at SciVerse ScienceDirect

Vacuum 89 (2013) 47e52

Contents lists available

Vacuum

journal homepage: www.elsevier .com/locate/vacuum

MALDI TOF and theoretical investigation of silver clusters obtained by gammairradiation

Jelena Cveti�canin a, Vasil Koteski a, Jelena Belo�sevi�c-�Cavor a, Aleksandra Radosavljevi�c a,Zorica Ka�carevi�c-Popovi�c a, Djordje Trpkov a, Danka A�cimovi�c a, Zorana Rogi�c Miladinovi�c a,Mario Cindri�c b, Olivera Ne�skovi�c a,*

aVin�ca Institute of Nuclear Sciences, P.O. Box 522, University of Belgrade, Mike Petrovi�ca Alasa 12-14, 11001 Belgrade, SerbiabRudjer Bo�skovi�c Institute, Bijeni�cka cesta 54, HR-10000 Zagreb, Croatia

a r t i c l e i n f o

Article history:Received 6 June 2012Received in revised form29 August 2012Accepted 30 August 2012

Keywords:Silver clustersMass spectrometryTheoretical investigation

* Corresponding author. Tel.: þ381 11 3408752; faxE-mail address: oliveran@vinca.rs (O. Ne�skovi�c).

0042-207X/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.vacuum.2012.08.016

a b s t r a c t

The aim of this study is to evaluate the size and stability of radiolytically produced silver clusters. Theclusters obtained by the gamma irradiation of a mixture of a silver salt and polyvinyl alcohol (PVA) werestudied both theoretically and by mass spectrometry. MALDI TOF mass spectrometry of the Agn clustersshowed the presence of clusters containing between 5 and 29 atoms. Both the theoretical and experi-mental results of this paper show that the most stable clusters are small clusters with an odd number ofatoms. The stability of these clusters is correlated with the cluster’s electronic configuration.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Clusters can be classified as a new type of material becauseclusters have properties that are fundamentally different from boththose of discrete molecules and the relevant bulk solid. The term“cluster” is defined as an aggregate of a countable number ofparticles, such as atoms or molecules. The study of small metalclusters has contributed extensively to the understanding of size-dependent and many-body character of nanoscale physics andchemistry. Important examples of size-dependent characteristicshave been observed in the measurement and theoretical calcula-tion of metal cluster melting [1,2], the transition of a cluster’sstructure from planar to three-dimensional [3,4], and the reactivityof gold-cluster nanocatalysts [5,6]. Resonant absorption frequen-cies have recently been measured for gas phase Agnþ clusters usingphotodepletion [7] and for Agn clusters in rare-gas matrices [8].Both studies showmaxima in the resonance absorption frequenciesthat are associated with closed-shell clusters.

Spectral analysis of both positive and negative ion silver clustersindicates that the intensity of peaks associated with clusters con-taining an odd number of atoms is larger than the intensity of peaksassociated with clusters containing an even number of atoms. This

: þ381 11 8066434.

All rights reserved.

phenomenon may be explained by simple electron counting.Clusters containing an odd number of atoms have an even numberof valence electrons, which results in spin pairing. Therefore, theseclusters have enhanced stability compared to those clusters with aneven number of atoms [9e12]. The mass spectra of silver clustersalso contain peaks for specific ‘‘magic numbers’’ [13], where a steepdecrease in ion intensity is observed after the corresponding peak.An explanation of this observation is based on the Jellium model[9e14] in which the stability of the cluster is determined by theenergy levels of the valence electrons bound in the cluster. At themagic numbers, the number of atoms in the cluster corresponds tothe number of electrons necessary for shell closing to occur in theelectron shell model when charge is taken into consideration [9e12]. For example, Ag3þ and Ag9þ, which respectively have 2 and 8valence electrons, appear as magic numbers. These clusters corre-spond to the closing of the first and second shell in the electronshell model.

Among the most fundamental properties of a cluster are theenergies required to ionize the neutral cluster either by the addi-tion or loss of an electron. As observed for silver clusters in theliterature [15], the addition of an electron is defined as the electronaffinity (EA), and the loss of an electron is defined as the ionizationenergy (IE). The oddeeven pattern discussed above for the ionintensities in the mass spectra of silver clusters also manifests itselffor the EAs and IEs for Agn clusters. Agn clusters with odd numbersof atoms are easier to ionize through both the addition of an

J. Cveti�canin et al. / Vacuum 89 (2013) 47e5248

electron or the removal of an electron, as these clusters have bothrelatively high electron affinities and relatively low ionizationenergies. It should be stressed that the matrix used must havea lower IE or higher EA than the cluster. Some matrices that may besuccessfully used for detecting silver clusters by the MALDI TOFtechnique have the required properties.

In this study, we investigate the structure and the stability in thegas phase of Agn silver clusters formed by gamma irradiation usingMALDI TOF(matrix assisted laser desorption/ionization, time-of-flight), MALDI TOFTOF and theoretical ab initio calculations. Thestability of Ag11 clusters was studied with the aid of the MALDITOFTOF, a tandem mass spectrometry method. In MALDI TOFTOF,the collision induced dissociation (CID) technique is coupled withTOF, which enables MSMS(tandem mass spectrometry) analysis. Inthe mass spectra for clusters of a characteristic size, the Ag isotopicabundance in nature is observed [16].

2. Material and methods

The formation of silver clusters using two different syntheticprocedures were investigated previously [17]. Twomixtures of 5 mlof distilled water, 1 ml of PVA (0.05 g/ml) and 144 mg of AgNO3were prepared using ultrasound. These mixtures were aerated for30 min under Ar or N2O in order to remove oxygen and wereexposed to gamma-irradiation (60Co radiation facility) underambient conditions. The irradiationwas performed at a dose rate of10 kGy/h until an absorbed dose of 55 kGywas reached. This dose isrequired for complete reduction of the Agþ ions in the solution.Oxidized species were not detected in the analyses of the massspectra. The intensity of the investigated cluster’s peaks was largerfor the sample prepared under N2O when using the same laserintensity.

2.1. Mass spectrometry analysis

Positive and negative ion MALDI TOF mass spectra wereacquired using a MALDI TOF (AB Applied Biosystems, Voyager-DEPRO, Framingham, USA). MALDI TOFTOF analyses were performedusing a 4700 Proteomics Analyzer (Applied Biosystems).

2.2. Preparation of MALDI matrix

A saturated solution of the matrix a-cyano-4-hydroxycinnamicacid (CHCA) was prepared by dissolving the compound in 50%acetonitrile with the addition of 0.1% trifluoroacetic acid. Thesolution was vortexed thoroughly and was sonicated in a waterbath for several minutes at room temperature. The 2,5-dihydroxybenzoic acid (DHB) matrix was prepared by dissolvingthe matrix in 50% acetonitrile with the addition of 0.3% trifluor-acetic acid [18]. The CHCA and DHB solutions were used for thepreparation of samples for MALDI TOF analysis. All employedchemicals were obtained from SigmaeAldrich (St. Louis, WI, USA).

MALDI TOFmass spectral analysis of the samples was performedin a linear and reflectron mode with delayed extraction. Massspectra were calibrated using the standard calibration mix2 in themass range of 170e4000 Da. MALDI TOFTOF experiments werecarried out using N2 gas for the study of fragmentation throughmetastable decomposition. The selection of the precursor ion alsoinvolved the selection of precursor fragments resulting from post-source decay between the ion source and the time-ion-selector.Typically, the precursor ion or (M þ Na)þ was selected ina window (�5 Da) that was centered on the first isotope. It wasverified that only the chosen precursor was selected in thiswindow. To achieve the best statistics, averages of mass spectra forover 100 different spots on the target, with each spot including 100

laser shots, were calculated. The laser fluence was adjusted toproduce an unsaturated signal for the selected ion precursor. Analiquot of each sample solution containing the internal standardwas combined in a 1:1 ratio with either the CHCA or DHB matrixand was mixed thoroughly. Aliquots (0.5 mL) of the mixtures werespotted on a 100 spots sample plate and were air-dried. Ions wereobtained by laser desorption using a 200 Hz frequency N2 pulselaser operating at 337 nm.

Samples were mixed in ratios of 1:10 v/v ratio. A concentrationof 10 mg/ml of CHCA was obtained by diluting the CHCA in a 1:1acetonitrile:water solution. After dilution, 0.5 mL aliquots of thesample were placed on a MALDI plate. Samples were subsequentlyair-dried and analyzed. Data Explorer version 4.9 was used for theanalysis of the recorded spectra. Ions of the CHCAmatrix were usedfor internal calibration, on the basis of the theoretically calculatedmasses of CHCAmonomers, dimers and trimers, respectively, atm/zvalues of 190.05 (molecular formula C10H7NO3), 379.09 (molecularformula C20H14N2O6) and 568.14 (molecular formula C30H21N3O9).The mass spectrum of the matrix was recorded in order to elimi-nate the signals generated by the matrix itself. For negative ions,a DHB matrix, which has the characteristic mass of 154.12 [19], wasused. The applied laser intensity was between 10 and 30% of themaximum available laser power.

CHCA matrix was used for positive and negative ions, DHBmatrix was used only for negative ions.

2.3. MALDI TOFTOF MSMS measurements

Each sample was measured by using the MSMS 1 kV positivemode in the 4000 Series Explorer Version 3.0 software. Theprecursor mass was the molecular mass of the selected clusters,and the mass window was �1.00 Da and þ5.00 Da of theprecursor mass. Molecular ions were extracted from the ionsource after a 460 ns delay. The suppressor for the metastable ionwas turned off. The accelerating voltage at the ion sourcewas þ8.0 kV, and the grid voltage was þ6.85 kV. The collision cellwas at þ7.0 kV. Fragment ions were extracted from the collisioncell after a 27.3 ms delay. The accelerating voltage from thecollision cell was þ15.0 kV. The sampling bin size was 0.5 ns withan input bandwidth of 500 MHz and a vertical full scale of200 mV. The reflectron detector voltage was þ2.1 kV. The pres-sure inside the collision cell was adjusted with atmospheric gas tothe level of 10�6 Torr before the initiation of the measurement.Each spectrum was automatically acquired by accumulating theresults of 5000 shots with random edge-biased positioning ofeach laser shot. By using the Data Explorer Version 4.6 software,the resulting mass spectrum was internally calibrated with themost abundant isotopic peaks of the precursor ion and a knownfragment ion. A Mass Spec Calculator Professional version 4.09demo program was used to assist in the identification of thefragment ions.

3. Results and discussion

The formation mechanism of silver clusters was explained inand our previous paper [17]. Silver clusters were produced fromAgClO4 by gamma irradiation [20]. Clusters were stabilized bysodium polyphosphate and analyzed by optical spectroscopy.However, our research was mass spectrometric characterization ofobtained clusters.

Samples produced by the gamma-irradiation of a mixture ofPVA and AgNO3 were analyzed by mass spectrometry. The resultsindicated that positive Ag cluster ions were formed in the presenceof a CHCA matrix in both working modes and that the resolutionwas greater when the measurement was performed in a reflectron

Fig. 1. MALDI TOF spectrum of cationic silver clusters is shown. The numbers above the mass values (5e29) indicate the number of silver atoms contained in a specific cluster. CHCAmatrix was used.

J. Cveti�canin et al. / Vacuum 89 (2013) 47e52 49

mode (Fig.1). Themost intense peaks in the spectrum are 538.5902,754.4372 and 970.2236 Da, which respectively correspond toclusters containing 5, 7 and 9 silver atoms. Both matrices were usedwhen working with negative ions, and the peak intensities werestronger when using the DHB matrix. For both matrices, thenumber of peaks in the negative ion spectrum is lower than in thepositive ion spectrum, and the peaks were more intense when theDHB matrix was utilized. In the negative ion spectrum, the clustersAg2 and Ag3 were observed (Fig. 2). MALDI TOF analysis of AgNO3and AgNO3/PVA using the CHCA matrix showed the presence ofpositive Ag ions but not the formation of silver clusters, as observedin Fig. 1. This confirms that the clusters are formed by gammairradiation and not by the laser desorption process. However, in

Fig. 2. MALDI TOF spectrum of silver cluster

literature, clusters formation was attributed to laser desorptionprocess and the usage DHB matrix [21,22].

The spectrum of the positive ions produced from the AgNO3/CHCA system is shown in Fig. 1S of the Supporting Information. Thespectrum of the Ag11eNa complex is shown in Fig. 2S of theSupporting Information.

4. Theory/calculation

4.1. Numerical investigations

Ab initio calculations were performed by utilizing the SIESTAcode [23] to solve the standard KohneSham equations [24] using

anions is shown. DHB matrix was used.

Fig. 4. Average bond lengths in Ǻ of the lowest-energy isomers of neutral, anionic andcationic silver clusters.

J. Cveti�canin et al. / Vacuum 89 (2013) 47e5250

a basis set of numerical atomic orbitals. We utilized standard norm-conserving scalar relativistic pseudopotentials [25] for the 11-electron valence configuration. For the exchange-correlationenergy functional, we adopted the generalized gradient approxi-mation (GGA) implemented by Perdew, Burke and Ernzerhof (PBE)[26]. In the SIESTA calculations, we utilized a split-valence double-zbasis set and an energy cut-off of 120 Ry.

The obtained equilibrium geometries of the lowest-energyisomers Agn with 3 � n � 20 are represented in Fig. 3. Fig. 4shows the relationship between the average bond lengths of thelowest-energy isomers and the number of atoms. The calculatedequilibrium geometries and the average bond lengths are in goodoverall agreement with recent calculations for Agn clusters with upto 14 atoms [27,28] and Ag20 cluster [27]. From our calculations wehave also predicted the equilibrium geometries of Agn clusters inthe range 15� n< 20, which have not been previously investigatedusing first principles calculations. The transition from 2D to 3Dgeometry for the silver clusters occurs when n � 6.

In Fig. 5, the relationship between the calculated binding ener-gies per atom of neutral and charged silver clusters Agmn (m¼�1, 0,1; n � 20) and the number of atoms in the cluster is shown. Thebinding energies were calculated according to formula (1) given ina prior study [27], in which the binding energies, ionizationpotentials (IP) and electron affinities (EA) for Agn clusters (n � 13)were reported. Herewe calculated the corresponding properties forAgn clusters with up to 20 atoms. In addition to the odd/eveneffects, we also observe an extra stability of clusters with n¼ 6þm.This stability can be observed from the peaks in the second

Fig. 3. Equilibrium geometries of the lowest-ener

difference of the total cluster energy (Fig. 6). These peaks arerelated to the peaks observed in the mass abundance spectra. Fig. 6indicates which neutral, anionic and cationic clusters are particu-larly stable. The stable neutral clusters were Ag8 and Ag14. Thestable anionic clusters were Ag3, Ag9 and Ag17, while the stablecationic clusters were Ag3 and Ag7. These calculated results are in

gy isomers of silver clusters with 3 � n � 20.

Fig. 5. The relationship between the calculated binding energies per atom of neutraland charged silver clusters and the number of atoms in the cluster is shown.

Fig. 6. The relationship between the second total energy differences and the numberof atoms for neutral, anionic and cationic silver clusters is shown.

Fig. 7. The calculated ionization energies (IE) and electron affinities (EA) of varioussmall silver clusters Agn (n � 20) are depicted.

Fig. 8. HOMO e LUMO gaps of the lowest-energy isomer of neutral, anionic andcationic Ag clusters are shown.

J. Cveti�canin et al. / Vacuum 89 (2013) 47e52 51

partial agreement with our experimental data, where we observedthat the most stable cationic silver clusters are those with 7 and 9atoms and the most stable anionic clusters are those with 3 atoms.

Ionization energies (IE) and electron affinities (EA) for varioussilver clusters are calculated as the energy difference between therelaxed neutral and ionic clusters and are reported in Fig. 7. Inaddition to the expected odd/even oscillations, we observe anenhanced EA peak for a cluster size n ¼ 17, which is in agreementwith the results presented in Fig. 6.

The HOMO e LUMO (HOMO e highest occupied molecularorbital, LUMO e lowest unoccupied molecular orbital) gaps of thelowest-energy isomer of neutral, anionic and cationic Ag clustersare depicted in Fig. 8. Generally, the gaps are notably sensitive tothe correlated effects of the cluster geometry and electronic densityof levels. As shown in Fig. 8, we obtained a peak for the neutralhexamer (Ag6), and themost relevant peaks for cationic and anionicsilver clusters are respectively observed at n ¼ 3, 9 and n ¼ 3, 7. Thecomputational and experimental results support the correlationbetween large HOMO e LUMO gaps and increased stability of thecorresponding cluster.

5. Conclusions

In this report, we describe an investigation of the structure andthe stability in the gas phase of Agn silver clusters formed bya novel, reproducible and simple gamma irradiation method.Clusters containing between 5 and 29 atoms were observed bymass spectrometry. The literature predicted the correlating magicnumbers, which are related to the electronic configuration of thecluster. We found that the most stable clusters are those withclosed electron shells. Analysis showed that clusters that corre-spond to the most intense peaks in the spectra were small clusterswith an odd number of atoms. This analysis has also beenconfirmed theoretically by ab initio calculations.

The results obtained from MSMS analysis of the different clus-ters clearly demonstrate that MALDI TOFTOF can be used todifferentiate unstable silver clusters with good reproducibility.

Acknowledgments

This work was financially supported by the Ministry of Educa-tion and Science of the Republic of Serbia under Project Nos.III45005 and III45018 and International Atomic Energy Agency,Vienna, project CRP: F23028 contract No. 15384.

J. Cveti�canin et al. / Vacuum 89 (2013) 47e5252

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.vacuum.2012.08.016.

References

[1] Haberland H, Hippler T, Donges J, Kostko O, Schmidt M, von Issendorff B. PhysRev Lett 2005;94:035701.

[2] Cleveland CL, Leudtke WD, Landman U. Phys Rev Lett 1998;81:2036.[3] Furche F, Ahlrichs R, Weis P, Jacob C, Gilb S, Bierweiler T, et al. J Chem Phys

2002;117:6982.[4] Häkkinen H, Moseler M, Landman U. Phys Rev Lett 2002;89:033401.[5] Heiz U, Sanchez A, Abbet S, Schneider W-D. J Am Chem Soc 1999;121:3214.[6] Sanchez A, Abbet S, Heiz U, Schneider W-D, Häkkinen H, Barnett RN, et al.

J Phys Chem A 1999;103:9573.[7] Tiggesbaumker J, Koller L, Lutz HO, Meiwes-Broer KH. Chem Phys Lett 1992;

190:42.[8] Harbich W, Fredigo S, Buttet J. Chem Phys Lett 1992;195:613.[9] Katakuse I, Ichihara T, Fujita Y, Matsuo T, Sakurai T, Matsuda H. Int J Mass

Spectrom Ion Proc 1985;67:229.[10] Katakuse I, Ichihara T, Fujita Y, Matsuo T, Sakurai T, Matsuda H. Int J Mass

Spectrom Ion Proc 1986;74:33.

[11] Selinger A, Schnabel P, Wiese W, Irion MP. Ber Bunsen-Ges 1990;94:1278.[12] Feldheim DL, Foss Jr CA, editors. Metal nanoparticles: synthesis, charateriza-

tion and applications. New York: Marcel Dekker Inc; 2002.[13] Huda MN, Ray AK. Phys Rev A 2003;67:013201.[14] Knight WD, Clemenger K, de Heer WA, Saunders WA, Chou MY, Cohen ML.

Phys Rev Lett 1984;52:2141.[15] Lias SG, Bartmess JE, Liebman JF, Holmes JL, Levin RD, Mallard WG, et al. Chem

Ref Data 1988;17(Suppl. 1).[16] Holden NE. In: Lide DR, editor. CRC handbook of chemistry and physics. 86th

ed. Boca Raton, FL: CRC Press; 2005.[17] Cveticanin J, Krkljes A, Kacarevic-Popovic Z, Mitric M, Rakocevic Z, Trpkov Dj,

et al. Appl Surf Sci 2010;256:7048.[18] Karas M, Hillenkamp F. Anal Chem 1988;60:2299.[19] Cohen LH, Gusev AI. Anal Bioanal Chem 2002;373:571.[20] Ershov BG, Abkhalimov EA, Sukhov NL. High Energ Chem 2005;39:55.[21] Sung-Seen C, Sung-Ho H. Bull Korean Chem Soc 2007;28:2508.[22] Szymanska-Chargot M, Gruszecka A, Smolira A, Bederski K. Acta Phys Pol A

2011;120:1012.[23] Soler JM, Artacho E, Gale JD, Garcia A, Junquera J, Ordejon P, et al. J Phys

Condens Matter 2002;14:2745.[24] Kohn W, Sham LJ. Phys Rev 1965;145:561.[25] Troullier N, Martins JL. Phys Rev B 1991;43:1993.[26] Perdew JP, Burke K, Ernzerhof M. Phys Rev Lett 1996;77:3865.[27] Fernandez EM, Soler JM, Garzon IL, Balbas LC. Phys Rev B 2004;70:165403.[28] Zhang H, Tian D. Comp Mater Sci 2008;42:62.