- Email: [email protected]
Photoelectrochemistry on a planar, interdigitated electrochemical cell
Electrochimica Acta 56 (2011) 8752–8757
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Photoelectrochemistry on a planar, interdigitated electrochemical cell M. Neumann-Spallart Groupe d’Études de la Matière Condensée, C.N.R.S. and Université de Versailles St. Quentin, 1, place Aristide Briand, 92195 Meudon CEDEX, France
a r t i c l e
i n f o
Article history: Received 5 April 2011 Received in revised form 20 July 2011 Accepted 20 July 2011 Available online 28 July 2011 Keywords: Interdigitated electrode Photoelectrocatalysis Titanium dioxide
a b s t r a c t The construction of a planar electrochemical cell with two interdigitated electrodes, with a finger spacing of 25–100 m and 200–50 pairs of fingers and gaps, respectively, is described. The working electrode consists of n-type semiconducting titanium dioxide (anatase). Under UV light, photocurrents are produced. No reduction in the photocurrent due to iR drop was observed, even in electrolytes of low ionic strength. This makes the device an interesting candidate for photoelectrocatalytic purification of drinking water. An example is shown in the degradation of the azo-dye AO7 (acid orange 7). k(photoelectrocatalysis) on a device with 50 finger pairs was 3.29 × 10−6 /s and the corresponding kinetic parameter, p = kFV/iphoto (rate constant normalized to unit volume and photocurrent), was 404 M−1 . © 2011 Elsevier Ltd. All rights reserved.
1. Introduction The removal of organic material of biological and industrial origin (bacteria, hormones, endocrine disruptors, textile dyes, dioxines, etc.) from waste- and drinking water is of enormous importance nowadays. There is a great need for new technologies serving this purpose. One way could be photocatalysis, involving semiconducting oxides under illumination. In a heterogeneous photocatalytic process, a solid semiconductor catalyst such as titanium dioxide or other transition metal oxides irradiated with near UV light generate strongly oxidizing species (valence band holes, ·OH radicals) which can destroy organic species present in the aqueous phase in contact with the surface of the catalyst. However, there are limitations to the efficiency of the process that are partly due to the semiconductor and partly due to various recombination processes and the technical outline of the reactor in which the reaction is carried out. Early work [1] showed how photocatalysis in semiconductor particulate suspensions can be understood on the basis of photoelectrochemical analysis of macroscopic electrodes. From this analysis it became also clear that the main obstacle to efficient use of absorbed photons is the recombination of photogenerated charge carriers, and that this can be suppressed by applying electrical bias to the semiconducting photocatalyst which must be immobilized on an electrically conducting substrate. This has also the advantage of making separation of the catalyst from the reac-
E-mail address: [email protected] 0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2011.07.089
tion mixture (the liquid phase) superfluous, once the reaction is over. Recent work on organic impurity degradation using TiO2 [2–6] and WO3 thin film electrodes [7–9] validated these ideas. In an electrochemical cell, iR drop is one of the factors limiting high current throughput at moderate bias. Especially in electrochemical treatment of additives (pollutants) in drinking water, having low ionic strength, where addition of supporting electrolyte is prohibited for obvious reasons, ways must be devised for limiting iR drop. (Typical tap water conductivities of 528 S/cm were observed in our laboratory.) One way is to use a parallel plate reactor with two opposite electrodes and a small space between them where the electrolyte is passed through. For practical reasons (if the electrodes are large) this distance may not fall below 1 mm, if energy investment for pumping is to be minimized. Such a reactor with 10 cm × 10 cm plates and a 1 mm gap, with a conductivity cell constant, , of 0.001 cm−1 was recently demonstrated [2], and had an electrolyte resistance for tap water (500 S/cm) of 2 , leading to an iR drop of only 40 mV for current densities of around 3 × 10−4 A/cm2 as observed for photocurrents on TiO2 produced by solar irradiation. However, as the pressure build-up is considerable in a module consisting of many such cells, another way is suggested in the present work: both electrodes are positioned in the same plane so that no flow restrictions apply. However, when large currents are to be passed through a planar device, it must be taken into account that the conductivity cell constant of such an arrangement is considerable higher than that of an equivalent arrangement where the electrodes face each other. In order to ensure low iR drop between the electrodes, they are arranged in an interdigitated way. Electrochemical cells with various planar interdigitated arrangements have been used for redox sensing and other electroanalytical
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
8753
applications [10–13]. Such structures can lead to very low cell constants if a large number of thin fingers and small spacing are used [14].
2. Experimental The schematics of a typical planar device with an interdigitated part of 0.5 cm × 1.5 cm are shown in Fig. 1. Fingers extending from the left side towards the centre consist of photoactive TiO2 deposited on platinum, whereas the finger family starting on the right side is the Pt counter electrode. Electric contacts were taken at the bottom to which the Pt collector areas are extended. The devices were fabricated by standard lithographic techniques using contact masks. The masks were prepared in the following way: the pattern was drawn as vector graphics on computer file using the program “Adobe Illustrator”. The patterns were then transferred either onto transparent plastic sheets using a laser printer, or directly onto photographic film. As these foils were subsequentially used as contact masks in the lithographic processes, high accuracy combined with mechanical stability for repeated use was sought, giving preference to patterns on photographic film. An example is given in the optical microscope pictures in Fig. 2 with details of the pair of masks used for the “100–25” device, i.e., a device with 100 m wide Pt and TiO2 fingers and a 25 m gap between the finger families: a mask used for Pt deposition (a) and a mask used during TiO2 deposition (b). The optical pictures show the excellent sharpness of the lithographic masks. Masks with different finger widths (100, 50, 25 m) and interfinger distances (100, 50, 25 m) were prepared, with a finger length of 5 mm. Devices with patterns deposited over a length of 15 mm and equal finger spacing of 25–100 m led to 200–50 pairs of fingers, respectively, and the same number of pairs of gaps (insulating parts made of SiO2 between the finger families), covering a total operating area of 0.75 cm2 , of which 0.1875 cm2 (one fourth) was rendered photoactive due to coverage with titanium dioxide.
Fig. 1. Schematics of an interdigitated device (in scale: interdigitated area 1.5 cm × 0.5 cm). Black part – insulating area (SiO2 ), light grey part – titanium dioxide, white part – Pt. Bottom left and right – area for applying external contacts.
Fig. 2. Inspection of the emulsion side of developed photographic film containing the masks. Details (optical microscope) of a pair of masks used for depositing a “100–25” device (100 m wide Pt and TiO2 fingers and 25 m gap between the finger families). Left – mask used for Pt deposition, right – mask used for revealing TiO2 .
8754
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
Fig. 3. Profilometric scan of a “100–100” device (100 m wide Pt and 5 + 100 + 5 m TiO2 fingers and 100 m gap between the finger families).
The substrates for the devices were 2 in. silicon wafers with a thick (2 m) insulating layer of silicon dioxide. A thin (7 nm) titanium layer, serving as an adhesion promoter, and a thicker (200 nm) platinum layer were then deposited by evaporation on a pre-patterned positive photoresist (using the first mask of a set of two – the “Pt-mask”). The metals were patterned by removing the pre-patterned photoresist, leaving an interdigitated arrangement consisting of both electrodes with their pads. The next step consisted of over-coating the whole wafer by a 100 nm thick layer of titanium dioxide by spin-coating of an organo-titanate precursor, followed by calcination. Details of this step were given elsewhere [15,16]. The titanium dioxide was of the anatase modification as revealed by XRD analysis. Patterning of the titanium dioxide film was done by protecting the appropriate areas using the second mask of a set (“TiO2 mask”, right side of Fig. 2) by a patterned photoresist and etching the exposed TiO2 areas by Reactive Ion Etching (RIE) followed by stripping the photoresist. The “TiO2 masks” were designed such that the TiO2 coverage extended on all sides by 5 m beyond the underlying Pt back-contact strips in order to prevent dark current leakage. A Bank Intelligent Control FM89 potentiostat was used for voltage control of the electrochemical cells consisting of the above mentioned devices and a drop of electrolyte applied to the centre of the device. Reference and counter electrode terminals of the potentiostat were connected to the exposed Pt-finger family, and the working electrode terminal to the Pt-finger family under the TiO2 side. As light sources, the output of a medium pressure Hg lamp or that of the same lamp combined with an interference filter
centred at 365 nm or that of an LED with 376 nm centre wavelength operated with 20–30 mA drive current were used. When the LED was used, it was immersed into the drop of electrolyte in order to maximize irradiance of the electrode. The model pollutant AO7 (“acid orange 7”, 4-(2-hydroxy-1naphthylazo)benzenesulfonic acid sodium salt, C16 H11 N2 NaO4 S) and sodium hydroxide were obtained from Sigma–Aldrich and used without further purification. Ultrapure water (Millipore) was used for preparing NaOH solutions. Various locally available mineral waters served as electrolytes having typical drinking water quality and covering a wide range of ionic strengths. The used electrolytes had the following conductivities: NaOH 0.1 M – 20,600 S/cm, mineral water “Hepar” – 2560 S/cm, “Chantereine” – 577 S/cm, “Volvic” – 195 S/cm, 1 × 10−4 M NaOH – 12 S/cm. 3. Results and discussion Devices with various finger widths and spacing were made. An example of a device resulting from the multi-step deposition process is shown in the profilometric scan in Fig. 3. The 100 m device (100 m TiO2 /100 m gap/100 m Pt) features sharp edges, well reproducing the lithographic masks. The heights of the two finger families (upper trace Pt, lower trace TiO2 ) are clearly outlined. The choice of the finger widths was based on the following considerations: the cell constants, , of interdigitated devices were calculated using a working curve published by Olthuis et al. [14] which allows to calculate as a function of finger width, finger length, inter-finger distance and number of fingers. Accordingly, we calculated for
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
8755
Fig. 4. Cell constant for a 1.5 cm × 0.5 cm area of 5 mm long interdigitated fingers of varying width.
our cells (interdigitated area 0.5 cm × 1.5 cm) as a function of finger width while taking a constant finger length of 0.5 cm. Fig. 4 presents the calculated results for finger width being identical to inter-finger distance, in terms of the number of fingers within the cell. When going from 1 mm wide fingers to 100 m wide fingers, the cell constant is expected to be reduced by a factor of 10. For 100 m wide fingers a of 0.04 cm−1 is obtained, which would lead to a resistance of 80 for tap water of 500 S/cm. With 80 and typical solar current densities as observed on TiO2 [2], a voltage drop of only a few mV would ensue, and a drop of 40 mV for 1 × 10−4 M NaOH, having a conductivity well below common mineral waters. These considerations are checked with the following experiments on an interdigitated device, 100 m TiO2 /100 m gap/100 m Pt, where a 0.5 cm × 0.5 cm area of the interdigitated part was covered by a drop of electrolyte (50 L), resulting in a surface area of 0.25 cm2 . The active area (semiconductor/electrolyte interface) was one fourth of that (0.0625 cm2 ), accounting for the area covered by the counter electrode (Pt fingers) and the insulating gaps between the electrodes. The collection of i–U (current–voltage) curves under illumination and in the dark (Fig. 5), using rectangularly chopped illumination from a medium pressure mercury lamp and electrolytes of different conductance (mineral waters and NaOH) showed the typical response of a polycrystalline n-type TiO2 (anatase) electrode [2]. Although a true plateau was not established, a region where the photocurrent–voltage curve flattens could be discerned. In this region, photocurrents of 7–9 × 10−6 A were observed using UV illumination, corresponding to a photocurrent density of up to 144 A/cm2 based on the part of the illuminated area covered by TiO2 . Small differences between the values of the photocurrents for the different electrolytes can be ascribed to variations of the area of the interface, as the electrolyte drop was not confined. The rise of the photocurrent is not very steep and the plateau region is not very well developed, invoking polarization stemming from the following sources: (i) in a two-terminal electrochemical cell, working and counter electrode current–potential characteristics are added up resulting in the observed i–U curve. The electrochemical reaction at the counter electrode is due to oxygen reduction (as air saturated electrolytes were used) and as the electrolyte was stagnant, this leads to concentration polarisation of the counter electrode. (ii) iR drop can arise from a built-in resistance of the device itself, i.e., limitation of current collection at the ends of the fingers (tip side) and/or contact resistance between TiO2 and Pt.
Fig. 5. Current–voltage curves (20 mV/s) of a 100 m device ((100 + 5) m TiO2 /100 m gap/100 m Pt) under chopped UV illumination from a medium pressure Hg lamp using different electrolytes.
When decreasing the conductivity of the electrolyte from 20,600 to 195 S/cm (Fig. 5 – going from top to bottom), the i–U curves do not rise less steeply, i.e., they do not suffer from increasing iR drop. It can therefore be concluded that a 100 m finger width and spacing is small enough in electrolytes having the low conductance used here. On a 50 m device, a four times higher light intensity than for the experiments in Fig. 5, corresponding to photocurrent densities as encountered for solar illumination, was used, and still no influence of iR drop was noticed for electrolytes down to 195 S/cm. Only for an electrolyte with 12 S/cm specific conductivity, a value hardly ever encountered with drinking water (except if obtained from molten snow or glacier ice), some influence is apparent (Fig. 6), which can partly be ascribed to difficulties for cathodic reactions on the counter electrode. The above considerations on the cell constant are for a liquid layer which is thick enough to accommodate for electric field lines leading to saturation of the conductance, i.e., a situation where further increase of the thickness of the electrolyte layer would not improve conductance. For this to occur, as easily can be shown, the height of the liquid column over an electrode pad must be several times larger than its width. Clearly, for 1 mm wide fingers, this would be more than the thickness of a free flowing liquid film, whereas 100 m wide fingers would require a film of hardly 1 mm to meet the condition. Therefore, such an arrangement can be used in a falling film reactor.
8756
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
Fig. 6. Current–voltage curves (20 mV/s) of a 50 m device (50 + 5 m TiO2 /50 m gap/50 m Pt) under chopped UV illumination using 0.1 M NaOH (thicker trace) or 1 × 10−4 M NaOH (thinner trace).
Fig. 7 shows i–U curves with monochromatic light sources for the electrolyte with a specific conductivity of 195 S/cm. The photocurrents are proportional to light intensity, and in the plateau region photocurrents of 7 × 10−7 , 4 × 10−7 , and 2 × 10−7 A are obtained for the LED emitting at 376 nm at 30 mA and 20 mA drive current, and monochromatized emission of the mercury lamp at 365 nm, respectively. For 376 nm an IPCE of around 0.01 was estimated based on light flux measurements with a calibrated photodiode, in agreement with measurements carried out on aerosol pyrolysis deposited anatase layers [17] where for a 100 nm thick TiO2 layer as used in the present study, the incident photon to current efficiency (IPCE) was only 0.0165, whereas with a thickness of 1000 nm an IPCE of 0.18 was reached. A substantial increase of photocurrents for thicker TiO2 deposits on the fingers can therefore be expected. From the results obtained so far on the interdigitated, planar device and known photoelectrocatalytic properties of spray [2] or sol–gel deposited semiconducting TiO2 , the devices are seen to be fit for oxidative impurity degradation in drinking water in a simple process without pressure build-up due to absence of flow velocity restrictions (recirculative pumping in a thick reactor or free streaming of a liquid film).
Fig. 7. Current–voltage curves (20 mV/s) with monochromatic light sources (376 nm LED at 20 and 30 mA drive current and 365 nm from Hg medium pressure lamp with interference filter) for the electrolyte with the lowest conductance used (195 S/cm).
Fig. 8. Degradation of AO7 as measured by its absorbance at 486 nm. 14 ml 1 × 10−4 M solution in Volvic (air saturated); open squares – illumination by a Hg medium pressure lamp, electrode device obscured, no bias; filled squares – illumination with Hg lamp, no bias; filled circles – illumination with Hg lamp, photocurrent of 11 A at 0.8 V bias.
An example is shown in the degradation of AO7. A “100–100” device was placed in the centre of a small glass vessel filled with 14 ml of 1 × 10−4 M AO7 dissolved in the mineral water Volvic. Irradiation with a medium pressure Hg lamp led to a photocurrent of 11 A at a bias of 0.8 V. In a “blank” experiment it was seen that without bias also some degradation took place. This could be due to photocatalysis (photochemical reaction at the unbiased electrode) or due to direct photolysis. A second blank experiment with the device obscured (turned 180◦ out of the light beam) led to the same degradation rate, showing that this contribution is entirely due to direct photolysis, as some light is absorbed by the reactant solution before reaching the electrode. (In a final implementation with a down-streaming liquid film (less than 1 mm thick) this contribution would be minimized and at the same time light absorption by the electrode increased.) In Fig. 8 results of the three types of degradation experiments are plotted logarithmically. From the linear dependences, apparent first order rate constants were obtained: k(direct photolysis) = 2.17 × 10−6 /s and k(photoelectrocatalysis + direct photolysis) = 5.16 × 10−6 /s. k(photoelectrocatalysis) was therefore 3.29 × 10−6 /s and the corresponding kinetic parameter, p (=kFV/iphoto , i.e., rate constant normalized to unit liquid volume and photocurrent) was 404 M−1 , in good agreement with results found for photoelectrocatalytical degradation of AO7 in a batch reactor under recirculation using a polycrystalline anatase electrode [2]. The deposition of such structures on glass may be envisaged, as TiO2 can be deposited at temperatures compatible with the softening temperature of glass [2,17]. As for building larger devices of this kind, by depositing multiple copies of the small building blocks described here and connecting them in parallel, ink-jet printing is within reach for down to 100 m features if reproducible alignment to the precision required by multiple lithographic steps can be achieved. Instead of printing masks, the active layers could be directly printed using organic/inorganic inks containing the required precursors with intermediate annealing steps. The photoelectrocatalytic process can be easily upscaled. A device consisting of interdigitated patterns repeated over a large area would be the active part of a falling film reactor dimensioned such that the required decrease of the concentration of organic impurities would be afforded in single pass or multiple pass (recirculation) mode.
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
Acknowledgements F. Süßl, MetaDesign AG, Berlin, Germany is thanked for transferring the original outlines of the lithographic masks onto computer files. R. Keller (malkasten Bildbearbeitungsstudio Ges.m.b.H., Wien, Austria) is thanked for transferring the contents of the computer files of the masks onto photographic film. The preparation of the devices by Y. Paz (Technion, Israel) is gratefully acknowledged. References [1] M. Neumann-Spallart, O. Enea, J. Electrochem. Soc. 131 (1984) 2767. [2] P.S. Shinde, P.S. Patil, P.N. Bhosale, A. Brüger, G. Nauer, M. Neumann-Spallart, C.H. Bhosale, Appl. Catal. B: Environ. 89 (2009) 288. [3] N. Philippidis, S. Sotiropoulos, A. Efstathiou, I. Poulios, J. Photochem. Photobiol. A: Chem. 204 (2009) 129. [4] P.A. Carneiro, M.E. Osugi, J.J. Sene, M.A. Anderson, M. Valnice, B. Zanoni, Electrochim. Acta 49 (22–23) (2004) 3807.
8757
[5] J. Marugán, P. Christensen, T. Egerton, H. Purnama, Appl. Catal. B: Environ. 89 (1–2) (2009) 273. [6] M.F. Brugnera, K. Rajeshwar, J.C. Cardoso, M. Valnice, B. Zanoni, Chemosphere 8 (5) (2010) 569. [7] M. Hepel, S. Hazelton, Electrochim. Acta 50 (2005) 5278. [8] J. Georgieva, S. Armyanov, E. Valova, Ts. Tsacheva, I. Poulios, S. Sotiropoulos, J. Electroanal. Chem. 585 (1) (2005) 35. [9] M. Neumann-Spallart, S.B. Sadale, J. New Mater. Electrochem. Syst. 13 (2010) 127. [10] Y. Iwasaki, M. Morita, Curr. Sep. 14 (1) (1995) 2. [11] A.E. Cohen, R.R. Kunz, Sens. Actuators B 62 (2000) 23. [12] X. Zhu, J.-W. Choi, C.H. Ahn, Lab Chip 4 (2004) 581. [13] K. Ueno, M. Hayashida, J.-Y. Ye, H. Misawa, Electrochem. Commun. 7 (2005) 161. [14] W. Olthuis, W. Streekstra, P. Bergveld, Sens. Actuators B 24–25 (1995) 252. [15] Y. Paz, Z. Luo, L. Rabenberg, A. Heller, J. Mater. Res. 10 (1995) 2842. [16] H. Haick, Y. Paz, ChemPhysChem 4 (2003) 620. [17] A. Belaidi, S.M. Chaqour, O. Gorochov, M. Neumann-Spallart, Mater. Res. Bull. 39 (2004) 599.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Photoelectrochemistry on a planar, interdigitated electrochemical cell M. Neumann-Spallart Groupe d’Études de la Matière Condensée, C.N.R.S. and Université de Versailles St. Quentin, 1, place Aristide Briand, 92195 Meudon CEDEX, France
a r t i c l e
i n f o
Article history: Received 5 April 2011 Received in revised form 20 July 2011 Accepted 20 July 2011 Available online 28 July 2011 Keywords: Interdigitated electrode Photoelectrocatalysis Titanium dioxide
a b s t r a c t The construction of a planar electrochemical cell with two interdigitated electrodes, with a finger spacing of 25–100 m and 200–50 pairs of fingers and gaps, respectively, is described. The working electrode consists of n-type semiconducting titanium dioxide (anatase). Under UV light, photocurrents are produced. No reduction in the photocurrent due to iR drop was observed, even in electrolytes of low ionic strength. This makes the device an interesting candidate for photoelectrocatalytic purification of drinking water. An example is shown in the degradation of the azo-dye AO7 (acid orange 7). k(photoelectrocatalysis) on a device with 50 finger pairs was 3.29 × 10−6 /s and the corresponding kinetic parameter, p = kFV/iphoto (rate constant normalized to unit volume and photocurrent), was 404 M−1 . © 2011 Elsevier Ltd. All rights reserved.
1. Introduction The removal of organic material of biological and industrial origin (bacteria, hormones, endocrine disruptors, textile dyes, dioxines, etc.) from waste- and drinking water is of enormous importance nowadays. There is a great need for new technologies serving this purpose. One way could be photocatalysis, involving semiconducting oxides under illumination. In a heterogeneous photocatalytic process, a solid semiconductor catalyst such as titanium dioxide or other transition metal oxides irradiated with near UV light generate strongly oxidizing species (valence band holes, ·OH radicals) which can destroy organic species present in the aqueous phase in contact with the surface of the catalyst. However, there are limitations to the efficiency of the process that are partly due to the semiconductor and partly due to various recombination processes and the technical outline of the reactor in which the reaction is carried out. Early work [1] showed how photocatalysis in semiconductor particulate suspensions can be understood on the basis of photoelectrochemical analysis of macroscopic electrodes. From this analysis it became also clear that the main obstacle to efficient use of absorbed photons is the recombination of photogenerated charge carriers, and that this can be suppressed by applying electrical bias to the semiconducting photocatalyst which must be immobilized on an electrically conducting substrate. This has also the advantage of making separation of the catalyst from the reac-
E-mail address: [email protected] 0013-4686/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2011.07.089
tion mixture (the liquid phase) superfluous, once the reaction is over. Recent work on organic impurity degradation using TiO2 [2–6] and WO3 thin film electrodes [7–9] validated these ideas. In an electrochemical cell, iR drop is one of the factors limiting high current throughput at moderate bias. Especially in electrochemical treatment of additives (pollutants) in drinking water, having low ionic strength, where addition of supporting electrolyte is prohibited for obvious reasons, ways must be devised for limiting iR drop. (Typical tap water conductivities of 528 S/cm were observed in our laboratory.) One way is to use a parallel plate reactor with two opposite electrodes and a small space between them where the electrolyte is passed through. For practical reasons (if the electrodes are large) this distance may not fall below 1 mm, if energy investment for pumping is to be minimized. Such a reactor with 10 cm × 10 cm plates and a 1 mm gap, with a conductivity cell constant, , of 0.001 cm−1 was recently demonstrated [2], and had an electrolyte resistance for tap water (500 S/cm) of 2 , leading to an iR drop of only 40 mV for current densities of around 3 × 10−4 A/cm2 as observed for photocurrents on TiO2 produced by solar irradiation. However, as the pressure build-up is considerable in a module consisting of many such cells, another way is suggested in the present work: both electrodes are positioned in the same plane so that no flow restrictions apply. However, when large currents are to be passed through a planar device, it must be taken into account that the conductivity cell constant of such an arrangement is considerable higher than that of an equivalent arrangement where the electrodes face each other. In order to ensure low iR drop between the electrodes, they are arranged in an interdigitated way. Electrochemical cells with various planar interdigitated arrangements have been used for redox sensing and other electroanalytical
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
8753
applications [10–13]. Such structures can lead to very low cell constants if a large number of thin fingers and small spacing are used [14].
2. Experimental The schematics of a typical planar device with an interdigitated part of 0.5 cm × 1.5 cm are shown in Fig. 1. Fingers extending from the left side towards the centre consist of photoactive TiO2 deposited on platinum, whereas the finger family starting on the right side is the Pt counter electrode. Electric contacts were taken at the bottom to which the Pt collector areas are extended. The devices were fabricated by standard lithographic techniques using contact masks. The masks were prepared in the following way: the pattern was drawn as vector graphics on computer file using the program “Adobe Illustrator”. The patterns were then transferred either onto transparent plastic sheets using a laser printer, or directly onto photographic film. As these foils were subsequentially used as contact masks in the lithographic processes, high accuracy combined with mechanical stability for repeated use was sought, giving preference to patterns on photographic film. An example is given in the optical microscope pictures in Fig. 2 with details of the pair of masks used for the “100–25” device, i.e., a device with 100 m wide Pt and TiO2 fingers and a 25 m gap between the finger families: a mask used for Pt deposition (a) and a mask used during TiO2 deposition (b). The optical pictures show the excellent sharpness of the lithographic masks. Masks with different finger widths (100, 50, 25 m) and interfinger distances (100, 50, 25 m) were prepared, with a finger length of 5 mm. Devices with patterns deposited over a length of 15 mm and equal finger spacing of 25–100 m led to 200–50 pairs of fingers, respectively, and the same number of pairs of gaps (insulating parts made of SiO2 between the finger families), covering a total operating area of 0.75 cm2 , of which 0.1875 cm2 (one fourth) was rendered photoactive due to coverage with titanium dioxide.
Fig. 1. Schematics of an interdigitated device (in scale: interdigitated area 1.5 cm × 0.5 cm). Black part – insulating area (SiO2 ), light grey part – titanium dioxide, white part – Pt. Bottom left and right – area for applying external contacts.
Fig. 2. Inspection of the emulsion side of developed photographic film containing the masks. Details (optical microscope) of a pair of masks used for depositing a “100–25” device (100 m wide Pt and TiO2 fingers and 25 m gap between the finger families). Left – mask used for Pt deposition, right – mask used for revealing TiO2 .
8754
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
Fig. 3. Profilometric scan of a “100–100” device (100 m wide Pt and 5 + 100 + 5 m TiO2 fingers and 100 m gap between the finger families).
The substrates for the devices were 2 in. silicon wafers with a thick (2 m) insulating layer of silicon dioxide. A thin (7 nm) titanium layer, serving as an adhesion promoter, and a thicker (200 nm) platinum layer were then deposited by evaporation on a pre-patterned positive photoresist (using the first mask of a set of two – the “Pt-mask”). The metals were patterned by removing the pre-patterned photoresist, leaving an interdigitated arrangement consisting of both electrodes with their pads. The next step consisted of over-coating the whole wafer by a 100 nm thick layer of titanium dioxide by spin-coating of an organo-titanate precursor, followed by calcination. Details of this step were given elsewhere [15,16]. The titanium dioxide was of the anatase modification as revealed by XRD analysis. Patterning of the titanium dioxide film was done by protecting the appropriate areas using the second mask of a set (“TiO2 mask”, right side of Fig. 2) by a patterned photoresist and etching the exposed TiO2 areas by Reactive Ion Etching (RIE) followed by stripping the photoresist. The “TiO2 masks” were designed such that the TiO2 coverage extended on all sides by 5 m beyond the underlying Pt back-contact strips in order to prevent dark current leakage. A Bank Intelligent Control FM89 potentiostat was used for voltage control of the electrochemical cells consisting of the above mentioned devices and a drop of electrolyte applied to the centre of the device. Reference and counter electrode terminals of the potentiostat were connected to the exposed Pt-finger family, and the working electrode terminal to the Pt-finger family under the TiO2 side. As light sources, the output of a medium pressure Hg lamp or that of the same lamp combined with an interference filter
centred at 365 nm or that of an LED with 376 nm centre wavelength operated with 20–30 mA drive current were used. When the LED was used, it was immersed into the drop of electrolyte in order to maximize irradiance of the electrode. The model pollutant AO7 (“acid orange 7”, 4-(2-hydroxy-1naphthylazo)benzenesulfonic acid sodium salt, C16 H11 N2 NaO4 S) and sodium hydroxide were obtained from Sigma–Aldrich and used without further purification. Ultrapure water (Millipore) was used for preparing NaOH solutions. Various locally available mineral waters served as electrolytes having typical drinking water quality and covering a wide range of ionic strengths. The used electrolytes had the following conductivities: NaOH 0.1 M – 20,600 S/cm, mineral water “Hepar” – 2560 S/cm, “Chantereine” – 577 S/cm, “Volvic” – 195 S/cm, 1 × 10−4 M NaOH – 12 S/cm. 3. Results and discussion Devices with various finger widths and spacing were made. An example of a device resulting from the multi-step deposition process is shown in the profilometric scan in Fig. 3. The 100 m device (100 m TiO2 /100 m gap/100 m Pt) features sharp edges, well reproducing the lithographic masks. The heights of the two finger families (upper trace Pt, lower trace TiO2 ) are clearly outlined. The choice of the finger widths was based on the following considerations: the cell constants, , of interdigitated devices were calculated using a working curve published by Olthuis et al. [14] which allows to calculate as a function of finger width, finger length, inter-finger distance and number of fingers. Accordingly, we calculated for
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
8755
Fig. 4. Cell constant for a 1.5 cm × 0.5 cm area of 5 mm long interdigitated fingers of varying width.
our cells (interdigitated area 0.5 cm × 1.5 cm) as a function of finger width while taking a constant finger length of 0.5 cm. Fig. 4 presents the calculated results for finger width being identical to inter-finger distance, in terms of the number of fingers within the cell. When going from 1 mm wide fingers to 100 m wide fingers, the cell constant is expected to be reduced by a factor of 10. For 100 m wide fingers a of 0.04 cm−1 is obtained, which would lead to a resistance of 80 for tap water of 500 S/cm. With 80 and typical solar current densities as observed on TiO2 [2], a voltage drop of only a few mV would ensue, and a drop of 40 mV for 1 × 10−4 M NaOH, having a conductivity well below common mineral waters. These considerations are checked with the following experiments on an interdigitated device, 100 m TiO2 /100 m gap/100 m Pt, where a 0.5 cm × 0.5 cm area of the interdigitated part was covered by a drop of electrolyte (50 L), resulting in a surface area of 0.25 cm2 . The active area (semiconductor/electrolyte interface) was one fourth of that (0.0625 cm2 ), accounting for the area covered by the counter electrode (Pt fingers) and the insulating gaps between the electrodes. The collection of i–U (current–voltage) curves under illumination and in the dark (Fig. 5), using rectangularly chopped illumination from a medium pressure mercury lamp and electrolytes of different conductance (mineral waters and NaOH) showed the typical response of a polycrystalline n-type TiO2 (anatase) electrode [2]. Although a true plateau was not established, a region where the photocurrent–voltage curve flattens could be discerned. In this region, photocurrents of 7–9 × 10−6 A were observed using UV illumination, corresponding to a photocurrent density of up to 144 A/cm2 based on the part of the illuminated area covered by TiO2 . Small differences between the values of the photocurrents for the different electrolytes can be ascribed to variations of the area of the interface, as the electrolyte drop was not confined. The rise of the photocurrent is not very steep and the plateau region is not very well developed, invoking polarization stemming from the following sources: (i) in a two-terminal electrochemical cell, working and counter electrode current–potential characteristics are added up resulting in the observed i–U curve. The electrochemical reaction at the counter electrode is due to oxygen reduction (as air saturated electrolytes were used) and as the electrolyte was stagnant, this leads to concentration polarisation of the counter electrode. (ii) iR drop can arise from a built-in resistance of the device itself, i.e., limitation of current collection at the ends of the fingers (tip side) and/or contact resistance between TiO2 and Pt.
Fig. 5. Current–voltage curves (20 mV/s) of a 100 m device ((100 + 5) m TiO2 /100 m gap/100 m Pt) under chopped UV illumination from a medium pressure Hg lamp using different electrolytes.
When decreasing the conductivity of the electrolyte from 20,600 to 195 S/cm (Fig. 5 – going from top to bottom), the i–U curves do not rise less steeply, i.e., they do not suffer from increasing iR drop. It can therefore be concluded that a 100 m finger width and spacing is small enough in electrolytes having the low conductance used here. On a 50 m device, a four times higher light intensity than for the experiments in Fig. 5, corresponding to photocurrent densities as encountered for solar illumination, was used, and still no influence of iR drop was noticed for electrolytes down to 195 S/cm. Only for an electrolyte with 12 S/cm specific conductivity, a value hardly ever encountered with drinking water (except if obtained from molten snow or glacier ice), some influence is apparent (Fig. 6), which can partly be ascribed to difficulties for cathodic reactions on the counter electrode. The above considerations on the cell constant are for a liquid layer which is thick enough to accommodate for electric field lines leading to saturation of the conductance, i.e., a situation where further increase of the thickness of the electrolyte layer would not improve conductance. For this to occur, as easily can be shown, the height of the liquid column over an electrode pad must be several times larger than its width. Clearly, for 1 mm wide fingers, this would be more than the thickness of a free flowing liquid film, whereas 100 m wide fingers would require a film of hardly 1 mm to meet the condition. Therefore, such an arrangement can be used in a falling film reactor.
8756
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
Fig. 6. Current–voltage curves (20 mV/s) of a 50 m device (50 + 5 m TiO2 /50 m gap/50 m Pt) under chopped UV illumination using 0.1 M NaOH (thicker trace) or 1 × 10−4 M NaOH (thinner trace).
Fig. 7 shows i–U curves with monochromatic light sources for the electrolyte with a specific conductivity of 195 S/cm. The photocurrents are proportional to light intensity, and in the plateau region photocurrents of 7 × 10−7 , 4 × 10−7 , and 2 × 10−7 A are obtained for the LED emitting at 376 nm at 30 mA and 20 mA drive current, and monochromatized emission of the mercury lamp at 365 nm, respectively. For 376 nm an IPCE of around 0.01 was estimated based on light flux measurements with a calibrated photodiode, in agreement with measurements carried out on aerosol pyrolysis deposited anatase layers [17] where for a 100 nm thick TiO2 layer as used in the present study, the incident photon to current efficiency (IPCE) was only 0.0165, whereas with a thickness of 1000 nm an IPCE of 0.18 was reached. A substantial increase of photocurrents for thicker TiO2 deposits on the fingers can therefore be expected. From the results obtained so far on the interdigitated, planar device and known photoelectrocatalytic properties of spray [2] or sol–gel deposited semiconducting TiO2 , the devices are seen to be fit for oxidative impurity degradation in drinking water in a simple process without pressure build-up due to absence of flow velocity restrictions (recirculative pumping in a thick reactor or free streaming of a liquid film).
Fig. 7. Current–voltage curves (20 mV/s) with monochromatic light sources (376 nm LED at 20 and 30 mA drive current and 365 nm from Hg medium pressure lamp with interference filter) for the electrolyte with the lowest conductance used (195 S/cm).
Fig. 8. Degradation of AO7 as measured by its absorbance at 486 nm. 14 ml 1 × 10−4 M solution in Volvic (air saturated); open squares – illumination by a Hg medium pressure lamp, electrode device obscured, no bias; filled squares – illumination with Hg lamp, no bias; filled circles – illumination with Hg lamp, photocurrent of 11 A at 0.8 V bias.
An example is shown in the degradation of AO7. A “100–100” device was placed in the centre of a small glass vessel filled with 14 ml of 1 × 10−4 M AO7 dissolved in the mineral water Volvic. Irradiation with a medium pressure Hg lamp led to a photocurrent of 11 A at a bias of 0.8 V. In a “blank” experiment it was seen that without bias also some degradation took place. This could be due to photocatalysis (photochemical reaction at the unbiased electrode) or due to direct photolysis. A second blank experiment with the device obscured (turned 180◦ out of the light beam) led to the same degradation rate, showing that this contribution is entirely due to direct photolysis, as some light is absorbed by the reactant solution before reaching the electrode. (In a final implementation with a down-streaming liquid film (less than 1 mm thick) this contribution would be minimized and at the same time light absorption by the electrode increased.) In Fig. 8 results of the three types of degradation experiments are plotted logarithmically. From the linear dependences, apparent first order rate constants were obtained: k(direct photolysis) = 2.17 × 10−6 /s and k(photoelectrocatalysis + direct photolysis) = 5.16 × 10−6 /s. k(photoelectrocatalysis) was therefore 3.29 × 10−6 /s and the corresponding kinetic parameter, p (=kFV/iphoto , i.e., rate constant normalized to unit liquid volume and photocurrent) was 404 M−1 , in good agreement with results found for photoelectrocatalytical degradation of AO7 in a batch reactor under recirculation using a polycrystalline anatase electrode [2]. The deposition of such structures on glass may be envisaged, as TiO2 can be deposited at temperatures compatible with the softening temperature of glass [2,17]. As for building larger devices of this kind, by depositing multiple copies of the small building blocks described here and connecting them in parallel, ink-jet printing is within reach for down to 100 m features if reproducible alignment to the precision required by multiple lithographic steps can be achieved. Instead of printing masks, the active layers could be directly printed using organic/inorganic inks containing the required precursors with intermediate annealing steps. The photoelectrocatalytic process can be easily upscaled. A device consisting of interdigitated patterns repeated over a large area would be the active part of a falling film reactor dimensioned such that the required decrease of the concentration of organic impurities would be afforded in single pass or multiple pass (recirculation) mode.
M. Neumann-Spallart / Electrochimica Acta 56 (2011) 8752–8757
Acknowledgements F. Süßl, MetaDesign AG, Berlin, Germany is thanked for transferring the original outlines of the lithographic masks onto computer files. R. Keller (malkasten Bildbearbeitungsstudio Ges.m.b.H., Wien, Austria) is thanked for transferring the contents of the computer files of the masks onto photographic film. The preparation of the devices by Y. Paz (Technion, Israel) is gratefully acknowledged. References [1] M. Neumann-Spallart, O. Enea, J. Electrochem. Soc. 131 (1984) 2767. [2] P.S. Shinde, P.S. Patil, P.N. Bhosale, A. Brüger, G. Nauer, M. Neumann-Spallart, C.H. Bhosale, Appl. Catal. B: Environ. 89 (2009) 288. [3] N. Philippidis, S. Sotiropoulos, A. Efstathiou, I. Poulios, J. Photochem. Photobiol. A: Chem. 204 (2009) 129. [4] P.A. Carneiro, M.E. Osugi, J.J. Sene, M.A. Anderson, M. Valnice, B. Zanoni, Electrochim. Acta 49 (22–23) (2004) 3807.
8757
[5] J. Marugán, P. Christensen, T. Egerton, H. Purnama, Appl. Catal. B: Environ. 89 (1–2) (2009) 273. [6] M.F. Brugnera, K. Rajeshwar, J.C. Cardoso, M. Valnice, B. Zanoni, Chemosphere 8 (5) (2010) 569. [7] M. Hepel, S. Hazelton, Electrochim. Acta 50 (2005) 5278. [8] J. Georgieva, S. Armyanov, E. Valova, Ts. Tsacheva, I. Poulios, S. Sotiropoulos, J. Electroanal. Chem. 585 (1) (2005) 35. [9] M. Neumann-Spallart, S.B. Sadale, J. New Mater. Electrochem. Syst. 13 (2010) 127. [10] Y. Iwasaki, M. Morita, Curr. Sep. 14 (1) (1995) 2. [11] A.E. Cohen, R.R. Kunz, Sens. Actuators B 62 (2000) 23. [12] X. Zhu, J.-W. Choi, C.H. Ahn, Lab Chip 4 (2004) 581. [13] K. Ueno, M. Hayashida, J.-Y. Ye, H. Misawa, Electrochem. Commun. 7 (2005) 161. [14] W. Olthuis, W. Streekstra, P. Bergveld, Sens. Actuators B 24–25 (1995) 252. [15] Y. Paz, Z. Luo, L. Rabenberg, A. Heller, J. Mater. Res. 10 (1995) 2842. [16] H. Haick, Y. Paz, ChemPhysChem 4 (2003) 620. [17] A. Belaidi, S.M. Chaqour, O. Gorochov, M. Neumann-Spallart, Mater. Res. Bull. 39 (2004) 599.