Hana Krysovaa,
Ladislav Kavan*a,
Zuzana Vlckova Zivcovaa,
Weng Siang Yeapb,
Pieter Verstappenb,
Wouter Maesbc,
Ken Haenenbc,
Fang Gaod and
Christoph E. Nebeld
aJ. Heyrovsky Institute of Physical Chemistry of the AS CR, v. v. i., Dolejškova 3, CZ-182 23, Prague 8, Czech Republic. E-mail: kavan@jh-inst.cas.cz; Fax: +420 2 8658 2307; Tel: +420 2 6605 3975
bHasselt University, Institute for Materials Research (IMO), B-3590 Diepenbeek, Belgium
cIMEC vzw, IMOMEC, B-3590 Diepenbeek, Belgium
dFraunhofer Institute for Applied Solid State Physics (IAF), Tullastrasse 72, D-79108 Freiburg, Germany
First published on 16th September 2015
Diamond foams composed of hollow spheres of polycrystalline boron-doped diamond are chemically modified with two donor–acceptor type molecular dyes, BT-Rho and CPDT-Fur, and tested as electrode materials for p-type dye-sensitized solar cells with an aqueous electrolyte solution containing methyl viologen as a redox mediator. Reference experiments with flat polycrystalline diamond electrodes evidence full blocking of the methyl viologen redox reaction by these dyes, whereas only partial blocking is observed for the diamond foams. This is ascribed to sp2-carbon impurities in the foam, viz. trans-polyacetylene and graphite-like carbon. Cathodic photocurrents under solar light illumination are about 3 times larger on foam electrodes compared to flat diamond. Long-term (1–2 days) illumination of the sensitized foam electrodes with chopped light at 1 sun intensity causes an increase of the cathodic photocurrent density to ca. 15–22 μA cm−2. These photocurrent densities represent the largest values reported so far for dye-sensitized diamond electrodes. The photoelectrochemical activation of the sensitized diamond electrodes is accompanied with characteristic changes of the dark voltammogram of the MV2+/MV+ redox couple and with gradual changes of the IPCE spectra.
Spectral sensitization of diamond surfaces by organic dyes, absorbing visible light, has been pioneered in 2008 by Zhong et al.12 who covalently anchored dicyanovinyl-bithiophene and C60-bithiophene on H-terminated BDD through Suzuki cross-coupling reactions. They observed photocurrent densities of ca. 120 nA cm−2 under white light illumination (150 W halogen lamp) in an aqueous electrolyte solution with methyl viologen acting as the electron carrier. Later on, photocurrent densities of ca. 4–6 μA cm−2 were observed in similar systems under 1 sun illumination.9 Sensitization of BDD by manganese phthalocyanine13,14 or Ru(SCN)2(pbca)2 (pbca = 2,2′-bipyridine-4,4′-dicarboxylate) (commonly known as the N3 dye)15 provided rather low photocurrent densities, typically of the order of 1–10 nA cm−2 under white light illumination with an intensity of about 1 sun. Krysova et al.16 reported on non-covalent anchoring of the 4-(bis{4-[5-(2,2-dicyanovinyl)thiophene-2-yl]phenyl}amino)benzoic acid dye (coded P1) through a polyethyleneimine linker. Although the P1 dye is known to be successful for the sensitization of p-NiO,17,18 the reported photocurrent densities on diamond electrodes were again low, about 100–150 nA cm−2 at white light intensity of 18 mW cm−2 (interestingly, the P1 dye turned out to be appropriate for the sensitization of n-TiO2, too,18 which is reminiscent of the applicability of the N3 dye in both TiO2 and diamond electrodes).15 Yeap et al.19 modified the diamond surface with different thiophene-based molecular wires through a combination of diazonium electrografting and Suzuki cross-coupling, and observed photocurrent densities of ca. 150 nA cm−2 under white light illumination (15 mW cm−2 intensity).
One of the reasons for the poor photoelectrochemical performance of sensitized diamond is obviously the low surface area (roughness factor) and the lack of mesoporous texture of the BDD films applied so far.16,19 The light-harvesting efficiency of a monolayer of dye molecules on a flat surface is inherently small. For instance, the maximum incident photon-to-electron conversion efficiency (IPCE) of the N3-dye monolayer on an atomically flat TiO2 anatase is theoretically 0.27% (and experimentally around 0.11%)20,21 while IPCEs exceeding 90% are commonly observed for a properly nanostructured anatase with a roughness factor of ca. 1000 (the roughness factor, Rf, being defined as the ratio of the physical surface area to the geometric electrode area).1 This finding raises the challenge to reproduce surface nanostructuring on diamond electrodes for p-DSC applications, too. The growth of nano-textured mesoporous diamond films has been attempted in the past by various protocols, such as inductively-coupled plasma etching,22,23 oxygen-plasma treatment,24 etching by superheated water in graphene nanobubbles25 and by templating with SiO2 fibers26 or SiO2 spheres.27,28 To the best of our knowledge, none of these porous diamond materials were used in p-DSCs yet.
To address this point, we present here our initial results observed upon replacing the traditional flat diamond films from chemical vapor deposition (CVD) growth by so-called ‘diamond foams’. This material was developed by Kato et al.27 and was successfully tested for applications in double-layer supercapacitors.27,28 It is grown by using SiO2 spheres (500 nm in diameter) as templates, on which a thin BDD layer is deposited by standard CVD growth. The template is subsequently etched away by HF solution, leaving hollow spheres of diamond which replicate the SiO2 template. The diamond-foam electrodes exhibit a relatively low specific capacitance (2.4–16 F g−1) compared to e.g. activated carbon, but they are attractive because of their large electrochemical window and, particularly, because of their capability of fast charging at rates as high as 1000 V s−1.28
To directly compare flat diamond films and diamond foams, we have functionalized the diamond foams with the same dyes as applied in our previous work on flat BDD films (‘flat’ means here a polycrystalline CVD-grown film which was deposited on a smooth substrate, such as a fused quartz plate16,19). More specifically, two different molecules were used as the diamond sensitizers, viz. (E)-2-{4-[2-(6-bromo-4,4-diethyl-4H-cyclopenta[1,2-b:5,4-b′]dithiophen-2-yl)vinyl]-3-cyano-5,5-dimethylfuran-2(5H)-ylidene}malononitrile (coded CPDT-Fur) and (Z)-2-{5-[(5′-bromo[2,2′-bithiophen]-5-yl)methylene]-4-oxo-2-thioxothiazolidin-3-yl}acetic acid (coded BT-Rho). These molecules were anchored by the Suzuki coupling protocol through a phenyl linker, which was attached to the originally H-terminated diamond surface by diazonium electrografting.19 The structures of the modified diamond surfaces are shown in Scheme 1.
![]() | ||
Scheme 1 Structures of the chemically modified diamond surfaces with (a) CPDT-Fur and (b) BT-Rho sensitizers covalently anchored through a phenyl linker. |
The diamond foam was synthesized as described before.28 Briefly, silica spheres (0.5 μm in diameter; Kisker Biotech GmbH & Co. Germany) were spin-coated from an isopropanol suspension onto a BDD diamond wafer (grown with the MWCVD method; 2% CH4 in H2, 3200 W, 50 mbar, 750 °C, B/C = 4000 ppm). Subsequently, the spheres were seeded with aqueous 1 wt% H-terminated nanodiamond colloid and dried. BDD coating on the spheres was performed by the MWCVD method (1% CH4 in H2; 2200 W, 40 mbar, 650 °C, B/C = 12000 ppm). Typically, 5–6 layers of diamond-coated spheres on top of a compact BDD film were prepared. The electrode was then cleaned by boiling in H2SO4 + HNO3 (3
:
1) for 1.5 h at 200 °C followed by treatment with HF to remove the SiO2 templates and finally rinsed with water and methanol. Before dye anchoring, the diamond foam was hydrogenated under the following conditions: (i) 500 sccm hydrogen flux, 40 mbar pressure with a 3500 W microwave power for 5 min, (ii) 500 sccm hydrogen flux, 20 mbar pressure with a 2500 W microwave power for 15 min, (iii) at the end of the plasma treatment, the microwave power was switched off and the samples were allowed to cool down under hydrogen flux for 1 hour (a longer treatment time was used for the foam than for the flat films to secure complete hydrogenation). SIMS analysis of a one-layer electrode indicated a B-concentration of (7–10) × 1019 atoms per cm3 for the foam layer and ≈1021 atoms per cm3 for the supporting compact BDD film (cf. Fig. S1 in ESI†).
The functionalization of the diamond surface started with diazoniation of the targeted aniline (4-aminophenylboronic acid pinacol ester) followed by electrochemical reduction of the in situ generated diazonium salt.12,19 More specifically, 5 mM of 4-aminophenylboronic acid pinacol ester was diazotized with an equimolar amount of NaNO2 in a (N2 gas purged) 0.5 M HCl solution, which was directly used for CV scanning between +0.5 and −0.8 V (vs. Ag/AgCl) at a rate of 100 mV s−1 for 5 scans. After the modification, the substrate was sequentially rinsed and sonicated in water, THF and n-hexane. The Suzuki cross-coupling reaction was performed in a glove box. A 15 mL ACE pressure tube (Sigma-Aldrich) containing a magnetic stirring bar and the functionalized diamond film was charged with the dye, Pd catalyst, base and solvent and heated at 80 °C for 18 h. An optimized base/solvent/catalyst was used, as described elsewhere.19
Electrochemical experiments were carried out in a one-compartment cell using an Autolab PGstat-302N controlled by GPES-4 software. The BDD film was used as a working electrode (Ag contact with Au wire insulated by TorrSeal epoxy coating), platinum mesh was used as the counter electrode and a Ag/AgCl electrode (sat. KCl) was applied as the reference electrode. All electrochemical measurements were performed under Ar atmosphere. For the photoelectrochemical experiments, the cell was equipped with a quartz optical window, and the electrode was illuminated by a white light source (Oriel Xenon lamp, model 6269). The solar radiation (direct and diffuse) was simulated by an Oriel AM 1.5 Global (81088) filter. The light intensity was measured by a standard Si photodiode (PV Measurements, Inc. USA). For the quantum efficiency measurements (IPCE), the light was monochromatized using a Newport 1/4 m grating monochromator (model 77200). Photoelectrochemical measurements were performed in an Ar-saturated 0.1 M Na2SO4 solution containing 5 mM methyl viologen (MV2+), pH ≈ 7. The counter electrode was platinum and the reference electrode was Ag/AgCl (sat. KCl). The photoelectrochemical cell was placed in a dark room and controlled by a potentiostat (micro-AutolabIII, Ecochemie, B.V.) with NOVA software.
Raman spectra of the BDD foam electrodes are shown in Fig. S3 (ESI†). In accordance with previous works,27,28 the sp3-carbon (diamond) peak is detected at ca. 1330 cm−1. The main impurity signals are assigned to trans-polyacetylene (1150 cm−1) and to graphite-like (sp2) carbon, which manifests itself by a broad and strong G-peak at ca. 1500 cm−1. The absence of Raman features around 500 cm−1 indicates removal of SiO2 as well as low B-doping. The latter is further confirmed by negligible Raman intensity near 1225 cm−1 where the Raman feature assigned to a Fano resonance line-shape of the diamond band normally occurs for heavily doped diamond.31 The dye anchoring causes no marked changes to the Raman spectra (Fig. S3†). For comparison, Raman spectra of flat BDD films on SiO2 or Si substrates and with different excitation wavelengths are shown in Fig. S4a and b (ESI†). Again, no significant difference between pure and dye-sensitized films is observed. The strong features around 500 and 1200 cm−1 confirm larger B-doping of the flat films. The frequency of the first peak is known to scale with acceptor concentration which allows determination of the doping level.32 The calculated concentration is 2 × 1021 atoms per cm3 which is near the expected value for the flat films (see Experimental section). The diamond foam has a smaller dopant concentration by one order of magnitude: (7–10) × 1019 atoms per cm3 (as found by SIMS, see above). Furthermore, the signal of sp2-carbon impurity is considerably smaller on the flat diamond compared to the diamond foam.
Kr adsorption isotherms at 77 K were evaluated by the Brunauer–Emmett–Teller (BET) method. This procedure is applicable for determination of the roughness factor (Rf).29,30 The detection limit of our method is about 20 in the Rf units33 which is above the assumed values of Rf for flat diamond (≈4, see Section 3.2). On the other hand, our BDD foam electrode provided good isotherm (Fig. S5; ESI†) from which the Rf equals 114.
i = dQ/dt = CdE/dt = Cv, | (1) |
![]() | (2) |
We further measured the cyclic voltammograms of the diamond electrodes (both flat films and foams) in the electrolyte solution, which was applied in the photoelectrochemical experiments (5 mM methyl viologen in 0.1 M Na2SO4, pH 7; see below). Fig. 2 shows the relevant data with a reference voltammogram obtained on an indium-tin oxide (ITO) electrode. The latter exhibits the expected redox waves of methyl viologen (redox potential of −0.65 V vs. Ag/AgCl).36 On a flat diamond electrode, these waves become highly irreversible, although the electrochemical responsiveness of BDD to methyl viologen is known to be quite good.37 Interestingly, the sensitized BDD flat film exhibits a total blocking of the MV2+/MV+ redox couple. This may indicate that reduction of our dyes requires more negative potentials than the reduction of MV2+ as indicated by the positions of the ELUMO levels (see Experimental section). This hypothesis is supported for BT-Rho, while CPDT-Fur has a reduction potential near that of methyl viologen.
The voltammetric response of the diamond foam electrodes is interestingly different. In contrast to the flat BDD film, the pristine foam shows quasi-reversible waves of the MV2+/MV+ redox couple like on ITO. The redox-blocking by the dyes is again seen, albeit it is not as perfect as on the flat films. We ascribe both effects to the presence of sp2-carbon impurities (graphite and trans-polyacetylene; see Fig. S3† and discussion of Raman spectra) which do not anchor dyes, and presumably to incomplete dye coverage of the hollow diamond spheres as shown in Fig. 1 and S2.† Fig. 2 (dashed color curves) further shows the voltammograms of sensitized electrodes, which passed the long-term photoelectrochemical test (several hours of chopped illumination with white light under negative bias). These voltammograms are discussed below.
The observed photocurrent densities on diamond foam are approximately 3-times larger than those on flat diamond, if we take into account the different light intensities in the reference work.19 However, that the photocurrent density is not enhanced proportionally to the increase of Rf, as it follows from the surface area enlargement (Section 3.2.). This is likely due to incomplete dye coverage, which is caused by the presence of sp2 impurities and the complicated morphology of our foam electrodes (incomplete dye coverage on the foam is also corroborated by the partial dislodging of the MV2+/MV+ redox reaction, see Fig. 2 and discussion thereof).
To investigate the long-term stability of the electrodes, we have repeated the experiment as in Fig. 3, but on a larger timescale and with a 5-times larger light intensity (100 mW cm−2; 1 sun). The corresponding plot is shown in Fig. 4 for the BT-Rho sensitizer. The response of the BDD foam electrode sensitized with CPDT-Fur was similar (Fig. S7, ESI†).
The blank (sensitizer-free) foam shows stable photocurrent densities of ca. 300 nA cm−2 at these conditions (light intensity, bias; see black curve in Fig. 4). Cathodic photocurrent densities of ca. 5–10 nA cm−2 were previously reported for a blank (non-sensitized) BDD electrode with flat surface and ca. 10-times smaller light intensity.16,19 Fig. S8 (ESI†) compares our pristine foam and flat electrodes. Under the applied experimental conditions (white light of 20 mW cm−2 power and bias of −0.3 V), the foam exhibits a stable cathodic photocurrent of 50–60 nA cm−2, while the flat diamond delivers ca. 20 nA cm−2. Interestingly, we observe a factor of ≈3 enhancement for the foam (as for the sensitized photocurrent, but this might be just a coincidence). The cathodic photocurrent under sub-bandgap illumination of blank BDD has been ascribed to either sp2 carbon impurities or to specific states related to B in the lattice.38 In spite of its unclear origin, the onset of the cathodic photocurrent transients scales with pH and the flatband potential of BDD.38 We should note that the diamond foam indeed contains a significant proportion of sp2 carbon impurities (trans-polyacetylene and graphitic carbon)27,28 (cf. Fig. S3†).
The most striking effect, observed in both Fig. 4 and S7† is the huge enhancement of photocurrent simply when the chopped illumination progresses for hours. After 1 day of this ‘aging-activation’, the photocurrent density on the BT-Rho sensitized foam (which was initially ca. 2 μA cm−2 for the fresh electrode) increased to ca. 15 μA cm−2 (Fig. 4). Our champion electrode exhibited a maximum photocurrent density of about 22 μA cm−2; the corresponding plot is shown in Fig. S9 (ESI†). Furthermore, sharp current transients are observed at the light on/off events, which were missing in the fresh samples (Fig. 3). The on/off current transients are often observed on dye-sensitized semiconductor electrodes and are attributed to sluggish charge-transport kinetics, in particular to slow diffusion of the molecules of the electrolyte redox-mediator (MV2+ in our case).1 The absence/occurrence of current spikes in Fig. 3 and 4, respectively, can be simply attributed to the different time scales of both experiments. This is illustrated in Fig. S10 (ESI†), showing that the current spikes in Fig. 4 develop during ca. 50 s, which is longer than the time interval of light on/off switching used in Fig. 3.
To the best of our knowledge, the observed photocurrent densities of 15–22 μA cm−2 at 1 sun illumination are the largest, among all values for sensitized diamond electrodes previously reported.9,12–16,19 This remarkable photoelectrochemical activity is ascribed to the unique morphology of the diamond foam electrodes with an enhanced surface area for light-harvesting.
Details of the above-described ‘aging-activation’ were further explored for electrodes at various stages of the photoelectrochemical treatment. Fig. 5 shows the plots for the BT-Rho sensitized foam electrode. The corresponding plot for the CPDT-Fur foam electrode is similar (Fig. S11 in ESI†). Again, the current transients are not expressed when the light chopping is fast (10 s dark/light). For the aging-activated electrodes, the photocurrent density scales linearly with the light intensity (between 0.1 and 1 sun, as shown in Fig. 5 and S11†). The effect of applied voltage is illustrated in Fig. S12 (ESI†) for an aged BT-Rho electrode. As expected,16,19 the photocurrent density increases with negative bias.
We further explored the question whether or not the photoelectrochemical ‘aging-activation’ is specific for the diamond foam only. To this purpose, the long-term behavior of the sensitized flat BDD films was analyzed as well. The samples were identical to those investigated in our earlier paper.19 Fig. S13 (ESI†) presents an example of a BT-Rho-modified flat diamond film. The photocurrent density increase is again obvious. This confirms that the photoelectrochemical ‘aging activation’ is not specific for the diamond foam, but it occurs on the flat films, too.
To further explore the effect of ‘aging-activation’ we measured the action spectra (IPCE vs. wavelength) at various stages of activation (Fig. 6). IPCE is defined by the equation:
![]() | (3) |
IPCE = ηinj(1 − 10−Γε) | (4) |
The IPCEs and photocurrent densities observed in this study are better than those, previously achieved for sensitized BDD electrodes.9,12–16,19 The blank (non-sensitized) foam expectedly shows negligible IPCEs over the whole spectral region studied from UV to near-IR. The fresh sensitized electrode exhibits an IPCE/wavelength spectrum which resembles the UV-Vis optical spectrum of the fresh BT-Rho dye. The pure dye has a main absorption peak at 445 nm and a second weaker one at 320 nm (in ethanolic solution), but the spectrum shows significant changes upon illumination with white light at 1 sun intensity (Fig. S14a in ESI†). The main visible peak attenuates, and the UV peak broadens toward shorter wavelengths. Similar spectral changes are observed also for a thin film of the BT-Rho dye (Fig. S14b in ESI†). The observed spectral variations resemble those of the IPCE/wavelength spectra upon aging (Fig. 6).
The maximum IPCEs gradually increase and shift towards the UV part of the spectrum. Qualitatively, this leads to a hypothesis that the structure of the dye and/or diamond-dye surface complex is changing, but an in-depth analysis of these structural changes and their effect on photocurrent generation is beyond the scope of this paper. We only note that the electrochemical properties of the dye-sensitized electrodes change too, upon aging. As discussed in Section 3.2, (Fig. 2), the aged electrodes were considerably more active towards the MV2+/MV+ redox couple as compared to the response of fresh sensitized diamonds. This effect is particularly expressed for the foam electrodes. However, the physical interpretation of the aging-activation of the dye-sensitized diamond surface is unclear, and it will be addressed by refined analytical and structural studies in the near future.
Chemical derivatization of the flat diamond surface by BT-Rho and CPDT-Fur caused a complete blocking of the methyl viologen (MV2+/MV+) redox reaction, whereas the redox-blocking effect was less perfect for the sensitized foam electrodes. This is ascribed to the presence of sp2-carbon impurities, such as trans-polyacetylene and graphite-like carbon, which are detected by Raman spectroscopy. Cathodic photocurrent densities of ca. 500–700 nA cm−2 were observed at −0.2 V bias for a fresh foam electrode sensitized with both CPDT-Fur and BT-Rho illuminated by white light (20 mW cm−2; simulated AM 1.5G solar spectrum). These photocurrent densities are approximately 3 times larger than those on flat diamond, which is attributed to the enhanced surface area of the foam electrodes.
Long-term (1–2 days) illumination of the sensitized foam electrodes with chopped white light at 1 sun intensity and −0.3 V bias caused a significant increase of cathodic photocurrent densities to values of ca. 15–22 μA cm−2. These are the largest photocurrent densities reported so far for dye-sensitized diamond electrodes. The photocurrent densities scale linearly with light intensity (between 0.1 and 1 sun). The activation is accompanied with characteristic changes of the dark voltammogram of the MV2+/MV+ redox couple and with gradual changes of the IPCE spectra. The latter initially resemble the optical UV-Vis absorption spectrum of the dye, but subsequently the maximum IPCEs increase and shift towards the UV. The photoelectrochemical ‘aging-activation’ is expressed also on flat diamond films, but detailed physical interpretation of the effect remains elusive.
Footnote |
† Electronic supplementary information (ESI) available: SIMS and Raman spectra of diamond foam electrodes, Kr-adsorption isotherm, UV-Vis spectra, additional SEM images and electrochemical data. See DOI: 10.1039/c5ra12413a |
This journal is © The Royal Society of Chemistry 2015 |