G

M

Ub

Ha

b

c

N

ARRAA

DS

KORMCD

1

dStcv(aabtcitcsf

e

1h

ARTICLE IN PRESS Model

ASPEC-14953; No. of Pages 10

International Journal of Mass Spectrometry xxx (2013) xxx– xxx

Contents lists available at SciVerse ScienceDirect

International Journal of Mass Spectrometry

j ourna l ho me page: www.elsev ier .com/ locate / i jms

nimolecular dissociation characteristics of cationic complexesetween nicotinic acid and Cu(II) and Ni(II)

éloïse Dossmanna, Carlos Afonsoa,b, Jean-Claude Tabeta, Einar Uggerudc,∗

Institut Parisien de Chimie Moléculaire, Université Pierre et Marie Curie-Paris 6, UMR 7201-FR2769, Place Jussieu, F-75252 Paris Cedex 05, FranceUniversité de Rouen, UMR CNRS 6014 COBRA, 1 rue Tesnière, 76130 Mont-Saint-Aignan, FranceMassespektrometrilaboratoriet og Senter for teoretisk og beregningsbasert kjemi (CTCC), Kjemisk institutt, Universitetet i Oslo, Postboks 1033 Blindern,-0315 Oslo, Norway

a r t i c l e i n f o

rticle history:eceived 15 April 2013eceived in revised form 28 May 2013ccepted 28 May 2013vailable online xxx

edicated to the memory of Detlefchröder.

a b s t r a c t

Cu2+ and Ni2+ form dimeric ML(L−H)+ complexes with nicotinic acid (M = Cu, Ni; L = nicotinic acid) uponelectrospray ionization. Quantum chemical calculations indicate thermochemical preference for coordi-nation of the carboxylate groups rather than the ring nitrogens to the central metal ion in both cases. Inanalogy to the dimeric metal complexes of amino acids the primary dissociation reaction upon collisionalactivation of ML(L−H)+ is the loss of CO2 in both cases. Further dissociation of the decarboxylated speciesshow preference for loss of a 3-pyridinyl radical for M = Cu and NiCO2 for M = Ni. This can be understoodin light of the redox properties of the two metals and from previous studies of similar complexes with

eywords:rganometallic chemistryeactivityass spectrometry

O2 activation

amino acids. Loss of the pyridinyl radical bonded to the carboxylate group in these cationic entities doesnot lead to M(�2-O2C) structures previously observed for similar anionic metal species.

© 2013 Elsevier B.V. All rights reserved.

ecarboxylation

. Introduction

Recently, a new bidentate structural binding motif of carbonioxide to various metals and metal ions has been described [1–4].uch tetragonal structures (Scheme 1) can be formed by elec-rospray ionization (ESI) of liquids containing suitable precursorompounds in combination with collisional activation and haveery interesting chemical properties. For example XMg(�2-O2C)−

X = Cl, Br or OH) formed from mixtures of MgX2 salts and oxaliccid has been shown to be a carbon nucleophile and may serve as

model system for carbon dioxide activation during organic andiological C C bond formation reactions, for example during pho-osynthesis. Carboxylation reactions of this kind are of significanturrent interest as a promising method for CO2 sequestration inndustrial processes for combined combustion and chemical syn-hesis [5]. In this respect it is of great interest to investigate how

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

hemical factors (the metal, the charge state, complexation andolvent) influence the stability of this binding motif. In our questor suitable model systems of this kind we have chosen a simple

∗ Corresponding author. Tel.: +47 22 85 55 37; fax: +47 22 85 54 41.E-mail addresses: heloise.dossmann@upmc.fr (H. Dossmann),

inar.uggerud@kjemi.uio.no (E. Uggerud).

387-3806/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.ijms.2013.05.030

strategy, namely to mix a salt of the metal in question with car-boxylic acids and study the collisionally induced decomposition ofthe ionic complexes formed in situ in the electrospray process. Inthe present context we chose to study the M(II) salts of nickel andcopper, since they are among the most interesting metals both forindustrial catalysis and in biological systems. On the other hand,copper and nickel have quite different redox properties. While cop-per normally displays oxidation state 0, I and II, oxidation state I israre for nickel. The former metal is characterized for its flexibility inoxygen processing and radical initiated processes, the latter by itscatalytic role in hydrogenation/dehydrogenation reactions (Raneynickel) and in C C activation.

The simplest conceivable Cu2+ and Ni2+ complex structuresformed with carboxylates are of the type (R-CO2)2M, providinga conceptual starting point for forming precursors that by directloss of R may lead to structures of the type (R-CO2)M(�2-O2C)upon activation, see Scheme 1 (the �2-O2C notation indicates thatboth oxygens are bonded to the metal). In order to form positivelycharged ions of such (R-CO2)2M complexes by ESI it is neces-sary that the ligand in addition to a carboxylic acid residue also

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

contains a sufficiently basic site, for example an amino group topick up an extra proton. On the other hand, amino groups alsohave affinity for cationic transition metal ions so within the ligandthere will be competition between metal complexation of the

ARTICLE ING Model

MASPEC-14953; No. of Pages 10

2 H. Dossmann et al. / International Journal of M

citcdbrce[

aloattucwc

2

2

2

Q4pt

2

teiHlowctrtwiwwoct

Scheme 1.

arboxylate and the amino groups, leading to structural complex-ty. This is a well-known situation for amino acids, which give riseo various monomeric, dimeric and trimeric complexes with bothopper and nickel in ESI mass spectrometry, and the unimolecularissociation of Cu(II) amino acid and carboxylic acid complexes haseen particularly well studied by mass spectrometry due to the richadical cation chemistry resulting from the open shell d9 electroniconfiguration of Cu2+ [6–13]. On the other hand Ni2+ is d8, an evenlectron species with lower tendency towards radical chemistry14–16].

The structural flexibility of amino acids allows for strong inter-ction between the metal ion and several functional groups of theigand. To overcome this issue we decided to investigate complexesf nicotinic acid since this ligand is both simpler and more rigid thanmino acids, avoiding simultaneous binding to the nitrogen andhe carboxylic acid/carboxylate group of the same ligand moleculeo the metal. Different mass spectrometric approaches have beensed in order to determine the binding patterns of the formedomplexes as well as their dissociation behaviour. These effortsere complemented by density functional theory (DFT)-based

alculations.

. Methods

.1. Experimental details

.1.1. GeneralAll chemicals have been purchased from Sigma–Aldrich (St.

uentin Fallavier, France) and used without further purification.0 �M solutions of MCl2 (M = Cu, Ni) and nicotinic acid were pre-ared in methanol. Working solutions were prepared by mixinghese two solutions in a 5:1 ratio (v:v).

.1.2. Ion trapExperiments were in part performed on a quadrupole ion

rap mass spectrometer (Esquire 3000, Bruker, Bremen, Germany)quipped with an orthogonal ESI source. Sample solutions werenfused with a syringe pump model 74900 (Cole-Parmer, Vernonills, IL) at a flow rate of 160 �L h−1. Nitrogen was used as nebu-

izing gas at a pressure of 6 psi, and as drying gas at a temperaturef 250 ◦C and a flow rate of 4 L min−1. Optimized source voltagesere as follows: capillary at −3.5 kV, end plate offset at −500 V,

apillary exit (CE) at +45 V, skimmer 1 at +15 V (providing a poten-ial difference of 30 V with CE), and skimmer 2 at +6 V. Theseelatively soft source conditions were used in order to preservehe rather weakly complexed ions. The low mass cut-off (LMCO)as fixed at 28% of the m/z of the precursor ions and the analyt-

cal scan range for mass spectra was m/z 50–600. The scan rateas set at 13,000 m/z s−1 (standard mode). Ion accumulation time

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

as automatically set with ion charge control (ICC) with a targetf 10,000 to limit space charge effects. For low-energy sequentialollisionally induced dissociation (CID) experiments, resonant exci-ation was used with an amplitude voltage of 0.70 Vp–p and an ion

PRESSass Spectrometry xxx (2013) xxx– xxx

isolation window of m/z 0.8 in order to obtain monoisotopic ionselection.

2.1.3. FT-ICRA hybrid quadrupole Fourier transform ion cyclotron resonance

(hQh-FT/ICR) mass spectrometer (Solarix, Bruker Daltonics, Bre-men, Germany) equipped with an actively shielded 7 T magnet wasused for accurate mass measurements. The samples were infusedin the electrospray ion source at a flow rate of 120 �L h−1 withthe assistance of N2 nebulizing gas. Ionization was performed inthe positive ion mode with an ESI high voltage of 4000 V. Voltageapplied in the desolvation zone were: capillary exit 200 V, deflec-tor 180 V, funnel 110 V, skimmer 50 V. Complexes of interest werefirst mass-selected with the quadrupole (m/z 5 window), accumu-lated for 0.5 s in a linear ion trap and then isolated in the ICR cell inorder to obtain monoisotopic precursor ion selection (13% of fre-quency sweep amplitude). Activation of the ions was performedusing sustained off-resonance irradiation collision-induced disso-ciations (SORI-CID) using argon as collision gas introduced througha pulse valve. The ions were excited using excitation amplitudeof 0.5 Vp–p with a frequency offset of −500 Hz applied for 200 ms.A pumping delay of 2 s was used before the excitation/detectionstep.

All mass spectra and collision activation spectra were acquiredwith Solarix control (version 1.5. Bruker Daltonics) in broadbandmode from m/z 21.5 to m/z 500. The image signal was amplified anddigitized using 2 M data point resulting in the recording of a 0.4 stime domain which was transformed into the corresponding fre-quency domain by Fourier transform (one zero fill and apodizationusing the sinbell function).

The ESI mass spectra were internally calibrated from unambigu-ous signals (single point calibration). Typically the precursor ionwas used as internal calibrant for CID spectra. Reported m/z ionswere compared to the theoretical m/z and ions with an error higherthan 5 ppm were not considered.

2.1.4. Ion mobility measurementsThe ion mobility experiments were carried out using a Synapt G2

HDMS (Waters, Manchester, United Kingdom). This instrument is ahybrid quadrupole/time-of-flight mass spectrometer, which incor-porates a travelling-wave ion mobility (TWIM) device, in whichlow-voltage waves push the ions through a gas filled cell [17]. Thegas flow (N2) was set at 90 mL min−1, the travelling wave height andvelocity were set to 40 V and 600 m s−1, respectively. For compari-son, theoretical collision cross section values were computed usingthe trajectory method (TM) method with the MOBCAL software[18,19]. The number of data points in the Monte Carlo integra-tions of impact parameter and orientation was set to 1000. IMScell calibration was carried out with polyalanines at the concentra-tion of 10 ng �L−1 using collision cross section (CCS) values usedfor calibration from the Clemmer lab database [20].

2.1.5. Computational procedureQuantum chemical calculations were carried out using the pro-

gram system GAUSSIAN 09 [21]. Geometry optimization and singlepoint energies were obtained using the B3LYP method with the6-31G(2df,p) basis set. All structures presented in this work werecharacterized either as minima or saddle points from vibrationalanalysis. Relative energies were corrected by including unscaledzero-point vibrational energies (ZPVE). The accuracy of the com-putational approach was probed by comparing calculated bond

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

dissociation energies of selected copper and nickel complexeswith experimental values found in the literature. The results showsatisfying agreement and are presented in the Supplementaryinformation.

IN PRESSG Model

M

al of Mass Spectrometry xxx (2013) xxx– xxx 3

3

3

sattmoTm

3i

iaa

Fi

ARTICLEASPEC-14953; No. of Pages 10

H. Dossmann et al. / International Journ

. Results

.1. Electrospray spectra

The positive mode ESI mass spectra of the MCl2/nicotinic acidolutions display significant peaks corresponding to ML(L−H)+

nd ML2(L−H)+ (M = Cu, Ni; L = nicotinic acid) in agreement withhe theoretical isotope masses and distributions. Depending onhe nicotinic acid concentration, we also observe the protonated

onomer and dimers of nicotinic acid. At higher concentrationsf CuCl2, we also identify a peak corresponding to Cu2L(L−H)•+.he metal complexes were also seen to form mono-adducts withethanol from the solvent.

.2. Binding patterns in the neat cluster ions, computationalnsight

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

In the present context, the molecular structures of the ML(L−H)+

ons (M = Cu, Ni) are of key interest. Nicotinic acid (L) and itsnion (L−H)− allow for various binding patterns, and the questionrises whether there is preference for binding the ligand L = NAH

Fig. 1. Geometry optimized structure of the most stable [Cu+2NAH−H]•+

ig. 2. Geometry optimized structures of the two most stable [Ni+2NAH−H]+ complexes

n degrees.

Scheme 2.

to the metal core via the ring nitrogen (N side) or the carboxylicacid/carboxylate group (A side), see Scheme 2. To investigate thiswe conducted quantum chemical calculations for all isomeric formsobtained by varying the complexation site (N or A) in both lig-ands and by varying the position of the movable proton (N or Ain both ligands). We also noted that the molecular geometry of

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

nicotinic acid does not allow for simultaneous binding of both Nand A to the metal, avoiding a complicating factor in more flexi-ble molecules containing the same functional groups (amino acids,peptides etc.). For both copper and nickel we find that the lowest

complexes (doublet). Bond lengths are in A and angles in degrees.

(triplet and singlet). Relative energies are in kJ mol−1, bond lengths in A and angles

ARTICLE ING Model

MASPEC-14953; No. of Pages 10

4 H. Dossmann et al. / International Journal of M

Table 1Experimental collision cross sections (CCS) (polyAla calibration) obtained for[M+2NAH−H]+ complexes (M = Cu, Ni).

Ion Drift time (ms) CCS (A2)

[2NAH−H+Cu]•+ 2.46 94.0[2NAH−H+Cu]•+ 2.68 100.3[2NAH−H+Ni]+ 2.72 101.4

Table 2Theoretical CCS determination of the two most stable calculated forms of each[M+2NAH−H]+ complex (M = Cu, Ni).

Ion CCS (A2)

2NA−Cu−ANH•+ 102.12NA−Cu−NAH•+ 102.2

emitismplt

ft(f

3

miis

F(p

significantly more compact structure. Unfortunately, the quan-tum chemical calculations could not help us in identifying thismore compact structure. CID experiments performed subsequent

1NA−Ni−ANH+ 101.03NA−Ni−ANH+ 102.3

nergy forms correspond to tetradendate carboxylate arrange-ents NA−M−ANH+, in which the non-deprotonated ligand exists

n its zwitterionic form, i.e. the moveable proton resides on one ofhe non-metalbonding nitrogen atoms. The situations are depictedn Figs. 1 and 2, which also show that the copper(II) complex isquare planar while the nickel(II) complex is tetrahedral in theost stable triplet electronic configuration, in good agreement with

revious observations [22–24]. The latter may also exist in a low-ying singlet state (14 kJ mol−1 higher than the triplet), for whichhe square planar arrangement is preferred.

A fuller account of the relative energies of possible isomericorms is given in the Supplementary information, which showshat the forms where nitrogen is coordinated to the metal, as inAN−M−ANH+) and (AN−M−NAH+), are clearly lower in energyor copper than for nickel.

.3. Ion mobility measurements

As outlined above, the [M+2NAH−H]+ complexes (M = Cu, Ni)ay present different binding patterns. To probe whether only one

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

somer is preferentially formed or a mixture of different isomerss present in the electrospray, we conducted ion mobility mea-urements (Figure S1 and Table 1). Experimental collision cross

ig. 3. MS2 (a) SORI-CID (FT-ICR) and (b) CID (ion trap) spectra of [Cu+2NAH−H]•+

m/z 308). In spectrum (b), some ions have captured a water molecule due to theresence of water in the trap. These water adducts are indicate by an asterisk.

PRESSass Spectrometry xxx (2013) xxx– xxx

sections of the detected ions were determined from travelling waveion mobility experiments. However, in most experiments only oneisomeric species could be resolved, but a minor isomeric com-plex (6.6%) was evidenced for the CuL(L−H)•+ ion (m/z 308) witha drift time of 2.46 ms (94 A2). The main CuL(L−H)•+ copper dimer(100%) was detected with a drift time of 2.68 ms (100.3 A2) and theNiL(L−H)+ nickel dimer presented a drift time of 2.72 ms (101.4 A2)(Table 1).

The calculated CCS values for the optimized geometries weredetermined using the trajectory method (Table 2). The obtainedvalues are consistent with the corresponding experimental val-ues for both nickel and copper complexes (Table 1). However,on the basis of the very similar cross sections of the isomericstructures located in quantum chemical calculations study, onewould in hindsight not expect them to be separable in the ionmobility experiment. On the other hand, the minor isomericcopper dimer with an experimental CCS of 94 A2 must have a

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

Fig. 4. MS2 (a) SORI-CID (FT-ICR) and (b) CID (ion trap) spectra of [Ni+2NAH−H]+

(m/z 303) and MS3 (c) SORI-CID and (d) CID spectra of [Ni+2NAH−H−CO2]+ (m/z259). In spectra (b) and (d), some ions have captured a water molecule due to thepresence of water in the trap. These water adducts are indicate by an asterisk.

ING Model

M

al of M

tsps

3

umSedfchcteletd

[lTe

ARTICLEASPEC-14953; No. of Pages 10

H. Dossmann et al. / International Journ

o the ion mobility separation showed that this minor component isignificantly more fragile towards dissociation than the major com-onent (data not shown), thereby indicating lower thermochemicaltability.

.4. Dissociation of [M+2NAH−H]+ complexes (M = Cu, Ni)

Activation of [M+2NAH−H]+ complexes (M = Cu, Ni) was donesing two activation methods. The first, low-energy CID experi-ents, was performed on a quadruple ion trap and the second,

ORI-CID activation, was carried out on a FT-ICR mass spectrom-ter. When the former gives usually access to the lower energyissociation pathways [25], the latter is known to enable slowragmentation processes, which may sometimes not be entropi-ally favourable. In fact, both techniques are considered as sloweating methods [26], the main differences is the high pressureonditions of the quadrupole ion trap yielding a fast cooling ofhe ions which limits consecutive dissociation and favour the lownergy dissociation pathways [25]. For the FT-ICR, because of theow-pressure condition no collisional cooling of the product ions isxpected yielding low kinetic shift [12,27]. Hence, comparing thesewo activation spectra may bring some information concerning theissociation processes observed.

The dominating dissociation reaction observed for both

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

Cu+2NAH−H]•+ (m/z 303) and [Ni+2NAH−H]+ (m/z 308) is theoss of a single CO2 molecule, leading to [M+2NAH−H−CO2]+.his is confirmed from the FT-ICR (SORI-CID) and ion trapxperiments (Figs. 3 and 4). The decarboxylated fragment ions

Scheme 3

PRESSass Spectrometry xxx (2013) xxx– xxx 5

[M+2NAH−H−CO2]+ (M = Cu, Ni) may fragment further but nowreflecting the influence of the metal centre. While the decarboxy-lated species may lose a second molecule of CO2 for both M = Cu andNi (Figs. 3 and 4c), the dominating secondary reaction for M = Ni isloss of NiCO2 while it is loss of the ring fragment radical C5H4N•

for M = Cu. The latter corresponds to the dissociation pattern seenfor amino acid copper(II) complexes of the same kind, describedin great detail in the literature [9]. For [Cu+2NAH−H−CO2]•+ wealso note the loss of CO2 + C5H4N•. All these observations are ver-ified by accurate mass measurements. The SORI-CID spectrum of[Ni+2NAH−H−CO2]+, Fig. 4c, also shows a peak for loss of CH2O2,probably either due to the successive H2O and CO losses or loss offormic acid. This is clearly a relatively high-energy process since itis not seen in the low energy collision spectra (ion trap), in anal-ogy with previous observations by Bouchonnet et al. for complexesbetween Cu(II) and amino acids [8].

3.5. Dissociation mechanisms

Possible dissociation pathways were investigated by meansof quantum chemical calculations. For the sake of clarity, weonly present schematic representation of involved ion struc-tures (Scheme 3) and geometries of transition structures here(Figs. 6 and 8). Optimized geometries and energies for all calculated

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

structures are presented in Supplementary information.A schematic potential energy diagram based on the quantum

chemical calculations for [Cu+2NAH−H]•+ is reproduced in Fig. 5.Several pathways may lead to the same product pairs (or isomeric

.

ARTICLE IN PRESSG Model

MASPEC-14953; No. of Pages 10

6 H. Dossmann et al. / International Journal of Mass Spectrometry xxx (2013) xxx– xxx

F elativep

pits

ig. 5. Potential energy diagram for the dissociation of 2[Cu+2NAH−H]•+ complex (rathways and black lines the non-observed ones.

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

roduct pairs) but only the lowest energy route for each processs indicated to keep the diagram simple. The diagram shows thathe facile decarboxylation reaction may occur directly from thequare planar global potential energy minimum NA−Cu−ANH•+

Fig. 6. Geometry optimized transition structures involved in dissociation of 2

energies in kJ mol−1). Red lines represent the experimentally observed dissociation

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

by a C C bond activation mechanism involving partial rotationof the protonated ligand ANH, thereby liberating the carboxylatein the form of CO2 from the metal centre via TSCu1 (Fig. 6). Thisrearrangement/dissociation mechanism leads to NA−Cu−NH•+ at

[Cu+2NAH−H]•+ complex. Bond lengths are in A and angles in degrees.

ING Model

M

al of M

4d

Nberh–atbralihntclsoCfiAfslp

Aw

Fd

ARTICLEASPEC-14953; No. of Pages 10

H. Dossmann et al. / International Journ

4 kJ mol−1, in which the carbon of the remaining NH unit is bondedirectly to Cu.

Now, starting from NA−Cu−NH•+, the 3-pyridinyl radical part ofA may be lost via TSCu3 by direct dissociation of the C C(O2(Cu))ond leading to a closed shell cationic species A−Cu−NH+. How-ver, the calculations indicate that the CO2 moiety of A−Cu−NH+ isather weakly bonded (CO2 loss requires 259 – 191 = 68 kJ mol−1)aving a OCO· · ·Cu−NH+ structure rather than (�2−CO2)−Cu−NH+

the latter would have contained the anticipated binding motif of Scheme 1. For the ultimate dissociation product, Cu−ANH+,he copper atom is formally Cu(I), meaning that the homolyticond dissociation resulting from pyridinyl loss leads to electroniceduction of the metal. The alternative to successive losses of CO2nd C5H4N• would be the direct loss of nicotinic acid radical. Thisoss may proceeds without any barrier to lead to the Cu−ANH+

on and the endothermicity of the reaction is 247 kJ mol−1. Thisigher energy process is not shown in Fig. 5. Interestingly, theicotinic acid radical turns out to be 13 kJ mol−1 less stable thanhe separated CO2 and C5H4N• fragments. According to the cal-ulations the alternative to pyridinyl loss from NA−Cu−NH•+ isoss of the second carboxylate group in the form of CO2 requireslightly higher energy, in good agreement with the experimentalbservations. Out of several possible mechanisms for this secondO2 loss, the one with the over-all lowest barrier correspondsrst to a rearrangement of the NA−Cu−NH•+ ion leading to theN−Cu−NH•+ ion via a 160 kJ mol−1 barrier (TSCu2) and then to the

acile loss of the CO2 moiety from this ion (through the transitiontructure TSCu2bis). Further dissociation of the doubly decarboxy-ated species N−Cu−NH•+ by 3-pyridyl radical loss to give Cu−NH+

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

roceeds without barrier and is predicted to require 259 kJ mol−1.The direct loss of 3-pyridinyl radical from NA−Cu−ANH•+ to give

−Cu−ANH+ was calculated to be without reverse critical energy,ith products at 180 kJ mol−1. The fact that the experiment only

ig. 7. Potential energy diagram for the dissociation of the triplet [Ni+2NAH−H]+ compleissociation pathways and black lines the non-observed ones.

PRESSass Spectrometry xxx (2013) xxx– xxx 7

shows a minor peak resulting from this reaction indicates that thecalculated value may be somewhat too low. Alternatively, furtherloss of CO2 from A−Cu−ANH+ to give Cu−ANH+ in rather favourable(at 247 kJ mol−1, not shown in the diagram). Also in A−Cu−ANH+,the terminal CO2 is weakly bonded, not possessing the sought-after(�2-CO2) moiety.

In conclusion, the computational data are consistent with theobserved dissociation patterns showing kinetic rather than ther-modynamic product control. Finally, we note that TSCu5 giving riseto CuCO2 loss from NA−Cu−ANH•+ is expected to be energeticallymore favourable than forming Cu−NH+. The latter is neverthelessshowing a slightly higher signal than the former in the CID spec-trum (m/z 142 vs m/z 157, Fig. 3b) and becomes clearly higher onthe SORI-CID spectrum. This observation is again consistent withkinetic control of the reaction favouring direct bond cleavage.

Fig. 7 displays the potential energy diagram computed for thevarious dissociation pathways for [Ni+2NAH−H]+ for the tripletelectronic state. The diagram describing the energetically slightlyhigher singlet state is shown in the Supplementary information. Inanalogy to the copper case described above, loss of CO2 occurs via aC C bond activation pathway via TS3Ni1 at 156 kJ mol−1, giving thesingly decarboxylated ion at 61 kJ mol−1. Comparing Figs. 5 and 7show that the energetic requirements are rather similar for thetwo cases. The same is apparently the case for the second CO2 loss,however despite the quite similar energetic requirement, TS3Ni2 iscompletely different from TSCu2. In the former case (Fig. 7) insertionof the nickel atom into the C (CO2) bond leads to a considerableelongation of that bond (bond length = 2.06 A) leading directly tothe fully dissociated products, while in the latter case the pro-

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

cess occurs in two steps: rearrangement to the N-coordinatedNA−Cu−NH+ to AN−Cu−NH+ followed by CO2 loss. Interestingly,N−Cu−NH+ is 99 kJ mol−1 more stable than N−Ni−NH+ in relativeterms, reflecting the stronger binding of nitrogen ligands to Cu(II)

x (relative energies in kJ mol−1). Red lines represent the experimentally observed

ARTICLE IN PRESSG Model

MASPEC-14953; No. of Pages 10

8 H. Dossmann et al. / International Journal of Mass Spectrometry xxx (2013) xxx– xxx

triple

cp

drmCpmamcanestf

Tms

Fig. 8. Geometry optimized transition structures involved in dissociation of

ompared to Ni(II), as also seen in the calculations of the dimericarent complexes described above (Supplementary information).

The experimentally observed loss of NiCO2 from the singlyecarboxylated ion NA−Ni−NH+ can be described as a two-stepseaction, first requiring transfer of the 3-pyridinyl moiety to theetal via TS3Ni4 leading to the ion–dipole complex [N−Ni−NH,

O2]+ followed by NiCO2 loss via TS3Ni5 to the fully dissociatedroducts at 132 kJ mol−1. (Also displayed in Fig. 7 is the path to for-ation of 1N NH+ via the more energy demanding loss of nickel

tom (TS3Ni3) starting from 3N Ni NH+ resulting as well in a for-al loss of NiCO2 from singly decarboxylated ion NA Ni NH+.) The

alculations predict that the NiCO2 entity is not covalently bondeds in motif a (Scheme 1), but is essentially a complex between aickel atom and CO2. It is, however, noteworthy that in the singletlectronic state the energy minimum of NiCO2 is predicted to be aide-on adduct in which the Ni atom binds to both C and one of theerminal O atoms. Interestingly, this dissociation is leading to theormation of a new C C bond, namely the 1N NH+ ion.

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

The fact that TS3Ni5 is computed to be 40 kJ mol−1 higher thanS3Ni2 could reflect the level of accuracy of the computationalodel since the abundances of the CID signals for the succes-

ive losses of two CO2 molecules compared to CO2 loss followed

t [Ni+2NAH−H]+ complex. Bond lengths are given in A and angles in degree.

by NiCO2 loss are similar. This observation could be also consis-tent with the fact that dissociation of this complex is driven bythermodynamic factors instead of kinetic ones, favouring thus for-mation of the most stable products. This would point out the largedifference between copper and nickel complexes by means of disso-ciation behaviour although from a first sight both complexes seemto roughly behave similarly.

Again starting from the intermediate NA Ni NH+, we per-formed calculations for the unobserved loss of the 3-pyridinylradical for comparison with the copper system, for which it isobserved. Perhaps not unexpected, this homolytic dissociationfrom the even electron nickel species display a transition stategeometry TS3Ni6 at 294 kJ mol−1 in contrast to the open shell coppercongener for which this only requires 192 kJ mol−1. In line with thisway of reasoning, we also note that the direct homolytic C C cleav-age from the parent NA Ni ANH+ complex is predicted to requiremore than twice the amount of energy as the corresponding pro-cess in NA Cu ANH•+. Finally, as already mentioned, the relatively

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

high-energy loss of CH2O2 would either be due to the successiveH2O and CO losses or loss of formic acid. The latter now seemsmore probable, based on result of our computational investigationsas discussed in more detail in the Supplementary information.

ING Model

M

al of M

4

sapasdddbtemtbcitrbscsettist

ctpuchTc

umctrCitri

A

ttto(spDmft

[

[

[

[

[

[

[

[

[

[

[

[

ARTICLEASPEC-14953; No. of Pages 10

H. Dossmann et al. / International Journ

. Discussion and conclusions

Previous mass spectrometric investigations [6–16] have demon-trated that ML(L−H)+ complex ions formed between amino acidsnd Cu(II) and Ni(II) typically give CO2 loss as the primary CIDathway. Following initial CO2 loss, radical fragment loss waslso known to be abundant but only for the odd electron copperpecies. The same applies to species where L = nicotinic acid, asemonstrated here, for which 3-pyridinyl loss is abundant from theecarboxylated complex NA Cu NH•+. We also see evidence forirect loss of 3-pyridinyl from the parent complex NA Cu ANH•+

ut this is a minor reaction. In both cases it was of great interesto investigate the product ion structures, since they were consid-red to have a potential for containing the key M(�2 O2C) bondingotif. However, on the basis of our observations and in particular

he quantum chemical calculations this type of structures seem toe of little relevance for cationic transition metals of this type, andarbon dioxide seems not to form particular strong bonds to coppern these cases. Instead of the oxidized state required for the simul-aneous complexation of the two oxygen sites, copper prefers theeduced even electron Cu(I) state. Interestingly, the isoelectronicare nickel atom behaves similarly, and seems not to give rise to atable Ni(�2 O2C) geometry either. In this manner the electroni-ally more flexible transition metal at least for the cationic speciestudied here, behave very different from anionic alkali and alkaliarth metals, which keep their oxidized form, thereby being ableo accommodate bent CO2 also when the C C bond to the bondo the central carbon is broken [1–4]. In the future, it will be verynteresting to study anionic forms of the same transition metalstudied here and also zinc, to see how adding electrons will alterhe structural and energetic landscapes.

One noteworthy reaction has nevertheless been observed and isharacteristic for the Ni complex: the C C bond formation leadingo the 1N NH+ ion. Many catalytic cycles involve nickel(II) com-lexes to form new C C bonds. The last step of these cycles issually an elimination of the metal leading to the new bond, whichorresponds to the reaction observed (elimination of NiCO2). Weave investigated here the gas-phase mechanism of this reaction.his could be the first step of a work dedicated to probe a catalyticycle involving Ni(II) and nicotinic acid.

We would also like to add a final note on the prospect of CO2ptake. Although M(�2 O2C) species may be of relevance for theechanism for CO2 sequestration in various applications, it is of

ourse possible to envisage reaction mechanisms not requiring thisype of intermediates. In the present case, we may imagine theeverse of the observed decarboxylation reactions as prototypicalO2 insertion reactions catalyzed by a transition metal. However,

n the absence of any stabilizing interactions, it seems clear that theransition states found would pose too high enthalpic and entropicequirements since the calculated energy barriers seem rather highn all cases encountered here.

cknowledgments

This work was supported by the Norwegian Research Council byhe Grant No. 179568/V30 to the Centre of Theoretical and Compu-ational Chemistry through their Centre of Excellence program andhe Norwegian Supercomputing Program (NOTUR) through a grantf computer time (Grant No. NN4654K). The TGE High field FT-ICRCNRS) is gratefully acknowledged for the access to the FT-ICR masspectrometer as well as Dr. E. Derat (IPCM) for providing us com-utational resources. C.A. acknowledges the European Regional

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

evelopment Fund (ERDF) No. 31708 and the région Haute Nor-andie for financial support. Finally, UPMC is also acknowledged

or the financial support to permit one of the authors (E.U.) to comeo Paris as a visiting professor.

[

[

PRESSass Spectrometry xxx (2013) xxx– xxx 9

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ijms.2013.05.030.

References

[1] A.B. Attygalle, F.U. Axe, C.S. Weisbecker, Mild route to generate gaseous metalanions, Rapid Commun. Mass Spectrom. 25 (2011) 681–688.

[2] S. Curtis, J. DiMuzio, A. Mungham, J. Roy, D. Hassan, J. Renaud, P.M. Mayer, Reac-tions of atomic metal anions in the gas phase: competition between electrontransfer, proton abstraction and bond activation, J. Phys. Chem. A 115 (2011)14006–14012.

[3] S. Curtis, J. Renaud, J. Holmes, P. Mayer, Old acid, new chemistry. Negativemetal anions generated from alkali metal oxalates and others, J. Am. Soc. Mass.Spectrom. 21 (2010) 1944–1946.

[4] H. Dossmann, C. Afonso, D. Lesage, J.-C. Tabet, E. Uggerud, Formationand characterization of gaseous adducts of carbon dioxide to magnesium,(CO2)MgX-(X = OH, Cl, Br), Angew. Chem. Int. Ed. 51 (2012) 6938–6941.

[5] A. Correa, R. Martín, Metal-catalyzed carboxylation of organometallic reagentswith carbon dioxide, Angew. Chem. Int. Ed. 48 (2009) 6201–6204.

[6] S.R. Wilson, A. Yasmin, Y. Wu, Bipyridyl amino acid-metal complexes and theircharacterization by electrospray mass spectrometry, J. Org. Chem. 57 (1992)6941–6945.

[7] C.L. Gatlin, F. Turecek, T. Vaisar, Gas-phase complexes of amino acids withCu(II) and diimine ligands. Part I. Aliphatic and aromatic amino acids, J. MassSpectrom. 30 (1995) 1605–1616.

[8] S. Bouchonnet, Y. Hoppilliard, G. Ohanessian, Formation and fragmentationsof organometallic complexes involving aliphatic �-amino acids and transi-tion metal cations – a plasma desorption mass spectrometry study, J. MassSpectrom. 30 (1995) 172–179.

[9] F. Turecek, Copper-biomolecule complexes in the gas phase. The ternary way,Mass Spectrom. Rev. 26 (2007) 563–582.

10] C. Afonso, A. Riu, Y. Xu, F. Fournier, J.-C. Tabet, Structural characterization offatty acids cationized with copper by electrospray ionization mass spectrom-etry under low-energy collision-induced dissociation, J. Mass Spectrom. 40(2005) 342–349.

11] P. Wang, G. Ohanessian, C. Wesdemiotis, Cu(II)-catalyzed reactions in ternary[Cu(AA)(AA−H)]+ complexes (AA = Gly, Ala, Val, Leu, Ile, t-Leu, Phe), Eur. J. MassSpectrom. 15 (2009) 325–335.

12] C. Afonso, D. Lesage, F. Fournier, V. Mancel, J.-C. Tabet, Origin of enantioselectivereduction of quaternary copper d,l amino acid complexes under vibrationalactivation conditions, Int. J. Mass Spectrom. 312 (2012) 185–194.

13] C.-K. Siu, Y. Ke, Y. Guo, A.C. Hopkinson, K.W. Michael Siu, Dissociations ofcopper(II)-containing complexes of aromatic amino acids: radical cations oftryptophan, tyrosine, and phenylalanine, Phys. Chem. Chem. Phys. 10 (2008)5908–5918.

14] D. Zhang, W.A. Tao, R.G. Cooks, Chiral resolution of d- and l-amino acids bytandem mass spectrometry of Ni(II)-bound trimeric complexes, Int. J. MassSpectrom. 204 (2001) 159–169.

15] R. Jirásko, M. Holcapek, L. Kolárová, M. Nádvorník, A. Popkov, Characterizationof Ni(II) complexes of Schiff bases of amino acids and (S)-N-(2-benzoylphenyl)-1-benzylpyrrolidine-2-carboxamide using ion trap and QqTOF electrosprayionization tandem mass spectrometry, J. Mass Spectrom. 43 (2008) 1274–1284.

16] A.R.M. Hyyryläinen, J.M.H. Pakarinen, F. Fülöp, P. Vainiotalo, Diastereochem-ical differentiation of some cyclic and bicyclic �-amino acids, via the kineticmethod, J. Am. Soc. Mass. Spectrom. 20 (2009) 34–41.

17] K. Giles, J.P. Williams, I. Campuzano, Enhancements in travelling wave ionmobility resolution, Rapid Commun. Mass Spectrom. 25 (2011) 1559–1566.

18] M.F. Mesleh, J.M. Hunter, A.A. Shvartsburg, G.C. Schatz, M.F. Jarrold, Structuralinformation from ion mobility measurements: effects of the long-range poten-tial, J. Phys. Chem. 100 (1996) 16082–16086.

19] A.A. Shvartsburg, M.F. Jarrold, An exact hard-spheres scattering model for themobilities of polyatomic ions, Chem. Phys. Lett. (1996) 86–91.

20] S.C. Henderson, J. Li, A.E. Counterman, D.E. Clemmer, Intrinsic size parametersfor Val. Ile. Leu. Gln. Thr. Phe. and Trp residues from ion mobility measurementsof polyamino acid ions, J. Phys. Chem. B 103 (1999) 8780–8785.

21] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato,X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada,M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O.Kitao, H. Nakai, T. Vreven, J.A. Montgomery, J.E. Peralta, F. Ogliaro, M. Bearpark,J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Normand, K.Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega,J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo,R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W.Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, J.J.Dannenberg, S. Dapprich, A.D. Daniels, J.B. Farkas, J.V. Foresman, J.C. Ortiz, D.J.Fox, Gaussian 09, Revision B.01, Gaussian, Wallingford, CT, 2009.

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

22] M.Y. Combariza, R.W. Vachet, Effect of coordination geometry on the gas-phasereactivity of four-coordinate divalent metal ion complexes, J. Phys. Chem. A 108(2004) 1757–1763.

23] J.R. Hartman, R.W. Vachet, W. Pearson, R.J. Wheat, J.H. Callahan, A comparison ofthe gas, solution, and solid state coordination environments for the copper(II)

ING Model

M

1 al of M

[

[

[trometry, J. Mass Spectrom. 32 (1997) 461–474.

ARTICLEASPEC-14953; No. of Pages 10

0 H. Dossmann et al. / International Journ

complexes of a series of aminopyridine ligands of varying coordination number,Inorg. Chim. Acta 343 (2003) 119–132.

Please cite this article in press as: H. Dossmann, et al., Int. J. Mass Spectrom

24] J.R. Hartman, R.W. Vachet, J.H. Callahan, Gas, solution, and solid state coordi-nation environments for the nickel(II) complexes of a series of aminopyridineligands of varying coordination number, Inorg. Chim. Acta 297 (2000) 79–87.

25] D.M. Black, A.H. Payne, G.L. Glish, Determination of cooling rates in aquadrupole ion trap, J. Am. Soc. Mass Spectrom. 17 (2006) 932–938.

[

PRESSass Spectrometry xxx (2013) xxx– xxx

26] S.A. McLuckey, D.E. Goeringer, Slow heating methods in tandem mass spec-

. (2013), http://dx.doi.org/10.1016/j.ijms.2013.05.030

27] J. Laskin, M. Byrd, J. Futrell, Internal energy distributions resultingfrom sustained off-resonance excitation in FTMS. I. Fragmentation ofthe bromobenzene radical cation, Int. J. Mass Spectrom. 195/196 (2000)285–302.