Gas-phase synthesis of charged copper and silver Fischer carbenes from diazomalonates: Mechanistic and conformational considerations in metal-mediated Wolff rearrangements

Article abstract of DOI:10.1021/ja028337j

Copper(I) and silver(I) Fischer carbenes are synthesized in the gas phase. Various diazomalonate-based compounds with an attached metal ion are introduced into the gas phase by electrospray ionization and subjected to collisional activation. Loss of N₂ generates a metastable Fischer carbene, which subsequently undergoes Wolff rearrangement and loss of CO. Further excitation leads to the loss of another CO molecule and the generation of a stable Fischer carbene. Isotopically labeled compounds are utilized to confirm the assignment of the products resulting from this process. DFT calculations are used to evaluate various mechanistic possibilities and to quantitatively assess the energetics of reactants and products. Silver(I) is shown to be more effective in facilitating Wolff rearrangement than copper(I), although both are more effective when compared to spectator charges such as sodium or a fixed quaternary nitrogen. Carbenes are not produced when copper(II), nickel(II), or a proton is used to form a quasi-molecular ion from the diazomalonate carbene precursor. Finally, trapping of the Fischer carbene by various functional groups attached through the open coordination site of the metal is investigated.

Full text of DOI:10.1021/ja028337j

Published on Web 03/19/2003
Gas-Phase Synthesis of Charged Copper and Silver Fischer
Carbenes from Diazomalonates: Mechanistic and
Conformational Considerations in Metal-Mediated Wolff
Rearrangements
Ryan R. Julian,† Jeremy A. May,‡ Brian M. Stoltz,*,‡ and J. L. Beauchamp*,†
Contribution from the Beckman Institute and The Arnold and Mabel Beckman Laboratories of
Chemical Synthesis, DiVision of Chemistry and Chemical Engineering, California Institute of
Technology, Pasadena, California 91125
Received August 29, 2002; E-mail: jlbchamp@its.caltech.edu, stoltz@caltech.edu
Abstract: Copper(I) and silver(I) Fischer carbenes are synthesized in the gas phase. Various diazomalonate-
based compounds with an attached metal ion are introduced into the gas phase by electrospray ionization
and subjected to collisional activation. Loss of N2 generates a metastable Fischer carbene, which
subsequently undergoes Wolff rearrangement and loss of CO. Further excitation leads to the loss of another
CO molecule and the generation of a stable Fischer carbene. Isotopically labeled compounds are utilized
to confirm the assignment of the products resulting from this process. DFT calculations are used to evaluate
various mechanistic possibilities and to quantitatively assess the energetics of reactants and products.
Silver(I) is shown to be more effective in facilitating Wolff rearrangement than copper(I), although both are
more effective when compared to spectator charges such as sodium or a fixed quaternary nitrogen. Carbenes
are not produced when copper(II), nickel(II), or a proton is used to form a quasi-molecular ion from the
diazomalonate carbene precursor. Finally, trapping of the Fischer carbene by various functional groups
attached through the open coordination site of the metal is investigated.
excellent carbene precursors for such studies.6 In these experi-
ments, a carbene is produced through the low-energy collision-
Introduction
The gas-phase synthesis of metallocarbenoid compounds has
a surprisingly long history.1 In early studies, the metallocar-
benoids were created by high-energy methods utilizing electron
impact ionization. The resulting ions were then used in
bimolecular reactions with various neutral molecules to study
the chemistry of and determine metal-carbon bond strengths
for these charged Fischer carbenes. The importance of studying
gas-phase species in determining mechanistic pathways for
olefin metathesis was also realized early on,1b and more recent
studies have taken advantage of electrospray ionization (ESI)
to continue this work.2 Gas-phase experiments have also enabled
mechanistic studies with other highly reactive metallocarbenoid
species.3
activated dissociation (CAD)7 of a diazo compound. The
resulting loss of N2 and generation of the carbene is achieved
under conditions sufficiently mild that noncovalently bound
complexes are not dissociated in the process. The highly reactive
carbene then inserts into the guest and converts the noncovalent
complex into a covalently bound molecule. All of the molecules
in our previous study were based on the diazomalonate core
with at least one 18-crown-6 crown ether attached. These
reagents have been appropriately named “molecular mouse-
traps.”6
Diazomalonates8 (such as 1 and 2) and related diazo ketones9
are known to undergo Wolff rearrangement in the gas phase as
The utility of diazo compounds as carbene and metallocar-
benoid precursors has been heavily exploited in solution-phase
chemistry.4 The use of diazo compounds as precursors in gas-
phase studies involving mass spectrometry (MS) remains less
explored.5 We have recently reported that diazo compounds are
shown in Scheme 1. Since its discovery in 1902, the Wolff
† Beckman Institute.
(4) Doyle, M. P.; McKervey, M. A.; Ye, T. Modern Catalytic Methods for
Organic Synthesis with Diazo Compounds; Wiley-Interscience: New York,
1998. (b) Moody, C. J.; Whitham, G. H. ReactiVe Intermediates; Oxford
University Press: New York, 1992; pp 26-50. (c) Straub, B. F.; Hofmann,
P. Angew. Chem., Int. Ed. 2001, 40, 1288-1290.
‡ Division of Chemistry and Chemical Engineering.
(1) Stevens, A. E.; Beauchamp, J. L. J. Am. Chem. Soc. 1978, 100, 2584-
2585. (b) Stevens, A. E.; Beauchamp, J. L. J. Am. Chem. Soc. 1979, 101,
6449-6450. (c) Halle, L. F.; Armentrout, P. B.; Beauchamp, J. L. J. Am.
Chem. Soc. 1981, 103, 962-963.
(5) Lebedev, A. T.; Bakulev, V. A.; Hayes, R. N.; Bowie, J. H. Rapid Commun.
Mass. Spectrom. 1991, 5, 234-237. (b) Lebedev, A. T. ReV. Mass Spectrom.
1991, 10, 91-132. (c) Armentrout, P. B.; Sunderlin, L. S.; Fisher, E. R.
Inorg. Chem. 1989, 28, 4436-4437.
(2) Adlhart, C.; Hinderling, C.; Baumann, H.; Chen, P. J. Am. Chem. Soc.
2000, 122, 8204-8214.
(3) Bertani, R.; Michelin, R. A.; Mozzon, M.; Traldi, P.; Seraglia, R.; Busetto,
L.; Cassani, M. C.; Tagliatesta, P.; D’Arcangelo, G. Organometallics 1997,
16, 3229-3233.
(6) Julian, R. R.; May, J. A.; Stoltz, B. M.; Beauchamp, J. L. Angew. Chem.,
Int. Ed. 2003, 42(9), 1012-1015.
9
4478
J. AM. CHEM. SOC. 2003, 125, 4478-4486
10.1021/ja028337j CCC: $25.00 © 2003 American Chemical Society

Synthesis of Charged Cu and Ag Fischer Carbenes
A R T I C L E S
Scheme 1
Professional Suite (Hypercube, Inc., Gainesville, FL). Molecular
dynamics simulations were carried out using CACHe Worksystem Pro
5.04 (Fujitsu, Inc., Beaverton, OR). Structures were heated to 300 K
for 10 ps with a 0.001-ps time interval utilizing the augmented MM3
parameters.
Synthesis. Reactions were performed in flame-dried glassware under
a nitrogen atmosphere. Solvents were dried and purified using activated
alumina columns. All other reagents were used as received from
commercial sources. Reaction temperatures were controlled by an
IKAmag temperature modulator. Thin-layer chromatography (TLC) was
performed using E. Merck silica gel 60 F254 precoated plates (0.25
mm) and visualized by UV and p-anisaldehyde staining. ICN silica
gel (particle size 0.032-0.063 mm) was used for flash chromatography.
1H NMR spectra were recorded on a Varian Mercury 300 spectrometer
(at 300 MHz) in CDCl3 and are internally referenced to the residual
chloroform peak (7.27 ppm) relative to Me4Si. Data for 1H NMR spectra
are reported as follows: chemical shift (δ ppm), multiplicity, coupling
constant (Hz), and integration. IR spectra were recorded on a Perkin-
Elmer Paragon 1000 spectrometer and are reported in frequency of
absorption (cm-1). Preparatory reversed-phase HPLC was performed
on a Beckman HPLC with a Waters DeltaPak 25 × 100 mm, 100-µm
C18 column equipped with a guard.
rearrangement10 has been the subject of numerous studies.11 In
the present work, we use the solvent-free environment of gas-
phase experiments to study the mechanism of multiple, consecu-
tive Wolff rearrangements observed in diazomalonates.8 The
effects that various coordinated metal ions and other charged
groups have on Wolff rearrangements are discerned from gas-
phase MS experiments. Theory is used to quantitatively assess
each intermediate for the proposed mechanism. Although the
solution-phase synthesis of stable copper(I) and silver(I) Fischer
carbenes has been known for some time,12 here we report the
first gas-phase synthesis of copper(I) and silver(I) Fischer
carbenes. Results for several intermolecular reactions of these
carbenes with molecules coordinated to the metal ion are
presented.
2-Diazodimethyl Malonate (1). 1 was prepared according to
previously established methods.14 The product was isolated as a yellow
oil (2.58 g, 16.29 mmol, 93% yield) with the same physical properties
as previously reported.
Experimental Section
2-Diazodibenzyl Malonate (2). A round-bottomed flask (10 mL)
was charged with dibenzylmalonate (77 µL, 0.308 mmol), MeCN
(3 mL), and p-acetamidobenzylsulfonyl azide (116 mg, 0.485 mmol).
Et3N (150 µL, 1.08 mmol) was then added, and the reaction was stirred
at room temperature for 10 h. TLC analysis (3:1 hexanes/EtOAc eluent,
Rf ) 0.46) showed the reaction to be complete. The solvent was
removed by evaporation under reduced pressure, and the crude mixture
was subjected to flash chromatographic purification (5:1 hexanes/EtOAc
eluent) to afford 2 as a yellow oil (75 mg, 0.242 mmol, 78% yield)
with the same physical properties as previously reported.15
Diazomalonate 25 (Chart 1). A round-bottomed flask (5 mL) was
charged with CH2Cl2 (1 mL), CD3OD (500 µL, 11.3 mmol), and Et3N
(100 µL, 0.717 mmol). This mixture was stirred rapidly while malonyl
dichloride (20 µL, 0.206 mmol) was added dropwise at room temper-
ature. TLC analysis (3:1 hexanes/EtOAc eluent, Rf ) 0.45) showed
complete conversion to the malonate ester within 2 h. The solvent and
excess reagents were removed by evaporation under reduced pressure,
and then MeCN (2 mL) and p-acetamidobenzylsulfonyl azide (103.5
mg, 0.431 mmol) were added to the flask. Et3N (100 µL, 0.717 mmol)
was added, and the solution was stirred for 24 h. The solvent was then
removed by evaporation under reduced pressure. The product was
purified by dissolving the residue in a minimal amount of CH2Cl2 (500
µL) and then precipitating the salts by addition of Et2O (5 mL).
Filtration through Celite removed the salts, and the solvent was removed
by evaporation under reduced pressure to afford 25 as a yellow oil
(11.9 mg, 0.073 mmol, 35% yield).
All mass spectra were acquired on a Finnigan LCQ Classic
quadrupole ion trap instrument utilizing a standard electrospray source.
Solutions of the reagents in the ∼30-80 µM range were electrosprayed
from a ∼80/20 (v/v) solution of methanol/water with a minimum of
0.1% MeCN added. Soft ionization settings that minimize energetic
collisions during sample collection were used to maximize the intensity
of noncovalently bound complexes.13 Ions of interest were isolated and
subjected to collisional activation until product peaks were observed.
Helium was used as the collision gas for all experiments. For each
MSn step, the peak of interest was reisolated prior to further dissociation.
All chemicals were purchased from Sigma-Aldrich and used without
further purification unless otherwise noted. Metal ion complexes were
formed by adding an appropriate salt to the solution. No counterion
effects were noted. For studies of intermolecular reactions, the desired
ligand (such as 5-hexynenitrile) was added directly to the solution in
severalfold excess.
Calculations. Candidate structures were evaluated initially at the
PM3 semiempirical level. Following minimization at the lower level
of theory, structures were optimized using density functional theory
(DFT). The DFT calculations were carried out using Jaguar 4.1
(Schro¨dinger, Inc., Portland, OR). Full geometry optimization was
performed at the B3LYP/LACVP** level of theory. Semiempirical PM3
MNDO-type calculations were carried out using the HyperChem 5.1
(7) Marzluff, E. M.; Beauchamp, J. L. In Large Ions: Their Vaporization,
Detection, and Structural Analysis; Baer, T., Ng, C. Y., Powis, I.; Eds.;
John Wiley & Sons Ltd.: New York, 1996; pp 115-143. (b) McLuckey,
S. A. J. Am. Soc. Mass Spectrom. 1992, 3, 599-614. (c) Hayes, R. N.;
Gross, M. L. Methods Enzymol. 1990, 193, 237-263.
(8) Richardson, D. C.; Hendrick, M. E.; Jones, M. J. Am. Chem. Soc. 1971,
93, 3790-3791.
(9) Marfisi, C.; Verlaque, P.; Davidovics, G.; Pourcin, J.; Pizzala, L.; Aycard,
J.-P.; Bodot, H. J. Org Chem. 1983, 48, 533-537.
Diazomalonate 26. To a stirred solution of H2C(13CO2H)2 (19.7 mg,
0.179 mmol) in Et2O (2 mL) in a scratch-free flask (25 mL) was added
ethereal diazomethane solution (0.2 M, 4.0 mL, 0.800 mmol). TLC
analysis (3:1 hexanes/EtOAc eluent, Rf ) 0.45) showed the reaction to
be complete. The solvent and excess reagents were removed by
evaporation under reduced pressure, and then MeCN (1 mL) and
p-acetamidobenzylsulfonyl azide (25.9 mg, 0.108 mmol) were added
to the flask. Et3N (41 µL, 0.294 mmol) was added, and the solution
was stirred for 24 h. The solvent was then removed by evaporation
under reduced pressure. The product was purified by dissolving the
(10) Wolff, L. Justus Liebigs Ann. Chem. 1902, 325, 129.
(11) Sudrik, S. G.; Chavan, S. P.; Chandrakumar, K. R. S.; Pal, S.; Date, S. K.;
Chavan, S. P.; Sonawane, H. R. J. Org. Chem. 2002, 67, 1574-1579 and
references therein. (b) McMahon, R. J.; Chapman, O. L.; Hayes, R. A.;
Hess, T. C.; Krimmer, H. P. J. Am. Chem. Soc. 1985, 107, 7597-7606.
(c) Fenwick, J.; Frater, G.; Ogi, K.; Strausz, O. P. J. Am. Chem. Soc. 1973,
95, 124-132. (d) Pomerantz, M.; Levanon, M. Tetrahedron Lett. 1991,
32, 995-998. Also see ref 4a Chapter 9.
(12) Arduengo, A. J.; Dias, H. V. R.; Calabrese, J. C.; Davidson, F. Organo-
metallics 1993, 12, 3405.
(14) Falorni, M.; Dettori, G.; Giacomelli, G. Tetrahedron: Asymmetry 1998, 9,
1419
(13) Julian, R. R.; Beauchamp, J. L. Int. J. Mass. Spectrom. 2001, 210, 613-
(15) Kametani, T.; Yukawa, H.; Honda, T. J. Chem Soc., Perkin Trans. 1 1990,
3, 571-577.
623.
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4479

A R T I C L E S
Julian et al.
Chart 1
residue in a minimal amount of CH2Cl2 (500 µL) and then precipitating
the salts by addition of Et2O (5 mL). Filtration through Celite removed
the salts, and the solvent was removed by evaporation under reduced
pressure to afford 26 as a yellow oil (17.4 mg, 0.108 mmol, 60% yield).
Diazomalonate 27. A round-bottomed flask (10 mL) was charged
with CH2Cl2 (2 mL), EtOH (25 µL, 0.687 mmol), and 2-bromoethanol
(30 µL, 0.421 mmol). This mixture was stirred rapidly while malonyl
dichloride (20 µL, 0.206 mmol) was added dropwise at room temper-
ature. The reaction yielded a statistical mixture of esters within 2 h as
determined by ESI-MS. The primary product was 2-bromoethyl
ethylmalonate. The solvent and excess reagents were removed by
evaporation under reduced pressure, and then MeCN (2 mL) and
pyridine (50 µL) were added to the flask. After 24 h, p-acetamido-
benzylsulfonyl azide (94.1 mg, 0.392 mmol) and then Et3N (150 µL,
1.076 mmol) were added. The solution was stirred for another 12 h.
The solvent was removed by evaporation under reduced pressure to
afford 27; the yield was qualitatively estimated to be ∼50% by MS
(m/z 264.3).
Diazomalonate 28. To a stirred, dry solution of penta(ethylene
glycol) (44.4 µL, 0.210 mmol), CH2Cl2 (20 mL), and Et3N (150 µL,
1.076 mmol) was added malonyl dichloride (24 µL, 0.247 mmol). The
mixture was heated to reflux for 5.5 h, cooled, and then the solvent
was removed by evaporation under reduced pressure. The product was
recovered by extraction of the crude residue with refluxing hexanes
(20 mL). The solvent was then removed by evaporation under reduced
pressure to leave the malonate crown ether. This malonate ester16 was
then dissolved in MeCN (3 mL). Stepwise addition of p-acetamido-
benzenesulfonyl azide (151.0 mg, 0.629 mmol) and Et3N (200 µL, 1.435
mmol) provided 28 after stirring for 10 h. The solvent was removed in
vacuo, the residue was dissolved in a minimal amount of CH2Cl2 (500
µL), and the undesired salts precipitated out of solution with the addition
of Et2O (5 mL). Filtration through Celite and removal of solvent by
evaporation under reduced pressure yielded 28 (59.3 mg, 0.179 mmol,
85% yield).
0.138 mmol, 87% yield) as a light yellow oil. A small sample (∼15
mg) was chromatographed to analytical purity by HPLC (0.1% (wt/v)
TFA in water, 8.0 mL/min, 0.30% acetonitrile/min, 82-85 min). FTIR
(thin film) 2879, 2142, 1755, 1689; 1H NMR (300 MHz, CDCl3) 4.45
(dd, J ) 3.85, 12.1 Hz, 1H), 4.31 (q, J ) 7.14 Hz, 2H), 4.27 (m, 1H),
3.85 (t, J ) 4.95, 1H), 3.80 (br s, 1H), 3.67 (br s, 21H), 1.32 (t, J )
7.14 Hz, 3H); MS m/z 435.2 (H+).
Results and Discussion
Copper(I) and Silver(I). The ESI-MS spectrum for a mixture
of copper(I) and 1 is shown in Figure 1a. The base peak
corresponds to [1 + MeCN + Cu]+ or 3 (for simplicity, the
corresponding structures are to the right of the mass spectra in
Figure 1). It can also be seen from Figure 1a that 1 has a high
affinity for Na+ (present as an impurity) and that Cu(I) has a
high affinity for acetonitrile. As seen in Figure 1b, isolation of
3 followed by CAD leads to two sequential losses of 28 Da.
The first loss of 28 Da corresponds to the loss of N2 from the
diazo functionality leading to structure 4. The second loss of
28 Da results from Wolff rearrangement of 4 accompanied by
a loss of CO yielding 5. In Figure 1c, structure 5 is isolated
and subjected to further CAD resulting in another loss of 28
Da. This is also attributed to Wolff rearrangement of 5 followed
by the loss of a second CO yielding the stable copper Fischer
carbene 6. Figure 1d shows the results of CAD of structure 6.
In this case, the noncovalent acetonitrile adduct is lost, followed
by the pickup of either water or methanol. This pickup is not
surprising given the vacant copper(I) coordination site and the
fact that the spectra were acquired from a water/methanol
solution.
DFT calculations at the B3LYP level using the LACVP**
basis set were performed to determine structures and relative
energetics for the products and probable intermediates in Figure
1. The structures and energetics are presented in the reaction
coordinate diagram in Figure 2. When a ligand such as N2 or
CO is lost, the energy of the minimized separated molecule is
added to compare with the complex prior to dissociation.
Structure 3 is found to be the global minimum, and there is
support for this binding mode from crystal structures.17 Direct
dissociation of N2 from 3 is unlikely because the copper(I) ligand
restrains the geometry of the molecule in a tripletlike conforma-
tion, whereas the thermal dissociation of N2 must yield the
singlet state.18 Furthermore, 1 has a singlet ground state,6 and
attempts to minimize structure 3 with the N2 removed were
Diazomalonate 29. To a stirred, dry solution of 18-crown-6-
methanol (50.0 µL, 0.159 mmol), CH2Cl2 (1.5 mL), and Et3N (33 µL,
0.237 mmol) was added ethylmalonyl chloride (28 µL, 0.219 mmol).
The mixture was heated to reflux for 8 h, the solution was cooled, and
then the solvent was removed by evaporation under reduced pressure.
The residue was dissolved in MeCN (750 µL) and treated with Et3N
(30 µL, 0.215 mmol). To this solution was added p-acetamidoben-
zenesulfonyl azide (53.1 mg, 0.221 mmol), and the mixture was stirred
for 10 h. The solvent was removed by evaporation under reduced
pressure, the residue was dissolved in a minimal amount of CH2Cl2
(500 µL), and the undesired salts were precipitated out of solution with
the addition of Et2O (5 mL). Filtration through Celite and removal of
solvent by evaporation under reduced pressure yielded 29 (59.8 mg,
(16) Lamb, J. D.; Izatt, R. M.; Swain, C. S.; Bradshaw, J. S.; Christensen, J. J.
J. Am. Chem. Soc. 1980, 102, 479-482.
(17) Dias, H. V. R.; Polach, S. A. Inorg. Chem. 2000, 39, 4676-4677.
9
4480 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003

Synthesis of Charged Cu and Ag Fischer Carbenes
A R T I C L E S
Figure 1. Structures for the peaks in the mass spectra (left) given in the reaction scheme (right). (a) MS of a mixture of 1 and Cu (I) in a 20:80 (v/v)
water/methanol solution with ∼0.1% MeCN. (b) MS2 on complex 3. N2 and CO are lost sequentially. (c) MS3 on complex 5 resulting in the loss of an
additional CO. (d) MS4 on 6 resulting in the exchange of the MeCN ligand for either water or methanol.
unsuccessful. Matrix isolation studies on similar systems suggest
that copper(I) insertion into the C-N bond of 3 is a more likely
alternative.19
Similarly, calculations reveal a ∆H of -43 kcal/mol for reaction
2, which is the binding energy of MeCN to the 1‚Cu complex.
This value is in reasonable agreement with similar experimental
and theoretical results.20 The MeCN ligand is retained through-
out the entire series of reactions shown in Figure 2 without
dissociating. Therefore, the reaction barriers for each step must
be below 43 kcal/mol, otherwise the MeCN ligand would simply
dissociate. Furthermore, the binding energy of the sodium ion
is much greater at 59 kcal/mol, yet the ion dissociates prior to
the loss of N2 and the generation of the carbene. This
demonstrates that copper(I) lowers the activation barrier for the
generation of a carbene by at least 16 kcal/mol relative to
complexation with sodium.
Further experimental and theoretical evidence also suggests
that the copper(I) ion mediates the generation of the carbene
from the diazo precursor. CAD experiments on [1 + Na]+ result
in the complete loss of signal without producing any observable
peaks, suggesting dissociation of the sodium ion (which has an
m/z ratio that is too small to detect). Similar experiments with
rubidium yield a peak corresponding to Rb+ exclusively from
the collisional activation of [1 + Rb]+. The calculated ∆H for
reaction 1 is -59 kcal/mol, which can be regarded as the binding
energy of the sodium ion to 1.
The reaction shown in Figure 2 proceeds by rearrangement
of structure 3 to structure 10 followed by Cu+ insertion as shown
1 + Na+ f 1‚Na+
1‚Cu + MeCN f 3
(1)
(2)
(19) Chang, S.-C.; Kafafi, Z. H.; Hauge, R. H.; Billups, W. E.; Margrave, J. L.
J. Am. Chem. Soc. 1987, 109, 4508-4513.
(20) Vitale, G.; Valina, A. B.; Huang, H.; Amunugama, R.; Rodgers, M. T. J.
Phys. Chem. A 2001, 105, 11351-11364. (b) Chu, Y.; Yang, Z.; Rodgers,
M. T. J. Am. Mass Spectrom. 2002, 13, 453-468.
(18) Wentrup, C. ReactiVe Molecules: The Neutral ReactiVe Intermediates in
Organic Chemistry; John Wiley & Sons: New York, 1984; pp 177-264.
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4481

A R T I C L E S
Julian et al.
Figure 2. Diagram illustrating the energetics of reaction intermediates in Figure 1. The energetics are calculated relative to structure 3 and are given in
parentheses (in kcal/mol). When a neutral gas molecule is lost, the energy of the minimized, separated molecule is added to that of the remaining structure.
The barriers at each step have not been calculated but are below the binding energy of the MeCN ligand (which is calculated to be -43 kcal/mol as
explained in the text).
in 11. Crystal structures of copper(I) Fischer carbenes have
C-Cu bond lengths similar to those in 11 and, in some cases,
demonstrate a high affinity for MeCN ligands.21 Copper
insertion is postulated as a prelude to the Wolff rearrangement
from 4 to 12, which must proceed through a singlet state. There
is poor agreement between theoretically and experimentally
determined barriers to Wolff rearrangement, with theory predict-
ing higher values than those observed experimentally.22 Theory
would predict a large barrier for the conversion of 4 to 12
because of the exothermicity of the reaction.22b The experimental
evidence in Figure 1b suggests that both rearrangement of 4 to
12 and subsequent loss of CO to produce 5 proceed with
minimal barriers. It should be pointed out that, given a low
barrier to rearrangement, the structure for the observed peak in
Figure 1b may be the rearranged product (12) rather than carbene
(4).23 In either case, the observed products suggest that the
presence of copper lowers the barrier to Wolff rearrangement
in addition to facilitating the generation of the initial carbene.
The Wolff rearrangement product of 5 undergoes copper
insertion without barrier upon minimization, suggesting that the
transition state may occur prior to rearrangement. Therefore,
the loss of the second CO likely occurs in a concerted fashion
without a true ketene intermediate. The data in Figure 1b and
c show that the initial two losses occur simultaneously, which
suggests a higher barrier to the loss of the second CO. However,
it should be noted that the noncovalently bound MeCN ligand
is retained throughout the entire process, limiting the reaction
barriers to the bond dissociation energy of the Cu-NCMe bond
(or 43 kcal/mol). Finally, a stable Fischer carbene (6) is
produced upon loss of the second CO. Subsequent excitation
of this complex results in the loss of the MeCN ligand followed
by attachment of either H2O or MeOH from the residual solvent
vapor present in the ion trap (Figure 1d).
In the presence of silver(I), structure 13 (Figure 3a) is formed
in high abundance. However upon collisional activation in
Figure 3b, the MeCN ligand is lost instead of N2 as was the
case with copper(I). This difference is due to the weak binding
of the third ligand to the silver(I) cation.24 Loss of the MeCN
adduct is followed by collisional cooling of the product (14),
which leads to some pickup of residual methanol (15) or water
(16) from the trap as seen in Figure 3b. Further CAD of
reisolated 14 leads to the exclusive loss of 56 Da, comprising
N2 and CO with no intermediate loss of N2 being observed.
Again these losses are followed by the pickup of methanol or
water. The absence of an intermediate loss of N2 suggests that
silver(I) is more efficient at facilitating the Wolff rearrangement
and subsequent loss of CO than copper(I), consistent with the
predominant use of silver(I) as the catalyst where Wolff
rearrangement is required for a synthetic route.4a Isolation and
activation of 17 leads to the expected loss of the second CO,
yielding structure 20 and ensuing pickup of water or methanol.
The silver Fischer carbenes 20, 21, and 22 are the final
observable carbene products in this sequence. Further CAD of
20 leads to the loss of the Ag+ cation, presumably generating
the neutral dimethoxy carbene in the process.
(21) Barluenga, J.; Lopez, L. A.; Lober, O.; Tomas, M.; Garcia-Granda, S.;
Alvarez-Rua, C.; Borge, J. Angew. Chem., Int. Ed. 2001, 40, 3392-3394.
(b) Tulloch, A. A. D.; Danopoulos, A. A.; Kleinhenz, S.; Light, M. E.;
Hursthouse, M. B.; Eastham, G. Organometallics 2001, 20, 2027-2031.
(22) Likhotvorik, I.; Zhu, Z.; Tae, E. L.; Tippmann, E.; Hill, B. T.; Platz, M. S.
J. Am. Chem. Soc. 2001, 123, 6061-6068. (b) Scott, A. P.; Platz, M. S.;
Radom, L. J. Am. Chem. Soc. 2001, 123, 6069-6076.
(23) Similarly, it is impossible to distinguish experimentally whether structures
5 and 17 have already undergone Wolff rearrangement prior to experimental
observation.
(24) Shoeib, T.; Aribi, H. E.; Siu, K. W. M.; Hopkinson, A. C. J. Phys. Chem.
A 2001, 105, 710-719.
9
4482 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003

Synthesis of Charged Cu and Ag Fischer Carbenes
A R T I C L E S
Figure 3. Spectra acquired from a mixture of silver(I) and 1. The structures are given to the right of the appropriate spectrum. (a) MS spectrum showing
complexation with Ag+. (b) MS2 spectrum of 13, where the loss of MeCN generates the most abundant product ion peak. (c) MS3 spectrum of 14 showing
the facile loss of N2 and CO. (d) MS4 spectrum of 17 showing loss of second CO and generation of final silver Fischer carbenes. (e) MS5 spectrum of 20,
further CAD results in the cleavage of the metal-carbon bond.
Labeling Experiments. Isotopic labeling experiments were
performed to confirm the proposed reaction pathways. The
results for the two labeled compounds, 25 and 26, are given in
Table 1. Structure 25 is labeled with three deuteriums on each
of the terminal methyl groups. Comparison of the data in Table
1 with that shown in Figure 1 reveals that the six deuteriums
are retained throughout the entire experiment. This indicates
that only interior carbons are lost as CO and that there is no
detectable scrambling with MeOH in the ion trap. To confirm
which carbons are lost and to determine the extent to which
rearrangement of the carbene occurs, structure 26 was synthe-
sized with 13C at the 1,3 positions. Experiments with 26 confirm
that the two interior 13C labeled carbons are lost, as shown in
Table 1. There is no apparent rearrangement of the carbene via
an oxirene intermediate as depicted in Scheme 2.11 These two
experiments serve to confirm that N2 is lost first, as is expected,
and that carbon in the 1 and 3 positions, specifically, are lost
as CO.
Divalent Metals, Protons, and Fixed Charges. The experi-
mental MS data for several related experiments on 1 are
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4483

A R T I C L E S
Julian et al.
Table 1. Summary of MS Results.
molecule
attached cation
initial ligands
observed peaksa
25
Cu+
MeCN
MS2 on 268f 193(12), 212(100), 240(15), 268(21)
MS3 on 212f 161(11), 184(100), 212(80)
MS4 on 184f 143(5), 161(43),b 175(8),b 184(100)
MS2 on 264f 207(100), 236(11), 264(2)
26
Cu+
MeCN
MS3 on 207f 178(100), 207(98)
1
1
1
1
2
Cu2+
Cu2+
Ni2+
H+
2(1)
2MeCN
2(1)
MS2 on 268.4f 189.5(100), 268.4(<1)
MS2 on 151f 140(100)*, 147(50)b
MS2 on 266f 187(30), 196(100),b 203(72),b 266(<1)
MS2 on 159) 55(80), 69(6), 87(100), 101(8), 127(93), 145(28), 159(<1)
MS2 on 414f 345(8), 373(100), 414(10)
Cu+
MeCN
MS3 on 373f 289(55), 317(37), 345(42), 373(100)
2
Ag+
MS2 on 417f 289(12), 317(17), 333(39), 345(25), 361(96), 389(100), 417(16)
MS3 on 389) 317(10), 333(26), 361(69), 389(100)
27
see structure
MS2 on 264f 113(8), 157(13), 236(100), 264(22)
MS3 on 236f 157(100), 180(42), 192(40), 208(61), 236(57)
a Peaks are given with accompanying relative intensities in parentheses. Some low-intensity peaks (under 15% relative intensity) that are not assigned
structures or discussed in the text are omitted for clarity. b Solvent molecule pickups.
Scheme 2. Potential Carbene Scrambling via Oxirene (Not
Observed)
The pattern of losses for molecule 27 contrasts sharply with
the results obtained for all of the copper(I) and silver(I) adducts.
Metal adducts of 1 lose N2 and the first CO in one step without
the addition of further excitation energy. The loss of the second
CO occurs upon further activation, but no other competitive
products are produced in significant abundance at any stage of
the experiment. In the case of 27, N2 loss occurs without any
accompanying loss of CO. Upon further excitation of [27 -
N2]+, Wolff rearrangement followed by the loss of the first and
second CO molecules occurs with the appearance of several
other competitive products. In other words, Wolff rearrangement
products are no longer favored in the absence of a metal ligand.
summarized in Table 1. For both copper(II) and nickel(II), the
metal is complexed with three molecules of 1 or [3(1) + M]2+
(where M ) Ni or Cu), with the ligands coordinating the metal
in a pseudooctahedral fashion. CAD of these complexes does
not lead to the loss of N2 but instead yields only the loss of an
entire coordinating molecule. A small amount of [1 + Cu +
2MeCN]2+ is also formed from the copper solution. However,
as the data show in Table 1, this complex simply loses one of
the MeCN ligands upon CAD. These results illustrate that
copper(II) and nickel(II) are not efficient at mediating the
formation of Fischer carbenes or subsequent Wolff rearrange-
ments. Protonated 1 does not lose N2 upon CAD either but
instead yields the assortment of fragments shown in Table 1.
The results presented thus far suggest that the coordinating
charge can dictate the resulting chemistry upon collisional
activation in diazo compounds. Structure 27 was designed to
investigate the energetics of N2 loss and Wolff rearrangements
in the absence of a coordinating charge. In 27, the charge is
provided by a fixed quaternary nitrogen from the pyridinium
group. As seen in Table 1, this compound loses N2 to yield the
most abundant product by CAD in the MS2 spectrum. Further
collisional activation of the resulting molecule demonstrates that
Wolff rearrangement and the loss of multiple CO molecules
does occur. However, this process is accompanied by other
losses not related to Wolff rearrangement. In fact, the loss of
pyridine yields the base peak in the MS3 spectrum. This suggests
that the energetics associated with Wolff rearrangement and loss
of CO for this reaction are similar to those for the loss of
pyridine.25
These results suggest that copper(I) and silver(I) facilitate the
Wolff rearrangement and loss of CO shown in Figures 1 and 2.
Effect of Benzyl Groups. In structure 2, the methyl groups
are replaced by benzyl groups, and with the addition of
copper(I), the ESI-MS contains a prominent [2 + Cu + MeCN]+
peak. As shown in Table 1, CAD dissociation of this peak leads
primarily to the loss of the MeCN ligand. No appreciable pickup
of water or methanol is observed following the loss of MeCN.
This suggests that the benzyl groups are able to coordinate the
metal ion sufficiently so that the lost ligand is not replaced.
There is precedence for this type of coordination from crystal
structure studies.26 Interestingly, further CAD of [2 + Cu]+ leads
to the sequential loss of N2, CO, and CO with each intermediate
being observed. No further collisional activation steps are
necessary to observe the loss of all three molecules. Similar
experiments with silver(I) and 2 yield very similar results (Table
1), but in this case, the [2 + Ag]+ peak is formed directly from
(25) Katritzky, A. R.; Watson, C. H.; Dega-Szafran, Z.; Eyler, J. R. J. Am. Chem.
Soc. 1990, 112, 2471-2478.
9
4484 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003

Synthesis of Charged Cu and Ag Fischer Carbenes
A R T I C L E S
Table 2. MS Data for Crown Compounds
molecule
attached cation
observed peaksa
28
Na+
MS2 on 355f 327(100), 355(6)
MS3 on 327f 255(13), 271(6), 299(100), 327(5)
MS4 on 299f 225(100), 255(40), 271(73), 299(18)
MS2 on 395f 339(26), 367(100), 395(18)
28
Cu+
MS3 on 367f 339(94), 367(100)
MS4 on 339f 265(16), 311(100), 339(38)
+
28
28
29
NH4
MS2 on 350f 173(11), 217(12), 261(100), 333(33), 350(72)
MS2 on 197.5f 175.5(100), 184.3(6),b 191.2(21),b 197.5(5)
MS2 on 452f 380(12), 424(100), 452(61)
Cu2+
+
NH4
MS3 on 424f 175(26), 203(21), 247(19), 291(19), 351(44), 424(100)
MS2 on 457f 385(3), 429(100), 457(9)
29
Na+
MS3 on 429f 299(10), 357(29), 373(100), 385(10), 401(20), 429(30)
a Peaks are given with accompanying relative intensities in parentheses. Some low-intensity peaks (under 15% relative intensity) that are not assigned
structures or discussed in the text are omitted for clarity. b Solvent molecule pickups.
solution in high abundance. Because the metal ion is coordinated
to 2 by the benzyl groups, the relative conformation of the diazo
functionality to the esters is different for copper(I) and
silver(I) adducts of 1 and 2. DFT calculations suggest the
silver(I) adduct structure shown in 30. The fact that Wolff
rearrangement proceeds more easily for 30 than for 3 suggests
that the conformational energy minimum is similar to the
preferred conformational orientation that leads to Wolff rear-
rangement for 30. Conformational effects have been observed
to influence Wolff rearrangements.11
Macrocycles. Crown ethers and other related macrocycles
are well suited to coordinate a variety of metal ion and charged
functional groups27 and often mitigate against the attachment
of other solvent adducts once in the gas phase. Therefore,
structures 28 and 29 are well suited to study the proximal effects
of various cations and different coordination motifs on the Wolff
rearrangement in the absence of weakly bound ligands. The
diazo group for 28 is included in the ring, bringing the charge
in close proximity to the diazo group. However, it should be
noted that the coordination geometry between the charge and
the diazo is significantly different for 1 and 28 due to
conformational constraints imposed by the macrocycle. Structure
29 is more flexible, allowing for either coordination or separation
of the metal and the diazo group.
The CAD results for 28 coordinated to copper(I) and sodium
are presented in Table 2. In both cases, the loss of N2, CO, and
CO occurs predominantly one step at a time, requiring the
acquisition of an MS4 scan to induce the loss of the second
CO. This suggests that the more conformationally restrained
macrocycle requires additional activation energy to induce Wolff
rearrangement but does not necessarily exclude participation
of the cation in the reaction. The resonance excitation rf voltages
applied at each step of the experiment are (for sodium) 0.67,
0.83, and 0.99 V and (for copper) 0.69, 0.82, and 0.71 V,
respectively. The excitation voltages are very similar for the
first two steps of the experiment for both cations, but copper(I)
clearly catalyzes the loss of the second CO more efficiently
than sodium (∆V ∼0.3 V). This is additionally confirmed by
the absence of competitive fragments in the copper(I) experi-
ment, whereas in the case of sodium, the base peak in the MS4
spectrum is not the product of Wolff rearrangement.
The copper(II) adduct of 28 can also be prepared in small
abundance, and the results for the CAD are found in Table 2.
Copper(II) catalyzes the loss of 44 m/z, which is presumably
the loss of ethylene oxide from the crown ether portion of the
molecule.28 The loss of N2 is not detected, again suggesting
that copper(II) does not catalyze the formation of Fischer
carbenes or the subsequent Wolff rearrangement observed with
other cations. CAD of the ammonium adduct of 28 is also
presented in Table 2. In this case, the loss of NH3 followed by
subsequent cleavage of protonated 28 is the only process
observed. This suggests that close proximity of a labile proton
can prevent the loss of N2 and subsequent reactions from
occurring. However, it should be noted that this problem is easily
corrected by complexation with a more basic amine such as
1-hexylamine. CAD of the hexylamine adduct leads primarily
to the loss of N2 (data not shown). Alternatively, weaker binding
by ammonium versus 1-hexylamine may explain the dissociation
in the case of ammonium. These experiments confirm the notion
that the cation can greatly influence the subsequent chemistry
upon excitation of these molecules.
Charge Proximity Effects. In the copper(I) and silver(I)
adducts of 2, the metal ion is coordinated to the benzyl groups
and is not in close proximity to the diazo group. It is not
surprising that the resulting CAD spectra are very similar for
both metals. In general, charges that are not in close proximity
to the diazo group have less influence on the resulting CAD
patterns. This is well illustrated by comparing 28 and 29.
Molecular dynamic simulations suggest that the collisional
heating of 29 leads to extended structures where the lariat
sidearm extends away from the charge. This leads to a large
separation between the charge and the diazo group. For structure
28, heating of the quasimolecular complex does not lead to
significant structural changes, leaving the charge in close
proximity to the diazo group. The net effect is that the diazo
group is ∼2 Å further away from the charge in 29 when
compared to 28. CAD of [28 + NH4]+ does not lead to the loss
of N2 or subsequent Wolff rearrangement products. By contrast,
when 29 is complexed with ammonium, CAD leads to the loss
of N2 and formation of a carbene (Table 2). This implies that
the ammonium ion does not interact with the diazo group in
29, but this does not rule out subsequent interactions between
the newly generated carbene and the cation. Interestingly,
subsequent CAD of [29 - N2 + NH4]+ does not lead to any
Wolff rearrangement products. This is attributed to the high
(26) Dattelbaum, A. M.; Martin, J. D. Inorg. Chem. 1999, 38, 6200-6205. (b)
Wen, M.; Munakata, M.; Suenaga, Y.; Kuroda-Sowa, T.; Maekawa, M.
Inorg. Chim. Acta 2002, 332, 18-24.
(27) Bradshaw, J. S.; Izatt, R. M.; Bordunov, A. V.; Zhu, C. Y.; Hathaway, J.
K. ComprehensiVe Supramolecular Chemistry; Gokel, G. W., Ed., Perga-
mon/Elsevier: Oxford, U.K., 1996; Vol. 1, pp 35-95.
(28) Maleknia, S.; Brodbelt, J. J. Am. Chem. Soc. 1993, 115, 2837-2843.
9
J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003 4485

A R T I C L E S
Julian et al.
Table 3. Intermolecular Reactions with Fischer Carbenes.
cleavages are complicated and it is not our intention to fully
describe them here. However, it is shown from the results that
alkynes are much more reactive than analogous alkenes. This
is not an unusual result given the greater reactivity of alkynes
in general. Similarly, the observed reactivity of the phenyl group
was unexpected. In the case of 2, the benzyl groups ultimately
displaced the ligand and prevented any reactions from occurring.
attached
cation
partial loss
of ligand
covalent bond
cleavagea
molecule
initial ligand
1
2
1
1
Cu+
Cu+
Cu+
Cu+
5-hexynenitrile
5-hexynenitrile
5-hexenenitrile
yes
yes
yes
yes
MS3
no
no
4-phenylbutyronitrile
MS3
a If covalent bond cleavage occurs either competitively with loss of the
ligand or exclusively, the step where this occurs is indicated.
It is also interesting to note that CAD of [1 + hexynenitrile
+ Cu]+ yielded the loss of a neutral methyl radical in the MS3
spectrum. This unusual loss was confirmed by experiments with
26 in which the loss of CD3 was observed. The loss of methyl
radical is accompanied by the pickup of water or methanol. It
is unlikely that the methyl group was a metal ligand prior to
dissociation, so an alternative explanation is required to explain
the pickup of water and methanol. One possible explanation
for this involves the oxidation of the copper(I) to copper(II)
accompanied by the reduction of the newly formed terminal
CO2. The oxidation to copper(II) would create a new vacant
ligand site and lead to the observed pickups. The exact role of
the alkyne in this process remains unclear, but it may serve to
stabilize the higher oxidation state of the copper.
proton affinity of the carbene,29 which reacts with the am-
monium and prevents Wolff rearrangement. CAD of 29 com-
plexed with sodium leads to the expected loss of N2 and the
formation of both Wolff rearrangement products in the MS3
spectrum. Alternatively, weaker binding of ammonium by 28
relative to 29 may explain the difference in experimental results.
Conformational Effects. The degree of conformational
restraint present in a molecule can favor or disfavor subsequent
Wolff rearrangement. It was established in the previous
paragraph that copper(I) and silver(I) adducts of 2 behave very
similarly with the simultaneous loss of all products, but it should
be noted that the CAD spectra of these two complexes are not
similar to any others obtained in the course of this study. We
attribute this to a conformational effect where the metal restrains
the molecule in a conformation that favors Wolff rearrangement.
The opposite effect is observed in [28 + Na]+, which is also
conformationally restrained. In this case, Wolff rearrangement
is not favored, and additional activation energy must be added
at each step. Interestingly, the barrier becomes much larger when
the diameter of the macrocycle is reduced. This supports the
proposition that conformational constraints lead to the larger
activation barriers required for the CAD of 28 bound to various
ions. Unconstrained systems with a remote charge that does not
react with the carbene such as 27 and [29 + Na]+ all
demonstrate similar behavior upon CAD.
Intermolecular Reactions. Given the spectator MeCN ligand
present in 3, the possibility for intermolecular reactions might
exist given the addition of the proper ligand. The results for
attaching three different ligands are summarized in Table 3. It
is easily seen in Table 3 that in each of these experiments the
ligand is lost (either competitively or exclusively) at some point
in the experiment. This suggests that the binding energy of the
ligand to the copper(I) cation is not greatly increased by the
presence of the additional functional groups. In particular,
5-hexynenitrile is not strongly bound to copper(I). Therefore,
if the loss of the weakly bound noncovalent ligand is not
observed, but a covalent bond cleavage occurs that interrupts
the established sequential rearrangement processes (i.e., loss of
N2, CO, and CO), then it is assumed that an intermolecular
reaction has occurred. The reactions that lead to subsequent
Conclusions
In summary, the synthesis of copper(I) and silver(I) Fischer
carbenes from various diazo malonates in the gas phase is
demonstrated for the first time. The carbenes are generated by
the facile loss of N2. The carbenes prepared by this process
can undergo multiple Wolff rearrangements in the gas phase,
which subsequently lead to the loss of multiple carbon mon-
oxides. Surprisingly, up to six different carbenes or metallo-
carbenoids can be produced in a single experiment. A series of
control experiments that elucidate the effects of conformation
and metal mediation on Wolff rearrangements are detailed.
Although the data have been gathered from gas-phase experi-
ments, the results reveal general trends that should be applicable
for either enhancing or deterring Wolff rearrangements in
solution. Coordinated metal ions profoundly mediate the
energetics of these reactions. Silver(I) is most efficient at
initiating Wolff rearrangements followed by copper(I), but no
mediation is required, and molecules labeled with nonparticipat-
ing charges undergo similar chemistry at higher energies.
Conformational effects are also found to be important in
determining the requisite energy for Wolff rearrangement.
Divalent metal ions and protons interfere with the loss of N2
and do not promote the formation of carbenes.
Acknowledgment. The authors gratefully acknowledge fund-
ing provided by NSF (Grant CHE-9727566), the Beckman
Institute Foundation, the California Institute of Technology, the
Camille and Henry Dreyfus Foundation (New Faculty Award
to B.M.S.), and Abbott Labs (graduate fellowship to J.A.M.).
(29) Vogt, J.; Beauchamp, J. L. J. Am. Chem. Soc. 1975, 97, 6682-6685. (b)
Pliego J. R.; De Almeida, W. B. J. Chem. Soc., Faraday Trans. 1997, 93,
1881-1883.
JA028337J
9
4486 J. AM. CHEM. SOC. VOL. 125, NO. 15, 2003