template for electronic submission to acs...
TRANSCRIPT
The role of electrode-catalyst interactions in en-abling efficient CO2 reduction with Mo(bpy)(CO)4 as revealed by vibrational sum-frequency generation spectroscopyGaia Neri,a Paul M. Donaldsonb* and Alexander J. Cowana *
AUTHOR ADDRESS. a Department of Chemistry and Stephenson Institute for Renewable Energy, Uni-versity of Liverpool, L69 7ZD, Liverpool, UK, [email protected], b Central Laser Facility, STFC Rutherford Appleton Laboratory, Harwell, Didcot, Oxfordshire, OX11 0QX, UK, [email protected]: Group 6 metal carbonyl complexes ([M(bpy)(CO)4], M = Cr, Mo, W) are potentially promising CO2 reduction electrocatalysts. However catalytic activity onsets at prohibitively negative potentials and is highly dependent on the nature of the working electrode. Here we report in-situ vibrational SFG (VSFG) measurements of the electrocatalyst [Mo(bpy)(CO)4] at platinum and gold electrodes. The greatly im-proved onset potential for electrocatalytic CO2 reduction at gold electrodes is due to the formation of the catalytically active species [Mo(bpy)(CO)3]2- via a 2nd pathway at more positive potentials, likely avoiding the need for the generation of [Mo(bpy)(CO)4]2-. VSFG studies demonstrate that the strength of the inter-action between initially generated [Mo(bpy)(CO)4].- and the electrode is critical in enabling the formation of the active catalyst via the low energy pathway. By careful control of electrode material, solvent and electrolyte salt it should therefore be possible to attain levels of activity with group 6 complexes equiva-lent to their much more widely studied group 7 analogues.
INTRODUCTION: The electrochemical reduction of carbon dioxide offers a sustainable route to generating useful C1 feedstocks such as carbon monoxide and carbon based fuels. However the direct one electron reduction of CO2 is challen-ging,1 leading to intense interest into both metal electrodes and transition metal complexes that can facilitate the multi-proton, multi-electron re-duction of CO2 at more positive applied poten-tials.2 Numerous examples of transition metal electrocatalysts exist in the literature for the re-duction of CO2 to CO based on groups 7-10, with complexes of Mn, Re, Fe, Ru, Co, Ni and Ir3–14 showing high levels of selectivity towards the pro-duction of CO, although in many cases an un-desirably large overpotential is required for cata-lysis to occur at a significant rate. In contrast group 6 transition metal (Cr, Mo, W) complexes have received significantly less attention as po-tential CO2 reduction catalysts.15 This is perhaps surprising as Mo and W are known to be import-ant metal centres in biological systems for CO2 chemistry such as formate dehydrogenase.16
Therefore synthetic catalysts based on group 6 metals could represent a potentially fruitful, but
under-explored area of research. Of the studies reported,17–21 complexes of the form [M(bpy)(CO)4] are of particular interest, having demonstrated high Faradaic efficiencies for the reduction of CO2 to CO.17 FTIR spectroelectrochemical studies by Clark et al. showed that initial reduction leads to the formation of a ligand based radical [M(bpy)(CO)4].- which can then also undergo a further one electron reduction and CO loss to yield [M(bpy)(CO)3]2-, the proposed active catalyst for CO2 re-duction. Notably [M(bpy)(CO)3]2- is formally iso-electronic with [Mn(bpy)(CO)3]- and [Re(bpy)(CO)3]-, the highly active catalysts generated fol-lowing the reduction of the widely studied group 7 [M’(bpy)(CO)3X] complexes (M’ = Mn, Re, X = Br-, Cl-).4,6,7 Indeed in this important first study turnover frequencies (TOF) for the group 6 cata-lysts were found to be ~ 2-4 s-1 on glassy carbon electrodes (GCE), which is similar to the analog-ous Re complexes in acetonitrile in the absence of an additional proton source.17 However the relat-ively negative reduction potential of [M(bpy)(CO)4].- (e.g. M = Mo, Ered = -2.14 Vsce) prevents study in the presence of a Brønsted acid, condi-tions under which the Re and Mn catalysts are most active (e.g. [Re(bpy-tBu)(CO)3(CH3CN)(OTf)]
TOF = 570 s-1 in the presence of 1.4 M 2,2,2- Tri-fluoroethanol7), as direct reduction of the acids takes place. Although calculations indicate that [M(bpy)(CO)3]2- has charge delocalised across both the ligand and metal centre similarly to Re analogues,17 the reduction of [M(bpy)(CO)4].- is primarily bipyridine based. Therefore subsequent studies have focused on the use of alternative bidentate ligands in an attempt to positively shift the reduction potential of [M(bpy)(CO)4].- .19,21
Whilst these synthetic modifications are leading to encouraging results, it is striking that the greatest improvements in onset potential for CO2 reduction have been achieved through simply changing the working electrode material.18 Tory et al., reported that with [M(bpy)(CO)4] and a Au working electrode a ca. + 0.6 V shift in the onset potential for CO2 reduction occurs compared to Pt or GCE. This is an exciting result, with the group 6 catalyst then having an onset potential compar-able to some leading Re analogues.22 The exact mechanism of enhanced activity remains un-known however. Using Au the onset in catalysis occurs at potentials close to the first reduction of [Mo(bpy)(CO)4] to [Mo(bpy)(CO)4].- therefore it has been suggested that a small quantity of [Mo(bpy)(CO)3].- can be present in equilibrium with [Mo(bpy)(CO)4].-.18 Supporting this hypothesis are cyclic voltammetry (CV) experiments that show that the oxidation potential of [Mo(bpy)(CO)3]2- oc-curs at potentials close to that of the initial reduc-tion of [Mo(bpy)(CO)4] making it feasible that any generated [Mo(bpy)(CO)3].- could be reduced at these potentials to form the catalytic active spe-cies, [Mo(bpy)(CO)3]2- via this 2nd more facile route. However such surface species have not been observed by FTIR spectroelectrochemical experiments, which are typically dominated by electrochemically generated species in the bulk solution.
-1.5
-1.0
-0.5
0.0
-3.0 -2.5 -2.0 -1.5
-1.0
-0.5
0.0
Cur
rent
den
sity
(mA
cm
-2)
Au electrode
0.6 V
Pt electrode
Potential (V vs. Fc/Fc+)
Ar CO2
Figure 1: CV of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1 M TBAPF6 recorded under an Ar (black) and CO2 (red) atmosphere at 200 mV s-1 using either a Pt (top) or Au (bottom) electrode. The solid grey arrows indicate onset potentials for CO2 reduction which oc-curs at a potential ca. 0.6 V positive on Au compared to Pt.
In contrast to FTIR spectroscopy IR-Vis vibra-tional sum frequency generation (VSFG) spectro-scopy is an intrinsically surface specific vibra-tional spectroscopy technique allowing the detec-tion of sub-monolayer concentrations of interme-diate surface species.23 This provides the opportunity to gain new insights into the chemistry and stability of species at electrode surfaces that are not readily obtainable by conven-tional methods. The basic theory and applications of VSFG spectroscopy have been reviewed in detail elsewhere.24,25 Briefly, short IR and Visible laser pulses are overlapped spatially and temporally at a sample surface. When the IR laser frequency matches that of a surface molecule’s vibrational mode, the vibrationally polarised molecules can interact with visible light via their electronic po-larisability, resulting in the generation of light at the sum of the incident IR and visible frequencies. In most cases, due to interference effects, the light from bulk media is completely extinguished, leaving a weak, but detectable signal from just the interface. The intensity of the VSFG signal is dependent upon the IR and Raman activity/strength of the modes and on their number dens-ity and average orientation relative to the sur-face. An SFG signal from the nonresonant (NR) in-teraction of both IR and Visible light beams with the interface (via its electronic polarizability) is often also present. VSFG is now beginning to be applied to complex multicomponent systems in a small number of reports. For example, to study the formation of CO2 adducts at silver and plat-inum electrodes, with measurements of the NR-SFG intensity providing a rationale of the role of an ionic liquid (EMIM-BF4) in controlling the onset potential for catalysis. 26–28 VSFG spectroscopy has also been used to study the orientation of several examples of Mn and Re carbonyl complexes at a range of surfaces including Au and TiO2 although in the absence of an applied potential.29–33 Here we report an in-situ VSFG study on the elec-trocatalytic reduction of CO2 using [Mo(bpy)(CO)4]. To the best of our knowledge this repres-ents the first in-situ study of a molecular elec-trocatalyst under potentiostatic control by VSFG spectroscopy. VSFG data recorded during CV ex-periments provide clear spectroscopic evidence of the nature of the intermediates formed on the electrode surface and report on the nature of the electrode-catalyst interactions with both Au and Pt. Validation of the previously proposed mechan-ism of enhanced activity18 is an important step for the community, demonstrating that control of the electrode-catalyst interaction offers an alternative
route to unlocking the potential of this under-studied class of CO2 reduction catalysts.
2100 2000 1900 1800 1700
SFG intensity
-2.14 V
-2.14 V
-2.83 V
-1.44 V
Wavenumber / cm-1
VFc
/Fc+
-10.00
40.00
90.00
140.0
190.0
240.0
290.0
340.0
390.0
-1.44 V
Figure 2: In-situ VSFG spectra of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1 M TBAPF6 recorded under an Ar atmosphere at a polycrystalline Au electrode dur-ing a CV measurement (5 mV s-1, -1.44 to -2.83 V, figure S3) averaging for 1 s per spectrum using a IR-Vis delay of 0.45 ps to suppress NR-SFG signals.
RESULTS AND DISCUSSION: Using polycrystal-line Au and Pt electrodes, similar electrochemical behaviour to that previously reported was ob-served (Figure 1).17,18 Under argon both elec-trodes give a reversible reduction of [Mo(bpy)(CO)4], with E1/2 -1.93 V (Au) and -1.96 V (Pt) to form [Mo(bpy)(CO)4].- and a subsequent irrevers-ible reduction leading to the formation, at least transiently, of [Mo(bpy)(CO)3]2- at -2.60 V (Au) and -2.76 V (Pt). All potentials are reported vs. Fc/Fc+
(CH3CN).
2100 2000 1900 1800 1700 1600
SFG
inte
nsity
(a.u
.)
A = 0.02
(h, SFGNR)
(g, -2.8 V)
(f, -2.5 V)
(e, -2.3 V)
(d, -2.1 V)
(c, -1.9 V)
(b, -1.5 V)
Abs
.
Wavenumber / cm-1
(a, FTIR)
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
B Lorentz Fi t o f Sheet1 B
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
B Lorentz Fit of Sheet1 B
2 4 6 8 1 0
0. 0
0. 5
1. 0
1. 5
2. 0
B
A
B Lor entz Fi t of Shee t1 B
2 4 6 8 100. 0
0. 5
1. 0
1. 5
2. 0
B
A
B Lo rentz F it of Shee t1 B
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
B Lorentz Fit o f Sheet1 B
2 4 6 8 100. 0
0. 5
1. 0
1. 5
2. 0
B
A
B Lorentz F it of Sheet1 B
2 4 6 8 1 00 .0
0 .5
1 .0
1 .5
2 .0
B
A
B L or en tz Fit o f S he et 1 B
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
B Lorentz Fi t o f Sheet1 B
2 4 6 8 10
0. 0
0. 5
1. 0
1. 5
2. 0
B
A
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A 2 4 6 8 100. 0
0. 5
1. 0
1. 5
2. 0
B
A
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A2 4 6 8 10
0.0
0.5
1.0
1.5
2.0
B
A2 4 6 8 10
0. 0
0. 5
1. 0
1. 5
2. 0
B
A2 4 6 8 10
0. 0
0. 5
1. 0
1. 5
2. 0
B
A
B Lore ntz Fit of Sheet 1 B
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
B Lor en tz Fi t o f Sh ee t1 B
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A2 4 6 8 10
0.0
0.5
1.0
1.5
2.0
B
A
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
2 4 6 8 100. 0
0. 5
1. 0
1. 5
2. 0
B
A
2 4 6 8 10
0 .0
0 .5
1 .0
1 .5
2 .0
B
A
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
2 4 6 8 10
0. 0
0. 5
1. 0
1. 5
2. 0
B
A
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
2 4 6 8 100. 0
0. 5
1. 0
1. 5
2. 0
B
A
2 4 6 8 100. 0
0. 5
1. 0
1. 5
2. 0
B
A
[Mo(bpy)(CO)4]
[Mo(bpy)(CO)4].-
[Mo(bpy)(CO)3]2-
[Mo(bpy-H)(CO)4]-
2 4 6 8 100.0
0.5
1.0
1.5
2.0
B
A
2100 2000 1900 1800 1700 1600
Wavenumber / cm-1
MoCO
CO
CO
CON
N
MoCO
CO
CO
CON
N
ca. -2.0 V, + e-
MoCO
CO
CO
CON
N
2-
MoCO
COCON
N
2-
Mo
CO
COCO
N
N
+ CO
- CO
ca. -2.6 V, + e-
CO loss first, Au only
reduction first, Au + Pt
- CO
ca. -2.2 V, + e-
Figure 3: Left: Solution phase FTIR (a) and in-situ VSFG spectra (b-g) of [Mo(bpy)(CO)4] (1 mM) at a Au elec-trode in CH3CN and 0.1 M TBAPF6 under Ar. The FTIR spectrum is recorded at open circuit and the VSFG spectra at the potentials indicated. The intensity of the NR background (dashed line, h) of the cell recorded with 0 ps time-delay between the IR and 800 nm laser pulses provides an approximation of the spectrum of the broad-band IR pulse. Spectra (b-g) are recorded with a 0.45 ps delay between the IR and asymmetric 800 nm laser pulses. Centre: pictorial representation of the proposed assignments generated by Lorentzian curves at the frequencies indicated in the text, see figure S4,5 for fittings on which this is based. The assignments at -2.3 and -2.5 V are tentative due to the presence of a complex mixture of species (3) with spectral features at similar positions. The grey SFG peak at ca. 2090 cm-1 is due to CO at the gold electrode, the purple peaks are assigned to [Mo(bpy-H)(CO)4]- (see main text). Right: Scheme showing the proposed mechanisms of formation of key species studied here.
Oxidation of [Mo(bpy)(CO)3]2- occurs at -2.20 and -2.15 V on Au and Pt respectively. Under CO2 a sig-nificant current enhancement with both electrode materials indicates that electrocatalytic CO2 re-
duction occurs. Importantly, in-line with past work, we observe a significantly earlier onset po-tential for CO2 reduction on Au electrodes (ca. 0.60 V, -2.30 V) when compared to Pt (Figure 1).
The GCE (Figure S1) and Pt CVs show similar be-haviour and here we use a Pt electrode for com-parison to Au as the reflective nature of the ma-terial simplifies the SFG experiment. The data shown in figure 1 uses the same electrodes as employed in our spectroelectrochemical cell.
The nature of the gold electrode-complex inter-actions under potentiostatic control were studied using in-situ SFG spectroscopy of [Mo(bpy)(CO)4], Figure 2. In this work identification of VSFG sig-nals with background-free undistorted lineshapes/centre frequencies is aided by the in-troduction of a time delay (typically 0.4 to 0.45 ps) between the femtosecond (fs) broad-band IR pulse (500 cm-1) and the fs derived time-asym-metric visible picosecond (ps) pulse to suppress the NR SFG response.34 Notably in figure 2 we ob-serve Lorenztian-shaped VSFG lines showing clear potential dependent changes in the VSFG spec-tra, with vibrational bands of complexes shifting linearly in frequency with applied potential due to vibrational stark shifting and the formation of new bands at potentials corresponding to expected re-duction/oxidations. It is therefore apparent that the experiment is suitably sensitive to observe changes within the electrochemical double layer with minimal data acquisition times (1 s per spec-trum). The potential dependence of the ν(CO) bands indicates that they are not arising due to phantom transitions28 which can occur in optically thick samples due to mode specific linear absorp-tion of the IR prior to SFG signal generation. This point is fully discussed in the supporting informa-tion (figure S2). We note that in this study, all ex-periments were conducted with ppp polarisation and that all analysis was based on identifying band frequencies, rather than assessing orienta-tion changes from band intensity changes.
At potentials prior to the first reduction, all four υ(CO) bands of [Mo(bpy)(CO)4] are distinguishable in the VSFG spectrum (Figure 3b). These are as-signed through comparison to the known FTIR spectrum (Figure 3a) : 2015 cm-1 (A12), 1904 cm-1
(B1), 1875 cm-1 (A11) and 1832 cm-1 (B2),35 how-ever the spectrum is dominated by the B1 out-of-phase axial stretching mode which appears at ca. 1905 cm-1 at -1.5 V. Between -1.9 V and -2.1 V the SFG spectra show a decrease in the intensities of the [Mo(bpy)(CO)4] modes concomitant with the growth of new features at ca. 1874 cm-1 and 1808 cm-1 and a weak band at 1850 cm-1 (Figure 3c,d and S4 for multi-lorentzian fitting), which are as-signed to the B1, B2 and A11 modes of the singly reduced complex [Mo(bpy)(CO)4].- at the Au elec-trode by analogy to past FTIR studies, which re-port υ(CO) at 1991, 1871, 1843, 1805 cm-1 in solu-tion (THF).17,18 At the most negative potentials studied (-2.3 to -2.8 V) SFG signals are observed at 2090, 1870 (br.), 1835, 1790 (br.) and 1735 cm-1, Figure 3e-g and S4 for multi-lorentzian fit-ting. Both Figure 1 and CV’s recorded during the
SFG experiment (Figure S3) indicate that at these potentials [Mo(bpy)(CO)4].- is reduced and the weak features at 1835 and 1735 cm-1 can be readily assigned to the υ(CO) modes of [Mo(bpy)(CO)3]2- at, or close to the Au surface.17,18 Supporting this assignment is the increase in the intensity of the feature at 2090 cm-1 due to linearly bound CO on Au,36 due to ligand loss following reduction of [Mo(bpy)(CO)4].-. A more detailed study on the potential dependent formation of [Mo(bpy)(CO)3]2- is pre-sented in the following sections.
The strong broad features at 1870 cm-1 and 1790 cm-1 at -2.5 and -2.8 V in figure 3(g) are sim-ilar to the frequency the υ(CO) modes of [Mo(bpy)(CO)4].-, however by -2.8 V we would expect a minimal concentration of this species to be present at the electrode surface based on the CV data recorded in-situ (Figure 1, S3). Tory et al.18
reported the presence of a tetracarbonyl anion with very similar υ(CO) frequencies at 1995, 1874, 1848 and 1807 cm-1 in N-methyl-2-pyrrolidone (NMP) upon the reduction of [Mo(bpy)(CO)4].-, which was assigned to [Mo(bpy-H)(CO)4]-. Similar behavior has also been seen with 2,2’-dipyridy-lamine derivatives.19 It was proposed that the wa-ter in NMP reacted with [Mo(bpy)(CO)3]2- which was then followed by rapid CO re-coordination. The cell design for our spectroelectrochemical studies prevents experiments from being carried out under strict anhydrous conditions. Here we also assign the broad SFG features at 1870 and 1790 cm-1 to a species formed following the inter-action of water with [Mo(bpy)(CO)3]2-. Supporting the assignment are studies under CO2 (vide infra).
2100 2000 1900 1800 1700
-2.10 V -2.30 V -2.35 V -2.50 V -2.80 V
SFG
inte
snity
(a.u
., vs
. -1.
5 V
)
Wavenumber / cm-1
Figure 4: Difference VSFG spectra of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1 M TBAPF6 under an
Ar atmosphere at the potentials indicated versus a spectrum recorded at -1.5 V using a Au electrode.
To assess the point at which [Mo(bpy)(CO)3]2- is formed at the Au electrode we have replotted the VSFG data recorded at potentials close to the first reduction as difference spectra (versus the spec-trum of [Mo(bpy)(CO)4] at -1.5 V), figure 4. This allows for observation of small changes in SFG in-tensity that might otherwise be masked by the strong modes of [Mo(bpy)(CO)4]. The growth of the υ(CO) modes of [Mo(bpy)(CO)3]2- can now be more clearly observed at 1735 cm-1 and as a shoulder at 1835 cm-1 between -2.3 to -2.8 V. It is therefore apparent that a concentration of the catalytically active [Mo(bpy)(CO)3]2-
is formed on the Au electrode surface at least 300 mV anodic of the expected reduction potential as measured by CV, and a small increase in SFG intensity at 1735 cm -1 as early as -2.1 V may also be occurring, figure 4. These surface sensit-ive SFG studies contrast previous spectroelectrochemical stud-ies which do not show the presence of [Mo(bpy)(CO)3]2- in the bulk solution until the 2nd reduction potential is reached and we confirm the previously hypo-thesised formation of [Mo(bpy)(CO)3]2- via a 2nd
more facile surface mediated pathway.18 This is likely to be the establishment of an equilibrium between [Mo(bpy)(CO)4].- and [Mo(bpy)(CO)3].- at the Au surface, avoiding the need for the genera-tion of the high energy species [Mo(bpy)(CO)4]2-. However direct evidence for the transient forma-tion of [Mo(bpy)(CO)3].- remains elusive, hence the stabilisation of [Mo(bpy)(CO)4]2- by Au cannot be completely ruled out. An additional VSFG fea-ture is observed at ca. 1760 cm-1 between -2.1 and -2.35 V in the difference spectrum, figure 4, however assignment of this single band is not possible as it is in the spectral region where both [Mo(bpy)(CO)4]2- and [Mo(bpy)(CO)3].- are expec-ted to have υ(CO) stretches.17
2100 2000 1900 1800 1700 1600
-1.76
-1.76
-2.36
-1.16
Wavenumber / cm-1
VFc
/Fc+
-20.00
16.00
52.00
88.00
124.0
160.0
196.0
232.0
268.0
304.0
340.0
-1.16
SFG intensity(a)
2100 2000 1900 1800 1700 1600
CO2
SFG
inte
nsity
(a.u
.)
Wavenumber / cm-1
Ar (b)
Figure 5: (a) In-situ VSFG spectra of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1M TBAPF6 recorded under a CO2 atmosphere at a Au electrode during a CV mea-surement (5 mV s-1, -1.44 to -2.83 V) averaging for 1 s per spectrum. (b) VSFG spectra of [Mo(bpy)(CO)4] under argon (black) and CO2 (red) at a potential of –2.35 V.
Figure 5 (a) shows VSFG spectra recorded dur-ing a CV measurement of [Mo(bpy)(CO)4] under a CO2 atmosphere using an Au working electrode (related CV in Figure S6). At open circuit the spec-trum shows the presence of both [Mo(bpy)(CO)4] (1907 cm-1, Fig S7,8) and CO on Au (2115, 1932 cm-1).36,37 Prior to the SFG study being carried out a CV was recorded to check for correct operation of the spectroelectrochemical cell, leading to some CO being present from CO2 reduction. By ca. -1.9 V we see a loss of the 1907 cm-1 feature of [Mo(bpy)(CO)4] and growth of a new band at lower frequency (ca. 1880 cm-1). At <-1.9 V an in-crease in intensity of the SFG mode of CO at Au (ca. 2080-2100 cm-1) is also observed which is as-signed to be due to a combination of CO loss from the Mo complex and to CO produced via CO2 re-duction, which onsets at this potential. It is feas-ible that the feature at 1880 cm-1 is due to a CO2 reduction intermediate, however to date we have been unable to detect the stretching modes that would be expected for bound CO2 species indicat-ing that such species are either short-lived or have a low residence time near the electrode sur-face. Therefore based on the similarity of the fre-quency of the mode to that observed under argon we assign the 1880 cm-1 feature to [Mo(bpy)(CO)4].-, Figure 5, S9.
In contrast to experiments under Ar, under CO2 we do not observe the presence of either [Mo(bpy)(CO)3]2- (1835 and 1735 cm-1 under Ar) or [Mo(bpy-H)(CO)4]- (1870, 1790 cm-1 under Ar), even at the most negative potentials studied (-2.35 V, Figure 5b). The smaller potential window studied under CO2 compared to under Ar is due to the large current density observed at these neg-ative potentials and the subsequent formation of bubbles, which could not be displaced from the thin pathlength (ca. 50 μm) of the SEC cell. The lack of observation of [Mo(bpy)(CO)3]2- is signific-ant, indicating that it is able to rapidly react with CO2, whilst the absence of spectral features as-signable to [Mo(bpy-H)(CO)4]- supports its pro-posed mechanism of formation via reaction of [Mo(bpy)(CO)3]2- with water in the absence of CO2.
2100 2000 1900 1800 1700 1600
-2.18
-2.18
-2.93
-1.43
VFc
/Fc+
Wavenumber / cm-1
-5.000
7.500
20.00
32.50
45.00
57.50
70.00
82.50
95.00
-1.43
SFG intensity
Figure 6: In-situ IR-Vis SFG spectra of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1M TBAPF6 recorded under an Ar atmosphere at a Pt electrode during a CV measurement (50 mV s-1) averaging for 1 s per spectrum using a delay of 0.4 ps to suppress NR sig-nals.
For the remainder of the paper we will focus on elucidating the factors that enable the formation of [Mo(bpy)(CO)3]2- on the Au surface at such pos-itive potentials. Control VSFG studies of a [Mo(bpy)(CO)4] solution at a Pt electrode are shown in figure 6. At the start of the CV measure-ment (-1.43 V, Fig S10,11) VSFG bands associated with [Mo(bpy)(CO)4] are observed at 1898 and 1872 cm-1. As [Mo(bpy)(CO)4] is reduced (-1.9 to -2.3 V) the SFG intensity at 1898 cm-1 decreases, leaving a single υ(CO) mode at 1867 cm-1 (-2.4 V, figure S12). Assignment on the basis of only a single vibrational mode is inevitably tentative, however it is reasonable to anticipate the forma-tion of [Mo(bpy)(CO)4].- at these potentials on Pt, the observed wavenumber being in-line with ex-pectations for this species.18 At potentials negat-ive of -2.6 V we see a decrease in the intensity of the υ(CO) mode of [Mo(bpy)(CO)4].- . It is notable that when a Pt electrode was used, no new SFG features that could be assigned to either [Mo(bpy)(CO)3]2- or [Mo(bpy-H)(CO)3]- were ob-served. Previous spectroelectrochemical studies
have shown that [Mo(bpy)(CO)3]2- is generated at a Pt electrode and that it has reasonable stabil-ity.18 The lack of observation of the active catalyst in this surface specific study is likely either due to (i) a rapid movement of [Mo(bpy)(CO)3]2- away from the Pt electrode following formation or (ii) a high degree of disorder of surface [Mo(bpy)(CO)3]2- leading to weaker VSFG signals.
-2.1 -2.2 -2.3 -2.4 -2.5 -2.6
1864
1866
1868
1870
1872
1874
1876
1878
1880
2.7 +/- 1.0 cm-1
Pt Au
Wav
enum
ber /
cm
-1
Potential (V vs. Fc/Fc+)
14.0 +/- 4.0 cm-1
Figure 7: Stark shift of a υ(CO) mode of [Mo(bpy)(CO)4].- measured during VSFG/CV measurements of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1M TBAPF6 recorded under an Ar atmosphere at a Pt (red) and Au (black) electrodes. For experiments using Au electrodes a shift of 13.3 ± 1.0 cm-1 V-1 (dashed fit line) can be obtained over the full potential window where the band is present (-2.1 to -2.6 V), however at potentials negative of -2.3 V [Mo(bpy)(CO)3]2- is formed and hence small quantities of [Mo(bpy-H)(CO)4]-, a complex with similar IR modes to [Mo(bpy)(CO)4].-, may be present which could lead to inac-curacies in the measured shift over this larger po-tential range.
The vibrational Stark effect provides a probe of the local electric field strength experienced by the complexes, hence insights into the double layer structure and the degree of electrode-com-plex interaction can be gained through its meas-urement.38,39 Figure 2 shows a strong potential de-pendence, with the υ(CO) modes of the Mo com-plexes at the gold electrode shifting as a function of applied potential, and therefore electric field strength. We focus on the analysis of the modes of [Mo(bpy)(CO)4].- as it will be shown below that the interaction between this complex and the electrode is critical. Using a Au electrode we find that the B1 out of phase axial stretching mode of [Mo(bpy)(CO)4].- shifts by 14.0 ± 4.0 cm-1 V-1
between -2.1 and -2.3 V, figure 7. In contrast the υ(CO) modes of species at the Pt electrode show minimal potential dependence (Figure 6) and we only measure a Stark shift of 2.7 ± 1.0 cm-1 V-1 for [Mo(bpy)(CO)4].- at Pt, figure 7. Changes in geo-metry of the complex at the two different elec-trodes would give rise to different Stark shifts, however it is notable that the spectra of [Mo(bpy)
(CO)4].- on both Pt and Au are similar, with the B1 the υ(CO) mode dominating. Instead the results can be rationalised through consideration of the electrode/electrolyte interface. The simplest model of the interface would assume a linear po-tential drop between the electrode surface and the outer Helmholtz layer, which is typically over a distance of ca. 1 nm or less.40 Assuming a 1 nm Helmholtz layer we arrive at estimated Stark tun-ing rate of ca. 1.4 x10-6 cm-1/(V/cm) for the υ(CO) mode of [Mo(bpy)(CO)4].- at an Au electrode, in line with typical values for small molecules at in-terfaces.38,41,42 In reality the microscopic structure of the electrical double layer is still debated and the Helmholtz model is undoubtedly an oversim-plification. Of those proposed the Gouy-Chapman-Stern model has been most widely applied to suc-cessfully predict the decrease in electric field strength with distance from the electrode sur-face.40,43 Here a compact Stern layer (< 1 nm) ex-ists consisting of solvated electrolyte ions and specifically adsorbed molecular species, over which the majority of the potential drop occurs and the magnitude of the Stark tuning rate indic-ates that [Mo(bpy)(CO)4].- is either very close to the Au surface or even specifically adsorbed. Not-ably we observe a large change in the NR signal following the reduction of [Mo(bpy)(CO)4] to [Mo(bpy)(CO)4].- in-line with a large reorganisation occurring of the interfacial structure at this poten-tial (Figure S13).28 Beyond the Stern layer lies an outer diffuse region within which molecules will experience a significantly decreased electric field due to screening effects. It is likely that with a Pt electrode [Mo(bpy)(CO)4].- persists at a greater distance from the electrode surface, being rapidly moved away from the electrode following forma-tion.
-4
-2
0
2
(b)
i (A
)
(a)
CC
N
H
HH
-2.01-2.01-2.01-2.01
SFG
inte
nsity
(a.u
.)
MoCO
CO
CO
CON
N
-2.76-2.76 -1.26 -1.26
V (vs. Fc/Fc+)
MoCO
CO
CO
CON
N
-1.26
Figure 8: (a) CV measurement (20 mV s-1) of [Mo(bpy)(CO)4] (1 mM) in CH3CN and 0.1 M TBAPF6 recorded under an Ar atmosphere at a Au electrode; (b) SFG intensity vs. potentials at 1904 (green), 1850 (blue) and 2242 cm-1 (red). Vibrational modes at these frequencies are assigned to the species in-dicated in the figure. The yellow box is a guide to the eye to highlight the region where [Mo(bpy)(CO)4].- is the dominant complex present.
To further interrogate the gold electrolyte inter-face and to provide evidence for the proposed surface interactions with the Au electrode, we also monitored VSFG signals relating to the solvent (CH3CN) during CV measurements. The intensity of the SFG signals at 2242, 1850 and 1905 cm -1 assigned to CH3CN, [Mo(bpy)(CO)4].-
and [Mo(bpy)(CO)4] respectively are shown in fig-ure 8. At 2242 cm-1 the absorbance of the bulk solvent is 0.55 (Figure S14) demonstrating that sufficient IR intensity can reach the electrode sur-face. We believe that the band at 2242 cm-1 is a VSFG mode and is unlikely to be a phantom trans-ition28 as the observed signal intensity does not mirror that of the NR signal (Figure S13) beyond the initial large change when [Mo(bpy)(CO)4].- is formed (ca. -2V). In contrast we find a clear in-verse correlation exists between the intensity of the SFG modes assigned to CH3CN and [Mo(bpy)(CO)4].- at the Au surface. This is in-line with a very close interaction between [Mo(bpy)(CO)4].-
and the Au electrode. The presence of [Mo(bpy)(CO)4].- leads to a reorganisation, disordering and potentially the expulsion of the solvent from the electrode surface decreasing the VSFG response at 2242 cm-1. No clear inverse correlations with the VSFG signal of CH3CN and [Mo(bpy)(CO)3]2-/[Mo(bpy-H)(CO)3]- have been identified. When [Mo(bpy)(CO)3]2-/[Mo(bpy-H)(CO)3]- are the domin-ant species in the electrochemical experiment (<-2.5 V) the VSFG signal of CH3CN increases, sug-gesting that reduction of [Mo(bpy)(CO)4].- leads to reforming of the structured solvent layer. Simil-arly when [Mo(bpy)(CO)4] is the dominant species (from -1.25 to -1.65 V) we observe a maximum in the VSFG response of CH3CN. It is therefore ap-parent that the ability to change the nature of the solvent structure and potentially displace solvent from the gold electrode surface is specific to [Mo(bpy)(CO)4].-.
It may be envisaged that the strength of the electrode-CO bond would be the critical factor in enabling CO loss from [Mo(bpy)(CO)4].- and sub-sequent reduction to form catalytically active [Mo(bpy)(CO)3]2-. However whilst we did observe the build-up of CO on Pt electrodes during experi-ments which could only be removed through oxid-ative stripping at anodic potentials, indicating a strong Pt-CO interaction in this system, the form-ation of [Mo(bpy)(CO)3]2- via the 2nd low potential pathway (scheme in figure 2) was not observed. Indeed a similarly low catalytic behaviour is
achieved with GCE (figure S1), a material with minimal CO affinity, suggesting that the proximity of [Mo(bpy)(CO)4].- to the electrode surface is in-stead the critical factor, in-line with our findings. The degree of interaction of an electrochemically generated species with the electrode is a complex balance that is dependent upon multiple (and in-terrelated) factors including the electrolyte salt choice and concentration, the nature of the solvent and substrate/product interactions with the electrode. This offers several simple routes to improve catalytic activities with these group 6 CO2 reduction catalysts. The role of the solvent has already been demonstrated, with the use of NMP greatly enhancing CO2 reduction catalytic currents compared to THF and CH3CN, despite the increased viscosity of NMP18 and VSFG studies to elucidate the nature of a wider range of the NMP-catalyst-electrode interactions are underway. Therefore it appears viable that careful control of solvent and supporting electrolyte will deliver en-hanced catalytic activity for this class of com-plexes at a greater range of electrode materials in the future.
CONCLUSIONS: The VSFG studies presented here provide evidence for the previously pro-posed18 formation of [Mo(bpy)(CO)3]2-, the active catalyst for CO2 reduction, via a 2nd pathway at more positive applied potentials, enabling cata-lytic onset potentials that are similar to the much more widely studied group 7 analogues. Our stud-ies show that [Mo(bpy)(CO)4].- has a strong inter-action with the Au electrode and this is critical in enabling CO loss from this relatively stable spe-cies in solution. The clear demonstration of a low energy pathway to forming active catalytic spe-cies is an important step, indicating that simple strategies such as electrolyte control, alongside synthetic strategies that encourage facile CO loss, are likely to be productive routes to develop-ing more efficient electrocatalysts for CO2 reduc-tion based on the comparatively abundant group 6 metals.
ASSOCIATED CONTENT Supporting Information. Details of the spectro-electrochemical apparatus, VSFG experiment, syn-thesis of [Mo(bpy)(CO)4], supporting CV experiments and support VSFG analysis are available in the sup-porting information. This material is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATIONCorresponding Author*[email protected], *[email protected] Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.
ACKNOWLEDGEMENTSAJC and GN acknowledge funding from the EPSRC (EP/K006851/1, EP/N010531/1). Prof. L. Hardwick (University of Liverpool) is acknowledged for valu-able discussions. The work was carried out at the Central Laser Facility (UK STFC Rutherford-Appleton Laboratory, Ultra, experiment numbers: 15130005, 16130016).
REFERENCES(1) Costentin, C.; Robert, M.; Savéant, J.-M. Chem.
Soc. Rev. 2013, 42, 2423–2436.(2) Khezri, B.; Fisher, A. C.; Pumera, M. J. Mater.
Chem. A 2017, 5, 8230–8246.(3) Takeda, H.; Cometto, C.; Ishitani, O.; Robert, M.
ACS Catal. 2017, 7, 70–88.(4) Bourrez, M.; Molton, F.; Chardon-Noblat, S.;
Deronzier, A. Angew. Chem. Int. Ed Engl. 2011, 50, 9903–9906.
(5) Sampson, M. D.; Kubiak, C. P. J. Am. Chem. Soc. 2016, 138, 1386–1393.
(6) Hawecker, J.; Lehn, J.-M.; Ziessel, R. J. Chem. Soc. Chem. Commun. 1984, 6, 328–330.
(7) Smieja, J. M.; Sampson, M. D.; Grice, K. a; Ben-son, E. E.; Froehlich, J. D.; Kubiak, C. P. Inorg. Chem. 2013, 52, 2484–2491.
(8) Costentin, C.; Robert, M.; Savéant, J.-M. Acc. Chem. Res. 2015, 48, 2996–3006.
(9) Johnson, B. A.; Maji, S.; Agarwala, H.; White, T. A.; Mijangos, E.; Ott, S. Angew. Chem. Int. Ed. 2016, 55, 1825–1829.
(10) Kang, P.; Zhang, S.; Meyer, T. J.; Brookhart, M. Angew. Chem. Int. Ed. 2014, 53, 8709–8713.
(11) Beley, M.; Collin, J.-P.; Ruppert, R.; Sauvage, J.-P. J. Chem. Soc. Chem. Commun. 1984, 19, 1315–1316.
(12) Neri, G.; Aldous, I. M.; Walsh, J. J.; Hardwick, L. J.; Cowan, A. J. Chem. Sci. 2016, 7, 1521–1526.
(13) Shen, J.; Kortlever, R.; Kas, R.; Birdja, Y. Y.; Diaz-Morales, O.; Kwon, Y.; Ledezma-Yanez, I.; Schouten, K. J. P.; Mul, G.; Koper, M. T. M. Nat. Commun. 2015, 6, 8177.
(14) Walsh, J. J.; Neri, G.; Smith, C. L.; Cowan, A. J. Chem. Commun. 2014, 50, 12698–12701.
(15) Grice, K. A. Coord. Chem. Rev. 2017, 336, 78–95.
(16) Reda, T.; Plugge, C. M.; Abram, N. J.; Hirst, J. Proc. Natl. Acad. Sci. 2008, 105, 10654–10658.
(17) L. Clark, M.; A. Grice, K.; E. Moore, C.; L. Rhein-gold, A.; P. Kubiak, C. Chem. Sci. 2014, 5, 1894–1900.
(18) Tory, J.; Setterfield-Price, B.; Dryfe, R. A. W.; Hartl, F. ChemElectroChem 2015, 2, 213–217.
(19) Franco, F.; Cometto, C.; Sordello, F.; Minero, C.; Nencini, L.; Fiedler, J.; Gobetto, R.; Nervi, C. ChemElec-troChem 2015, 2, 1372–1379.
(20) Grice, K. A.; Saucedo, C. Inorg. Chem. 2016, 55, 6240–6246.
(21) Sieh, D.; Lacy, D. C.; Peters, J. C.; Kubiak, C. P. Chem. – Eur. J. 2015, 21, 8497–8503.
(22) Smieja, J. M.; Kubiak, C. P. Inorg. Chem. 2010, 49, 9283–9289.
(23) Shen, Y. R. Nature 1989, 337, 519–525.
(24) Shen, Y. R. J. Phys. Chem. C 2012, 116, 15505–15509.
(25) Wang, H.-F. Prog. Surf. Sci. 2016, 91, 155–182.(26) Rosen, B. A.; Haan, J. L.; Mukherjee, P.; Braun-
schweig, B.; Zhu, W.; Salehi-Khojin, A.; Dlott, D. D.; Masel, R. I. J. Phys. Chem. C 2012, 116, 15307–15312.
(27) García Rey, N.; Dlott, D. D. J. Phys. Chem. C 2015, 119, 20892–20899.
(28) Braunschweig, B.; Mukherjee, P.; Haan, J. L.; Dlott, D. D. J. Electroanal. Chem. 2017, 800, 114-125.
(29) Anfuso, C. L.; Snoeberger, R. C.; Ricks, A. M.; Liu, W.; Xiao, D.; Batista, V. S.; Lian, T. J. Am. Chem. Soc. 2011, 133, 6922–6925.
(30) Wang, J.; Clark, M. L.; Li, Y.; Kaslan, C. L.; Ku-biak, C. P.; Xiong, W. J. Phys. Chem. Lett. 2015, 6, 4204–4209.
(31) Anfuso, C. L.; Xiao, D.; Ricks, A. M.; Negre, C. F. A.; Batista, V. S.; Lian, T. J. Phys. Chem. C 2012, 116, 24107–24114.
(32) Clark, M. L.; Rudshteyn, B.; Ge, A.; Chabolla, S. A.; Machan, C. W.; Psciuk, B. T.; Song, J.; Canzi, G.; Lian, T.; Batista, V. S.; Kubiak, C. P. J. Phys. Chem. C 2016, 120, 1657–1665.
(33) Calabrese, C.; Vanselous, H.; Petersen, P. B. J. Phys. Chem. C 2016, 120, 1515–1522.
(34) Lagutchev, A.; Lozano, A.; Mukherjee, P.; Ham-bir, S. A.; Dlott, D. D. Spectrochim. Acta. A. Mol. Biomol. Spectrosc. 2010, 75, 1289–1296.
(35) Antonín Vlček, J.; Baumann, F.; Kaim, W.; Grev-els, F.-W.; Hartl, F. J. Chem. Soc. Dalton Trans. 1998, 2, 215–220.
(36) Chen, D.-J.; Allison, T. C.; Tong, Y. J. J. Phys. Chem. C 2016, 120, 16132–16139.
(37) Chang, S. C.; Hamelin, A.; Weaver, M. J. J. Phys. Chem. 1991, 95, 5560–5567.
(38) Baldelli, S. J. Phys. Chem. B 2005, 109, 13049–13051.
(39) Bishop, D. M. J. Chem. Phys. 1993, 98, 3179–3184.
(40) Bard, A. J.; Faulkner, L.R. Electrochemical Meth-ods: Fundamentals and Applications, 2nd Edition; Wiley, New York, 2001.
(41) Fried, S. D.; Boxer, S. G. Acc. Chem. Res. 2015, 48, 998–1006.
(42) Brewer, S. H.; Franzen, S. J. Chem. Phys. 2003, 119, 851–858.
(43) Brown, M. A.; Goel, A.; Abbas, Z. Angew. Chem. Int. Ed. 2016, 55, 3790–3794.
For Table of contents only
-
Mo
Mo
Electrod
e
φ1
Potenti
al
Distance
---
--
--
--
--
--
------
φ2
-
-
-
10