An intramolecular N-H?Co hydrogen bond and a structure correlation study of the pathway for protonation of the Co(CO)3L- anion (L = CO, PR3)

Article abstract of DOI:10.1016/S0022-328X(00)00377-6

A low temperature X-ray crystal structure of the zwitterionic organometallic compound Co(CO)₃{PPh₂(C₆H₄NHMe₂)} (2) provides the first example of an intramolecular N-H?Co (d¹⁰) hydrogen bond. Synthesis and characterization of such intramolecularly hydrogen-bonded systems have proven much more elusive than their counterparts that exist as intermolecularly hydrogen-bonded salts. The molecular structure of 2 taken together with a series of previously characterized N-H?M (d¹⁰) hydrogen-bonded salts provides a series of snapshots of the changes in geometry at the metal center as one progresses along the pathway for protonation of the anion Co(CO)₃L^- (L = CO, PR₃) to yield the metal hydride HCo(CO)₃L.

Full text of DOI:10.1016/S0022-328X(00)00377-6

www.elsevier.nl/locate/jorganchem
Journal of Organometallic Chemistry 609 (2000) 36–43
An intramolecular NꢀH···Co hydrogen bond and a structure
correlation study of the pathway for protonation of the
Co(CO)3L− anion (L=CO, PR3)
Lee Brammer *, Juan C. Mareque Rivas, Christopher D. Spilling
Department of Chemistry, Uni6ersity of Missouri-St. Louis, 8001 Natural Bridge Road, St. Louis, MO 63121-4499, USA
Received 25 April 2000; received in revised form 6 June 2000
Abstract
A low temperature X-ray crystal structure of the zwitterionic organometallic compound Co(CO)3{PPh2(C6H4NHMe2)} (2)
provides the first example of an intramolecular NꢀH···Co (d10) hydrogen bond. Synthesis and characterization of such
intramolecularly hydrogen-bonded systems have proven much more elusive than their counterparts that exist as intermolecularly
hydrogen-bonded salts. The molecular structure of 2 taken together with a series of previously characterized NꢀH···M (d10)
hydrogen-bonded salts provides a series of snapshots of the changes in geometry at the metal center as one progresses along the
pathway for protonation of the anion Co(CO)3L− (L=CO, PR3) to yield the metal hydride HCo(CO)3L. © 2000 Elsevier Science
S.A. All rights reserved.
Keywords: Structure correlation study; Protonation pathway; Intramolecular NꢀH···Co bond
reverse hydrogen bond D···HꢀA may be able to form.
By contrast, proton transfer arising from a DꢀH···M
hydrogen bond results in a metal hydride species
(MꢀH) that is unable to form the reverse hydrogen
bond, D···HꢀM [5], due to the hydridic nature of the
hydrogen atom. Much of our own work in this area has
involved the study of NꢀH···Co hydrogen bonds in salts
of the type R3NH+Co(CO)3L− (L=CO [3b,c,e,g], tri-
arylphosphine [3e,f]). Consistent with the above com-
ments, and pertinent to our own work, is the
observation by Norton, Sweany and co-workers that,
despite its notable thermodynamic acidity, HCo(CO)4
does not form CoꢀH···A hydrogen bonds [6]. A further
distinction between DꢀH···M hydrogen bonds and
more conventional hydrogen bonds is evident when one
considers the contributions to the hydrogen bond en-
ergy. A recent report describes ab initio (MP2 level)
calculations on OꢀH···PtL4 hydrogen bonds and sug-
gests that, in contrast to conventional hydrogen bonds,
there is a substantial dispersion contribution to the
hydrogen bond energy [4n]. Our own ab initio calcula-
tions on a variety of DꢀH···M hydrogen bonds suggest
a greater polarization contribution than is present in
conventional hydrogen bonds [7].
1. Introduction
Over the past decade in particular there has been an
emergence of research concerning an array of hydrogen
bond types variously categorized as ‘weak’ or ‘non-con-
ventional’ [1], the archetype of which is the CꢀH···O
hydrogen bond [2]. A class of non-conventional hydro-
gen bonds that arises only in inorganic systems, notably
within organometallic and coordination chemistry, is
the DꢀH···M [3,4] hydrogen bond (D=hydrogen bond
donor, A=hydrogen bond acceptor, M=transition
metal). In earlier work we have established spectro-
scopic, geometric and chemical criteria for identifying
DꢀH···M hydrogen bonds [3a,d]. In many respects they
resemble conventional hydrogen bonds. An important
difference arises when one considers the hydrogen bond
as an incipient proton transfer reaction. In a conven-
tional hydrogen bond, DꢀH···A, the hydrogen atom is
protonic in nature before, during and after transfer
from hydrogen bond donor to acceptor. Indeed, the
* Corresponding author. Tel.: +1-314-5165345; fax: +1-314-
5165342.
E-mail address: lee.brammer@umsl.edu (L. Brammer).
0022-328X/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved.
PII: S0022-328X(00)00377-6

L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
37
In earlier work on hydrogen-bonded R3NH+
Co(CO)3L− salts, the focus was entirely on intermolec-
ular hydrogen bonds [3b,c,e,f]. The present report
describes efforts to prepare an intramolecularly hydro-
gen-bonded analog of these systems. While intramolec-
ular DꢀH···M hydrogen bonds are known for square
planar d8 metal centers [4c,d,f,g] and for d6 metallo-
carbinols [4h–j], prior to this report there had been
only one such example involving a tetrahedral d10 metal
center [4b]. Here we report the reaction of an N,P-
an additional hour in the ice bath and then was allowed
to warm to r.t. (20 min). The mixture was re-cooled in
ice and the reaction was cautiously quenched by the
slow addition of a large excess of water (5 ml). The
reaction mixture was partitioned between 2 M HCl (100
ml) and Et2O (100 ml). The Et2O was re-extracted with
HCl (100 ml) and the combined HCl extracts were
adjusted to pH 12 with NaOH. The resulting aqueous
solution was extracted with CH2Cl2 (3×75 ml) and the
combined extracts were dried, filtered and evaporated
in vacuo to give an orange oil. (7.53 g, 94%). 31P-NMR
l −14.6 p.p.m. indicates 95% purity in accord with a
previous literature report [9b]. The X-ray crystal struc-
ture of the salt 1·HCl has been determined, but will be
reported elsewhere.
chelating ligand PPh2(o-C6H4CH2NMe2)
1
with
HCo(CO)4 in which displacement of a CO ligand by the
phosphine coupled with transfer of the metal-bound
hydrogen to the amine yields the desired hydrogen-
bonded system, 2, which has been characterized by low
temperature X-ray diffraction.
Accumulated structural data on a series of NꢀH···Co
hydrogen-bonded salts, including 2, have then been
used to map out the reaction pathway for protonation
of the Co(CO)3L− anion to yield the hydride complex
HCo(CO)3L (L=CO, PR3).
2.3. Synthesis of
Co(CO)3{PPh2(o-C6H4CH2NHMe2)} (2)
Co2(CO)8 (1.52 g, 4.4 mmol) was dissolved in hexane
(35 ml) and to the stirring solution at r.t. an excess of
dimethylformamide (8 ml) was added. For 1 h the
mixture was stirred after which two layers appeared, a
colorless upper layer and a pink lower layer. The
solution was then cooled to −78°C and 8 ml of 1 M
HCl was slowly added. After addition was completed
the mixture was allowed to warm to 0°C and stirred for
20 min. After that time, the organic layer turned yellow
(HCo(CO)4) and the aqueous layer blue (CoCl2). The
aqueous layer was removed and the yellow solution was
washed with two equal portions of water (10 ml) and
afterwards dried over P2O5 at −78°C for 30 min. The
dried solution containing HCo(CO)4 was slowly added
to a solution of 1 (1.4 g, 4.4 mmol) in toluene (15 ml)
at −196°C. After the addition was completed, the
mixture was allowed to warm to −78°C and stirred for
15 min, yielding an orange solution and orange precip-
itate. The orange solution was transferred to another
flask and slow evaporation overnight yielded yellow
crystals of 2 suitable for X-ray diffraction, during
which time the solution turned red.
2. Experimental
2.1. General synthetic procedures and instrumentation
All manipulations of oxygen- and/or water-sensitive
materials were carried out under an atmosphere of
argon either using Schlenk-line techniques or in a Vac-
uum Atmospheres drybox. NMR spectra were recorded
on a Varian Unity-Plus 300 MHz or Bruker ARX-500
spectrometer.
The orange precipitate was dried under vacuum to
give an orange solid. This solid contained a mixture of
species, which proved impossible to separate. H-NMR
2.2. Synthesis of
o-diphenylphosphino-N,N-dimethylbenzylamine (1)
1
data for this material in C6D5CD3 revealed resonances
indicative of NꢀH protons (in the range l 10–14),
suggesting the possible presence of 2 together with
other related products. Dissolution of the yellow crys-
tals of 2 repeatedly yielded NMR spectra consistent
only with the ligand 1. Ultimately, definitive
spectroscopic characterization of 2 did not prove feasi-
ble, despite numerous repetitions of the synthesis. One
common by-product, the red dimer [Co(CO)3-
{PPh2(o-C6H4CH2NMe2)}]2 (3) with monodentate
phosphine-bound ligands, was later identified and crys-
tallographically characterized. The analogous by-
Following literature precedent [8], to a stirring solu-
tion of n-BuLi in hexanes (2.5 M, 12 ml, 0.03 mol) at
r.t. was added dimethylbenzylamine (3.75 ml, 0.025
mol). The reaction mixture was diluted with freshly
distilled Et2O (38 ml) and allowed to stand for 24 h.
The lithiated benzylamine deposited on the flask as a
mass of yellow crystals leaving an orange solution. The
crystalline mass was re-suspended in solution with vig-
orous stirring. The solution was cooled in an ice water
bath and diphenylchlorophosphine (5.4 ml, 0.03 mol)
was added slowly. The reaction mixture was stirred for

38
L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
product, [Co(CO)3(PPh3)]2, was observed in earlier
work using PPh3 in the preparation of (DABCO)H+
Co(CO)3PPh−3 (DABCO=1,4-diazabicyclooctane) [3e].
associated with the major twin component was ob-
tained. This matrix was used for integration of the full
data set without further refinement of the matrix during
integration. Symmetry-equivalent data showing poor
internal agreement in intensity were examined by hand.
For these groups of reflections, either the outliers or the
entire set of symmetry-equivalent reflections were re-
moved. A total of 371 reflections were thus eliminated.
No absorption correction was applied. A weighting
scheme was chosen to provide an approximately con-
stant analysis of variance across all groups of reflec-
tions and to downweight the contribution of the
weakest reflections, a number of which are somewhat
overestimated in intensity due to the residual effects of
the twinning. The oxygen atom displacement parame-
ters were constrained to be 1.4 times that of the at-
tached carbonyl carbons, and the distances in ring
C(11)ꢀC(16) were restrained to have mutually similar
displacement parameters. Hydrogen atoms were in-
cluded in calculated positions with fixed isotropic dis-
placement parameters, and refined using a riding
model. All other non-hydrogen atoms were refined
freely with anisotropic displacement parameters. Refin-
ement converged to a sensible geometry with physically
reasonable anisotropic displacement parameters. Thus,
despite somewhat low data quality for 2, the use of
appropriate refinement methods has led to structure
determination of adequate accuracy for unambiguous
identification of 2, and discussion of its geometry with
modest precision. The crystal structure was solved by
direct methods and refined to convergence against F2
data [F2] −3|(F2)] using the SHELXTL suite of pro-
grams [9]. Further experimental details are provided in
Table 1 and in CIF format as supplementary data.
2.4. X-ray crystal structure determination of
Co(CO)3{PPh2(o-C6H4CH2NHMe2)} (2)
Crystalline compound 2 decomposes rapidly in air.
Thus, crystals were first coated in hydrocarbon oil in
the drybox, and subsequently mounted on glass fibers
and rapidly transferred to the cold nitrogen stream of
the diffractometer. A number of intensity data sets were
collected on different crystals of 2 using a Bruker
SMART CCD-based area detector diffractometer. The
refinement based upon the best of these data sets is
presented herein. However, even the best data set ob-
tained was from a crystal that exhibited unresolved
twinning. As efforts to obtain an accurate description
of the twinning were unsuccessful, the intensity data
were integrated so as to minimize the effects of twin-
ning. Thus, a single orientation matrix that was a very
good fit to the positions of 500 diffraction maxima
Table 1
Data collection, structure solution, and refinement parameters for 2
and 3 a
2
3
Crystal color
Yellow
Red
Crystal size (mm)
0.33×0.30
×0.03
0.08×0.07×0.03
Crystal system
Space group, Z
Triclinic
Monoclinic
P21/n, 2
10.590(9)
15.513(13)
14.052(11)
90
101.36(2)
90
2263(3)
1.357
(
P1, 2
,
a (A)
7.9437(1)
8.0781(1)
18.1070(1)
98.437(1)
100.579(1)
98.241(1)
1112.36(3)
1.383
,
b (A)
,
c (A)
2.5. X-ray crystal structure determination of
[Co(CO)3{PPh2(o-C6H4CH2NMe2)}]2 (3)
h (°)
i (°)
k (°)
3
,
V (A )
In the solid state this compound decomposes only
over prolonged exposure to air (days). However, as a
precaution, crystals were first coated in hydrocarbon oil
in the drybox, and subsequently mounted on glass
fibers and rapidly transferred to the cold nitrogen
stream of the diffractometer. The intensity data set for
3 was collected using a Bruker SMART CCD-based area
detector diffractometer. An absorption correction was
applied using empirical methods [10] based upon the
method of Blessing [11].
The crystal structure was solved by direct methods
and refined to convergence against F2 data [F2] −
3|(F2)] using the SHELXTL suite of programs [9]. All
non-hydrogen atoms were refined anisotropically. Hy-
drogen atoms were included in calculated positions with
fixed isotropic displacement parameters, and refined
using a riding model. Further experimental details are
Dcalc (g cm−3
)
Temperature (K)
v(Mo–Ka) (mm−1
q range (°)
133(5)
0.868
1.16–25.00
11409
208(5)
0.853
1.98–25.00
29628
)
Reflections collected
Independent reflections, n, 3922 (0.093)
(Rint
3988 (0.149)
)
Reflections used in
refinement
3551
3963
L.S. parameters (p)
Restraints (r)
253
15
0.172
0.441
1.18
271
0
0.083
0.170
1.36
R1(F) b, [I\2.0|(I)]
wR2(F2) b, all data
S(F2) b, all data
a Esds in cell parameters for 2 are unrealistically small due to the
large number of data used in the calculation. These values represent
random and not systematic errors in the parameters.
b R1(F)=S(ꢀFoꢀ−ꢀFcꢀ)/SꢀFoꢀ;
wR2(F2)=[Sw(Fo2−Fc2)2/SwF4o]1/2
;
S(F2)=[Sw(F2o−Fc2)2/(n−p)]1/2
.

L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
39
Co(ꢀI). The previous example, [Ni{NH(CH2CH2-
PPh2)3}CO]BPh4 (4), reported by Midollini and co-
workers [4b], was not explicitly described by the au-
thors as having an NꢀH···Ni hydrogen bond, but
exhibits all the hallmarks of analogous interactions
involving d10 Co centers [3b,c,e,f, 4a], including 2.
Indeed, extended Hu¨ckel calculations reported for 4
suggest an electrostatic component to the NꢀH···Ni
interaction is important (as is typical of a hydrogen
bond). The hydrogen bond [N···Co, 3.31(2); H···Co,
,
2.27 A; NꢀH···Co, 166.1°] [12] in 2 is longer than that
,
in 4 [N···Ni, 3.098(8); NꢀH 1.15(9); H···Ni, 1.95(9) A;
NꢀH···Ni, 171(6)°], consistent with the fact that three
carbon atoms separate the nitrogen and phosphorus in
2, but only two carbon atoms provide the link in 4. The
hydrogen bond in 2 is similar in geometry to that of the
intermolecular hydrogen bonds in (DABCO)H+
Fig. 1. Molecular structure of 2 shown with 30% probability ellip-
soids for non-hydrogen atoms.
Co(CO)3PPh−3 (5) [3.294(6), 2.25 A, 169.6°] [3e, 12],
,
[(DABCO)H+][Co(CO)3PPh2(p-tol)−]·CH3CN 6·CH3-
+
,
CN [3.347(6), 2.30 A, 175.5°] [3f, 12] and [(DABCO)H
][Co(CO)3P(p-tol)3−]·CH3CN 7·CH3CN [3.371(2), 2.32
,
A, 177.4°] [3f, 12] (p-tol=para-tolyl). However, while
the phosphine lies trans to the hydrogen bond in 5–7, a
cis arrangement, of course, facilitates the intramolecu-
lar hydrogen bond in 2. In 5–7, the nitrogen sub-
stituents of the ammonium cation are typically
staggered about the hydrogen bond with respect to the
(equatorial) carbonyl ligands of the anion (torsion an-
gle CꢀN···CoꢀC of ca. 60° [13]), whereas in 2 the
constraint of the intramolecular hydrogen bond results
in a mean CꢀN···CoꢀL torsion angle of only 16°.
Similarly, in 5–7, the (OC)3Co and PR3 moieties are
Fig. 2. Molecular structure of 3 shown with 30% probability ellip-
soids for non-hydrogen atoms.
Table 2
a
,
Selected interatomic distances (A) and angles (°) for 2 and 3
provided in Table 1 and in CIF format as supplemen-
tary data.
2
3
CoꢀCo
CoꢀP
CoꢀC(1)
CoꢀC(2)
CoꢀC(3)
Co···N
2.702(2)
2.218(2)
1.798(8)
1.799(8)
1.766(8)
2.194(5)
1.75(2)
1.76(2)
1.71(2)
3.31(2)
1.13(3)
1.16(2)
1.19(2)
1.84(2)
1.83(2)
1.87(2)
3. Results and discussion
b
3.1. Crystal structures of 2 and 3
C(1)ꢀO(1) b
C(2)ꢀO(2)
C(3)ꢀO(3)
PꢀC(11)
PꢀC(21)
PꢀC(31)
1.153(7)
1.153(7)
1.160(7)
1.827(6)
1.852(6)
1.833(6)
The molecular structures of 2 and 3 are shown in
Figs. 1 and 2, respectively. Crystallographic data are
summarized in Table 1 and selected interatomic dis-
tances and angles are provided in Table 2.
PꢀCoꢀC(1)
PꢀCoꢀC(2)
PꢀCoꢀC(3)
C(1)ꢀCoꢀC(2)
C(1)ꢀCoꢀC(3)
C(2)ꢀCoꢀC(3)
95.7(8)
115.4(6)
112.9(7)
104.7(10)
109.0(12)
116.5(9)
96.1(2)
93.1(2)
95.7(2)
119.5(3)
118.0(3)
120.3(3)
The molecular structure of 2 provides the second
example of an intramolecular NꢀH···M hydrogen bond
involving a tetrahedral d10 metal center, the first for
a For compound 3, label Co corresponds to atoms Co(1) and/or
Co(1a) in Fig. 2 and label P corresponds to atom P(1) Fig. 2.
b Axial CO in 2.

40
L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
staggered about the CoꢀP bond, whereas in 2, the
intramolecular hydrogen bond constrains this torsion
angle to ca. 28° (mean of three CꢀCoꢀPꢀC angles). No
significant differences between 2 and 5–7 in CoꢀC,
CoꢀP or CꢁO distances are observed.
with HCo(CO)4 as per the procedure described for the
preparation of 2. In both cases the product strongly
1
implicated from H-NMR data (doublet at ca. l −10)
was
the
monophosphine-hydride
compound,
HCo(CO)3L (L=8 or 9). No evidence for the forma-
tion of NꢀH···Co hydrogen-bonded products was
found. The likely explanation is either the nitrogen is
not basic enough (8) to be protonated by HCo(CO)3L
or the phosphine is so electron donating that the hy-
dride ligand is less easily deprotonated by the amine
base (9). Of course, these explanations amount to the
same result, but provide a different emphasis.
A common decomposition pathway for HCo(CO)3L
(L=CO, PR3) leads to formation of the symmetric
CoꢀCo bonded dimer Co2(CO)6L2 [14], as we have
observed previously for L=PPh3 [3e]. We were able to
crystallographically characterize the corresponding
dimer, 3, where L=PPh2(o-C6H4CH2NHMe2) (1). The
two phosphine ligands are coordinated in a monoden-
tate manner through phosphorus, and lie trans to the
CoꢀCo bond. This is consistent with all previously
reported structures of Co2(CO)6L2 dimers, where L is a
monodentate ligand [15]. Interestingly the benzyl amine
group in 3, bearing an unprotonatated nitrogen, adopts
a conformation in which the amine nitrogen lies ap-
proximately in the plane of the aryl ring [C(23)ꢀ
C(22)ꢀC(4)ꢀN(1) −23.3 versus C(15)ꢀC(16)ꢀC(17)ꢀN
100.2° in 2]. This conformation facilitates an in-
tramolecular CꢀH···N hydrogen bond (Fig. 2), which
satisfies the hydrogen bond acceptor capability of the
3.3. Pathway for protonation of the anion Co(CO)3L−
(L=CO, PR3)
An instructive way to think of a hydrogen bond is as
an incipient proton transfer. This provides a link be-
tween structure and reaction, protonation in this case.
Taken collectively, the crystal structures of 2, 5–7 and
other R3NH+Co(CO)−4 structures provide a series of
snapshots of the protonation of the Co(CO)3L− anion
by an ammonium cation, R3NH+. Arranging these
molecular structures sequentially one can use the Struc-
ture Correlation Principle [16] to map out the reaction
pathway in terms of the changes in geometry of the
R3NH+Co(CO)−4 unit. In this case it is most instructive
to follow the change in geometry around the metal
center as a function of H···Co separation. The former is
most readily quantified in terms of the mean
LeqꢀCoꢀLeq or LaxꢀCoꢀLeq angles, where Leq and Lax
are the equatorial and axial ligands most readily
defined in the pseudo-trigonal bipyramidal geometry of
the metal hydride product of protonation, HCo(CO)3L.
In addition to the NꢀH···Co hydrogen-bonded struc-
tures, the initial point of the reaction pathway is repre-
sented by ammonium and Co(CO)3L− ions situated at
,
amine nitrogen [H···N 2.306 A and CꢀH···N 101.8°,
,
using CꢀH 1.083 A]. Metalꢀmetal and metalꢀligand
bonds in 3 are similar to, but slightly longer than those
of (six) related structures [15], viz. CoꢀCo 2.645–2.671,
cf. 2.702(2) A in 3, CoꢀP 2.135–2.178, cf. 2.218(2) A in
3, and CoꢀC 1.754–1.774, cf. 1.788 A (average) in 3.
,
,
,
3.2. Spectroscopic characterization of 2
As noted in the experimental section, spectroscopic
characterization of 2 proved elusive in contrast to the
intermolecularly hydrogen-bonded analogs 5–7, despite
numerous attempts to obtain NMR or IR data on the
pure material. Typical 1H-NMR data (in toluene-d8) for
the crude product showed a number of resonances in
the region l 10–14, consistent with NꢀH groups, but
indicative of more than one product, and inconclusive
in identifying 2. NMR spectra obtained from dissolu-
tion of the crystals of 2 (in toluene-d8) were consistent
only with the presence of the ligand, 1. IR data, for
which the w(CO) stretches have been valuable in char-
acterizing previous intermolecularly hydrogen bonded
systems 5–7, did not exhibit a pattern of carbonyl
bands consistent with either 2 or likely by-products, but
suggested a mixture of carbonyl containing products.
These observations suggest considerable instability of
,
approximately van der Waals separation [H···Co 3.0 A].
The metal hydride endpoint of the protonation reaction
is represented by the structures of HCo(CO)4 [17],
HCo(CO)3L (known only for L=PCy3 [18]) and the
isoelectronic d10 anion present in PPN+HFe(CO)−4 [19],
the latter being isoelectronic and structurally similar to
HCo(CO)4. Data for the more constrained, and in-
tramolecularly NꢀH···Ni hydrogen-bonded, system, 4,
again involving a d10 metal center, are also included in
the analysis. Scheme 1 depicts all structures, shown
sequentially, together with pertinent geometric data.
Fig. 3 shows the Structure Correlation analysis of the
protonation pathway [20].
2
relative to 5–7, in particular indicating that
aminophosphine loss from 2 is quite facile in solution.
In this light the definitive crystallographic characteriza-
tion of 2 is all the more remarkable. In an effort to
obtain species similar to 2, but with greater stability,
two other aminophosphines, Ph2Ppy (py=2-pyridyl)
(8), Ph2PCH2CH2NEt2 (9), were prepared and reacted
A number of important points arise from examina-
tion of the Structure Correlation plot (Fig. 3). The
geometric changes that occur over the course of the
protonation reaction can be followed along either of
the solid lines, which although fitted mathematically to

L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
41
Scheme 1. Structures whose geometries were used in the Structure Correlation study to determine the geometric changes that define the pathway
for protonation of Co(CO)3L–. Average values Žr are listed for the angles with esds (|) representing the scatter in these angles (i.e.
|=[Si(Žr−ri)2/(n−1)]1/2, where ri=ith observation, n=number of observations). These e.s.d.s do not necessarily correlate with those of
,
individual angles. HꢀM distances for 12–15, 7, 6, 2, and 5 are calculated after normalizing the X-ray determined NꢀH distance to 1.054 A, as
determined by neutron diffraction in 11. Original references for the X-ray structures are 10 [model geometry, this work], 11 ([21], neutron
diffraction), 12 [3c], 13 [3e], 14 [3f], 15 [4a], 7 [3f], 6 [3f], 2 (this work), 5 [3e], 4 [4b], 16 [19], 17 ([17], electron diffraction), 18 [18]. See also Ref.
[22] for reports of the synthesis and IR characterization, but not crystal structures, of HCo(CO)3PPh3, HCo(CO)3P(OPh)3 and Et3NH+
Co(CO)3PPh−3
.
Thus, at this stage of the reaction one could argue that
the metallate anion has adopted a geometry consistent
with receiving the proton in the final stage of the
reaction in which it is transferred to the metal, i.e. it is
now ‘prepared’ for proton transfer.
the data points should be used primarily to guide the
eye in interpreting the plot. Progression along the reac-
tion pathway (right to left) may be described as follows.
Approach of the NꢀH group of the cation to one face
of the tetrahedral anion and formation of the NꢀH···Co
hydrogen bond results initially in little perturbation of
the metal coordination geometry. Thus, by the time
that the H···Co separation has decreased to ca. 2.4,
It is noticeable that, with the exception of 4, to which
we will return shortly, no structures are represented in
this last stage of the reaction for in which MꢀH separa-
,
,
tions lie in the range 2.2–1.6 A. This observation is
from the approximately 3.0 A at van der Waals separa-
consistent with the report by Norton, Sweany and
co-workers that hydrogen-bonded structures of the type
(CO)4CoꢀH···NR3 are not formed [6]. Indeed our own
efforts to prepare systems with shorter Co···H separa-
tions than in 5 by using trialkylphosphines have re-
sulted in the formation of cobalt hydride species [24]. In
this light, Midollini’s NꢀH···Ni hydrogen-bonded struc-
ture, 4, warrants further examination. Firstly the hy-
drogen atom appears to be primarily located on the
nitrogen, based upon the X-ray crystal structure,
tion [23], the interligand angles have changed relatively
little from 109.5 to LeqꢀCoꢀLeq 112 and LaxꢀCoꢀLeq
107°, typified by the structure of 14. This corresponds
to a slight opening up of the anion geometry in an
‘umbrella-like’ fashion. From this point on the reaction
coordinate a much more substantial change in metal
coordination geometry occurs as the H···Co separation
,
decreases from ca. 2.4 to 2.2 A resulting in a geometry
best represented by the structure of 5. At this point the
metal coordination geometry (at least in terms of the
carbonyl and phosphine ligands) already closely resem-
bles that of the final product, namely the metal hydride
(viz. LeqꢀCoꢀLeq 115.6 and LaxꢀCoꢀLeq 102.3 in 5 ver-
sus LeqꢀCoꢀLeq 117.3 and LaxꢀCoꢀLeq 99.7° in 17).
,
though a refined NꢀH distance of 1.15(9) A suggests a
more weakened NꢀH bond relative to the preceding
NꢀH···Co cases, and thus a stronger H···M interaction.
The cage-like geometry of the (Ph2PCH2CH2)3NH+

42
L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
tripod ligand (Hnp+3 ) restricts the metal coordination
geometry leading to the unusual situation in which the
H···Ni separation is extremely short, but the metal
geometry is still tetrahedral (both LeqꢀCoꢀLeq and
LaxꢀCoꢀLeq have average values of 109.5°). The data
point for 4 clearly lies well off the most probable
reaction pathway, but illustrates that in such con-
strained systems very short NꢀH···M hydrogen bonds
can be achieved even when the d10 metal center is not
suitably ‘prepared’ for proton transfer [25]. The Struc-
ture Correlation study clearly demonstrates that 4 is a
remarkable structure.
binary system linked through an NꢀH···Co hydrogen
bond. Rather is a five component hydrogen-bonded
chain consisting of two outer Co(CO)−4 anions, each of
which is linked to
a central N-methylpiperazine
molecule via an NꢀH···Co hydrogen bond and an
NꢀH···N hydrogen bond, each originating from the
secondary ammonium center of an N-methylpiper-
azinium cation.
4. Summary
The structures of 11 and 12 also represent geometries
that lie slightly off the main reaction pathway. Specifi-
cally they can be described as having a larger distortion
away from tetrahedral geometry at the metal than
anticipated given their Co···H separations. In the case
of 11, this makes sense in view of the much greater
steric bulk of the cation relative to the ammonium
groups used (e.g. (DABCO)H+) in the other examples.
For 12, such an argument is not as strong. However, it
should be noted that this structure is not just a simple
The low temperature X-ray crystal structure of an
elusive organometallic zwitterion containing an in-
tramolecular NꢀH···Co (d10) hydrogen bond has been
reported together with one of the common byproducts
of its synthesis. A Structure Correlation study has been
undertaken to derive a geometric model for the reaction
pathway for protonation of the Co(CO)3L− anion to
yield the hydride complex HCo(CO)3L anion (L=CO,
PR3).
5. Supplementary material
Crystallographic data for the structural analyses have
been deposited with the Cambridge Structural Data
Centre, CCDC nos. 141607 and 141608 for compounds
2 and 3, respectively. Copies of the information may be
obtained free of charge from The Director, CCDC, 12
Union Road, Cambridge CB2 1EZ, UK (fax: +44-
1223-336033; e-mail: deposit@ccdc.cam.ac.uk or www:
http://www.ccdc.ca.ac.uk).
Acknowledgements
Financial support from the Donors of the Petroleum
Research Fund, administered by the American Chemi-
cal Society, the Missouri Research Board, and an
UMSL Research Award, is acknowledged. J.C.M.R.
acknowledges further support in the form of fellow-
ships from the UM-St. Louis Chemistry Alumni, the
UM-St. Louis Graduate School and Mallinkrodt
Chemical Co. We thank Mr Michael Groaning for the
synthesis of the Ph2PCH2CH2NEt2 ligand. Purchase of
the Bruker (Siemens) SMART diffractometer was funded
in part by NSF grant no. CHE-9309690.
References
Fig. 3. Structure Correlation study, modelling the pathway for proto-
nation of Co(CO)3L− anion at the metal center. Error bars on
LꢀMꢀL angles at 1| level (see Scheme 1 for calculation of |).
[1] G.R. Desiraju, T. Steiner, The weak hydrogen bond, in: IUCr
Monographs on Crystallography, No. 9, Oxford University
Press, Oxford, 1999.

L. Brammer et al. / Journal of Organometallic Chemistry 609 (2000) 36–43
43
[2] (a) T. Steiner, Crystallogr. Rev. 6 (1996) 1. (b) G.R. Desiraju,
Acc. Chem. Res. 24 (1991) 270. (c) G.R. Desiraju, Acc. Chem.
Res. 29 (1996) 441. (d) R. Taylor, O. Kennard, J. Am. Chem.
Soc. 104 (1982) 5063.
ton, A. Moroz, R.L. Sweany, S.L. Whittenburg, Organometallics
10 (1991) 2357.
[7] P. Sherwood, L. Brammer, unpublished results.
[8] (a) K.P. Klein, C.R. Hauser, J. Org. Chem. 32 (1967) 1479. (b)
T.B. Rauchfuss, F.T. Patino, D.M. Roundhill, Inorg, Chem. 14
(1975) 652.
[3] (a) L. Brammer, J.M. Charnock, P.L. Goggin, R.J. Goodfellow,
A.G. Orpen, T.F. Koetzle, J. Chem. Soc. Dalton Trans. (1991)
1789. (b) L. Brammer, M.C. McCann, R.M. Bullock, R.K.
McMullan, P. Sherwood, Organometallics 11 (1992) 2339. (c) L.
Brammer, D. Zhao, Organometallics 13 (1994) 1545. (d) L.
Brammer, D. Zhao, F.T. Ladipo, J. Braddock-Wilking, Acta
Crystallogr. Sect. B 51 (1995) 632. (e) D. Zhao, F.T. Ladipo, J.
Braddock-Wilking, L. Brammer, P. Sherwood, Organometallics
15 (1996) 1441. (f) L. Brammer, J.C. Mareque Rivas, D. Zhao,
Inorg. Chem. 37 (1998) 5512. (g) J.C. Mareque Rivas, L. Bram-
mer, Coord. Chem. Rev. 183 (1999) 43. (h) L. Brammer, in:
J.A.K. Howard, F.H. Allen, G.P. Shields (Eds.), Implications of
Molecular and Materials Structure for New Technologies,
Kluwer, Dordecht, The Netherlands, 1999, pp. 197–210.
[4] (a) F. Calderazzo, G. Fachinetti, F. Marchetti, P.F. Zanazzi, J.
Chem. Soc. Chem. Commun. (1981) 181. (b) F. Cecconi, C.A.
Ghilardi, P. Innocenti, C. Mealli, S. Midollini, A. Orlandini,
Inorg. Chem. 23 (1984) 922. (c) I.C.M. Wehman-Ooyevaar,
D.M. Grove, H. Kooijman, P. van der Sluis, A.L. Spek, G. van
Koten, J. Am. Chem. Soc. 114 (1992) 9916. (d) I.C.M. Wehman-
Ooyevaar, D.M. Grove, P. de Vaal, A. Dedieu, G. van Koten,
Inorg. Chem. 31 (1992) 5484. (e) S.G. Kazarian, P.A. Hamley,
M. Poliakoff, J. Am. Chem. Soc. 115 (1993) 9069. (f) A. Albi-
nati, F. Lianza, B. Mu¨ller, P.S. Pregosin, Inorg. Chim. Acta 208
(1993) 119. (g) A. Albinati, F. Lianza, P.S. Pregosin, B. Mu¨ller,
Inorg. Chem. 33 (1994) 2522. (h) Y.S. Shubina, L.M. Epstein, J.
Mol. Struct. 265 (1992) 367. (i) E.S. Shubina, A.N. Krylov, A.Z.
Kreindlin, M.I. Rybinskaya, L.M. Epstein, J. Mol. Struct. 301
(1993) 1. (j) E.S. Shubina, A.N. Krylov, A.Z. Kreindlin, M.I.
Rybinskaya, L.M. Epstein, J. Mol. Struct. 465 (1994) 259. (k) D.
Braga, F. Grepioni, E. Tedesco, K. Biradha, G.R. Desiraju,
Organometallics 16 (1997) 1846. (l) Y. Gao, O. Eisenstein, R.H.
Crabtree, Inorg. Chim. Acta 254 (1997) 105. (m) G. Orlova, S.
Scheiner, Organometallics 17 (1998) 4362. (n) J. Kozelka, J.
Berge`s, R. Attias, J. Fraitag, Angew. Chem. Int. Ed. 39 (2000)
198.
[9] SHELXTL-5.0, Bruker Analytical X-ray, Inc., Madison, WI, USA
1995.
[10] G. Sheldrick, SABABS, version 2, University of Go¨ttingen, 1995.
[11] R.H. Blessing, Acta Crystallogr. Sect. A 51 (1995) 33.
[12] The H···Co distance and NꢀH···Co angle are calculated by
extending the NꢀH distance along the determined NꢀH vector to
,
a value of 1.054 A, as determined by neutron diffraction for
Et3NH+Co(CO)−4 [3b]. For 2, as the ammonium hydrogen could
not be located in the difference map, the NꢀH vector was also
calculated, assuming sp3 hybridization at nitrogen.
[13] In 5 and 6, the (DABCO)H+ cation is disordered over two
orientations, the minor orientation being eclipsed (CꢀN···CoꢀC
ca. 0°) rather than staggered with the anion. Furthermore the
structure of (Quinuclidine)H+Co(CO)4− exhibits an ordered
eclipsed geometry about the hydrogen bond [3f].
[14] (a) This is particularly well documented for L=CO, where
decomposition has been shown to be autocatalytic resulting from
’
the presence of Co(CO)4 radicals [14b]. (b) R.W. Wegman, T.L.
Brown, J. Am. Chem. Soc. 102 (1982) 2494.
[15] (a) F.W.B. Einstein, R. Kirkland, Acta Crystallogr. Sect. B 34
(1978) 1690. (b) J.A. Ibers, J. Organomet. Chem. 14 (1968) 423.
(c) R.A. Jones, M.H. Seeberger, A.L. Stuart, B.R. Whittlesey,
T.C. Wright, Acta Crystallogr. Sect. C 42, (1986) 399. (d) T.
Bartik, B. Bartik, B.E. Hanson, K.H. Whitmire, I. Guo, Inorg.
Chem. 32 (1993) 5833. (e) P. Braunstein, D.G. Kelly, Y.
Dusausoy, D. Bayeul, M. Lanfranchi, A. Tiripicchio, Inorg.
Chem. 33 (1994) 233. (f) D.H. Farrar, A.J. Lough, A.J. Poe,
T.A. Stromnova, Acta Crystallogr. Sect. C 51 (1995) 2008.
[16] (a) H.-B. Bu¨rgi, J.D. Dunitz, Acc. Chem. Res. 1983, 16, 153. (b)
H.-B. Bu¨rgi, J.D. Dunitz (Eds.), Structure Correlation, VCH,
Weinheim, Germany, 1993.
[17] E.A. MacNeill, F.R. Scholer, J. Am. Chem. Soc. 99 (1977) 6243.
[18] L.S. Leigh, K.S. Whitmire, Acta Crystallogr. Sect. C 45 (1989)
210.
[5] (a) This is not to say that there are no examples of bona fide
MꢀH···A hydrogen bonds (for examples, see Refs. [5b–f]). Such
cases have been characterized, but these typically involve
cationic metal hydrides, in which the positive charge associated
with the hydrogen bond donor may compensate for the fact that
the hydrogenꢀmetal bond is polarized Ml+ꢀHd−. An unusual
case is the neutral intramolecular hydrogen bond reported by
Pickett and co-workers [5d]. (b) M.A. Adams, K. Folting, J.C.
Huffman, K.G. Caulton, Inorg. Chem. 18 (1979) 3020. (c) L.M.
Epstein, E.S. Shubina, A.N. Krylov, A.Z. Kreindlin, M.I. Rib-
inskaya, J. Organomet. Chem. 447 (1993) 227. (d) S.A.
Fairhurst, R.A. Henderson, D.L. Hughes, S.K. Ibrahim, C.J.
Pickett, J. Chem. Soc. Chem. Commun. (1995) 1569. (e) E. Peris,
R.H. Crabtree, J. Chem. Soc. Chem. Commun. (1995) 2179. (f)
D. Braga, F. Grepioni, E. Tedesco, K. Biradha, G.R. Desiraju,
Organometallics 15 (1996) 2692.
[19] M.B. Smith, R. Bau, J. Am. Chem. Soc. 95 (1973) 2388.
[20] A preliminary version of Figure 3 has appeared in slightly
different form in Ref. [3h].
[21] L. Brammer, M.C. McCann, R.M. Bullock and R.K. McMullan,
unpublished results for neutron structure at 123 K. See also Ref.
[3b] for report of neutron structure at 15 K.
[22] E.J. Moore, J.M. Sullivan, J.R. Norton, J. Am. Chem. Soc. 108
(1986) 2257.
[23] A.J. Bondi, J. Chem. Phys. 68 (1964) 441.
[24] D. Zhao, L. Brammer, unpublished results.
[25] (a) Some may have cause to question the assignment of an
NꢀH···Ni hydrogen bond in 4. However, spectroscopic data
(1H-NMR l 14.5 NꢀH and IR w(NꢀH) 2130 cm−1 br) and
computational results (EHMO) support this assignment. Perhaps
,
most convincing is the N···Ni separation in 4 at 3.092(8) A is
substantially shorter than in the deprotonated analog
,
[6] Either no hydrogen bond is formed or proton transfer occurs to
(np3)Ni(CO) at 3.25(1) A [24b]. (b) C.A. Ghilardi, A. Sabatini,
L. Sacconi, Inorg. Chem. 15 (1976) 2763.
give AꢀH+ and Co(CO)−4 . See, S.S. Kristja´nsdo´ttir, J.R. Nor-
.