Thin films of pure vanadium nitride: Evidence for anomalous non ...

16 downloads 6647 Views 1MB Size Report
surface layer. Such an amount of electrons confined in a nanosized. layer can behave as quasi 2D electron gas, which can result in. quantized capacitance [13].
Journal of Power Sources 324 (2016) 439e446

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Thin films of pure vanadium nitride: Evidence for anomalous nonfaradaic capacitance langer c, Eider Goikolea a, Oleksandr Bondarchuk a, *, Alban Morel b, c, d, Daniel Be Thierry Brousse b, d, Roman Mysyk a gico, c/Albert Einstein 48, 01510 Min ~ ano, Alava, Spain CIC energiGUNE, Parque Tecnolo Institut des Mat eriaux Jean Rouxel (IMN), CNRS UMR 6502 e Universit e de Nantes, 44322 Nantes Cedex 3, France c   Montr Departement Chimie, Universit e du Qu ebec a eal, Case Postale 8888, Succursale Centre-Ville, H3C 3P8 Montr eal, Qu ebec, Canada d  Reseau sur le Stockage Electrochimique de l’Energie, CNRS FR 3459, 80039 Amiens Cedex, France a

b

h i g h l i g h t s  Pure oxygen-free VN films can deliver surface capacitance up to ~3 mF/cm2 in KOH.  In contrast to vanadium oxynitrides No redox reactions are involved.  An alternative mechanism e subsurface space charge accumulation e is proposed.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 27 November 2015 Received in revised form 19 May 2016 Accepted 20 May 2016

An impressive gravimetric capacitance of 1300 F g1 (surface capacitance ~3.3 mF cm2) reported by Choi et al., 2006 for nanosized vanadium nitride has stimulated considerable interest in vanadium nitride as a potential electrode material for energy storing systems e supercapacitors. The postulated mechanism of charge storage in vanadium nitride materials involves redox reactions in the thin surface layer of vanadium oxide while the core vanadium nitride serves exclusively as a conducting platform. In this study we have synthesized pure oxygen-free vanadium nitride films and have found that they are capable of delivering a surface capacitance of up to ~3 mF cm2 at a potential scan rate of 3 mV s1 and ~2 mF cm2 at a potential scan rate of 1 V s1 in aqueous electrolytes. Combining electrochemical testing with X-ray photoelectron spectroscopy characterization has revealed that redox reactions play no or little role in the electrochemical response of pure VN, in contrast to the common wisdom stemming from the electrochemical response of oxygen-containing films. An alternative charge storage mechanism e space charge accumulation in a subsurface layer of ~100 nm e was put forward to explain the experimentally observed capacitance of VN films in aqueous electrolytes. © 2016 Elsevier B.V. All rights reserved.

Keywords: Supercapacitors Vanadium nitride Thin films X-ray photoelectron spectroscopy

1. Introduction Vanadium nitride has recently attracted considerable attention because of its possible application in electrochemical energy storage devices esupercapacitors. Choi et al. have demonstrated that nanosized VN can deliver an impressive gravimetric capacitance of up to ~1300 F g1 at a scan rate of 2 mV/s in 1M KOH electrolyte. The demonstrated capacitance is comparable with the performance of

gico, c/Albert Einstein * Corresponding author. CIC energiGUNE, Parque Tecnolo ~ ano, Alava, Spain. Tel.: þ34 945 297 108. 48, 01510 Min E-mail address: [email protected] (O. Bondarchuk). http://dx.doi.org/10.1016/j.jpowsour.2016.05.093 0378-7753/© 2016 Elsevier B.V. All rights reserved.

the benchmark - RuO2 [1e3]. However, in contrast to the latter, VN is less expensive and is more environmentally viable. In the seminal works of P.N. Kumta’s group it has been postulated that the observed remarkable capacitance of nano-VN stems from reversible redox reactions occurring at the thing vanadium oxide shell [4,5]:

VNx Oy þ OH  4VNx Oy k OH þVNx Oy  OH

(1)

where component VNxOykOH- represents electrical double layer (EDL) formed by the hydroxyl groups adsorbed on the surface and component VNxOy eOH represents oxidation of the VNxOy surface by the hydroxyl groups. The latter is considered as the major contributor to the observed specific (pseudo)capacitance [6]. Since

440

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

then this pseudocapacitive mechanism of charge storage has been attributed to capacitance observed in various VN based materials studied later on [5] [7e10]. In this paradigm, a VN nanoparticle’s core serves exclusively as a conductive platform for vanadium oxide surface phases, which are generally poor electrical conductors. But the question remains what would be a contribution of vanadium nitride if it was not encapsulated into the vanadium oxide shell? To the best of the authors’ knowledge, the electrochemical activity of pure VN, in whatever form, has never been reported. The synthesis of oxygen-free VN is a challenging task. Indeed, for thermodynamic reasons, the formation of vanadium oxide phases is much more favorable than that of vanadium nitride (enthalpy of formation is 217 kJ mol1 for VN vs e1491e1550 kJ mol1 for V2O5 and e1093e1218 kJ mol1 for V2O3 and e371e432 kJ mol1 for VO [11]. Moreover, the widely used ammonia-based synthesis of VN can also lead to the presence of an uncontrollable amount of hydrogen in the synthesized material, which, in turn, can affect the electronic properties of VN surface [12]. To elucidate the role of vanadium nitride in the charge storage mechanism we have chosen to use VN thin films as model systems. In this paper we report on a successful synthesis of oxygen-free VN thin films of various thicknesses employing ammonia-free process e direct nitridation of in-vacuo deposited metallic vanadium by annealing it in pure nitrogen at high pressure (1bar) and high temperature (800  C). We have also shown that pure VN films grown on tantalum foil are electrochemically active demonstrating an unexpectedly high surface capacitance of ~2 mF cm2 when tested at a high potential sweep rate of 1 V s1. The observed surface capacitance exceeds the common values of EDL capacitance (~102 mF cm2) by two orders of magnitude while no redox reactions on the VN films have been evidenced neither by cyclic voltammetry nor by recording X-ray photoelectron spectra before and after electrochemical testing. We argue that another capacitance mechanism, not considered before, is to be invoked to explain the observed surface capacitance of pure VN e space charge accumulation (SCA), when the accumulated charge (~1016 electrons cm2) is stored in an up to 100 nm-thick surface layer. Such an amount of electrons confined in a nanosized layer can behave as quasi 2D electron gas, which can result in quantized capacitance [13]. 2. Experimental section The experiments were carried out in an ultra high vacuum multichamber multitechnique system (SPECS GmbH) shown and described in Fig. 1. 2.1. Vanadium nitride film growth The VN films were grown on Ta foil in 2-step process e pure vanadium metal deposition on the supporting Ta film followed by direct nitridation. 2.1.1. Vanadium deposition The pieces of Ta foil (Alfa Aesar) with size of 10  10  0.125 mm3 were used as substrate for vanadium deposition. In this work we have done deposition of metallic vanadium in two ways: e-beam assisted evaporation in the UHV chamber (insitu) and D.C. magnetron sputtering (ex-situ) with deposition rate ~0.01 Å/sec and 0.5e0.7 Å/sec correspondently. In both cases the deposition rate was monitored by quartz crystal microbalance positioned in vicinity of the substrate. In the XPS system vanadium was evaporated from :2 mm vanadium rod (99.8%, Goodfellow) by means of an e-beam evaporator EBE-4 (SPECS GmbH). The deposition of thick metal films (30 nm and more) was performed ex-situ in a magnetron sputtering system Classic 500SP (Pfeiffer Vacuum).

Vanadium target (99.5%) from Kurt J. Lesker was used as a deposition source. Thickness of the deposited thick vanadium films was estimated by measuring the thickness of the “witness” film grown on a piece of masked Si wafer placed next to the Ta substrate during vanadium deposition. A piece of narrow Si bar (~3 mm  10 mm) cut out from the Si wafer was used as a growth mask clamped down to the silicon substrate. The depth of the trough created on the Si wafer upon vanadium deposition was measured with the help of stylus profiler DektakXT (Bruker). The precision of the described above method for film thickness measurements of films thicker than 10 nm was better than 20%. The substrate was kept at room temperature during vanadium deposition. In the case of ex-situ vanadium deposition the vanadium film was coated with a 1e2 nm graphite layer, using also DC sputtering before transferring it further to the XPS chamber for nitridation. This carbon coating was quite effective to prevent oxidation of the film during transferring it from the deposition chamber into the UHV chamber. The carbon coating layer was sputtered away by argon ions upon introducing the sample into the UHV chamber. 2.1.2. Direct nitridation Vanadium metal films were directly nitridated in the HPC by exposure to ~1bar of N2 (5X, Praxair) at elevated temperature of 800  C for 30 min. To prevent unwanted oxidation of vanadium the HPC and gas line were thoroughly outgassed by prolonged bakeout until pressure 1  109 mbar or less was reached. The sample temperature during nitridation was controlled using a K-type thermocouple integrated into the sample holder (SPECS) to which the Ta foil was spotwelded. Upon heating and dwelling at high temperature the sample was cooled down to room temperature in nitrogen atmosphere. Films with different thicknesses were grown. The thickness of the films varied from 1 nm to 400 nm. The surface capacitance was determined using formula (2) from the corresponding cyclic voltammetry (CV) recorded at a scan rate of 30 mV s1 in 1M KOH electrolyte. In the case of the 1 nm film, the film was grown on an HOPG substrate. Ex-situ AFM imaging revealed clusters, which were 50e80 nm in diameter and 12e20 nm in height. All the CVs were recorded in the potential range of the electrochemical stability window, which was empirically established for each tested film separately. 2.2. Characterization of samples 2.2.1. Electrochemical characterization Electrochemical characterizations of the synthesized VN films were performed in the Electrochemical Cell chamber (ECC) integrated to the XPS system. The ECC is schematically depicted in Fig. S1(A). During the electrochemical testing the ECC was backfilled with continuously flowing Ar gas. The 3-electrode setup for electrochemical tests is schematically depicted in Fig. S1(B). As a reference electrode was used an Ag/AgCl electrode (Sensortechnik Meinsberg). The counter electrode was made out of a Pt plate which measured 25  12.5  0.3 mm3. The electrolytes used for testing were: 0.1M KOH, 0.3M KOH, 1M KOH, 3M KOH, 0.6M K2SO4, 1M LiOH and 1M Li2SO4. Electrochemical measurements were controlled by an SP200 potentiostat (Bio-Logic). 2.2.2. X-ray photoelectron spectroscopy (XPS) characterization XPS characterization was used to follow changes in the film’s composition and oxidation state of the film’s constituent elements after electrochemical testing. The XPS spectra were acquired by means of a hemispherical electron energy analyzer PHOIBOS 150 (SPECS) and a twin Al/Mg anode X-ray source XR50 (SPECS) operated at 12 keV and power 100 W. The XPS spectra were acquired in

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

441

Fig. 1. Ultra high vacuum multichamber and multitechnique system (SPECS GmbH) with base pressure below 5$1010 mbar. The system consists of an analysis chamber (AC), a sample preparation (PrC) chamber and also include integrated high pressure (HPC) and electrochemical cells (EC). Samples under study can be transferred between different compartments of the system for treatment and characterization without being exposed to the atmosphere. The Analysis Chamber features instruments for X-ray Photoelectron Spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS) measurements e hemispherical electron energy analyzer FOIBOS-150-3D DLD (SPECS GmbH) and Mg/Al dual anode non-monochromated X-ray source XR50 (SPECS GmbH) and Al/Ag monochromated source FOCUS 500 (Specs GmbH), ultraviolet radiation source UVS20 (SPECS GmbH), Ar ion sputter gun IQE-12 (SPECS GmbH), 4-coordinates sample manipulator with the integrated e-beam sample heater. The preparation chamber is equipped with e-beam evaporator EBE-4 (SPECS GmbH), a quartz crystal microbalance (Inficon), low energy electron diffraction system ErLEED100 (SPECS GmbH), residual gas analyzer Dycor LC-D100 (Ametek) and an X-Y-Z-rotation manipulator with an integrated e-beam heater. In the PrC the substrate was cleaned by cycles of sputtering/annealing before metal deposition. High Pressure Cell HPC20 (Specs GmbH) allows for sample radiative heating by an external halogen lamp up to temperature of 800  C while being pressurised up to 20 bar.

the fixed analyzer transmission mode with pass energies 20 eV and 60 eV for detecting elemental spectral lines with high resolution and survey spectra respectively. The XPS spectra were peak-fitted using CasaXPS data processing software. Quantification has been done using sensitivity factors provided by the elemental library of CasaXPS [14]. Every time after taking the sample out from the electrolyte and before proceeding for XPS measurements the sample were thoroughly rinsed with DI water and dried under Ar flow, and at no point was the sample exposed to air. Drying without water rinsing always led to precipitation of electrolyte’s components on the film’s surface. In the case of KOH XPS spectra evidenced the presence of potassium and vanadium in 5þ oxidation state which can be explained by formation of potassium vanadate in accordance to the Pourbaix diagram for vanadium in water [15]. Once detected, potassium traces could be washed away by additional rinsing of the surface. XPS always detected traces of oxygen and carbon on VN surface after contacting with electrolyte as seen in Fig. S2. XPS depth profiling of the grown films was performed by means of a sputter gun IQ12 (SPECS) operated in scanning mode with Ar ion beam of 2 keV kinetic energy. The estimated sputter rate was ~0.01e0.02 nm s1. 2.2.3. Structure determination X-ray diffractometer D8 (Bruker) was used to determine the structure of the thick VN films. The spectrometer was operated in grazing mode.

2.2.4. Morphology characterization Atomic force microscope (Agilent M5500) operated in noncontact mode was deployed to assess surface topography of the grown VN films. Cross sectional analysis of the films morphology was performed by means of scanning electron microscope Quanta 200 (FEI). 3. Results and discussion 3.1. Vanadium nitride films growth In this work we have developed a procedure of VN thin film synthesis via direct nitridation of the pure vanadium film deposited on the tantalum foil in vacuum. Vanadium nitridation was achieved by annealing in 1bar of N2 at 800  C. The details of the synthesis are described in the experimental section. VN films of 3 nm, 10 nm, 30 nm, 80 nm, 160 nm, 240 nm and 400 nm thicknesses were grown on Ta foil. The 1 nm thick film was grown on the highly oriented pyrolitic graphite (HOPG). Ex-situ atomic force microscopy inspection of the 1 nm film revealed formation of nanoclusters of 50e80 nm in diameter and ~20 nm in height (Fig. S3(B)). Importantly, design of the used vacuum system (Fig. 1) ensured that the electrochemical testing and x-ray photoelectron spectroscopy characterization could be conducted without ambient exposure of as-grown films, thus preventing their oxidation. Fig. 2A shows XPS spectra of 160 nm as-prepared vanadium nitride film. The survey spectrum reveals little contamination of the

442

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

arrows. The positions of V2p3/2 peak and correspondent satellite are close to 513.5 eV and 516 eV reported for the VN films epitaxially grown on Pt(111) respectively [12]. The shape and positions of the V2p spectra did not reveal features characteristic of vanadium reduction. XPS derived atomic concentrations of V and N (numbers in parenthesis in Fig. 2A) were 52 at% and 48 at% respectively. These concentration values varied less than 5% from film to film. Thus the ratio of the atomic concentrations V:N ¼ 52:48 z 1 which is close to the stoichiometric value of the cubic d-VN phase. XRD measurements (Fig. S4 in Supporting Information) confirmed the film’s cubic crystal structure. The XPS and XRD data did not reveal formation of another known vanadium nitride phase eV2N. Time prolonged depth profiling proved that nitridation propagated all the way through the film for all used film thicknesses and that vanadium oxide phase, if any, was mainly located at the top surface. Effect of VN film’s disruption can be also ruled out, at least for 10 nm films or thicker, as the XPS signal from the tantalum substrate was fully screened by the on-top grown VN film. For the 3 nm thick VN film the XPS signal from Ta substrate was detectable. The Ta 4f XPS signal was attenuated by factor of ~6.5e7, which corresponds to the effective VN film thickness ~1.8e1.9 lMFP z 3.6e3.8 nm, where: lMFP ~ 2 nm is the mean free path for Ta 4f electrons with kinetic energy ~1230 eV when excited by the Mg Ka [16]. Values of XPS derived film thickness and film thickness by growth correlate quite well thus supporting that the 3 nm VN film was rather continuous. In Fig. 2B is shown high resolution V2p and O1s photoelectron spectra (XPS) of another VN 160 nm film before and after electrochemical testing. Freshly prepared VN films get slightly oxidized upon introduction into the electrochemical cell apparently due to non-zero humidity in the electrochemical cell (water vapor from the electrolyte and also argon gas flowing through the electrochemical cell contains up to ~1 ppm of water). The oxidation is evidenced by a small peak centered at ~530 eV of the as-grown film (solid line) which represents a typical 1s XPS spectral line of oxygen in metal oxides. The corresponding VOx spectral signature in the V2p spectrum is overshadowed by the shake-up satellite peaks. The total content of vanadium atoms bound to oxygen is estimated to be 5e10 at% to the most for all studied films. As though, we have proven that virtually oxygen-free VN films can be synthesized via direct nitridation of metallic vanadium.

Fig. 2. (A) Survey spectrum of the VN 160 nm film as grown. V 2p and N 1s spectral lines were used to estimate atomic concentrations presented in the parenthesis. The inset shows V2p and O1s spectral range indicating virtually no oxygen. The arrows labeled with (s) mark shake-up satellites characteristic for VN. (B) V2p and O1s XPS spectra of another VN 160 nm film before and after electrochemical testing. Solid line depicts V2p and O1s XPS spectral range for the as-grown film which was exposed to the environment of the argon filled electrochemical cell (before contacting the electrolite). The XPS spectra revealed ~5 at% of oxygen. Dotted and dashed lines represent spectra taken after charging to 0.4 V vs the Ag/AgCl electrode in 1M KOH (positive cut-off potential) and after charging to e1 V vs Ag/AgCl electrode in 1M KOH (negative cut-off potential) respectively. The evolution of oxygen 1s line reflects accumulation of contaminations: spectrum at 0.4 V was taken before the spectrum at e1 V.

film. The inset depicts a high resolution XPS spectrum in the energy region of V2p and O1s spectral lines. It is clear that there is virtually no oxygen present in the film. The V2p spectra contain the 2p3/2 emission at 513.3e513.7 eV and broad satellites on the high binding energy sides of the V2p peaks labeled with “s” and marked by

Fig. 3. CVs of VN film grown on Ta foil in comparison with CV of the bare Ta foil (black line). The VN film thickness was 400 nm, the Ta foil size was 10  5 mm2, electrolyte: 1M KOH, the counter electrode was a Pt plate with size 25  12.5  0.3 mm3.

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

3.2. Electrochemical testing of vanadium nitride films Fig. 3 depicts cyclic voltammograms of the 400 nm thick VN film measured in 1M KOH electrolyte at different scan rates. The voltammogram of the bare Ta substrate taken at 30 mV s1 is plotted on the same graph for comparison. For each cyclic voltammograms in Fig. 3 the measured current was normalized to the correspondent voltage scan rate. The normalized curves in Fig. 3 overlap very nicely for all used scan rates, meaning nearly ideal capacitive behavior of the VN film. The electrochemical stability window was limited to ~0.45 V and was found to lie between 0.65 V and 0.2 V. The shape of all cyclic voltammograms within the electrochemical window is close to rectangular. Further increasing the scan rate leads to a more resistive shape of the voltammograms as commonly expected [3]. The value of scan-rate-normalized current of the VN supported film is clearly much higher than that of the bare Ta substrate. The SEM and AFM (Fig. S3) inspections of the film’s morphology did not reveal roughness or porosity of the grown VN film, which could explain the enhanced capacitance of VN-coated Ta foil as compared to bare Ta foil. Noteworthy, in contrast to the results of other studies on VN films, powders, nanowires and nanoclusters no well-defined redox peaks have been observed in the CVs for all the used scan rates [4,7e10,17,18]. There is only one wide hump centered at ~0.4 V. This hump makes up about 4e5% of total area under the CV curve and it was found the hump’s position depends on the films thickness as it can be seen by comparing CVs in Fig. 3 for 400 nm film with the

443

ones for 240 nm film in Fig. 4A, where the hump is centered at ~0.4 V and ~0.75 V correspondently. Moreover, no redox peaks have been observed for all the fabricated and examined films. Additionally, the shape of cyclic voltammograms settles down upon the first couple of cycles, showing no further changes for at least 10,000 cycles in 1M KOH. To shed more light on the mechanism behind the surface capacitance of pure VN films we turned to the X-ray photoelectron spectroscopy (XPS). In this work we performed XPS measurements on in-situ grown VN films before and after electrochemical cycling. To probe different charging states with XPS, the potential scanning was stopped at different potentials e at the upper and low limits of the potential window and in between (for instance at 0.4 V where a hump in the CV of the 400 nm film was observed). It was dwelled on the corresponded potential till the electrode current has equilibrated. Once the electrode current stabilized at the level of couple microamperes or less (leakage current), the sample was ejected from the electrolyte, thoroughly rinsed with DI water inside the electrochemical cell and then transferred to the analysis chamber for XPS measurements without exposing to the ambient atmosphere. Fig. 2B shows V2p and O1s XPS spectra of the VN 160 nm film: i) as grown (solid line), ii) after charging to 1V vs Ag/AgCl (dashed line) and iii) after charging to 0.4 V (dotted line) vs Ag/ AgCl. Noteworthy, the V2p spectrum did not undergo visible changes in the shape and positions of the peaks upon electrochemical cycling. No changes caused by electrochemical processes have also been detected in the N1s spectra. XPS measurements did not detect signals from potassium neither for the charged state nor for the discharged state. The detected small increase in C1s XPS signal and an additional component in the O1s spectrum with the binding energy ~532 eV (see the dashed and dotted spectra in Fig. 2B) are due to unavoidable surface contamination after exposure to the electrolyte. In case of the films with high oxygen content, observed for samples stored on air for long time (Fig. S2, red curve), the VOx component in the as-grown film usually dissolved upon electrochemical testing, which was evidenced by a decrease in the corresponded feature at ~515e517 eV in the V2p spectrum and by concomitant decrease in the O1s feature at ~530 eV. The dissolution of surface vanadium oxide species in KOH is in line with observation reported by Choi et al. [1]. From the XPS results we can conclude that, unlike VOx, vanadium in pure VN films does not virtually change its oxidation state after electrochemical cycling whatever the charge state (cut-off potential) it has attained [4,5]. As such, one can assign the observed electrochemical response not to vanadium oxide or oxynitride, but primarily to VN. To our knowledge, this is the first time that the electrochemical activity of pure VN material has been observed. Surface capacitance of the VN films was estimated by formula:

Z

t2

IðtÞdt C¼

Fig. 4. Effect of electrolyte. (A) CVs for VN 240 nm thick film taken in KOH electrolyte with different concentrations: 0.1M, 0.3M, 1M and 3M KOH. (B) CVs for VN 160 nm thick film taken in different electrolytes: 1M KOH and 0.6M K2SO4. The scan rate in all cases was 100 mV/s. For comparison purposes the CVs were scanned in the same potential range as the practical potential stability window determined in 1M KOH electrolyte, i.e., from 1V through 0.5V. By contrast, the potential stability window in 0.6M K2SO4 was found to be much wider: from [1.0: 0.2]V vs Ag/AgCl.

t1

½Eðt2Þ  Eðt1ÞS

;

(2)

where: E(t2) and E(t1) are the positive and negative cut-off potentials, S is the area of the electrode. The surface area estimated from the topography images (Figure S3(A)) differed from the projected area by less than 5%. Therefore, the geometrical area of the Ta substrate was used to quantify capacitance per unit area. The obtained values of surface capacitance are in the range of 2e3 mF cm2 when tested in 1M KOH electrolyte. To elucidate the role of the electrolyte we have tested the VN films in KOH solutions with different concentrations and additionally in neutral electrolyte K2SO4. The effect of varying the KOH concentration in the range of 0.1M KOH to 3M KOH is shown in Fig. 4A. Clearly, varying the concentration by an order of magnitude

444

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

e from 0.3M KOH to 3M KOH, did not virtually affect the shape of the voltammograms recorded at 100 mV s1. Further decreasing the KOH concentration down to 0.1M leads to a more resistive CV shape (the blue curve in Fig. 4A), which can be explained by the low conductivity of diluted electrolyte. Also no substantial changes in the CVs of the VN films have been detected when basic KOH electrolyte was replaced by neutral K2SO4electrolyte, as evidenced by Fig. 4B. Interestingly, when 0.6M K2SO4 electrolyte was used VN films showed to be stable in a much wider potential window: [1.0;0.2] V (not shown here). Similarly, changing electrolyte for 1M LiOH, 1M Li2SO4, 1M NaOH and 1M Na2SO4 had little effect on the measured surface capacitance of the VN films compared to 1M KOH. Summarizing, by varying KOH concentration by an order of magnitude and by experimenting with different basic and neutral aqueous electrolytes we have clearly shown that OH ions do not play a dominating role in the charge storage mechanism in pure VN films. This distinguishes the electrochemical behavior and charge storage mechanism of pure VN films from those of vanadium oxynitride reported up to now. Most importantly, the question arises of what is the mechanism behind the huge measured surface capacitance of 2e3 mF cm2 while at the same time Faradaic reactions are ruled out? Indeed, the observed capacitance of VN films exceeds by two orders of magnitude the typical surface capacitance of carbon-based materials and is even higher than the value of ~0.1 mF cm2 reported for some metallic surfaces [19]. Such a capacitance value means accumulation of ~103 C cm2 of charge or ~1016 charged particles per surface unit area of 1 cm2. Assuming one electron per unit cell (the unit cell size in cubic VN is ~0.4 nm) the thickness of the charged layer is estimated to be ~60 nm. That necessarily leads one to the idea that the charge storage process propagates into the bulk of material - a conclusion opposite to the redox surface charge storage postulated in publications of Choi et al. and Hanumantha et al. [4,5]. 3.3. Capacitance dependence on the VN film thickness To verify the mechanism of charge storage in a subsurface layer of finite thickness, the value of surface capacitance was measured as a function of VN film thickness. Interestingly, the potential window was found to vary irregularly by 0.1e0.2 V when the film thickness changed. For all the film thicknesses the potential window was found substantially narrower than that of the water-based electrolyte (1.23 V). Fig. 5 exhibits the main finding of this study e the dependence of surface capacitance on the VN film thickness. We have found that surface capacitance measured at a scan rate of 30 mV/s increases with the film thickness until it reaches a plateau at ~2e3 mF cm2 when the film thickness has grown up to ~100 nm. The experimental data can be fitted with function yðxÞ ¼ A  Bex=t shown in Fig. 5 (the dashed line). The fitting gives a characteristic value of parameter t~30 nm ± 10 nm. Thus, the film thickness dependence of surface capacitance shown in Fig. 5 confirms that the charge accumulates in a subsurface region of finite thickness. The characteristic thickness of ~30 nm can be thought of as a charge screening length. We will try to relate these two experimentally determined quantities: the effective thickness of 30 nm and the accumulated charge density Nsc ~1016 per cm2. Another noteworthy observation made in the course of measuring the thickness dependence is the result for the 1 nm thick film e the thinnest film used in this study. In the latter case VN was deposited on the surface of highly oriented pyrolytic graphite (HOPG). The AFM topography image in Fig. S3(B) reveals clusters partially covering the HOPG substrate. The height of the clusters was in the range 10e20 nm and the lateral sizes were around 100 nm. The measured surface capacitance of the vanadium nitride clusters supported on the HOPG was 0.13 mF cm2 in contrast to

Fig. 5. Specific capacitance of VN films plotted as a function of film thickness. The surface capacitance (solid rectangular) was determined from the cyclic voltammograms recorded at the potential scan rate 30 mV/s in 1M KOH. The cyclic voltammogram of very thin film shows resistive character when was recorded at high scan rates. The fitting function (dashed line) reveals characteristic thickness ~30 nm. The 1 nm thick film was grown on highly oriented pyrolytic graphite. The gravimetric capacitance (solid stars) was calculated as: gravimetric capacitance (F g1) ¼ C/S  h  r, where: C e is the surface, S ¼ 1 cm2 is the geometric area of the electrode, h is the film thickness, r ¼ 6.13 g cm3 is the theoretical density of VN.

0.015e0.018 mF cm2 determined for pure HOPG surface in 1M KOH electrolyte. The exposed surface area in Fig. S3(B) is 6% larger than the corresponded projected area. Assuming only the EDL charge storage mechanism the expected surface capacitance of the 1 nm vanadium nitride grown on HOPG should be ~0.02 mF cm2 but the experimental value is six times larger! No characteristic peaks in the CV curves of the 1 nm VN were observed e similar to the thicker VN films grown on tantalum. Thus one can speculate that the same charge storage mechanism takes place for VN clusters as it does for continuous VN films. This preliminary conclusion is subject to prove in the course of the authors’ ongoing research of the electrochemical properties of supported VN clusters. The latter can be considered as a model system for the powder form of VN, which is the practically relevant form of material, but is indeed the most difficult to obtain oxygen-free as finely divided nanoparticles. In addition, the capacitance measured for very thin films is close to 275 F g1. This value drastically drops down upon increasing the film thickness (Fig. 5). However, the capacitance values measured at a scan rate of 30 mV s1 for 5 nm vanadium oxynitride powder (700 F g1) [4] are 3 times this value, which also emphasizes the role of oxygen in charge storage.

3.4. Quantitative consideration Let us assume that the charge is distributed in the 30 nm subsurface layer. The accumulated charge density estimate will be ~1021 cm3. The last value puts the material in-between highly doped semiconductors (carrier concentration ~1018 cm3) and metals (electron concentration ~10221023 cm3). For metals the Thomas-Fermi approach for the electron screening of an external field should be considered [20]. Let us verify whether the ThomasFermi approximation is applicable in the case of a VN film, i.e., if the inequality kFd [ 1 is true. Here kF ¼ (3p2/n)1/3 is the wave vector of an electron at the Fermi level, n is electron concentration and d is the characteristic thickness of the charged volume. For n ¼ 1021 cm3 and d~30 nm the product kFd is ~92[1. For a 3 nm thick film the product kFd is ~7.5. Thus the Thomas-Fermi approximation should apply practically for all the used film

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

445

thicknesses. In the Thomas-Fermi approach the density of the charge induced in material by external electric field rind(x) and potential profile 4(x) in the material are related as:

rind ðxÞ ¼ e2

vn 4ðxÞ; vm

(3)

 3=2 2 where n is electron concentration, m ¼ 3p2 ðZ =2mÞn3=2 is the chemical potential of the electron gas, m and e are the electron mass and charge, respectively, x is the position in space. Here the external electric field is perpendicular to the electrode-electrolyte interface and the electron concentration depends only on the distance from the interface: n ≡ n(x). Thus, using Gauss’s law: d2 4=dx2 ¼ rind =εε0 ; , and by replacing rind with en(x), we get a differential equation for the potential:

pffiffiffi  3=2 d2 4 1 3=2 8e me ¼ 4 ; εε0 dx2 3p2 Z2

(4)

where ε is the relative permittivity of a material, ε0 ¼ 8.85  1012 F m1 is the permittivity of vacuum. To solve Equation (4) we adopted solutions derived by Reich et al. and Lüth [13,21]:

Z2 4ðxÞ ¼ A1 e me2

nðxÞ ¼ A2

Z2 me2

!3 4

ðεε0 Þ2 ðx þ lÞ

(5)

ðεε0 Þ3 ðx þ lÞ6

(6)

!3

where: A1 ¼ 450 p4, A2 ¼ 9000 p4 and l is the characteristic length. The value of l can be obtained by imposing conditions for concentration profile n(x):

Z∞ Nsc ¼

nðxÞdx;

(7)

0

where Nsc z 1016 cm2 is the experimentally determined number of charged particles accumulated in VN at saturation (for film thicknesses  100 nm). Using (6) and (7) we get:

l½nm ¼ A3

4pZ2 ε0 me2

!3  5

ε3 Nsc

15

1 ε3 5 ¼ A3 ðaÞ z0:15ε3=5 Nsc 3 5



(8)

where: A3 ¼ (225p/32)1/5, az 53 p.m. is the Bohr radius. By integrating Equation (6) it is easy to show that a subsurface layer with thickness l confines 97% of the stored charge. Table 1 shows clearly that relative electric permittivity ε should be about 104 to reconcile the theoretical predictions for screening length with the experimental data. It is instructive to estimate the order of magnitude for ε by modeling the VN-electrolyte interface with a simple in-series capacitor model shown in Fig. 6A where Cel and Csc represent the charges accumulated in the electrolyte and in the VN film, respectively. Loosely assuming that capacitance Csc is comparable, at least by the order of magnitude, with the experimentally

Table 1 Effective screening length l as a function of electric permittivity calculated using Equation (8). ε

10

100

1000

10,000

100,000

l, nm

0.6

2.4

9.5

37.7

150

Fig. 6. (A) Charge storage at the electrolyte-semiconductor interface. The charge accumulates in a subsurface layer of characteristic thickness~30 nm. Total measured 2 1 1 capacitance at saturation C1 tot ¼ Csc þ Cel , here Ctot is ~2e3 mF/cm . This capacitance values correspond to ~1016 cm2 accumulated charges. (B) ~1016 cm2 charges accumulated in a 30 nm thick subsurface layer can be viewed as a quasi 2D electron gas confined in a potential well with partially filled energy bands. The pattern of energy levels in the potential well will depend on the film thickness for films thinner than the characteristic thickness. Therefore, the position of the electrode’s Fermi level wrt the redox potential of electrolyte will vary with the film thickness. Hence the size and the cut offs of the potential window will also depend on the film thickness for very thin films.

determined surface capacitance, we get an estimate for ε: ε ¼ Csc d=ε0 z105 . Here: Csc  3 mF cm2 and d ¼ 30 nm. Such a gigantic value of relative permittivity of pure VN thin films equates VN with conjugated polymers, which combine metallic conductivity and high dielectric constants of up to ~105 [22]. In the latter case, giant polarization is conferred by long-range orbital delocalization e a polarization mechanism different from the conventional dipole orientation mode. It remains to be investigated what mechanism is behind such huge dielectric constant determined for the VN thin films. Conversely, one should keep in mind that the value of relative permittivity was inferred from the experimental data, assuming a linear relationship between the electric field and the polarization of the material. Hence, such a huge value of the electric permittivity estimate can be an indication that nonlinear polarization needs to be taken into account. Indeed, a typical strength of electric field in the interface area is ~105 V cm1. This can bring the interface into the conditions when nonlinear dielectric response is observed for ferroelectrics and polymers [23e25].

446

O. Bondarchuk et al. / Journal of Power Sources 324 (2016) 439e446

For nonlinear dielectric response of SrTiO3 Reich et al. used a nonlinear relationship between the electric field E and electric in~ 3 , resulting in the following form for the duction D [13]:EzAD charge density profile in the material: nðxÞ∝ðx þ lÞ12=7 . That means the induced charge tails off in the bulk of the material on the much larger distance than predicted by Equation (6) for linear dielectric response. A simple integration of the n(x) profile shows that only 39% of the charges are located within the l-layer and a layer of ~50l is needed to accommodate 95% of charges when nonlinearity is introduced into the model. The polarization mechanism at the vanadium nitride-electrolyte interface is subject to further studies and goes beyond the scope of the current publication. To summarize, the observed anomalous surface capacitance of the VN films could be explained by space charge accumulation accompanied by polarization of the material. The latter could exhibit even non-linear behavior. Preliminary, by probing the valence band of the VN films by means of ultraviolet photoelectron spectroscopy (UPS) we have found irreversible reduction of the work function from 4.6 eV of the pristine film down to 4.3 eV upon electrochemical testing. By comparing the DFT calculations with the UPS results obtained for epitaxial VN films, Glaser et al. concluded that surface hydrogenation could create such an effect on the electronic structure of the vanadium nitride films [12]. It is also worth noting that ~1016 charges per 1 cm2 confined in the ~30 nm thick layer can be thought of as 2D electron gas. As though, quantum effects could also play a role. Fig. 6B schematically shows a quantum well formed at the electrode-electrolyte interface. The Fermi level of the partially filled band can vary its position with respect to the redox level of the electrolyte as the film thickness changes, thus affecting the electrochemical response of the film. The observed fluctuation of the cut-off potentials in the range of 0.2e0.3 eV when switching from one film thickness to another could be a manifestation of the quantum nature of the electrochemically active interface layer. Moreover, the above-mentioned hump observed in the CVs could be also a manifestation of the non-monotonic energy dependence of the surface density of states due to energy quantization. 4. Conclusion To summarize, pure vanadium nitride films with thickness from 1 nm to 400 nm were used in this work as model systems in order to get insight into the details of the electrochemical activity of vanadium nitride based materials. It has been found that oxygenfree vanadium nitride films can be synthesized via direct oxidation of metallic vanadium, prepared in vacuo. These films can deliver a surface capacitance of ~3 mF/cm2 at a potential scan rate of 3 mV s1 and ~2 mF cm2 at a potential scan rate of 1 V s1 in basic (1M KOH, 1M LiOH) or neutral (Li2SO4, K2SO4) electrolytes while no redox reactions can be assigned to the charge/discharge process. By measuring the film thickness dependence of capacitance we have found that there is an active subsurface layer with a charge-accumulating effective thickness of up to ~100 nm. Combining electrochemical cycling with surface characterization tools, we have shown that the charge storage mechanism, which is postulated earlier for the VNxOy materials, involving reduction of vanadium cations by hydroxyl groups, does not apply to pure VN. Another charge storage mechanism, although different from the classical EDL, appears to be behind the observed surface capacitance in the VN films. We invoked the space charge accumulation mechanism, known to occur at the junction between p- and n-type semiconductors, as the alternative to the common EDL and surface

faradaic reactions (pseudocapacitance). The synthesis of oxygen free VN films was crucial to unveil the space charge mechanism which was apparently obscured in the VNxOy materials by redox reactions occurring on the surface. The space charge accumulation is the major contributor to the substantial surface capacitance of VN films in aqueous electrolytes. This result can have more general implications for the range of metal nitrides which are mainly good electrical conductors, chemically stable, and environmentally viable. Thus, they could be good candidates as electrode materials in the energy-efficient high-rate energy storage systems. Acknowledgment The work was funded by the Basque Government under the Etortek Energigune’12 Program. O.B. thanks Dr. Shamil Shaikhutdinov and Dr. Igor lyubinetsky for fruitful discussion of this work. We would like to thank all of the anonymous Reviewers for their helpful comments and suggestions. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.05.093. References [1] J. Zheng, T. Jow, J. Electrochem. Soc. 142 (1995) L6eL8. [2] J. Zheng, P. Cygan, T. Jow, J. Electrochem. Soc. 142 (1995) 2699e2703. [3] B.E. Conwey, Electrochemical Supercapacitors: Scientific Fundamentals and Technological Applications, Springer, New York, 1999. [4] D. Choi, G.E. Blomgren, P.N. Kumta, Adv. Mater. 18 (2006) 1178. [5] P.J. Hanumantha, M.K. Datta, K.S. Kadakia, D.H. Hong, S.J. Chung, M.C. Tam, J.A. Poston, A. Manivannan, P.N. Kumta, J. Electrochem. Soc. 160 (2013) A2195eA2206. langer, J.W. Long, J. Electrochem. Soc. 162 (2015) [6] T. Brousse, D. Be A5185eA5189. [7] R. Lucio-Porto, S. Bouhtiyya, J.F. Pierson, A. Morel, F. Capon, P. Boulet, T. Brousse 141 (2014) 203e211. [8] C.M. Ghimbeu, E. Raymundo-Pinero, P. Fioux, F. Beguin, C. Vix-Guterl, J. Mater, Chem. 21 (2011) 13268e13275. [9] L. Zhang, C.M.B. Holt, E.J. Luber, B.C. Olsen, H. Wang, M. Danaie, X. Cui, X. Tan, V.W. Lui, W.P. Kalisvaart, D. Mitlin, J. Phys. Chem. C 115 (2011) 24381e24393. [10] A.M. Glushenkov, D. Hulicova-Jurcakova, D. Llewellyn, G.Q. Lu, Y. Chen, Chem. Mat. 22 (2010) 914e921. [11] P.J. Linstrom, W.G. Mallard, NIST Chemistry WebBook, NIST Standard Reference Database Number 69, National Institute of Standards and Technology, Gaithersburg MD, 2015, 20899. [12] A. Glaser, S. Surnev, M.G. Ramsey, P. Lazar, J. Redinger, R. Podloucky, F.P. Netzer, Surf. Sci. 601 (2007) 4817e4823. [13] K.V. Reich, M. Schecter, B.I. Shklovskii, Phys. Rev. B 91 (2015) 115303. [14] N. Fairley, @Casa Software ltd, Copyright (C) 1999-2010, 2010. [15] F.M. Al-Kharafi, W.A. Badawy, Electrochimica Acta 42 (1997) 579e586. [16] A. Jablonski, C.J. Powell (Eds.), NIST Electron Inelastic-mean-free-path Database- Version 1.2, Nation Institute of Standards and Technology, Gaithersburg, MD, 2010. [17] R.L. Porto, R. Frappier, J.B. Ducros, C. Aucher, H. Mosqueda, S. Chenu, , T. Brousse, Electrochimica Acta 82 (2012) B. Chavillon, F. Tessier, F. Chevire 257e262. [18] X. Lu, M. Yu, T. Zhai, G. Wang, S. Xie, T. Liu, C. Liang, Y. Tong, Y. Li, Nano Lett. 13 (2013) 2628e2633. [19] S. Trasatti, J. Electroanal. Chem. Interfacial Electrochem. 123 (1981) 121e139. [20] N.W. Aschcroft, N.D. Mermin, Solid State Physics, Brooks/Cole, New York, 1976. [21] H. Lüth, Solid Surfaces, Interfaces and Thin Films, sixth ed., Springer, Berlin, 2015. [22] H. Pohl, J. Electron. Mater. 15 (1986) 201e203. [23] Z. Luo, X. Lou, F. Zhang, Y. Liu, D. Chang, C. Liu, Q. Liu, B. Dkhil, M. Zhang, X. Ren, H. He, Appl. Phys. Lett. 104 (2014) 142904. [24] S. Ikeda, H. Kominami, K. Koyama, Y. Wada, J. Appl. Phys. 62 (1987) 3339e3342. [25] K.L. Ngai, J. Chem. Phys. 142 (2015) 114502.