Redox Grafting of Diazotated Anthraquinone as a Means of Forming

ARTICLE pubs.acs.org/Langmuir

Redox Grafting of Diazotated Anthraquinone as a Means of Forming Thick Conducting Organic Films Antoine Bousquet,† Marcel Ceccato, Mogens Hinge, Steen Uttrup Pedersen,* and Kim Daasbjerg* Department of Chemistry, Aarhus University, Langelandsgade 140, DK-8000 Aarhus C, Denmark Interdisciplinary Nanoscience Center (iNANO), Aarhus University, Ny Munkegade 120, DK-8000 Aarhus C, Denmark

bS Supporting Information ABSTRACT: Thick conductive layers containing anthraquinone moieties are covalently immobilized on gold using redox grafting of the diazonium salt of anthraquinone (i.e., 9,10-dioxo9,10-dihydroanthracene-1-diazonium tetrafluoroborate). This grafting procedure is based on using consecutive voltammetric sweeping and through this exploiting fast electron transfer reactions that are mediated by the anthraquinone redox moieties in the film. The fast film growth, which is followed by infrared reflection absorption spectroscopy, atomic force microscopy, X-ray photoelectron spectroscopy, ellipsometry, and coverage calculation, results in a mushroom-like structure. In addition to varying the number of sweeps, layer thickness control can easily be exerted through appropriate choice of the switching potential and sweep rate. It is shown that the grafting of the diazonium salt is essentially a diffusion-controlled process but also that desorption of physisorbed material during the sweeping process is essentially for avoiding blocking of the film due to clogging of the electrolyte channels in the film. In general, sweep rates higher than 0.5 V s
1 are required if thick, porous, and conducting films should be formed.

’ INTRODUCTION Functionalization of conductive and semiconductive surfaces is important in many fields such as bioelectronics,1 photovoltaic,2 or chemical and biological sensing.3 In particular, modification of carbon and metal electrodes by means of electroreduction of aryldiazonium salts has attracted attention4
6 since the first report on this method appeared 20 years ago.7 The electrochemical reduction promotes the formation of aryl radicals capable of attacking the surface and forming a covalent bond with it.8
11 Because of the high reactivity of aryl radicals, the resulting film usually consists of disordered and electrically insulating multilayers.8,12 The electrochemically driven grafting process is self-limiting with the final film thickness seldom exceeding 10 nm. The aryldiazonium salt approach offers a versatile approach to introduce various functional groups such as nitro,13 alcohol,14 and carboxylic acid15 functionalities at the surface. However, for a number of applications, including sensors and conducting films, the creation of thick layers containing a large number of redox units attached at the surface would be important. In recent years a few methodologies have been developed to create thicker organic layers (e.g., larger than 20 nm). This kind of electrografting, which we coin redox grafting, is carried out either under potentiostatic conditions using long electrodeposition times at high overpotential16
19 or by using repetitive cyclic voltammetry, exploiting in both cases the redox activity of already immobilized electrochemically active groups.19
24 r 2011 American Chemical Society

In our previous work we used redox grafing of aryldiazonium salts pertaining to anthraquinone [i.e., 9,10-dioxo-9,10dihydroanthracene-1-diazonium (AQD)], 4-nitrobenzene, and benzophenone (i.e., 4-benzoylbenzene diazonium) to show that conducting films with thicknesses, even in the micrometer range, could be formed in a controlled manner on glassy carbon, gold, and stainless steel.24 The redox moieties in terms of anthraquinone (AQ), 4-nitrobenzene, and benzophenone are the central units serving to transport electrons from the electrode surface through the film created to the aryldiazonium molecules at the outer layer during the grafting process. The grafting efficiency is dependent on the ability of the radical anion of the redox unit to reduce the corresponding aryldiazonium salt while at the same time having the required chemical stability against being irreversibly protonated by the residual water present in the grafting medium (usually 0.1 M Bu4NBF4/CH3CN). The development of this kind of film using redox grafting is not only of interest from a mechanistic point of view but also because of the potential prospect of using, in particular, the AQ system as supercapacitor25
27 or electrocatalyst toward the reduction of oxygen.28
30 Ultimately, the successful development of these two applications would be dependent on having methodologies enabling the introduction of a large number of electroactive Received: September 19, 2011 Revised: November 11, 2011 Published: December 16, 2011 1267

dx.doi.org/10.1021/la203657n | Langmuir 2012, 28, 1267–1275

Langmuir Scheme 1. Formation of AQ-Based Films Obtained from Electroreduction of AQD at Three Different Potentials Corresponding to the Reduction of the Diazonium Salt Itself and the First and Second Reduction of the AQ Moiety

AQ moieties in the film. So far, electrografting of AQD has been mainly used to deposit thin organic layers on carbon,25,29
33 gold,34 and nickel35 electrodes as well as on carbon particles.36 With the 4-nitrobenzenediazonium salt already investigated,24 we wished at this point to shed further light on the characteristics of the thick films created by redox grafting of AQD on Au using repetitive cyclic voltammetry. Elucidation of the role of the sweep rate, the number of sweeps, and the switching potential is carried out with the important notion that the switching potential selected decides if a one-electron reduction (of
N2+), twoelectron reduction (of
N2+ and AQ f AQ•
), or threeelectron reduction (of
N2+ and AQ f AQ•
f AQ2
) takes place during sweeping, as illustrated in Scheme 1. In any of the cases the result of the sweeping is the covalent attachment of multilayers of AQ at the surface upon expulsion of dinitrogen but with film thicknesses that are very much dependent on the switching potential, in that the layer becomes substantially thicker, if the redox activity of AQ is exploited. Cyclic voltammetry, infrared reflection absorption spectroscopy (IRRAS), ellipsometry, X-ray photoelectron spectroscopy (XPS), and atomic force microscopy (AFM) were employed in the characterization of the produced films.

’ EXPERIMENTAL SECTION Materials. Acetonitrile (anhydrous, 99.9%, Lab-Scan) was dried prior to electrochemical experiments by passing it through a solvent purification system (MB SPS-800; MBRAUN). After this treatment the water content was determined to be 18
20 ppm by Karl Fischer titration (851 Titrando; Metrohm). Water was triple distilled. Tetrabutylammonium tetrafluoroborate (Bu4NBF4) was synthesized using standard procedures. 9,10-Dioxo-9,10-dihydroanthracene-1-diazonium tetrafluoroborate (AQD) was synthesized from the corresponding amine according to standard procedures.37 Further purification consisted in a recrystallization from acetonitrile/diethyl ether, filtration, vacuum drying, and storing at
18 °C. 1-Aminoanthracene-9,10-dione, ferrocene, ferricyanide, and ferrocyanide (Sigma-Aldrich) were used without further purification. Substrates. Gold plates (glass plates coated with 20 nm Ti and 200 nm Au using physical vapor deposition, 10 mm  10 mm  1 mm, Polyteknik, Denmark) were immersed in a piranha solution, which

ARTICLE

comprises 1:3 (v:v) 30% H2O2:H2SO4 (caution: piranha solution is a very strong oxidant and should be handled with extreme care), at room temperature for 5 min. Subsequently, the plates were rinsed several times in triple distilled water, followed by sonication in acetone and ethanol (10 min in each case). Finally, the plates were dried in a stream of argon. Gold rods (diameter = 1 mm) were embedded in epoxy resin, sanded (with sandpaper grid P2000), and carefully polished with diamond suspensions (Struers, grain sizes 9, 3, 1, and 0.25 μm) and successively with 0.05 μm alumina slurries prepared in Milli-Q water on a microcloth pad. After removal of trace alumina from the surface by rinsing with water and cleaning in an ultrasonic bath with ethanol, the electrodes were cleaned with a piranha solution for 2 min. An electrochemical cleaning in 0.01 M H2SO4 was carried out by cycling the electrodes between
0.2 and 1.45 V, until a reproducible voltammogram was obtained. Before derivatization, the cleaned electrodes were rinsed with triple distilled water and ethanol (99.99%) and successively dried under a stream of argon. Cyclic Voltammetry. A standard three-electrode electrochemical setup (CH Instrument 660B or CH Instrument 601C) consisting of a gold electrode as working electrode (geometrical surface area = 0.0079 cm2 for the gold rods and 0.6 ( 0.1 cm2 for the part of the gold plate being immersed into the solution), a platinum wire as auxiliary electrode, and a Ag/AgNO3 as reference electrode was used in the electrochemical experiments. At the end of each experiment performed in CH3CN, the standard potential of the ferrocenium/ferrocene couple, 0 +, was measured, and all potentials were referenced against SCE using EFc 0 38 + = 0.41 V vs SCE in CH3CN. a previous determination of EFc Electrografting was performed by way of cyclic voltammetry at a sweep rate = 1 V s
1 using repetitive cycles going usually from 0.55 V vs SCE to a predetermined switching potential Eλ =
0.35,
1.5, or
2.1 V vs SCE. The grafting solution consisted of 2 mM AQD in dry 0.1 M Bu4NBF4/CH3CN (degassed using argon). After electrografting, the gold disk electrodes were ultrasonicated for 10 min in acetone and dried under a stream of argon. Unfortunately, this cleaning procedure could not be easily applied in the case of the gold plates because of the risk of delaminating the gold sputtered layer from the glass plate. Therefore, the plates were rinsed several times in acetonitrile and dried in a stream of argon. Importantly, in the few experiments, where ultasonication was successfully applied, the resulting decrease in film thickness was less than 10%. Determination of Charge and Surface Coverage. The electroactivity of the anthraquinone-grafted gold disk electrodes was evaluated from the first wave in a cyclic voltammogram recorded in 0.1 M Bu4NBF4/CH3CN. The total charge (Q) used for this oneelectron reduction of the surface attached anthraquinone was obtained by numerical integration of the background subtracted electrochemical response recorded in cyclic voltammetry. First, the background was subtracted using a fourth-degree polynomial (obtained from a fitting of the data on either side of the wave) in order to describe the steadily rising of the background signal usually occurring while surpassing the potential region of the peak. The charge Q was determined by numerical integration of the electrochemical signal, thereby allowing a calculation of the surface coverage ΓAQ = Q/nFA, where F is the Faraday constant, A the electrode area, and n the number of electrons (n = 1). It is important to note that for film-coated electrodes there is an uncertainty associated with establishing an appropriate baseline since the properties of the film change during sweeping. For this reason the uncertainty could be as high as 20%.

Infrared Reflection
Absorption Spectroscopy (IRRAS). IRRAS spectra were recorded on a Bio-Rad FTS 65A (Randolph, MA) FTIR spectrometer equipped with an external experiment module with a narrow band mercury
cadmium
telluride (MCT) detector cooled in liquid nitrogen. The infrared beam was p-polarized by a gold wire 1268

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir

ARTICLE

analyzer ellipsometer (Dre, Germany). The gold plates were measured at 75°. The ellipsometric parameters of the bare (Δs, ψs) and grafted (Δg, ψg) substrates were measured in air at ambient temperature, where Δ is the phase shift and tan(ψ) is the amplitude ratio upon reflection. The complex refractive index of the bare substrate was calculated from the measured Δs and ψs values. A three-layer optical model consisting of a substrate with a complex refractive index, the grafted layer given by its refractive index and thickness, and the surrounding medium (air) was used to calculate the overall reflection coefficients for in-plane (Rp) and out-of-plane (Rs) polarized light. The real and the imaginary parts of the refractive index of the bare substrate were obtained by measuring the clean plates prior to modification. Ellipsometric measurements were performed on the same area of the plates before and after electrografting. Because the measurements are carried out on a dried and thus collapsed film, the refractive index of the layer is fixed at a constant value (real = 1.55; imaginary = 0), independent of the thickness. The average and the standard deviation values reported correspond to data points obtained from measuring three spots on each plate. X-ray Photoelectron Spectroscopy (XPS). A Kratos Axis UltraDLD instrument operated with a monochromatic Al Kα X-ray source at a power of 150 W was employed for the XPS analysis. Survey scans were acquired by accumulating two sweeps in the 0
1100 eV range at a pass energy of 160 eV. High-resolution scans were acquired at a pass energy of 20 eV. The pressure in the main chamber during the analysis was in the range of 10
11 bar. The generated XPS data were processed using the CasaXPS software. Atomic surface concentrations were determined by fitting the core spectra using Gaussian line shapes and a linear background. Binding energies of the components in the spectra were determined by calibrating against the C
H/C
C peak in the C 1s spectra at 285.0 eV. The systematic error is estimated to be of the order of 5
10%. Atomic Force Microscopy (AFM). AFM analysis was carried out in air using a NanosurfeasyScan 2 AFM instrument (Nanosurf, Switzerland) operated in noncontact tapping mode with resonant frequency of 146
236 kHz. PPP-NCLR-20 and AFM cantilevers were purchased from Nanosensors (Switzerland). The spring constant of the cantilever was 21
98 N/m. The images were recorded with a scan rate of 170
256 pixels/s, resolution of 256 pixels/line, and scan size of 5  5 μm2. Average values of root-mean-squared roughness, Rq, were determined from three selected areas (5  5 μm2), using the Nanosurf Easyscan 2 software.

’ RESULTS AND DISCUSSION Figure 1. Ten consecutive voltammograms recorded at a gold plate on 2 mM AQD using ν = 1 V s
1, Ei = 0.55 V vs SCE, and Eλ = (A)
0.35, (B)
1.5, and (C)
2.1 V vs SCE in 0.1 M Bu4NBF4/CH3CN. For the sake of clarity, only the 1st (solid), 2nd (dashed), 3rd (dot), and 10th (dash
dotted) cyclic voltammograms are shown in (A). polarizer. The spectral resolution and number of scans averaged were 4 cm
1 and 800, respectively, and the substrates were irradiated with an incident grazing angle of 80°. The p-polarized reflectivity of the film, Rp(d), was divided with the reflectivity of the bare substrate, Rp(0), and presented as IRRAS absorbance,
log[Rp(d)/Rp(0)], after baseline correction using the facilities of the Digilab Resolution Pro 4.0 program. The area of the CdO stretching absorbance band was calculated by the program as well. All spectra were recorded at room temperature in dry atmosphere. Ellipsometry. Thicknesses of films (dried under argon flow after ultrasonification) in the dry state were measured using a rotating

I. Effect of Switching Potential. Cyclic Voltammetry. Figure 1 shows 10 consecutive cyclic voltammograms recorded on a 2 mM AQD solution at a gold plate using a sweep rate, ν, = 1 V s
1 and with the sweeping ranging from an initial potential, Ei, = 0.55 V vs SCE, to three different switching potentials, Eλ, = (A)
0.35, (B)
1.5, and (C)
2.1 V vs SCE in 0.1 M Bu4NBF4/CH3CN. These cases correspond to the three situations, where the sweeping is extended progressively to include the reduction of the diazonium salt moiety, the first reduction wave of AQ , and finally the second AQ reduction wave. As seen in Figure 1A, the reduction wave of the diazonium salt (appearing at the reduction peak potential, Ep,r, = 0.18 V vs SCE) is the only one present on the first sweep for Eλ =
0.35 V vs SCE. On the successive cycles the current signal decreases and ultimately disappears which is consistent with progressive blocking of the electrode surface because of the formation of an insulating film. 1269

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir In Figure 1B with Eλ =
1.5 V vs SCE one of the characteristic features, in addition to the signal arisen from the reduction of the diazonium salt, is the appearance of the first redox wave (Ep,r,1 =
1.12 V vs SCE and oxidation peak potential, Ep,o,1, =
0.84 V vs SCE on the first cycle) pertaining to the AQ moiety. Taking the average of the peak potentials as an approximation for the formal 0 0 potential, E0 , gives E01 =
0.98 V vs SCE in agreement with the literature value 0for the one-electron reduction of AQ to AQ•
in acetonitrile (E01 =
0.96 V vs SCE).39 Notably and in contrast to the behavior seen in Figure 1A, the reduction wave of the diazonium salt is still present after 10 cycles, although it is diminished compared to that on the first cycle. The concomitant increase in the intensity of the first redox wave of a AQ moiety can be attributed to both a mediated reduction of the diazonium salt and a steady increase in the number of redox groups attached to the surface. Actually, the prewave appearing at
0.8 V vs SCE is believed to originate from a total catalysis situation,40 where the mediated electron transfer to AQD occurs so fast that the diazonium salt becomes completely consumed in the diffusion layer. Finally, in Figure 1C characterized by the most extended sweeping, i.e., Eλ =
2.1 V vs SCE, the second redox wave of the 0 AQ moiety appears (Ep,r,2 =
1.76, Ep,o,2 =
1.42, and E02 =
1.59 V vs SCE on the first cycle). In this case also, an increase in the intensity of, in particular, the first redox wave as a function of the number of sweeps is noticed, pointing again to the presence of the mediating effect along with an increase in the number of electroactive groups attached to the surface. The relatively smaller current increase seen at the second wave shows that the mediating effect vanishes at this point because of the diazonium salt being depleted in the diffusion layer at the first wave. The reduction wave pertaining to the diazonium salt once again shows up at each sweep and with a signal size that is comparatively larger than that seen in Figure 1B. We believe that this can be attributed to an inability to graft certain areas of the gold surface or, at least, that immobilized material at these regions are easily removed by the sweeping process, in particular, if Eλ =
2.1 V vs SCE.23 As discussed below, XPS measurements showing traces of Au for the thick films substantiate this view. In a recent study Shewchuk and McDermott also reported that more than 70% of a diazonium-modified gold substrate could be replaced in an exchange reaction with an alkylthiol.41 Interestingly, if the redox grafting is carried out at glassy carbon, where the covalent attachment at the surface is much more efficient going through the stronger C
C rather than Au
C bonds, we noticed that the wave pertaining to the diazonium salt essentially disappeared already on the second sweep. Film Thickness. Measurement of film thickness, d, by ellipsometry gave 5.2 ( 0.3, 52.4 ( 3.5, and 32.7 ( 5 nm for the AuAQ plates grafted at Eλ =
0.35,
1.5, and
2.1 V vs SCE, respectively, using 10 consecutive sweeps at ν = 1 V s
1. The two first observations are in complete agreement with the results expected on the basis of our previous work.24 In the first case characterized by Eλ > Ep,r,1, the film growth comes to a halt at an early stage as for most electrograftings of aryldiazonium salts4 due to the fact that the disordered organic layer formed is insulating in this potential range. On the other hand, if Ep,r,2 < Eλ < Ep,r,1, the radical anion of AQ is created during sweeping, which allows the AQ groups to serve as redox relay stations. This facilitates transport of electrons throughout the layer to the outer surface, where AQD becomes

ARTICLE

Figure 2. Cyclic voltammograms of Au-AQ disk electrodes (area = 0.0079 cm2) recorded at ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN. The electrodes were prepared from 2 mM AQD using 2 (solid), 5 (dot), 8 (dashed), and 10 (dash-dotted) voltammetric sweeps with Ei = 0.5 V vs SCE, Eλ =
1.5 V vs SCE, and ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN.

reduced. At the same time the AQ groups fulfill another important purpose upon charging, in that a continuous expulsion of physisorbed material may take place as the potential is swept negative to Ep,r,1.23,24 Thus, physisorbed species are removed from the layer to leave defects in the film, in which supporting electrolytes and solvent can penetrate and facilitate the electron transfer from the electrode to the solution through the AQ moieties. The result is that the organic film can become 10 times as thick as if created under “ordinary” conditions, where only the
N2+ group is reduced.24 On this basis it seems surprising that d is becoming smaller, if the switching potential is pushed further negative, that is, Eλ < Ep,r,2. However, this is presumably due to the immobilization being less efficient under these conditions as seen also for the grafting of, e.g., triaryl- and alkyldiphenylsulfonium salts,42 where the aryl radical to an appreciably greater extent may be further reduced to the corresponding anion, a nongrafting agent. Furthermore, for electron transfer reactions occurring in mixed valence systems, the AQ2
species with its high charge localization may actually be a less efficient electron donor than the comparatively more delocalized AQ•
.43 Note also that a more efficient desorption of adsorbed material may take place at the more negative potentials applied, although an actual detachment of covalently attached materials is not occurring (Figure S1, Supporting Information). In the further studies of the grafting process, we decided to use exclusively Eλ =
1.5 V vs SCE to diminish the risk of reducing the aryl radical to the aryl anion, if more extreme potentials were to be employed, not to mention the general risk of protonating the generated dianion of AQ by the residual water always present in CH3CN. Electrochemical Response of Au-AQ Electrodes. Figure 2 shows the superposition of voltammograms recorded of AuAQ electrodes (redox grafted using 2, 5, 8, and 10 cyclic voltammetric sweeps) in a pure electrolyte solution. The 0 presence of a distinct one-electron bell-shaped wave with E01 ≈
0.97 V vs SCE (peak separation ≈ 78 mV) for the four cases confirms that AQ indeed has become attached at the surface. Clearly, the intensity of both the reduction and the oxidation signals pertaining to the first wave of surface attached AQ 1270

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir

Figure 3. Cyclic voltammograms recorded at a bare Au (dashed) and a Au-AQ (line) disk electrode on 5 mM ferrocene using ν = 0.1 V s
1 in 0.1 M Bu4NBF4/CH3CN. The Au-AQ gold disk electrode was prepared from 2 mM AQD employing 10 consecutive sweeps with Ei = 0.55 V vs SCE, Eλ =
1.5 V vs SCE, and ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN.

increases as a function of the number of sweeps applied in the redox grafting procedure. It may be noted that the size of the current signals in Figures 2 and 1B becomes comparable, if the ratio of 76 for the two electrode areas (0.6 cm2 for the plate used in Figure 1B and 0.0079 cm2 for the disk electrode used in Figure 2) are corrected for. Still, in the latter case there are additional contributions from the mediated electrochemical reduction of AQD as well as from adsorption/desorption phenomena occurring for this or other related species in the film. As expected for Au-AQ , a second one-electron reduction wave can be seen at more negative potentials, and despite this being chemically irreversible, repeated sweeping induces essentially no chemical or electrochemical degradation of the surface (Figure S1, Supporting Information). This underlines the prospect in using this kind of electrodes in the further development of capacitors and electrocatalysts. Redox Probe Experiments. Figure 3 shows the response of the Au and Au-AQ electrodes in a solution containing the ferrocene redox probe. In spite of the thick film on the Au-AQ electrode (d = 52.4 nm), it is still possible to detect the oxidation and reduction waves of ferrocene. The peak separation, ΔEp ≈ 110 mV, is slightly larger than that recorded at a bare Au substrate (ΔEp ≈ 70 mV), indicating a surprisingly small effect on the ferrocene redox properties. This may be attributed to the presence of ungrafted areas on the gold surface along with a high permeability of the porous AQ film structure toward ferrocene. At the same time this behavior is in line with the persistency of the reduction signal of AQD in cyclic voltammetry (Figure 1B, C) which further would imply that the aryl radical generated during grafting rather than reacting with the ungrafted areas of the surface, which would block the entire surface, is doing hydrogen abstraction from the solvent. Indeed, such a scenario is plausible, considering that the more reactive kinks and defects sites on the Au surface already have reacted and through this been deactivated after the first sweep(s). The alternative explanation given by Pinson et al.17 for thick unsubstituted but highly conducting phenylene films is that the inner layer simply acts as a part of the electrode with the electrons easily being transported to the outer film/electrode surface.

ARTICLE

Figure 4. Film thickness, d, of Au-AQ samples against number of sweeps, N, for Eλ =
0.35 (2) and
1.5 (b) V vs SCE. The inset shows the relationship between the coverage of the corresponding Au-AQ electrodes, ΓAQ, and d for Eλ =
1.5 V vs SCE (9). Note that measurements of ΓAQ and d pertain to disk electrodes and plates, respectively. For both kinds of electrodes electrografting was carried out on 2 mM AQD using cyclic voltammetry with Ei = 0.55 V vs SCE and ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN.

II. Effect of Number of Sweeps. Film Thickness and Surface Coverage. One of the strengths of the redox grafting methodol-

ogy described herein is the ability to control the final thickness of the layer by applying consecutive potential sweeping. To investigate the effect of varying the number of sweeps, N, the thickness of the films grafted on Au plates at Eλ =
0.35 and
1.5 V vs SCE, respectively, was measured for the recording of until 10 cyclic voltammograms. Figure 4 shows the development in d as a function of N using either Eλ =
0.35 or
1.5 V vs SCE. For the samples grafted without exploiting the mediating effect of the AQ groups, d is essentially constant with 3 ( 1 nm. In contrast, for redox grafting carried out at Eλ =
1.5 V vs SCE the values of d vary from a few nanometers for 1 or 2 cycles to >50 nm, if 10 cycles are employed. Noteworthy, the plot shows an upward curvature with a steadily rise in d as N increases. This could be interpreted as if a hyperbranched structure is formed during the grafting process with every new “layer generation” bringing in more redox active groups than the previous one. In this respect, it is interesting that multilayer formation induced by the electrografting procedure is usually assumed to proceed by way of an aromatic homolytic substitution reaction, in which the grafting agent, the aryl radical, attacks already grafted aryl layers relatively fast.4 Considering that this attack may occur in several positions of the benzene ring, a branched structure would be expected to arise. At the same time the electrochemical response of Au-AQ samples (modified at Eλ =
1.5 V vs SCE using 2, 5, 8, and 10 cycles) was recorded in 0.1 M Bu4NBF4/CH3CN (see Figure 2). This enables the calculation of the surface coverage of anthraquinone, ΓAQ, by charge integration under the first voltammetric peak. The respective values of ΓAQ were 1.4, 4.2, 7.3, and 10.3 nmol cm
2 with the last value being, to the best of our knowledge, the highest reported one in the literature for electrode modification using AQD.28,31,32,34 The inset of Figure 4 shows that a linear relationship between ΓAQ and d exists with a proportionality constant = 2.0  10
10 mol cm
2 nm
1. The parallel increases in ΓAQ and d are also 1271

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir

ARTICLE

Figure 6. IRRAS spectra of Au-AQ plates prepared from 2 mM AQD using 2, 4, 6, 8, and 10 voltammetric sweeps with Ei = 0.55 V vs SCE, Eλ =
1.5 V vs SCE, and ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN.

Figure 5. AFM images of (A) a bare Au plate and Au-AQ plates prepared from 2 mM AQD using (B) 2, (C) 4, (D) 6, (E) 8, and (F) 10 voltammetric cycles with Ei = 0.55 V vs SCE, Eλ =
1.5 V vs SCE, and ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN.

evident, if both these parameters are plotted against N as demonstrated in the Supporting Information (Figure S2). The presence of this linear relationship suggests that the same fraction of the AQ units introduced is electrochemically active as each new layer is added to the film. Interestingly, no such proportionality between ΓAQ and d was observed for the corresponding 4-nitrobenzene-containing thick films,24 where a relatively large fraction of the redox activity was lost as the layer grew. However, for the very thin films (d < 4 nm) prepared under the usual conditions for diazonium grafting a proportionality constant = 3.2  10
10 mol cm
2 nm
1 was obtained.44 This number is, by and large, in agreement with our result for the AQ-based layers, taking into account the different sizes of the two redox units. Surprisingly, conversion of the proportionality constant of 2.0  10
10 mol cm
2 nm
1 to molar volume (= 500 cm3 mol
1) gives a ∼3 times larger value than the 159 cm3 mol
1 calculated from the density of crystalline anthraquinone (δ = 1.308 g cm
3) or derived from studies on thin anthraquinone films.33 At this point we do not know the exact reasons for the discrepancy, but we intend to address this issue in future studies by investigating in detail the layer morphology, the implication of using two different types of gold substrates in the measurements as herein (disk electrodes for the determination of ΓAQ and plates for d) and the extent to which electrochemically

silent and therefore unaccounted anthraquinone units are present in the film. Layer Morphology. AFM was employed to characterize the five films obtained after 2, 4, 6, 8, and 10 successive sweeps with a bare Au substrate serving as reference. The AFM images in Figure 5 show a globular layer structure as also observed for thin film grafting of 4-nitrobenzenediazonium salt23 and AQD34 on gold. These globular features become more pronounced as the number of sweeps is increased and along with this an increase in the surface roughness occurs (Table S1, Supporting Information). In fact, for the thickest films the height of the globular features is twice as large as the average thickness measured by ellipsometry, indicating that ungrafted surface areas of gold can easily be hidden underneath such features. These images support the interpretation propounded on the basis of the thickness and coverage measurements. Because of the high porosity of the globular layer structure, solvent and supporting electrolyte can penetrate the film during the grafting, hence allowing the redox moieties to be charged24 and becoming electrochemically active. In this manner the high porosity becomes a key factor in sustaining the high conductivity of the organic layer. At the same time in the noncharged state of the film the aryldiazonium salt will be allowed to diffuse all the way toward the surface to be reduced at the ungrafted areas of the gold surface as evidenced by the reappearance of the first reduction wave of AQD for the cyclic voltammetric sweeps in Figure 1B, C. Infrared Reflection
Absorption Spectroscopy. IRRAS was employed to investigate the chemical structure of the growing film during cyclic voltammetric sweeping. Figure 6 shows the IRRAS spectra of five Au-AQ plates in the region 2000
1000 cm
1 (full spectra shown in Figure S3; Supporting Information) with the 1680 cm
1 band corresponding to the stretching of CdO (aryl ketone), the 1595 and 1585 cm
1 bands to the CdC aromatic stretching, and the 1320 and 1275 cm
1 bands to the ring stretching and C-H bending, respectively.45 These signals are consistent with the presence of an organic layer consisting of AQ units solely with no indication of the corresponding alcohols as evidenced by the lack of O
H signals at 3600
3100 cm
1 (Figure S3, Supporting Information). Two main observations from these spectra can be noted. First, the increase of ΓAQ and d as a function of N is confirmed by the progressive increase in the intensity of all the characteristic 1272

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir

ARTICLE

Table 1. Surface Chemical Composition Determined by XPS of Bare Au and Au-AQ Plates Modified Using 2, 4, 6, 8, and 10 Voltammetric Cyclesa

b

Table 2. Thickness of Au-AQ Samples Grafted under Various Conditionsa entry

ν (V s
1)

N

Nν
1/2 (V
1/2 s1/2)

d (nm)b

atomic concentration (%)

1

0.5

5

7.1

33.4 ( 0.2

C

2 3

0.5 1

10 10

14.1 10

66.7 ( 3.4 52.4 ( 3.5

N

Au

O

Au

6.7

33

0

Au-AQ2

9.4

81.3

3

6.1

8.6

7.0 ( 1.1

Au-AQ4

10.7

85.5

2.5

1.3

8.0

16.1 ( 2.0

Au-AQ6

10.0

86.7

3.2

0.22

8.7

26.5 ( 2.2

Au-AQ8 Au-AQ10

10.9 10.7

86.0 85.8

2.8 3.2

0.12 0.03

7.9 8.0

38.5 ( 2.2 52.4 ( 3.5

59

C/O

d (nm)

sample

4.9

N.D.

a

Electrografting was carried out on 2 mM AQD using cyclic voltammetry with Ei = 0.55 V vs SCE, Eλ =
1.5 V vs SCE, and ν = 1 V s
1 in 0.1 M Bu4NBF4/CH3CN. b The index number denotes the number of voltammetric sweeps employed in the grafting procedure.

bands. Second, the ratio of CdO/CdC intensities (∼2.4) is independent of N, indicating that the layer has the same chemical composition and average orientation throughout the growth process. X-ray Photoelectron Spectroscopy. The surface chemical composition determined by XPS of bare Au and Au-AQ plates modified using 2, 4, 6, 8, and 10 cyclic voltammetric sweeps in the grafting process is summarized in Table 1. First, it is noted that the Au content decreases as the film thickness increases in accordance with expectations. However, even for a 50 nm thick sample traces of gold are detectable. Considering that the analysis depth of XPS is 10 nm the most plausible interpretation of this phenomenon is that areas of ungrafted gold surface exist, which also would be in line with the conclusions drawn from the cyclic voltammetric and AFM experiments. From the inhibition of the gold signal recorded at 84 eV (Au 4f7/2), using the bare substrate as reference and assuming the photoelectron escape depth to be 4.2 nm, as reported for alkanethiols on Au (111),46 the thickness of the films was estimated.47 In particular, it was found that d increases from 9.5 to 31.9 nm going from Au-AQ2 to Au-AQ10, which should be compared with the corresponding ellipsometrically determined values of 7.0 and 52.4 nm (see Table 1). As seen, the consistency in these measurements is far from being optimal, which most likely can be attributed to the high uncertainty introduced in the XPS experiment by the large surface roughness. In agreement with the IRRAS results, the XPS data show that the chemical composition of the organic layer is essentially the same throughout the film with ∼10% oxygen, ∼85% carbon, and ∼3% nitrogen. An experimental C/O ratio of 8 is a little higher than the theoretically 7 expected for a film consisting of AQ units only. Our hypothesis is that some of the supporting electrolyte (Bu4NBF4) may be trapped in the layer which is substantiated by the finding of a small amount of F in some samples. The nitrogen content may to some extent also originate from the supporting electrolyte, but most likely the largest contribution stems from azobenzenes formed because of diazo coupling reactions taking place during grafting.13,48,49 III. Combined Effect of Sweep Rate and Number of Sweeps. To elucidate the combined effect the ν and N parameters would exert on the film formation, we carried out a more systematic study, in which several Au-AQ samples were prepared by redox grafting using Ei = 0.55 and Eλ =
1.5 V vs SCE and

4

2

10

7.1

48.1 ( 3.3

5

2

20

14.1

107 ( 8c

6

5

10

4.5

24.7 ( 0.5

7

5

50

22.4

138 ( 8c

8

d

1.5 ( 0.3

a

Electrografting carried out on 2 mM AQD using cyclic voltammetry with Ei = 0.55 V vs SCE and Eλ =
1.5 V vs SCE in 0.1 M Bu4NBF4/ CH3CN. b Measured by ellipsometry. c Measured also by profilometry. d Electrolysis carried out at
1.5 V vs SCE for 40 s.

Figure 7. Film thickness, d, of Au-AQ plates against Nν
1/2. Electrografting was carried out on 2 mM AQD using repititive cyclic voltammetry with Ei = 0.55 V and Eλ =
1.5 V vs SCE in 0.1 M Bu4NBF4/ CH3CN.

with ν varying from 0.5 to 5 V s
1 and N from 5 to 50. The resulting d values showed an overall variation from 33 to 138 nm. A single experiment employing potentiostatic electrolysis conditions was also conducted, in the case of which a film thickness of 1.5 nm obtained. These results are collected in Table 2. In Figure 7, we have plotted the values of d versus Nν
1/2 (entries 1
7), and as seen a reasonably linear relationship is found. This finding may be understood on account of the following considerations. From the linear relationship between d and ΓAQ (inset of Figure 4; vide supra) and further the observation that the integrated charge underneath the curves recorded during grafting in Figure 1B increase as a function of N (like d and ΓAQ do), it may be deduced that d is proportional to the overall charge consumed in the reduction of AQD. This conclusion is very important in itself, but the results may be taken a bit further by assuming in a first rough approximation that d is linearly dependent on N, which would correspond to a linearization of the curved plot in Figure 4. The implication then is that R d should become proportional to NQ = N i dt, where Q denotes the charge consumed in a single cycle, i is the corresponding current, and t is the time. Considering that a substantial part of Q is consumed under diffusion-controlled conditions as the sweep potential passes Ep,r,1, i is expected to be dependent on t
1/2 1273

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir √ because of the t-dependent growth of the diffusion layer.50 Hence, it may be argued √ that d as a first approximation becomes proportional to N t or Nν
1/2 as indeed suggested by the overall correlation in Figure 7. This being said there is no doubt that the picture is more complex than so. In addition to the uncertainty arisen because of the many assumptions underlying the derived relationship, the sets of data point pertaining to Nν
1/2 = 7.1 (entries 1 and 4) and 14.1 (entries 2 and 5) show another effect, in that a 50
60% thicker layer is obtained, if one uses a 4 times higher ν along with a doubling of N. We suggest as in our first study24 and as mentioned above that this phenomenon is due to the sweeping of the potential being essential for promoting desorption of physisorbed material during the electrografting—material which otherwise will clog the layer and block the film growth. These specific results therefore show that the essential removal of adsorbed species requires employment of relatively high sweep rates that exploit the fact that relatively thinner layers are grafted and need to be “purified” at each individual sweep. In contrast, at low sweep rates e0.5 V s
1 the film formation can be severely disturbed because of this problem. Noteworthy, if the grafting is carried out by way of a potentiostatic electrolysis at
1.5 V vs SCE for 40 s (entry 8) rather than the usual potential sweeping approach the film created has as low thickness as 1.5 nm. This is in line with the thin 4 nm nitrophenyl-containing films formed if applying potentiostatic electrolysis to the 4-nitrobenzenediazonium salt13 but stands in contrast to diazonium salts of ruthenium tris(bipyridine) complexes,20 where deposition can be performed at constant applied potentials without complications. This indicates that the adsorption phenomena noted herein pose no or only small problems for the larger organometallic systems.

’ CONCLUSION The properties of anthraquinone generated films on Au using redox grafting of AQD are strongly dependent on the switching potential used. If the switching potential is selected just slightly negative to the reduction wave of the diazonium salt, a thin layer of a few nanometers thickness is formed. However, pushing the switching potential further negative to encompass the first reduction wave of the anthraquinone moiety provides an efficient way to prepare thick anthraquinone-containing layers. The film growth that is based on the mediating effect of the AQ groups is fast with the thickness easily exceeding 100 nm upon carrying out repetitive sweeping. The morphology of these layers is porous and mushroom-like and with the existence of ungrafted areas on the gold surface molecules such as ferrocene or AQD itself can easily penetrate the film in acetonitrile and reach the surface. The surface coverage of the electrochemically active AQ redox units follows the thickness trend, suggesting that the same high fraction of AQ units in the film is electrochemically detectable throughout the film and that the exact size of the electrochemical signal (as required by a given application) can easily be fine-tuned using the sweeping procedure. An analysis of the development of the film thickness as a function of sweep rate and number of sweeps shows that the growth of the layer essentially occurs under diffusion-controlled conditions. However, on top of this another effect comes through, in that relatively thicker layers can be created upon applying higher sweep rates. This is probably due to the fact that the restructuration occurring during the sweeping with layer

ARTICLE

opening/closing, entrance/expulsion of solvent/electrolyte, and, in particular, desorption of physisorbed material promotes the formation of the thick layers of high conductivity. Employment of sweep rates >0.5 V s
1 is optimal for creating these anthraquinone-containing redox layers, the thickness of which subsequently can be controlled through the number of sweeps. Importantly, the chemical composition is the same during the growth with no sign of degradation processes taking place. Hence, it should be possible to apply the redox grafting method in the preparation of supercapacitors and high-quality sensors.

’ ASSOCIATED CONTENT

bS

Supporting Information. Electrochemical stability measurements, plots of d and ΓAQ vs N, IRRAS spectra, and surface roughness measurements of Au-AQ samples. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Authors

*E-mail: [email protected] (S.U.P.); [email protected] (K.D.). Present Addresses †

IPREM UMR CNRS 5254, Universit
e de Pau et des Pays de l’Adour, H
elioparc 2 avenue du Pr
esident, Angot, 64053 Pau cedex 9, France.

’ ACKNOWLEDGMENT The Danish Agency for Science, Technology and Innovation, Grundfos A/S, and SP Group A/S are gratefully acknowledged for financial support. ’ REFERENCES (1) Willner, I.; Baron, R.; Willner, B. Biosens. Bioelectron. 2007, 22, 1841–1852. (2) Gr€atzel, M. Inorg. Chem. 2005, 44, 6841–6851. (3) Gooding, J. J. Electroanalysis 2008, 20, 573–582. (4) Pinson, J.; Podvorica, F. Chem. Soc. Rev. 2005, 34, 429–439. (5) Onclin, S.; Ravoo, B. J.; Reinhoudt, D. N. Angew. Chem., Int. Ed. 2005, 44, 6282–6304. (6) Love, J. C.; Estroff, L. A.; Kriebel, J. K.; Nuzzo, R. G.; Whitesides, G. M. Chem. Rev. (Washington, DC, U. S.) 2005, 105, 1103–1169. (7) Delamar, M.; Hitmi, R.; Pinson, J.; Sav
eant, J. M. J. Am. Chem. Soc. 1992, 114, 5883–5884. (8) Combellas, C.; Kanoufi, F.; Pinson, J.; Podvorica, F. I. Langmuir 2005, 21, 280–286. (9) Nowak, A. M.; McCreery, R. L. Anal. Chem. 2004, 76, 1089–1097. (10) Boukerma, K.; Chehimi, M. M.; Pinson, J.; Blomfield, C. Langmuir 2003, 19, 6333–6335. (11) Bernard, M. C.; Chauss
e, A.; Cabet-Deliry, E.; Chehimi, M. M.; Pinson, J.; Podvorica, F.; Vautrin-Ul, C. Chem. Mater. 2003, 15, 3450–3462. (12) Kariuki, J. K.; McDermott, M. T. Langmuir 1999, 15, 6534– 6540. (13) Ceccato, M.; Nielsen, L. T.; Iruthayaraj, J.; Hinge, M.; Pedersen, S. U.; Daasbjerg, K. Langmuir 2010, 26, 10812–10821. (14) Iruthayaraj, J.; Chernyy, S.; Lillethorup, M.; Ceccato, M.; Ron, T.; Hinge, M.; Kingshott, P.; Besenbacher, F.; Pedersen, S. U.; Daasbjerg, K. Langmuir 2011, 27, 1070–1078. (15) Boland, S.; Barriere, F.; Leech, D. Langmuir 2008, 24, 6351–6358. (16) Kariuki, J. K.; McDermott, M. T. Langmuir 2001, 17, 5947–5951. (17) Adenier, A.; Combellas, C.; Kanoufi, F.; Pinson, J.; Podvorica, F. I. Chem. Mater. 2006, 18, 2021–2029. 1274

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275

Langmuir

ARTICLE

(18) Jousselme, B.; Bidan, G.; Billon, M.; Goyer, C.; Kervella; Guillerez, S.; Hamad, E. A.; Goze-Bac, C.; Mevellec, J. Y.; Lefrant, S. J. Electroanal. Chem. 2008, 621, 277–285. (19) March, G.; Reisberg, S.; Piro, B.; Pham, M. C.; Fave, C.; Noel, V. Anal. Chem. 2010, 82, 3523–3530. (20) Agnes, C.; Arnault, J. C.; Omnes, F.; Jousselme, B.; Billon, M.; Bidan, G.; Mailley, P. Phys. Chem. Chem. Phys. 2009, 11, 11647–11654. (21) Bureau, C.; Levy, E.; Viel, P. US 2004/0248428 A1, 2004. (22) Bureau, C.; Levy, E.; Viel, P. PCT Int. Appl. WO 03018212, 2003. (23) Haccoun, J.; Vautrin-Ul, C.; Chauss
e, A.; Adenier, A. Prog. Org. Coat. 2008, 63, 18–24. (24) Ceccato, M.; Bousquet, A.; Hinge, M.; Pedersen, S. U.; Daasbjerg, K. Chem. Mater. 2011, 23, 1551–1557. (25) Algharaibeh, Z.; Liu, X.; Pickup, P. G. J. Power Sources 2009, 187, 640–643. (26) Kalinathan, K.; DesRoches, D. P.; Liu, X.; Pickup, P. G. J. Power Sources 2008, 181, 182–185. (27) Pognon, G.; Brousse, T.; Demarconnay, L.; B
elanger, D. J. Power Sources 2011, 196, 4117–4122. (28) Kullapere, M.; Seinberg, J. M.; M€aeorg, U.; Maia, G.; Schiffrin, D. J.; Tammeveski, K. Electrochim. Acta 2009, 54, 1961–1969. (29) Tammeveski, K.; Kontturi, K.; Nichols, R. J.; Potter, R. J.; Schiffrin, D. J. J. Electroanal. Chem. 2001, 515, 101–112. (30) Salimi, A.; Banks, C. E.; Compton, R. G. Phys. Chem. Chem. Phys. 2003, 5, 3988–3993. (31) Sarapuu, A.; Vaik, K.; Schiffrin, D. J.; Tammeveski, K. J. Electroanal. Chem. 2003, 541, 23–29. (32) Reilson, R.; Kullapere, M.; Tammeveski, K. Electroanalysis 2010, 22, 513–518. (33) Li, Q.; Batchelor-McAuley, C.; Lawrence, N. S.; Hartshorne, R. S.; Compton, R. G. New J. Chem. 2011, 35, 2462–2470. (34) Kullapere, M.; Marandi, M.; Sammelselg, V.; Menezes, H. A.; Maia, G.; Tammeveski, K. Electrochem. Commun. 2009, 11, 405–408. (35) Kullapere, M.; Tammeveski, K. Electrochem. Commun. 2007, 9, 1196–1201. (36) Pandurangappa, M.; Lawrence, N. S.; Compton, R. G. Analyst 2002, 127, 1568–1571. (37) Allongue, P.; Delamar, M.; Desbat, B.; Fagebaume, O.; Hitmi, R.; Pinson, J.; Sav
eant, J.-M. J. Am. Chem. Soc. 1997, 119, 201–207. (38) Daasbjerg, K.; Pedersen, S. U.; Lund, H. In General Aspects of the Chemistry of Radicals; Alfassi, Z. B., Ed.; Wiley: Chichester, UK, 1999; pp 385
427. (39) Ajloo, D.; Yoonesi, B.; Soleymanpour, A. Int. J. Electrochem. Sci. 2010, 5, 459–477. (40) Andrieux, C. P.; Sav
eant, J. M. J. Electroanal. Chem. 1986, 205, 43–58. (41) Shewchuk, D. M.; McDermott, M. T. Langmuir 2009, 25, 4556–4563. (42) Vase, K. H.; Holm, A. H.; Norrman, K.; Pedersen, S. U.; Daasbjerg, K. Langmuir 2008, 24, 182–188. (43) Hankache, J.; Wenger, O. S. Chem. Rev. (Washington, DC, U. S.) 2011, 111, 5138–5178. (44) Brooksby, P. A.; Downard, A. J. Langmuir 2004, 20, 5038–5045. (45) Pecile, C.; Lunelli, B. J. Chem. Phys. 1967, 46, 2109–2118. (46) Bain, C. D.; Whitesides, G. M. J. Am. Chem. Soc. 1989, 111, 7164–7175. (47) Laforgue, A.; Addou, T.; B
elanger, D. Langmuir 2005, 21, 6855–6865. (48) Doppelt, P. H. G.; Pinson, J.; Podvorica, F.; Verneyre, S. Chem. Mater. 2007, 19, 4570–4575. (49) Toupin, M. B. D. J. Phys. Chem. C. 2007, 111, 5394–5401. (50) Daasbjerg, K., Pedersen, S. U., Vincenzo, B., Ed.; Wiley-VCH Verlag GmbH: Weinheim, 2001; Vol. I.

1275

dx.doi.org/10.1021/la203657n |Langmuir 2012, 28, 1267–1275