J . Phys. Chem. 1985, 89, 4950-4956
4950
hydrocarbon molecules. Note Added in Proof. Recently, theoretically predicted vibrational frequencies for MgH2(X1Zg+)have appeared in the literature,35 with u,, u2, and u3 at 1641, 475, and 1667 cm-’, (35) J. A. Pople, B. T. Luke, M. J. Frisch, and J. S. Binkley, J . Phys. Chem., 89, 2198 (1985). (36) In this paper the periodic group notation 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 elements 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.. 111 3 and 13.)
--.
respectively, which are in good agreement with the infrared data reported here.
Acknowledgment. The financial assistance of the Natural Sciences and Engineering Research Council of Canada’s Operating and Strategic Grants Programme is gratefully appreciated; J.M.P. acknowledges NSERC for a graduate scholarship and W.H.B. expresses his gratitude to the Guggenheim Foundation for a scholarship during his sabbatical leave. Registry No. MgH2, 7693-27-8; CH3MgH, 63533-51-7; CH4, 7482-8; H,, 1333-74-0; Mg, 7439-95-4; Xe, 7440-63-3.
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on August 31, 2015 | http://pubs.acs.org Publication Date: November 1, 1985 | doi: 10.1021/j100269a013
Porphyrin Thin Film Cells in Ultrahigh Vacuum: The Requirement of Water and Oxygen for Photovoltaic Response Billy J. Stanbery,+ Martin Gouterman,* and Robert M. Burgess Department of Chemistry, University of Washington, Seattle, Washington 981 95 (Received: June 28, 1984; In Final Form: J u l ~ 12, ’ 1985)
Porphyrin sandwich cells [substrate/AI/MgTPP/Ag] (MgTPP = magnesium tetraphenylporphyrin) were prepared and studied in ultrahigh vacuum. It was found that an oxide layer is needed on the AI for a functioning Ag electrode to be formed, and without the oxide layer the Ag is islanded. With an oxide layer present, two devices with C = 20 nF were prepared and carefully studied. Resistance is low until shunts are burned out by pulse biasing the device. R , (defined as V / I at 1 V) varied from 1 kR to 1 MR depending on the history. I-Vcurves are nonlinear and symmetric, curving upward at higher voltages, until the devices are exposed to both oxygen and water vapor at pressures comparable to that of ambient air. Neither rectification nor a photovoltaic response is observed prior to exposure to both oxygen and water vapor; in one case a photovoltaic response was observed in the absence of a rectifying I-V curve. Capacitance/voltage measurements are consistent with its characterization as a MIS (metal-insulatorsemiconductor) device. These observations are explained by a photoelectrochemical process involving water-catalyzed AI oxidation by O2 and photogenerated porphyrin cation charge carriers.
Introduction In a previous paper,’ Kampas and Gouterman (KG) reported on the photovoltaic properties of porphyrin thin film devices. These devices were prepared at high vacuum (=lo” torr) by vapor deposition on a glass substrate. The devices can be represented as [substrate/A1(=200 A)/porphyrin(=2000 A)/Ag(=150
A)] (1)
The KG paper reported photovoltaic action spectra. Peak quantum yields (electrons/photon absorbed) were 0.9% for tetraphenylporphyrin and 3.4% for octaethylporphyrin. In the model that described these devices the Al/porphyrin interface was taken as a blocking contact where a Schottky barrier forms while the porphyrin/Ag interface was taken as ohmic. This type of device has been used by various other authors.*-’ Subsequent to the KG paper, Kampas, Yamashita, and Fajer (KYF) tested some 30 different porphyrins.8 In their most studied configuration the devices consisted of porphyrin vapor deposted on an A1 electrode. However, for an ohmic contact a Pt electrode in a solution of ferri-ferrocyanide was used, Le., a Wang cell.9 KYF reported a wide variation in quantum yield (0.2-10-4) depending on the porphyrin. “The best results were obtained with magnesium tetraphenylporphyrin (MgTPP) and cadmium porphin which yielded open circuit voltages of = l V, quantum yield of -0.2, and =l% power conversion efficiency.” Since the gram price of tetraphenylporphyrin is comparable to the milligram price of porphine, MgTPP was chosen by us for the studies reported here,
’
Present address: Advanced Device Technology, Boeing Aerospace Co., P.O. Box 39999, MS 88-43, Seattle, WA 98124.
0022-3654/85/2089-4950$01.50/0
in which device structures 1 were prepared and studied in ultrahigh vacuum (UHV). The UHV studies to be described here had two motivations. (i) Surface effect: At lod torr a monolayer can cover a surface in ==2s, if the sticking coefficient (Le., the probability of an incident molecule remaining on the surface)l0 is unity. Since any residual oxygen should rather effectively stick to Al, there is every reason to believe that the A1 electrode has a thin oxide coat when the devices are not made in UHV. (ii) Ambient atmosphere effect: KG devices exhibited rectifying I-Vcurves in which the A1 donates electrons to the porphyrin. It was thought that O2in the porphyrin film might provide the sites where electrons are trapped, since earlier work on phthalocyanines,” porphyrins,I2 and xanthine dyes’) showed that photocurrent depended on 0, pressure. As will be described below, there is indeed a surface effect (i) and ’ Kampas, F. J.; Gouterman, M. J . Phys. Chem. 1977, 81, 690. Usov, N. N.; Benderskii, V. A. Sou. Phys.-Semicond. (Engl. Transl.) 2, 580. (3) Lyons, L. E.; Newman, 0. M. G. Ausf. J . Chem. 1971, 24, 13. (4) Fedorov, M. I.; Benderskii, V. A. Sou. Phys.-Semicond. (Engl. Transl.) 1971, 4, 1198. (5) Ghosh, A. K.; Morel, D. L.; Feng, T.; Shaw, R. F.; Rowe, Jr., C. A. J . Appl. Phys. 1975, 63, 957. (6) Tang, C. W.; Albrecht, A. C . J . Chem. Phys. 1975, 62, 2139; 1975, 63. - - ,957. (7) Faulkner, L. R.; Fan, F. R. J. Chem. Phys. 1978, 69, 3334, 3341. (8) Kampas, F. J.; Yamashita, K.; Fajer, J. Nature (London) 1980, 284, 40. (9) Wang, J. W. Proc. Natl. Acad. Sci. U.S.A. 1969, 62, 653. (10) Roth, A. “Vacuum Technology”; North-Holland: Amsterdam, 1976; P 3. (1 1) Day, P.; Price, M. G. J . Chem. SOC.A 1969, 236. (12) Meshkova, G. N.; Meshkov, A. M.; Vartanyan, S. T. Sou. Phys.Semicond. (Engl. Trans.) 1969, 3, 605. (13) Vartanyan, A. T. Opt. Techno/. 1970, 37, 279.
0 1985 American Chemical Society
Porphyrin Thin Film Cells
The Journal of Physical Chemistry, Vol. 89, No. 23, 1985 4951 picoammeter. When we realized that the device resistance was comparable to the feedback resistance of the amplifier, so that changes in photoconductivity could be mistaken for a low photocurrent, another detection system was built. The device was then used as the feedback element of an operational amplifier, which is used as a null detector, and the high resistance and capacitance of the device tend to stabilize the circuit. (See Appendix.) Gases and vapors were introduced into the vacuum chamber to study their influence on electronic properties of the devices. Low-pressure oxygen was admitted via a Varian variable leak valve. Higher pressures of nitrogen, oxygen, and water vapor were measured with a manometer. For experiments requiring photoexcitation, light was supplied by a tungsten lamp. The light passed through a water filter, a Bausch & Lomb 250-mm grating monochromator, and various lenses to enter a port on the vacuum system and illuminated the device on the Al electrode side. Since the KG experiments showed that photocurrent was sublinear with light intensity, action spectra should be taken at constant-photon flux. Figure 6A was obtained by using a Newport Research Corp. dual-shearing attenuator driven by a stepping motor under computer control. A United Detector Technology, Inc., calibrated photodiode is placed at the sample site while a small portion of the exciting light is diverted to a RCA 7265 phototube. This calibration run provided data so that during the experiment light diverted to the phototube could be used to control the position of the attenuator. For Figure 7, we made use of the fact that the logarithm of light intensity varies linearly with the position of the attenuator.
Figure 1. Configuration of DV-17, drawn to scale. The vertical dimension has been expanded lo4 times the horizontal dimensions.
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on August 31, 2015 | http://pubs.acs.org Publication Date: November 1, 1985 | doi: 10.1021/j100269a013
an effect due to the ambient atmosphere (ii), but the observed effects are rather different from what was anticipated. Experimental Section Ultrahigh vacuum was achieved by using a Varian system with torr) provided by sorption pumps and the rough pumping ( torr) by triode getter-ion and titanium ultrahigh vacuum ( sublimation pumps. The substrates used were Homosil quartz plates (1 in. X 1 in. X 'IL6 in.) obtained from Hereaus Amersil Inc. Before use the cleaning procedure included immersing in concentrated nitric acid, followed by a water rinse, immersing in a 1:1:2 ",OH, H202, H 2 0 solution, followed by a water rinse and blow drying with nitrogen. The contact pads, a 90% indium and 10% silver alloy, and Cu leads were attached to the substrate, which was then placed in the upper portion of a Soxhlet extractor filled with 2-propanol and refluxed overnight. The substrates were removed from the extractor and immediately mounted in the UHV. Aluminum and silver were deposited from a degassed tungsten filament, and the MgTPP from a radiation-shielded quartz crucible. MgTPP was purchased from Strem Chemicals, and degassed over a =30-h period at temperatures below the sublimation temperature, to a base pressure of torr or less. While the initially obtained samples from Strem were relatively pure, some latter batches contained substantial amounts of free base. For our last device the MgTPP was purified by using the following procedure. The MgTPP was separated from the free base by chromatography and recrystallization. The chromatography proceeded via addition of a saturated solution of MgTPP in CH2CI2(=I-2 mL) to a 2 X 15-cm dry-packed column of A1203(Woehlm, Activity Grade I, neutral). Bands corresponding to free base were the first off the column with CH2C12elution, and the column was washed with this solvent until no further color was obtained in the eluant. MgTPP was then collected by elution with acetone. After removal of the acetone, the MgTPP was crystallized from a boiling constant volume mixture of CH2C12and methanol by replacing distilled CH2C12with methanol. The crystals were air dried and showed the characteristic MgTPP optical spectrum. During the film deposition the substrate was mounted horizontally with the vapor stream moving vertically upward. Three masks were used in an arrangement similar to KG. The geometry of a typical device is shown in Figure 1. Film thicknesses were measured with a Leybold-Hereaus Inficon XTM quartz crystal thickness monitor with an accuracy of 0.5% of the total thickness. A density of 1.0 g/cm3 was assumed for the MgTPP. After all vaporizations were complete, the device was rotated to a vertical position, where the following types of in situ measurements were performed in the dark. (1) Capacitance was measured by an ESI 252 impedance meter at 1 kHz. (2) I-V curves were obtained in several manners, with +Vdefined as A1 negative in all figures (i.e., forward bias): (a) for "simple I-V curves", voltages were scanned over a range -Vto +Vwith Vno more than 2.5 V; (b) "hysteresis I-V curves" were obtained by scanning from zero to +V to -V to zero. Voltages were supplied by a D/A converter. and current was measured by a Keithley 417
*
Surface Effects During initial studies, seven devices were made in UHV with configuration: [quartz/Ag(100-350 A)/MgTPP( 1250-3700 A)/ Al( 1 15-600 A)] (2) Device structures with A1 as the top electrode have been used b e f ~ r e . ' ~The ~ ' ~capacitance of devices 2 were very small, all under 8 pF. This contrasts to the 9.8-nF capacitance previously shown by device structure 1 fabricated at low vacuum,' which is in better agreement with the calculated capacitance based on our porphyrin film thicknesses. In response to these low capacitances, we changed to fabricating devices with Al as the layer on the quartz, i.e., device structure 1. However, with one exception (see next section), the devices made with clean A1 showed capacitances of only a few picofarads. We then studied one of these devices carefully to diagnose the problem. DV-17 (device 17) was prepared with the following thicknesses: [quartz/A1(320 A)/MgTPP(2800 A)/Ag(500 A)] (DV-17) This device showed capacitance of only 4.3 pF. From the basic geometry of the device (Figure 1) it can be seen that, in addition to the main triple layer system, there are three regions with single layers [qtz/Al, qtz/Ag, qtz/MgTPP] and two double-layer regions [qtz/Al/MgTPP and qtz/MgTPP/Ag]. Several features of the film regions were observed macroscopically. (1) The Ag layer is continuous electrically from the qtz/Ag to the qtz/MgTPP/Ag regions. It is discontinuous from the qtz/MgTPP/Ag region to the triple layer region. (2) The Ag in the triple layer region is dark and nonmetallic in appearance. (3) In the region qtz/ MgTPP, the porphyrin shows a faint red fluorescence; in the qtz/Al/MgTPP region it shows a strong red fluorescence; it is nonfluorescent in the triple layer region except for a few fluorescent points, presumably flaws. In light of earlier studies,I6 the strong fluorescence is indicative of greater crystallinity. Thirteen scanning electron microscope (SEM) photographs were taken covering five of the six regions of the devices. The following (14) Ashwell, J. J.; Bonham, J. S.; Lyons, L. E. Aust. J . Chem. 1980,33, 1619. (15) Murti, D. K.; Brillson, L. J.; Slowik,J. H.J . VUC.Sci. Technol. 1982, 20,233. (16) Kampas, F. J.; Gouterman, M. J . Lumin. 1978,17, 439.
4952 The Journal of Physical Chemisrry. Vol. 89, No. 23. 1985
Stanbery et al.
-241
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on August 31, 2015 | http://pubs.acs.org Publication Date: November 1, 1985 | doi: 10.1021/j100269a013
0.64r
Fiiurc 2. Scans, ?,, L’ i. t i ~n n i ;ro.hqx phtm,.ivr.lph .I( \I \I$I I)l’ \g rcgwn. shou 1.6 :n lhir 1 hc rn.~~nifi;.,t m 1s Indiidird ~n !he tefi. uhere we rhou 11 h r n i ‘The i\I:#n> V,, the middle term becomes negligible, but the last term does not. In order for both terms to make a negligible contribution, the following inequality must additionally be met: IdRf >> Ve(Rf/Rd) which is equivalent to v,
