The Grignard Reagent
Preparation, Structure, and Some Reactions
Milton Orchin
University of Cincinnati, Cincinnati, OH 45221
The Grignard reagent, t h e product resulting from the re-
action of an alkyl or aryl halide with elemental magnesium
(I),
RX + Mg - RMgX
was named after Victor Grignard (1871-1935; Nobel Prize,
shared with Paul Sabatier, 1912) and is probably the single
most important organometal reagent used in the laboratory
synthesis of organic compounds in this century. Indeed a
case can he made for the proposition that the remarkable
versatility of the Grignard reagent and its ease of prepara-
tion delayed the interest in organo-transition metal chemis-
try for decades. l'heavai~abili;~ of pure magnesium required
for the laboratory preparation of the Grignard reagent was
no doubt enhanced hy the commercial interest in magne-
sium, principally for use as an incendiary. The "Elektron"
was a 10-lb incendiary bomb developed by the Germans in
1918 and consisted of almost pure magnesium.
The successful preparation of the Grignard reagent in
good yield requires an ether solvent, usually diethyl ether or
tetrahvdrofuran (or less commonly EtzN), whose solubiliz-
ing prbperties arise from complcx~tionot the ether oxygen
at the maanesium center. Although practically all Grignard
reactions can be rationalized on t h e basis o f the strGture
RMgX, it was realized for many years that such a represen-
tation is a gross oversimplification and that the formation of
the reagent and its structure and reactions are very complex
(2).
Structure of the Grlgnard Reagent
I t is now generally accepted that in diethyl ether and
tetrahydrofuran solution, Grignard reagents are best repre-
sented by an equilihrium mixture:
2RMgX + R,MG + MgXz
1 2
RzMg can be conveniently separated from the RMgX and
MgXz (which are highly associated in EtzO) by addition of
dioxane, which precipitates all the halide. No substrate is
known that will react exclusively with RMgX hut not RzMg
and vice versa, and the reaction with either alone or the
mixture always produces the same product albeit with occa-
sional differences in rate (3). Dimeric and polymeric struc-
tures make minor contributions, especially in dilute EtzO
solution (less than 0.3 M) and hence the equilibrium he-
tween 1 and 2, the so-called Schlenk (4) equilibrium repre-
sents the hest formulation of the Grignard reagent. This
equilihrium is consistent with the exchange observed he-
tween radioactive 28MgBr~ and (CzH5)~Mg (5a) and between
25MgBr~ and (CzH5)zMg (56) when carefully purified magne-
sium is used. No exchange is reported t o occur (5c) when less
pure magnesium is used. The reason for the difference re-
mains a mystery. Impurities in the range of parts per million
and/or traces of oxygen have been implicated but not dem-
onstrated. I t is more likelv that ex~erimentaltechniques
were imperfect in the no-exchange experiments. I.'nequivo.
ral evidence for the Schlenk equilihrium,
586 Journal of Chemical Education
CH3MgBr Etp -320
CHsMgBr THF 3.5
CzHsMgBr Et20 482
C2HsMgBr THF 5.09
C2HsMgCI THF 5.52
comes from 'H NMR variable temperature studies @a). At
room temperature only a single methyl signal is observed,
hut a t about -100 'C two methyl signals appear, indicating
that the rapid equilibrium a t room temperature is frozen out
a t the low temperature.
There are several dimeric structures that can be written
for the Grignard reagent inEtzO when X = F, CI, Br, or I and
in T H F where X = F. Of these, the symmetrical, 3, and
unsymmetrical, 4, dimers are the most probable. However,
the dimeric structure that best explains (66) the intermedi-
ate responsible for the Schlenk equilihrium is the alkyl
bridging dimer, 5.
The dramatic effect of solvent on the composition of the
Grignard reagent is illustrated by experimental values of K
(see table) found (7) for the Schlenk equilibrium in the
solvents diethyl ether and tetrahydrofuran. The large differ-
ence in K as a function of solvent is ascribed to solvation
effects, especially those associated with solvation of the
MgBrz; in THF, four solvent molecules are involved, but in
EtzO, only two are involved.
Formation of the Grignard Reagent
There appears to be general agreement that Grignard for-
mation is initiated by a single electron transfer (SET) from
surface magnesium to an adiacent RX, formina a radical
anion that is associated witha surface kagnesi"m radiral-
cation. There also is general agreement that free radicals R.
are formed somewh&e in the process and that RMgX is
formed before RzMg with which i t rapidly equilibrates.
There remains, however, some controversy as to whether the
free radicals formed first diffuse directly and completely
into solution and return t o the surface to form RMgX or
whether some fraction of the population of radicals remains
associated with the maenesium and combine to form RMeX
without any substantii liferime as free radicals in solutih.
The two pathways to HMgX areshown in Figure 1 where the
subscripts indicate a surface species (s) and a species in
solution (soh), respectively.
Perhaps the most convincing evidence for the suggestion
that radical formation can occur on the surface to form
RMgX (pathway b, d) without formation of free R. (pathway
c, e, d) is the behavior of chiral l-hromo-l-methyl-2,2-di-
phenylcyclopropane, 6.
The reaction of the bromide 6 with Me eives the corres~ond-
ing RMgBr with an optical purity of ;rout 12%(8). were all
the bromide to proceed to reaction exclusivelv throueh the
free radical 7 inbo~ution,complete racemization wouli have
occurred since in inde~endentstudies it has been shown (9)
that the rate of inversibn of the methylcyclopropyl radical (7
without the phenvl - . is about 101's-I at 71 O C . Fur-
groups)
thermore when 7 i s generated in solution from a variety of
chiral precursors, other than the Grignard reagent, the prod-
uct is racemic (lo), and side products resulting from dimer-
ization and disproportionation that are characteristic of free
radical reactions in solution are present in such reactions
(11) but are absent in the preparation of RMgBr from 6.
The ~ . r e.~ a r a t i oofnthe Grienard reaeent from chiral nre-
cursors isonly one of a number of probes for assessing radical
formation. Other ~robesconsistof CIDNP studies (12).radi-
cal trapping expe;iments (13), and, particularly, the isomer-
ization of rearraneement-prone radicals. This latter Drobe
has been particul&ly instructive because the rate ofrear-
rangement of several radical species are known from inde-
-
pendent experiments and the extent of isomerization RMgX
R'MgX during Grignard formation can be determined
experimentally as well.
Extensive experiments have been performed with 6-bro-
mo-l-hexene, 8. When the Grignard reagent is prepared
from this halide, two RMgX compounds are isolated:
-
The rate of isomerization R- .R' of the putative intermedi-
ate cyclohexenyl radical is known (14) to be about lo5 s-',
and preparation of the Grignard reagent from 8 leads to
about 95% of 9 and 5%of 10. When a log-log plot of the ratio
of R'MgXIRMgX to the rate constant for the rearrangement
is constructed using this point and two others similarly ob-
tained, a straight line with a slope of 0.5 can be used to
connect these points (16). This relationship is to be expected
from a diffusion theory model (15).Hence it was proposed
that in Grignard formation the radical R. is formed first,
diffuses into solution, has time to moreor less rearranee. and
both b a n d .R' return to thesurfare toreact with ~ g ~ , , a n d
form HMgX and R'MgX. This is the process of pathway c, e,
d, Figure I . The presently available evidence indicates that
Figure 1 . The mechanism of Orignard formation
this pathway as well as pathway d operate competitively.
Relatively little work has been done on the rate of forma-
tion of the Grignard reagent and the identification of the
kinetically important rate-determining step although the
initial one-electron reaction on the magnesium surface is a
reasonable candidate. The rate increases with surface area of
themagnesium (16) but does not seem to beaffected by trace
imnurities. In one studv the rate of reaction in ether of
cyhopentyl bromide with magnesium metal in the form of a
variable meed rotatine disk (17) was determined. The rate
was foundto be propo;tional the square root of the angu-
lar velocity of the disk demonstrating that the reaction is
mass transfer limited.
ReaCtlon of Grlgnard Reagents
There are literally thousands of examples of the reactions
of Grignard reagents with a variety of substrates. Perhaps
one of the most detailed studies involves the reaction with
benzophenone, e.g.:
In a careful study of this reaction using both CH3MgBr and
(CH3)3CMgBr,a dual pathway mechanism, Figure 2, was
proposed (18) that rationalized the rates and extent of con-
version of the ketone to 11, the precursor to the tertiary
alcohol, and 12, the precursor to the pinacol, the two major
products. As shown in Figure 2, the former develops via a
polar pathway, and the latter via a single electron radical
pathway (Fig. 2).
The partitioning between the polar and the one electron
transfer pathways depends on the nature of the ketone, the
solvent, the R group, and the purity of the magnesium metal.
Thus the reaction of CH3MgBr in EtnO (prepared from pure
single crystal Mg in excess CH3Br) with PhzCO proceeds
exclusivelyvia the polar mechanism, while tert-butylmagne-
sium chloride reacts with P h E O in EtzO by single electron
transfer. Interestingly, when 0.05 M FeC13 is added to the
CH3MgBr reaction the mechanism is converted from the
polar pathway to the single electron transfer mechanism (see
below) with production of the pinacol.
Figure 2. me two p h w a y s fw ma Grignard-ketone reaction.
Volume 66 Number 7 July 1989 587
The Effect of Transition Metals and Thelr Halldes on the
Grignard Reagent and Its Reactions
When catalytic quantities of CoC12are added to a solution
of a Grignard reagent containing free excess alkyl halide,
rapid coupling occurs:
P.M... r\ c0.1.. 4 c,.,

RMgX R X+ ac4
-
RR' MgX, + Figure 3. Free radical catalysis of the Grlgnard reaction.

as well as formation of RR, Rf-H, and R'R'. The steps in the

reaction may be written as follows using the reaction be-
tween ethyl hromide and phenylmagnesium bromide: RMgCl MCI, + - RMCI + MgC1,
CoCI, + PhMgBr - Ph. + CoCI + MgBrCl and with excess RMgX
2RMgCI + MClz - R,M + 2MgC1,
2Ph - Ph-Ph The most electro~ositivemetal (in this case Me) ends UD
comhined with halogen. ~ l e m e n t a magnesium
l Lduces th'e
halides of all metals below it in the electromotive series. and
similarly RMgX converts the inorganic halides of all metals

C,Hs + Solvent - C,H,

below magnesium to the alkyl derivative of the element.

Ph. + Solvent - C,H,
The generation of the free radical occurs in the first step.
The mechanism of this initiation can be viewed as exactly
analogous to the one electron transfer pathway in the benzo-
phenone reaction with methylmagnesium bromide shown
earlier:
-
Thus with CuXz the reaction may he written: CuClz
+
+
2RMgX RzCu 2MgXz followed by reductive elimina-
tion: RzCu R-R +
Cu. This represents an overall two elec-
tron reduction of the metal, but this explanation of the
coupling has received relatively littlesupport,and it is much
more likely the product arises through the radical mecha-
nism shown in the catalytic cycle with CoCl?.
Acknowledgment
Discussions with E. C. Ashby, H. M. Walhorsky, and J. L.
Belletire were very helpful.

A catalytic cycle involving CoC1 t o generate P h and CzHs

radicals in the above reaction is shown in Figure 3. One
would expect the radical pathway to PhC2H5to be competi-
tive with its formation from the S N displacement
~ in a
Wurtz-type coupling reaction:
BrMgPh + C&Br - +6 -6 +6
[BrMg. .Ph.. .CzHs.. Br]
-6

In the above discussion i t was assumed that some RX is

present in excess in the Grignard solution containing the
catalytic CoClz (or other transition metal halide such as
CuClz, FeCla etc). However, it is known that if CoClz is
added to a Grignard reagent in the absence of free RX,
coupling of the Grignard still proceeds. Presumably the
mechanism shown for the one electron transfer involving
RMgX and CoClz that leads to radical formation is again
operative. However, another possibility should be consid-
ered since i t is known (19) that RMgX can convert a transi-
tion metal halide to the alkyl derivative:

588 Journal of Chemical Education