Cyclic voltammetry study of the electrocatalysis of carbon dioxide reduction by bis(polyazamacrocyclic) nickel complexes

Electrochimica Acta 45 (2000) 2061 – 2074 www.elsevier.nl/locate/electacta

Cyclic voltammetry study of the electrocatalysis of carbon dioxide reduction by bis(polyazamacrocyclic) nickel complexes Chanaka de Alwis a, Joe A. Crayston a,*, Thomas Cromie a, Tanja Eisenbla¨tter a, Robert W. Hay a, Ya.D. Lampeka b, Lyudmila V. Tsymbal b b

a School of Chemistry, Uni6ersity of St Andrews, St Andrews, Fife KY16 9ST, UK L.V. Pisarzhe6sky Institute of Physical Chemistry, National Academy of Sciences of Ukraine, Prospekt Nauki 31, Kie6 252039, Ukraine

Received 23 July 1999; accepted 17 November 1999

Abstract Two series of binuclear macrocyclic nickel(II) complexes with varying lengths of the chain linking the two macrocyclic rings were characterised by cyclic voltammetry under argon and CO2. The first series consisted of binuclear complexes [Ni2L2 – 6]4 + containing pentaaza macrocycles with (CH2)n bridges (n= 2, 3, 4, 6) or a p-xylyl linkage (L6). In general, the two nickel sites in the binuclear complexes behave independently with the currents corresponding to the simultaneous transfer of two electrons. The redox potentials are remarkably constant along this series, but the peak separations increase, reflecting slower electron transfer due to more effective adsorption on the electrode. Electrochemical data for the electrocatalytic reduction of CO2 in MeCN/10% H2O revealed catalytic waves for CO2 reduction with E cp close to −1.7 V and catalytic currents (i cp) which are about half those of the mononuclear complex, proposed to be due to steric constraints allowing strong interaction of only one nickel centre of the binuclear one on the surface. The catalytic currents increased slightly as the linking chain length increased as the stereochemical constrains were relaxed somewhat. There was also a splitting in the catalytic peaks of the bismacrocyclic complexes which could reflect two types of adsorbed catalyst sites. In the more sterically crowded series of complex, [Ni2L7]4 + along with the series of linked heptaaza macrocyclic complexes [Ni2L9 – 11]4 + much more positive redox potentials were observed due to both alkylation of the coordinated nitrogen atoms, which decreases the ligand field, and the introduction of steric barriers to axial coordination. These steric barriers prevented strong electrode interaction and led to a lower catalytic activity. Indeed, the complex [Ni2L7]4 + did not even show any interaction with CO2 in dry acetonitrile. The complexes showed well separated peaks due to solution and surface catalytic activity, and the surface catalytic currents were now comparable to mononuclear complexes at the same effective concentration. We proposed that the less effective absorption on the electrode arising from ligand steric interactions places far fewer stereochemical constraints on the adsorption of both nickel centres to the same extent as the binuclear complex, and hence the catalytic currents for binuclear complex and mononuclear complex are comparable. © 2000 Elsevier Science Ltd. All rights reserved. Keywords: Binuclear; Macrocycles; Nickel(II); Carbon dioxide; Electrocatalysis

* Corresponding author. Fax: + 44-1334-463808. E-mail address: [email protected] (J.A. Crayston) 0013-4686/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 3 - 4 6 8 6 ( 9 9 ) 0 0 4 2 7 - 2

2062

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

1. Introduction Increased CO2 concentration in the atmosphere is now a global environmental problem. Therefore, the electrocatalytic reduction of carbon dioxide by macrocyclic nickel complexes is a topic of considerable interest and has recently been reviewed [1]. The complex [Ni(cyclam)]2 + ((1) cyclam =1,4,8,11-tetra-azacyclotetradecane, Chart Scheme 1) has proved to be an excellent catalyst for the reduction of CO2 to CO [2,3]. In the search for other efficient catalysts we have found that the nickel(II) complexes of 3,10-dimethyl-

1,3,6,8,10,12-hexa-azacyclotetradecane (2) [4] and 3-(2%hydroxyethyl)-1,3,5,8,12-penta-azacyclotetradecane (3) [5] also show very promising catalytic properties. Binuclear metal complexes of polyazamacrocycles have attracted considerable attention, [6] because in many ways they display specific behaviour differentiating them from monomacrocyclic relatives [7 – 9]. Here we wish to highlight only their role in the stabilisation of mixed-valence species with less common oxidation states, such as NiIII and CuIII [10 – 12]. The redox properties of binuclear complexes are controlled by the nature and length of the link between the metal centres,

Scheme 1. Recently described catalysts for CO2 electroreduction.

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

2063

Normally, several steps are required in the synthesis of the bismacrocyclic ligands [16 – 19]. On the other hand, the template condensation of amines and formaldehyde has been extensively used in the preparation of mononuclear complexes of saturated macrocycles [20 – 26], such as the pentaazamacrocyclic complex, [NiL1]2 + [27] (Chart Scheme 2). Recently, this method has been extended to synthesise bismacrocyclic metal complexes [28 – 30]. In the present paper we have used the approach developed by Rosokha and Lampeka [30] to prepare the binuclear nickel(II) complexes ([Ni2L2 – 5]4 + ) with polymethylene (CH2)n as well as [Ni2L6]4 + with a p-xylylene bridge(Chart Schemes 2 – 4). For comparison purposes we investigated also the properties of the monomacrocyclic complex [NiL1]2 + substituted at

Scheme 2. Complexes studied in this paper.

the link’s conformational flexibility and the nature of the bridgehead atoms. Moreover, bis(macrocyclic) dinickel [13] and dipalladium [14] complexes are of interest in CO2 reduction. The presence of two co-ordination sites that might both be close enough to interact simultaneously with small molecules or the products of their redox transformation can also lead to potentially different reaction pathways than for the mononuclear complex itself. For instance, [Ni2(biscyclam)]4 + (4) [13] (Chart Scheme 1) displayed greater efficiency towards H2O reduction, but lower activity for CO2 reduction in comparison to (1). Replacement of one of the 14-membered rings by a 13-membered ring has been shown to impair the catalytic activity for H2O reduction [15]. These few available results indicate that the catalytic behaviour of binuclear nickel complexes can be quite different from their mononuclear analogues. Another possible attraction to the use of binuclear complexes is that they may bring about the reduction of the CO2 beyond CO, the usual two-electron product. Not only would this be more thermodynamically efficient for fuel cell applications and lead to more useful products, but it would also help to alleviate the ‘poisoning effect’, in which the reduced nickel macrocycle reacts irreversibly with the carbon monoxide produced and adsorbs strongly to the electrode, leading to a decrease in the catalytic current [3].

Scheme 3.

Scheme 4.

2064

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

the non-coordinating nitrogen by an n-propyl group (Chart Scheme 2). Closely related to the above series is the binuclear complex [Ni2L7]4 + the behaviour of which was compared to the monomacrocyclic complex [NiL8]2 + (Chart Scheme 2), although this was not an ideal ‘mononuclear’ analogue in that it shows a greater degree of alkylation of the amine nitrogens and, possibly, greater steric crowding of the metal center. In addition a series of binuclear heptaaza complexes [Ni2L9 – 11]4 + were examined with (CH2)n bridges (n=2, 3, 5) (Chart Scheme 2). The mononuclear complex [NiL12]2 + was used as a reference compound for the [Ni2L9 – 11]4 + bismacrocyclic complexes.

2. Experimental

2.1. Synthesis The monomacrocyclic complexes [NiL1](ClO4)2 [27], [NiL8](ClO4)2 [25] and [NiL12](ClO4)2 [31] were prepared as previously described. The binuclear Ni(II) complexes of ligands L2 – 5 as the perchlorate salts were prepared according to [30] via the reaction of [Ni(2,3,2tet)](ClO4)2 (2,3,2-tet=1,9-diamino-3,7-diazanonane) with the a, v-diamines and formaldehyde. The same synthetic procedure was applied for the preparation of [Ni2L6](ClO4)4 except that p-xylylenediamine instead of aliphatic diamines was used as locking fragment in cyclization. The yield of the products of this type appear to be a function of the chain length of the bridging diamine, and the lowest yield was obtained using 1,2-diaminoethane. Bis(macrocyclic) complex [Ni2L7](ClO4)4 was prepared as described in [29]. For the synthesis of complexes of ligands containing the diazabicyclononane rings L9 – 11 the procedure developed in [31] was used. All complexes gave satisfactory elemental analyses and their spectral characteristics corresponded well to those described in the literature.

2.2. Electrochemistry Cyclic voltammograms were recorded using an EG and G PARC 273A potentiostat/galvanostat controlled by version 4.11 of the Electrochemistry Research software running on a PC. The measurements were carried out using a three-necked glass cell using a saturated calomel electrode (SCE) as the reference with a Luggin capillary and a Pt wire as the counter (auxiliary) electrode. The working electrode was either a hanging mercury drop electrode (Metrohm, area =0.0087 cm2), a mercury-plated silver disk electrode (0.0314 cm2) or a glassy carbon electrode (BAS MF-2012, area =0.0707 cm2). The mercury-plated silver disk electrode was prepared by dipping the freshly-polished silver electrode

into triply-distilled mercury for 10 min [32]. Periodically, the hydrogen evolution background current in pH 7 buffer was checked to be \ 10 mA at −1.8 V, and re-dipped if necessary. A Pt micro-disc electrode (radius 68 mm) was used in chronoamperometry measurements. The solution was purged with argon for 20 min before each measurement and no IR compensation was used. A peak separation of 66 mV was observed for ferrocene under the same conditions. The water employed was triply distilled and acetonitrile (Aldrich HPLC) grade was used as received. The solutions were 0.1 mol dm − 3 in acetonitrile with tetra-n-butylammonium hexafluorophosphate (TBAH, Fluka, electrochemical grade) as the supporting electrolyte. Bulk electroysis was carried out in a specially designed air-tight cell with separated compartments for counter and reference electrodes. A mercury pool was used as a working electrode with 20 cm3 exposed area. The cell volume was 125 cm3 of which 90 – 110 cm3 was occupied by gaseous products. A typical electrolysis would use 20 cm3 of 10% v/v MeCN/H2O/0.1 mol dm − 3 TBAH with 0.5 mmol dm − 3 of binuclear complex at − 1.7 V using a 20 cm2 mercury pool electrode. Gas samples (100 ml) were removed through a septum using an syringe fitted with an air-lock, and then injected into the GC (Pye-Unicam) fitted with a carbosphere column (Alltech) running at 100°C and TCD detector, and with a flow-rate (Ar) of 4 cm3 min − 1.

3. Results and discussion

3.1. Electrochemistry of penta-aza binucelar complexes [Ni2L 2 – 5] 4 + 3.1.1. Cyclic 6oltammetry at a glassy carbon electrode in acetonitrile The CVs of all the complexes clearly showed two reversible NiII/I and NiIII/II waves (see Fig. 1(a) for [Ni2L4]4 + ). The potentials of the waves are very similar to (1) itself, despite the expected small positive shift due to interaction with the non-bonded tertiary nitrogen atom and/or different molecular mechanics characteristics of CN bonds as compared to CC ones [26,33,34]. The NiIII/II data are similar to those reported previously by us [30]. The potentials do not vary significantly as the length of the connecting chain is varied (Table 1). These potentials are comparable to the mononuclear complex [NiL1]2 + , in contrast to the binuclear complexes prepared by alkylation of one of the coordinating N atoms [35] which show more positive potentials compared to the mononuclear complex. However, in the latter case this is due to the poorer ligand field of the tertiary N atom and/or steric interactions with the coordination of anions in the NiIII state [35,36].

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

2065

Fig. 1. Cyclic voltammograms (100 mV s − 1) at GC electrode in CH3CN/0.1 mol dm − 3 TBAH for (a) 1 mmol dm − 3 solutions of 1 ( – – ) and [Ni2L4]4 + (bold) under Ar and (b) 1 mmol dm − 3 solutions of 1 ( – – ) and [Ni2L4]4 + (bold) under CO2.

Electrochemical data for the reduction of NiII centres in (4) in which cyclam units are linked via a single CC bond have been reported by two separate research

groups [13,15], as has data on the closely-related tetramethyl derivative (5) [14]. The results obtained by these groups differ as to whether the NiII/I wave is

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

2066

irreversible in aqueous media or not. Because of the low solubility of the compounds under study, we have found that the NiII/I wave in aqueous media is obscured by hydrogen evolution at the Hg/Ag electrode. Therefore we have studied the redox behaviour of the NiII/I couple in acetonitrile or acetonitrile/water mixtures under argon. The similarity of the redox potentials of both the mononuclear complex and binuclear complexes sugTable 1 Cyclic voltammetry data for binuclear and mononuclear complexes in acetonitrilea Complex

NiII/I

NiIII/II

E1/2 (V)

DEp (mV)

E1/2 (V)

DEp (mV)

1 [NiL1]2+ [Ni2L2]4+ [Ni2L3]4+ [Ni2L4]4+ [Ni2L5]4+ [Ni2L6]4+ [Ni2L7]4+ [NiL8]2+ [Ni2L9]4+ [Ni2L10]4+

−1.440 −1.43 −1.44 −1.44 −1.435 −1.44 −1.39 −1.12 −1.12 −1.30 −1.30

68 73 98 120 126 139 176 83 68 78 82

+1.008 +0.99 +1.00 +1.00 +1.01 +1.03 +1.02 +1.27 +1.28 +1.45 +1.36

84 76 84 85 128 186 285 76 74 318 120

[Ni2L11]4+

−1.24

92

+1.44

362

[NiL12]2+

−1.31

83

+1.41

296

a Conditions: 0.5 mmol dm−3 binuclear complexes and 1 mmol dm−3 mononuclear complexes in acetonitrile with 0.1 mol dm−3 TBAH as the supporting electrolyte (glassy carbon electrode, area =0.071 cm2). Scan rate 100 mV s−1.

Table 2 Chronoamperometrically-determined diffusion coefficients (D) and n valuesa Compound

D/10−6 cm2 s−1

Ferrocene 1 [NiL1]2+ [Ni2L4]4+ [Ni2L7]4+ [NiL8]2+ [Ni2L9]4+ [Ni2L10]4+ [Ni2L11]4+ [NiL12]2+

10.1 5.3 14.5 2.5 3.9 8.6 2.54 0.72 1.09 1.68

a

90.4 90.7 90.36 90.14 90.9 90.3 90.34 90.96 90.15 90.22

n 1.13 0.8 0.95 2.4 1.6 1.07 2.02 2.29 2.02 0.74

90.06 90.15 90.03 90.20 90.6 90.05 90.38 90.43 90.38 90.14

Conditions: for 1 mmol dm−3 solutions of the complexes in acetonitrile with 0.1 mol dm−3 tetra-n-butylammonium hexafluorophosphate as the supporting electrolyte. obtained at a Pt micro-disk electrodelectrode radius = 68 mm.

gests that the two nickel centres in the binuclear complex can behave independently. The oxidation and reduction peak heights of the CVs (Fig. 1(a)) indicate that two electrons are exchanged in each redox transformation. This was proved by employing a Pt microelectode and chronoamperometry method (stepping over the NiIII/II wave) to calculate independently the diffusion coefficients (D) and the number of electrons transferred (n) as described previously [37]. Before examining our complex we measured D and n values for ferrocene and (1), obtaining values in good agreement with published data [38]. The results are given in Table 2. The obtained value of the diffusion coefficient for binuclear complex was significantly less than for complex (1). This helps to explain why the currents observed appear to be somewhat less than twice the mononuclear complex currents in the CV. The number of electrons transferred in the oxidation – reduction process is higher (2.4 +6 0.2) than the expected two electrons which is possibly caused by the double layer charging current (the data was not corrected for the background), or could reflect current in between two non-interacting one-electron processes(n= 2) and a concerted two-electron transfer (n=2.82). Indeed the observed value is quite close to the value expected for two species accidentally having the same redox potential [39]. The observed peak separations for the NiIII/II couple are somewhat higher than the predicted theoretical values for electrochemically reversible waves indicating a quasi-reversible process for both mononuclear complex and binuclear complex. It is interesting to compare our DEp (DEp = E ap − E cp) peak separation trends (ranging from 76 to 285 mV) with those reported for other binuclear complexes. Can some of the broadening be attributed to partial resolution of the mixed-valent redox couples, e.g. in the anodic scan resolution of the [NiIIINiIII]/[NiIIINiII] and [NiIIINiII]/[NiIINiII] couples? In (4) the wave on the positive scan has a peak separation DEp of 130 – 150 mV corresponding to an (unresolved) potential difference in the [NiIIINiIII]/[NiIIINiII] and [NiIIINiII]/[NiIINiII] couples of 66 – 75 mV, respectively [40]. In contrast, the corresponding wave of (6) has a DEp of 100 mV [11], which according to [41] corresponds to a difference in potentials of about 80 mV for the [NiIIINiIII]/[NiIIINiII] and [NiIIINiII]/ [NiIINiII] couples. If the two cyclam groups are joined by alkylation of a coordinating N atom then the metal centres will be closer together. Thus Ciampolini et al [16] showed that there was significant extra peak broadening (DEp) compared to the mononuclear complex (due to differences in the [NiIIINiIII]/[NiIIINiII] and [NiIIINiII]/[NiIINiII] couples), but this extra peak separation became negligible when the chain was greater than 3 methylene units long. Similar results were noted in binuclear Schiff base complexes directly linked via a

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074 Table 3 Electrochemical data for the electrocatalytic reduction of CO2a Complex

Under Ar

Under CO2

E1/2 (V)

DEp (mV)

ic/mA

Ep (V)

−1.43 −1.47 −1.49 −1.45 −1.48 −1.47 −1.13 −1.16 −1.33

130 226 256 298 230 260 78 78 129

800 375 368 376 397 447 63 27 21,36

[Ni2L10]4+ −1.29

137

28,43

[Ni2L11]4+ −1.34

104

19, 47

80

18, 40

−1.92 −1.73 −1.72 −1.77 −1.76 −1.78 −1.49 −1.56 −1.39, −1.73 −1.43, −1.68 −1.41, −1.82 −1.41, −1.68

[NiL1]2+ [Ni2L2]4+ [Ni2L3]4+ [Ni2L4]4+ [Ni2L5]4+ [Ni2L6]4+ [Ni2L7]4+ [NiL8]2+ [Ni2L9]4+

[NiL12]2+

−1.34

a Conditions: 0.5 mmol dm−3 solutions of binuclear complexes and 1 mmol dm−3 solutions of mononuclear complexes in acetonitrile/10% H2O/0.1 mol dm−3 TBAH. (Scan rate 100 mV s−1).at a mercury-plated silver electrode (area = 0.0314 cm2). E cp = reduction potential of CO2. i cp = catalytic current.

coordinating N atom [42], and 5-coordinate complexes with saturated ligands [43]. In face-to-face macrocycles the metal–metal distances are too long to observe any unusual effects due to metal–metal interaction [44]. Another factor which could be used to explain the effect observed is the possible dependence of the quilibrium between four- (square-planar) and six- (octahedral) coordinate nickel(II) on the length of the bridge, in other words on the structure of the bismacrocyclic molecule as a whole. It is well known that the E1/2 values of the NiII/I and especially the NiIII/II couples are strongly dependent on the solvent and the electrolyte. It has been reported that for bismacrocyclic complexes joined through coordinating nitrogen atoms this equilibrium shifts to the octahedral form if the metal–metal distance becomes smaller, because this reduces the electrostatic repulsion between the metal centres [11,45]. On the other hand, our spectroscopic study on the complexes [Ni2L2 – 5]4 + [30,46] revealed that the length of the bridge has no influence on such equilibria in acetonitrile or perchlorate-containing aqueous solution, though it is of importance in sulfate-containing aqueous media. This was explained by a specific two-centered (metal and NH group) coordination of hydrogensulfate by nickel complexes, but in this case the opposite tendency is observed: the shorter the bridge the higher the proportion of square planar form persists in solution.

2067

At the same time, spectroscopic data have revealed that [Ni2L9 – 11]4 + shows no tendency to bind axial ligands in acetonitrile, irrespective of the structure of the bridge [31]. Therefore we conclude that neither metal – metal interactions nor equilibria between four- and six-coordinated forms of nickel(II) are likely to be responsible for the large peak separations observed. As a matter of fact, we observe a large increase in the peak separations for both redox transitions as n increases. Thus the more probable explanation is that the rate of charge transfer to the larger molecules is slower, because a greater degree of rearrangement is necessary prior to electron transfer, and/or because the molecule is strongly adsorbed on the electrode, blocking fast electron transfer. In the presence of CO2 in acetonitrile the NiII/I redox couple became irreversible for both 1 and the binuclear complexes indicating reaction of CO2 with reduced Ni(I) centres (Fig. 1(b)). The data cannot, however, be used to determine the number and mode of binding of CO2 to the Ni(I) centres of the binuclear compound.

3.1.2. Catalytic acti6ity towards CO2 reduction on mercury electrodes: cyclic 6oltammetry at a mercury-plated sil6er electrode in acetonitrile/10% 6/6 water Catalytic reduction of CO2 with surface-confined catalyst is best observed at mercury electrodes rather than GC electrodes due to less extensive adsorption onto the latter as explained elsewhere [47]. The mercury-coated silver electrode was used in experiments designed to compare catalytic activity (Table 3). An example of the catalytic current observed in the presence of a binuclear complexes is shown in Fig. 2(a) for [Ni2L4]4 + . It should be noted that (1) itself shows only an increasing background current (presumably due in large measure to hydrogen production). From our studies of a wide range of complexes it seems that (1) is unique in this respect. The currents for the binuclear complexes are generally about half that for the pentaaza mononuclear complex [NiL1]2 + at the same concentration of nickel sites in solution (Fig. 2(b)). It is tempting then to assume that the binuclear complexes have only one active nickel centre: we know from many studies that the catalysis takes place predominantly at adsorbed nickel complex sites and that once the surface is fully covered (which it will be at the concentrations used in the determination of catalytic currents) then the catalytic current remains constant. If the binuclear complex (in half the concentration in solution) is adsorbing with both nickel sites active then the surface coverage of catalytic sites, and therefore the catalytic current will be the same as the mononuclear complex. Instead, the observed lower currents for the binuclear complexes indicate that the activity of adsorbed nickel centres on the electrode is lower, and the halving of the current

2068

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

favours the proposition that only one of the nickel centres maintains full activity, the other failing to adsorb on the electrode and act as an effective catalyst for CO2 reduction. The complex (4) was also about half as active as the mononuclear complex, and this was explained using similar arguments [13]. The much larger catalytic current for the p-xylylene linked chain supports this view since this more rigid linker can help both nickel centres achieve planarity on the electrode.

The complexes possessing tetra-, hexamethylene and xylyl bridges show larger splittings in their catalytic waves than the mononuclear complex; in particular there is a shoulder on the high potential side of the main catalytic peak and the peak itself is split into two (Fig. 2(a)). We have measured the concentration dependence of the peak for the [Ni2L4]4 + complex and compared it with the mononuclear complex [NiL1]2 + (Fig. 3) [48].

Fig. 2. Cyclic voltammograms (100 mV s − 1, SCE reference electrode) at a mercury – silver amalgam electrode for (a) 0.5 mmol dm − 3 solution of binuclear complex [Ni2L4]4 + under Ar ( – – ) and under CO2 (bold) in CH3CN/0.1mol dm − 3 TBAH with 10% H2O and (b) the same for 1 mmol dm − 3 mononuclear complex [NiL1]2 + .

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

2069

Fig. 3. The concentration dependence of the catalytic currents for mononuclear [NiL1]2 + (triangles) and the peak catalytic current of the binuclear Ni complex [Ni2L4]4 + (diamonds) in saturated CO2/CH3CN/0.1mol dm − 3 TBAH/10% H2O solution.

The non-linear dependence is typical for a surfaceconfined catalytic process. These peaks have quite similar concentration dependence and could obey a Langmuir– type isotherm for the current to give pseudo-absorption coefficients of 0.046 and 0.084 mmol dm − 3 for the binuclear and the mononuclear complexes, respectively. The catalytic current increases slightly as the connecting chain length (n) increases (Table 3), reaching about 6% higher for the [Ni2L5]4 + complex, despite the longer chain occupying a larger area on the electrode and thus lowering the effective electrode coverage. We believe that the trend is instead due to greater conformational flexibility in the complexes with longer spacer units, allowing more effective interaction of both nickel sites with the surface.

3.1.3. Controlled-potential electrolysis of [Ni2L 4] 4 + So far, we have carried out bulk CO2 electrolysis and product analysis studies of [Ni2L4]4 + only. Electrolysis of 20 cm3 of 10% v/v MeCN/H2O/0.1 mol dm − 3 TBAH at − 1.7 V using a 20 cm2 mercury pool electrode in a divided cell yielded CO and H2 as the major products in the headspace gas (monitored by GC). After about 10% electrolysis (as defined as the amount of charge passed relative to the calculated charge based on the mmoles of dissolved CO2 present being reduced by 2e per molecule to CO), we observed a sudden rise in the current and a corresponding increase in the amount of H2 (Fig. 4). This behaviour is also observed for (1) and probably indicates deactivation of the catalyst perhaps due to nickel deposition. The electrolysed solution was analysed by electrospray mass spectrometric analysis of the soluble

products. There was no evidence for either formate or oxalate formation. We may compare our results on the catalytic activities of the complexes under consideration with those of (4) [13] and (5) [14] under similar conditions. In the former work, the bismacrocyclic complex was half as active as the mononuclear complex, and in addition produced a small amount of H2 ( B3% Faradaic yield). No irreversible decomposition of the catalyst was detected. For complex (5) [14] the focus was on the photochemical reduction of CO2 in solution, in which the binuclear complex (5) was up to 8 times more active than the mononuclear complex and 12 times more active than (4). It was also more selective for CO versus H2. The electrocatalytic behaviour was also reported as a footnote, and in that case the binuclear complex was 10 times less active than the mononuclear complex. These observations appear to highlight the important role of the surface coverage and steric barriers to strong adsorption of both nickel centres when comparing binuclear and mononuclear complex catalytic activity.

3.2. Electrochemistry of [Ni2L 7] 4 + The CV data of [Ni2L7]4 + in CH3CN at a glassy carbon electrode shows that the E1/2 values for the redox couples are similar to those of its mononuclear complex analogue [NiL8]2 + (included in Table 1). The potentials of the waves are very much more positive than the binuclear complexes [Ni2L2 – 5]4 + because of the decrease in the ligand field strength on alkylation of the donor nitrogen atoms and the increased steric crowding

2070

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

of the axial sites of the metal ion [36]. The peak heights of the CV and microelectrode chronoamperometry data of [Ni2L7]4 + in CH3CN indicate that two electrons are exchanged in the redox processes. The observed D and n values for [Ni2L7]4 + and [NiL8]2 + are presented in Table 2. The CV recorded with a mercury-plated silver electrode under CO2 in CH3CN/10% water is shown in Fig. 5(a). An irreversible peak was seen near − 1.49 V. Most interestingly, the NiII/I couple remained unaffected under a CO2 atmosphere (Fig. 5(a), compare mononuclear complex, [NiL8]2 + (Fig. 5(b)). A sharp anodic peak was observed near −0.14 V on the reverse scan which is evidence for the formation of an insoluble carbonyl adduct during CO2 reduction [3]. The catalytic currents observed under CO2 in comparison to [NiL8]2 + are included in Table 3. The [Ni2L7]4 + produced a higher catalytic current than its mononuclear analogue, probably because the analogue is much more sterically hindered at the nickel than the binuclear complex. Both [Ni2L7]4 + and [NiL8]2 + displayed a much poorer catalytic activity for carbon dioxide reduction when compared to the unhindered mononuclear hexaazacyclam complex (2) under similar conditions [4].

3.3. Electrochemistry of hepta-aza bismacrocyclic complexes [Ni2L 9 – 11] 4 + and comparison with [NiL 12] 2 + 3.3.1. Cyclic 6oltammetry at a glassy carbon electrode in acetonitrile The CVs of all the complexes clearly showed one reversible or partially reversible wave for both NiII/I and

NiIII/II redox transformations (see Fig. 6(a) for [Ni2L9]4 + ). Chronoamperometric measurements confirmed the two-electron nature of these waves (Table 2). For reasons mentioned in connection with [Ni2L7]4 + the potentials of the waves are very much more positive than for the bis(pentaazamacrocyclic) [Ni2L2 – 5]4 + described above or indeed complex (1) itself. Additionally, the tetraazabicyclononane fragment corresponds to ethylenediamine in terms of the space occupied around the nickel centre. As such, these complexes could be considered as 13-membered macrocycles, rather than 14-membered ones, and this would result in an anodic shift in potentials because the hole of 13-membered macrocycle is too small to accommodate the nickel(I) ion [31]. As observed above for the pentaaza complexes, there is a large increase in the peak separation as n, i.e. the linking chain length, increases. This reflects a slower rate of charge transfer to the larger molecules, probably because a greater degree of rearrangement is necessary prior to electron transfer.

3.3.2. Catalytic acti6ity towards CO2 reduction: cyclic 6oltammetry at a mercury-plated sil6er electrode in acetonitrile/10% 6/6 water Under CO2 this family of complexes, as well as the mononuclear model complex, show quite different behaviour to that of the binuclear complexes described above and (1) (data summarised in Table 3). Instead of a single catalytic peak, there are two peaks (Fig. 6(a) for [Ni2L9]4 + ), one very close to the NiII/I potential and the other at more negative potentials, with a position close to the catalytic CO2 reduction current peak for (1)

Fig. 4. Time evolution of CO (circles) and H2 (squares) during bulk electrolysis in saturated CO2/MeCN/10% H2O/0.1mol dm − 3 TBAH at −1.7 V for [Ni2L4]4 + (0.5 mmol dm − 3).

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

2071

Fig. 5. (a)CVs for [Ni2L7]4 + binuclear complex (1 mmol dm − 3) under Ar ( – – ) and CO2 (bold). (b) CVs for the mononuclear complex [NiL8]2 + (1 mmol dm − 3) under Ar( – – ) and CO2 (bold). (MeCN/10% H2O/0.1mol dm − 3 TBAH, 100 mV s − 1at a Hg/Ag electrode).

in this solvent system. We assign these processes to the reduction of CO2 by the Ni2 + centres of the binuclear complex in solution and the surface catalytic current peak, respectively. The concentration dependence of these peaks confirms this assignment. Note it is only because the catalytic currents are so small (ca. 10% of the penta-aza binuclear complexes) that we are able to clearly observe the NiII/I peak due to soluble complex

catalysis. This demonstrates that the nickel sites are somewhat less hindered than in [Ni2L7]4 + . Closer examination of the mononuclear complex CV (Fig. 6(b)) reveals similar behaviour but the two peaks are less well resolved. The lower catalytic currents observed for this series of mononuclear and binuclear complexes is almost certainly due the steric crowding of the axial nickel sites. We speculate that the weaker interaction

2072

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

with the electrode arising from ligand steric interactions places far fewer stereochemical constraints on the adsorption of both nickel centres to the same extent as the binuclear complex, and hence the catalytic currents for binuclear and mononuclear complex are comparable. Nevertheless, the surface catalytic currents of the binuclear complexes are observed to increase by about 20% as the connecting chain length increases from two to five methylene chains (Table 3).

4. Conclusions In this paper we have shown that steric considerations in dinickel complexes can have a powerful influence on the efficiency of CO2 reduction mediated by these complexes. In the first series of binuclear complexes [Ni2L2 – 6]4 + studied the catalytic waves for CO2 reduction are about half that of the mononuclear complex. The most likely explanation is that steric con-

Fig. 6. (a)CVs for 0.5 mmol dm − 3 [Ni2L9]4 + binuclear complex under Ar ( – – ) and CO2 (bold). (b) CVs for the mononuclear complex [NiL12]2 + (1 mmol dm − 3) under Ar( – – ) and CO2 (bold) (MeCN/10% H2O/0.1mol dm − 3 TBAH, 100 mV s − 1 at a Hg/Ag electrode).

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

straints permit strong interaction of only one nickel centre of the binuclear complex on the surface. Further evidence for the presence of at least two catalytic centres comes from the observed splitting in the catalytic peaks of the binuclear complexes. In the more sterically crowded complex, [Ni2L7]4 + there was no evidence even of any interaction of the reduced complex with CO2 in dry acetonitrile. These steric barriers prevented strong electrode interaction and led to a much lower catalytic activity. The series of heptaaza bismacrocyclic complexes [Ni2L9 – 11]4 + showed well-separated peaks due to solution and surface catalytic activity, and the surface catalytic currents were now comparable to mononuclear comples at the same effective concentration. We proposed that the weaker interaction with the electrode arising from ligand steric interactions places far fewer stereochemical constraints on the adsorption of both nickel centres to the same extent as the binuclear complex, and hence the catalytic currents for bis-and monomacrocyclic complexes are comparable in this series.

[4] [5] [6]

[7]

[8] [9] [10]

Acknowledgements [11]

Dedicated to the memory of Professor R.W. Hay. We thank CVCP for the award of an Overseas Research Studentship to D.C.L. de Alwis. JAC thanks the University of St Andrews and the Fulbright Commission for the award of a visiting scholarship to Caltech, and also COST for a cooperative travel grant. The work has also been supported by a grant from the Royal Society Joint Projects with the Former Soviet Union as well as the Foundation of Basic Research of the Ministry of the Ukraine for Science and Technology.

[12] [13] [14]

[15] [16]

References [1] (a) J. Costamagna, G. Ferraudi, J. Canales, J. Vargas, Coord. Chem. Rev. 148 (1996) 221. (b) D.J. Darensbourg, M.W. Holtcap, Coord. Chem. Rev. 153 (1996) 155. (c) W. Leitner, Coord. Chem. Rev. 153 (1996) 257. (d) Electrochemical and Electrocatalytic Reactions of Carbon Dioxide, B.P. Sullivan., K. Krist, H.E. Guard (editors), Elsevier, Amsterdam, 1993. (e) P. Christensen, A. Hamnett, A.V.G. Muir, J.A. Timney, J. Chem. Soc. Dalton Trans., 1455 (1992), and references therein. (f) P.G.S. Tamburini, P.A. Vigato, Coord. Chem. Rev. 139, (1995) 17. (g) K. Rajeshwar, J.G. Ibanez, G.M. Swain, J. Appl. Electrochem. 24 (1994) 1077. (h) E. Fujita, B.S. Brunschwig, T. Ogata, S. Yanagida, Coord. Chem. Rev. 132 (1994) 195. (i) M.A. Halcrow, G. Christou, Chem. Rev. 94 (1994) 2421. [2] (a) M. Beley, J.P. Collin, R. Ruppert, J.P. Sauvage, J. Chem. Soc. Chem. Comm. (1984) 1315. (b) M. Beley, J.P. Collin, R. Ruppert J.P. Sauvage, J. Am. Chem. Soc. 108 (1986) 7461. [3] (a) G.B. Balazs, F.C. Anson, J. Electroanal. Chem. 322

[17] [18] [19] [20] [21] [22] [23] [24] [25] [26]

[27]

2073

(1992) 325. (b) G.B. Balazs, F.C. Anson, J. Electroanal. Chem. 361 (1993) 149. C.I. Smith, J.A. Crayston, R.W. Hay, J. Chem. Soc. Dalton Trans, (1993) 3267. R.W. Hay, J.A. Crayston, T.J. Cromie, P. Lightfoot, D.C.L. de Alwis, Polyhedron 16 (1997) 3557. (a) P.V. Bernhardt, G.A. Lawrence, Coord. Chem. Rev. 104 (1990) 297. (b) L.F. Lindoy, Adv. Inorg. Chem. 45 (1998) 75. (c) G. DeSantis, L. Fabbrizzi, M. Licchelli, C. Mangano, P. Pallavicini, A. Poggi, D. Sacchi, A. Taglietti, Supramol. Chem. 6 (1996) 239. G.J. Bridger, R.T. Skerl, S. Padmanabhan, S.A. Markellucci, G.W. Henson, M.J. Abrams, H.C. Joao, M. Witvrouw, K. De Vreese, R. Pauwels, E. De Clerq, J. Med. Chem. 39 (1996) 109. R. Hetlich, H.-J. Schneider, J. Am. Chem. Soc. 119 (1997) 5638. G. DeSantis, L. Fabbrizzi, M. Licchelli, P. Pallaricini, Coord. Chem. Rev. 120 (1992) 237. (a) J.P. Collman, P.S. Wagenknecht, J.E. Hutchison, Angew. Chem. 33 (1994) 1537. (b) J.P. Collman, J.E. Hutchison, M.A. Lopez, A. Tabard, R. Guilard, W.K. Seok, J.A. Ibers, M. L’Her, J. Amer. Chem. Soc. 114 (1992) 9869. (c) F.C. Anson, C. Shi, B. Steiger, Acc. Chem. Res. 30 (1997) 437. A. McAuley, S. Sumbramanian, Inorg. Chem. 36 (1997) 5376. S.K. Mandal, L.K. Thompson, K. Nag, J.-P. Charlaud, E.J. Gabe, Inorg. Chem. 26 (1987) 1391. J.-P. Collin, A. Jouaiti, J.-P. Sauvage, Inorg. Chem. 27 (1986) 1988. (a) K. Mochizuki, S. Manaka, I. Takada, T. Kondo, Inorg. Chem. 35 (1996) 5132. (b) K. Mochizuki, H. Gotoh, M. Suwabe, T. Sakakibara, Bull. Chem. Soc. Jpn. 64 (1991) 1750. B.D. Steffey, C.J. Curtis, D.L. DuBois, Organometallics 14 (1995) 4937. M. Ciampolini, L. Fabbrizzi, A. Perotti, A. Poggi, B. Seghi, F. Zanobini, Inorg. Chem. 26 (1987) 3527. K. Weighardt, I. Tolksdorf, W. Herrmann, Inorg. Chem 24 (1985) 1230. R. Das, F. Nag, Inorg. Chem. 30 (1991) 2833. L. Fabbrizzi, L. Montagna, A. Poggi, T.A. Kaden, L.C. Siegfried, J. Chem. Soc. Dalton Trans. (1987) 2631. M.P. Suh, S.-G. Kang, Inorg. Chem. 27 (1988) 2544. S.-K. Jung, S.-G. Kang, M.P. Suh, Bull. Korean Chem. Soc. 10 (1989) 362. M.P. Suh, W. Shin, S.-G. Kang, M.S. Lah, T.M. Chung, Inorg. Chem. 28 (1989) 1602. M.P. Suh, J. Choi, S.-G. Kang, W. Shin, Inorg. Chem. 28 (1989) 1763. S.-G. Kang, S.-K. Jung, J.R. Kweon, Bull. Korean Chem. Soc. 11 (1990) 431. M.P. Suh, S.-G. Kang, V.L. Goedken, S.-H. Park, Inorg. Chem. 30 (1991) 365. L. Fabbrizzi, A.M.M. Lanfred, P. Pellavicini, A. Perotti, A. Taglietti, F. Ugozzoli, J. Chem. Soc. Dalton Trans. (1991) 3262. (a) R.W. Hay, A. Danby, P. Lightfoot, Ya. D. Lampeka, Polyhedron 16 (1997) 2777. (b) A. Danby, Ph. D. Thesis, University of St Andrews (1996).

2074

C. de Alwis et al. / Electrochimica Acta 45 (2000) 2061–2074

[28] S-G. Kang, S.-K. Jung, J.K. Kweon, Bull. Korean Chem. Soc. 12 (1991) 219. [29] S.G. Kang, S.-K. Jung, J.K. Kweon, M.-S. Kim, Polyhedron 12 (1993) 353. [30] (a) S.V. Rosokha, Ya.D. Lampeka, J.Chem.Soc. Chem.Commun. (1991) 1077. (b) S.V. Rosokha, Ya.D. Lampeka, I.M. Maloshtan, J. Chem. Soc. Dalton Trans. (1993) 631. [31] S.V. Rosokha, I.M. Maloshtan, Ya.D. Lampeka, Zhurnal Neorg. Khimii 38 (1993) 1012. [32] M. Ciszkowska, M. Donten, Z. Stojek, Anal. Chem. 66 (1994) 4112. [33] F. Abba, G. De Santis, L. Fabbrizzi, M. Licchelli, A.M.M. Lanfredi, P. Pallavicini, A. Possi, F. Ugozzoli, Inorg. Chem. 33 (1994) 1366. [34] S.V. Rosokha, Ya.D. Lampeka, Teor. Eksper. Khimya 28 (1992) 436. [35] (a) R. Schneider, A. Riesen, T.A. Kaden, Helv. Chim. Acta 69 (1986) 53. (b) A. Urfer, T.A. Kaden, Helv. Chim Acta 77 (1994) 23. [36] E.K. Barefield, G.M. Freeman, D.G. Van Derveer, Inorg. Chem 25 (1986) 552. [37] G. Denuault, M.V. Mirkin, A.J. Bard, J. Electroanal. Chem. 308 (1991) 27.

.

[38] D.T. Pierce, T.L. Hartfield, E.J. Billo, Y. Ping, Inorg. Chem. 36 (1997) 2951. [39] A.J. Bard, L.R. Faulkner, Electrochemical Methods, Wiley, New York, 1980. [40] L. Fabbrizzi, L. Montagna, A. Poggi, T.A. Kaden, L.C. Siegfried, J. Chem. Soc. Dalton Trans, (1987) 2631. [41] D.E. Richardson, H. Taube, Inorg. Chem. 20 (1981) 1278. [42] K. Mochizuki, Bull. Chem. Soc. Jpn. 61 (1988) 1023. [43] S.J. Brudenell, L. Spiccia, A.M. Bond, P.C. Mahon, D.C.R. Hockless, J. Chem. Soc. Dalton Trans, (1998) 3919. [44] D.H. Busch, G.G. Christoph, L.L. Zimmer, S.C. Jackels, J.J. Grzybowski, R.C. Callahan, M. Kojima, K.A. Holler, J. Mocak, N. Herron, M. Chavan, W.P. Schammel, J. Am. Chem. Soc. 103 (1981) 5107. [45] K. Mochizuki, A. Iijima, Y. Endoh, Y. Ikeda, Bull. Chem. Soc. Jpn. 63 (1990) 565. [46] L.V. Tsymbal, S.V. Rosokha, Ya.D. Lampeka, J. Chem. Soc. Dalton Trans. (1995) 2633. [47] C. de Alwis, J.A. Crayston, in preparation. [48] Note that the CO2 electrocatalytic activity and concentration dependence of the related N-methyl complex has been reported in Ref. [33].