J. Phys. Chem. B 2005, 109, 12837-12844
12837
Electrochemical and Electrochromic Properties of Layer-by-Layer Films from WO3 and Chitosan Fritz Huguenin,*,† Ernesto R. Gonzalez,‡ and Osvaldo N. Oliveira Jr.§ Departamento de Quı´mica, Faculdade de Filosofia, Cieˆ ncias e Letras de Ribeira˜ o Preto - UniVersidade de Sa˜ o Paulo, 14040-901 Ribeira˜ o Preto (SP), Brazil, Instituto de Quı´mica de Sa˜ o Carlos - UniVersidade de Sa˜ o Paulo, 13560-970 Sa˜ o Carlos (SP), Brazil, and Instituto de Fı´sica de Sa˜ o Carlos - UniVersidade de Sa˜ o Paulo, CP 369, 13560-970 Sa˜ o Carlos (SP), Brazil ReceiVed: January 24, 2005; In Final Form: May 13, 2005
The design of improved materials for electrochromic applications now involves extensive use of novel composites, thus requiring an investigation of the mechanisms responsible for electrochromism in these structures. Using films of WO3 and chitosan produced with the layer-by-layer (LBL) technique, we demonstrate that characteristics such as the number of electrochemical active sites (K), the molar absorption coefficient (), and the electrochromic efficiency (η) can be obtained using the quadratic logistic equation (QLE). The complexation ability between chitosan and WO3 allowed the growth of visually uniform multilayers of the composite, with the same amount of material adsorbed in each deposition cycle. By fitting the absorbance changes (∆A) resulting from the electronic intervalence transfer from W(V) to W(VI) sites in four-bilayer LBL films of WO3/chitosan and WO3/chitosan with ethanol in the precursor dispersion, K was estimated to be ca. 5.5 × 10-8 mol cm-2 and 3.6 × 10-8 mol cm-2, respectively. The molar absorption coefficient and electrochromic efficiency vary with the charge injected because of the saturation of W(V) sites and the dissipation and feedback effects implicit in the QLE associated with ion-network interactions, such as the proton trapping effect. The LBL film of WO3/chitosan showed a smaller molar absorption coefficient and electrochromic efficiency than that containing ethanol because of a greater proton trapping effect for the LBL film with no ethanol. This enhanced trapping effect was seen as a decrease in the electronic flux involved in intervalence transfer in electrochemical impedance spectroscopy experiments.
Introduction WO3 has been extensively investigated for a number of applications, including electrocatalysis,1 photoelectrochemistry,2,3 energy storage,4 photocatalysis,5 sensors,6 gas chromism,7 and photochromism.8 Other applications exploit the electrochromic properties9 because tungsten ions from WO3 exhibit different oxidation states, and the intervalence electron transfer from W(V) to W(VI) states produces a broad absorption in the red region of the visible spectrum. Thus, WO3 displays a color change from white to blue when reduced, which is affected by the number of ions inserted into the oxide matrix, through the process
WO3 + xM+ + xe- f MxWO3
(1)
where M+ can be H+, Li+, K+, and so forth. For each electron injected (removed) into (from) the conduction band, a cationic species is inserted (removed) into (from) the structure to compensate for the charge. The Coulombic reversibility and the coloration efficiency observed for this material are used with advantages in display devices and smart windows.10,11 It is therefore important to analyze the dependence of the WO3 optical properties with respect to the ionic insertion/ deinsertion process. Models such as the site-saturation model and the quadratic logistic equation (QLE) have been used to * Corresponding author. E-mail:
[email protected]. † Departamento de Quı´mica. ‡ Instituto de Quı´mica de Sa ˜ o Carlos. § Instituto de Fı´sica de Sa ˜ o Carlos.
predict the change in absorbance (∆A) and the electrochromic efficiency (η, absorbance change per unit charge) as a function of charge injected (q).12,13 The QLE has been used to describe population growth in biological systems and to model physicochemical processes. The differential form of the QLE is
dx x ) sx 1 dt K
(
)
(2)
where x is the population at time t, dx/dt is the rate of change of the population, the constant K is associated with the system’s ability to sustain the population growth, and s is a positive constant. The (1 - x/K) term opposes the temporal growth of the population and is associated with the feedback effect. The QLE has been used to assess temporal changes in the properties of electrochemical systems.14-17 For instance, it has been used to rationalize the dissipation and feedback effects associated with the transport of charge. In fuel cells, the QLE was used to interpret limitations in charge transport,16 which was shown in a modeling study to be due mainly to the transport of protons in the ionic exchange membrane.18 The concepts used in that work were extended to interpret experimental results in intercalation processes,17 as in the intercalation of Li ions in V2O5 films.13 In the context of this paper, the QLE is employed to describe the evolution of electron transfer in WO3/chitosan LBL films, which are donor-bridge-acceptor (DBA) systems. Oxygen atoms between the tungsten ions are the bridges, and the W(V) and W(VI) sites are the donors and acceptors, respectively. The mathematical expression that relates the change in absorbance to the charge injected is a direct application of
10.1021/jp0504165 CCC: $30.25 © 2005 American Chemical Society Published on Web 06/10/2005
12838 J. Phys. Chem. B, Vol. 109, No. 26, 2005
Huguenin et al.
the QLE in its difference form:
(
∆A ) rq 1 -
q K
)
(3)
The constant K corresponds to the system’s ability to sustain the increase in absorbance, representing the maximum value of q, and r is a positive constant. The term (1 - q/K) represents the effects from dissipation and feedback, which increase with q. In comparing eq 2 with eq 3, it is implied that the charge injected is the population and ∆A is related to dq/dt. In fact, the magnitude of the absorbance in the visible range depends on the frequency of the charge (electron) transfer between W(V) and W(VI) sites, in addition to the dependence on the amount of charge injected. The electrons that participate in this chargetransfer process are those injected electrochemically. As the charge injected increases so does the number of transferred electrons from W(V) to W(VI) sites and the number of intercalated ions, which are transported simultaneously to maintain electroneutrality. This increase leads to interactions between the ions and to interactions between the ions and the network with the appearance of dissipative forces. At the same time, the dissipation effect increases with increasing numbers of charge carriers, which contributes to the feedback effect. That is to say, the dissipative forces opposing the electron transfer increase with the number of charge carriers, decreasing the electrochromic efficiency. For these DBA electron-transfer systems, features such as the bridge chemical structure, the DBA energy gap, and the donor-acceptor distance determine the electron-transfer rate.19 Parameters related to these features are included in the constant r of eq 3. The feedback effect is also caused by the decrease in the number of sites available for the intervalence transfer, as described in the site-saturation model.12 This model also assumes that the absorbance is due to an electron transfer from one site to a neighboring empty site. Therefore, as the number of electrons and ions increases in the film, a higher fraction of neighboring sites from a given electron is occupied, thus decreasing the electrochromic efficiency. In the experiments described here, during the insertion of protons, W(V) sites are formed, and ∆A increases because of a higher intervalence transfer W(V) f W(VI) flux. Upon further increasing the charge injected, the intervalence transfer W(V) f W(VI) flux decreases because of an excess of W(V) sites in comparison with the number of W(VI) sites. In this work, the QLE is employed to characterize the optical and electrochemical properties of novel composites from WO3 and chitosan obtained with the layer-by-layer (LBL) technique. Proton trapping is discussed on the basis of electrochemical impedance spectroscopy data and is associated with dissipation and feedback effects, which are described by the QLE. From the analysis, we can predict the molar absorption coefficient () and the electrochromic efficiency, which may be important factors for the design of layer-by-layer structures with improved properties particularly because of the possibility of molecularlevel manipulation of the LBL films. Indeed, recent works have indicated that LBL films are promising for microbatteries and electrochromic devices because of enhanced charge-storage capability and uniformity.20-22 The choice of chitosan to build the LBL films was based on its ability to form complexes with oxides in solutions,23,24 which allows uniform growth as a function of the number of bilayers. Chitosan is colorless for the small amount deposited in the LBL method and does not affect the electrochromic properties of the transition-metal oxide
Figure 1. UV-visible spectra of LBL films from WO3/chitosan. In the inset, the absorbance at 590 nm is plotted vs the number of bilayers for WO3/chitosan.
because the complexes formed from chitosan can ensure a high proton conductivity.25 Experimental Section WO3 xerogel was synthesized following a sol-gel method.26,27 Briefly, a 0.25 mol L-1 aqueous solution of sodium tungstate (Na2WO4) was added to a proton-exchange resin at room temperature, leading to a yellowish liquid containing tungstic acid (H2WO4). Protonation leads to hydroxy species. Then, condensation occurs via oxolation, and polyanions are formed during the first condensation steps. This dispersion became turbid as time evolved and became a gel. A precipitate was formed after a few hours. Because of the fast condensation of oxide particles, 1 cm3 of ethanol was added to some of the samples as a stabilizing agent. The dispersion also became turbid, but the process was slower than without ethanol. These samples are referred to as et-WO3/chitosan. Chitosan was purchased from Aldrich. It is the deacetylated form of chitin, which is soluble, and a polycation in acid media because of the protonation of the -NH2 groups.25 The aqueous dispersion of chitosan was obtained with the addition of 1.6 g of chitosan in 1 L of HCl solution at pH 2. Layer-by-layer (LBL) films were assembled onto indium-tin oxide (ITO)-coated glass purchased from Flexitec (Curitiba, Brazil), with a sheet resistance e20 Ω and a geometrical area of 1 cm2. The layers were obtained via ionic attraction of oppositely charged materials by the alternating immersion for 3 min of the ITO substrate into the polycationic (chitosan, 1.6 g L-1) and anionic (WO3) dispersions. This WO3 dispersion was used 2 h after Na2WO4 was made to pass through the ion-exchange resin and before gel formation. After the deposition of each layer, the substrates were rinsed for 30 s in HCl solution (pH 2) and dried under a nitrogen flow at room temperature. The film growth was monitored by measuring the ultravioletvisible (UV-vis) absorption spectra of the samples with a Hitachi U-3501 spectrophotometer. The absorbance change showed that the amount of WO3 and chitosan adsorbed in a WO3/chitosan LBL film increases linearly with the number of bilayers. For the electrochemical and spectroelectrochemical experiments, the counter electrode was a platinum sheet with an area of 10 cm2, and the quasi-reference electrode was saturated calomel. An electrolytic solution of HCl (pH 2) was used in all electrochemical experiments that were carried out with an Autolab PGSTAT30 potentiostat/galvanostat. Chromogenic analysis was carried out simultaneously with the electrochemical experiments using a microprocessor-controlled
Properties of Layer-by-Layer Films
J. Phys. Chem. B, Vol. 109, No. 26, 2005 12839
Figure 2. (s) Current density, j, and (O) absorbance changes, ∆A, at (a) 470, (b) 525, (c) 590, (d) 623, and (e) 660 nm for four-bilayer LBL films of WO3/chitosan. V ) 50 mV s-1.
solid-state light source (WPI, Inc.). Plastic fiber optic cables up to 1 mm in diameter were used to deliver red light (660 nm) from the instrument to a PDA1 photodiode amplifier (WPI, Inc.). The films were placed in a cell made of optical glass for the transmission experiments, where the light beams at fixed wavelengths were transmitted across the film during cyclic voltammetry. The experimental data were fitted using Origin software, with the values of r and K in QLE as fitting parameters. Results and Discussion The absorption spectra of as-prepared WO3/chitosan LBL films for varying numbers of bilayers are presented in Figure 1 and correspond to films in their bleached state. The profile is similar to that of a cast WO3‚xH2O film in the bleached state.28 The absence of an absorption band above 550 nm due to electron intervalence transfer between W(V) and W(VI) sites (observed in colored WO3 films)10,29,30 indicates that the as-prepared WO3/
chitosan LBL films contain a negligible number of W(V) sites. The inset in Figure 1 shows the absorbance at 590 nm versus the number of bilayers, which indicates that the amount of WO3 and chitosan adsorbed in the film increases linearly with the number of bilayers. This is a clear indication of the outstanding influence of chitosan for the building of uniform LBL films containing metal oxides. Figure 2a-e shows five identical voltammograms obtained consecutively at 50 mV s-1 for four-bilayer LBL films from WO3/chitosan, which depicts the proton insertion process during the negative potential scan and the proton deinsertion process during the positive potential scan. The charge associated with the deinsertion process decreased ca. 29% after 1000 voltammetric cycles. These voltammograms were taken simultaneously with in situ measurements of absorbance changes (∆A), also shown in Figure 2, for different wavelengths: (a) 470, (b) 525, (c) 590, (d) 623, and (e) 660 nm. The largest absorbance occurs at 590 nm, which was then used in the subsequent experiments.
12840 J. Phys. Chem. B, Vol. 109, No. 26, 2005
Huguenin et al.
Figure 4. (a) (s) Current density, j, and (O) absorbance changes, ∆A, at 590 nm for four-bilayer LBL films of WO3/chitosan. V ) 50 mV s-1. (b) Plot of the (s) theoretical and (O) experimental ∆A data at 590 nm as a function of charge.
Figure 3. (a) Current density, j, and (b) absorbance changes, ∆A, at 590 nm for various scan rates for four-bilayer LBL films of WO3/ chitosan: V ) 10, 20, 50, 100, and 200 mV s-1. The arrow points to increasing scan rates from 10 to 200 mV s-1. (c) ∆A for different scan rates normalized by ∆A at 10 mV s-1.
Part a of Figure 3 displays the cyclic voltammograms, and part b of Figure 3 displays the absorbance changes at 590 nm for four-bilayer LBL films from WO3/chitosan, obtained using several scan rates (V). The voltammetric profiles depend on the scan rate, with shifts of the peak currents and of the potential corresponding to the maximum ∆A. There are potential ranges in which reduction and oxidation processes are thermodynamically allowed, but these processes do not take place instantaneously because of the slow proton diffusion into the material. Therefore, the current peak is shifted as the scan rate is increased. Despite this slow diffusion, chitosan appears to allow the migration/diffusion of protons to compensate for the charge injected into the structure because ∆A, which is associated with the number of W(V) and protons inserted, increases for lower scan rates (longer time). The migration/diffusion of protons in chitosan has been reported in the literature.25,31 In addition, the ∆A data show that the protons inserted are deinserted in the reverse process. The difference in ∆A before and after each of the cycles is negligible. We could also use the charge (q) profile
for these analyses, but q includes a contribution from the water reduction reaction and the non-faradaic charging that affect the results. Figure 3c shows how ∆A/∆AV)10 mV/s (∆A for different scan rates normalized with respect to ∆A at 10 mV s-1) varies with the scan rate. The curves were fitted with an empirical exponential function A0e-V/B, where A0 () 0.29992) and B () 58.12265 s mV-1) are constants and V is the scan rate. Extrapolating the curve for an infinite scan rate, the value of ∆A/∆AV)10 mV/s tends to that of the preexponential A0 factor. This approximation indicates that ca. 30% of the faradaic charge at 10 mV s-1 (0.54 mC.cm-2) is on the film surface, which contains 1.67 × 10-9 mol cm-2 of electroactive W(VI) sites. The total number of electroactive oxide sites can be probed by monitoring the light absorption associated with the intervalence transition between W(V) and W(VI) sites. Here, it is assumed that the number of W(IV) sites is very small and that the electronic transition occurs between W(V) and W(VI) sites, as mentioned in the literature.11,32,33 The energy distribution in these materials and the slow ionic diffusion in the solid state avoid the total reduction of W(VI) sites for potentials more positive than that for the reduction of H2O. This implies a mixed reaction, which hinders the determination of the number of electroactive W(VI) sites. However, the method employed here avoids the influence from the reduction of water. The K value is obtained by extrapolating the curve for a potential range higher than that for the water reduction reaction. Figure 4a shows the cyclic voltammograms and ∆A data from -0.5 to 0.3 V at 50 mV s-1 for a four-bilayer WO3/chitosan LBL film. The corresponding absorbance is shown in Figure 4b as a function of charge. In this case, the charge during the deinsertion process
Properties of Layer-by-Layer Films
J. Phys. Chem. B, Vol. 109, No. 26, 2005 12841
Figure 6. Plot of the theoretical electrochromic efficiency and the theoretical molar absorption coefficient at 590 nm as a function of charge for (s) WO3/chitosan and (‚ ‚ ‚) et-WO3/chitosan.
Figure 5. (a) (s) Current density, j, and (O) absorbance changes at 590 nm vs applied potential for four-bilayer LBL films of et-WO3/ chitosan. V ) 50 mV s-1. (b) Plot of the (s) theoretical and (O) experimental ∆A data at 590 nm as a function of charge for a fourbilayer LBL films of et-WO3/chitosan.
was plotted, which avoids the contribution from the water reduction reaction. A wider potential range (from -0.5 to 0.3 V) was used to probe the nonlinear profile in a more refined way. The estimated values of K and r were 5.3 mC cm-2 and 38 cm2 C-1, respectively, from which it can be estimated that the number of electroactive sites in the composite is ca. 5.5 × 10-8 mol cm-2. It is known that the condensation process of tungstic acid H2WO4 is too fast, and a gel or precipitate is obtained within a few hours.27 To avoid this fast condensation, ethanol was used to stabilize the colloidal dispersion. This is essential for preparing LBL films with negligible light-scattering properties due to the formation of small particles. Even though the amount of oxide increases linearly with the number of bilayers (as shown in Figure 1), films with several bilayers exhibit nonlinear growth due to aggregation in the WO3 precursor colloidal dispersion. Furthermore, ethanol alters the dielectric constant of the medium and, consequently, the interaction between chitosan and WO3. Figure 5a displays the cyclic voltammogram (a) and absorbance changes at 590 nm (b) for LBL films from et-WO3/chitosan. The results are similar to those for the WO3/chitosan film, except for the lower current density. The potentials at the current peaks in the reduction process are -0.39 and -0.43 V for et-WO3/ chitosan and WO3/chitosan, respectively. In the oxidation process, these potentials are -0.21 and -0.18 V for et-WO3/ chitosan and WO3/chitosan, respectively. These results indicate that proton intercalation/deintercalation is more reversible (faster) in et-WO3/chitosan than in WO3/chitosan, which means that diffusion/migration effects are not responsible for the difference in current density. The difference is attributed to the
smaller amount of oxide in the et-WO3/chitosan in comparison to that in the WO3/chitosan film due to the addition of ethanol, which changes the dielectric constant of the medium and the interaction between WO3 and chitosan. The number of electroactive sites in et-WO3/chitosan was also determined using the QLE, presented in eq 3, as shown in Figure 5b. The curves exhibit the same profile, regardless of the range investigated. K and r were 3.5 mC cm-2 and 50 cm2 C-1, respectively. With this value of K, the number of electroactive W(VI) sites in the et-WO3/chitosan composite was calculated to be 3.62 × 10-8 mol cm-2, which is smaller than that for WO3/chitosan as calculated above. In addition, r is also increased when ethanol is added to the precursor colloidal dispersion, which indicates a higher electrochromic efficiency. Assuming that the ∆A data follow the Beer-Lambert law, the molar absorption coefficient () and the electrochromic efficiency can be obtained from
∆A ) l∆C
(4)
where l is the film thickness in cm and C is the concentration of the absorbent species in mol cm-3, which can be associated with the charge injected
q ) znF
(5)
where n is the number of mols of redox species per unit area, z is the charge of this species, and F is the Faraday constant. Therefore, q/zF is equivalent to l∆C, and and η can be related by34
η)
∆A ) q zF
(6)
Considering that absorbance changes follow eq 3 and using the Beer-Lambert law in the differential form, we can write
dA 2rq ) )rdq zF K
(7)
Thus, the molar absorption coefficient and the electrochromic efficiency are given by eqs 8 and 9, respectively:
(
) zFr 1 -
(
η)r1-
2q K
)
2q K
)
(8) (9)
12842 J. Phys. Chem. B, Vol. 109, No. 26, 2005
Huguenin et al.
Figure 7. (a) Nyquist diagrams, (b) phase angle as a function of frequency logarithm, and (c) complex plane representation of the capacitance obtained by applying dc potentials of (-9-) 0.3, (-0-) 0.1, (-O-) 0.0, (-b-) -0.1, and (-4-) -0.3 V for WO3/chitosan. (d) Nyquist diagrams, (e) phase angle as a function of frequency logarithm, and (f) complex plane representation of the capacitance obtained with the same sequence of dc potentials for et-WO3/chitosan.
The constant r corresponds to the electrochromic efficiency in the absence of a negative feedback effect (2q/K , 1). The value of r is in the range quoted in the literature for other preparation methods such as electrodeposition35 and reactive dc magnetron sputtering.10 Considering a feedback effect, a contribution of -2rq/K should be added to the electrochromic efficiency. Figure 6 shows the molar absorption coefficient and the electrochromic efficiency for WO3/chitosan and et-WO3/ chitosan, obtained from eqs 8 and 9. The difference in the molar absorption coefficient and the electrochromic efficiency for the two composites is discussed below, together with the impedance data. Interfacial and bulk processes can be studied with impedance spectroscopy. For the LBL films investigated here, Nyquist
diagrams, the phase angle as a function of log frequency, and the complex plane representation of the capacitance have been obtained by applying dc potentials of 0.3, 0.1, 0.0, -0.1, and -0.3 V with a 5-mV-amplitude ac signal superimposed. The results in Figure 7a for WO3/chitosan show semicircles associated with resistive and capacitive processes in parallel. One of these coupled processes is associated with the interface, which can be represented by the charge-transfer resistance (Rct) and the double-layer capacitance (Cdl). This process is observed for high frequencies in Figure 7b, which shows the phase angle as a function of frequency for WO3/chitosan. There is only a coupled parallel RctCdl at 0.3 V. However, bulk processes appear at low frequencies (
