Journal of Nanoparticle Research 4: 417–422, 2002. © 2002 Kluwer Academic Publishers. Printed in the Netherlands.

Mass spectrometric analysis of water-soluble gold nanoclusters L. Maya1 , C.H. Chen2 , K.A. Stevenson2 , E.A. Kenik3 , S.L. Allman2 and T.G. Thundat2 Chemical and Analytical Sciences Division and Center for Advanced Engineering Science Advanced Research, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6119, USA (Fax: 865 574 4939; E-mail: [email protected]); 2 Life Sciences Division, 3 Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA

1

Received 18 March 2002; accepted in revised form 27 June 2002

Key words: particle size, measurements, scanning electron microscopy, transmission electron microscopy, scanning probe microscopy, aggregates

Abstract Batches of water-soluble gold nanoclusters of nominal 2.0 or 3.5 nm diameter were prepared to evaluate particle size determinations by a number of techniques such as transmission electron microscopy or atomic force microscopy and to validate estimates derived by mass spectrometric analysis using matrix-assisted laser desorption ionization (MALDI). Good agreement was found and MALDI lends itself to analyses even in the presence of aggregates. Introduction Knowledge of the particle size is a common concern in synthesizing or manipulating nanoparticles. The most common techniques in use utilize microscopic examination by means of scanning electron microscopy (SEM), transmission electron microscopy (TEM) and scanning probe microscopy either atomic force or tunneling (AFM or STM). An additional readily available estimate may be derived from the determination of the angular spread of the reflection from crystallites in a powder X-ray diffractogram (XRD) to obtain a quantitative measure of the mean particle size of the sample. Additional techniques involve scattering measurements that, given the size range involved, are conducted with neutrons or X-rays and sedimentation measurements, that would require an ultracentrifuge given the size range. Finally, recent developments have made possible the use of mass spectrometers to establish the size of particles up into the megadalton size range. The present study collected size estimates for a given gold nanocluster preparation as derived by XRD, TEM, AFM to compare and assess the use of mass spectroscopy as a tool to obtain size estimates of colloidal systems initially present in an aqueous

medium. The answer is relevant to establishing mass spectroscopy as an additional tool to obtain size information on metallic nanocluster preparations. There are precedents to the use of mass spectrometric analysis to obtain particle size of nanoclusters; representative examples are that of Alvarez et al. (1998) who examined aerosols containing salts or passivated sodium clusters. In studies more specifically related to gold nanoclusters, Whetten et al. (1996) studied the separation and mass analyses of alkylthiol coated gold nanoclusters. Finally, Schaaff et al. (1998) characterized a water-soluble glutathione coated gold cluster having a molecular weight of 10.4 kDa.

Materials and methods The reagents, hydrogen tetrachloroaurate, tetrakis (methoxy) phosphonium chloride and thioctic acid as well as methanol were obtained from Aldrich Chemical Company and were used without any additional purification. Purified and filtered low conductivity water, 18 M
, was processed in a Barnstead deionizing unit. Microscopes used were a Philips Tecnia 20 at 200 keV for TEM, and a Digital Instruments Nanoscope III

418 operating in the tapping mode for AFM. Particle size in the AFM micrograph is obtained from a section analysis that gives the profile of the surface and thus the height of the particles. The lateral dimensions are not a reliable measure of the particle because of convolution effects stemming from the finite size of the microscope tip. X-ray diffractograms were collected on a Philips XRG 3100 wide-angle X-ray diffraction system. Finally, mass spectra were obtained to measure the particle weight that can be used to estimate the size of the nanoparticle. In the past, it has been very difficult to measure particle weights higher than 100,000 Da due to the ultra-low vapor pressure and detection efficiency. These problems were overcome through the contributions of Hillenkamp and colleagues (Karas et al., 1987) who developed matrix-assisted laser desorption ionization (MALDI) to successfully measure molecular weight of large biomolecules. Typically, the sample is prepared by mixing solutions of large biomolecules with relatively small organic molecules; subsequently, the sample is dried and crystallized on a metal substrate. The molar ratio of the large molecules to small molecules is typically less than 1%. Thus, the small molecule can serve as matrix to surround large biomolecules. A laser beam with the wavelength selected for strong absorption by the small matrix molecules but not the larger ones is used to desorb the sample. The small matrix molecules absorb the laser photons and are vaporized. During the vaporization process, some of the large molecules are carried out from the substrate by the matrix molecules without fragmentation. Due to the collision process between desorbed matrix ions and large biomolecules, protonation and deprotonation processes occur to produce biomolecular ions, which can be subsequently mass selected and detected by a mass spectrometer. The efficiency of protonation and deprotonation process is a strong function of the matrix compound and the type of biomolecule. The selection of an appropriate matrix compound is very critical for the success of MALDI in obtaining reliable mass spectra. For most mass spectra, the charged ions are typically detected by a charged particle detector such as an electron multiplier, a channeltron or a microchannel plate. Secondary electrons are produced when the surfaces of this charged particle detector are hit by the charged particles. However, the efficiency of the ejection of secondary electrons strongly depends on the velocity of the charged particles. For large molecules, with energies less than 30 keV, the secondary electron ejection efficiency is very low. Thus, the overall detection efficiency for a mass spectrometer

to detect large molecules is quite low. Measurements of nanoparticles are subject to the same constraint. Recently, MALDI has also been applied to measure the weight of nanoparticles. However, it has been limited to very small nanoparticles with particle weight typically less than 100,000 Da. In this work, we use specific matrix for nanoparticles to succeed in measuring weight of nanoparticles higher than 200,000 Da. Gold nanoclusters of a nominal 2 nm diameter were prepared following the procedure described by Duff et al. (1993), and clusters of nominal 3.5 nm diameter, following the procedure of Chen and Kimura (1999). The smaller clusters are coated with a phosphine monolayer while the larger ones, are coated with thioctic acid which is a disulfide previously used in our work (Maya et al., 2000). Mass spectrometric analysis was performed by a linear time-of-flight mass spectrometer. (Voyager II, PerSeptive Biosystem with modification for pulsed extraction) A nitrogen laser was used for the desorption and ionization of the samples. The laser fluence used was between 20 and 100 mJ/cm2 . The laser fluence was always adjusted to optimize the nanoparticle ion signals. The pressure of the chamber of time-of-flight mass spectrometer was typically at 1.5×10−7 Torr. The polarity of the sample-biased voltage can be changed for both positive and negative ion measurements. The ion energy was set ∼26 keV. A microchannel plate was used as detector of particle ions. Several different compounds including 2,5-dihydroxybenzoic acid (2,5 DHB), sinapinic acid, 2,4,6-trihydroxyacetophenone (THAP), α-cyano-4-hydroxy cinnamic acid (α CHCA), 2-(4-hydroxy-phenylazo)-benzoic acid (HABA) and 3-hydroxypicolinic acid (3 HPA) have been tested for obtaining mass spectra of nanoparticles. It was found that 3 HPA was the best matrix among them. The ratio of particles to matrix was varied to optimize the mass spectra signals. The typical practice was to use 1 to 0.01% of particle solution to mix with various quantity of saturated 3 HPA solution for MALDI sample preparation. The samples were then dried by room air from a fan. Mass spectra were obtained by averaging of 8–128 laser shots to reduce background.

Results and discussion A batch of gold clusters produced as a phosphinepassivated sol in an aqueous medium was purified through dialysis to eliminate sodium chloride impurity formed as a by-product from the synthesis. The purified

419 product was examined by XRD, TEM, AFM and subjected to mass spectrometric analysis. The results of the examination are given in Figures 1–4, respectively. Particle size as determined by XRD showed an average

Figure 1. X-ray diffraction pattern of phosphine-coated gold nanoclusters. Line broadening is so pronounced that the 111 and 200 reflections are not resolved.

Figure 3. Atomic force micrograph of gold nanoclusters passivated with phosphine spin coated on substrate. Micrograph is 1 µm on the side.

Figure 2. Transmission electron micrograph of an evaporated sample of phosphine-coated gold nanoclusters. Size bar corresponds to 10 nm.

crystallite size of 2.0 nm while TEM showed an average of 2.59 nm and a standard deviation of 0.75. The AFM showed an average particle size of 1.88 nm with a standard deviation of 0.57. The mass spectrometric analysis showed two maxima with values of 56.121 kDa and 28.568 kDa corresponding to the single and double charged ions, respectively. The peak intensity in the spectrum for doubly charged ions is comparable to singly charged ions. It is mostly due to the higher velocity of doubly charged ions. The velocity of doubly charged ion is about 40% higher than singly charged ions. Since the secondary electron ejection efficiency is very low at this velocity range, the difference of detection efficiency for doubly charged ions compared to singly charged ion can compensate the population difference between singly charged ions and doubly charged ions. The spectrum in Figure 4 doesn’t necessarily indicate the population of doubly charged ions is about the same as singly charged ions.

420

Figure 5. X-ray diffraction pattern of gold nanoclusters coated with thioctic acid as the ammonium salt.

Figure 4. Mass spectrometer analysis of phosphine-coated gold nanoclusters.

The size of the gold core in the mass spectrometric analysis is derived after taking into consideration that the gold content of the passivated clusters is 80 wt% as determined on an independently prepared batch under the same conditions. The result of this calculation corresponds to a cluster with 228 atoms that would have a diameter of 1.95 nm, assuming to have the theoretical density of bulk gold. The agreement among all the techniques examined is reasonably good considering the limitations of each technique. In the case of TEM, as noted by Wilcoxon et al. (2000), determining the boundary of a cluster particularly under 2 nm is challenging and leads to an uncertainty of about 10–15%; in addition to this, aggregation is possible under the influence of the electron beam as well as in the process of drying the sample on the grid. In the case of AFM the measurements correspond to the gold core, apparently because of compressibility effects on the passivating coating. This has been noted by Brown and Hutchinson (1997), and in the analysis of DNA samples by Thundat et al. (1992). An additional sample was examined to confirm the validity of the mass spectrometric analysis using a material with clusters significantly larger than the previous one. In this case the sample was coated with thioctic acid prepared as the ammonium salt to avoid

the presence of sodium which under certain conditions interferes with the mass spectrometric analysis. It has been known that sodium ions can attach to biomolecules to cause poor mass resolution. For example, when a single sodium ion attaches to a protein with molecular weight of 100,000, the mass resolution (M/M) needs to be higher than 4500, which is extremely difficult for a simple linear time-of-flight mass spectrometer, to resolve these two peaks. When there are various ions attached to the parent ion, the mass resolution can become much lower. This was also found to apply, in our experience, to gold nanoparticles. Nordhoff et al. (1993), demonstrated that the use of an ammonium salt to replace sodium ions in solution improved the mass resolution. The XRD showed crystallites in the thioctic acid coated sample of 3.35 nm as an average size. This average is derived from estimates of line broadening for the 111 and 220 reflections, which give values of 3.5 and 3.2 nm, respectively. In this case, given the relatively larger crystallites, the 220 reflection at a 2-theta value of 64.6 is evident in the diffractogram (see Figure 5) and can be used to derive a crystallite size. The mass spectrometric analysis gave a value of 230 kDa (see Figure 6) which after considering the gold content of this material of 83.2 wt%, gives a value of 3.2 nm as the average for the diameter of the gold core of these nanoclusters. The synthetic procedure used to prepare the ammonium salt of the thioctic acid-coated clusters departed from the typical procedure that yields the sodium salt since sodium borohydride is the reducing agent in the preparation. Conversion to the ammonium salt in this study involved acid addition to the water-soluble

421

Figure 7. AFM micrograph, 1 µm on the side, of cluster aggregates from the thioctic acid-coated gold particles. Darker features are about 15–20 nm in diameter. Figure 6. Mass spectrometer analysis of thioctic acid-coated clusters.

Conclusions sodium form. This causes precipitation of the insoluble free thioctic acid-coated clusters. Centrifugation to isolate the solid and water washing was followed by the addition of enough ammonium hydroxide to solubilize the solid thus producing the typical deep red wine coloration of gold colloids. Attempts were made to examine the ammonium salt solution by TEM and AFM. In both cases individual particles could be found but a large proportion of the material was in the form of multicluster aggregates that precluded a reasonable assessment of particle size. This is illustrated by the AFM micrograph shown in Figure 7. The micrograph shows agglomerated particles to form clusters as big as 15 nm in diameter. Some improvement was noted after examining a sample of the solution that was subjected to sonication but the proportion of aggregates was still large and some small crystallites of about 1 nm in diameter became evident. Apparently aggregates are formed in the phase change involved in the precipitation of the free acid. Presumably the more energetic process employed in the mass spectrometric analysis that places the particles in the vapor phase is more than adequate in breaking the aggregates. The XRD analysis is not sensitive to the presence of aggregates so this technique and the mass analysis are in a relatively good agreement.

It is seen that mass spectrometric analyses of gold particles in the range of 50–250 kDa provide a reliable measure of particle size in aqueous sols of gold nanoclusters as compared with other techniques. Use of mass spectrometry involving MALDI, is also useful even in the presence of aggregates that would preclude the use of other conventional techniques such as TEM or AFM. Acknowledgements Research was sponsored partially by the Engineering Research Program of the Office of Basic Energy Sciences, U.S. DOE, and partially by the Laboratory Directed Research and Development Program of the Oak Ridge National Laboratory (ORNL), managed by UT-Batelle, LLC, for the U.S. DOE under Contract No. DE-AC05-00OR22725. Support for E.A.K. from the Division of Materials Sciences and Engineering, DOE. References Alvarez M.M., I. Vezmar & R.L. Whetten, 1998. J. Aerosol. Sci. 29, 115.

422 Brown L.O. & J.E. Hutchinson, 1997. J. Am. Chem. Soc. 119, 12384. Chen S. & K. Kimura, 1999. Langmuir 15, 1075. Duff D.G., A. Baiker & P.P. Edwards, 1993. Langmuir 9, 2301. Karas M., D. Beckmann, U. Bahr & F. Hillenkamp, 1987. Int. J. Mass Spectrom. Ion Processes 78, 53. Maya L., G. Muralidharan, T.G. Thundat & E.A. Kenik, 2000. Langmuir 16, 9151.

Nordhoff N., R. Cramer, M. Karas, F. Hillenkamp, F. Kirpekar, K. Kristensen & P. Roepstorff, 1993. Nucleic Acid Res. 21, 3347. Schaaff T.G., G. Knight, M.N. Shafigullin, R.F. Borkman & R.L. Whetten, 1998. J. Phys. Chem. B. 102, 10643. Whetten R.L., J.T. Khoury, M.M. Alvarez, S. Murthy, I. Vezmar, Z.L. Wang, P.W. Stephens, C.L. Cleveland, W.D. Luedke & U. Landman, 1996. Adv. Mater. 8, 428. Wilcoxon J.P., J.E. Martin & P. Provencio, 2000. Langmuir 16, 9912.