J. Electroanal. Chem.

, 199 (1986) 93-100 93

Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands

STUDIES USING MICROELECTRODES OF THE Mg(II)/Mg COUPLE IN

TETRAHYDROFURAN AND PROPYLENE CARBONATE

J. DAVID GENDERS and DEREK PLETCHER

Department of Chemtstry, The Umverstty, Southampton SO9 5NH (Great Brrtam)
(Received 4th September 1985; in revised form 21st October 1985)

ABSTRACT

The deposition and dissolution of ma~esium metal from various tetrahydrofurau and propylene
carbonate solutions has been investigated using microelectrode techniques. It is demonstrated that from a
solution of a Grignard reagent, CaHsMgBr (0.5 mol dme3) in THF, it is possible both to electroplate
magnesium onto copper and to redissolve it anodically with high coulombic efficiency. The equilibrium
potential of the Mg/Mg(iI) couple in this medium is estimated to be + 850 mV vs Li/Li+ and the couple
is also shown to be quite reversible with an exchange current denstty of 1 mA cm-‘. In contrast we were
unable to deposit Mg cathodically from solutions of MgBr, and Mg(CIO,), in THF or PC even at 338 K
and/or in the presence of a base electrolyte (CH~)~NCIO~ or LiAsF,; Indeed, the presence of MgBra
inhibits substanttally the deposition of lithium metal. The addttion of MgBr, to the solution of Gngnard
reagent, however, enhances the rate of depositton of Mg metal. During long time-scale experiments.
magnesium metal also results from a chemical reaction between the electrodepostted lithium and
magnesium ion in solution.

The electrodeposition and dissolution of magnesium have potential applications

in surface finishing, battery technology and even electroforming or metal extraction.
It is therefore not surprising that, from time to time, papers have appeared which
describe attempts to electroplate magnesium [l-5] or to study its corrosion 161 in
non-aqueous solvents. Most investigations of the metal deposition have used only
constant current electrolysis with analysis of the cathode deposit and these give no
information about kinetics or mechanism. Other groups [6-lo] have studied the
Mg(II)/MgHg couple where a more complete description of the thermodynamics
and kinetics is possible. For example, in dimethylformamide the standard potential
[7] was found to be - 2.47 V vs. a Ag/Ag+ reference electrode and the standard rate
constant [6] was appro~mately 10Sg cm s-i.
There is convincing evidence for the deposition of magnesium from solutions of
Grignard reagents in ether solvents [1,2,4] while the reduction to the metal of a
magnesium halide species in a solution of a Grignard reagent in an ether was an
integral part of the commercial Nalco process for the electrolytic production of lead

~22-0728/86/SO3.50 0 1986 Elsevier Sequoia S.A.

94

alkyls [5]. The deposition of magnesium from solutions of Mg(BH,), or MgBrz +

LiBH, in ether has also been reported [3] but others have not been successful in
electroplating magnesium [10,13.14]. We have recently developed microelectrode
techniques [11,12] for the study of the Li/Li+ couple in ether solvents and now wish
to report their extension to the Mg/Mg(II) couple. Both solutions of Grignard
reagents and magnesium salts with and without a base electrolyte have been
investigated.

EXPERIMENTAL

The construction of the microelectrodes (40 pm and 80 pm diameter Cu discs),

the instrumentation and the experimental techniques have been described previously
[11,12].
The THF (East Anglia Chemicalsi was refluxed under a Nz atmosphere and
above equimolar amounts of freshly cut sodium and benzophenone and then
distilled immediately prior to the experiment. PC (Aldrich) was distilled under
reduced pressure. Prior to experiments, solutions were degassed with argon. Electro-
chemical Grade lithium hexafluoroarsonate (USS Agri-Chemicals), anhydrous mag-
nesium perchlorate (BDH) and magnesium bromide etherate (Aldrich) were used as
supplied while tetramethylammonium perchlorate and the Grignard reagent were
prepared by standard procedures.
The potentials in this paper are reported versus a Li/Li+ (0.5 mol dm-“)
electrode in the solvent used for the experiment.

RESULTS AND DISCUSSION

Figure la shows a cyclic voltammogram recorded at 293 K and at 50 mV s-’ for

a 80 pm diameter copper disc electrode in a solution of the Grignard reagent,
C,HgMgBr (0.5 mol dmv3), in THF but without any added electrolyte. It was
recorded in a cell where the second (reference) electrode was a Li foil in THF +
LiAsF, (0.8 mol dmp3) and the two solutions were separated by a glass sinter. Cyclic
voltammograms were also run in a cell without a sinter and with a Mg rod in the
Grignard reagent solution as the reference electrode; the responses are identical
except that with the Mg rod reference electrode the whole voltammogram is shifted
along the potential axis so that the reverse scan crosses the I = 0 axis at E = 0.0 V.
The cyclic voltammogram of Fig. 1 has a similar form to that reported for the
deposition and dissolution of Li in THF [ll]. On the sweep from + 1300 mV, no
current is observed until + 780 mV when the cathodic current rises steeply to the
potential limit. On reversing the direction of scan, the reduction process continues
until +850 mV; at this potential, the current crosses the potential axis and a
symmetrical “stripping peak” is seen. The total anodic and cathodic charges are
equal and the voltammogram is clearly that for a metal deposition and dissolution
process.
 95

-2
0.4 mA cm
f

r’
+1300 mV
I
+1300 mv +820 mV

I
Fig I. Cychc voltammograms run at polished, 80 pm diameter, copper rmcroelectrodes. Potential scan
rates 50 mV s-‘. (a) C,H,MgBr (0.5 mol dme3) in THF at 293 K: (b) C2H,MgBr (0.5 mol dmv3) m
refluxmg THF; (c) C,H,MgBr (0.5 mol dme3) +MgBrzEt20 (0.5 mol dme3) in refluxing THF.

Potential steps from + 850 mV to the region + 780 to 650 mV led to well-formed
rising I-t transients. Typical transients are shown in Fig. 2a. The timescales of the
transients decreased while the current densities increased as the applied potential
was made less positive. Such data confirms that the cathodic process involves the
nucleation and growth of a new phase on the copper surface. Figure 2(b) shows the
transients for the oxidation of an electrodeposited Mg layer. The “steady state”
currents read from a series of both deposition and dissolution transients were used
to construct a Tafel plot [ll] and a value of 1 mA crnm2 for the exchange current
density was estimated. Hence the Mg/Mg(II) couple in the Grignard solution is
quite rapid although slower than the Li/Li+ couple in the same solvent 1121.
It should be noted that the nature of the electroactive magnesium(I1) species in
the Grignard solution is unclear although the species present in solutions of
Grignard reagents lead to unexpectedly high conducti~ties and have been discussed
in terms of the Schlenk equilibrium [15]. Certainly it is to be expected that the
Mg(I1) is effectively complexed by the alkyl group and hence the equilibrium
potential for the Mg/Mg(II) couple, + 850 mV vs. Li/Li+, is surprisingly positive.
In aqueous solution it has been calculated that the standard potentials for Mg/Mg’+
and Li/Li+ are -2.73 V and -3.06 V vs. SHE respectively [16]. But the much
larger difference in THF may result from a negative shift of the Li/Li+ couple due
to the strong interaction of the cation with ethers rather than a positive shift in the
Mg/Mg(II) couple.
96

In the Nalco process, the electrolysis is carried out at elevated temperatures and,
although the electrolyte contains Grignard reagent, it is essential to the overall
chemical change in the cell that the magnesium formed at the cathode results from
the reduction of a magnesium halide species formed at the anode. Hence, it was

I/mA cm-’
t

t fs
5 10 15 20

Fig 2. Current-time transients for the depoxtion and dissolutmn of Mg from C,H,MgBr (1.0 mol
dme3) in THF at 293 K. Potential-ttme profiles: (a) +850 mV to the value shown; (b) +850 mV to
+630 mV for 12 s, then to the value shown.
 97

considered of interest to investigate the influence of temperature and added mag-

nesium bromide on the cyclic voltammetry. Figure lb shows the voltammogram for
the solution of Grignard reagent (0.5 mol dmP3) in refluxing THF at 338 K. The
I-E response is shifted to more positive potentials; nucleation is observed at + 890
mV and on the reverse sweep, reduction continues to +900 mV. Iience it appears
that, thermodynamically, the deposition of magnesium is easier at the higher
temperature. On the other hand, very surprisingly, the shape of the I-E curve on the
reverse sweep (there is a more pronounced region of zero current) would suggest that
the kinetics of the Mg/Mg(II) couple are slower at the higher temperature, cf.
Li/Li+ [12]. This is, of course, difficult to accept and it may be that a change in the
composition of the Grignard reagent solution with temperature [15] or some film
formation on the magnesium surface occurring at 338 K leads to the observed
changes in response. The changes in the voltammogram are quite reproducible on
cycling the temperature between 293 and 338 K. Figure lc shows the Z-E response
for C,HsMgBr (0.5 mol dmm3) +MgBr,Et,O (0.5 mol dme3) again in refluxing
THF (the MgBr,Et,O is not soluble below 333 K). The shape of the voltammogram,
Fig. lc, is very similar to that in Fig. lb except it is clear that the currents for Mg
deposition are higher and hence the charges associated with both the cathodic and
anodic processes are larger. Hence it must be concluded that the deposition of Mg
from MgBr, does occur in this system.
On the basis of the above data, it was expected that the deposition of Mg from
solutions of other magnesium(I1) species, especially salts, would be straightforward.
This proved not to be the case. The first difficulty is to identify anhydrous
magnesium salts which are soluble in THF; anhydrous magnesium hexa-
fluoroarsonate and related salts are not manufactured and of the salts available. only
“dry” Mg(ClO,), and MgBr,O(C,H,), were found to dissolve (the latter only at
elevated temperatures). But cyclic voltammograms, run for Mg(ClO,), (0.2 mol
dmF3) in THF at room temperature and for MgBr,O(C,H,), (0.3 mol dme3) in
refluxing THF with either 80 pm Cu or 8 pm C electrodes, did not show any
features expected for metal deposition. Indeed, little current was observed within a
reasonable potential range.
It was thought that this failure to deposit Mg might be due to a deficiency of
ionic species in solution and hence the experiments with MgBr,O(C,H,)z were
repeated but with LiAsF, (0.5 mol dme3) present in the refluxing THF. Again, the
cyclic voltammograms contained no evidence for the cathodic deposition of Mg
metal. On the other hand, on extending the potential scan to values negative to the
Li/Li+ couple, Li metal deposition was observed. Interestingly, the curves show that
the presence of MgBr, inhibits both Li deposition and dissolution. For example,
with a solution of MgBr, (0.3 mol dmP3) + LiAsF, (0.5 mol dmm3) in refluxing
THF, no cathodic current is observed until -0.43 V (cf. -0.05 V in the absence of
MgBr,) when the current increase is extremely rapid (similar to that which would
occur for film breakdown) while on the reverse sweep only a small anodic peak is
observed (there is a large imbalance). On the other hand, the possibility that the
anodic peak is due to Mg dissolution is essentially ruled out because it is almost
98

0.5 V negative to the dissolution peak in the Grignard solution. Nor is the change in
this cyclic voltammogram for lithium deposition and dissolution due to the presence
of bromide ion, since cyclic voltammograms for a LiBr solution were very similar to
those for LiAsF6, with only a small negative shift in the equilibrium potential. This
would suggest that the difficulty with Mg deposition is not associated with the poor
kinetics of the Mg/Mg(II) couple but more due to the formation of an inhibiting
layer on the electrode surface; adsorbed Mg2+ or more likely a hydrolysis or
solvolysis product of the Mg(I1) species in solution are possibilities.
The inhibition of lithium deposition by MgBr, is also seen in PC, see Fig. 3.
When MgBr, is present, lithium deposition does not commence until quite negative
potentials and the stripping peak for lithium oxidation is also shifted markedly to
positive potentials. Indeed, the Z-E response on the reverse sweep is totally different
after addition of MgBr,. Similar experiments in PC using solutions of Mg(CIO,),
and MgBr, in the presence of (CH,),NClO, also led to no evidence for Mg
deposition.
On the other hand, Mg metal can be formed, but only as powder in solution or a
poorly adhering deposit, during long time-scale electrolyses of solutions containing
both LiAsF, and MgBr20(C,H,)z in THF. The magnesium almost certainly arises
from a chemical reaction between Li metal and a Mg(I1) species in solution as
already reported by Rieke et al. [17]. Moreover, we believe that such a mechanism
via Li metal would explain the obse~ations of Connor et al. [3]; most of their
solutions would contain sufficient Li’ ion, either added as a background electrolyte

IaJ
J 4 mA cm
-2 IbJ

I
+500 mV +lOOO mV -800 mv

Fig 3. Cyclic voltammograms in PC at 293 K. 80 pm Cu disc electrode. Potential scan rate 50 mV s-‘.
(a) LLAsF, (0.65 mol dm..‘); (b) LIAsF, (0.65 mol dm-‘1 + MgBr, (sat).
 99

or present as an impurity due to their method of preparation of the magnesium salts.

With such a mechanism, it is probably inevitable that the deposit adheres only
poorly.
Some preliminary experiments have examined the possibility that a catalytic
amount of Grignard reagent might be sufficient to permit deposition of magnesium
from THF/MgBr,Et 2O solutions. A cyclic voltammogram for a solution of
MgBrzEt,O (0.3 mol dmw3)+CZH5MgBr (0.1 mol dmm3) at 338 K still shows
deposition of the metal although the nucleation overpotential is increased to about
300 mV. On the other hand, the peak for the dissolution of magnesium is very
smeared out indicating that the kinetics of the anodic corrosion are poor in this
medium and the stripping efficiency apparently drops to about 40%. Moreover a
similar effect could be seen in a solution containing only the Grignard reagent (0.1
mol dm-‘). The reasons for the inhibition of magnesium corrosion in low concentra-
tions of Grignard reagent are not clear and further studies of both the deposition
and corrosion reactions are necessary to understand the electrochemistry of solutions
containing both C2H,MgBr and MgBr,Et?O.
We therefore conclude that the electrodeposition of magnesium does not occur
when only simple Mg(II) species are present in solution in THF or PC. Magnesium
metal can be formed via electrodeposited Li but this does not adhere to the electrode
surface and is present largely as powder free in solution. In addition, Mg deposition
and anodic dissolution are possible and, indeed, appear to proceed well when the
catholyte contains a high concentration of Grignard reagent, and the organomag-
nesium species appears able to catalyse the reduction of MgBr,. For example, it
could be that the reducible species is an organomagnesium species but the mag-
nesium bromide in solution is able to “trap” the organic fragment after reduction to
reform an electroactive species. There is evidence in the literature that aluminium
deposition is catalysed by hydride and organoaluminium intermediates [lS]. When
only low concentrations of Grignard reagent are present. the deposition and particu-
larly the anodic dissolution reactions are slowed down. When the stripping efficiency
is much less than loo%, the interpretation of the cyclic voltammetric data becomes
less certain and additional approaches should be used to investigate magnesium
deposition.

ACKNOWLEDGEMENT

The authors would like to thank ICI Organics Division (Grangemouth) for
support of this work.

REFERENCES

1 L.W. Gaddum and H.E. French, J Am. them Sot.. 49 (1927) 1295.
2 W. Evans. J. Am. Chem. kc.. 64 (1942) 2965.
3 J.H. Connor, W.E. Reid and G.B. Wood, J. Eiectrochem. Sot.. 104 (1957) 38.
4 A. Brenner and J.L Sligh, Trans. Inst. Met. Fimshmg. 49 (1971) 71.
100

S M.M. Baizer and H. Lund (Ed%), Organic Electrosynthesis, Marcel Dekker, New York, 1983.
6 Q.R. Brown and R. McIntyre. Electroclum. Acta, 30 (1985) 627.
7 OR. Brown and R, McIntyre, Electrochim. Acta, 29 (1984) 995.
8 W.R. Fawcett and J.S. Jaworskir, J. Chem. Sot. Faraday Trans. 1. 78 (1982) 1971.
9 T. Psarras and R.E. Dessy. J. Am. Chem. Sot., 89(196~) 5132.
10 J. Bittrich, R, Landsberg and W. Gaube, Wiss. Z. Tech. Hochsch. Chem. “Carl Schorlemmer”
Lenna-~erseburg. 2 (1959/60) 449.
11 J.D. Genders, W. Hedges and D. Fletcher, .I. Chem. Sot. Faraday Trans. 1, 80 (1984) 3399.
12 W. Hedges and D. Fletcher, J. Chem. Sot. Faraday Trans. 1, m press.
13 T.P. Dirkse and H.T. Briscoe, Met. Ind., 36 (1938) 284.
14 A. Brenner, Adv. Electrochem. Electrochem. Eng., 5 (1967) 205.
15 EC. Ashby, Quart. Rev. Chem. Sot., 21 (1967) 259.
16 W.M. Latimer, The Oxidation States of the Elements and Thetr Potentials in Aqueous Solution.
Prentice Hall, Englewood Cliffs, 1952.
17 R.D. Rreke, P.T. Li, T.P. Burns and S.T. Uhm, J. Org. Chem., 46 (1981) 4323.
18 M.W.M Graef, J. Electrochem. Sot., 132 (1985) 1039.