Langmuir 1985,1,605-607 cules/surface MgO moiety. This suggests that the active sites are catalytic and that the phosphorus residues are mobile and move away from the active site so as not to poison it, and thereby must move into the bulk, or the MgO matrix is continually disrupted and fresh active sites are exposed (a stoichiometric reaction). These decomposition reactions are essentially stoichiometric, and in the most favorable cases about one phosphorus molecule is decomposed for every two MgO molecules present in the bulk or on the surface.
Experimental Section Materials. MgO and CaO were purchased from ROC/RIC and are listed as 99.9949999% pure. The procedures for washing, drying, and thermally activating have been described previously! Carbon monoxide (Linde, 99.999%) and deuterium gas (Stohler, 99.7%) were passed through a -196 OC trap just prior to adsorption. Deuterium oxide (Merck, 99.8%) was freeze-thaw d e g d before use. Phosphorus compounds were purchased from Strem Chemical Co. and were freeze-thaw degassed just before use. IR Spectroscopy. A Perkin-Elmer 1330 equipped with a CDS 0.d.) were Data Station 1300was used. The MgO pellets (15" prepared from dry, non-activated MgO at 2000 psig. They were placed in a special cell described earlieru where they were activated under vacuum at 700 OC for 14 h. After it was cooled to room temperature the pellet was exposed to several torr of pressure of the desired phosphorus compounds for several minutes. The torr, excess phosphorus compound was pumped out down to the pellet guided into the NaCl window section of the apparatus, and IR spectra taken. (15) Matauo, K.; Klabunde, K. J. J. Org. Chem. 1982,47, 843-848.
605
GC-MSAnalyses. A Finnigan 4021 GC-MS system with an INCOS data system and Finnigan 9610 gas chromatograph was used. Volatile compounds from phosphorus compound decompositions were separated on a l/* in. X 25 f t stainless steel column packed with 20% bis(Bmethoxyethy1) adipate on Chromasorb P-AW SO/€@at 40 OC and 25 mL/min helium gas flow rate. Mass spectra were taken at 20-70 eV where the optimum was selected to maximize the molecular ion for each compound. Elemental Analyses. The (CH3CH20)3P/Mg0residue was analyzed after baking at 220 OC for 2 h and vacuum removal of volatiles. Anal. Calcd for CzH6P03(Mg0)l.9:C, 13.01; H, 2.73; P, 16.78; Mg, 25.02. Found C, 12.89;H, 2.58; P, 16.91;Mg, 25.22. Adsorption of Phosphorus Compounds. Thermally activated oxide (1.0 g) was exposed to a known amount (PV measurements) of phosphorus compound until the pressure in the closed system (of known volume) was constant for 30 min. The amount adsorbed was then calculated by PV measurements. Decomposition of Phosphorus Compounds. In these experiments excess phosphorus compound was condensed onto a 1.0-g MgO sample, and the sample was isolated from vacuum and warmed to the desired temperature for a prescribed time (usually 2 h). Afterward the sample was cooled to -78 OC, and volatiles were vacuum evaporated and trapped in a -196 OC trap. They were diluted with p-xylene with cyclopentane as an internal standard and analyzed by GC and GC-MS as described above. Acknowledgment. This work was supported by the Army Research Office under Grant DAAG29-84-K-0051. Registry No. MgO, 1309-48-4; CaO, 1305-78-8; (CH3CH2O),P=O, 78-40-0; (CH3CH2)3P,554-70-1; (CH,O),P, 121-45-9; (CHsCH20)3P,122-52-1;(CH&H2CH20)3P,923-99-9; ((CH&CHO)3P, 116-17-6;(C1CH2CH20),P,140-08-9;(CF3CH20),P,37069-4; (CBHb)(CH3CH20)2P, 1638-86-4; (CH&H2)2(CH3O)P, 13506-71-3.
Photoluminescent Response of Palladium-Cadmium Sulfide and Palladium-Graded Cadmium Sulfoselenide Schottky Diodes to Molecular Hydrogen Michael K. Carpenter, Hal Van Ryswyk, and Arthur B. Ellis* Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin 53706 Received March 27, 1985 The bulk photoluminescence (PL) of Schottky diodes constructed from P d and n-type CdS (Pd-CdS) and from n-type, graded CdS,Sel, (0 Ix I1;2: = 1at surface) (Pd-CdS,Sel,) is sensitive to molecular Hz. For the Pd-CdS diode, exposure to Hz (31, N2/Hz mixture) significantly enhances the P L intensity 510 nm) relative to the intensity in air; by use of a dead-layer model, the of edge emission (Amm corresponding reductions in depletion width and Schottky barrier height can be estimated. For the Pd-CdS,Sel, diode, Hz changes the spectral distribution; P L from this material, color coded to spatially resolve e--h+ pair recombination, indicates the depth over which the electric field is reduced in the semiconductor. These phenomena demonstrate optically coupled chemical sensing of hydrogen.
-
Introduction The bulk photoluminescence (PL) of semiconductorshas proven to be a useful probe of Schottky barrier characteristics in both semiconductor/metal diodes' and photoelectrochemicalcells (PEGS)? In particular, the electric field in the semiconductor can be estimated from PL intensity using a dead-layer model: Electron-hole (e--h+) (1) Hollingsworth, R. E.;Sites, J. R. J.Appl. Phys. 1982,53,6367 and references therein. (2) Hobeon, W. S.; Ellis, A. B. J. Appl. Phys. 1983, 54, 6956 and references therein.
0743-7463/85/2401-0605$01.50/0
pairs formed within a distance from the interface on the order of the depletion width do not contribute to PL. While the importance of surface properties on bulk PL has been described?~~ their interrelationship has not been examined experimentally in detail. I n choosing a system for such studies, our interest was drawn to Schottky diodes constructed with Pd because of the known sensitivity of their current-voltage (i-V) properties to gaseous H,.In general, Schottky barriers (3) Mettler, K. Appl. Phys. 1977, 12, 75. (4) Stephens, R. B. Phys. Rev. B. 1984,29, 3283.
0 1985 American Chemical Society
Carpenter, V a n Ryswyk, and Ellis
606 Langmuir, Vol. 1, No. 5, 1985
resulting from junctions of P d with a variety of n-type semiconductors decreased upon exposure of the diode to Hz, reflecting diminution of the P d work function by the gas.g The effect is sufficiently sensitive to Hz concentration that Pd-CdS5 and Pd-TiOz6 Schottky diodes have been proposed as Hz detectors. Variation in Schottky barrier height with Hz appears to also strongly influence the efficiency with which metal-coated semiconductor electrodes evolve Hz in PEC's.l0 We report herein that P L intensity from Pd-CdS Schottky diodes is substantially perturbed by exposure of the diode to H,. Moreover, the P L properties are consistent with a dead-layer model, allowing calculation of the variation in electric field thickness with exposure to H,, Schottky diodes constructed from Pd and a graded CdS,Sel-x (0 Ix I 1; x = 1 at surface) semiconductorl1J2 (Pd-CdS,Se,,) evince a change in spectral distribution upon exposure to H,; the color-coded nature of the light emitted from this semiconductor can be used to map the effective electric field in this solid and its perturbation by H p Besides illustrating the coupling of surface interactions to bulk PL, these studies demonstrate the feasibility of constructing optically coupled chemical sensors.
Experimental Section Samples of n-type, single-crystal,CdS and CdSe c-plates (1-mm thickness; - 2 0 cm resistivity) were purchased from Cleveland Crystals, Inc., and cut into pieces of -0.25-cm2 area. Graded samples of n-CdS,Sel.., (0 Ix 5 1)where the graded zone had a thickness of 1.0 gm were prepared by vapor-phase diffusion of S into CdSe and characterized as described previous1y.l2 Prior to Pd deposition on its 0001 Cd-rich face, CdS was etched in Br2/MeOH (1:lO v/v); the graded CdS,Se,, samples were not etched owing to the thinness of the graded layer. Deposition utilized Pd foil (50 X 50 X 0.1 mm; >99.997% metallic purity; Aesar Co., Seabrook,NH) and a SPI Super Sputter apparatus. Sputtering was conducted at 2 X lo4 torr Ar pressure and 60-/LAbeam current for -45 s; a parallel deposition onto Pt foil was used in conjunction with electrochemical stripping (0.75 V vs. SCE in 1 M HC1 aqueous electrolyte) to estimate the Pd layer thickness to be -100 A. After deposition, electrical contact was made to the translucent Pd layer with Ag epoxy and to the back surface with Ga/In eutectic and Ag epoxy; Cu wires were connected to the Ag epoxy and current-voltage properties obtained with a PAR Model 173 potentiostat and Model 175 programmer. PL spectra were recorded using 457.9- and 488.0-nm excitation from a CR-12 Ar+ laser and an Aminco-Bowman spectrometer equipped with a Hamamatsu R446S PMT. The sample was enclosed in a cell which permitted dry N2,a 3:l N2/H2mixture (Air Products tank of H, mixed with in-house Nz), or air to bathe the sample; flow rates of -0.5 L/min were employed. Optical properties of the Pd film were examined by depositing the metal on a microscope slide; transmission and reflectivity of 457.9-, 488.0-, and 514.5-nm light were measured in air and the N2/H2 medium.
-
Results and Discussion The P L spectrum from a Pd-CdS sample in air, illuminated through the metal with 457.9-nm ultraband gap
tI
1
1
q J
g 1':
I
550 WAVELENGTH /nm
500
Figure 1. Uncorrected PL spectra of a Pd-CdS Schottky diode
in air (solid line) and in a 3:1, N2/Hzatmosphere (dashed line). The sample w a excited with the same intensity (-40 mW/cm2) of 457.9-nm light in both experiments, using an identical sample-detection optics geometry.
-
(E, 2.4 eV) light, is characterized by band edge emission at -510 nm,13 as shown in Figure 1. When a 3:l mixture of Nz/Hz is passed over the sample, Figure 1reveals that the P L intensity is enhanced by approximately 70%; the enhancement occurs over -30 s with the -100-A-thick P d layer. P L enhancement requires the presence of P d but does not appear to derive from changes in the metal's optical properties: optical transmission and reflectivity of the metal film, deposited on a microscope slide, were insensitive to the presence of H2 After flushing the sample cell with Nzand then with air, the PL intensity returns to its original value. This effect is reversible over at least 10 cycles. If more penetrating 488.0-nm light is used for excitation, the PL intensity is augmented by -40% with exposure to H2. From current-voltage data, the Pd-CdS structure exhibits typical diode behavior in air and in the N2/Hz medium. These spectral changes are consistent with a reduction in Schottky barrier height resulting from the dissolution of Hz in Pd, a well-studied phenomenon."1° Qualitatively, the PL intensity is expected to rise because the smaller electric field in the semiconductor allows a larger fraction of e--h+ pairs to radiatively recombine. By regarding the region supporting the electric field as being completely nonemissive, i.e., a dead layer, a quantitative expression for relative PL intensity can be obtained, eq 1; this $Pair/$Hz
= exp(-a'D)
(1)
447. ..
treatment assumes that the CdS surface recombination velocity in the diode is either very large or insensitive to H223 In eq 1,&% and $& are the radiative quantum yields in H2 and in air, D is the difference in dead-layer thickness between the two media, and a' = (a+ 8) with a and /3 the solid's absorptivities for the exciting and emitted light, respectively. For CdS, a for E l c polarized light is 6 X lo4 and -(9-10) X lo4 cm-l for 488.0- and
(9)Dus,R. Surf. Sci. 1973, 42, 324. (10) Heller, A. Science (Washington,D.C.) 1984, 223, 1141. (11) Streckert, H. H.; Ellis, A. B. J.Phys. Chem. 1982,86, 4921. (12) Carpenter, M. K.; Streckert, H. H.; Ellis, A. B. J. Solid State Chem. 1982, 45, 51.
(13) Streckert, H. H.; Tong, J.; Carpenter, M. K.; Ellis, A. B. J. Electrochem. SOC.1982, 129, 772.
(5) Steele, M. C.; MacIver, B. A. Appl. Phys. Lett. 1976, 28, 687. (6) Yamamoto, N.; Tonomura, S.;Matsuoka, T.; Tsubomura, H. Surf. Sci. - .. 1980. -. .. 92. 400. --I
- - 7
(7) Yamamoto, N.; Tonomura, S.; Matauoka, T.; Tsubomura, H. J. Appl. Phys. 1981,52,6227. (8) Lundstram, I,;Shivaraman, M. S.; Svensson, C. Surf. Sci. 1977,64,
__
Langmuir, Vol. 1, No. 5, 1985 607
Photoluminescent Response of Schottky Diodes
457.9-nmlight, respectively; /3 is 7 X 103 cm-' at 510 nm.14 The P L enhancements seen for the two excitation wavelengths employed give a consistent value for ALI of 500-600 A. This value for the contraction of the electric field upon exposure to H2 can be used to calculate a reduction in Schottky barrier height if the initial height is known. Literature estimates for the Pd-CdS barrier height in air range from -0.5 to 0.8 eV;597our i-V curves yield an estimated height of 0.6 eV. The depletion width W is related to barrier height qV by eq 2, where eo is the per-
-
w = (2€€ov/qN~)'/~
(2)
mittivity of free space, q is the electronic charge, and t and ND are the semiconductor's dielectric constant and charge carrier density; for our samples, e and ND are -10 and 9 X 1015 ~ m - ~ respectively.16 , Substitution into eq 2, equating D with W, leads to an estimated decline in barrier height of 0.2 eV. Literature values vary from ~ 0 . to 5~ -0.2 eV (extrapolated).' The range of values likely refleds variations in sample preparation.16 A complementary system with H2-sensitive P L is the Pd-CdS,Sel, Schottky diode. The P L spectrum in air is shown in Figure 2 and consists of edge emission from all of the CdS,Sel-, compositions which comprise the 1.0pm-thick graded zone, from CdS at the surface to the CdSe (Eg 1.7 eV;,,A 720 nm) substrate.12 A linear correlation exists between composition x and the emission band maximum, eq 3.13 In conjunction with AES/Ar+ X,,(nm) = 718 - 210x (3) N
-
-
sputter etch data, eq 3 provides a map of radiative recombination in the solid The PL is color coded to indicate the depth from the surface a t which e--h+ pair recombination occurs. Perturbation of P L by applied potential reflects changes in the effective electric field (EEF) in the solid;"J2 we use the term "EEF" to reflect the fact that the electric field of this solid is complex and contains contributions, e.g., from band-edge gradients, in addition to the field arising from the Schottky barrier. Exposure of a Pd-CdS,Sel, diode to H2 results in asymmetric enhancement of as much as 50% in the blue (14) Dutton, D.Phys. Rev. 1968,112, 785. (15) The value for e waa obtained from: Berlincourt, D.; Jaffe, H.; Shiozawa, L. R. Phys.Reu. 1963,129,1009. For ND,a mobility value of -350 cmx(V 8 ) waa used. See: Devlin, S. S. In "Physicsand Chemistry of II-VI ompounds"; Avert, M., Prener, J. S., Eds.; North-Holland Publishing Co.: Amsterdam, 1967; Chapter 11. (16) Aspnes, D.E.;Heller, A. J . Phys. Chem. 1983,87,4919.
I
500
600
WAVELENGTH I nm
I
I
700
BO(
Figure 2. Uncorrected PL spectra of a Pd-CdS,Sel, Schottky diode in air (solid line) and in a 3:1, Nz/H2 atmosphere (dashed line). The sample was excited with the same intensity (-40 mW/cmz) of 457.9nm light in both experiments, using an identical sample-detection optics geometry.
end of the PL spectrum, Figure 2; the Gffect corresponds to a modest color change. The material has diode i-V characteristics in both environments. Qualitatively, the P L enhancement a t shorter wavelengths indicates a reduction in the EEF in the near-surface region of the semiconductor, since it is the near-surface, S-rich compwitions that give rise to the emission. A more quantitative estimate of the affected region is afforded by the cessation of spectral changes for X 1 600 nm. From eq 3 and AES/Ar+ sputter etch data, exposure to H2 influences the EEF to a depth of -0.1 pm (1000 A) from the surface. In summary, bulk P L from Pd-CdS and Pd-CdS,Sel-, Schottky diodes provides a sensitive probe of changes in the electric field of the semiconductors resulting from a surface interaction with H2, The ability to transform molecular surface interactions into a change in bulk PL intensity (Pd-CdS) or color (Pd-CdS,Sel,) could have applications to the design of optically coupled chemical sensors. Further experiments designed to couple analyte sensitivity to PL are under way in our laboratories.
Acknowledgment. We are grateful to the Office of Naval Research, the Army Research Office, and the 3M Co. for generous support of this research. R. K. No11 and J. Jacobs are thanked for their assistance with the Pd deposition process. Dr. Nils Blom is acknowledged for experimental assistance and helpful discussions. Registry No. CdS, 1306-23-6;Hz, 1333-74-0;Pd, 7440-05.3.
