Laser Desorption and Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry of 29-Kda Au:SR Cluster Compounds

Anal. Chem. 2004, 76, 6187-6196

Laser Desorption and Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry of 29-kDa Au:SR Cluster Compounds

T. Gregory Schaaff*

Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6131

Positive and negative ions generated by laser-based for electron microscopy2,3 or bright fluorescent microscopy4-6 ionization methods from three gold:thiolate cluster com- probes. Since the optical and electronic properties of cluster and pounds are mass analyzed by time-of-flight mass spec- nanocrystal compounds are inexorably linked to the size of the trometry. The three compounds have similar inorganic inorganic core, many future applications rely on the ability to core masses (∼29 kDa, ∼145 Au atoms) but different synthesis and isolate compounds with either a narrow core size n-alkanethiolate ligands associated with each cluster distribution or those that are molecularly pure (i.e., one single compound (Au:SR, R ) butane, hexane, dodecane). structure). Giant (nanometer-scale) cluster compounds have been 7 Irradiation of neat films (laser desorption/ionization) and isolated for selected systems recently (e.g., Pd145 metallic clusters 8 films generated by dilution of the cluster compounds in and semiconductor Ag2S cluster compounds ). However, due to an organic acid matrix (matrix-assisted laser desorption/ their inherent compositional and structural complexity, these types ionization) with a nitrogen laser (337 nm) produced of compounds represent a significant challenge for routine distinct ion abundances that are relevant to different analytical chemical techniques, even those developed for other structural aspects of the cluster compound. Laser de- macromolecular systems. sorption/ionization of neat Au:SR compound films pro- Because chemical properties are derived from the organic duces ions consistent with the inorganic core mass (i.e., ligands, one of the unique aspects of the Au:SR compounds devoid of original hydrocarbon content). Matrix-assisted is the ability to isolate and accumulate cluster compounds with distinct inorganic core sizes through fractionation,9-11 laser desorption/ionization produces either ions with m/z chromatography,11-13 supercritical extraction,14 and electrophoretic values consistent with the core mass of the cluster methods.15 As a result of such separations, it was possible to map compounds or ions with m/z values consistent with the the evolution of optical properties from bulklike (broad plasmon approximate molecular weight of the cluster compounds, resonance excitation)10 to molecular-like (discrete electronic depending on ionization conditions. The ion abundances, transitions)11,15 metallic electronic structure. In addition, other and ionization conditions under which they are detected, interesting properties have been discovered for these cluster provide insight into desorption/ionization processes for these unique cluster compounds as well as other analytes (2) Hainfeld, J. F.; Furuya, F. R. J. Histochem. Cytochem. 1992, 40, 177-184. typically studied by matrix-assisted laser desorption/ (3) Powell, R. D.; Halsey, C. M. R.; Hainfeld, J. F. Microsc. Res. Technol. 1998, 42,2-12. ionization. (4) Dubertret, B.; Skourides, P.; Norris, D. J.; Noireaux, V.; Brivanlou, A. H.; Libchaber, A. Science 2002, 298, 1759-1762. (5) Bruchez, M.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P. Science 1998, Gold:thiolate (Au:SR) cluster compounds (or monolayer pro- 281, 2013-2016. tected cluster compounds) constitute a special subset of metallic (6) Chan, W. C. W.; Nie, S. M. Science 1998, 281, 2016-2018. nanostructures that have been the subject of numerous studies (7) Tran, N. T.; Powell, D. R.; Dahl, L. F. Angew. Chem., Int. Ed. 2000, 39, 4121-4125. 1 in recent years. Conceptually, these compounds consist of a dense (8) Wang, X. J.; Langetepe, T.; Persau, C.; Kang, B. S.; Sheldrick, G. M.; Fenske, metallic gold core surrounded by a shell of thiolate ligands (See D. Angew. Chem., Int. Ed. 2002, 41, 3818-3822. (9) Whetten, R. L.; Khoury, J. T.; Alvarez, M. M.; Murthy, S.; Vezmar, I.; Wang, Chart 1.). The measured optical properties are dominated by the Z. L.; Stephens, P. W.; Cleveland, C. L.; Luedtke, W. D.; Landman, U. Adv. electronic structure associated with metallic bonding within the Mater. 1996, 8, 428-433. inorganic core, while the gross chemical properties are derived (10) Alvarez, M. M.; Khoury, J. T.; Schaaff, T. G.; Shafigullin, M. N.; Vezmar, I.; Whetten, R. L. J. Phys. Chem. B 1997, 101, 3706-3712. with the organic (or biologic) ligands attached to that core. Similar (11) Schaaff, T. G.; Shafigullin, M. N.; Khoury, J. T.; Vezmar, I.; Whetten, R. L.; metallic and semiconductor cluster compounds are finding ap- Cullen, W. G.; First, P. N.; GutierrezWing, C.; Ascensio, J.; JoseYacaman, - plications in biologic imaging, such as strongly scattering centers M. J. J. Phys. Chem. B 1997, 101, 7885 7891. (12) Song, Y.; Jimenez, V.; McKinney, C.; Donkers, R.; Murray, R. W. Anal. Chem. 2003, 75, 5088-5096. * Phone: (865) 574-2297. Fax: (865) 574-9771. E-mail: [email protected]. (13) Jimenez, V. L.; Leopold, M. C.; Mazzitelli, C.; Jorgenson, J. W.; Murray, R. Current address: BWXT Y-12, P.O. Box 2009, MS 8189, Oak Ridge, TN 37821- W. Anal. Chem. 2003, 75, 199-206. 8189. (14) Clarke, N. Z.; Waters, C.; Johnson, K. A.; Satherley, J.; Schiffrin, D. J. (1) Templeton, A. C.; Wuelfing, M. P.; Murray, R. W. Acc. Chem. Res. 2000, Langmuir 2001, 17, 6048-6050. 33,27-36. (15) Schaaff, T. G.; Whetten, R. L. J. Phys. Chem. B 2000, 104, 2630-2641.

compounds, which also show strong size dependencies or From these studies, three to four molecular-like ions were quantum size effects (e.g., electrochemical charging of the cluster observed for a class of gold-phosphane cluster compounds, which core,16,17 chiroptical effects in gold clusters with biologically had previously been thought to contain a single component. derived ligands,15,18 and solid-state molecular crystal structure19). Recently, one such compound has been separated in high purity Laser desorption/ionization mass spectrometry has been by low-pressure chromatographic techniques and analyzed by highly efficient for analyzing the core mass of the cluster matrix-assisted laser desorption/ionization mass spectrometry.23 compounds to provide both rapid monitoring of size separation In addition, other metallic-phosphane cluster compounds have techniques and optimization of reaction parameters to produce recently been studied by electrospray ionization mass spectrom- specific cluster compounds in high yield and the qualitative etry.24 Despite many refinements in the methods for preparation observations regarding further reactions of the cluster com- and isolation (or separation) of the gold-thiolato metal cluster pounds.20,21 Owing to the central role of laser desorption/ionization compound materials, they have never been conclusively demon- (LDI)-MS, many reports have differentiated the separated cluster strated to exist as molecularly defined substances (e.g., single- compounds by their respective core masses. For example, 29- crystal structure determination). However, previous studies allude kDa Au:SC4 refers to a gold cluster compound that, upon to a molecular system composed of cluster species with nearly irradiation with UV irradiation, produces a group of ions centered identical composition and molecular weight. Clearly, an advanced at m/z 29 000 and has an associated ligand shell composed of mass spectrometry investigation, utilizing ultrasensitive and ul- butanethiolate. While LDI-MS mass spectrometry has served a trasoft (gentle) ionization methods, should enable a direct, rapid central role in isolating Au:SR cluster compounds, implementation determination of the distribution of assemblies present after of low-fragmentation ionization methods (e.g., electrospray ioniza- synthesis and isolation. tion and matrix-assisted laser desorption/ionization, MALDI) has This report presents (among other things) progress toward only achieved sporadic success. this goal and illuminates some of the mechanisms and challenges While gold:thiolate cluster compounds have received increased inherent to the (MA)LDI processes, as applied to such materials. attention in recent years due to the ease of preparation, the first Results from both LDI and MALDI mass spectrometry are analysis of gold cluster compounds by mass spectrometry was presented for three gold cluster compounds with the same Au performed on gold:phosphane cluster compounds by McNeal and core mass: 29-kDa Au:SC4, Au:SC6, and Au:SC12. The abundance co-workers using 252Cf-plasma desorption mass spectrometry.22 of structurally relevant ions under both LDI and MALDI conditions was dependent on both the irradiance delivered (on a single-shot (16) Chen, S. W.; Ingram, R. S.; Hostetler, M. J.; Pietron, J. J.; Murray, R. W.; basis) and the total power delivered to the sample. Unlike many Schaaff, T. G.; Khoury, J. T.; Alvarez, M. M.; Whetten, R. L. Science 1998, 280, 2098-2101. other macromolecular structures, structurally relevant high m/z (17) Chen, S. W.; Murray, R. W. J. Phys. Chem. B 1999, 103, 9996-10000. ions were produced from Au:SR cluster compounds under many (18) Schaaff, T. G.; Knight, G.; Shafigullin, M. N.; Borkman, R. F.; Whetten, R. ionization conditions (i.e., desorption from neat films or from L. J. Phys. Chem. B 1998, 102, 10643-10646. (19) Whetten, R. L.; Shafigullin, M. N.; Khoury, J. T.; Schaaff, T. G.; Vezmar, I.; clusters diluted in organic matrixes). The ability to detect these Alvarez, M. M.; Wilkinson, A. Acc. Chem. Res. 1999, 32, 397-406. ions under many ionization conditions provides insight into likely (20) Alvarez, M. M.; Khoury, J. T.; Schaaff, T. G.; Shafigullin, M.; Vezmar, I.; Whetten, R. L. Chem. Phys. Lett. 1997, 266,91-98. (23) Gutierrez, E.; Powell, R. D.; Furuya, F. R.; Hainfeld, J. F.; Schaaff, T. G.; (21) Schaaff, T. G.; Shafigullin, M. N.; Khoury, J. T.; Vezmar, I.; Whetten, R. L. Shafigullin, M. N.; Stephens, P. W.; Whetten, R. L. Eur. Phys. J. D 1999, 9, J. Phys. Chem. B 2001, 105, 8785-8796. 647-651. (22) McNeal, C. J.; Winpenny, R. E. P.; Hughes, J. M.; MacFarlane, R. D.; (24) Kawano, M.; Bacon, J. W.; Campana, C. F.; Winger, B. E.; Dudek, J. D.; Pignolet, L. H.; Nelson, L. T. J.; Gardner, T. G.; Irgens, L. H.; Vigh, G.; J. P. Sirchio, S. A.; Scruggs, S. L.; Geiser, U.; Dahl, L. F. Inorg. Chem. 2001, 40, Fackler, J. Inorg. Chem. 1993, 32, 5582-5590. 2554-2569.

6188 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 desorption/ionization processes for the gold cluster compounds. In addition, this unique system may provide further detail concerning proposed mechanisms for MALDI of typical biologic analytes.

EXPERIMENTAL SECTION The 29-kDa Au:SR cluster compounds used for these studies were synthesized and separated using procedures described in detail elsewhere.21 Briefly, the cluster compounds were prepared by a procedure based on methods described by Brust et al.,25 which are optimized to yield (in high abundance) the compound that produces ions under LDI conditions centered at m/z 29 000 for gold:butanethiolate, gold:hexanethiolate, and gold:dodecanethi- olate clusters (Au:SC4, Au:SC6, and Au:SC12, respectively). The compounds were separated by fractional crystallization, which is effected by slow addition of acetone to a concentrated toluene solution. Neat films of the mixture and separated cluster compounds (for LDI-MS) were prepared by pipeting 1 µLofa concentrated solution (5 mg/mL in toluene) onto a sample plate and allowing the solution dry under ambient conditions. Separation of the compounds was monitored by LDI-MS, as shown below. As described by many reports dealing with matrix-assisted laser desorption/ionization, the ability to cocrystallize the analyte Figure 1. Low m/z (a) and high m/z (b) regions of negative ion in a matrix crystal is highly advantageous, requiring a common laser desorption ionization mass spectrum obtained by irradiating a solvent. For this reason, methylene chloride (HPLC grade, Baker neat film of a mixture of gold:thiolate cluster compounds with N2 laser Scientific) was used as a common solvent for the cluster at 2.4 MW/cm2. compounds and the organic matrixes. Four matrixes (Sigma- Aldrich) were tested: 3,5-dihydroxybenzoic acid (DHB), 3,5- dimethoxy-4-hydroxycinnamic acid (sinapinic acid), 1,8,9-an- controlled settings, the beam was diverted outside the vacuum thracenetriol (dithranol), and trans-3-indoleacrylic acid. Films of chamber with a 90% transmission prism into a J3-09 pyroelectric/ the matrix-diluted cluster compounds were generated by diluting silicon Joulemeter (Molectron). The laser power settings in the 20 µLofa10µg/mL methylene chloride solution of Au:SR cluster instrument control software were changed, and the output from compounds 1:1 (v/v) with a saturated matrix solution in methylene the Joulemeter (fluence) was monitored with a digital storage chloride. Of the four matrix molecules, sinapinic and trans-3- oscilloscope. The irradiance values listed below were calculated indoleacrylic acid produced comparable mass spectra, while DHB for a focus of 200 µm (diameter) on the sample plate and corrected and dithranol did not seem to reduce fragmentation of the Au:SR for additional transmission losses due to the focusing lens and cluster compounds as well (i.e., most spectra were similar mass quartz window installed on the vacuum chamber. Reported values spectra obtained by irradiating neat films). MALDI mass spectra were obtained by averaging the measured fluence from 20 shown in this report are derived from ions produced by irradiation individual laser shots. of cluster compounds diluted in a sinapinic acid matrix. Laser desorption and matrix-assisted laser desorption/ioniza- RESULTS The Au:SR cluster compounds (where R ) butane, C4; hexane, tion mass spectra were obtained using Perseptive Biosystems C6; and dodecane, C12), produced different types of high m/z Voyager DE linear time-of-flight mass spectrometer operating in ions upon irradiation with a pulsed UV laser (N , 337 nm) under delayed extraction mode with an accelerating voltage of 20 kV. 2 various ionization conditions. To differentiate these two conditions The postdesorption delay time and space-focusing conditions were in the results to follow, LDI refers to irradiation of neat (thick) optimized using laser-desorbed cations and anions from the neat films generated by depositing 1-2 µL of a concentrated (10-20 film by resolving the m/z 32 spacing in the ions detected at m/z mg/mL) solution of Au:SR cluster compounds and MALDI refers values centered at 29 000. While these ions could not be baseline to irradiation of cluster compounds dispersed within a sinapinic resolved, the periodic spacing of m/z 32 could still be used to acid matrix. ensure optimum resolution conditions and the calibration was LDI mass spectra of gold:thiolate were obtained by irradiating correct under different ionization conditions (e.g., LDI vs MALDI) neat films of the Au:SR cluster compounds, typically at an for this region of the mass spectrum. irradiance of 2-3 MW/cm2. Figure 1 shows different regions of The Voyager DE mass spectrometer is equipped with a the negative ion LDI mass spectrum generated from a mixture of VSL-337 ND pulsed N2 laser (Laser Science, Newton, MA) with a Au:SR cluster compounds (R ) C12). Two distinct types of anions 4-ns pulse width at 337 nm. To determine the magnitude of the are produced when neat Au:SR films are irradiated. Ions in the laser irradiance delivered to the sample under different computer- low m/z region (Figure 1a) are extremely intense compared to - (25) Brust, M.; Walker, M.; Bethell, D.; Schiffrin, D. J.; Whyman, R. J. Chem. those at higher m/z values (on the order of 10 100 times signal Soc., Chem. Commun. 1994, 801-802. intensity) and can produce large background signals (due to

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 6189 detector relaxation) that extends to m/z ∼5000. In addition to the m/z 197 [Au]- anion, groups of anions are detected with nominal differences in m/z of 229. The lowest m/z ions in each group - corresponds to the “bare” [AuNSM] anion (see labels in Figure 1a). Higher m/z ions in each group are detected at m/z values that are separated by either m/z 12 or 13, which presumably corresponds to addition of elemental carbon or the combination of carbon and hydrogen. The relative intensity of these ions increases with increasing laser irradiance and are invariant with respect to ligand composition. Therefore, the remainder of this report concentrates on the higher m/z ions found at lower relative abundance that have a direct relation to the structure of the cluster compound as it exists in the condensed phase. Similar to previous reports,9,11,20 the higher m/z region of LDI mass spectra shown in Figure 1b is composed of peaks corresponding to negative ions in distinct regions centered about m/z 8000, 15 000, 22 000, and 29 000. The spectrum represents the average of 512 individual mass spectra obtained at an irradiance of 4.1 MW/cm2. To lower the contribution of noise arising from the low m/z ions, a low-mass cutoff was used to blank the mass spectrum below m/z 4000. The mass spectrometer resolution (e.g., space focus and delayed extraction) was optimized for the mean m/z shown in the Figure (m/z ∼15 000) to obtain the best Figure 2. Positive (a) and negative (b) ion laser desorption/ ionization mass spectra (at 2.4 MW/cm2) from a Au:SC4 cluster resolution for the entire window. The effect of delayed extraction compound after separation from smaller and larger size cluster on mass resolution and measured ion abundances is well docu- compounds, as seen in Figure 1. The insets in (a) and (b) show the mented26 and has been found to produce LDI mass spectra from narrower m/z range centered at m/z 29 000. The peaks detected at mixtures of Au:SR’s that do not always reflect the true abundances higher m/z values correspond to dimer, trimer, etc., ions of the base of components in the mixture being analyzed.27 Thus, when ion at m/z 29 000. monitoring separations or optimizing reaction conditions, LDI After fractionation and isolation, the 29-kDa (core mass) Au: mass spectra are acquired with static acceleration and fixed space SC4 cluster compound was separated from its other compounds focusing to obtain a qualitative, but more accurate representation having disparate core masses. Figure 2 shows the negative and of component abundance. positive ion LDI mass spectra for this compound obtained with When mixtures of Au:SR cluster compounds are irradiated, an irradiance of 4.1 MW/cm2. The spectra shown are the average the negative ions (and positive ions, not shown) detected have of mass spectra obtained from 32 laser shots. The acquisition of periodic spacing from m/z ∼6000 until it is impossible to resolve 32 shots has been observed to aid in the ability to resolve the the spacing due to instrumental resolution limitations. The ions m/z 32 spacing. When a larger set of spectra are averaged, the detected can be described as having major and minor m/z spacing m/z 197 and 229 spacing is resolved well (as seen in Figure 1b), consistent with the composition of the inorganic cluster core. The but the m/z 32 spacing cannot be resolved in the m/z 30 000 minor spacing in each group of ions corresponds to an m/z region of the mass spectrum. This is likely due to slight changes difference of 32 (S). The major m/z spacing between the most in laser power, electronic jitter, or both, which can become more abundant ion in each adjacent group corresponds to a difference pronounced when averaging larger sets of spectra. Comparison of either 197 (Au) or 229 (AuS). The N, M labels in the inset of the positive and negative ion mass spectra indicates that similar correspond to ions having the general formula [AuNSM]. Ions of groups of ions are produced at what seems to be approximately - the same general AuNSM composition were observed by Arnold the same intensity in both positive and negative ion modes. As and Reilly for the separated compound that produces negative can be seen in the insets of Figure 2, the same general m/z 197 ions centered at m/z 15 000.28 Abundant ions follow this general and 229 spacing is prevalent. The distribution of ion abundances progression from approximately m/z 6000 until the m/z 32 spacing starts as an abrupt rise at m/z ∼26 500, and ions at lower m/z cannot be resolved. While Au is monoisotopic, S has four natural values with the characteristic m/z 32 and 197 spacing are not isotopes: 32S (95.02), 33S (0.75), 34S (4.21), and 36S (0.02). The detected below that cutoff. The distribution of ions reaches a resolution of the mass spectrometer is not sufficient to resolve maximum at m/z ∼29 000 and decreases until ions with the the isotopic abundances; thus assignments and calibration were characteristic m/z 32 and 197 spacing are not detected (m/z made using the exact mass of 197Au (196.966 55) and the average 32 000). Peaks detected at higher m/z values correspond to mass of S (32.065). multimers of the m/z 29 000 ions (i.e., 2 × 29 000, 3 × 29 000, etc.). These multimer ions have been observed up to 10 × 29 000 (26) Reilly, J. P.; Colby, S. M. Anal. Chem. 1996, 68, 1419-1428. under negative ion operation and up to 8 × 29 000 under positive (27) Schaaff, T. G.Preparation and Characterization of Thioaurite Cluster ion operation. Compounds. Ph.D. Dissertation, School of Chemistry and Biochemistry, Georgia Institute of Technology, 1998. Threshold irradiance for detection of positive ions from the (28) Arnold, R. J.; Reilly, J. P. J. Am. Chem. Soc. 1998, 120, 1528-1532. Au:SC4 cluster compound was measured at 1.5 MW/cm2, which

6190 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 Figure 3. Positive ion laser desorption/ionization mass spectra Figure 4. Positive ion matrix-assisted laser desorption/ionization obtained from the separated cluster compound shown in Figure 2 at mass spectra obtained from the separated Au:SC4 cluster compound increasing irradiance (from top to bottom). The numbers above each shown in Figure 2 at increasing irradiance (from top to bottom). The mass spectrum correspond to irradiance in MW/cm2. numbers above each mass spectrum correspond to irradiance in MW/ cm2. The * above spectra at 2.1-3.0 MW/cm2 corresponds to m/z 28 400. produces a broad, featureless peak (similar to the top spectrum in Figure 3) at a signal-to-noise level of 2. Figure 3 illustrates the mass spectra on the irradiance when the 29-kDa (core mass) changes in positive ion abundance as a function of irradiance. At Au:SC4 cluster compound is diluted in a matrix of sinapinic acid. slightly above threshold for production of positive ions from neat Because it is well known that MALDI matrixes can produce so- films, no m/z 32 or 197 spacing is observed in the ion abundances called “hot spots” (where the signal is much stronger in specific (1.7 MW/cm2, top spectrum). In addition, at this lower laser areas of the sample), the mass spectra shown in Figure 4 were power, the peak apex is centered at m/z ∼30 000. As irradiance obtained by continuously repositioning the laser spot over the is increased, not only do the ion abundances start to show the entire sample surface while acquiring data. characteristic m/z 32 and 197 spacing, but the apex of the ion At lower irradiance (below 3 MW/cm2), the abundances of abundances shifts to lower m/z values (∼29 000). At high ions centered at m/z 29 000 wwere similar to those measured from irradiance (>5 MW/cm2), the m/z 32 spacing cannot be resolved LDI-MS of neat films (top 3 spectra, Figure 4), though the ions and the peak apex has shifted still lower, to m/z ∼28 500. A similar detected correspond to slightly lower m/z values. At near- dependence of ion abundances on irradiance was observed threshold irradiance, the ions detected correspond to a broad, negative ions generated from neat films of the 29-kDa Au:SC4 featureless peak with an apex at m/z 29 200, similar to the near- cluster compounds. Similar to high-fragmentation conditions for threshold LDI from neat films but at lower m/z values. At slightly MALDI, shown below, it should be noted that the abundance of higher irradiance (2-3 MW/cm2), the m/z spacing is resolved, ions detected at m/z ∼29 000 were relatively invariant with respect also consistent with resolution of these characteristic ions under to chainlength of the thiolate ligand. For example, similar ion LDI conditions at higher irradiance. While the m/z 32 spacing is abundances were detected for both LDI-generated positive and not resolved to the degree found in LDI from neat films, the m/z negative ions from Au:SC4, Au:SC6, and Au:SC12 cluster com- 197 spacing is partially resolved for approximately three or four pounds with core masses of 29 kDa. groups of ions. The first group of ions detected (i.e., the first peak MALDI has been used to mitigate fragmentation of the Au: in the progresses of m/z 197 spacing, denoted by * in Figure 4) SR cluster compounds during ionization repeatedly with a number corresponds to m/z 28 400. of typical organic matrix molecules. To date, the best matrix has At 3 MW/cm2, the ions corresponding to m/z values between been determined to be sinapinic acid. However, the use of trans- 28 000 and 30 000 remain relatively unchanged, while a distribu- 3-indoleacrylic acid (typically used for MALDI of organic poly- tion of ions corresponding to the broad featureless peak at m/z mers) was found to have performance comparable to that of 32 000 increases in relative abundance. Above 3.0 MW/cm2, the sinapinic acid. Unlike many other MALDI analytes, it appears that low m/z peaks again appear featureless and the ions correspond- two desorption/ionization regimes exist for MALDI of Au:SR ing to m/z 32 000 are detected at much higher relative abundance. cluster compounds (high and low fragmentation), which are The absolute intensity of the ions detected does not change dependent on the incident irradiance and total irradiance delivered appreciably, but the relative signal-to-noise ratio increases slightly. to the matrix/cluster mixture. Figure 4 shows the dependence of Reproducible features are partially resolved, superimposed on the

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 6191 Figure 6. Evolution of positive ion matrix-assisted laser desorption/ Figure 5. Positive ion matrix-assisted laser desorptionionization ionization mass spectra obtained from the Au:SC4 cluster compound mass spectra obtained at 2.1 MW/cm2 from the Au:SC4, Au:SC6, at irradiance of 4.0 MW/cm2. The numbers above each mass and Au:SC12 cluster compounds (from top to bottom, respectively). spectrum correspond to the number of laser shots delivered to the Again, the * above the spectra corresponds to m/z 28 400, common matrix/analyte sample. among all three compounds. which appeared optically dense, and triggering the laser (off and broad peak centered at m/z 32 000, but resolution at this m/z on) manually generated mass spectra such as those shown in range is not sufficient to make qualified assignments of ion Figure 6. The irradiance used for the spectra in Figure 6 was 4.0 composition, nor do the m/z values contain recognizable spacing MW/cm2. The mass spectrum obtained in the first few laser shots (e.g., consistent with loss of whole ligand molecules from the ion). is similar to both low-irradiance MALDI (Figures 4 and 5) and Ion abundances detected for the 29-kDa Au:SC6 and Au:SC12 LDI from neat films. During these first few shots, the distribution cluster compounds showed similar irradiance-induced effects. of ions has an apex at m/z ∼28 000, but the characteristic m/z Figure 5 shows the average of 32 mass spectra obtained at low 197 spacing is only partially resolved across the distribution of irradiance (2.4 MW/cm2) from the three 29-kDa cluster com- ion abundances detected. After the first 10-20 laser pulses, the pounds (top to bottom): Au:SC4, Au:SC6, and Au:SC12. With only ion abundances centered at m/z 29 000 decreases and the higher slight differences, the three compounds produce remarkably m/z ions are detected at an increased relative abundance centered similar groups of ions with m/z 197 spacing, but the m/z 32 at m/z 32 000. At ∼30 laser shots, ions centered at m/z 29 000 spacing was not resolved for these compounds. While mass and 32 000 are approximately the same abundance. Finally, at ∼75 resolution limitations preclude unequivocal assignment of the ions laser shots, the ions at m/z 29 000 are completely suppressed and detected, it is clear that the position of the first abundant group the ions detected are centered at m/z 32 000. In addition, initial of ions (with the characteristic m/z 197 spacing) is measured at mass spectra (first 10-20 shots) had ion abundances centered at approximately the same m/z (28 400) for all three cluster m/z 58 000, 87 000, etc., corresponding to the multimers of the compounds (denoted by * in Figures 4 and 5). The distribution ions centered at m/z 29 000 as seen in LDI mass spectra obtained of ions in this region is narrower than those detected with LDI. from irradiating neat films. The relative abundance of multimers As seen in Figure 5, the rise in abundance under this ionization remained constant with respect to the abundance of the ions at condition occurs at m/z 27 500, compared to m/z 27 000 for LDI m/z 29 000. However, no ion abundances were detected at the (inset, Figure 1a), and the ion abundance returns to near baseline m/z values corresponding to multimers of the ions centered at at m/z ∼31 000 (compared to m/z ∼32 000 for LDI). m/z 32 000 (see discussion below). Upon repeated experiments to evaluate changes in ion abun- After determining the changes in the abundance of desorbed dances as a function of irradiance for the three compounds, it ions under different laser conditions for the 29-kDa Au:SC4 was apparent that the relative ion abundances were also affected compound, the other two 29-kDa cluster compounds were by the total power delivered to the sample. Mass spectra obtained investigated using similar ionization conditions. Figure 7 shows from repeated pulses delivered to dense sample areas (i.e., areas the low-fragmentation MALDI mass spectra from three different with large crystalline matrix structures) changed with subsequent 29-kDa cluster compounds (from top to bottom): Au:SC4, Au: laser shots, while those obtained from sparse or thin areas on SC6, and Au:SC12. The position of the peak apex and tailing edge the sample were consistently similar to those seen in Figure 5. changes consistent with the longer chain length ligands of the Focusing the laser on an area in the matrix/cluster sample region, different cluster compounds. It is also interesting to note that the

6192 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 properties similar to hydrophobic polymers). As would be expected for a complex molecular structure, many other subtle properties must be considered for application of MALDI to metallic cluster compounds. For example, the near-IR, visible, and ultra- violet optical properties are quite different from that of typical MALDI analytes. The extinction coefficient (at 337 nm) for the 29-kDa cluster compounds is on the order of that for the sinapinic matrix molecules in which they are dispersed. In addition, the hydrophobicity of the cluster compounds may preclude their dispersal into various matrixes. Overcoming these differences and understanding specific ionization processes in these complex compounds are required before soft ionization techniques can be applied to this class of cluster compounds, or more importantly, used to develop similar techniques for other types of cluster compounds (e.g., semiconductor and other metallic cluster compounds). The lack of mass spectrometry-based studies of cluster compounds is likely due to a number of factors, with one of the most important being sample purity of available cluster compounds. To draw on the biologic MALDI analogy, understanding and applying MALDI to metallic cluster compounds before Figure 7. Positive ion matrix-assisted laser desorption/ionization isolation by cluster size would be similar to developing MALDI mass spectra obtained by irradiating the matrix/sample area at an for proteins by starting with an extract of all proteins from a whole irradiance of 4.0 MW/cm2 for 60-70 laser pulses and then acquiring cell without any prior chemical separations. As shown in Figure mass spectra for 32 additional laser pulses. The mass spectra (from top to bottom) correspond to ions generated from the Au:SC4, Au: S1 (Supporting Information), the changing abundances of ions SC6, and Au:SC12 cluster compounds, which produced m/z ∼28 400 as a function of laser irradiance and number of laser shots produce ions shown in Figure 5. The arrows in each mass spectrum cor- mass spectra with peaks superimposed from different size nanoc- respond to the expected m/z for a molecular-type ion from an intact rystal cores or, more detrimental, at m/z values that normally cluster compound with an Au:SR ratio of 2.57:1, as determined by correspond to ions generated from LDI of neat films. Thus, without elemental analysis. separations, it is impossible to determine which ions are produced under which conditions. Of the gold:thiolate cluster compounds, full width half-maximum (fwhm) of the distribution of ions does the compound that produces ions centered at m/z 29 000 by LDI not change appreciably with different chain lengths. A previously is the best understood compound (structurally and electroni- isolated 29-kDa Au:SC12 cluster compound was ionized by MALDI cally),21 which is why this compound (with different ligand with 3,5-dihydroxybenzoic acid as a matrix.21 However, in that molecules) was chosen for these studies. case, the ion distribution extended from m/z 28 000 to 40 000. The measurement of ion abundances corresponding to the In summary, the three 29-kDa Au:SR compounds (R ) C4, approximate core mass of the cluster compound is consistent with C6, and C12) produced nearly identical mass spectra both at low other analytical techniques more commonly used to analyze irradiance and during the initial irradiation of the matrix/cluster cluster sizes, e.g., X-ray diffraction,19 electron microscopy,11 and sample. The ions produced under this particular ionization scanning probe microscopy.29,30 Under LDI conditions, the fwhm condition were centered at m/z 29 000, with the first detected of the distribution of ions centered at m/z 29 000 is ∼2500. This (resolved) ion being m/z 28 400 for all three compounds. For the in turn translates into a core diameter dispersion (typically Au:SC4, Au:SC6, and Au:SC12, higher irradiance or additional laser determined by high-resolution electron microscopic analysis) of pulses produced higher m/z ions that were centered at m/z 32 000, 1.67 ( 0.02 nm using the formula 33 200, and 37 400, respectively. In previous reports, the Au:S ratio for the 29-kDa (core mass) Au:SR compounds was determined N (1/3) ) Au 21 D ( ) (1) by elemental analysis to be 2.57:1. The arrows in Figure 7 denote eq (π/6)d the m/z value that would correspond to an intact ion assuming the peak at m/z 28 400 corresponds to the number of gold atoms in the cluster core. which is based on the density (d ) 59 atoms/nm3) of gold in its native fcc structure. While the relative error associated with the DISCUSSION mass spectrally derived diameter (0.02 nm) clearly cannot translate While matrix-assisted laser desorption/ionization has proven into a physical parameter, it does illustrate the precision associated extremely useful for the analysis of thermally labile macromol- with the measurement, which cannot be achieved by either ecules, the implementation of this “soft” ionization method to electron microscopy or powder X-ray diffraction measurements. inorganic cluster compounds still faces significant challenges. Considering the structure of the cluster compounds (i.e., inorganic (29) Harrell, L. E.; Bigioni, T. P.; Cullen, W. G.; Whetten, R. L.; First, P. N. J. Vac. Sci. Technol. B 1999, 17, 2411-2416. core surrounded by a hydrophobic monolayer), there are similari- (30) Bigioni, T. P.; Harrell, L. E.; Cullen, W. G.; Guthrie, D. E.; Whetten, R. L.; ties to typical MALDI analytes (similar in size to proteins, chemical First, P. N. Eur. Phys. J., D 1999, 6, 355-364.

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 6193 From the seminal work by Tanaka and co-workers31 and others to follow,32,33 it is well documented that metallic colloids can effect liberation of large molecular biologic ions from a glycerol matrix to produce “fragmentation-free” mass spectra similar to those obtained routinely with organic matrixes. Bulk gold and silver also share similar photochemically driven reactions that occur under intense UV irradiationsselective cleavage of the S-C bond.34,35 The propensity to selectively cleave the S-C bond under UV irradiation provides a mechanism similar to those proposed for the irradiation of organic matrix crystals (i.e., violent disruption of a specific crystal structure into an expanding plume) to liberate molecular ions from biopolymers. Presumably, this violent disruption of the Au:SR cluster compounds can allow liberation of “intact” cluster cores. Thus, the detection of high m/z ions such as those shown in Figure 1 may be the consequence of the Au:SR clusters acting both as a poor matrix and an analyte. The mechanism for desorption and ionization of these large structurally relevant ions from neat films of Au:SR cluster compounds is likely similar to those proposed for MALDI. Presumably, the high m/z ions are generated in a dense, expanding plume populated by both high m/z ions and low m/z ions. Both fragmentation of these high m/z ions and aggregation with themselves (e.g., multimers in Figure 1b) and with low m/z ions (as shown in Figure 1a) likely occurs within this dense plume to add to the distribution of ions detected in the final LDI mass spectrum. Reactions within a dense plume (under LDI conditions) would be consistent with the narrower distribution of ions detected under certain MALDI conditions (plume is likely not as dense with MALDI prepared samples). The gold:thiolate cluster compounds are unique (compared to other thermally labile macromolecules) in that they produce structurally relevant high m/z ions under all ionization conditions. Many have reported on a phenomenon in MALDI in which it takes a few laser pulses to observe ions being produced from the matrix/analyte sample.36-38 This has been previously attributed Figure 8. Illustration of ion formation of Au:SR cluster compounds to a so-called “cleaning off” effect, where protein molecules not dispersed in an organic matrix under high- and low-fragmentation conditions (left and right images, respectively). The left image incorporated in the matrix crystal or amorphous material is ablated corresponds to the processes that occur at low irradiance or initial from the face of well-cocrystallized matrix/analyte samples and irradiation at high-irradiance levels to produce spectra similar to those protein ions are not formed. While this effect is difficult to observe generated from neat films due to either a neat film on the matrix/ directly in the analysis of proteins (because they form either analyte crystal or an amorphous layer, which allows for extensive metastable ions or no ions at all during this process), this effect fragmentation. The right illustrates a situation in which the amorphous (or neat) layer has been ablated and the cluster compounds are is observed more clearly in MALDI of the cluster compounds. ejected (or liberated) from the matrix/analyte crystal similar to other Considering the changes in mass spectra under different macromolecules. ionization conditions, it is likely that gold cluster compounds are present both as a surface (or amorphous) layer and cocrystallized within the matrix crystals. Even though the cluster compounds and the matrix molecules are both soluble in methylene chloride, the differences in hydrophobicity can likely still cause a high (31) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T. Rapid degree of segregation during the evaporation of the methylene Commun. Mass Spectrom. 1988, 2, 151-156. (32) Lai, E. P. C.; Owega, S.; Kulczycki, R. J. Mass Spectrom. 1998, 33, 554- chloride solvent. For example, the Au:SR cluster compounds with 564. R ) C4, C6, and C12 are only soluble in nonpolar solvents or (33) Schurenberg, M.; Dreisewerd, K.; Hillenkamp, F. Anal. Chem. 1999, 71, slightly polar solvents. On a microscopic scale, cocrystallization 221-229. (34) Lewis, M.; Tarlov, M.; Carron, K. J. Am. Chem. Soc. 1995, 117, 9574- with a highly polar molecule (e.g., an organic acid) would 9575. presumably produce either a highly amorphous solid or a neat (35) Rieley, H.; Price, N. J.; Smith, T. L.; Yang, S. H. J. Chem. Soc., Faraday film of the cluster compounds on the matrix crystal (See Figure Trans. 1996, 92, 3629-3634. (36) Bolbach, G.; Riahi, K.; Spiro, M.; Brunot, A.; Breton, F.; Blais, J. C. Analusis 8.). At low irradiance, erosion of the amorphous or neat layer 1993, 21, 383-387. would presumably require many shots to finally reach a portion (37) Perera, I. K.; Kantartzoglou, S.; Dyer, P. E. Int. J. Mass Spectrom. Ion Processes of the matrix/analyte crystal in which the clusters are adequately 1996, 156, 151-172. (38) Pittenauer, E.; Schmid, E. R.; Allmaier, G.; Puchinger, L.; Kienzl, E. Eur. dispersed within the crystal structure. This cleaning-off period Mass Spectrom. 1996, 2, 247-262. would be substantially less at higher irradiance, which is consis-

6194 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 desorbed molecules ions from a diluted state within the organic matrix (Figure 8b). Complete mitigation of fragmentation during ionization still seems elusive for this class of macromolecular compounds, but the combination of different ionization conditions provides a consistent description of the molecular structure of the Au:SR cluster compounds. Considering the plume density is likely lowest for the high-fragmentation MALDI conditions (less propensity for broadening the distribution of ions due to aggregation with low m/z ions), this method would provide the more accurate determination for the core mass of the cluster compound at m/z 28 400 (though this number likely includes a small contribution from remaining sulfur). In addition to sharing a remarkably similar core mass, the number of ligands is also similar. Assuming the m/z 28 400 represents the lowest fragment of the inorganic core, the core size is estimated at 144 gold atoms. With this assumption, the number of ligands associated with all three compounds (i.e., mass difference between 28 400 and arrows shown in Figure 7) suggests all three compounds have ∼53-56 ligand molecules associated with the condensed-phase structure. While it is still not possible to unequivocally assign a true “molecular weight”, the approximate molecular weights of the Au:SC4, Au:SC6 and Au:SC12 cluster compounds is determined to be ∼33 500, 35 000, Figure 9. Positive ion matrix-assisted laser desorption/ionization and 39 000, respectively. mass spectra for the Au:SC4 cluster compound obtained by irradiating Of course, an alternative interpretation considers the mass 2 the matrix/sample area at an irradiance of 4.0 MW/cm for the initial distribution of ions produced by MALDI under both high and low 64 laser pulses (a) and the average of mass spectra from 32 fragmentations. The fwhm of ions produced under all MALDI subsequent laser pulses (b). The dimer ion of the ions centered at m/z 29 000 is detected at m/z 58 000 in the first 64 laser pulses. Upon conditions is approximately the same for all three compounds. suppression of the m/z 29 000 ions, a distribution of ions at m/z 16 000 That is, a similar distribution of ions is detected regardless of is detected in the mass spectrum corresponding to final 32 laser ligand or ionization conditions. Considering the distribution of ions pulses, presumably the doubly charged species of the group of ions in this manner would lead to the assumption that an “island of producing the distribution centered at m/z 32 000. stability” exists for gold cluster compounds, which is centered at ∼145 atoms. This is also a plausible explanation, considering tent with the observation that, under low irradiance (and the first different theoretical models that have predicted a number of gold few laser shots under high irradiance), the mass spectra obtained cluster structures of exceptional stability between 140 and 150 from the matrix/analyte sample is similar to that obtained from atoms.39,40 From this interpretation, the peak positions and widths neat films. After the ablation of the neat film or amorphous can be treated statistically to determine level of “purity” or size/ material, the ions are instead generated from a well-cocrystallized molecular weight dispersion, somewhat analogous to a polymer matrix/analyte structure in which the cluster compounds are system. Fitting the distribution of ion abundances under different highly diluted to produce molecular-like ions. ionization to a typical Gaussian distribution function can provide Because ions are formed under many ionization conditions, it statistical information regarding constraints on the assembly is possible to probe the transition from neat film (or amorphous composition. For this set of compounds, the core mass is 29 100 matrix/cluster crystallinity) with studies such as those shown in ( 800 Da, corresponding to a Au core number of 147 ( 4. Figure 6. Several aspects of the complete mass spectra obtained Considering some sulfur content is still present, the apex estimate in this intermediate stage provide additional information in support is likely skewed toward slightly higher core mass, and the of ablation of a neat or amorphous film. Figure 9 shows an distribution is likely larger than that in the intact cluster com- 2 extended mass spectrum obtained at 4.0 MW/cm during the pound. The molecular weights and associated distributions of the changeover from high- to low-fragmentation regimes. In addition Au:SC4, Au:SC6, and Au:SC6 compounds obtained by the same to the two distributions of ions centered at m/z 29 000 (high type of fitting are determined to be 32 000 ( 700, 33 000 ( 1100, fragmentation) and 32 000 (low fragmentation), two additional and 37 000 ( 1000, respectively. distributions of ions are detected centered at m/z 16 000 and Regardless of the interpretation (e.g., based on assumption of 58 000. The ion at m/z 58 000 is the characteristic dimer ion single core and fragmentation or a statistical treatment of the produced under LDI conditions, but the m/z 32 000 ion does not distribution of ion abundances), it is clear the overall molecular produce a corresponding dimer at m/z 64 000. The peak at m/z assembly is highly uniform in structure when the masses of the 16 000 presumably corresponds to the doubly charged species of respective components are considered. (i.e., gold 197 Da and the ions that contribute to the peak detected at m/z 32 000. The detection of the two additional peaks centered at m/z 58 000 and (39) Cleveland, C. L.; Luedtke, W. D.; Landman, U. Phys. Rev. Lett. 1998, 81, 2036-2039. 16 000 are consistent with the ablation of an amorphous or neat (40) Cleveland, C. L.; Landman, U.; Schaaff, T. G.; Shafigullin, M. N.; Stephens, layer (Figure 8a) followed by desorption and ionization of P. W.; Whetten, R. L. Phys. Rev. Lett. 1997, 79, 1873-1876.

Analytical Chemistry, Vol. 76, No. 21, November 1, 2004 6195 thiolate ligands C4 89 Da, C6 117 Da, and C12 201 Da). The The data obtained by mass spectrometry represent the statistical similarity of the MW distributions to the core mass distribution average of the complete ensemble of cluster compounds. While though, brings into question this type of treatment. For example, X-ray diffraction represents the statistical average of the com- if MALDI is producing intact ions from compounds with a pounds, it cannot provide information regarding the relative distribution of core sizes, there is no (or only slight) statistical abundance of different cluster sizes in mixtures or offer qualitative deviation in the number of ligand molecules associated with the information regarding the sample purity. For high-resolution different core sizes. It should also be noted that the approximate electron microscopy, typical electron micrographs (and histograms compositions extracted from the statistical approach (Au143-151SC433, generated from them) represent only a few hundred clusters, Au143-151SC632-34,Au143-151SC1238-40) are both inconsistent with which could be considered a statistically valid representation of elemental analysis and counterintuitive (e.g., more SC12 ligands the entire ensemble on cluster compounds. However, the time in associated with assembly than SC6 or SC4). The single core and which the structurally relevant information is extracted from these fragmentation discussed in the previous paragraph are both two methods is usually on the order of hours, not minutes. consistent with elemental analysis and intuitively attractive, With the increasing interest in nanostructured materials, there considering the isolation of the Pd145 cluster compound by Tran is a distinct need for efficient and reliable analytical tools for their and co-workers.7 analysis. Mass spectrometry, combined with ionization techniques developed for other macromolecules, is promising because of its CONCLUSIONS high sensitivity and ever-increasing mass range and mass resolu- Cluster compounds are an increasingly important class of tion. While it is still problematic to obtain completely fragmenta- macromolecular structures. In addition to providing excellent tion-free MALDI mass spectra for the gold:thiolate and gold: model systems for understanding fundamental quantum effects phosphine cluster compounds, these compounds still constitute in nanostructures, many applications could be potentially impacted the bulk of any studies involving mass spectrometry and metallic by their further development: catalysis, sensor development, clusters. Advances in both ionization of the cluster compounds biologic tags, and molecular-scale electronics. Because the optical and application of these advances to other inorganic cluster and electronic properties of clusters and nanocrystal compounds compounds will rely on further studies of ionization processes are inexorably linked to the core size of the compound, realization and optimization of ionization conditions for these selected types of these proposed applications requires that methods and tech- of cluster compounds and other related nanocrystals. niques be developed to accurately synthesize and isolate the The author thanks Robert L. Whetten and Robert L. Hettich compounds with distinct inorganic core sizes. Accompanying this for their advice and suggestions regarding the drafting of the need for isolation of molecular-like structures on the nanometer manuscript and Gregory B. Hurst for use of the time-of-flight scale, high-throughput analytical methods need to address separa- instrumentation. Research was supported by the Division of tions and determination of purity, in much the same way as they Chemical Sciences, Geosciences and Biosciences, Office of Basic are used for other macromolecular structures across chemistry Energy Sciences, U.S. Department of Energy at Oak Ridge and biology. National Laboratory, managed and operated by UT-Battelle, LLC The ability to quickly “size” the gold:thiolate cluster through under Contract DE-AC05-00OR22725. monitoring ions generated by LDI-MS of neat films allowed for efficient optimization of synthetic parameters, as well as a SUPPORTING INFORMATION AVAILABLE convenient technique for monitoring size separations. There are Additional information as noted in text. This material is two advantages to the mass spectrometry approach over traditional available free of charge via the Internet at http://pubs.acs.org. inorganic materials analytical methods (e.g., X-ray diffraction and TEM) for analysis of these nanostructured materials: speed and statistics. The typical time required to analyze the cluster Received for review November 14, 2003. Accepted April compound by MALDI (or LDI) mass spectrometry is on the order 23, 2004. of 2-5 min from sample preparation to data collection and analysis. AC0353482

6196 Analytical Chemistry, Vol. 76, No. 21, November 1, 2004