Electrochemical Investigation of Adsorption of Single‐Wall Carbon Nanotubes at a Liquid/Liquid Interface

Abstract There is much interest in understanding the interfacial properties of carbon nanotubes, particularly at water/oil interfaces. Here, the adsorption of single‐wall carbon nanotubes (SWCNTs) at the water/1,2‐dichloroethane (DCE) interface, and the subsequent investigation of the influence of the adsorbed nanotube layer on interfacial ion transfer, is studied by using the voltammetric transfer of tetramethylammonium (TMA+) and hexafluorophosphate (PF6 −) as probe ions. The presence of the interfacial SWCNT layer significantly suppresses the transfer of both ions across the interface, with a greater degree of selectivity towards the PF6 − ion. This effect was attributed both to the partial blocking of the interface by the SWCNTs and to the potential dependant adsorption of background electrolyte ions on the surface of the SWCNTs, as confirmed by X‐ray photoelectron spectroscopy, which is caused by an electrostatic interaction between the interfacial SWCNTs and the transferring ion.


Introduction
Carbon nanotubes (CNTs) have attracted ag reat deal of attention owing to their interesting optical, mechanical, and electrical properties. [1] Their potentiala pplicationi navariety of areas such as nanoelectronics, [2] field effect transistors, [3] electrochemicala nd sensor devices, [1a, 4] and as catalyst support [5] has been demonstrated. However,f or most applications in nanoscience and technology,the processing and subsequent formation of stable assemblies of these nanostructures is highly important. [6] Liquid/liquid interfaces, particularly the oil/water interface, have become increasingly popular for the assembly of aw ide range of nanostructures, such as metal nanoparticles and two dimensional (2D) semiconducting nanomaterials. [7] The self-assembly of both single-wall carbon nanotubes (SWCNTs) and multi-walled carbon nanotube (MWCNTs) at avariety of liquid/liquid interfaces hasa lso been explored greatly as an alternative methodt og eneratef unctional CNT films. [8] Typically, the material to be assembled at the interface (i.e. CNTs, in this case) is suspended in one of the bulk liquid phases.T he suspension is then contacted with the second liquid phase and assembly is subsequently induced by mechanical agitation or addition of an inducing solvent. [9] Althought he majority of these studies-in the case of CNTsfocus more on the assembly process, few have investigated the properties of the CNT layers/films in situ at these interfaces. For example, Matsui et al. [8c] fabricated ultrathin films, or 2D layers, of SWCNTsa tt he water/n-hexane interface and char-acterized their opticala nd electrical properties ex situ, after transfer of the films onto as ilicon wafer.T he use of SWCNTst o transporte nzymes from ab ulk aqueous phase to aw ater/organic interface, and the subsequentc haracterizationo ft he biocatalytic activity of the resulting SWCNT-enzymei nterfacial layer has been examined, with an enhancement in the rate of biotransformation observed with the interfacial layer. [10] This was interpreted in terms of the high intrinsic surface area provided by the SWCNTsa nd the absence of intraparticle diffusion limitations. Zhang et al. [11] obtained aflexible thin film of imidazolium-functionalized SWCNTs( Im-SWCNTs) at an on-polarized water/chloroform interface and attempted electrochemical characterization of the resultanti nterfacial layer by using scanning electrochemical microscopy (SECM). Witho nly the oxidized form of the redox species[ Ru(NH 3 ) 6 3 + ]p resenti nt he aqueous phase, it was shown that, at the "bare" water/chloroform interface, an egative feedback current was generated as the tip approached the interface, owing to the interface acting as an insulator;w hereas, in the presence of an Im-SWCNTsi nterfaciall ayer,apositive feedback current was generated at the tip, indicating that the Im-SWCNTsf ilm was electroactive. However, as there was no redox species in the chloroform phase, no charge-transfer reaction occurred between the two immiscible liquids.
In the presence of appropriate electrolytes dissolved in each liquid phase, the liquid/liquid interface is referred to as the interface between two immiscible electrolyte solutions( ITIES). This special class of liquid/liquid interface can be polarized by the applicationo fa ne xternale lectric field, thus allowing both ion-and electron-transferr eactions to be readily studied by using electrochemical methods. [12] The modification of the ITIES with adsorbed solids has been shown to be av iable meansf or studying the properties of interfacially adsorbed ma-There is much interest in understanding the interfacial properties of carbon nanotubes, particularly at water/oil interfaces. Here, the adsorption of single-wall carbon nanotubes (SWCNTs ) at the water/1,2-dichloroethane( DCE) interface, and the subsequent investigation of the influence of the adsorbed nanotube layer on interfacial ion transfer,i ss tudied by using the voltammetric transfer of tetramethylammonium (TMA + )a nd hexafluorophosphate (PF 6 À )a sprobe ions. The presence of the in-terfacialS WCNT layer significantly suppressest he transfer of both ions across the interface, with ag reater degree of selectivity towards the PF 6 À ion. This effect was attributed both to the partial blockingo ft he interface by the SWCNTsa nd to the potentiald ependant adsorption of background electrolyte ions on the surfaceo ft he SWCNTs, asc onfirmed by X-ray photoelectron spectroscopy,w hich is caused by an electrostatic interaction between the interfacial SWCNTsa nd the transferring ion.
terials,s uch as membrane porosity [13] and catalytic activity of metal nanoparticles. [14] Recent studies in our laboratory have utilizedt his approach to probe the electrochemical properties of graphitic carbon nanostructures (CNTsa nd few-layer graphene) adsorbed at the ITIES. It was shown that interfacially assembled SWCNT/graphene layers serve as electron mediators, aiding heterogeneous electron transfer between aqueous and organic redox couples, which remain isolated in their respective phases. [15] This was utilized to functionalize interfacial SWCNT and graphene layers with metal nanoparticles by reducing aqueous metal salts using an organic electron donor, [15a, 16] and ac onducting polymerp oly(pyrrole), [17] through oxidation of the pyrrole monomer dissolved in the organic phase by an aqueous oxidizing agent. Similarly,t he electron-transfer-mediating properties of pristine liquid-phasee xfoliated graphenea tt he water/organic interface were found to result in ac atalytic effect on the heterogeneous oxygen reduction reaction. [18] Furthermore, the electrochemical doping of the interfacial SWCNTsw as investigated by using in situ Ramanspectroelectrochemistry. [15b] The objective of the current work is to investigate the electrical properties of SWCNTsa dsorbed at the water/DCE interface through analysis of their effect on the kinetics of ion transfer across the interface. The permeability of the films formed at differentS WCNTsc oncentration by the ionic species is also described.

SWCNT Adsorption at Water/DCE Interface
Interfacial SWCNT layers were formed following a1 0min bath sonication of cells containing aD CE dispersion of SWCNTsa nd an aqueous phase solution. The SWCNT film located between the bulk phases was visible af ew minutes after sonication. However,o wing to some emulsification of both the water/DCE interface and the bulk liquid phases, caused by the sonication, the cells were left to stand for 12 ht oa llow the emulsion droplets in the bulk phases to coalescebefore carrying out any electrochemical measurements. Figure 1a shows at ypical cell 12 ha fter sonication.
The resultant interfacial SWCNT filmwas characterized in situ by using optical microscopy and ex situ by using scanning electron microscopy (SEM). Figure 1b shows an optical micrograph of aS WCNT film obtained by using aD CE dispersion concentration of 3mgL À1 .A sc an be seen, the interfacial film was composed of multiple emulsified dropletss tabilized by SWCNTs. These droplets wereo bserved to be stable for up to 7days when left undisturbed. Longer time stability was not studied here. The droplets were also found to be stable when the water/DCE interface was polarized, ase xemplified by the microscopy images in Figure 1c-e. The images showing the morphology of the interfacial layer at different appliedp otential differencesa cross the water/DCE interface (Df)w ere captured during ac yclic voltammetry experiment, where Df was swept from À0.24 to + 0.46 V. The only effect observedw as the movement of the whole interfacialf ilm towards one side of the interface on positive polarization (indicated by the arrows, showing that as Df wass wept from À0.24 to + 0.46 V, the space between the film and the wall of the glass increases) and vice-versa on reversep olarization. This movement of the SWCNT film may be connected with the movement of individual interfacial SWCNTs, as previously highlighted by the Girault group.
[5a] At extreme positive interfacial potentiald ifference (Df! + 0.46 V), corresponding to background ion transfer,t he interfacial film rotates clockwiseo npositive scan and anticlockwise on reverse( negative) scan. Figure 1e was taken during this rotation, which is the reason it appear like slightly out of focus. Figure 2s hows ex situ SEM images of interfacial SWCNTs films prepared at different SWCNT dispersion concentrations. The interfacial films were carefully transferred ontoaSi/SiO 2 substrate prior to SEM measurement. As can be seen, the interfacial preparation methodr esulted in two types of SWCNT film morphologies depending on the initial dispersion concentration:a tl ow SWCNT concentration (1 mg L À1 ), the SWCNTsw ere predominately bent into rings (Figure 2a), with only af ew straight or partially bent tubes, whereas at the higher SWCNT concentrations of 6, 12, and 18 mg L À1 ,p orous interfacial films were formed, composed of random networks of multilayer SWCNTs ( Figure 2b-d). The density of these multilayer films can be seen to increase with increasing nanotube concentration, althought he density of the films obtained with 12 and 18 mg L À1 CNT concentrations were very similar. The observed concentration-dependent transition of SWCNT morphologies from rings to straight tubes is similar to the findings of Wang et al. for aw ater/DCB Pickering emulsion system stabilized by SWCNTs. [19]

Ion Transfer across SWCNTI nterfacial Films
The composition of the cell employed for all electrochemical measurements is outlined in Scheme 1. Figure 3s hows the cyclic voltammograms (CVs) obtained in the presence of only the background electrolytesa tt he bare water/DCE interface and with interfacial SWCNT films prepared from two different bulk SWCNT dispersion concentrations( 6a nd 18 mg L À1 ). It can be seen that, in the presence of the interfacial SWCNT layers, there was as light increase in the capacitive current. Thisc an be attributed to an increase in liquid/liquid interfacial roughness, owing to the presence of multiple emulsion droplets formed at the interface when the SWCNTsa re adsorbed (Figure 1). [20] An alternative explanation is to consider the relative capacitances of the "free" and "blocked" parts of the interface;h owever,s uch an argument would lead to ad ecrease in the net capacitance, owing to the low capacitance of carbon nanotubes. [21] Furtheri ncreasesi nt he bulk SWCNT concentration resulted in very little increase in the capacitive current. Additionally,t he interfacial SWCNT films affected the magni-tude and shape of the background electrolyte ion transfer peaks (Li + and Cl À ), which limit the potential window on the positive and negative ends, respectively.T he current magnitudes werer educed and the transfer peaks became broader, indicating that the presence of an interfacial SWCNT film makest he ion transfer more difficult.

TMA + + and PF 6 À Ion Transfer
The blocking effect of SWCNTsi nterfacialf ilms on iont ransfer was further investigated by employing TMA + and PF 6 À as probe ions. First, the transfer of each ion was performed in the absence of SWCNT layers and then repeated in the presence of interfacial SWCNT films of differing thickness. The CVs shown in Figure 4w ere obtained for TMA + and PF 6 À ions in the presence of SWCNT films prepared from bulk SWCNT concentrations of 1mgL À1 .A lso shown in the figure are CVs obtaineda t the unmodified interfacef or comparison. It can be seen that the responses of both TMA + and PF 6 À ions were very similar to those obtained in the absence of the interfacialS WCNTsf ilms. There was only as mall increase in peak separation (DE p )a nd as light reduction in the peak current magnitudes.
However,w hen the nanotube dispersion concentration used for the film preparation was increased to 6mgL À1 ,t he response obtained in all cases was significantly altered, as compared to those obtained at the bare interface ( Figure 5);b oth forwarda nd reverse transfer peaks were broadened and shifted away from each other and their magnitudesdecreased. This behavior indicates that increasing the SWCNT dispersion concentration leads to ah igheri nterfacial surfacec overage, resulting in ag reater part of the interface available for ion transfer being blocked by the SWCNTs. Figure 6a shows ag rapho ft he dependence of forward peak height (I pf ), measured at bare and at SWCNT-covered interfaces,a gainst the square root of the scan rate (n 1/2 )f or both TMA + and PF 6 À .I ne ach case, I pf was linearly relatedt on 1/2 and the decrease in I pf was similarf or both ions. Using the Randles-Ševčik relation for the data collected from the bare water/DCE interface, the aqueous diffusion coefficient( D w )o f each ion was calculated. D w values obtained forT MA + (1.2 Figure 2. Ex situ SEM images showingt he morphologieso ft he SWCNTs films formed at the liquid/liquid interface using a C SWCNT of a) 1mgL À1 , b) 6mgL À1 ,c)12mgL À1 L, andd)18mgL À1 .
Scheme1.Schematicoft he electrochemical cell used in ion transfer studies. Yise ither TMA + or PF 6 À . 10 5 cm 2 s À1 )a nd PF 6 À (1.4 10 5 cm 2 s À1 )w erei na greement with the literaturev alues of 1.2 10 5 cm 2 s À1 [22] for TMA + and 1.5 10 5 cm 2 s À1 [23] for PF 6 À .F igure 6b shows the change in DE p for each ion as af unctiono fs can rate in the presence of SWCNTs. It can be seen that the change in DE p is greater for PF 6 À compared to TMA + ,w hich indicates that the kinetics of the PF 6 À transfer were more inhibited by the interfacial SWCNT film.
To rationalize this observed ion selectivity,t he possible adsorptiono fe ither the probe ions or the organic background electrolyte ions on the assembled SWCNT filmwas investigated by using chronoamperometry and XPS. Firstly,p otential step experiments were performed for each probe iont ransfer.T he interfacial potential was stepped from ap otentialw here no ion transfer occurs (À0.1 and + 0.25 Vf or TMA + and PF 6 À ,r espectively) to ap otentialw here ion transfer from the water to organic phase takes place (+ 0.25 and À0.18 Vf or TMA + and PF 6 À ; . spectively). The interfacial potential was held at the ion transfer potential for 10 min, after which the SWCNT filmw as carefullyt ransferred onto aS i/SiO 2 wafer.T he transferredl ayers were then washedi ne thanol, isopropanol, and acetone and dried before subsequent XPS analysis. Ac ontrol sample was treated in as imilar way,w ith the exception that neither the probe ions nor the supporting electrolyte ions were present and no interfacial potentialwas applied. Figure 7a presentsX PS spectra of the aforementioned SWCNT films. The spectra show the presence of B, N, Cl, and P in films obtained with either TMA + or PF 6 À present, butn ot in the controls ample. The appearance of signals attributable to B, Cl, and Pi nt he TMA + and PF 6 À samples is indicative of ad-   sorptiono ft he aromatic cation, BTPPA + ,a nd anion, TPBCl À ,o f the organic supporting electrolyte on the SWCNT surface, as substantial amounts of these elements couldo nly be reasonably attributedt ot he supporting electrolyte ions. The absence of as ignificants ignal for fluorine in any sample, and particularly in the sample obtained with PF 6 À present, suggests that the PF 6 À was only weakly adsorbed or not adsorbed at all. The percentage atomic concentrations of B, N, Cl, and Pd etermined from the survey spectra of TMA + and PF 6 À samples were normalized to that of C, and the results are summarizedi nF igure 7b.T he P/N ratio of approximately 2:1i sc onsistent with the stoichiometric compositiono fB TPPA + .T he absence of additional Ni nt he TMA + sample could suggest that the TMA + ion only weakly adsorbs, or not at all, on the SWCNT surface. Similarly,t he B/Cl ratio was found to be close to the 1:4e xpected for the TPBCl À anion. The slight excess of Bw as attributed to overlap of the B1sa nd P2sp eaks, which made it difficult to accurately subtract the contributiono ft he P2ss ignal. Nevertheless, the XPS data clearly demonstrate the preferential adsorption of BTPPA + and TPBCl À on the SWCNTss urfaceo ver the TMA + and PF 6 À ions, which is plausiblec onsidering that both BTPPA + and TPBCl À are charged and could also interact with the SWCNTst hrough p-p stacking. [24] The structure of these electrolyte ions are shown in Figure 7. Furthermore, the XPS data presented in Figure 7b show ap otential-dependent adsorption of the supporting electrolyte ions on the interfacial SWCNTs, as illustrated by the relative intensities of the components obtained for the samples containing TMA + and PF 6 À .I t can be seen that at the TMA + transfer potential (+ 0.25 V), the Na nd Pp eaks, attributable to BTPPA + ,a re lower in intensity compared to those measured at the PF 6 À transfer potential (À0.18 V), whereas the intensities of the Ba nd Cl peaks from TPBCl À. al ower for the PF 6 À transfer potentialt han that of TMA + .O verall, the XPS data suggestt hat asymmetrica dsorption of the supportinge lectrolyte ions occurs on the interfacial SWCNTs, introducing an et negative or positive surfacec harge on the SWCNTsa tt he TMA + and PF 6 À transfer potentials, respectively,t hereby resulting in the retardation of ion transfer acrosst he interface through electrostatic attraction between the transferring ion and the adsorbed supporting electrolyte counter ion. Thed ifference in the extent of charge-transfer suppression between the two probe ions is associated with the relative positions of the transfer potentials of the probe ions with respect to the potentialo fz eroc harge (PZC)i nt he presence of the modified SWCNTs. The fact that the kinetics of PF 6 À ion transfer is affected more than that of the TMA + ion implies that the SWCNT film has ah igherc harge density at the PF 6 À transfer potential, causing more electrolyte ions to adsorb on its surfaceand, consequently,increasing the electrostatic attraction between the PF 6 À ion and the adsorbed BTPPA + cation. For the TMA + ion to be less hindered, its transfer potential should be closer to the PZC of the system, which results in less attraction between the TMA + and the adsorbed TPBCl À anion.

Kinetics of Ion Transfer
The apparent rate constant (k 0 app )o fT MA + and PF 6 À ion transfer in the presenceo fS WCNTsf ilms was determined by using the Nicholson method. [25] DE p valuesm easured at scan rates higher than 25 mV s À1 were used. Figure 8s hows the k 0 app valueso btained. Increasingt he SWCNT dispersion concentration in the organic phase resulted in ad ecrease in k 0 app for the TMA + ion, owing to the greater surface coverage by the SWCNTs ( Figure 2). This can be explained by invoking Amatore's theory of voltammetry [26] at ap artially blocked electrode if we assumet hat the SWCNTsh ave transformed the single continuous interfacial area into al arge number of smaller randomly distributed micro-/nanopores, the size and/or density of which decreases with increased interfacial coverage. According  to the theory, [26] under conditions of total overlap of the diffusion layers, k 0 app is lowered by af actor of (1Àq)[ Eq. (1)]: where q is defined as the fractional area covered by the blocking nanotube film. The fact that the voltammetric profile of TMA + transfer exhibited ap eak-shaped response rather than as igmoidal one indicates that an overlapping linear diffusion field was achieved. The interfacial coverage was estimated from the SEM data to be about 77.5, 88.0, and 91.7 %w hen the SWCNT concentration used in film preparation was 6, 12, and 18 mg L À1 respectively.T herefore, applying the (1Àq)c orrection factor gave an average k 0 value of 1.0 AE 0.1 10 À2 cm s À1 .
Aside from the slower kinetics displayed by the PF 6 À ion compared to TMA + transfer in the presence of interfacially assembled SWCNTs ( Figures 6b and 8), it is also clear from Figure 8t hat the negative probe ion also show al ess clear dependence of k 0 app on SWCNT concentration. The ion transfer and the XPS data indicate that there is ap otential-dependent change in surfacec omposition of the nanotubes, which in turn suggestst hat the nanotubes adsorbo nthe interface from the organic phase, that is, they constitute part of organic double layer.T his effect is then associated with the high surface charge density exhibited by the interfacial SWCNTsa tt he PF 6 À transfer potential, which leads to the attainment of maximum blockagea tt he SWCNT concentration of 6mgL À1 ,a sa gainst the TMA + ion.

Conclusions
In the present study,w eh ave demonstrated the use of ion transfer voltammetry at the liquid/liquid interface to characterize the electrical properties of SWCNTsa dsorbed at aw ater/ DCE interface. In the presence of adsorbed SWCNT layers of varying density/thickness, transfer of the positively charged TMA + ion across the interface was found to be less inhibited than the corresponding negative PF 6 À ion. The retardationo f ion transfer by the nanotube layer was analyzed by using the theory of voltammetry at partially blocked electrodes, and the selectivity between TMA + and PF 6 À ions was attributed to the potential-dependent adsorption of the organic supporting electrolyte ions on the interfacial SWCNTs, asi ndicated by XPS measurements, which caused electrostatici nteraction between the transferring ion and the SWCNT surface and, thereby, inhibited the ion transfer.

Methods
SWCNT dispersions in DCE were prepared by sonication. Pristine SWCNTs( 22 mg) were placed in a5 00 mL flat-bottom glass bottle containing DCE (100 mL). The contents were bath sonicated for 24 hb yu sing an Elmasonic P70 Hs onicator (Elma GmbH &C o. KG) at 37 KHz and 30 %p ower setting. The as-prepared dispersion was stable for months. Aliquots of the dispersion were taken, diluted, and used to determine the extinction coefficient (a)b yu sing UV/ Visa bsorption spectroscopy.T he value of a obtained at 660 nm was 39 AE 0.9 10 2 mg À1 mL m À1 and agrees with 41.00 AE 0.4 10 2 mg À1 mL m À1 reported previously for CVD-grown SWCNTsd ispersed in DCE. [16] Self-assembly of the SWCNTsa tt he water/DCE interface was achieved by following the procedure reported previously in our laboratory. [15] Briefly,a na liquot of the SWCNT dispersion in DCE was mixed with the organic supporting electrolyte and an equal volume of the aqueous phase was placed on top of this organic phase. Assembly was then induced by a1 0min bath sonication (37 kHz and 40 %p ower). Cyclic voltammetry and potential step experiments were carried out with an Autolab potentiostat PGSTAT20 (Metrohm-Autolab) operated in af our-electrode configuration mode with IR compensation applied during all cyclic voltammetry measurements. The applied potential was converted to Galvani potential difference (Df)b yu sing the standard ion transfer of TMA + ion (D w 0 f)t aken as + 160 mV for the water/DCE system. [28] The electrochemical cell used had ag eometric area of either 0.69 or 1.0 cm 2 and was similar to that reported elsewhere. [15a] Optical images of SWCNT interfacial films were recorded with as tereozoom microscope (SMZ168, Motic) connected to ad igital live camera (GXCAM-9, GX Optical). SEM images were obtained by using an FEI XL30 Environmental SEM-FEG operated under highvacuum state with an accelerating voltage of 15 keV.X PS was performed by using aK -Alpha X-ray photoelectron spectrometer (Fisher scientific) located at the EPSRC NEXUS facility,N ewcastle University,U K. The survey spectra were taken at 0.4 eV step size at three different locations on each sample. All experiments were carried out at room temperature.