Substrate effects on the mechanism of enantioselective hydrogenation using ruthenium bis(phosphine) complexes as catalyst: A mechanistic investigation of the hydrogenation of α,β-unsaturated acids and esters based on deuterium labeling studies

Article abstract of DOI:10.1016/j.ica.2005.10.026

The mechanism of ruthenium-bis(phosphine) catalyzed enantioselective hydrogenation of olefins was examined using [Ru((R)-BINAP)(H)(MeCN)_n(sol)_{3 - n}]BF₄ (n = 0-3, sol = solvent used in reaction) as catalyst. Tiglic and angelic acids were used as standard α,β-unsaturated acid substrates; (Z)-methyl α-acetamidocinnamate and dimethyl itaconate were used as standard α,β-unsaturated ester substrates. Isotopic labeling studies (deuterium scrambling) indicate that two distinct mechanisms are in operation for α,β-unsaturated acids versus α,β-unsaturated esters. In each case, 5-membered metallocycle intermediates are formed via olefin-hydride insertion. The mechanisms, however, deviate primarily in the activation of dihydrogen, which is strongly affected by the nature of the substrate. Hydrogenation of α,β-unsaturated acids proceed via heterolytic cleavage of dihydrogen, whereas hydrogenation of α,β-unsaturated esters proceed via homolytic cleavage of dihydrogen. A full discussion of the mechanisms is presented.

Full text of DOI:10.1016/j.ica.2005.10.026

Inorganica Chimica Acta 359 (2006) 2760–2770
www.elsevier.com/locate/ica
Substrate effects on the mechanism of enantioselective
hydrogenation using ruthenium bis(phosphine) complexes as catalyst:
A mechanistic investigation of the hydrogenation of
a,b-unsaturated acids and esters based on deuterium labeling studies
a
b
c,
*
Christopher J.A. Daley , Jason A. Wiles , Steven H. Bergens
a
Department of Chemistry, Western Washington University, 516 High Street, Bellingham, WA 98225, USA
b
Achillion Pharmaceuticals, Inc., 300 George Street, New Haven, CT 06511, USA
Department of Chemistry, University of Alberta, Edmonton, Alta., Canada AB T6G 2G2
c
Received 30 September 2005; accepted 17 October 2005
Available online 5 December 2005
Dedicated to Professor Brian R. James.
Abstract
The mechanism of ruthenium-bis(phosphine) catalyzed enantioselective hydrogenation of olefins was examined using [Ru((R)-
BINAP)(H)(MeCN)n(sol)3 ꢀ n]BF4 (n = 0–3, sol = solvent used in reaction) as catalyst. Tiglic and angelic acids were used as standard
a,b-unsaturated acid substrates; (Z)-methyl a-acetamidocinnamate and dimethyl itaconate were used as standard a,b-unsaturated ester
substrates. Isotopic labeling studies (deuterium scrambling) indicate that two distinct mechanisms are in operation for a,b-unsaturated
acids versus a,b-unsaturated esters. In each case, 5-membered metallocycle intermediates are formed via olefin-hydride insertion. The
mechanisms, however, deviate primarily in the activation of dihydrogen, which is strongly affected by the nature of the substrate. Hydro-
genation of a,b-unsaturated acids proceed via heterolytic cleavage of dihydrogen, whereas hydrogenation of a,b-unsaturated esters pro-
ceed via homolytic cleavage of dihydrogen. A full discussion of the mechanisms is presented.
ꢀ 2005 Elsevier B.V. All rights reserved.
Keywords: Ruthenium BINAP; Tiglic acid; (Z)-Methyl a-acetamidocinnamate; Asymmetric catalysis; Hydrogenation; Mechanism; Isotopic-labeling
study
1. Introduction
thalene (BINAP) and related species as the chiral ligand [1–
4]. The result is not only systems with exceptional enanti-
oselectivity, but systems that are also capable of extremely
high turnover frequencies (TOFs) and turnover numbers
(TONs).
Several groups, most notably that of Halpern et al.
[5], have reported on the mechanism of enantioselective
hydrogenation of a,b-unsaturated acids/esters in alcohol
solvents using ruthenium-bis(phosphine)-dicarboxylate
species as catalyst [6–10]. The general features of Halp-
ernÕs proposed mechanism are presented in Scheme 1
with tiglic acid as substrate. The mechanism follows four
main steps: first, a rapid equilibrium between the catalyst
The controlled hydrogenation of prochiral substrates by
chiral catalysts into enantiopure materials is an important
process in academia and industry. Over the past decade,
many ruthenium(II) bis(phosphine) complexes have been
determined to be highly active and selective catalysts for
the enantioselective hydrogenation of numerous unsatu-
rated organic substrates. The more successful catalysts
often incorporate 2,20-bis(diphenylphosphino)-1,10-binaph-
*
Corresponding author. Tel.: +780 492 9703.
E-mail address: steve.bergens@ualberta.ca (S.H. Bergens).
0020-1693/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2005.10.026

C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
2761
Scheme 1. Mechanism proposed by Halpern [5] for ruthenium bis(phosphine) catalyzed hydrogenation. Key steps are heterolytic cleavage of H2, H–D
exchange between 2 and solvent, and solvolysis to complete the catalytic cycle.
and tiglic acid by carboxylate ligand exchange to gener-
ate 1; second, heterolytic cleavage of dihydrogen to
generate the ruthenium monohydride species 2; third,
olefin-hydride insertion to generate the 5-membered het-
erometallacycle 3; and fourth, protonolysis (solvolysis)
of the Ru–C r bond in 3 to complete the cycle. Use
of methanol-d4 as solvent resulted in significant deute-
rium incorporation at the b-carbon of the hydrogenation
product, but little incorporation at the a-carbon. Halpern
et al. thereby concluded that significant H–D exchange
did not occur between the methanol-d4 solvent and free
H2, in the presence of catalyst and also that the rate
of H–D exchange between the monohydride species 2
and methanol-d4 must also be slow relative to the rate
that solvolysis of a ruthenium–carbon r bond does not
occur during this catalytic hydrogenation, and he pro-
posed a mechanism that involves oxidative addition,
not heterolytic cleavage, of dihydrogen gas (Scheme 2).
To account for the deuterium distributions observed by
both his group and HalpernÕs, Brown proposed that
H–D exchange with solvent can occur at two points in
the catalytic cycle: before, or immediately following ole-
fin-hydride insertion, Scheme 2: A and B, respectively.
The rates of H–D exchange with solvent may or may
not compete with the rates of olefin-hydride insertion
and/or reductive elimination, depending on the nature
of the catalyst and substrate used. BrownÕs mechanism
thereby provides a rational for the H–D exchange
observed by both groups.
As part of our ongoing efforts to elucidate the mecha-
nism of chirality transfer during enantioselective hydroge-
nation, we set out to clarify the mechanism of
hydrogenation of a,b-unsaturated acids and esters using
the same catalyst for a variety of studies. We previously
reported on the enantioselective hydrogenation mecha-
nisms of a,b-unsaturated esters [12–15], (i.e., (Z)-methyl
a-acetamidocinnamate (MAC, 13) [13–15], and methyl
a-acetamidoacrylate (MAA, 14) [12]) and ketones (i.e.,
dialkyl 3,3-dimethyloxaloacetate) [11] using [Ru((R)-
of olefin-hydride insertion, i.e., Scheme 1: k3 ꢁ k and
ꢀ1
k2. Reports by Noyori [6], Achiwa [7], and Chan [8] have
reaffirmed the overall proposed mechanism.
Brown et al. [9,10] reported that the catalytic hydroge-
nation of the a,b-unsaturated ester 7, using a different
ruthenium-bis(phosphine) catalyst than that used by
Halpern and Noyori but under similar conditions
(1 atm H2, 25 ꢁC, 24 h, methanol-d4), resulted in essen-
tially no deuterium incorporation into either the a- or
b-positions of the hydrogenation product, dimethyl 2,3-
dimethylsuccinate (8, Scheme 2). Brown thereby showed

2762
C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
Scheme 2. Brown [9] mechanism for ruthenium bis(phosphine) catalyzed hydrogenation. Predominant exchange process at step B for ruthenium
bis(phosphine) reductions of a,b-unsaturated acids. Predominant exchange process at step A for ruthenium bis(phosphine) reductions of a,b-unsaturated
esters.
BINAP)(H)(MeCN)n(sol)3 ꢀ n]BF4 (15: n = 0–3, sol = sol-
vent used in reaction) [16] as the active catalyst.1 We now
report the mechanistic changes between hydrogenations
of the a,b-unsaturated tiglic (5), and angelic (16) acids, as
compared to the a,b-unsaturated esters dimethyl itaconate
(17), MAC (13), and MAA (14), all with 15 as the active
catalyst. In addition, solvent effects are described. We also
disclose the isolation and characterization of the putative
catalytic intermediates observed during the hydrogenation
of dimethyl itaconate, namely [Ru((R)-BINAP)(MeCN)-
((R)- and (S)-C[CH3][CO2CH3]-[CH2CO2CH3])](BF4) (32a
and 32b).
Table 1
Pattern of deuterium incorporation in tiglic acid (5) hydrogenation
catalyzed by 15 and 19 under various conditions
H(trans)
CO2H
H(cis)
CH3 (e)
(d) H3C
H(c)
Catalyst, conditionsa
Percent deuterium incorporation
Hcis
Htrans
Hc
Hd
He
15 (H2, CD3OD)
15 (D2, CH3OH)
19b (H2, CD3OD)
19b (D2, CH3OH)
a
85
8
100
5
70
33
30
85
2. Results and discussion
Reaction conditions. Catalyst 15, 4 atm pressure, ambient temperature,
catalyst 19, 3 atm gas pressure, ambient temperature.
2.1. Ruthenium-BINAP catalyzed enantioselective
hydrogenation of a,b-unsaturated carboxylic acids
b
Reported by Takaya and Noyori [6].
Table 1 compares the results of the deuterium-labeling
studies of Takaya and Noyori where [Ru((R)-BINAP)(j2-
O2CCH3)2] (19) [17] was used as catalyst with our results
using 15 as catalyst. Use of 15 as catalyst resulted in a
greater degree of deuterium scrambling in 2-methylbutyric
acid (6) than that reported by Halpern, Takaya, and Noy-
ori. Under 3 atm of H2 in methanol-d4 solution, tiglic acid
was hydrogenated to yield 6 in 90% ee (R), with an abun-
dance of deuterium located at both the a-(Hc) and the b-
(Hcis) positions. The reaction performed under 3 atm of
D2 in methanol solution showed the expected reversal of
deuterium incorporation.
1
15 is efficient catalyst precursor for enantioselective hydrogenation of
The mechanism proposed by Halpern et al. can account
for these results if H–D exchange with the solvent occurs at
a,b-unsaturated carboxylic acids and ester derivatives, and is comparable
in efficiency to other ruthenium-BINAP catalyst systems reported.

C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
2763
either of two stages. The first possibility is if the rate of H–
Table 2
Pattern of deuterium incorporation in angelic acid (16) hydrogenation
catalyzed by 15 as catalyst under various conditions
D exchange between the solvent and H2 gas is comparable
to the rate of olefin hydride insertion (i.e., Scheme 1, kꢀ1 is
comparable to k3). We note, however, that unlike 19, 15
does not catalyze the H–D exchange between H2 and meth-
anol-d4 in the absence of added substrate [14] (i.e.,
H(cis)
CO2H
(trans) H
H(c)
CH3 (e)
(d) H3C
k
ꢀ1 = 0). This possibility is thereby unlikely. The second
possibility is if the rate of H–D exchange between solvent
and a species such as 2, Scheme 1 is comparable to the rate
of olefin-hydride insertion. Thus to account for the
observed exchange pattern, the Halpern mechanism
requires that a complex formed from the heterolytic cleav-
age of H2 gas (such as 21, Scheme 3) undergoes H–D
exchange with the solvent. Studies on the hydrogenation
of the thermodynamically less stable isomer of tiglic acid,
angelic acid (16) [18], were carried out to provide further
evidence for this sequence.
Hydrogenation of 16 (Table 2) under 3 atm of H2 in
methanol-d4 solution resulted in deuterium incorporation
at the b-position (Hcis = 62%), the a-position (Hc = 43%),
the other b-position (Htrans = 8%), and the b-methyl posi-
tion (Hd = 43%, assuming CH2D). The hydrogenation per-
formed under 3 atm of D2 in methanol solution resulted in
little-to-moderate deuterium incorporation at either the a-
position (Hc = 30%) or the b-position (Hcis = 7%). Small
amounts of deuterium incorporation also occurred, how-
ever, at the other b-position (Htrans = 8%) and at the b-
methyl position (Hd = 5%, assuming CH2D).
Catalyst, conditionsa
Percent deuterium incorporation
Hcis
Htrans
Hc
Hd
He
15 (H2, CD3OD)
15 (D2, CH3OH)
8
5
62
8
43
30
43
5
a
Reaction conditions. [Ru] = 2.6 mM, 3 atm gas pressure, ambient
temperature.
mechanism like HalpernÕs with both the heterolytic cleav-
age of H2 and a 5-membered heterometallocycle intermedi-
ate as shown in Scheme 3.
In this modification of the Halpern mechanism, the cat-
alyst 15 enters the catalytic cycle by reaction with substrate
to generate the carboxylate compound 20 and dihydrogen
(vide infra). Heterolytic cleavage of H2 results in formation
of the hydride 21, and rapid H–D exchange between 21 and
methanol-d4 leads to 21-d. Olefin-hydride or -deuteride
insertion leads to the 5-membered metallacycle 22, and sol-
volysis of the ruthenium–carbon r bond leads to 23.
The existence of 5-membered heterometallacycle inter-
mediates is supported by the deuterium-labeling studies
with angelic acid (16) via the sequence of steps shown in
Schemes 4 and 5. Heterolytic cleavage of dihydrogen leads
While deuterium is scrambled more when using 15 than
19 as catalyst, the results of the labeling studies of tiglic (5)
and angelic (16) acids in methanol solution supports a
+
H
Ru
O
D
Ru
O
Sol
+
H2
5
H+
H
P
Sol
O
P
Sol
O
P
Sol
O
P
Sol
Sol
Ru
O
*
*
*
MeOD
MeOH
Ru
Sol
*
P
P
P
P
H2
21-d
20
15
21
CO2H
D/H
H/D
6
5
K
CO2H
+
Sol
P
*
P
Sol
O
H/D
Solvolysis
Ru
O
*
O
P
Sol
O
+ or D+
*
Ru
Sol
H
P
*
H/D
H/D
22
23
Scheme 3. Proposed catalytic cycle for the hydrogenation of a,b-unsaturated acids using 15 as catalyst. This cycle is proposed to be generic for ruthenium
bis(phosphine) catalysts.

2764
C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
Scheme 4. Proposed mechanism to account for deuterium incorporation at Ha, Htrans, and Hd observed in angelic acid (16) hydrogenation with 15 as
catalyst.
Scheme 5. Proposed mechanism to account for deuterium incorporation at Ha, Hcis, and Hd observed in angelic acid (16) hydrogenation with 15 as
catalyst.
to the species 24 (Scheme 4). From this intermediate, ole-
fin-hydride insertion can occur through the si-face to form
the 5-membered metallacycle (R)-25.2 This species can
undergo b-deuteride elimination from either the a-position
(Ha) or the b-methyl group (Hd). The removal of the deu-
teride at the a-position (Ha) is a moot issue as it is the
reverse reaction of the olefin-deuteride insertion. Hydride
abstraction from the b-methyl group (Hd) will produce
(R)-26. H–D exchange with solvent would yield the deute-
ride (R)-26-d1, that then undergoes insertion to form (R)-
25-d1, with deuterium now located at the b-methyl position
(Hd) as observed (Scheme 4). Solvolysis of the ruthenium–
carbon r bond in (R)-25-d1, followed by ligand exchange
with 16 completes the catalytic cycle. We propose a net
isomerization of the C@C bond during the sequence of
steps shown in Scheme 4. Double bond isomerizations dur-
ing a catalytic transformation has precedent with a number
2
Defined as R configuration assuming replacement of ruthenium atom
by hydrogen atom as in final hydrogenation product.

C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
2765
of ruthenium bis(phosphine) catalysts [19].3 Isomerization
of the C@C bond of angelic acid (16) is not surprising as
its geometric isomer, tiglic acid (5), is thermodynamically
more stable [18]. Rapid, prior isomerization of angelic acid
to tiglic acid during this catalytic hydrogenation does not
go to completion, however, as the hydrogenation of angelic
acid proceeds with 46% ee (S), while the hydrogenation of
tiglic acid proceeds with 90% ee (R).4 These hydrogena-
tions would both occur with the same ee if they proceeded
this base is required for activity in an aprotic solvent sug-
gests that the role of triethylamine is to act as proton scav-
enger during the heterolytic cleavage of dihydrogen. The
role of the conjugate acid Et3NH+ is then likely to act as a
proton source for protonolysis of the ruthenium–carbon r
bond.
In a previous report, we showed that the stoichiometric
reaction between tiglic acid (5) and 15 in acetone proceeds
by immediate protonolysis of the Ru–H bond with forma-
tion of the Ru–tiglato complex 20 (Eq. (1)) [20]. The stoi-
chiometric reaction between the deuteride 15-d, prepared
by reaction between 15 and D2, and tiglic acid also forms
the Ru–tiglato complex (20), along with an equivalent of
H–D. The reaction between tiglic acid and one equivalent
of triethylamine in acetone solution, followed by reaction
through
a common intermediate formed via olefin
isomerization.
Scheme 5 shows how deuterium incorporation into the
b-methyl group can occur from (R)-26-d1. (R)-26-d1 can
undergo Ca–Cb bond rotation to form (R)-28-d1. Olefin-
deuteride insertion would form the 5-membered heterome-
tallacycle (R)-25-d1, with deuterium incorporation into the
b-methyl group (Hd) (Scheme 5). Solvolysis, followed by
exchange between the hydrogenated substrate (6) and 16
complete the catalytic cycle.
While these isotope substitution patterns do not dis-
count BrownÕs mechanism, we point out that should the
reaction proceed with dihydrogen oxidative addition
followed by insertion and reductive elimination, as BrownÕs
1
with one equivalent of 15 also yields 20 as shown by H
and 31P NMR.5 Further, reaction between 15 and tiglic
acid yields 20 in the protic solvent methanol. And finally,
20 is the only ruthenium complex detected on NMR anal-
ysis of the catalytic reactions (in all cases). These data are
strong evidence that the active catalyst is 20 under these
conditions.
ð1Þ
mechanism would require, one would not expect the use of
aprotic solvents to suppress the reaction. Indeed, the hydro-
genation of tiglic acid (5) was extremely slow in the aprotic
solvent acetone, yielding less than 10% conversion to 6
under the same conditions that gave 100% conversion in
methanol (4 atm H2, 25 ꢁC, 20 h). This observation favors
a mechanism similar to HalpernÕs that involves solvolysis
of a ruthenium–carbon r bond, and not a mechanism that
involves H2 oxidative addition followed by reductive elimi-
nation. Further, the hydrogenation of 5 in acetone did pro-
ceed in good rates to completion in the presence of one
equivalent of triethylamine per substrate acid group
(4 atm H2, 25 ꢁC, acetone; NEt3:5 = 1:1, 95% ee (R)). That
2.2. Ruthenium-BINAP catalyzed enantioselective
hydrogenation of a,b-unsaturated carboxylic esters
The above observations strongly support the general
Halpern mechanism (Scheme 1) for a,b-carboxylic acids.
We now discuss the mechanistic results we have obtained
using esters as substrates. In previous reports, we showed
that the mechanistic pathway operating with 15 as catalyst
for the hydrogenation of the a,b-carboxylic esters MAC
(13) [13–15] and MAA (14) [12] is different from the Halp-
ern mechanism. MAC is hydrogenated by 15 with 92% ee
(R) in acetone and with 87% ee (R) in methanol solvents
5
It is postulated that the coordination geometry about the ruthenium
3
Examples of isomerizations of olefinic substrates: Wiles et al. [16];
center is in the D configuration on the basis of the X-ray structures and
steric arguments previously reported for the related complexes Ru((R)-
BINAP)(O2CCMe@CHMe)2 and Ru((S)-BINAP)(O2CC(Me)3)2. It is
believed that the relative configuration of the phenyl rings on the
phosphines of the BINAP ligand impose a steric environment that
preferentially coordinates the tiglato ligands in a D configuration for (R)-
BINAP and a K configuration for (S)-BINAP complexes.
Saburi et al. [10].
4
In contrast, the products of hydrogenation of (Z)- and (E)-MAC (13)
were formed in similar ee (and formation of the same major enantiomer,
R), implying that the hydrogenation of the thermodynamically less
favored isomer (E)-MAC proceeds via a rapid prior isomerization to (Z)-
MAC. See [13] for details.

2766
C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
Scheme 6. Structural identification of catalytic intermediate (29) in ruthenium-BINAP catalyzed enantioselective hydrogenation of MAC (13).
Table 3
[20,21]. Use of MeOH-d4 did not result in significant
Deuterium incorporation in hydrogenation of 7 and DMI (17) with 31
amounts of deuterium labeling of the product, showing
(BrownÕs catalyst) and 15 as catalyst, respectively
the mechanism did not involve solvolysis of a ruthenium–
carbon r bond.
R
MeO2C
For these hydrogenations, only one ruthenium species
CO2Me
was observed in 31P NMR spectroscopic studies of the
operating catalytic reaction. The stoichiometric reaction
of [Ru((R)-BINAP)(H)(Sol)n(MeCN)3 ꢀ n](BF4) (n = 0–3;
Sol = acetone, 15) and MAC resulted in a single species
with identical NMR characteristics to those of the species
observed in the operating catalytic reaction. Isolation and
complete identification (NMR spectroscopic and X-ray
crystallographic analyses) showed it to be the single diaste-
reomer (RuMAC; 29) formed via olefin-hydride insertion
into the Ru–H bond (Scheme 6) [13–15]. Similar results
were observed with MAA [12] as substrate, except that
the stoichiometric reaction of 15 with MAA results in
two diastereomers in 72:28 ratio. Isolation and complete
identification (solution NMR spectroscopic and X-ray
crystallographic analyses) of the major diastereomer
(RuMAA; 30) [12] showed the insertion to have occurred
from the same face as with RuMAC (29) [15]. For both
these substrates then, olefin-hydride insertion positions
the hydride at the b-position, not at the a-position as
observed with tiglic, angelic, and other a,b-unsaturated
carboxylic acid substrates. Also the major diastereomer
formed upon stoichiometric reaction with 15 and both
these substrates corresponded to the major enantiomers
formed by the catalytic hydrogenation. These results, with
others obtained previously, showed that the hydrogena-
tions of MAC and MAA proceed via reaction of 15 and
substrate to make an olefin adduct, olefin-hydride insertion
to make the ruthenium-alkyl complexes 29 and 30, fol-
lowed by hydrogenolysis of the ruthenium–carbon r bonds
in 29 and 30 to complete the hydrogenations and regener-
ate 15. This mechanism more closely resembles that pro-
posed by Brown than HalpernÕs.
H
α
CH3
β
8 R = CH3 18 R = H
Catalyst, conditions, substratea
Ha
Hb
15, (H2, MeOD), 17
15, (D2, MeOH), 17
31, (H2, MeOD), 7
31, (D2, MeOH), 7
a
Reaction conditions. Catalyst 15, [Ru] = 2.6 mM, 3 atm gas pressure,
ambient temperature. Catalyst 31 (Brown; Ru(BINAP)(g2-MeC(O)CH-
C(O)Me)2(g2-allyl)), 1 atm gas pressure, ambient temperature.
30
65
10
70
–
80
10
60
Deuterium-labeling studies were performed and the results
are listed in Table 3 along with the proton assignments.
Under 3 atm of dihydrogen gas at 25 ꢁC in methanol-d4,
deuterium incorporation occurred solely and to 30%
extend at Ha of the product 18. Under 3 atm of dideute-
rium gas at 25 ꢁC in methanol, the deuterium incorpora-
tion was largely reversed, with deuterium at Ha (ꢂ65%)
and at Hb (ꢂ80%, assuming formation of CH2D) [22].
These results are in agreement with those observed by
Brown, where significant deuterium incorporation from
dideuterium gas is found at both the a-position (Ha)
(Brown: 70% D; here: 65% D) and b-methyl position
(Hb) (Brown: 60%; here: 80%) [22].
31P NMR analysis of the operating catalytic hydrogena-
tion of DMI (17) showed the presence of only two species
in detectable quantities. Further, the stoichiometric reac-
tion of DMI with 15 yielded the same two species, with
31P NMR spectra consistent with those observed with the
catalytic hydrogenation. Analysis of these NMR data
along with comparison to those of RuMAC (29) and
RuMAA (30) allows identification of these species as the
olefin-hydride inserted diastereomers 32a and 32b. Fig. 1
depicts the major isomer; 32a with is structurally analogous
to RuMAC (29) and RuMAA (30).
We investigated the mechanism for hydrogenation of the
diester dimethyl itaconate (DMI; 17) with the catalyst 15
because DMI has the same distribution of carbonyl and
olefin groups as MAC (13), MAA (14), and BrownÕs sub-
strate (7). DMI is hydrogenated in 95% ee (S) in both
methanol or acetone solution using 15 as catalyst [20].

C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
4. Experimental
2767
4.1. Methods and materials
The solvents tetrahydrofuran (K, Ph2CO), diethyl ether
˚
(K, Ph2CO), triethylamine (CaH2), acetone (3 A sieves),
and methanol (Mg) were distilled from drying agents under
argon gas. The argon and dinitrogen gases were passed
through a bed of Drierite drying agent. Unless stated other-
wise, commercial reagents were used without further puri-
fication, except dimethyl itaconate (distilled under argon
atmosphere), and all operations were performed under an
inert atmosphere of argon or dinitrogen gas. The com-
pounds [Ru((R)-BINAP)(1-3;5,6-g-C8H11)(MeCN)](BF4)
(33)7 and angelic acid (16) [23] were synthesized via the lit-
erature procedures.
Fig. 1. Isolation and structural identification of putative catalytic inter-
mediate (major species 32a) in ruthenium bis(phosphine) catalyzed
enantioselective hydrogenation of 17.
All 1H, 13C, and 31P NMR spectra were measured with a
Bruker AM-400 NMR spectrometer operating at 400.13,
The stoichiometric hydrogenolysis of the ruthenium–
carbon r bonds in the mixture of diastereomers 32a and
32b produced the hydrogenation product, 18, in 92% ee
(S). This stoichiometric ee is nearly identical to the catalytic
ee obtained with the hydrogenation of DMI, 95% ee (S).
We observed the same similarities between the stoichiome-
tric and catalytic eeÕs for the analogous ruthenium-alkyl
intermediates RuMAC (29) and RuMAA (30). These
investigations have thus yielded results that largely parallel
those we obtained previously with the hydrogenations of
MAC and MAA. These results are thereby strong evidence
that 32a and 32b are intermediates in the catalytic hydroge-
nation of DMI, and that the hydrogenation of DMI pro-
ceeds by the same mechanism as MAC and MAA. Not
by the mechanism proposed by Halpern.6
1
100.61, and 161.97 MHz, respectively. H and 13C NMR
chemical shifts are reported in parts per million (d) relative
to tetramethylsilane using the solvent as an internal refer-
ence. 31P NMR chemical shifts are reported in parts per mil-
lion (d) relative to an 85% H3PO4 external reference. All 13
C
and 31P NMR spectra are 1H decoupled unless stated other-
wise. Mass spectra were measured using Kratos MS50.
Microanalyses were performed at the University of Alberta
Microanalysis Laboratory. Optical rotations were mea-
sured with a Perkin–Elmer 241 polarimeter at 589 nm
(sodium D line) using 1.0 dm cells. Specific rotations, [a]D,
are reported in degrees per decimeter at 25 ꢁC, and the con-
centration (c) is given in grams per 100 mL.
3. Conclusion
4.2. Hydrogenations of tiglic acid (5) with D2 in methanol or
H2 in methanol-d4 using 33 as catalyst precursor
The mechanism of the ruthenium bis(phosphine) cata-
lyzed enantioselective hydrogenation of olefins has been
complicated by discrepancies in deuterium-labeling studies
reported in the literature for various substrates. Here, we
have shown that two mechanisms can account for the
observed differences. In each case, 5-membered metallocy-
cle intermediates are formed via olefin-hydride insertions.
However, the nature of the substrate affects the choice of
mechanistic pathway utilized primarily in the activation
of dihydrogen. a,b-Unsaturated acids, are prone to pro-
ceed with heterolytic cleavage of dihydrogen gas while
the a,b-unsaturated esters are prone to proceed with homo-
lytic cleavage of dihydrogen gas. The favored pathway may
result from electronic factors, from the disposition of the
functional group in the substrate, or from a combination
of both. A complete investigation using a wide variety of
olefins must be performed in order to determine which fac-
tors prejudice which mechanistic pathway.
The catalyst precursor 33 (10.0 mg, 1.04 · 10ꢀ5 mol) was
transferred to a glass bomb along with 100 equiv of tiglic
acid (5) (104.1 mg, 1.04 · 10ꢀ3 mol) in the glove box. The
dry, deoxygenated methanol, or methanol-d4, was then
added (4.0 mL) and the solution was stirred for 5 min.
The atmosphere was then flushed with dihydrogen gas, or
dideuterium gas, and once flushed the bomb was sealed
under 3 atm of dihydrogen or dideuterium. The solution
was stirred for 20 h at 25 ꢁC to complete hydrogenation.
Once completed, the solvent was removed under reduced
pressure, and the product was dissolved in CH2Cl2 and
passed through a Florisil plug to remove the catalyst.
The solvent was then removed under reduced pressure
and the product analyzed by NMR.
Residue from hydrogenation with H2 in methanol-d4:
1H (400.1 MHz, acetone-d6): d 0.9(ÔdÕ, Hd, 0% D incorp.),
1.10 (ÔsÕ, He, 0% D incorp.), 1.43 (m, Htrans, 0% D incorp.),
1.62 (m, Hcis, 85% D incorp.), 2.34 (m, Hc, 70.5% D
incorp.), 10.45 (br s, Hacid).
6
The 5-membered metalloheterocycle is formed from the substrate C1
through to the b-carbonyl (amido carbonyl of MAC and MAA) of
dimethyl itaconate. The 4-membered metalloheterocycle is formed from
the C1 through to the a-ketone carbonyl.
7
Examples of isomerizations of olefinic substrates: Wiles et al. [16].

2768
C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
1
5
Residue from hydrogenation with D2 in methanol: H
(400.1 MHz, acetone-d6): d 0.9 (ÔtÕ, Hd, 0% D incorp.),
1.10 (ÔdÕ, He, 0% D incorp.), 1.43 (m, Htrans, 0% D incorp.),
1.62 (m, Hcis, 7.5% D incorp.), 2.34 (m, Hc, 33% D incorp.),
10.45 (br s, Hacid).
3JH–H = 7.2 Hz, JH–H = 1.2 Hz), 2.01 (s, CH3CN), 2.03
3
(s, CH3CN), 6.45 (dq, CH3CHC(CH3)CO2Ru, JH–H
=
4
7.2 Hz, JH–H = 1.5 Hz), 6.58–7.89 (m, 32 H, BINAP).
31P (161.9 MHz, acetone-d6):
d
52.6 (dd, JP–P =
2
37.5 Hz), 56.9 (d, JP–P = 37.5 Hz). 13C (100.6 MHz, ace-
tone-d6): d 2.4 (CH3CN), 4.65 (CH3CN), 10.72 and 13.90
(CH3CHC(CH3)CO2Ru), 123.99 (CH3CN), 124.14
(CH3CN), 124.42–141.45 (BINAP and CH3CHC-
(CH3)CO2Ru), 185.46 (CH3CHC(CH3)CO2Ru).
2
4.3. Hydrogenations of angelic acid (16) with D2 in
methanol or H2 in methanol-d4 using 33 as catalyst precursor
The catalyst precursor 33 (12.0 mg, 1.25 · 10ꢀ5 mol) was
transferred to a glass bomb along with 100 equiv of angelic
acid (16) (125.5 mg, 1.25 · 10ꢀ3 mol). The dry, deoxygen-
ated methanol, or methanol-d4, was then added (4.8 mL)
and the solution was stirred for 5 min. The atmosphere
was then flushed with dihydrogen gas, or dideuterium
gas, and once flushed the bomb was sealed under 3 atm
of dihydrogen or dideuterium. The solution was stirred
for 20 h at 25 ꢁC to complete hydrogenation. Once com-
pleted, the solvent was removed under reduced pressure,
and the product was dissolved in CH2Cl2 and passed
through a Florisil plug to remove the catalyst. The solvent
was then removed under reduced pressure and the product
analyzed by NMR.
4.5. Stoichiometric reaction of [Ru((R)-BINAP)(H)-
(MeCN)n (Sol)3 ꢀ n](BF4) (n = 0–3) with tiglic acid in
methanol-d4
Compound 33 (20.0 mg, 2.09 · 10ꢀ5 mol) was partially
dissolved in methanol-d4 (ꢂ0.8 mL) in an NMR tube
under an argon atmosphere. At room temperature, the
tube was flushed with H2, pressurized (1–2 atm), and sha-
ken until a deep orange-yellow solution was generated
(ꢂ5 min). The H2 atmosphere was replaced by argon
gas and the solution was then transferred to an NMR
tube containing tiglic acid (2.10 mg, 2.09 · 10ꢀ5 mol)
under an argon atmosphere. To this solution was added
excess CD3CN (2.0 lL, 3.83 · 10ꢀ5 mol) via a gas-tight
syringe. The solution was immediately analyzed by 1H
NMR. The product was found to be identical to that
formed upon stoichiometric reaction of tiglic acid with
15 in acetone solution.
Residue from hydrogenation with H2 in methanol-d4:
1H (400.1 MHz, acetone-d6): d 0.9 (ÔdÕ, Hd, 43% D incorp.),
1.10 (ÔsÕ, He, 0% D incorp.), 1.43 (m, Htrans, 62% D incorp.),
1.62 (m, Hcis, 8% D incorp.), 2.34 (m, Hc, 43% D incorp.),
10.45 (br s, Hacid).
1
Residue from hydrogenation with D2 in methanol: H
(400.1 MHz, acetone-d6): d 0.9 (ÔdÕ, Hd, 5% D incorp.),
1.10 (ÔsÕ, He, 0% D incorp.), 1.43 (m, Htrans, 7.5% D
incorp.), 1.62 (m, Hcis, 5% D incorp.), 2.34 (m, Hc, 30%
D incorp.), 10.45 (br s, Hacid).
4.6. Hydrogenations of dimethyl itaconate (17) with D2 in
methanol or H2 in methanol-d4 using 15 as catalyst
The catalyst precursor 33 (11.5 mg, 1.20 · 10ꢀ5 mol) was
transferred to a glass bomb along with 75 equiv of dimethyl
itaconate 17 (126 lL, 9.00 · 10ꢀ4 mol) via gas-tight syr-
inge. The methanol, or methanol-d4, was then added
(4.6 mL) and the solution was stirred for 5 min. The atmo-
sphere was then flushed with dihydrogen gas, or dideute-
rium gas, and once flushed the bomb was sealed under
3 atm of dihydrogen or dideuterium. The solution was stir-
red for 20 h at 25 ꢁC to complete hydrogenation. Once
completed, the solvent was removed under reduced pres-
sure, and the product was dissolved in CH2Cl2 and passed
through a Florisil plug to remove the catalyst. The solvent
was then removed under reduced pressure and the product
analyzed by NMR.
4.4. Stoichiometric reaction of [Ru((R)-BINAP)(H)
(MeCN)n (Sol)3 ꢀ n](BF4) (n = 0–3) with tiglic acid and 1
equivalent of NEt3 in acetone-d6
Compound 33 (20.0 mg, 2.09 · 10ꢀ5 mol) was partially
dissolved in acetone-d6 (ꢂ0.6 mL) in an NMR tube under
an argon atmosphere. At room temperature, the tube was
flushed with H2, pressurized (1-2 atm), and shaken until a
deep orange-yellow solution was generated (ꢂ5 min). The
H2 atmosphere was replaced by argon gas and the solu-
tion was then transferred to an NMR tube containing tig-
lic acid (2.1 mg, 2.09 · 10ꢀ5 mol) and NEt3 (2.9 lL,
2.1 mg, 2.09 · 10ꢀ5 mol) in acetone-d6 (0.2 mL) under an
argon atmosphere. To this solution was added excess
CD3CN (2.0 lL, 3.83 · 10ꢀ5 mol) via a gas-tight syringe.
The solution was immediately analyzed by 1H NMR.
The product was found to be identical to that formed
upon stoichiometric reaction of tiglic acid with 15 in the
absence of added NEt3 in acetone solution. The product
readily loses MeCN in vacuo. NMR spectroscopic data
of the complex: 1H (400.1 MHz, acetone-d6): d 1.45
Residue from hydrogenation with H2 in methanol-d4:
1H (400.1 MHz, CD2Cl2): d 1.18 (d, Hb, 0% D incorp.),
2.46 (dd, 0% D incorp.), 2.68 (dd, 0% D incorp.), 2.88
(m, Ha, 30% D incorp.), 3.62 (s, OCH3, 0% D incorp.),
3.63 (s, OCH3, 0% D incorp.).
1
Residue from hydrogenation with D2 in methanol: H
(400.1 MHz, CD2Cl2): d 1.18 (d, Hb, 80% D incorp. assum-
ing CH2D), 2.46 (dd, 0% D incorp.), 2.68 (dd, 0% D
incorp.), 2.88 (m, Ha, 65% D incorp.), 3.62 (s, OCH3, 0%
D incorp.), 3.63 (s, OCH3, 0% D incorp.). Actual amounts
4
(apparent t, CH3CHC- (CH3)CO2Ru, JH–H = 1.5 Hz,
5JH–H = 1.2 Hz), 1.56 (dd, CH3CHC(CH3)CO2Ru,

C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
2769
of product were determined by 13C NMR chemical shifts
based on reported assignments: 18 (5%), 18-a-d1 (15%),
18-b-d1 (30%), 18-a,b-d2 (50%).15
C53H46O4NP2Ru: C, 62.98; H, 4.59; N, 1.39. Found: C,
62.47; H, 4.67; N, 1.31%.
4.8. Stoichiometric hydrogenation of [Ru((R)-BINAP)-
(MeCN)(C[CH2(D)]-[CO2CH3][CH2CO2CH3])]-
(BF4) (32a-d1/32b-d1)
4.7. Synthesis of [Ru((R)-BINAP)(MeCN)-
(C[CH3][CO2CH3] [CH2CO2CH3])](BF4) (RuDMI;
32a/32b)
To a glass bomb was added 32a-d1/32b-d1 (100.8 mg,
9.96 · 10ꢀ5 mol), made exactly like 32a/32b only with
dideuterium gas replacing dihydrogen gas. The bomb was
flushed with argon gas for 20 min to ensure no oxygen
gas was present. To the bomb was added acetone
(7.5 mL) via gas-tight syringe. The bomb was then flushed
with dihydrogen gas for 2 min, then it was sealed under
4 atm of dihydrogen gas. The reaction was stirred at room
temperature for 10 min and then the bomb was depressur-
ized. The solution was transferred to a flask and the solvent
was removed under reduced pressure. The residue was then
dissolved in CH2Cl2 (0.2 ml) and passed through a Florisil
plug with Et2O (ꢂ5 mL) to remove the catalyst. NMR
analysis showed complete conversion to dimethyl methyl-
succinate in 93% ee (S), with D–H exchange (ꢂ35%) occur-
ring at the b-methyl position corresponding to reversibility
of formation of 32a/32b on time-scale of irreversible
hydrogenolysis of Ru–Ca bond. Similar reversibility was
observed in the studies with 14 as substrate and 15 as
catalyst.
To a 50 mL solvent Schlenk flask was added 33
(102.2 mg, 1.07 · 10ꢀ4 mol). The Schlenk flask was placed
under reduced pressure and purged with argon gas several
times to remove traces of oxygen. To the Schlenk flask was
added acetone (7.5 mL) via gas-tight syringe. The solution
was stirred for 5 min then dihydrogen gas was bubbled
through the solution until complete hydrogenation of 33
was achieved (2 min). The solution was then bubbled with
argon gas to remove all traces of excess dihydrogen gas
(5 min). Once all traces of dihydrogen gas were removed,
dimethyl itaconate (17) was added (17.4 lL, 1.23 ·
10ꢀ4 mol) via gas-tight syringe. The golden orange solution
turned bright orange-yellow immediately. The flask was
shaken for 5 min, then it was placed under reduced pres-
sure to remove the solvent. The residue was dissolved in
a minimal amount of acetone (ꢂ1.0 mL) and then precipi-
tated out of solution with Et2O (40 mL). The solution was
filtered to dryness, the solid washed with Et2O (3 · 10 mL),
dried under a stream of argon gas (20 min), and finally
dried under high vacuum overnight. The sample was then
transferred to a vial and stored at ꢀ30 ꢁC in the glove
box: yield = 75.6 mg (70.2%). Attempts to obtain X-ray
quality crystals by slow diffusion of a variety of solvents
(CH2Cl2/Et2O, CH2Cl2/pentane, CH2Cl2/n-Bu2O, ace-
tone/Et2O, acetone/n-Bu2O, MeOH/Et2O, MeCN/Et2O,
benzene/Et2O) were unsuccessful. 1H NMR of major prod-
Acknowledgment
This work was supported by the Natural Sciences and
Engineering Research Council of Canada and by the Uni-
versity of Alberta.
uct (85%) (R)-32a (400 MHz, 25 ꢁC, acetone-d6): d 0.56
2
References
(d, 1H, Ru-C(CH3)(CO2CH3)(CH2CO2CH3), JH–H
=
19.5 Hz), 1.49 (d, 3H, RuC(CH3)(CO2CH3)(CH2CO2CH3),
4JH–Pa = 5.8 Hz), 1.95 (s, 3H, CH3CN), 2.28 (dd,
1H, RuC(CH3)(CO2CH3)(CH2CO2CH3), 2JH–H = 19.5
[1] R. Noyori, Angew. Chem., Int. Ed. Engl. 41 (2002) 2008.
[2] R. Noyori, Asymmetric Catalysis in Organic Synthesis, Wiley, New
York, 1994.
4
Hz, JH–Pa = 6.0 Hz), 3.83 (s, 3H, OCH3), 4.00 (s, 3H,
[3] R. Noyori, Tetrahedron 50 (1994) 4259.
OCH3), 6.45–8.15 (m, 32H, BINAP, overlap with
minor). 13C (400 MHz, 25 ꢁC, CD2Cl2): d 4.6 (s, CH3CN),
20.7 (s, RuC(CH3)(CO2CH3)(CH2CO2CH3)), 38.4 (dd,
[4] H. Takaya, T. Ohta, R. Noyori, in: I. Ojima (Ed.), Catalytic
Asymmetric Synthesis, VCH, New York, 1993 (Chapter 1).
[5] M.T. Ashby, J. Halpern, J. Am. Chem. Soc. 113 (1991) 589.
[6] T. Ohta, H. Takaya, R. Noyori, Tetrahedron Lett. 31 (1990)
7189.
RuC(CH3)(CO2CH3)(CH2CO2CH3),
2JC–P = 42.0 Hz,
2JC–P = 3.0 Hz), 40.8 (d, RuC(CH3)(CO2CH3)(CH2-
[7] K. Yoshikawa, M. Murata, N. Yamamoto, K. Inoguchi, K. Achiwa,
Chem. Pharm. Bull. 40 (1992) 1072.
3
CO2CH3), JC–P = 4.0 Hz), 52.8 (s, OCH3), 55.8 (s,
[8] A.S.C. Chan, C.C. Chen, T.K. Yang, J.H. Huang, Y.C. Lin, Inorg.
Chim. Acta 234 (1995) 95.
[9] J.M. Brown, M. Rose, F.I. Knight, A. Wienland, Recl. Trav. Chim.
Pays-Bas 114 (1995) 242.
[10] M. Saburi, H. Takeuchi, M. Ogasawara, T. Tsukahara, Y. Ishii, T.
Ikariya, T. Takahashi, Y. Uchida, J. Organomet. Chem. 428 (1992)
155.
[11] C.J.A. Daley, S.H. Bergens, J. Am. Chem. Soc. 124 (2002)
3680.
[12] J.A. Wiles, S.H. Bergens, V.G. Young Jr., Can. J. Chem. 79 (2001)
1019.
[13] J.A. Wiles, S.H. Bergens, Organometallics 18 (1999) 3709.
[14] J.A. Wiles, S.H. Bergens, Organometallics 17 (1998) 2228.
OCH3), 126.4–142.8 (m, BINAP and CH3CN), 167.2 (s,
3
3
CO2CH3, JC–P = 3 Hz), 190.0 (d, CO2CH3, JC–P
=
6.5 Hz). 31P (MHz, 25 ꢁC, acetone-d6): d 34.64 (d, 1Pa,
2
2JP–P = 21.5 Hz)*, 56.65 (d, 1Pb, JP–P = 21.5 Hz). Note,
Pa is trans to the Ru–C and Pb is cis to the Ru–C. Complete
identification of 1H and 13C of minor complex (15%,
(S)-32b) could not be performed due to lack of concen-
tration and overlap with major product signals. 31P
2
(MHz, 25 ꢁC, acetone-d6): d 40.8 (d, JP–P = 36.5 Hz),
2
63.6 (d, JP–P = 36.5 Hz). ESIMS. Calc. for C53H46O4N-
P2Ru (M ꢀ BF4)+: 924.194. Found: 924.1. Anal. Calc. for

2770
C.J.A. Daley et al. / Inorganica Chimica Acta 359 (2006) 2760–2770
[15] J.A. Wiles, S.H. Bergens, V.G. Young Jr., J. Am. Chem. Soc. 119
(1997) 2940.
(b) Y. Sun, C. LeBlond, J. Wang, D.G. Blackmond, J. Laquidara,
J.R. Sowa jr., in: R.E. Malz (Ed.), Catalysis of Organic Reactions,
Marcel Dekker, New York, 1996, p. 167;
(c) B.M. Trost, R.J. Kulawiec, J. Am. Chem. Soc. 115 (1993) 2027;
(d) T. Takaya, T. Ohta, S. Inoue, M. Tokunaga, M. Kitamura, R.
Noyori, Org. Synth. 72 (1993) 74;
(e) H. Takaya, T. Ohta, N. Sayo, H. Kumobayashi, S. Akutagawa, S.
Inoue, I. Kasahara, R. Noyori, J. Am. Chem. Soc. 109 (1987) 1596;
(f) S. Inoue, M. Osada, K. Koyano, H. Takaya, R. Noyori, Chem. Lett.
(1985) 1007.
[16] As we reported previously, there is one MeCN per Ru in solutions of
15 (prepared via hydrogenolysis of [Ru((R)-BINAP)(1–3;5,6-g-
C8H11)(MeCN)]BF4). Because the MeCN ligand exchanges rapidly
among all of the Ru centers in solution at room temperature, we
denote this mixture of hydrides generally as [Ru((R)-BINAP)(H)
(MeCN)n(sol)3 ꢀ n]BF4]. J.A Wiles, C.E. Lee, R. McDonald, S.H.
Bergens, Organometallics 15 (1996) 3782.
[17] Synthesis T. Ohta, H. Takaya, M. Kitamura, K. Nagai, R. Noyori,
J. Org. Chem. 52 (1987) 3174.
[20] C.J.A. Daley, J.A. Wiles, S.H. Bergens, Can. J. Chem. 76 (1998)
1447.
[21] J.A. Wiles, C.J.A. Daley, R.J. Hamilton, C.G. Leong, S.H. Bergens,
Organometallics 23 (2004) 4564.
[18] (a) Thermodynamically less stable than tiglic acid R.E. Buckles, G.V.
Mock, J. Am. Chem. Soc. 72 (1950) 680;
(b)
The Merck Index: S. Budavari, J. OÕNeil, A. Smith, P.E.
Heckelman (Eds.), 11th ed. Merck & Co. Inc., New Jersey, 1989, p.
102 (angelic acid), p. 1486 (tiglic acid).
[19] (a) Examples of isomerizations with allylic alcohols Y. Sun, J. Wang,
C. LeBlond, R.A. Reamer, J. Laquidara, J.R. Sowa Jr., D.G.
Blackmond, J. Organomet. Chem. 548 (1997) 65;
[22] The assumption of only CH2D being formed is appears to be correct.
The 13C NMR data show that there is solely formation of CH2D
based on the 13C chemical shift data reported D.J. Hardick, I.S.
Blagbrough, V.L. Potter, J. Am. Chem. Soc. 118 (1996) 5897.
[23] R.E. Buckles, G.V. Mock, J. Org. Chem. 15 (1950) 680.