Langmuir 2007, 23, 12413-12420
12413
Effect of Pt Particle Size and Distribution on Photoelectrochemical Hydrogen Evolution by p-Si Photocathodes Ilaria Lombardi,*,† Stefano Marchionna,† Giovanni Zangari,‡ and Sergio Pizzini† Department of Materials Science, UniVersity of Milan-Bicocca, Via Cozzi 53, Milan, 20125 Italy, and Department of Materials Science and Engineering, UniVersity of Virginia, CharlottesVille, Virginia 22904 ReceiVed June 1, 2007. In Final Form: September 7, 2007 In this work, we investigate the effect of the average size and density of Pt clusters on silicon on the photoelectrochemical production of hydrogen. The metallization of Si is performed via electroless deposition from aqueous HF solutions and from water-in-oil microemulsions. The first method enables control of the average diameter and density of Pt clusters by properly changing the deposition parameters like HF concentration and immersion times. However, on one hand, size dispersion is relatively wide and particles agglomeration may occur with this deposition technique. On the other hand, Pt islands with smaller dimensions at the nanoscale as well as with a narrower size distribution are deposited from reversed micellar solutions. Photoelectrochemical experiments show that the effect of Pt morphology on photoconversion efficiency strongly depends on light intensity. At low power of illumination (10 mW/cm2), Pt islands with a mean diameter of 100 nm and a density of 15 particles/µm2, which can be obtained via electroless deposition from a HF-based solution, provide the best photoelectrochemical performance. Nevertheless, this configuration of Pt clusters yields an abrupt collapse of photoconversion efficiency from 31% to 11.8% when the light power is increased up to 100 mW/cm2. At this light intensity, Pt islands with a mean size and density of ∼40 nm and 75 particles/µm2, respectively, obtained via the microemulsion method, allow photoconversion efficiency as high as 20% to be achieved.
Introduction The world rate utilization of fossil fuels has dramatically increased in recent decades and it is foreseen that this trend of consumption will not be sustainable indefinitely. The dramatic scenario arising from the shortage of conventional fuels is pushing the scientific community to find alternative energy sources. Exploitation of solar energy represents an appealing option due to its abundance and cleanliness, and significant research efforts worldwide are focused on capturing solar power and converting it into electrical or chemical energy. The photoelectrochemical splitting of water by semiconductors has been known for a long time and opens up interesting opportunities for the creation of a hydrogen-based energy policy.1-3 Among the semiconductor materials, silicon possesses a suitable band gap for efficient sunlight collection and the position of its conduction band is well-located with respect to the electrochemical potential of water reduction; thus, it can be profitably employed as a photocathode.4 Due to the sluggish kinetics of water electrolysis occurring at a semiconductor surface, a discontinuous layer of an electrocatalyst such as Pt is required to accelerate the reaction of hydrogen evolution while maintaining reasonable photovoltages; in addition, it has been found that such layer protects the Si surface from corrosion.5,6 Among the available deposition techniques of metal islands on semiconductors, electroless deposition is highly advantageous * To whom correspondence
[email protected]. † University of Milan-Bicocca. ‡ University of Virginia.
should
be
addressed.
due to its simplicity and low cost. The metallization of Si via electroless coating consists of an electrochemical process in which the anodic and cathodic reactions occur simultaneously on the silicon surface. Metal reduction on silicon is accompanied by silicon oxidation and subsequent etching as a result of the presence of HF which forms stable SiF62-complexes and thus provides electrons for metal ion reduction. The electroless deposition of Pt on p-Si has been investigated in a number of works, even if a systematic study of the morphologies of Pt islands as a function of the process parameters such as reactants concentration and time of deposition is still lacking.7-9 On the other hand, parameters such as the size as well as the spatial distribution of Pt islands critically affect the photoelectrochemical behavior of the Si photocathode, as their presence can strongly influence the overall barrier height, the extension of the depletion layer at the electrode/ electrolyte interface, and the recombination rate of photogenerated carriers. Theoretical models and experimental measurements have shown that the effective barrier height of the composite electrode surface consisting of the combination of low barrier height regions (Pt islands/p-Si Schottky contacts) and high barrier ones (unmetallized p-Si/electrolyte interfaces) greatly depends on the size of the low barrier height areas. The effect of the low barrier phase is gradually reduced as the size of the Pt/Si contacts becomes smaller than the depletion width behind the high barrier areas (∼100 nm in our case).10-12 Ideally, high barrier heights and therefore high photovoltages are predicted for Pt clusters which are 5 nm in size and separated by 20 nm.13 Hence, close control
E-mail:
(1) Gerisher, H. Pure Appl. Chem. 1996, 100, 13061. (2) Heller, A. Science 1984, 223, 1141. (3) Lewis, N. S.; Nocera, D. G. Proc. Nat. Acad. Sci. 2006, 103, 15729. (4) Fahrenbruch, A. L.; Bube, R. H. Fundamentals of Solar Cells: PhotoVoltaic Solar Energy ConVersion; Academic Press: New York, 1983. (5) cccNakato, Y.; Yano, H.; Nishiura, S.; Ueda, T.; Tsubomura, H. J. Electroanal. Chem. 1987, 228, 97. (6) Menezes, S.; Heller, A.; Miller, B. J. Electrochem. Soc. 1980, 127, 1268.
(7) Gorostiza, P.; Servat, J.; Morante, J. R.; Sanz, F. Thin Solid Films 1997, 275, 12. (8) Gorostiza, P.; Diaz, R.; Servat, J.; Sanz, F. J. Electrochem. Soc. 1997, 144, 909. (9) Gorostiza, P.; Allongue, P.; Diaz, R.; Morante, J. R.; Sanz, F. J. Phys. Chem. B 2003, 107, 6454. (10) Tung, R. T. Appl. Phys. Lett. 1991, 58, 2821. (11) Lewis, N. S. Appl. Phys. Lett. 2000, 77, 2698. (12) Lewis, N. S. J. Electroanal. Chem. 2001, 508, 1. (13) Nakato, Y.; Ueda, K.; Yano, H.; Tsubomura, H. J. Phys. Chem. 1988, 92, 2316.
10.1021/la7016165 CCC: $37.00 © 2007 American Chemical Society Published on Web 10/17/2007
12414 Langmuir, Vol. 23, No. 24, 2007
Lombardi et al.
over the size and distribution of Pt islands on Si is essential for any metal deposition process utilized in the fabrication of semiconductor-driven water electrolysis devices. In this work, we intend to gain better insight on the electroless deposition of Pt onto p-type Si to be employed for the fabrication of efficient photocathodes. Specifically, we explore the possibility of tuning the average size and density of Pt particles on Si by properly changing the deposition conditions such as immersion time and HF concentration. Electroless deposition from waterin-oil reversed micelle microemulsions is also employed to achieve tighter control over the dimensions and mutual distances of Pt nanoislands. The photoelectrochemical activity of Ptdecorated photocathodes is also presented and related to the dimensions and spatial distribution of Pt clusters. Experimental Section Single-crystal, p-type Si (Wacker, CZ, (111), ca. 1 Ω cm) substrates are diced into pieces with an area of approximately 1 cm2. Before immersion in the electroless solution, samples are sonicated in isopropyl alcohol for 15 min and subsequently etched in 10% HF (48%) for 10 min to achieve hydrogen termination. The electroless solution for Pt deposition contains 1 mM Na2PtCl6‚6H2O while HF concentration varies in the range of 0.25-1 M. Immersion times range from 2 to 10 min. Depositions are carried out at 25 °C without stirring of the solution. To achieve better control over Pt cluster size, electroless deposition from water-in-oil reversed micelle microemulsions is employed as well. The reversed micelle microemulsions are prepared by mixing a water-based Pt plating solution with n-heptane and a surfactant, sodium bis(2-ethylhexyl)sulfosuccinate (AOT, C20H37O7SNa). The water-based solution consists of 0.5 M HF and 10 mM Na2PtCl6‚6H2O. The AOT/heptane solution is made by dissolving 0.33 M AOT in n-heptane. The two solutions are mixed in appropriate ratios to obtain microemulsions with different R values, where R is the ratio of the molar concentration of H2O to that of AOT in the final microemulsions (R ) [H2O]/[AOT]). This latter parameter defines the size of the micelle which, in turn, determines the size of the deposited metal nanoclusters.14,15 Morphological characterization of the Pt-decorated Si substrates is performed with a scanning electron microscope (SEM, Tescan VEGA TS5136XM) while an image processing program (UTHSCSA ImageTool) is used to carry out the size distribution analysis of the Pt islands. For the electrochemical measurements, an InGa alloy is applied on the wafer back side for the ohmic contact and the sample is then mounted onto a PVC holder which exposes a round area of 0.44 cm2 to the solution. Water photoelectrolysis is carried out in 0.1 M H2SO4 (pH ) 1) using a three electrode cell with platinized Si as the working electrode and saturated calomel electrode (SCE) and Pt as the reference and counter electrodes, respectively. For the photoelectrochemical measurements, the potential of the working electrode is linearly swept at a rate of 20 mV/s from the open circuit potential to -1 V vs SCE and the photocurrent measured (Amel Instruments, Model 2049). To compare the photocurrent density (J) - potential (V) curves for the different photocathodes prepared in this work, we used a tungsten halogen lamp with a light intensity of approximately 10 mW/cm2 (measured with a calibrated thermopile) as illumination source. Photoconversion efficiencies of the photoelectrodes are evaluated at several light intensities using a solar simulator (Oriel Corporation, model 81150) which emits a simulated solar AM1.5 spectrum.
Results and Discussion Effect of the Deposition Conditions on the Morphology of Pt Islands. Electroless Deposition of Pt from Aqueous HF Solutions. The effect of deposition time on the size and distribution (14) Arcoleo, V.; Turco Liveri, V. Chem. Phys. Lett. 1996, 258, 223. (15) Magagnin, L.; Bertani, V.; Cavallotti, P. L.; Maboudian, R.; Carraro, C. Microelectron. Eng. 2002, 64, 479.
Figure 1. SEM images of Pt particles on Si obtained from solution of [Na2PtCl6‚6H2O] ) 1 mM, [HF] ) 0.5 M and a deposition time of (a) 2 min, (b) 6 min, and (c) 10 min. Scale bar is 1 µm. Table 1. Geometric Parameters of Pt Particles Depicted in Figure 1
2 min 6 min 10 min
avg diameter (nm)
diameter STD %
Pt density (counts/µm2)
Pt coverage %
96 105 98
48.3 39 35
4.2 11.1 15.2
3 9.6 11.5
of Pt particles obtained at constant HF concentration (0.5 M) is shown in Figure 1. The main dimensional parameters of the Pt clusters deposited after 2, 6, and 10 min of immersion in the electroless solution are reported in Table 1. The metal islands have an average size of ∼100 nm which remains basically unchanged regardless of the deposition time. The size dispersion is relatively wide as the standard diameter deviations (STDdia) are 48%, 39%, and 35% for immersion times of 2, 6, and 10 min, respectively. Furthermore, the density of the Pt nuclei increases
Photoelectrochemical Hydrogen EVolution by p-Si
Langmuir, Vol. 23, No. 24, 2007 12415 Table 2. Geometric Parameters of Pt Particles Depicted in Figure 2
0.25 M 0.5 M 1M
avg diameter (nm)
diameter STD %
Pt density (counts/µm2)
Pt coverage %
85 105 304
37 39 38
16.1 11.1 0.86
9.1 9.6 5.7
obtained for HF concentrations of 0.25, 0.5, and 1 M. It can be clearly observed that an increase in HF molarity yields an increase of the mean Pt cluster size up to about 300 nm and a simultaneous decrease of particle density. To account for this deposition behavior, we analyzed in more detail the electroless process involved in Pt island formation. As known, the overall deposition process consists of the coupling of two half-cell electrochemical reactions:7,8 on the anodic side, silicon is etched according to the dissolution reaction given by
Si + 6HF + 4hVB f SiF62- + 6H+
(reaction 1)
where, formally, four valence band holes (hVB) are required to drive silicon oxidation at the Si surface. In the cathodic half-cell reaction, Pt ions are reduced at the p-Si surface according to the following reaction:
PtCl62- f Pt + 6Cl- + 4hVB
Figure 2. SEM images of Pt clusters on Si from solutions containing [Na2PtCl6‚6H2O] )1 mM and [HF] equal to (a) 0.25 M, (b) 0.5 M, and (c) 1 M. Deposition time is 6 min. Scale bar is 1 µm.
in time, showing that film growth proceeds mainly through the generation of new nuclei. Progressive nucleation thus plays a major role during Pt growth on Si, in agreement with other reported data.7 The fact that the maximum size of Pt particles is basically unaffected by deposition time is not clear at this stage of research. On the other hand, the narrowing of size distribution of the Pt particles with the increase in deposition time indicates a diffusion control regime for their further growth.16 It is also worth noting that prolonged deposition times favor a higher degree of particle agglomeration, as observed in Figures 1b and 1c. Similar trends are recorded for samples obtained at different HF molarities. Figure 2 compares the morphologies of Pt islands on Si obtained by electroless deposition from solutions containing different concentrations of HF at a fixed deposition time of 6 min. Table 2 summarizes the main dimensional parameters of the Pt clusters (16) Liu, H.; Favier, F.; Ng, K.; Zach, M. P.;. Penner, R. M. Electrochim. Acta 2001, 47, 671.
(reaction 2)
When the two reactions occur at the same electrode, a mixed potential is established during the galvanic displacement process. At this potential, the rate of silicon dissolution equals that of Pt reduction. Figure 3a shows that the increase of HF molarity does not substantially change the mixed potential of the overall reaction as a maximum shift of 20 mV in the anodic direction is recorded. Further information on the deposition mechanism can be inferred by considering the two partial reactions independently. To this end, the anodic process is separated from the cathodic one by using a two-chamber cell in which the cathode (p-Si) is immersed in the cathodic solution (containing water and the Pt salt, without HF) while the anode (p-Si) is in contact with the anodic solution (containing water and HF, without Pt ions). A ceramic, porous diaphragm is utilized to keep the two electrolytes separated. First, it is observed that the open circuit potential of the anode (-0.22 V vs SCE) is not affected by the change of HF concentration in the anolyte, meaning that the use of different HF concentrations does not alter the thermodynamic driving force of the reaction.17 On the other hand, the deposition current recorded after short-circuiting the electrodes in the two-halfcells varies with the amount of the HF present in the system as shown in Figure 3b. Similarly to electrodeposition transients obtained under a constant potential,18,19 the observed current transients consist of a reduction peak followed by the approach to a diffusion-limited current. The occurrence of diffusion regimes associated with the supply of metal ions to the reaction sites have been observed also for electroless systems.7-9 Our system is an electroless one, but can be in part assimilated to the former since growth occurs under an approximately constant driving force. The transients in Figure 3b exhibit a decrease in the peak current imax and an increase in the time at which the peak current is observed, tmax, with increasing HF concentration. It is worth noting that for HF concentrations of 0.25 and 0.5 M (Figure 3b) (17) El-Raghy, S. M.; Abo-Salama, A. A. J. Electrochem. Soc. 1979, 126, 171. (18) Southampton Electrochemistry Group. Instrumental Methods in Electrochemistry; Ellis Horwood: London, 1990. (19) Oskam, G.; Long, J. G.; Natarajan, A.; Searson, P. C. J. Appl. Phys. 1998, 31, 1927.
12416 Langmuir, Vol. 23, No. 24, 2007
Figure 3. (a) Mixed potential of p-Si immersed in the Pt-based solution at different HF concentrations. (b) Short-circuit current densities which are measured after the coupling of the two half-cell electrodes at different HF concentrations.
the current tends to slightly increase again after the peak. This increase in current may be ascribed to the occurrence of the competing reduction reactions of protons or dissolved oxygen at the Pt islands. These reactions become gradually more favorable as the deposition of Pt proceeds and more Pt surface is available, as in the case of low and intermediate HF concentrations.20 The rising part of the transients approximates well the instantaneous nucleation behavior at small [HF], while progressive nucleation describes better the behavior observed with [HF] ) 1 M.19 The transition to progressive nucleation with increasing [HF], which leads to the formation of larger and less dense Pt particles, may be explained in terms of the increased percentage of fluorine termination of the Si surface with increasing HF concentration.21 The Si-F bonds in fact are thermodynamically more stable than the Si-H bonds22 and this could favor the dissolution of Si in already etched areas, leading to the growth of existing Pt nuclei rather than to further nucleation events at higher HF concentration. Additionally, the slow elimination of Si-F bonds generates new (20) Xia, X. H.; Ashruf, C. M. A.; French, P. J.; Kelly, J. J. Chem. Mater. 2000, 12, 1671. (21) Takahagi, C.; Nagai, I.; Ishitani, I.; Kuroda, H. J. Appl. Phys. 1988, 64, 3516. (22) Zhang, X. G. Electrochemistry of Silicon and its Oxide; Kluwer Academic/ Plenum Publishers: New York, 2001.
Lombardi et al.
nucleation sites during the electroless process, inducing progressive nucleation. Electroless Deposition of Pt from ReVersed Micelle Microemulsions. As already mentioned in the Introduction, it is critically important to control the size and the separation of the Pt islands at the electrode surface, given that sparsely distributed ultrafine metal contacts are predicted to provide high effective barriers at the mixed contact surface and consequently high conversion efficiencies.11,12 Although the HF-assisted electroless deposition of Pt on Si allows one to obtain fairly good control over the mean particle size, less control is achieved on particles separation and distribution, especially for long deposition times. Furthermore, it would be desirable to deposit Pt islands with a size in the deep nanometric regime and with a tighter size distribution. The electroless deposition from water-in-oil reversed micelle microemulsions offers a powerful method to fulfill this goal in a simple and inexpensive way; this method has been successfully utilized for example in the deposition of Au nanoclusters on Si.15,23 The water-in-oil system consists of monodisperse spherical droplets of water stabilized by a surfactant, AOT, and dispersed in a continuous alkane phase. The size of metal clusters displaced on Si is found to be determined by the micelle size which, in turn, depends on the microemulsion parameter R. A linear relationship exists between R and the micelle radius.14 Additionally, the stabilizing effect of the aqueous nanodroplets by AOT facilitates the deposition of well-separated metal clusters during the electroless process.23 Figure 4 displays the SEM images of Pt nanoclusters on Si obtained from a microemulsion with a R value of 50 and for immersion times of 1, 10, and 20 min. As shown in Figure 4a, after 1 min of deposition time, the silicon substrate appears uniformly covered with very thin metal islands approximately 35 nm in size even if the scarce light contrast of the SEM micrograph does not allow a more detailed analysis of their size and density. As the deposition time is raised, the metal islands grow in volume and become gradually more distinguishable (see Figures 4b and 4c and the corresponding geometrical parameters in Table 3). This is in agreement with the reaction scheme proposed by Magagnin et al.,15 suggesting that the process starts with the collision and the subsequent pinning of the micelle at silicon surface defects. Pt is displaced on Si from the pinned micelle which, in turn, reintegrates its Pt ions content by colliding and exchanging matter with other micelles from the bulk, without changing its size.14 Thus, the metal cluster on Si will be allowed to grow until it reaches the original micelle volume. The sample obtained after 20 min of immersion in the microemulsion (see Figure 4c) shows a reduced Pt cluster density while several pores are created. This phenomenon is well-known and relates to the formation of porous silicon via metal-assisted chemical etching in HF solutions.24 Interestingly, the size of the pores is strictly related to the dimensions of the Pt clusters as the corrosion mechanism is localized at the micelle sites on the surface. Therefore, it can be presumed that porous silicon layers with pores of controlled dimensions can be fabricated with this method. The partial removal of Pt islands due to the silicon etching at the Si/Pt contacts decreases the particles density, while a slight enlargement of the mean particles diameter is recorded. Finally, long deposition times favor particle agglomeration beside the formation of nanoholes on Si. Photoelectrochemical Measurements. In this section, we present the results of the photoelectrochemical measurements performed on the two sets of electrodes, starting from those (23) Gao, D.; He, R.; Carraro, C.; Howe, R. T.; Yang, P.; Maboudian, R. J. Am. Chem. Soc. 2005, 127, 4574. (24) Li, X.; Bohn, P. W. Appl. Phys. Lett. 2000, 77, 2572.
Photoelectrochemical Hydrogen EVolution by p-Si
Langmuir, Vol. 23, No. 24, 2007 12417
Figure 5. Photocurrent density vs potential for platinized Si obtained at (a) [HF] ) 0.5 M and at 2, 6, and 10 min deposition times (see SEM pictures in Figures 1a-1c); (b) 6 min of deposition times and 0.25, 0.5, and 1 M HF concentration (see SEM pictures in Figures 2a-2c). Illumination is performed with a tungsten halogen lamp at ∼10 mW/cm2.
Figure 4. SEM pictures of Pt particles on Si deposited from a microemulsion with R ) 50 and immersion times of (a) 1 min, (b) 10 min, and (c) 20 min. Scale bar is 500 nm. Table 3. Geometric Parameters of Pt Particles Depicted in Figure 4
10 min 20 min
avg diameter (nm)
diameter STD %
Pt density (counts/µm2)
Pt coverage %
38 47
25 21
75 34
8.5 5.9
obtained by electroless deposition in aqueous HF solutions. Figures 5a and 5b show the J-V characteristics under illumination with a tungsten halogen lamp of Pt-activated Si photocathodes prepared at constant HF concentration and different immersion times (see SEM images in Figure 1), and for samples obtained at different HF concentrations and fixed deposition time (see SEM images in Figure 2), respectively. As a general characteristic, it can be observed that most of the S-shaped curve resides at
potential more positive than the reversible H+/H2 electrode potential (RHE), shown on the graphs as a vertical dotted line. The best photoelectrode performance shows an open circuit photopotential (Eaoc) ∼ 400 mV more positive than RHE, similarly to other reported results.25 The increase of deposition time, Eaoc, as well as the occurrence of hydrogen evolution reaction (HER) shifts to more anodic values thus implies that higher photovoltages are attained by increasing the density of Pt particles whose mean diameter is roughly the same. However, prolonged deposition times, though beneficial for establishing higher photovoltages, cause a simultaneous decrease in saturation photocurrents, as demonstrated by the sample obtained after 10 min of deposition in the electroless solution. This is largely expected since the increase of metal coverage lowers the illuminated Si area and, consequently, the amount of photogenerated carriers. Additionally, carrier recombination phenomena may be accelerated with aggregation of Pt (25) Nakato, Y.; Yano, H.; Nishiura, S.; Ueda, T.; Tsubomura, H. J. Electroanal. Chem. 1987, 228, 97. (26) Maier, C. U.; Specht, M.; Bilger, G. Int. J. Hydrogen Energy 1996, 21, 859.
12418 Langmuir, Vol. 23, No. 24, 2007
Lombardi et al.
particles and enlargement of the Pt/Si contact areas, thus causing a reduction of the observed photocurrents.27 As for the samples obtained at different HF concentrations, higher open circuit potentials are measured under illumination for smaller and denser Pt particles (lower and intermediate HF concentrations) while the Pt morphology characterized by large and less dense islands (higher HF concentration) gives lower Eaoc. Moreover, it can be clearly observed that the sample with larger Pt islands (obtained at [HF] ) 1 M) has a saturation current density which is lower than that expected from its Pt coverage. In fact, if we compare for instance the sample of Figure 2b with that of Figure 2c, we find that despite the latter having a remarkably smaller Pt coverage, its saturation photocurrent is approximately 12% lower than that of the former. Hence, it is confirmed that recombination of photogenerated carriers is sensitive to the size of the Pt clusters, and the greater their mean diameter, the greater the tendency of the system to recombine carriers and reduce the photocurrent and photovoltage. It should however be pointed out that the actual width of the Pt-Si contacts is generally lower than the mean diameter of the Pt particles prepared in this work, as the effective contact area between Pt and Si may present undercuts (see Figure 2 in ref 8). On the other hand, the voltage drops due to irreversibilities and ohmic losses, determined from the slopes of the J-V curves, tend to decrease with the simultaneous decrease of the Pt particle density and increase of the average particle size. Therefore, the optimal morphology of Pt particles on Si able to perform the photoelectrochemical splitting of water in an efficient manner should balance opposing factors: on one side, small and dense Pt islands provide higher Eaoc but voltage drops tend to increase as the number of contacts is raised; on the other side, larger and less dense Pt clusters split water more readily (i.e., with smaller overvoltages) but carriers recombination reduces the attainable photocurrent. The different photoelectrochemical behavior of the electrodes investigated so far under simulated solar radiation at various light intensities may be better visualized in terms of photoconversion efficiency. The photoconversion efficiency η of light energy to chemical energy in the presence of an external applied potential is calculated according to the following equation:28
η (%) ) jp
(E0rev - |Eapp|) × 100 I0
positive constant. For relatively strong illuminations such that bI0 . 1, eq 1 can be reduced to
where jp is the photocurrent density, E0rev is the standard reversible potential of the water splitting reaction (1.23 V vs NHE), I0 is the intensity of incident light (power density), and |Eapp| is the absolute value of the applied potential. This latter value is given by Eapp ) (Eaoc - Emeas), Emeas being the potential at which jp is measured while Eaoc is the open circuit potential of the working electrode under the same illumination. For a given sample, the increase of light intensity raises the Eaoc values in a logarithmic fashion as displayed in Figure 6a. The increment in Eaoc with power illumination may be estimated by the following equation,22
∆Eaoc )
(
)
kT 1 + bI02 ln e 1 + bI01
Figure 6. (a) Experimental and calculated values of open circuit potential Eaoc at different light intensities. (b) Variation of the saturation current density with light intensity. Data are referred to sample of Figure 1b.
(1)
where ∆Eaoc is the gain in open circuit photopotential which is produced by increasing light intensity from I01 to I02 and b is a (27) Yae, S.; Nakanishi, I.; Nakato, Y.; Toshima, N.; Mori, H. J. Electrochem. Soc. 1994, 141, 3077. (28) Khan, S. U. M.; Al-Shahry, M.; Ingler, W. B. Science 2002, 297, 2243.
∆Eaoc )
()
kT I02 ln e I01
(2)
Equation 2 is used to calculate the shifts in Eaoc with light power using the experimental value at 125 mW/cm2 as reference point. As shown in Figure 6a, the theoretical data fit very well the experimental points for high values of light intensities (down to 50 mW/cm2) while lower light intensities worsen the agreement between the measured and calculated values. This is not surprising since the validity of eq 2 is limited to the case of strong illuminations. On the other hand, the variation of the saturation current density with power of illumination shows a linear relationship, as displayed in Figure 6b. As known, the increase of light intensity induces a greater tendency for the systems to recombine carriers as band bending is progressively reduced and the electric field separating electron-hole couples across the space charge layer becomes smaller.29 Consequently, η diminishes with increasing light power density as shown in Figure 7a. (29) Lewis, N. S. Inorg. Chem., 2005 44, 6900.
Photoelectrochemical Hydrogen EVolution by p-Si
Langmuir, Vol. 23, No. 24, 2007 12419
Figure 7. (a) Values of photoconversion efficiency as a function of light intensity and electrode measured potential. Data are referred to sample of Figure 1b. (b) Results of photoconversion efficiency at 10 mW/cm2 for various samples depicted in Figures 1 and 2.
Figure 8. (a) Photocurrent density vs potential of samples obtained from microemulsions with R ) 50 and different deposition times (see SEM pictures of Figure 4). (b) Comparisons of photoconversion efficiencies between Si sample of Figure 1c and that of Figure 4b at 10 and 100 mW/cm2 of light power.
Furthermore, the curve peak at which the maximum efficiency is located shifts in the anodic direction by lowering illumination power, thus implying that the HER reaction is favored at low light intensities. The graph reported in Figure 7b shows the η values of some samples whose SEM images are displayed in Figures 1 and 2 at an illumination of 10 mW/cm2. The best performance is that of the electrode b of Figure 1c for which a maximum photoconversion efficiency of 31% is recorded at approximately -0.3 V vs SCE. More interestingly, high photoconversion efficiencies are recorded at potentials significantly more positive than RHE, thus enabling the use of these Pt-decorated silicon electrodes as efficient photocathodes for hydrogen production. Figure 8 shows the J-V characteristics of p-Si electrodes on which Pt nanoislands are deposited via the microemulsion method with R ) 50 (see SEM pictures of Figure 4). The sample obtained after 10 min of deposition is that with the highest photovoltage, with an Eaoc value of 300 mV more anodic than RHE. Conversely, Pt nanodots formed after 1 min of immersion have poor catalytic activity toward HER, likely due to the small amount of deposited Pt. Similarly, the J-V performance of the sample of Figure 4c obtained after 20 min of deposition and characterized by less
dense Pt nanoislands and nanoholes is worse than that fabricated after 10 min of immersion. To further reduce the Pt clusters mean diameter, we used microemulsions with a lower R value. However, samples obtained with R ) 20 and different deposition times show negligible, often no, photocurrent at potentials more positive than NHE. To establish which kind of morphology of the Pt islands is to be preferred for our purposes, we have compared the photoconversion efficiencies of electrodes obtained with the two deposition techniques investigated in this work. In particular, the behavior of the sample of Figure 1c (providing the best performance with electroless deposition from aqueous HF solutions) is compared with that of the sample of Figure 4b (providing the best performance with the electroless deposition from micellar solutions). Results are displayed in Figure 8b. It is interesting to notice that highly dense and fine Pt islands obtained with the microemulsion method (Figure 4b) tend to improve the electrode photoconversion efficiency η as the light intensity is increased. Specifically, the deposition of uniformly distributed, nanometric islands with controlled size (∼40 nm) allows for sensitive mitigation of the worsening in η which is recorded upon increase of light intensities for electrodes
12420 Langmuir, Vol. 23, No. 24, 2007
characterized by larger (∼100 nm) and less uniform Pt particle sizes (sample of Figure 1c). Hence, we demonstrated that the effect of the size and distribution of Pt clusters on the photoconversion efficiency of p-Si varies with the intensity of the light striking the electrode. The use of small, well-controlled Pt nanoislands on Si, which can be obtained with the microemulsion method, enables reduction of the recombination losses at high power of photons flux. On the other hand, when less control over the Pt particle size and distribution is achieved, the electrode rapidly degrades its photoelectrochemical performance with the increase of the photons flux, as recombination of photogenerated carriers tends to become dominant.
Conclusions In this work, the photoelectrochemical activity of p-Si photocathodes is investigated as a function of the size and distribution of Pt catalytic islands displaced on Si via electroless deposition from aqueous and reversed micellar solutions. The main conclusions can be summarized as follows. First, the average size and density of the Pt particles can be tuned by properly choosing the HF concentration and deposition times from aqueous solutions. The use of water-in-oil microemulsions enables the deposition of Pt islands with enhanced monodispersity as well
Lombardi et al.
as smaller dimensions on a nanometric scale. Second, higher photovoltages are obtained by increasing the density of the Pt particles even if voltage drops tend to augment with the number density of the Pt/Si contacts. On the other hand, the enlargement of the average size of Pt islands raises recombination losses. Finally, the effect of the size and distribution of the Pt particles on photoconversion efficiency is found to be strongly dependent on the power of illumination. In particular, the increase of light intensity degrades the photoconversion efficiency of the electrodes. The use of fine (∼40 nm), dense (∼75 µm-2), and welldispersed Pt islands on Si is crucial for reducing recombination losses at high power of illumination (100 mW/cm2), while larger (∼100 nm) and less dense (∼15 µm-2) Pt particles provide better photoelectrochemical performances of the silicon photocathodes at low light intensities (10 mW/cm2) without the need for tight control over particles size dispersion and interparticle distances. Acknowledgment. This work was carried out in the framework of the Project “Development of nanostructured electrodes for the production of hydrogen via the photoelectrochemical splitting of water” sponsored and granted by Fondazione Cariplo, Milano. LA7016165
