this light it is clear that the action spectra of this dye in Figure 2 is the result of sensitization by the trans form of the dye on the electrode surface. This would indicate that the surface is essentially nonpolar in character and lacks any substantial hydrogen bonding as is expected for clean surfaces of these compounds. It is clear that the sensitization of current by the dye proceeds through the injection of an electron from the excited state of the dye into the conduction band of the semiconductor, since the dye produces anodic photocurrent. Given the measured oxidation potential of the dye, its excitation energy, and the semiconductor flatband potentials, this was the expected behavior. Estimating the ground-state oxidation potential of the dye to be +0.650 V, an excitation energy of 1.1eV would place the excited-state oxidation potential a t -0.45 V. The donor state of the excited dye is positive of this value by an amount equal to the dye’s rearrangement energy, which can be 200-300 mV, but this still places the donor level of the dye at or above the -0.14 V flatband potential of the electrode and enables it to inject electrons into the conduction band. A t the low solution concentrations of these experiments, the dye adsorbs flat onto the electrode surface as is typical for the planar thiacyanine dyes.1° This adsorption orientation is driven by the mutual hydrophobicity of the dye and the surface as well as the van der Waals interaction of the conjugated T system of the dye with the sulfur and selenium surfaces of these compounds. A t 2.5 X M dye in the electrolyte, coverage of the electrode by this

monomer form of the dye is nearly complete after several hours as the data of Figure 2A indicate. The action spectrum for dilute dye of Figure 2A is that of the monomer whereas that of the more concentrated solution shows a blue shift indicative of aggregates. These are most likely formed by adsorption from the 1 X M solution of additional dye onto and reorganization of the dye monolayer already present on the WSe2surface. The only other chemical species demonstrated to adsorb strongly on these surfaces is the triiodide ion through an unusual donor-acceptor interaction,” The action spectrum of this multilayer should be compared with the action spectrum of Figure 5 which was recorded immediately after the experiment of Figure 2 following cleavage of the electrode to provide a clean surface. In Figure 5 the maximum is shifted to 1100 nm and ripples appear on the envelope of the absorption band. We believe that this result is also indicative of aggregates, probably of small size such as different forms of dimers or trimers. These new aggregates would form from monomers adsorbed perpendicular to the surface and parallel to one another like bookends, with sufficient excitonic interaction to cause the small splittings seen in Figures 5 and 7. We believe that the nucleation of the perpendicular orientation of these aggregates would be mostly highly favored a t step sites on the surface of these layer compounds where the adsorptive interaction of the dye with the surface would consist of lateral as well as perpendicular components. It is clear that the adsorbed dye layer can consist of a mixture of monomer and aggregate species. However, the nucleation of these aggregates a t the steps could help explain the potential dependence of the action spectra of Figure 4 since step sites are known to be effective recombination centers which drastically affect the dark and light current-voltage curves at electrodes of these layer compounds.12 There could also be energetic grounds for such a potential dependence of the action spectrum. Each of the different forms of the dye, monomer, dimer, or trimer, will have a different energy for the level of the donor state relative to the substrate conduction band leading to a different penetration depth of the injected electrons into the crystal bulk before thermalization. I t is known from studies of organic insulating electrodes13that an electron with a large thermalization length will have less probability of surface recombination than one with a shorter thermalization length. The applicability of such image force models of charge transfer at insulators to the dye-semiconductor system of this work is presently being exp10red.l~ The quantum yield measurement made for the data of Figure 2 amounted to 1.1%of the incident light a t 1080 nm. To obtain the efficiency of current production, one requires an estimate of the amount of incident light which is absorbed by the adsorbed dye layer. A monolayer of adsorbed rhodamine B, which has the same extinction coefficient and about half of the surface area of this dye, absorbs 2.6% of the incident radiation at its spectral m a x i m ~ m . With ~ these figures, the quantum yield for electron injection can be estimated to be about 0.8. In Figure 6, sensitization of the WSe2 electrode with multi~~

~

(11) Turner, J. A.;Parkinson, B. A. J. ElectroanaL Chem. 1983,150, 611. (12) Parkinson, B.A.;Furtak, T. E.; Canfield, D.; Kam, K. K.; Kline, G. Faraday Discuss. Chem. Soc. 1980,70,233. Lewerenz, H.J.;Heller, A.; DiSalvo, F. J. Am. Chem. Soc.1980,102, 1877. Kautek, W.; Gerischer, H.; Tributsch, H. Ber. Bunsenges. Phys. Chem. 1979,83,1000. Furtak, T . E.;Canfield, D.; Parkinson, B. A. J. Appl. Phys. 1980, 51, 6018. (13) Willig, F.Adu. Electrochem. Electrochem. Eng. 1981, 12, 1. (14)Spitler, M. T., manuscript in preparation.

Langmuir 1986,2, 553-558 layers of the dye adsorbed from the high-concentration solutions of the dye resulted in a quantum efficiency of 4.4% based on incident radiation. With the upper bound of absorption by the dye layer being put a t 7%, a lower bound on the quantum yield for an aggregated adsorbed dye layer is 0.6. These estimates of the quantum yield in the range 0.6-0.8 confirm our premise that a clean, defect-free electrode surface can serve as an excellent interface for the study of dye-sensitized surfaces. Indirectly, these results support the contention of the literature that surface decomposition products and hydroxide layers can serve as efficient recombination centers in the quenching of dyesensitized photocurrents a t semiconductor electrodes.2

553

In conclusion, with its near ideal behavior, this system should prove useful for modeling the sensitization process. A mathematical treatment for the process using a onedimensional Onsager model is currently under exploration14 as is the extension of experimentation to other layered semiconductor/dye systems. Acknowledgment. Much of this work was done a t the Solar Energy Research Institute where support was from the Divisions of Chemical Sciences and Advanced Energy Projects, Office of Energy Science, USDOE. Registry No. WSz, 12138-09-9;WSe2, 12067-46-8; 3,3'-diethyl-9,11,15,17-dineopentylene-2,2-thiapen~c~b~~ine iodide, 15979-19-8;hydroquinone, 123-31-9.

Adsorption of Methoxy on Cu(100) Daniel Zerokaf and Roald Hoffmann* Department of Chemistry and Materials Science Center, Cornell University, Ithaca, New York 14853 Received September 17, 1985. I n Final Form: June 4, 1986 In order to develop a clear understanding of the adsorption of CH30 on metal surfaces we have studied the adsorption of CH30 on the Cu(100) surface since details on the adsorption site preference of CH30 have been recently determined by surface EXAFS. The observed site preference is the 4-fold hollow site with a Cu-O distance of 1.97 A. Through extended Huckel tight binding solid-state calculations we observe that the main interaction of CH30with surface copper atoms is through the 2e HOMO level. Effectively no contribution is made by any unoccupied MO due to the high energy of these. This interaction is in marked contrast to CO adsorption where substantial electron density is transferred to the 2r LUMO level. The electron redistribution that occurs on adsorption of CH30 is analyzed through projected density of states diagrams, changes in populations of molecular orbitals and atomic orbitals of the CH30 adsorbate, and changes in electron populations of atomic orbitals of surface atoms. Introduction We are interested in the electron redistribution that occurs when methoxy, CH30, bonds or chemisorbs on a metal surface. Our interest has been triggered by a number of studies. Our original interest was generated from the studies of Ho and co-workers' of CH30 on the Ni(ll0) surface. These studies involved high-resolution electron energy loss spectroscopy (HREELS), time-resolved electron energy loss spectroscopy (TREELS), and thermal desorption spectroscopy (TDS). Subsequent interest followed from several reports24 of the adsorption geometry of CH30 on a Cu(100) surface. Studies5 related to CH30 on Cu(ll0) surface are of related interest. At the same time the realization that the reverse process to CH30 adsorption could be an important step in methanol synthesis6 was another factor in the importance of studying CH30 adsorbed on a metal surface. Indeed there are a plethora of other metal substrates that are candidates for study and whose study would be of interest. In this report we focus on a highly specific example, the chemisorption of methoxy on a Cu(100) surface. The reason for this choice is that the adsorption geometry of CH30 on this surface was thought to be very well characterized. A recent report2 of a SEXAFS (surface-extended X-ray absorption fine structure) study indicated +On sabbatical leave for Spring 1985 from the Department of Chemistry, Lehigh University, Bethlehem, P A 18015.

0743-7463/S6/2402-0553$01.50/0

that CH30 bonds oxygen end down in a 4-fold hollow site with a Cu-O distance of 1.97 8,. This adsorption geometry was entirely consistent with a SEXAFS study' of 0 on Cu(100) where the Cu-0 distance is 1.94 8,and the oxygen atom adsorbs in a 4-fold hollow site. In addition, vibrational spectra4*of surface species indicate that the C-0 bond should be upright on the surface. (1) (a) Richter, L. J.; Gurney, B. A.; Villarrubia, J. S.;Ho, W. Chen. Phvs. Lett. 1984. 111. 185-189. (b) Bare. R. S.:Stroscio. J. A.: Ho. W. Sui/. Sci. 1985, i50,399-418; 1985; 155, L281-L291. (c)'Richt&, L. J.; Ho, W. J.Vac. Sci. Technol. A 1985,3, 1549-1553. (d) Richter, L. J.; Ho, W. J. Chem. Phys. 1986,83, 2569-2582. (2) The following article is concerned with HCOz adsorbed on Cu(100) but includes the first details of the Cu-0 bond distance for CH30 on Cu(100). Stijhr, J.; Outka, D. A.; Madix, R. J.; Dobler, U. Phys. Rev. Lett. 1985,54, 1256-1259. (3) (a) Stohr, J.; Gland, J. L.; Eberhardt, W.; Outka, D.; Madix, R. J.; Sette, F.; Koestner, R. J.; Doebler, U. Phys. Rev. Lett. 1983, 51, 2414-2417. (b) Outka, D. A.; Madix, R. J.; Stohr, J. Surf. Sci., unpublished results. (4) (a) Sexton, B. Surf. Sci. 1980, 88, 299-318. (b) Andersson, S.; Persson, M. Phys. Reu. B 1981, 24, 3659-3662. (c) Ryberg, R. Chem. Phys. Lett. 1981, 83, 423-426. (d) Ryberg, R. Phys. Reu. E 1985, 31, 2545-2547. (e) Ryberg, R. J. Chem. Phys. 1985,82,567-573. (5) (a) Wachs, I. E.; Madix, R. J. J. Catal. 1978, 53, 208-227. (b) Bowker, M.; Madix, R. J. Surf. Sci. 1980, 95, 19C-206. (c) Carlson, T. A.; Agron, P. A.; Thomas, T. M.; Grimm, F. A. J . Electron Spectrosc. Relat. Phenom. 1981,23, 13-24. (d) Sexton, B. A.; Hughes, A. E.; Avery, N. R. Surf. Sci. 1985,155,366-386. (e) Prince, K. C.; Holub-Krappe, E.; Horn, K.; Woodruff, D. P. Phys. Reu. E 1985, 32, 4249-4251. (0 Bader, M.; Puschmann, A.; Haase, J. Phys. Reu. E 1986,33, 7336-7338. (6) Klier, K. Adu. Catal. 1982, 31, 243-313. (7) Dobler, U.; Baberschke, K.; Stohr, J.; Outka, D. A. Phys. Reu. B. 1985, 31, 2532-2534.

0 1986 American Chemical Society