lable at ScienceDirect

Journal of Power Sources 284 (2015) 524e535

Contents lists avai

Journal of Power Sources

journal homepage: www.elsevier .com/locate/ jpowsour

A molecular hybrid polyoxometalate-organometallic moieties and itsrelevance to supercapacitors in physiological electrolytes

Selvaraj Chinnathambi, Malika Ammam*

Faculty of Mathematics and Natural Sciences, Products and Processes for Biotechnology-Institute for Technology, Engineering & Management,University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands

h i g h l i g h t s

* Corresponding author.E-mail addresses: m78ammam@yahoo.fr, m.amma

http://dx.doi.org/10.1016/j.jpowsour.2015.03.0340378-7753/© 2015 Elsevier B.V. All rights reserved.

g r a p h i c a l a b s t r a c t

� Ruthenium bipyridineephosphomo-lybdate is formed with/without po-tassium iodide.

� Hybrids characterized by CHN anal-ysis, TEM, FT-IR, TGA, XRD andelectrochemistry.

� The hybrid synthesized with KI hasexcess Ru(bpy) moieties.

� Hybrid with excess Ru(bpy) is betterfor capacitors in physiologicalelectrolyte.

� 125 F g�1 and 68 F g�1 are obtainedfor hybrids formed with and withoutKI.

a r t i c l e i n f o

Article history:Received 12 November 2014Received in revised form19 January 2015Accepted 5 March 2015Available online 6 March 2015

Keywords:PolyoxometalatesInorganiceorganic hybridsPhysiological electrolytesElectrochemistrySupercapacitors

a b s t r a c t

Supercapacitors operating in physiological electrolytes are of great relevance for both their environ-mentally friendly aspect as well as the possibility to be employed for powering implantable micro-electronic devices using directly biological fluids as electrolytes. Polyoxometalate (POMs) have beenproven to be useful for supercapacitors in acidic media. However, in neutral pH, POMs are usually notstable. One relevant alternative is to stabilize POMs by pairing them with organic moieties to form hy-brids. In this study, we combined K6P2Mo18O62$12H2O (P2Mo18) with Ru(bpy)3Cl2.6H2O (Ru(bpy)). Thesynthesis was carried out with and without the mild reducing agent KI. The hybrids were characterizedby CHN analysis, TEM, FT-IR, XRD, TGA and cyclic voltammetry. CHN elemental analysis revealed that onemole [P2Mo18O62]6� is paired with 3 mol [Ru(bpy)3]2þ to form [Ru(bpy)3]3PMo18O62$nH2O. With KIpresent, [P2Mo18O62]

6� is linked to 3.33 mol to yield [Ru(bpy)3]3.33PMo18O62$mH2O. Excess of Ru(bpy) in[Ru(bpy)3]3.33PMo18O62$mH2O was further confirmed by TEM, FT-IR, XRD, TGA and cyclic voltammetry.In turn, hybrid composition is found to strongly influence the supercapacitor behavior. The hybrid rich inRu(bpy) is found to perform better for supercapacitors in physiological electrolytes. 125 F g�1 and68 F g�1 are the capacitance values obtained with [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3P-Mo18O62$nH2O, respectively. In terms of specific energy densities, 3.5 Wh kg�1 and 2 Wh kg�1 wereobtained for both hybrid simultaneously. The difference in supercapacitor performance between bothhybrids is also noticed in impedance spectroscopy which showed that [Ru(bpy)3]3.33PMo18O62$mH2O haslower electron transfer resistance if compared to [Ru(bpy)3]3PMo18O62$nH2O. Finally, if compared ofparent K6P2Mo18O62$12H2O, the stability of both hybrids is found to be highly improved.

© 2015 Elsevier B.V. All rights reserved.

m@rug.nl (M. Ammam).

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535 525

1. Introduction

Supercapacitors are energy storage devices, with the ability todeliver huge amount of energy in a short period of time [1].Unlike batteries, supercapacitors have low energy densities(<10 Wh kg�1) [2,3]. Hence, current research in supercapacitor isfocused on exploring new materials that could increase the en-ergy density. There have been lots of reports dealing with highsurface carbon based supercapacitors [4e9]. Although high spe-cific capacitance was achieved for some materials, energydensity is still not as comparable with batteries. Transitionmetal oxides are another promising material called pseudoca-pacitors as they can store energy in the form of redox reactionsin addition to the double layer capacitance [10e15]. These ma-terials can deliver higher energy densities through Faradaicreactions occurring on the electrode surface. Among thenumerous investigated metal oxides, RuO2 [16] and MnO2 [17]showed to be the most relevant for supercapacitors. Though,metal oxide based pseudocapacitors could deliver high energydensities than carbon based double layer capacitors, they stillhave problems with cycle life and stability of the materials. Also,these materials require strong acidic or alkaline electrolytes andmarginal activities are expected in neutral and physiologicalelectrolytes.

Nowadays, there is huge interest to create micro sized energystorage devices to power implantable biomedical devices likesensors transmitters or micro-artificial organs. For that, it isimportant to investigate the performance of supercapacitors inphysiological electrolytes. Transition metal oxides are not asactive in physiological electrolytes as they are in acidic or alkalinesolutions. Hence, it is important to explore new materials thatmay perform better in physiological electrolytes. Poly-oxometalates (POMs) are molecular oxides formed betweentransition metal centers and oxygen [18]. Most POMs are highlystable in acidic media and undergo fast, quasi-reversible, multiredox reactions [19,20]. In acidic media, POMs have been shownto be relevant for supercapacitors either as electrolytes or evenwhen they are deposited on electrodes [21e26]. Available reportson POMs for supercapacitor studies involved specially composites

Fig. 1. Schematic representation of the two formed hybrids [Ru(bpy)3]3PMo18O62$nH2O (a) anRu(bpy) and P2Mo18 via hydrogen bonding is represented in dotted line.

with conducting polymers, and carbon based materials like car-bon nanotubes and graphene tested and studied in acidic solu-tions. For example, by anchoring POMs onto electrochemicallygrown polyaniline, Romero et al. reported a stable capacitance of120 F g�1 [23,24]. The same group utilized two different con-ducting polymers polyaniline and PEDOT to yield a stablecapacitance of 168 F g�1 and 80 F g�1, respectively [25]. Similarly,Michael et al. anchored POMs on electrochemically depositedpolypyrrole and PEDOT, they assembled a cell in asymmetricconfiguration that resulted in 30 F g�1 at 10 mA and energydensity of 3.9 Wh kg�1 at 1 kW kg�1 [26]. Recently Cui et al.functionalize the graphene with polyaniline and Keggin basedPOM (PMo12) to obtain higher capacitance of 587 F g�1 at0.1 mA g�1 [27]. In other studies, POMs are immobilized ontographene surface through ionic liquid to yield a stable capaci-tance of 408 F g�1 at a current density of 0.5 A g�1 [28]. However,it is important to remind that all these studies are carried outusing strong acidic electrolytes, where higher capacitance valuesare expected. On the other hand, it is well known that when pHincreases, most POMs tend to decompose and their structuralarrangements collapse and hence lose their properties [20]. Thelatter can be improved by combining POMs with organic moietieslike ionic liquids to form hybrids [29e31]. These hybrids areusually more stable at neutral pH but not yet investigated inphysiological electrolytes and serum. On the other hand, ahandful of recent reports appeared on various materials inphysiological electrolytes or simply at neutral pH of whichdepending on the material perform more or less for super-capacitor applications [32e36].

In this paper, we have investigated the possibility for super-capacitor electrodes in the physiological electrolyte using hybridsformed between the Dawson type K6[P2Mo18O62]$12H2O andRu(bpy)3$6H2O. A Ru based organic moiety is selected because Ru isusually relevant for supercapacitor application and may enhancethe activity of the POM. The hybrids were synthesized and char-acterized by various analytical techniques including CHN analysis,TEM, FT-IR, XRD, TGA and cyclic voltammetry. The performance ofthe deposited hybrids on electrodes are evaluated for super-capacitor use in physiological electrolytes.

d [Ru(bpy)3]3.33PMo18O62$mH2O (b) estimated from CHN analysis. Interaction between

Fig. 2. TEM images of the molecular hybrid [Ru(bpy)3]3.33PMo18O62$mH2O (a) and[Ru(bpy)3]3PMo18O62$nH2O (b). Presence of pores is marked by black circles.

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535526

2. Experimental

2.1. Chemicals

Ultrapure water milliQ grade was used for all experiments.WelleDawson K6[P2Mo18O62]$12H2O (abbreviated as P2Mo18) issynthesized and characterized according to the procedure reportedby Briand et al. [37], by replacing ammonium by potassium salt.Tris(bipyridine)ruthenium(II) chloride (Ru(bpy)3$Cl2.6H2O) purissgrade was purchased from Aldrich. Potassium iodide (KI), Na2SO4,NaCl, MgCl2, CaCl2, KH2PO4, K2HPO4 analytical grade and sodiumsalt of Alginic acid and Chitosan were obtained from Sigma orAldrich. Glassy carbon (GC) electrode 3 mm diameter was from CHinstruments and stainless steel foil from Aldrich.

2.2. Synthesis of the hybrid materials and preparation of theiraqueous suspensions

The synthesis of Ruthenium bipyridineephosphomolybdate iscarried out by mixing Ru[(bpy)3]Cl2$6H2O with K6P2Mo18O62$12H2O. Under constant stirring, 0.260 g of Ru(bpy)3$Cl2 wasdissolved in ultrapure water and 0.245 g of K6P2Mo18O62$12H2Owas added to the solution. An orange colored precipitate is formedimmediately upon addition of K6P2Mo18O62$12H2O to the Ru[(bpy)3]Cl2$6H2O solution. The precipitate was filtered off andwashed several times with ultrapure water and dried in oven at60 �C. The second sample was prepared by similar procedure but byadding excess of the mild reducing agent potassium iodide (KI).Precisely,1.5 g of KI was added to the solution containing Ru[(bpy)3]Cl2$6H2O and K6P2Mo18O62$12H2O, and the original orange colorturned to brown. The precipitate was filtered off and washedseveral times with ultrapure water and dried in oven at 60 �C.

2.3. Equipment and methodology

The hybrid materials were analyzed by CHN elemental analysis(Intertek Chemicals and Pharmaceuticals, USA). The structures ofthe hybrids were estimated from the average carbon, nitrogen andhydrogen content. Transmission Electron Microscopy (TEM) wascarried out using CM12 from Philips working at 120 Kev. Imageswere recorded on slow scan CCD camera, samples were made oncarbon coated 400 mesh grids. Spectrum 2000 FT-IR from PerkinElmerwas used for the FT-IR studies. The samples were grinded andthen scanned from 4000 to 500 cm�1. Powder X ray diffraction(XRD) was performed with Bruker e D8-Advance with LYNXEYEdetector. Thermogravimetry analysis (TGA) was carried out withPerkin Elmer TGA 7 Thermogravimetric analyzer. The samples werehold for 0.5 min at 20 �C then heat started from 20 �C to 600 �C inan inert atmosphere at a heating rate of 10 �C/min. Mass mea-surements of the deposited hybrids were carried out with a mi-crobalance with a precision of 1 mg (KERN, ABT 100-5M).

The electrochemical measurements were performed with CH-Instruments (CH600 and CH760) connected to a computer fordata acquisition. A three-compartment electrochemical cell wasused. The side arms contained a reference electrode (Ag/AgCl, 1 MKCl) and a platinum counter electrode. The materials were alsotested in a two electrode system to evaluate the capacitive behaviorof the material in conventional real system. However, it is impor-tant to note that the final goal of these electrodes is the in vivoapplication for which conventional two electrode system used inindustry may not be appropriate in its current design. The elec-trochemical studies were mainly performed in phosphate buffersaline (PBS) solution containing Mg2þ, Ca2þ, Naþ, Kþ to mimic thephysiological conditions. All the experiments were performed inthis media, expect when it is mentioned otherwise.

The deposition of the hybrid materials was carried out followingthe route: under ultrasonic and stirring conditions, 10 mg of eachhybrid is suspended in 500 mL of distilled water with some chargingagent (Sodium alginate or Chitosan). The electrode was thenimmersed in the suspension and subjected to a voltage in order todrive the suspended hybrid particles towards the electrode anddeposit.

3. Results and discussion

3.1. Characterization of the hybrid materials

3.1.1. CHN elemental analysisIn this work, we combined the Dawson anion [PMo18O62]6�with

[Ru(bpy)3]2þ cation to form a molecular hybrid. CHN analysis of thehybrid preparedwithout KI showed that onemole of [P2Mo18O62]6�

react with 3 mol of [Ru(bpy)]2þ to form a hybrid molecular material

2000 1750 1500 1250 1000 750 500

4000 3500 3000 2500 2000 1500 1000 500

[Ru(bpy) ] PMo O .nH O

[Ru(bpy) ] PMo O .mH O

wave number cm-1

P Mo

Inte

nsity

/ a.

u.

(e)

(d)

(c)

(b)

Ru(bpy)

Inte

nsity

/ a.

u.

(a)

[Ru(bpy) ] PMo O .mH O

[Ru(bpy) ] PMo O .nH O

wave number cm-1

Fig. 3. IR spectra of the parent reagents P2Mo18, Ru(bpy), the hybrids [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O as well as a zoom the FT-IR spectra of bothhybrids in the region 2000e500 cm�1.

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535 527

following the reaction:

3[Ru(bpy)3]Cl2$6H2O þ K6[P2Mo18O62]$12H2O / [Ru(bpy)3]3PMo18O62$nH2O þ 6KCl (1)

In presence of KI, it is shown that excess of 1/3 mol Ru[(bpy)3]2þ

is somehow linked to [P2Mo18O62]6� to form [Ru(bpy)3]3.33P-Mo18O62$mH2O. This can be interpreted by presence of 1 extramolecule of [Ru(bpy)3]2þ shared between 3 blocks of [Ru(b-py)3]3PMo18O62$nH2O as schematically depicted in Fig. 1. In thisrespect, the linkage between the POM and Ru(bpy) may probablybe achieved through hydrogen bonding. n and m represent thenumber of watermolecules left in the structural arrangement of thePOM in both hybrids, which according to results from other char-acterization techniques might be different (n s m).

3.1.2. TEMFig. 2 shows TEM images of the hybrid materials synthesized

with and without KI. It can clearly be seen that the morphology ofthe hybrid material synthesized without KI depict smalleraggregated particles with average sizes of 20e30 nm (Fig. 2b). Bycomparison, the hybrid synthesized with KI illustrates bigger

particles with average sizes that may be over 100 nm (Fig. 2a).The agglomeration is also less and the morphology looks moreporous (circles). It is worth noting that the results are repro-ducible for all areas analyzed by TEM for both samples. In otherwords, the difference is certainly not related to artifacts fromTEM. The difference between both materials would probably re-sults from presence of the 1 extra molecule of Ru[(bpy)3]2þ

shared between 3 blocks of [Ru(bpy)3]3PMo18O62$nH2O, whichspreads or extends the structural network in the hybrid materialthat in turn makes it look like a polymer. By contrast, in thesample synthesized without KI, there is probably no linkage be-tween [Ru(bpy)3]3PMo18O62$nH2O blocks, hence resulting insmaller packed agglomerated particles rather than extendedpolymerized sheets.

3.1.3. FT-IRFT-IR spectra are practical to figure out structural and bonding

changes in the POM unit in the hybrid materials. Ru[(bpy)3]Cl2.6H2O (Fig. 3b) shows characteristic peaks at 3685e3337 cm�1

(H2O), 2975e2865 cm�1 (CeH str.) 1593 cm�1 (CeC, C]CeH str.),1455e1415 cm�1 (C]CeH str.), 1312e1156 cm�1 (NeCeH, C]CeHstr.), 1061e1009 cm�1 (C]CeH, CeC, NeC str.), 897 cm�1 (CeH str.)

Fig. 4. A comparison of XRD spectra of P2Mo18, Ru(bpy), and hybrids [Ru(b-py)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O.

Fig. 5. (a) TGA curves of the parent reagents P2Mo18 and Ru(bpy). (b) TGA curves ofboth resulting hybrids [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O.

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535528

and 772e728 cm�1 (CeH, NeH str.). The ligand 2,2-bpy can fartherbe identified by the spectral features located in the region of1630e1423 cm�1 and 758e728 cm�1 [38]. As shown in Fig. 3(a), theparent P2Mo18 Dawson unit shows characteristic bands at3567e3287 cm�1 (H2O), 1609 cm�1 (OeH bending),1058e1035 cm�1 (PeO str.), 946 cm�1 (MoeO terminal str.),880 cm�1 (edge sharing MoeOeMo str.) and 737 cm�1 (cornersharing MoeOeMo str.) [30]. As shown in Fig. 3c and d, the P2Mo18ion cluster in the hybridmaterials preparedwith andwithout KI areclearly marked in the stretching region of 1058e746 cm�1, indi-cating that the primary Dawson structure is retained in the hybrids.Careful analysis of Fig. 3e in the 2,2-bpy stretching regions of1630e1423 cm�1 and 758e728 cm�1 reveals that the relative peakintensities are stronger for the hybrid synthesized with KI ([Ru(b-py)3]3.33PMo18O62$mH2O) if compared to hybrid synthesizedwithout KI [Ru(bpy)3]3PMo18O62$nH2O). These results are repro-ducible and are not due to error in measurements. Since nature ofbonding in both hybrids are actually the same but the number isslightly higher in [Ru(bpy)3]3.33PMo18O62$mH2O if compared to[Ru(bpy)3]3PMo18O62$nH2O, the slight increase in the relative in-tensity observed in 2,2-bpy stretching regions of 1630e1423 cm�1

and 758e728 cm�1 actually confirms that [Ru(bpy)3]3.33P-Mo18O62$mH2O contains more Ru-bpy rings. In turn, this corrobo-rates results from CHN analysis and TEM. Finally, FT-IR data confirmthat the basic structure and geometry of the Dawson anion in bothhybrids [Ru(bpy)3]3PMo18O62$nH2O and [Ru(bpy)3]3.33P-Mo18O62$mH2O are preserved but the water content is nearlyabsent.

3.1.4. XRDPowder X-ray diffraction was utilized to analyze the structures

of the parent compounds and the hybrids. Fig. 4 shows a compar-ison between the XRD pattern of the parents hydratedK6[P2Mo18O62]$12H2O, Ru[(bpy)3]Cl2.6H2O and hybrid materials[Ru(bpy)3]3PMo18O62$nH2O and [Ru(bpy)3]3.33PMo18O62$mH2Osynthesizedwithout andwith KI. XRD of K6[P2Mo18O62]$12H2O andRu[(bpy)3]Cl2$6H2O exhibit sharp high intensity peaks, referring toa high crystalline nature of the materials [20]. By contrast, thehybrid materials [Ru(bpy)3]3PMo18O62$nH2O and

[Ru(bpy)3]3.33PMo18O62$mH2O show broader peaks revealing semi-amorphous structures. This is due to interactions between Ru[(bpy)3]2þ cations and [P2Mo18O62]6� anion, which form agglom-erated particles in [Ru(bpy)3]3PMo18O62$nH2O and polymerizedsheets in [Ru(bpy)3]3.33PMo18O62$mH2O, instead of single crystalsin K6[P2Mo18O62]$12H2O or Ru[(bpy)3]Cl2.6H2O. Careful analysis ofFig. 4 reveals presence of few sharper peaks in the spectrum of[Ru(bpy)3]3.33PMo18O62$mH2O if compared to [Ru(bpy)3]3P-Mo18O62$nH2O. Comparison with XRD spectra of the parent re-agents affirms that the peaks fit with those of Ru[(bpy)3]Cl2$6H2O.The latter support the results of CHN analysis, TEM and FT-IR ofwhich all showed presence of higher amounts of Ru(bpy) in thehybrid synthesized with KI ([Ru(bpy)3]3.33PMo18O62$mH2O).

3.1.5. TG analysisThe thermogravimetric behavior of the hybrids synthesized

without and with KI present and the parent reagentsK6[P2Mo18O62]$12H2O and [Ru(bpy)3]Cl2$6H2O are illustrated inFig. 5. [Ru(bpy)3]Cl2$6H2O shows an initial decline of mass up to100 �C due to loss of water molecules. Then major weight loss from300 to 600 �C in two distinct steps attributed to burning of the

Fig. 6. Cyclic voltammetry curve of dissolved P2Mo18 in different proton concentrations of pH 1 (a), pH 3 (b) and pH 7 (c) (Mo based redox waves are marked as i, ii, and iii). (d)Represents the cyclic voltammetry curve of Ru(bpy) at pH 7. (e) Stability of P2Mo18 recorded soon after addition and after 2 min. (f) Comparison of cyclic voltammetry curve ofP2Mo18 and Ru(bpy) at pH 7. Scan rate 100 mV s�1. Electrolytes: pH 1 (1 M H2SO4), pH 3 (0.25 M Na2SO4 þ H2SO4), pH 7 (0.25 M total salt containing 0.05 M KH2PO4, 0.05 M K2HPO4,0.1 M NaCl, 0.025 M MgCl2 and 0.025 M CaCl2).

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535 529

organic part. Thermal analysis of K6[P2Mo18O62]$12H2O shows aninitial weight loss from 25 to 100 �C and amajor loss from 500 �C to600 �C (Fig. 5a) [30]. The first change in mass would be related towater loss from K6[P2Mo18O62]$12H2O, which includes the 12constitution water molecules present in the crystal structure. Thesecond loss may be attributed to decomposition of the Dawsonunit. TGA curve of the hybrid synthesized without and with KIdepicts in both materials an initial decline of mass till 100 �C,related to loss of water molecules (Fig. 5b). However, interestingly,the water loss from [Ru(bpy)3]3.33PMo18O62$mH2O looks signifi-cantly low if compared to ([Ru(bpy)3]3PMo18O62$nH2O or theparent K6[P2Mo18O62]$12H2O. This means that the hybrid synthe-sized with KI contains far less constitution water molecules ifcompared to ([Ru(bpy)3]3PMo18O62$nH2O or the parent

K6[P2Mo18O62]$12H2O (m < n). The latter could result from for-mation of I2 gas (2I� þ 2e� ¼ I2), which after interaction with theconstitutionwater molecules may lead to their dislocation from thecrystal structure. This in turn will free the oxygen centers in[P2Mo18O62]6� and facilitate formation of hydrogen bonding be-tween O of [P2Mo18O62]6� and H of [Ru(bpy)3]2þ, hence allowingthe insertion of one extra Ru(bpy)3 shared between 3 blocks of[Ru(bpy)3]3PMo18O62 to form [Ru(bpy)3]3.33PMo18O62. In the regionof 100e300 �C, [Ru(bpy)3]3.33PMo18O62$mH2O looks also thermallymore stable than [Ru(bpy)3]3PMo18O62$nH2O and K6[P2Mo18O62]$12H2O. This means that the Dawson unit [P2Mo18O62]6� is morestable when surrounded by 4 Ru(bpy) as in [Ru(bpy)3]3.33P-Mo18O62$mH2O rather than by 3 like in [Ru(bpy)3]3PMo18O62$nH2Oor 0 in K6[P2Mo18O62].12H2O. Finally, above 300 �C, weight loss in

Fig. 7. (a) Comparison between the cyclic voltammetry curves of parent materials P2Mo18, and Ru(bpy) and the resulting hybrids [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(b-py)3]3PMo18O62$nH2O. (b), (c) and (d) represent the cyclic voltammetry curves of the hybrids [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O at pH 1, pH 3 and pH 7,respectively. Mo based redox waves are marked as (i, ii, and iii) and Ru based wave is marked as (iv).

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535530

both hybrids becomes significant as the organic part start to burnout. But, even at higher temperatures [Ru(bpy)3]3.33PMo18O62$

mH2O is still relatively more stable than ([Ru(bpy)3]3PMo18O62nH2O).

3.1.6. Cyclic voltammetryFig. 6 illustrates the cyclic voltammetry curve of dissolved

K6[P2Mo18O62]$12H2O at pH 1 (Fig. 6a), pH 3 (Fig. 6b) and pH 7(Fig. 6c) and Ru[(bpy)3]Cl2 at pH 7 (Fig. 6d). In all media,K6[P2Mo18O62]$12H2O undergo three consecutive two-electronredox reactions [30]. At pH 1, three redox peaks appearatþ470mV (i),þ344 mV (ii), andþ176mV (iii), respectively. WhenpH increases, these peaks undergo a shift towards negative po-tentials, indicating that the redox behavior is pH dependent.However, if compared to pH 3, the shift in potential of the first

Table 1Reduction potential of the Mo and Ru based redox waves registered at different pH.

Material pH Peak (i) Peak (ii) Peak (iii) Peak (iv)

K6[P2Mo18O62]$12H2O pH 1 470 344 176 e

[Ru(bpy)3]3.33PMo18O62$mH2O 381 308 128 470[Ru(bpy)3]3PMo18O62$nH2O 397 266 108 517K6[P2Mo18O62]$12H2O pH 3 310 142 �39 e

[Ru(bpy)3]3.33PMo18O62$mH2O 226 111 �83 460[Ru(bpy)3]3PMo18O62$nH2O 234 112 �80 495K6[P2Mo18O62]$12H2O pH 7 287 �83 �236 e

[Ru(bpy)3]3.33PMo18O62$mH2O 46 �116 �350 287[Ru(bpy)3]3PMo18O62$nH2O 46 �118 �328 303

redox wave (i) obtained in the physiological electrolyte pH 7 looksnot very significant if compared to displacement of the two otherswaves (ii and iii), which show little influence of pH on the firstredox wave (i). Another notable difference is that whileK6[P2Mo18O62]$12H2O shows a decent stability at low pH, in thephysiological electrolyte, the structure tends to collapse in a shortperiod of time. This is evidenced by Fig. 6e, where it can beobserved that after few minutes, the cyclic voltammetry curve ofK6[P2Mo18O62]$12H2O appears flat, indicating a complete disinte-gration of the compound. On the other hand, the cyclic voltam-metry curve of [Ru(bpy)3]Cl2$6H2O in the physiological electrolytepH 7 shows 1 reversible wave located at 380 mV (Fig.6d), corre-sponding to oxidation of Ru2þ to Ru3þ. This wave is located in thesame region as the first Mo based redox wave of K6[P2Mo18O62]$12H2O, but the intensity of Ru2þ/Ru3 is low (Fig. 6f). This is un-derstandable since Mo based first wave of K6[P2Mo18O62]$12H2Oinvolves two-electrons while Ru based wave of [Ru(bpy)3]Cl2$6H2Ois a one-electron oxidation.

Fig. 7a represents a comparison between the cyclic voltammetrycurves of both hybrids formed without and with KI present([Ru(bpy)3]3PMo18O62$nH2O and [Ru(bpy)3]3.33PMo18O62$mH2O)as well as the parent reagents K6[P2Mo18O62]$12H2O and[Ru(bpy)3]Cl2$6H2O in the physiological electrolyte pH 7. It will benoted that all redox waves observed for both parentsK6[P2Mo18O62]$12H2O are present in the cyclic voltammetry curvesof the hybrids, but the potentials are overall shifted towardsnegative values. This may stem from the fact the Mo and Ru based

Fig. 8. (a) Photographs of a [Ru(bpy)3]3.33PMo18O62$mH2O suspension inwater without and with presence of charging agent, (b) Picture of a chronoamperometry curve obtained fordeposition of the molecular hybrid on electrodes. Images of glassy carbon (c) and stainless steel (d) after deposition of the molecular hybrid [Ru(bpy)3]3PMo18O62$mH2O.

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535 531

centers in the hybrids experience an alkaline like environment ifcompared to the parent reagents. This would come from thepresence of pyridine rings in the hybrid materials. It can also beobserved that intensity of the Ru based redox wave is quite high in[Ru(bpy)3]3.33PMo18O62$mH2O if compared to [Ru(bpy)3]3P-Mo18O62$nH2O, confirming previous results from CHN analysis, FT-IR, XRD and TGA. The difference between the electrochemicalbehavior of both hybrids is further studied and compared at pH 1, 3and 7 (Fig. 7b, c and d). In all media, one can notice that the Mo andRu based waves are present, but undergo a displacement of po-tential towardsmore negative values as the pH increases. The redoxpotentials of Ru and Mo based waves are gathered in Table 1.Abundance of Ru in [Ru(bpy)3]3.33PMo18O62$mH2O with respect to[Ru(bpy)3]3PMo18O62$nH2O is confirmed in all media of pH 1, 3 and7. Finally, it is worth noting that unlike K6[P2Mo18O62]$12H2Owhich is instable at pH 7, cycling of the hybrids in various pHmediashowed that they are highly stable even in the physiological elec-trolyte pH 7 confirming that [Ru(bpy)3]3.33PMo18O62$mH2O and[Ru(bpy)3]3PMo18O62$nH2O molecules are quite stable if comparedto the parent POM. This confirms that there is a strong interactionbetween [P2Mo18O62]6� unit and Ru[(bpy)3]2þ moieties in bothhybrids network, which stabilizes [P2Mo18O62]6� in the hybrids andprevent them from collapsing.

3.2. Supercapacitor study

3.2.1. Preparation of the supercapacitor electrodesFig. 8 displays a picture of a suspension based hybrid in absence

and presence of the charging agent (Fig. 8a), a typical chro-noamperometry signal that is used to deposit the hybrid from awater based suspension to either glassy carbon or stainless steel foil(Fig. 8b) and images of the resulting deposited hybrid [Ru(b-py)3]3PMo18O62$mH2O (Fig. 8c and d). It can be seen that thedeposited film is thick and homogenous, which is a key factor insuccessful prepared supercapacitor electrode. For that, it is impor-tant to carry out the deposition from a stable suspension (Fig. 8aright), which require presence of some charging agent in the sus-pension in order to keep the hybrid particles well suspended andprevent their sedimentation. Two charging agents have beentested: Alginate and protonated Chitosan. These biopolymers wereselected because of their physiologically friendly aspect. However,it is observed that Chitosan based suspensions is highly stable ifcompared to Alginate. Afterward, the electrode (glassy carbon orstainless steel) is immersed in the hybrid suspension and subjectedto a voltage (Fig. 8b) and after fewmin, a thick film is deposited. Theelectrode was then washed gently with ultrapure water, dried thentested as a supercapacitor electrode in the physiological electrolytepH 7 (containing phosphates, Naþ, Mg2þ and Ca2þ ion).

3.2.2. Evaluation of the hybrids based electrodes forsupercapacitors in physiological electrolytes

Fig. 9a depicts the cyclic voltammetry curve of a deposited[Ru(bpy)3]3PMo18O62$nH2O and [Ru(bpy)3]3.33PMo18O62$mH2Ofrom suspensions containing Chitosan as charging agent. Comparedto the unmodified GC electrode, substantial current densities areobtained testifying that both hybrids are active for supercapacitor.

Fig. 9. (a) Comparison of cyclic voltammetry curves of the electrode before and after modification with [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O. (b) Cyclicvoltammetry curve of [Ru(bpy)3]3.33PMo18O62$mH2O modified GC electrode and Chitosan modified GC. The Insert shows a zoom of the curve of Chitosan (charging agent). (c) Cyclicvoltammetry curves of GC/[Ru(bpy)3]3PMo18O62$nH2O electrode at different scan rates. (d) Chargeedischarge curve of GC/[Ru(bpy)3]3.33PMo18O62$mH2O and GC/[Ru(bpy)3]3P-Mo18O62$nH2O at 0.2 A g�1. Scan rate for (a) and (b) is 100 mV s�1. The film deposited in (b) is thicker than the film deposited in (a). Electrolyte: pH 7 (0.25 M total salt containing0.05 M KH2PO4, 0.05 M K2HPO4, 0.1 M NaCl, 0.025 M MgCl2 and 0.025 M CaCl2).

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535532

Also, it is clear that the hybrid with [Ru(bpy)3]3.33PMo18O62$mH2Operforms better in terms of current density if compared to [Ru(b-py)3]3PMo18O62$nH2O. The latter may results from presence of theextra Ru(bpy) in [Ru(bpy)3]3.33PMo18O62$mH2O. It can also benoticed that the shape of the cyclic voltammograms deviate fromthe rectangular form of typical supercapacitor material, confirmingthe pseudo capacitor nature of the hybrids. Effect of the Faradaicreactions appears in the voltage range of Ru and Mo waves (0.8to �0.2 V vs. Ag/AgCl). However, if compared to Fig. 7, one canclearly see that pseudocapacitance due to the redox reactions is farless. This results from nature of the deposited film, which is thickand deposited from particles suspension in Fig. 9a. In order to findout if there is any contribution to supercapacitor from the chargingagent, Chitosan alonewas deposited on GCwithout the hybrids andthe results are shown in Fig. 9b. It can be observed that the currentvalues of Chitosan is negligible if compared to [Ru(bpy)]3.33P2-Mo18O62$mH2O, indicating that all the activity is due to the hybrids.In terms of performance with scan rate, Fig. 9c depicts that there isan increase of the current of the voltammograms with the increasein scan rate. At elevated scan rates, the cyclic voltammetry curvesretain their rectangular shape, demonstrating a rapid currentresponse on voltage reversal at the vertex potentials.

The specific capacitance of [Ru(bpy)3]3PMo18O62$nH2O and[Ru(bpy)3]3.33PMo18O62$mH2O based electrodes is estimated fromthe galvanostatic chargeedischarge cycling (Fig. 9d). The currentdensity used for the charge and discharge cycling was 0.2 A g�1.There is a quasi-linear variation of the potential during the chargeand discharge process. Thus, the capacitance can be estimated us-ing the following equation [29].

C ¼ It=Em (2)

where C is the specific capacitance (F g�1), I is current, t is thedischarge time, E is the potential range during discharge and m isthe mass of the deposited hybrid material. At a current density of0.2 A g�1, [Ru(bpy)3]3.33PMo18O62$mH2O delivered 125 F g�1, while[Ru(bpy)3]3PMo18O62$nH2O gave only 64 F g�1. This suggests thatthe hybrid synthesized with KI present ([Ru(bpy)3]3.33P-Mo18O62$mH2O) generates twice capacitance if compared to thehybrid synthesized without KI ([Ru(bpy)3]3PMo18O62$nH2O). Thereason for this may have to dowith the difference inmicrostructureof both materials. As can be seen in TEM images, [Ru(bpy)3]3.33P-Mo18O62$mH2O showed a porous structure that could intercalatemore cations into the structure if compared to

Fig. 10. (a) Represents chargeedischarge curves of GC/[Ru(bpy)3]3.33PMo18O62$mH2Oat different current densities, (b) Dependence of specific capacitance on current den-sity of chargeedischarge cycling. (c) Cyclic life of GC/[Ru(bpy)3]3.33PMo18O62$mH2Oelectrode at 0.3 A g�1. Electrolyte: pH 7 (0.25 M total salt containing 0.05 M KH2PO4,0.05 M K2HPO4, 0.1 M NaCl, 0.025 M MgCl2 and 0.025 M CaCl2).

Fig. 11. (a) Galvanostatic chargeedischarge curve at 0.2 A g�1, (b) Ragon plot and (c)Nyquist plot of [Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O in twoelectrode configuration. Electrolyte: pH 7 (0.25 M total salt containing 0.05 M KH2PO4,0.05 M K2HPO4, 0.1 M NaCl, 0.025 M MgCl2 and 0.025 M CaCl2). Impedance spec-troscopy is carried out in the frequency range of 1 kHz to 0.01 Hz with Ac amplitude of5 mV.

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535 533

[Ru(bpy)3]3PMo18O62$nH2O that has agglomerated particles, henceintercalation/deintercalation of ions would be difficult to occur.Finally, it is worth noting that the obtained capacitance valuescompare well with some capacitances obtained with either POM orRu based materials (oxides) studied in acidic or alkaline medium,

despite of the complex nature of the physiological electrolyteemployed in our study [11,15,21,22,24]. On the other hand, ifcompared to the handful studies carried out recently in physio-logical electrolytes, the present hybrids compare well or better[35,36]. Because [Ru(bpy)3]3.33PMo18O62$mH2O performs better

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535534

than [Ru(bpy)3]3PMo18O62$nH2O, the rest of the study was carriedout with [Ru(bpy)3]3.33PMo18O62$mH2O.

Further chargeedischarge cycling at different current densitiesis carried out with the [Ru(bpy)3]3.33PMo18O62$mH2O based elec-trodes and the results are gathered in Fig. 10a. As expected, thecapacitance value decreases as the current density increases. Atcurrent densities of 0.27, 0.33, 0.4, 0.5 and 0.7 A g�1, [Ru(b-py)3]3.33PMo18O62$mH2O based electrode is able to deliver specificcapacitances of 58, 51, 45, 40 and 37 F g�1, demonstrating the abilityof this hybrid to deliver substantial capacitance values in thephysiological electrolyte and at higher applied current densities.The variation of specific capacitance versus current density isshown in Fig. 10b. At low current densities, more number of ionscould intercalate into the hybrid material and more Ru and Mocenters undergo redox reactions. When the current density in-creases, less number of ions are available to Ru and Mo centers toundergo redox reactions, in turn there is a decrease in the capaci-tance value. With respect to stability, Fig. 10c depicts that thefaradic efficiency of [Ru(bpy)3]3.33PMo18O62$mH2O electrode de-creases by about 11% over 400 cycles. This compares well or betterwith supercapacitor electrodes studied in physiological electrolytes[33e35]. Furthermore, because the buffer solution acquires anorangeebrown turbid coloration with time, this indicates that theloss is mainly due to delamination of the deposited [Ru(b-py)3]3.33PMo18O62$mH2O film from the electrode surface. The lattercould be improved by employing stabilizers like polymers appliedas outer layers to prevent the deposited film fromdelamination andhence increasing the stability of the electrode for longer periods.The latter will be the subject of future investigations.

In order to evaluate the supercapacitor performance of the hy-brids in real system, two electrode cells were assembled in sym-metric configuration. Fig.11a shows chargeedischarge curve of thecell assembled with [Ru(bpy)3]3.33PMo18O62.mH2O and [Ru(b-py)3]3PMo18O62$nH2O. At 0.2 A g�1, the cell assembled with[Ru(bpy)3]3.33PMo18O62$mH2O delivered 80 F g�1, while [Ru(b-py)3]3PMo18O62$nH2O delivered only 44 F g�1. Energy densities andpower densities values are calculated and Ragone plot are shown inFig. 11b. At 0.16 kW kg�1, [Ru(bpy)3]3.33PMo18O62$mH2O deliveredan energy density of 3.5 Wh g�1 versus 2 Wh kg�1 for the cellassembled with [Ru(bpy)3]3PMo18O62$nH2O. Electrochemicalimpedance spectroscopy study is carried out for both cells at fre-quencies ranging from 1 kHz to 10 mHz and Nyquist plots aredepicted in Fig. 11c. Zoom of Fig. 11c, illustrates that Nyquist plotsare composed of two parts: a semicircle at high frequencies andlinear curve at low frequencies. The semicircle at higher fre-quencies consists of electrolyte solution resistance and chargetransfer resistance. The electrolyte solution resistance (Rs) is thesame for both cells, but notable differences occurred in the chargetransfer resistance (Rct) value. The intercept of the semicircle withreal axis gives the value of the charge transfer resistance. The radiusof the semicircle of the cell assembled with [Ru(bpy)3]3.33P-Mo18O62$mH2O is about 1.5 k U, which is smaller than that of thecell assembled with [Ru(bpy)3]3PMo18O62$nH2O 4.5 k U. The dif-ference in the charge transfer resistance confirms the morpholog-ical advantage of [Ru(bpy)3]3.33PMo18O62$mH2O if compared to[Ru(bpy)3]3PMo18O62$nH2O. With reduced charge transfer resis-tance, the percolation of electrolyte into the material is facilitated.With respect to cycle life of the materials in the two electrodesystem, 13% and 43% are decreases in capacitance obtained with[Ru(bpy)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O,respectively over 500 cycles. This illustrates that the life cycle ofRu(bpy)3]3.33PMo18O62$mH2O is more stable than [Ru(bpy)3]3P-Mo18O62$nH2O. However, these stability data should be takencarefully since no actual stabilizer is utilized and the decrease maysimply result from delamination of film from the electrode surfaces.

4. Conclusions

In this study, we paired K6P2Mo18O62$12H2O (P2Mo18) withRu(bpy)3Cl2$6H2O (Ru(bpy)) to form a molecular hybrid. The syn-thesis was carried out without and with presence of the mildreducing agent KI. The hybrids were characterized by CHN analysis,TEM, FT-IR, XRD, TGA and cyclic voltammetry. In absence of KI, CHNanalysis revealed that one mole of [P2Mo18O62]6� is paired with3 mol [Ru(bpy)3]2þ to form [Ru(bpy)3]3PMo18O62$nH2O. With KIpresent, the Dawson unit is paired to 3.33 mol to yield [Ru(b-py)3]3.33PMo18O62$mH2O. The latter can be interpreted by presenceof one extra molecule of Ru[(bpy)3]2þ shared between 3 blocks of[Ru(bpy)3]3PMo18O62$nH2O. Excess of Ru(bpy) in [Ru(bpy)3]3.33P-Mo18O62$mH2Owas further confirmed by TEM, FT-IR, XRD, TGA andcyclic voltammetry. In turn, composition of the hybrid was found tostrongly influence the supercapacitor behavior. The hybrid rich inRu(bpy) ([Ru(bpy)3]3.33PMo18O62$mH2O) was found to performbetter for supercapacitors in physiological electrolytes. 125 F g�1

and 68 F g�1 are the capacitance values delivered by [Ru(b-py)3]3.33PMo18O62$mH2O and [Ru(bpy)3]3PMo18O62$nH2O, respec-tively at a current density of 0.2 A g�1. In terms of specific energydensities, 3.5 Wh kg�1 and 2 Wh kg�1 were obtained for bothhybrid simultaneously. The advantage of [Ru(bpy)3]3.33P-Mo18O62$mH2O over [Ru(bpy)3]3PMo18O62$nH2O for super-capacitor is also confirmed in difference between the electrontransfer resistances measured with impedance spectroscopy.Approximately 1.5 and 4.5 k U are the values of charge transferresistances obtained with [Ru(bpy)3]3.33PMo18O62$mH2O and[Ru(bpy)3]3PMo18O62$nH2O, respectively. Furthermore, the stabil-ity of both hybrids is found to be highly improved if compared tothe parent K6P2Mo18O62$12H2O. Future work will focus on opti-mization of the supercapacitor electrodes, study of syntheticmechanisms and eventually carry out in vitro and in vivo testing ofthe supercapacitor electrodes.

Acknowledgment

The authors would like to thank the University of Groningen forsupport through the Roseland Franklin Fellowship. We also thankIntertek for CHN analysis, Dr. Marc. D. A. Stuart for TEM, G.O.R.Alberda van Ekenstein for TGA, Jacob Baas for training on XRD and J.H. Marsman for training on FT-IR.

References

[1] B.E. Conway, J. Electrochem. Soc. 138 (1991) 1539e1548.[2] A.G. Pandolfo, A.F. Hollenkamp, J. Power Sources 157 (2006) 11e27.[3] Y. Zhang, H. Feng, X. Wu, L. Wang, A. Zhang, T. Xia, H. Dong, X. Li, L. Zhang, Int.

J. Hydrog. Energy 34 (2009) 4889e4899.[4] L. Zhang, X.S. Zhao, Chem. Soc. Rev. 38 (2009) 2520e2531.[5] R. Kotz, M. Carlen, Electrochim. Acta 45 (2000) 2483e2498.[6] C. P�ean, C. Merlet, B. Rotenberg, P.A. Madden, P.L. Taberna, B. Daffos,

M. Salanne, P. Simon, ACS Nano 8 (2014) 1576e1583.[7] J. Li, X. Cheng, A. Shashurin, M. Keidar, Graphene 1 (2012) 1e13.[8] E.R. Pi~nero, F. Leroux, F. B�eguin, Adv. Mater. 18 (2006) 1877e1882.[9] E. Frackowiak, F. B�eguin, Carbon 39 (2001) 937e950.

[10] S. Faraji, F.N. Ani, J. Power Sources 263 (2014) 338e360.[11] J. Jiang, A. Kucernak, Electrochim. Acta 47 (2002) 2381e2386.[12] M.W. Mehrens, J. Schenk, P.M. Wilde, E. Abdelmula, P. Axmann, J. Garche,

J. Power Sources 105 (2002) 182e188.[13] C.C. Hu, T.W. Tsou, J. Power Sources 115 (2003) 179e186.[14] G. Wang, L. Zhang, J. Zhang, Chem. Soc. Rev. 41 (2012) 797e828.[15] N.L. Wu, S.Y. Wang, C.Y. Han, D.S. Wu, L.R. Shiue, J. Power Sources 113 (2003)

173e178.[16] J.P. Zheng, T.R. Jow, J. Electrochem. Soc. 142 (1995) L6eL8.[17] S. Devaraj, N. Munichandraiah, J. Phys. Chem. C 112 (2008) 4406e4417.[18] C.W. Baker, D.C. Glick, Chem. Rev. 98 (1998) 3e49.[19] D.E. Katsoulis, Chem. Rev. 98 (1998) 359e387.[20] M. Ammam, J. Mater. Chem. A 1 (2013) 6291e6312.[21] P.G. Romero, M. Chojak, K.C. Gallegos, J.A. Asensio, P.J. Kulesza, N.C. Pastor,

M. Lira-Cantu, Electrochem. Commun. 5 (2003) 149e153.

S. Chinnathambi, M. Ammam / Journal of Power Sources 284 (2015) 524e535 535

[22] A.M. White, R.C.T. Slade, Synth. Metals 139 (2003) 123e131.[23] A.C.C. Gallegos, R.M. Rosales, M. Baibarac, P.G. Romero, M.E. Rincon, Electro-

chem. Commun. 9 (2007) 2088e2092.[24] A.K.C. Gallegos, M.L. Cantu, N.C. Pastor, P.G. Romero, Adv. Funct. Mater. 15

(2005) 1125e1133.[25] J. Vaillant, M.L. Cantu, K.C. Gallegos, N. Casanpastor, P.G. Romero, Solid State

Chem. 34 (2006) 147e159.[26] G.M. Suppes, C.G. Cameron, M.S. Freund, 157 (2010) A1030eA1034.[27] Z. Cui, C.X. Guo, W. Yuan, C.M. Li, Phys. Chem. Chem. Phys. 14 (2012)

12823e12828.[28] M. Yang, B.G. Choi, S.C. Jung, Y. Han, Y.S. Huh, Sang Bok Lee, Adv. Funct. Mater.

24 (2014) 7301e7309.[29] M. Ammam, J. Fransaer, J. Electrochem. Soc. 158 (2011) A14eA21.[30] M. Ammam, E.B. Easton, Electrochim. Acta 56 (2011) 2847e2855.

[31] M. Ammam, E.B. Easton, Sens. Actuators B 161 (2012) 520e527.[32] M. Ammam, J. Fransaer, Chem. Commun. 48 (2012) 2036e2038.[33] K. Kume, N. Kawasaki, H. Wang, T. Yamada, H. Yoshikawa, K. Awaga, J. Mater.

Chem. A 2 (2014) 3801e3807.[34] S.R. Shin, C.K. Lee, I. So, J.H. Jeon, T.M. Kang, C. Kee, S.I. Kim, G.M. Spinks,

G.G. Wallace, S.J. Kim, Adv. Mater. 20 (2008) 466e470.[35] J. Hur, Kyuhyun, S. Hwang, B.L. Choi, S. Kim, S. Hwang, N. Park, K. Kim, Sci. Rep.

3 (2013) 1282.[36] C.A. Amarnath, N. Venkatesan, M. Doble, S.N. Sawant, J. Mater. Chem. B 2

(2014) 5012e5019.[37] L.E. Briand, G.M. Valle, H.J. Thomas, J. Mater. Chem. 12 (2002) 299e304.[38] J. Song, Z. Luo, H. Zhu, H. Zhu, Z. Huang, T. Lian, A.L. Kaledin, D.G. Musaev,

S. Lense, K.I. Hardcastle, C.L. Hill, Inorg. Chim. Acta 363 (2010) 4381e4386.