Langmuir 1992,8,1845-1850
1845
Identification of Surface Films Formed on Active Metals and Nonactive Metal Electrodes at Low Potentials in Methyl Formate Solutionst Yair Ein Ely and Doron Aurbach* Department of Chemistry, Bar-Ilan University, Ramat Gan 52100, Israel Received November 5,1991. In Final Form: January 10, 1992
The electrochemical behavior of methyl formate (MF) and MFICOdLi salt solutions with active metals (lithium,calcium)and nonactive metals (nickel,gold) was investigated. The salts included calcium,lithium, and tetrabutylammonium perchlorates and tetrafluoroborates, and LiAsFe. Surface-sensitive Fourier transform infrared (FTIR) spectroscopy and X-ray microanalysis were used in order to characterize the variouselectrode surfaces,in addition to cyclicvoltammetry. It was found that the metal formates (HCOOLi or (HCOO)zCa)are the major surface species formed both on the active and on the nonactive electrodes at potentials below 1.5 V (Li/Li+)and that they passivate them. When COZand Li salt are present in MF solutions, films formed on both active and nonactive electrodes at low potentials contain both lithium formate and lithium carbonate.
Introduction There is an increasing interest in methyl formate (MF) as a solvent or cosolvent for secondary lithium batteries. In a recent report related to Li-LiCoCz rechargeable cells, MF/COz solution is described as a superior electrolyte for this system.' The conductivity of MF/LiAsFG solutions is around 30-50 m W c m and higher by an order of magnitude compared to propylene carbonate or ethereal solutions. Other reports related to Li-vanadium oxide cells also mention MF solutions as a good electrolyte for these systemse2 The compatibility of MF for secondary Li systems is surprising since esters are supposed to be very reactive with lithium. Recent studies with the cyclic ester y-butyrolactone (BL) revealed that this solvent readily reacts with lithium to form carboxylate species? and therefore very poor Li cycling efficiency was obtained with this s01vent.~In any event, it was also found that, in spite of the high reactivity of BL toward lithium, Li cycling efficiency in this solvent was strongly affected by the presence of trace additives such as reactive gases.4 Indeed, as previously reported,l the presence of CO2 remarkably improves cycling efficiency of Li anodes in MF solutions. However, it is very unclear how the presence of COZinfluences Li cycling efficiency in a solvent such as MF which is supposed to be very reactive with lithium. It is generally accepted that Li cycling efficiency depends mostly on the Li surface chemistry in solution^.^^ C02 reduction mechanisms in nonaqueous systems are not fully ~nderstood,~ and it is not at all clear if COZdissolved in MF is able to affect Li surface chemistry in solutions.
* To whom correspondence should be addressed.
+ This paper was originally presented at the Fall 1991 Meeting of
the Electrochemical Society held in Phoeniz, AZ (pv92-15pp 145156, 1992). (1)Plichta, E.; S h e , S.;Uchiyama, M.; Salomon, M.; Chua, D.; Ebner, W. B.; Lin, H. W. J. Electrochem. SOC.1989, 136, 1865. (2) Uchiyama, M.; S h e , S.; Plichta, E.; Salomon,M. J . Power Sources
The present study aims to investigate the electrochemical behavior of MF and MF/COz solutions with active (lithium, calcium) and nonactive metal (gold, platinum, nickel) electrodes. As it is generally accepted that the electrochemistry of lithium and probably other active metals in aprotic solvents is surface film controlled, a special effort was made to understand the surface chemistry of lithium, calcium, and nonactive metals at low potentials in MF solutions, using Fourier transform infrared (FTIR) spectroscopy and X-ray microanalysis. In addition, the voltammetric behavior of MF and MF/ COZ solutions was also studied using nonactive metal electrodes and was compared to that of other polar aprotic systems previously s t ~ d i e d It . ~was ~ ~interesting to find to what extent the voltammetric behavior of MF solutions with noble metals is surface film controlled and whether similar films formed on active metals are also formed on nonactive metals at low potentials. The effect of the COZ pressure on the electrodes surface chemistry was also studied.
Experimental Section MF (Aldrich,anhydrous)was distilled twice over Pz05 under argon at room temperature. LiClO4 (Aldrich) was used after dehydration and drying in vacuum (150 "C, 3 days); LiAsF6 (Lithco)and LiBF4 (Tomiyama)were used as received. All the preparation for the spectroscopic studies as well as the electrochemical measurements were carried out under high purity argon atmosphere in gloveboxes. The glovebox operation and electrochemical and spectroscopicinstrumentation (includingtransfer systems) are described elsewhere.3,4,6,8,10,11FTIR measurements were carried out ex situ, using mirrorlike lithium, calcium, and nonactive metal surfaces prepared as described previously.8JoJ1 Lithium or calcium mirrors were stored for predetermined periods of time in solutionsfollowed by washing (pure solvent) and drying (vacuum). Then, they were covered with KBr plates as described already.1° The protected surfaces were studied by FTIR, external reflectance mode at a grazing angle,
1987, 20, 279. (3) Aurbach, D. J. Electrochem. SOC.1989, 136, 1606. (4) Aurbach, D.; Gofer, Y.; Langzam, Y. J. Electrochem. SOC.1989,
using the appropriate accessories.3 For the study of MF/COz solutions, mirrorlike lithium and nonactive metal (gold, nickel) electrodes were studied in three
Interfacial Electrochem. 1990, 282, 73. (7) Taniguchi, I. In Modern Aspects of Electrochemistry; Bockris, J. OM., White, R. E., Conway, B. E., Eds.; Plenum Press: New York and London, 1991; Vol. 20, pp 327-344.
(8)Aurbach, D.; Daroux, M. L.; Faguy, P.; Yeager, E. J . Eleetroanal. Chem. Interfacial Electrochem. 1991,297, 225. (9) Aurbach, D.; Gottlieb, H. E. Electrochim. Acta 1989, 34, 141. (10) Aurbach, D.;Daroux, M.;Faguy,P.; Yeager, E. B. J.Electrochem. SOC.1987,134, 1611. (11)Aurbach, D.; Skaletaky, R.; Gofer, Y. J . Electrochem. SOC.1991, 138, 3536.
136,3198. ( 5 ) Peled,E. InLithi~mBatteries;Gabano,J. P., Ed.; Academic Press: New York, 1983; Chapter 3. (6) Aurbach, D.; Malik, Y.; Meitav, A.; Dan, P. J . Electroanal. Chem.
Q743-7463/92/2408-1S~5~03.00/0 0 1992 American Chemical Society
1846 Langmuir, Vol. 8, No.7, 1992
Ely and Aurbach Scheme I
I
I
HCOOCH3
0
CH3’
-
e-
-
!
+
Li+ or Ca2+ H ‘or C H ~ * I-
II ! HCOCHJ
0
II
HCOCH3
HCOO-
+
0
II
HCO-
CH4
t
+
CH: HCOOLi 1
or C 2 H b f
V.
Li/MF pure
4003
3600
29;l
zq”27)(1
I
3200
2800
2400
2000
1600
1200
600
4
WAVENUMBER
Figure 1. FTIR spectra obtained ex situ from lithium surfaces stored in MF and MF solutions: (a) exposure to MF vapor for 1 h; (b) MF/LiAsF8 0.5 M (3 h); (c) pure MF (3 h).
I T
8z 8 (L
8 a
0
Ilooe3 0224
0.7‘
z
..U
b
c.0
I
Ca+ MF
c
Co /MF/Co
d
CO +MF/LIAsF,
(C104)2
UCH
,
2111 1121
WAVENUMBER
Figure 2. FTIR spectra obtained ex situ from calcium surfaces (protected with KBr plates) after being stored in pure MF and MF solutions: (a) exposure to MF vapor (4h); (b) MF/Ca(ClO4)2 0.5 M (3 days); (d) MF/LiAsFe 0.5 M (3 days). electrode cells placed in aluminum pressure vessels. These were loaded with electrochemical cells (including MF solutions) in the glovebox and were then pressurized with COz (ultrahigh purity, Oxygen Center, Israel) to different pressures. In a typical experiment, a nickel mirrorlike electrode (area of 8-10 cm2)was polarized in solutions (using the appropriate cell in the required atmosphere) from open circuit voltage (OCV)to a predetermined lower potential. The electrode was held at that potential for 15-30 min. Then, the cell was disconnected from the potentiostat and the electrode was removed, washed (pure solvent), vacuum dried, and measured spectroscopically (ex situ FTIR, external reflectance mode, at a grazing angle) as described before.8J2 Cyclic voltammetry of MF solutions was measured using gold wire electrodes in three electrode glass cells (lithium counter and reference electrodes). The gold wire (0.5 mm in diameter and 10 mm length) was embedded in glass on both sides in order to avoid edge effects. X-ray microanalysis of lithium and calcium samples was obtained with the eXL system (Link, England) attached to a JEOL 840-JSM electron microscope, using a transfer technique which was previously de~cribed.~
Results and Discussion a. MF Solutions. Figures 1and 2 show FTIR spectra obtained from lithium and calcium electrodes (respectively) exposed to various MF environments. The spectra
of Figures l a and 2a were obtained from samples exposed to MF vapor, and the other spectra of Figures 1and 2 were obtained from samples stored in MF solution for a few hours as indicated. All the spectra of Figures 1and 2 have peaks around 2970,2860,2730cm-l (CH stretching), 1620 ( C 4 stretching), and 1380and 790 cm-l (COO- bending) which are typical of formate anion (HC00-).12 This proves that both lithium and calcium have similar surface chemistry in MF solutions, producing metal formate as a major surface species. MF is probably reduced by the active metals, as outlined in Scheme I. However, as proved in Figures 1and 2, lithium or calcium formates are not the only surface species formed in these systems. Other peaks around 1460,1100-1000, and 700 cm-l appear in part of the spectra of Figures 1and 2. Their appearance depends on the experimental condition and the presence of salts in solutions. For instance, these additional peaks are present in the spectra of Figure la,c, obtained from Li treated in pure MF, but absent in the spectrum of Figure lb, obtained from Li stored in MF/LiAsFe solution, which is a clean lithium formate spectrum. It is interesting to note that while the spectra obtained from calcium surfaces exposed to MF vapor have typical CaC03 peaks (Figure 2a, as indicated), spectra obtained from calcium surfaces identically prepared and stored in MF are mostly calcium formate spectra (Figure 2b). CaC03 may be unavoidably present on calcium surfaces prepared in the glovebox due to the presence of trace 02 and COZ in the glovebox atmosphere.ll Therefore, both CaCO3 and calcium formate may be present on calcium surfaces exposed to MF vapor. These studies prove that, in solutions, the initial CaCO3 film is unstable and is readily substituted by the solvent reduction products. The aforementioned additional peaks appearing in the spectra of Figures 1and 2 (in addition to the metal formate peaks) may be attributed to alkoxy (LiOR or Ca(OR)2) species.12 Naturally, one would suspect lithium or calcium methoxide to be formed as well, due to a cleavage of MF radical anion to the HCO radical and CH30-. Figure 3a shows FTIR spectra obtained from LiOCH3 stored for a few days in MF (dried and pelletized with KBr). The spectrum is identical to that of the starting alkoxide. No lithium formate was formed, which proves that lithium methoxide is stable in MF. None of the LiOCH3 peaks appear in the spectra of Figures 1 and 2. The spectrum in Figure 3b, which was obtained from lithium surface exposed to methanol vapor followed by storage in MF, has both LiOCH3 and LiOOCH peaks. The spectrum in Figure 3c obtained from lithium surface stored in MF/MeOH 0.01 M solution has only the typical lithium formate peaks. These experiments obviously prove that methoxide is not formed in these systems and that if it was formed on lithium (12)Aurbach, D.; Youngman-Chusid, 0.; Gofer, Y.; Meitav, A.Electrochim. Acta 1990,35,625.
Surface Films in Methyl Formate Solutiom
I .
I
uc-0
vC H
lob9
261
r
Langmuir, Vol. 8, No. 7,1992 1847 Lithium electrodes stored in MF/COp ( 3 0 t m )
I
VCI-0
b LiAsFs 1M
I
C. loo43
LitMF/MeOH 005M VCH
JJdC211-') I
4000
3600
3200
2800
2400 2000 WAVENUMBER
1600
1200
800
400
Figure 3. (a) FTIR spectrum of LiOCH3 stored for a few days in MF (pelletized with KBr after drying); (b) FTIR spectrum obtained from lithium surface exposed to methanol vapor (5 min) followed by storage in MF (3 h); (c) FTIR spectrum obtained from Li surface stored for a few hours in MF/MeOH 0.05 M solution.
I
1
K)
3600
3200
2800
2400 2000 WAVENUMBER
1600
1200
800
400
Figure 4. (a)FTIR spectrum of KBr pelletized LiOH; (b) FTIR spectrum obtained from LiOH stored for a few days in MF (pelletizedwith KBr after drying); (c) FTIR spectrum obtained from lithium surface stored in wet MF (0.01 M HzO) for a few days; (d) FTIR spectrum of Ca(OH)*stored for a few days in MF (KBr pelletized after drying). in MF, it should be stable and detectable. Also, methanol, if present as a contaminant in MF, does not react with the active metal, probably because of the high reactivity of the solvent which prevents competingsurface reactions of contaminants. Similar experiments with similar results have been performed with calcium methoxide and calcium surfaces. Another contaminant that may be unavoidably present in MF is trace water which may react with active metals to form insoluble hydroxides. Parts a and b of Figure 4 show FTIR spectra of anhydrous LiOH and LiOH stored over MF, respectively. The solid formed by the reaction of MF and LiOH was vacuum dried after storage. (The spectra of this product and LiOH were obtained pelletized with KBr.) The spectrum in Figure 4b is typical of LiOOCH, which means that LiOH reacts completelywith MFto form lithium formate and methanol. The spectrum in Figure 4c obtained from lithium stored in MF/HzO 0.01 M solution is also typical of lithium formate. No LiOH is formed, as indicated by the absence of the typical 3675-cm-l LiOH peak. This proves that the surface chemistry of lithium in water-contaminated MF is similar to that of dry MF since LiOH, if formed, also readily reacts with MF to form lithium formate. In contrast to that, as proven by the
4000
3600
3200
2800
2400 2000 WAVENUMBER
1600
1200
800
400
Figure 5. Typical FTIR spectra obtained from lithium surfaces (protected with KBr plates), after being stored for 24 h in MF/ COz (3 atm) solutions containing Li salts. (a) 1M LiClO,; (b) 1 M LiAsF6. spectrum in Figure 4d, Ca(0H)z if formed on calcium in wet MF only partly reacts with the solvent. The calcium formate formed probably passivates the hydroxide and in this case avoids a complete reaction. Among many polar aprotic solvents previously studied, including ethers, BL, propylene carbonate, and ethylene carbonate, MF seems to be the most reactive one toward active metals. Lithium formate could be detected on Li surfaces exposed to MF vapor for a few minutes. Also, calcium formate could be detected on calcium surfaces stored for a few hours in MF. By comparison, in order to detect BL reduction products on calcium,ll storage of a few days was required. Other aprotic solvents such as propylene carbonate (PC) or ethers hardly react with calcium.ll The high reactivity of MF toward lithium and calcium is supposed to affect possible salt reduction; however, the FTIR spectra of Figures 1 and 2 discussed above prove that the presence of salt anions such as AsF6- or C104- influence the surface chemistry of the active metal in solutions as well. The presence of BF4salts affect MF reduction on lithium even more strongly (aswill be shown later). However, X-ray microanalysis of lithium or calcium samples stored in Clod-, AsFe-, or BFrsalt solutions could not detect fluorine, chlorine, arsenic, boron, or chlorine peaks. In contrast, similar samples stored in these salt solutions in PC, ethylene carbonate (EC),or ethers and measured identically show pronounced F or C1 peaks, which proves that salt reduction also contributes to the buildup of surface films on active metals in these solvent solutions. These results obviously provide further proof of the superior reactivity of MF as compared to other polar aprotic solvents. Thereby, salt reduction (which probably also occurs in MF solutions to some extent) is less involved in the surface chemistry of lithium in MF than in the other aforementioned solvents. b. MF/COz Solutions. The surface chemistry of lithium in MF/COz solutions at different COZpressures (1-6 atm) was investigated by ex situ FTIR spectroscopy of either mirrorlike lithium surfaces or polished nickel foils. It was found that the surface films formed on nickel electrodes polarized in solutions to potentials close to that of Li bulk deposition are similar to those formed on lithium but of a better quality. Nickel has the advantage over noble metals such as gold, platinum, or silver as it does not alloy with lithium. Hence, surface film formation on nickel at low potentials in these solutions is not interfered with by Li alloying or Li under potential deposition (which occurs at potentials above Li deposition on noble metal electrodes). Figure 5 shows FTIR spectra obtained from lithium electrodes stored for 24 h in MF/COz (3 atm), LiClOd
1848 Langmuir, Vol. 8, No. 7, 1992 and LiAsF6 solutions (a and b, respectively). Both spectra have lithium formate peaks (1640-1620, 1380, and 780 cm-l) and Li2CO3 peaks (1500-1450 and 880-865 cm-l 9J0). The spectrum in Figure 5a also has a pronounced band around 1100 cm-l which is attributed to a C1-0 bond (similar to that of LiC1049. This may be related to lithium perchlorate reduction products such as LiCl02 or LiC103 that should be much less soluble in MF and therefore precipitate on the surface. Formation of such chlorine species on lithium stored in LiC104/PC was indicated previously by X-ray photoelectron spectroscopic (XPS) studies.13 The spectrumin Figure 5b also has, in addition to lithium formate and LizCO3, peaks around 1700,1300,1150, and 900 cm-l. On the basis of other studies with carbon electrodes which enabled extraction of surface species in an amount suitable to NMR and MS studies, this spectrum is attributed to species formed by complicatedcondensation of C02 and MF reduction products containing alkoxy groups. These spectral studies show that the surface chemistry of lithium and noble metals at low potentials in MF/CO2 solutions depends on both potential and C02 pressure. As the C02 pressure is higher, the major surface species formed are lithium formate and Li2CO3. CO2 reduction in both aqueous and nonaqueous systems was, in recent years, intensively investigated. C02 reduction products such as HCOO- were identified in protic ~olvents.~ In aprotic solvents C02 reduction seems to be complicated and some reports list oxalate anion (C2Od2-),carbonate anion (Cos2-), and CO as ita possible p r ~ d u c t s . ~ The present study obviously proves that C02 dissolved in MF reacts in the presence of Li salts on both lithium and noble metals to form Li2CO3. The lithium formate detected also is obviously formed only by solvent reduction since C02 cannot be reduced to formate in the absence of protic substance^.^ When the salt is LiC104, Li2CO3 may also be formed due to a reaction between Liz0 and C02. Since it is known that LiC104 is reduced in aprotic media to LiC1,6J2J3Liz0 is an obvious coproduct. This oxide readily reacts with C02 to form lithium carbonate. However, the fact that a similar product is formed when the salt is LiAsF6 proves that C02 itself is also reduced in spite of the competition of the solvent reduction discussed above. On the basis of mechanisms proposed in the literature7 for C02 reduction in nonaqueous media, C02 is probably first reduced to C02*-, followed by further reaction steps in which another C02'- moiety and maybe another CO2 molecule are involved. The overall reaction may be regarded as a disproportionation of two C02'- species to which precipitates in the presence of Li cations form co32-, as Li2CO3 and CO gas. Hence, the improvement of Li cyclingefficiency obtained in MF-based electrolyte due to addition of CO2l should be attributed to a better passivation of the lithium electrodes due to Li2CO3 formation. It should be noted that enhancement of Li2CO3 formation over organic surface species on Li electrodes in PC-based electrolyte solutions (for instance, by an addition of trace H2O to PC solutions) also increases Li cycling efficiency in these systems.4 113) Garreau, M.;Thevenin, J.; Warin, D. B o g . Batteries Sol. Cells 1979,2,54.Froment, M.;Garreau, M.; Thevenin, J.; Warin, D. J. Mic r o s ~Spectrosc. . Electron. 1979,4 , 111, 483. (14)Youngman-Chusid, 0.; Babai, M.; Ein Ely, Y.; Aurbach, D. The electrochemical behavior of carbon electrodes in polar aprotic systems containing Li salts; in preparation.
Ely and Aurbach
, 4000
3600
3200
2800
2400
2000
1600
I200
800
4 )0
WAVE NUMBER
Figure 6. FTIR spectra obtained ex situ from nickel electrodes polarized to 0.05 V (Li/Li+)in various LiBF4/MF solutions: (a) LiBF4 1 M/MF; (b) LiBF4 1 M/MF/C02 (3 atm); (c) LiAsF6 0.5 M/LiBF40.5 M/MF; (d) LiAsF6 0.5 M/LiBF4 0.5 M/MF/C02 (3 atm). The electrodes were held at this potential for 30 min followed by washing (pure solvent) and drying (vacuum).
However, the best MF-based electrolyte system for rechargeable Li batteries was reported to be MF/LiAsFe/ LiBFdC02 solution.' Therefore, special attention was given to this system. Figure 6 shows FTIR spectra obtained from nickel electrodes polarized in MF/LiBFdC02 and MF/LiBF$ LiAsFdCO2 solutions to low potentials (0.05 V, Li/Li+). Spectra obtained from samplestreated in similar solutions containing no C02 are also present for a comparison (spectra obtained from lithium samples stored in these solutions were quite similar but of a worse quality). Comparing these spectra to those of Figures 1-5 shows that the presence of LiBF4 in solutions affects the surface chemistry remarkably. Except for the spectrum of Figure 6c obtained from LiAsFs/LiBFd solution, none of the spectra have the clear, high-resolution lithium formate and/or Li2CO3 peaks appearing in the previous figures. The pronounced peak around 1100 cm-l appearing in the spectra of Figure 6a,d may be attributed to the B-F bond, as the LiBF4 spectrum has a similar peak. The 1620-cm-1 peak appearing in all the spectra of Figure 6 probably proves that lithium formate is formed but it is obviously not the major surface species. Hence, one would conclude that the presence of BF4changes the surface chemistry lithium electrodes, perhaps due to BF4- reduction to form insoluble products which block the surface and suppress the aforementionedsurface reactions in solutions containing no LiBF4. Hence, a new combination of surface films is obtained which occasionally leads to better passivation of the lithium electrodes and probably a more homogeneous Li deposition and dissolution, which may explain the high performance of this electrolyte system.' c. Voltammetric Behavior of MF Solutions. Since MF solutions were found to be more reactive toward lithium and calcium than other polar aprotic solvents previously studied,*lO it was interesting to compare their voltammetric behavior with noble metal electrodes to that of other polar aprotic systems previously s t ~ d i e d . ~ ~ ~ Figures 7 and 8 show typical first cycle voltammograms obtained with gold electrodes and MF/LiC104 and MF/ LiAsF6 solutions, respectively (dashed lines). The effect of COz atmosphere (3 atm) is also shown in these figures
Langmuir, Vol. 8, No. 7, 1992 1849
Surface Film in Methyl Formate Solutions Au Electrode L1C104 lM/MF
MF/Ca ( C 1 0 4 ) 2 A u Electrode 20mv/sec
I
,
T
ocv I
3
2 I
I
1
4 I
t
5v
1
Figure 7. Typical first cyclevoltammogramsobtainedwith gold electrodesand MF/LiClO, solutions(20mV/s): dashed line,argon atmosphere; solid line, COz atmosphere (3 atm).
Figure 8. Typical first cycle voltammogramsobtainedwith gold electrodesand MF/LAFe solutions(20mV/s): dashed line,argon atmosphere; solid line, COz atmosphere (3 atm).
O C V 4 1 6 V Lt/LI*
4000
3600
3200
2800
2400
2000
I600
1200
800
V Ca/Cat+
Lkj
400
WAVENUMBER
Figure 9. FTIR spectra obtained ex situ from gold electrodes treated in MF/LiAsFe 0.5 M solution. The potential was swept from open circuit voltage (OCV) to predetermined potentials as indicated in spectra a-d in the figure. The electrodes were held at this potential for 15 min, followed by washing (pure solvent) and drying (vacuum). (solid lines). In both cases, the currents measured during the cathodic sweep at low potentials are at least 2-fold higher than those measured with other polar aprotic solvents containing Li salts in similar experiments.699 However, these currents become much smaller in consecutive sweeps, which means that the electrodes become passivated due to precipitation of reaction products. The typical peak of lithium UPD on gold and its related anodic stripping peak which characterizes voltammograms of other nonaqueoussystems containing Li salts618arescarcely visible in the voltammograms of Figures 7 and 8. The peak around 1.25 V appearing in both figures is attributed to water reaction, as addition of water to the solutions leads to its increase. Massive solvent oxidation occurs at potentials above 5 V and the oxidation products do not passivate the electrodes. The anodic peaks between 4 and5 V are related to an oxidation of surface films formed during the cathodic
-2
I
0
I
2
3
Figure 10. Typicalvoltammogramsobtainedwith MF/Ca(ClO& 0.5 M solutions and gold electrodes (20 mV/s). The solid line and the dashed line are related to first and second cycles, respectively. sweep as proven by the spectroscopic results to be presented below. Figure 9 shows FTIR spectra obtained ex situ from gold electrodes treated at different potentials in MF/LiAsF6 solutions. In each experiment the potential was swept from open circuit voltage (OCV) (-3 V, Li/ Li+)toa predetermined low potential (followedby washing with pure MF and drying prior to the measurement). Spectrum a related to 1.6 V is attributed to either traces of partly hydrated salt or salt reduction products as the peak around 720 cm-l is related to the As-F bond and the peaks around 3600-3500 and 1630cm-l are typical of trace water hydrating Li salts. However, as the potential becomes lower, typical lithium formate peaks characterize the spectra (Figure 9b,c), and their height increases as potential decreases. As proven by the spectrum in Figure 9d, measured after the potential was swept from OCV to 0.1 V and then to 4.5 V, the lithium formate surface species are oxidized around 4.5 V. Similar spectral results were obtained from electrodes treated in LiC104/MF. Figure 10 shows voltammograms measured with gold electrodes treated in MF/Ca(C10& solution. In this case the reference electrode was a calcium foil in solution, which is about 0.7 V in the Li/Li+ scale. FTIR measurements of electrodes treated in this system at different potentials seem to indicate that the cathodic currents at potentials below 1 V (Ca/Ca2+)are related to solvent reduction to calcium formate that precipitates on the electrode surface and partly passivates it. Hence, the surface chemistry of noble metal electrodes at low potentials in MF solutions of Ca and Li salts is quite similar to that of lithium or calcium in these systems. The solvent is reduced at potentials below 1.5 V (Li/Li+) to calcium or lithium formate which passivate the electrodes due to precipitation. Lithium can be easily deposited through the lithium formate films. In contrast, calcium cannot be deposited, probably due to the poor mobility of the bivalent Ca2+through the surface films. As shown in Figures 7 and 8, the presence of C02 changes the voltammetricbehavior. In the case of LiClO~solutions, lower currents are measured at low potentials (compared to solutions under argon atmosphere). When the salt is LaSF6, a pronounced irreversible reduction wave is measured at potentials below 0.5 V. In order to understand the voltammetric behavior of MF/C02 solutions with these two Li salts, the potential-dependent surface chemistry of the electrodes was measured using ex situ FTIR (external reflectance mode). Figures 11and 12 present FTIR spectra obtained from nickel electrodes treated at various potentials in MF/C02 (3 atm) solutions of LiAsF6 and LiC104, respectively.
Ely and Aurbach
1850 Langmuir, Vol. 8, No. 7, 1992 Nickel Electrodes Treated in LIAsF,/MF /coz ( 3 a t m )
4000
3600 3200
2800
2400
2000
1600
1200
800
4
WAVE NUM BER
Figure 11. FTIR spectraobtainedex situ from nickel electrodes polarized in LiAsF8 1 M/MF/COz (3 atm) solutions: (a-d) the potential was stepped from open circuit voltage (OCV) (-3 V, Li/Li+)to 1.5,0.75,0.5,and 0.25 V, respectively;(e) the potential waa scanned from OCV to 0.05 V (Li/Li+). Nickel Electrodes treated in LiC10&4F/C02 (3otm)
uci-0
W
u
z a
m LL
0 v)
m
a
c 0 0 5 V ( V S LI/L,*)
41
)O
3600
3200
2800
2400 2000 WAVE NUMBER
1600
1200
800
400
Figure 12. FTIR spectraobtained ex situ from nickel electrodes polarized in LiClO4 1 M/MF/C02 (3 atm) solutions: (a-d) the potential was stepped from open circuit voltage (OCV) (-3 V, Li/Li+)to a certainlow potential followedby washing (pure MF) and drying; (a) 1.5 V; (b) 1 V; (c) 0.05 V.
The spectra of Figure 11prove that MF is reduced on nonactive metals in the presence of Li salts around 1.5 V, but C02 reduction requires lower potentials. As the potential is lowered,Li2COsis formed (in addition to lithium formate) and becomes the major surfacespecies. However, at potentials close to that of Li deposition potentials (Figure 12c)new products, whose structure will be discuesed e1sewhere,14are formed. The reduction wave in Figure 8 is probably due to this process. The surface chemistry of the electrodes treated in LiClOdMF/C02 solutions is complicated. The spectrum in Figure 12a related to 1.5 V has both lithium formate and LizC03peaks. However, since the spectrum in Figure l l a related to LiASF6 solution has no Li2CO3 peaks, it is
obvious that the source of the lithium carbonate formed on the electrode in MF/C02/LiC104 around 1.5 V is not C02 reduction but rather a secondary reaction. Hence, these results seem to prove that, in spite of the competition of solvent reduction, LiC104 is also reduced around 1.5 V. An obvious product is Liz0 (in addition to chlorides) which readily reacts with C02 to form Li2CO3. Figure 12b related to 1V also has, in addition to lithium formate and Li2CO3 peaks, ROCOzLi peaks (1320 and 820 cm-1 and the shouldersaround 1650and 1090 cm-9. These species could be formed due to reaction between alkoxy species and COZ.~ Figure 12c related to 0.05 V shows that, in contrast with LiAsFs/MF/COz, the relative amount of Li2CO3 does not increase as potential decreases (compare with Figure llb,c). Hence, the passivation of the electrodes at low potentials in LiClOJMF solutions due to the presence of CO2 as indicated in Figure 7 may be explained by the above secondary reactions of C02 with surface species formed by salt and solvent reduction, which suppress direct COZ reduction, probably due to electrode blocking by their products.
Conclusion Methyl formate is one of the most reactive polar aprotic solvents toward active metals. It is reduced on active metals (such as lithium and calcium) to metal formate as a major product which precipitates on the metal surface and passivates it. MF reduction on noble metals in the presence of Li or Ca salts occurs at potentials below 1.5 V (Li scale) to also form metal formate, which forms surface films and passivates the electrodes. The presence of two expected contaminants-water and methanol in solutions at trace amounts-does not affect the surface chemistry of active metals or noble metals at low potentials in MF since traces of methanol do not seem to react on the electrodes, and metal hydroxide formed by water reduction reacts with MF to form metal formate. The compatibility of MF/C02/Li salt solutions for the secondary Li system reported, which is surprising in light of the high reactivity of MF, is due to the effect of CO2 on the surface chemistry. COz in MF reacts on lithium and on nonactive metals at low potential to form Li2CO3; therefore, passive films formed on lithium or noble metals at low potentials in MF/C02 contain both lithium formate and lithium carbonate. However, when the salt is LiAsFe and at low C02 pressures, MF/C02 solutions are reduced on lithium to a more complicated species, probably polymeric, whose identification has not yet been completed. Acknowledgment. Partial support for this work was obtained from BSF Binational US-Israel Science Foundation. Registry No. MF, 107-31-3;Con,124-38-9;Li, 7439-93-2;Ca, 7440-70-2;Ni, 7440-02-0; Au, 7440-57-5;LiC104, 7791-03-9;CaClO,, 13477-36-6; BUNClO,, 1923-70-2; LiBF4, 14283-07-9; Ca(BF4)2,13814-93-2;BhNBF,, 429-42-5; L b F 8 , 29935-35-1; HCOOLi, 556-63-8; (HCOO)&a, 544-17-2;LizCOS, 554-13-2.
