J. Phys. Chem. C 2007, 111, 2409-2415
2409
Visible Laser Desorption/Ionization Mass Spectrometry Using Gold Nanorods Lee C. Chen,† Takamasa Ueda,† Michihiro Sagisaka,† Hirokazu Hori,*,† and Kenzo Hiraoka‡ Interdisciplinary Graduate School of Medical and Engineering, and Clean Energy Research Center, UniVersity of Yamanashi, 4-3-11 Takeda, Kofu 400-8511, Japan ReceiVed: August 26, 2006; In Final Form: December 9, 2006
In this report, we demonstrate the novel application of gold nanorods in the laser desorption/ionization mass spectrometry. The gold nanorod substrate was fabricated by electrodepositing the gold into the pores of the porous alumina template. The embedded nanorods with diameter of ∼15 nm were exposed to the surface by partial removal of the alumina template. The vertically aligned gold nanorods were irradiated by a frequencydoubled/-tripled Nd:YAG laser and the desorption/ionization was found to be favored by the use of 532 nm visible laser, which is in the range of the localized surface plasmon resonance. The present technique offers a potential analytical method for the low-molecular-weight analytes which are rather difficult to handle in organic matrix-assisted laser desorption/ionization mass spectrometry.
Introduction Matrix-assisted laser desorption/ionization (MALDI) mass spectrometry is a very effective and soft method in obtaining mass spectra for synthetic and biological samples such as peptides and proteins with less molecular fragmentation.1,2 Depending on the matrixes, laser wavelengths ranging from ultraviolet (UV) to infrared (IR) have been employed. The nitrogen lasers (337 nm) or the frequency-tripled Nd:YAG lasers (355 nm) are employed for UV-absorbing matrix, while the Er: YAG lasers (2.94 µm) and TEA-CO2 lasers (10.6 µm) are commonly used in the IR-MALDI. For the popular organic matrixes such as R-cyano-4-hydroxycinnamic acid (R-CHCA) and 2,5-dihydroxybenzoic acid (DHB), the strong matrix-related ion signals also appear in the low-mass region of the mass spectrum. Although the large biomolecules (>700 Da) are usually unaffected, the detection for the analytes of low molecular weight is rather difficult due to the matrix interferences. Direct laser desorption/ionization (LDI) without adding any particular matrix becomes an alternative to MALDI for small analytes as the interference with the matrix ions can be avoided. However, direct laser desorption/ionization is a “harder” method and the direct absorption of the intense UV light by the molecules may lead to molecular fragmentation. In LDI, the choice of substrate, its surface morphology, and the sample preparation/deposition method become the options to enhance and “soften” the desorption/ionization. UV laser desorption/ionization without using organic matrix has been extensively studied on various substrates, for example, cobalt,1 graphite,3 silicon,4-7 metal-coated porous alumina,8 and titania sol-gel.9 Silicon substrates include porous silicon,4,5 silicon nanowires,6 and ordered silicon nanocavity.7 Matrix-less IR laser desorption/ionization was also reported on a flat silicon surface.10 Despite some attempts that used graphite substrate,11 rhodamine 6G, and Neutral Red as matrixes,12,13 there has been little progress in the development of visible laser mass spectrometry. * To whom correspondence should be addressed. Tel.: 055-220-8676. Fax: 055-220-8682. E-mail:
[email protected]. † Interdisciplinary Graduate School of Medical and Engineering. ‡ Clean Energy Research Center.
When the metallic nanostructure is irradiated with the laser of certain wavelength at the appropriate polarization, the surface plasmon resonance will be excited, of which all the free electrons within the conduction band will oscillate in-phase.14-16 For the metal such as silver and gold, the surface plasmon resonance takes place at the visible light and it leads to a huge concentration of optical near-field at a small volume which is usually called the “hot spot”. The optical near-field at the hot spot is so strong that even a single molecule bound to the metallic nanoparticles can produce a measurable signal of the surfaceenhanced Raman spectroscopy.17,18 Due to the strong electron oscillation, the surface plasmon induced nonthermal desorption had been reported for several metals. By irradiation of the roughened silver surface with visible or near ultraviolet laser, two prominent peaks were observed in the kinetic-energy distribution of Ag+ ions produced from the surface.19 In another experiment where the surface plasmon was coupled using the attenuated-total-reflection method, similar results were also obtained for the metal atoms desorbed from the Au, Al, and Ag films.20 While the lower energy peak was generally referred to as thermal peak, the authors attributed the peak of the higher kinetic energy to the nonthermal electronic process. Nonthermal visible laser desorption of alkali atoms were also reported for sodium particles and sodium film.21,22 Because the surface plasmon resonance is strongly damped, the local heating which is due to the joule losses on the metal surface could take place. For the gold nanoparticles which have small heat capacity, the heat transfer was estimated to be in the picosecond time scale, and the high lattice temperature can be reached rapidly.16 Although the gold nanoparticles, e.g., gold colloids, have a strong optical extinction at ∼520 nm due to the surface plasmon resonance,16 the colloids aggregate when mixed with certain analytes and the plasmon frequency red-shifts drastically to a longer wavelength. This poses a problem to a nontunable laser for excitation. When used as the matrix, the chemical contents in the colloids buffer solution can also contribute to the background noise. In this work, we demonstrate the visible laser desorption/ ionization of biomolecules, for example, peptides and neutral carbohydrates, from the gold nanorod substrate using a frequency-
10.1021/jp065540i CCC: $37.00 © 2007 American Chemical Society Published on Web 01/20/2007
2410 J. Phys. Chem. C, Vol. 111, No. 6, 2007
Chen et al.
Figure 1. Fabrication processes of the gold nanorod substrate: (a) Porous alumina (Al2O3) template fabricated by anodic oxidation using 20 wt % sulfuric acid. (b) Pulsed electrodeposition of gold within the pores of the porous alumina. (c) Partial removal of the alumina template using 8% v/v aqueous solution of phosphoric acid. (d) SEM inspection of the fabricated gold nanorods (plan view). (e) The gold nanorods viewed at 45° tilt angle.
doubled Nd:YAG laser at 532 nm. The vertically aligned gold nanorod arrays were fabricated by filling the pores of a porous alumina template with gold using the electrodeposition method.23,24 The diameter of the gold nanorods follows the pores of the alumina template and the aspect ratio can be controlled through the deposition time. The porous alumina template with ordered nanopore arrays can be easily fabricated using anodic oxidation, and the pore diameter can be tuned from ∼10 nm to >100 nm depending on the electrolyte and the anodization voltage.25,26 Compared to the colloids aggregates and the electrochemically roughened surface, which are of random structure, the surface morphology of the gold nanorod arrays can be better controlled and fabricated reproducibly. Experimental Section Fabrication of Gold Nanorods. The gold nanorod substrate can be made from aluminum sheet or the aluminum film coated on the glass/silicon as the starting material. A schematic describing the fabrication processes is depicted in Figure 1. Sulfuric acid (∼20 wt %) was used as the electrolyte for the anodic oxidation of aluminum. Platinum counter electrode was used in the anodic oxidation as well as in the electrodeposition of gold. The aluminum was oxidized at the anodization voltage of ∼12 V for 5-10 min to form porous alumina. The pore diameter was in the range of ∼15 nm. As shown in Figure 1a, a thin barrier layer was also formed at the bottom of the pores. Although it was possible to remove the barrier layer using an etching method, at ∼12 V anodization voltage, the barrier was thin enough that the gold could be electrodeposited directly within the pores at moderate voltage. The aqueous solution of
40 mM chlorauric acid (HAuCl4) was used as the working electrolyte. The pulsed electrodeposition was conducted at 12 V with the duty cycle of ∼1/10 and pulse repetition rate of 1 s-1. After several minutes, the deposited surface became ruby red in color, and the grown nanorods were embedded inside the porous alumina as illustrated in Figure 1b. To trap the analyte molecules on the gold surface, the embedded nanorods were partially exposed to the surface (Figure 1c) by chemical etching of the alumina template. The etching was done by using aqueous solution of ∼8% v/v phosphoric acid. Figure 1d and 1e show the scanning electron micrographs (SEM) of the fabricated gold nanorods. The substrate was viewed at a 45° tilt angle in Figure 1e. The diameter of the gold nanorods was ∼15 nm and the lengths could be fabricated in the range of ∼50 to ∼200 nm depending on the deposition condition. All the rods were oriented in the same direction with their major (long) axis perpendicular to the surface. Thus, the oscillation direction of the localized plasmon resonance could be selectively excited by the TM- or TE-polarized light. The optical electric field was perpendicular to the substrate surface (i.e., along the major axis of the gold nanorods) when it was TM-polarized, and vice versa when it was TE-polarized. Unless otherwise stated, the standard substrate used in this study consisted of gold nanorods with length of ∼100 nm and diameter of ∼15 nm. Sample Preparation. All chemicals and analytes were obtained commercially and used without further purification. Bovine insulin was prepared in the aqueous solution of 1% trifluoroacetic acid (TFA). Lys-Lys, bradykinin, and melittin were dissolved in water. Lactose was prepared in the aqueous
Visible LDI Mass Spectrometry Using Gold Nanorods
Figure 2. Normalized specular reflectance of the gold nanorod substrate irradiated by (a) TE-polarized and (b) TM-polarized laser at 45° incidence angle.
solution of sodium chloride (∼10 ppm) to promote cationization. The citric buffer was prepared by mixing the aqueous solution of 10 mM citric acid with the diammonium citrate (10 mM) at the ratio of 1/2. Working stocks containing the analyte were prepared in the concentration of 1-10 pmol/µL. About 0.2-1 µL of the working stock was pipetted onto the gold nanorod substrate and the droplet was gently dried using a warm air blower. When the droplet was dried, the gold nanorod substrate loaded with the analytes was transferred into the vacuum chamber of the time-of-flight mass spectrometer. Time-of-Flight Mass Spectrometer. The laser desorption/ ionization experiment was performed with a 2.5 m time-of-flight mass spectrometer (JEOL 2500) with delayed ion extraction. The instrument can be operated in linear or reflectron mode. The acceleration voltage for ions was 20 kV. The vacuum pressures in the ion source and the detector were 7.5 × 10-5 and 5 × 10-7 Torr, respectively. The primary laser source for
J. Phys. Chem. C, Vol. 111, No. 6, 2007 2411 the desorption/ionization experiment was a linearly polarized frequency-doubled Nd:YAG laser which was operated at 532 nm wavelength and pulse width of 4 ns. The optical polarization was adjusted using a half-waveplate. The gold nanorod substrate was attached to a modified target plate and was irradiated by the laser at 60° to the surface normal. For comparison, the frequency-tripled output of the Nd:YAG laser (355 nm wavelength) was also used to investigate the wavelength dependence. The laser spot size on the target substrate was about 0.5 mm in diameter. The relative optical intensity of the 532 nm visible laser and the 355 nm UV laser were measured using a highspeed photodetector in arbitrary units. In the following section the relative optical intensity for the 532 nm laser will be referred as “Vis. Intensity” and “UV Intensity” for the 355 nm laser. The typical laser fluence for the gold nanorods was estimated to be around few 10 mJ/cm2. The mass spectra shown were acquired from the accumulation of 40-60 single-shot mass spectra. Results and Discussion The specular reflectance of the gold nanorods measured at 45° incidence angle is depicted in Figure 2. The normalized reflectance in Figures 2a and 2b show the surface plasmon absorption at ∼520 nm for both TM- and TE-polarized light. This visible absorption band coincides spectrally with the surface plasmon of the spherical nanoparticles and can be excited by a frequency-doubled Nd:YAG laser of 532 nm. This absorption band is also referred to as the transverse surface plasmon resonance of the gold nanorods.16,24 For TM polarization, there is also an increase in the near-infrared absorption which is likely due to the longitudinal surface plasmon resonance (free electrons oscillate along the major axis of the gold nanorods). At the wavelength of 532 nm, the dependence of optical absorption on the laser polarization becomes more apparent with the increase of incidence angle. Figure 3a shows the normalized
Figure 3. (a) Specular reflectance of gold nanorods (b), aluminum (1), and gold film (9), measured at ∼60° incidence angle using 532 nm laser with different polarization angle. The reflected intensity was normalized to that of TE polarization. (b) and (c) show the mass spectrum of lactose (342 Da) obtained from gold nanorods using TE-polarized (b) and TM-polarized laser (c). Both are of the same laser fluence (vis. intensity ) 60). (d) shows the TM-polarized LDI of lactose with higher laser fluence (vis. intensity ) 75).
2412 J. Phys. Chem. C, Vol. 111, No. 6, 2007
Chen et al.
Figure 4. Mass spectra of 5 pmol melittin (2847 Da) acquired from the gold nanorods of different etching times. To denotes the time at which the chemical etching of the alumina template had just reached the gold nanorods. The nanorods started to protrude from the template after To. For this substrate, the optimum condition was achieved after ∼13 min from To. The laser polarization was TE (vis. intensity ) 85).
specular reflectance of the gold nanorod substrate measured using the 532 nm laser with difference optical polarization. The measurement was taken at a 60° incidence angle which was close to our LDI experimental condition. The reflectance of flat aluminum and gold film are included for comparison. The reflected optical intensity was normalized to that of TE polarization. Unlike the flat metal surface in which the reflectivity is minimum for TM-polarized light,27 the gold nanorod substrate has a higher optical absorption for the 532 nm laser at TE polarization. Irradiating the nanorods with the optimized laser polarization, and hence lower laser fluence, was found to be advantageous in reducing gold ions and the background noise. The desorption/ionization of lactose (10 pmol) from the gold nanorods using the 532 nm laser with TE and TM polarization are shown in Figures 3b and 3c, respectively. Both mass spectra were obtained with the same laser fluence. The analyte ion signal was maximum when the laser was TE-polarized. The mass spectrum acquired by the TM-polarized laser with slightly higher laser fluence is shown in Figure 3d. Although the sodiated lactose ion signal in Figure 3d was comparable to that of TE polarization in Figure 3b, the Au+ ion signal was more intense and Au3+ started to appear as well as the background noise. The chemical etching of the alumina template was also a key process in obtaining good mass spectrum of the deposited analyte. The desorption/ionization efficiency was found to be quite dependent on the etching time. Figure 4 shows the mass
spectra of 5 pmol melittin (2847 Da) acquired from the gold nanorods prepared with different etching times. In Figure 4, To denotes the time at which the chemical etching of the alumina template has just reached the gold nanorods. The SEM images of the gold nanorod substrate taken after the LDI experiment are shown in Figure 5. The analyte ion signals increased as the gold nanorods started to emerge from the template. The optimum condition was achieved at about To + 13 min when the ∼100 nm nanorods were almost completely released from the template. Further etching of the alumina template caused the nanorods to topple (see Figure 5c) and the ion signal started to diminish. Excessive etching detached some of the nanorods from the alumina template (Figure 5d) and it became difficult to observe ion signals. The laser wavelength dependence was investigated by using the frequency-tripled output of the Nd:YAG laser at 355 nm, which is away from the surface plasmon resonance. The mass spectra for ∼6 pmol Lys-Lys (274 Da) acquired from the gold nanorods and stainless steel target plate by using 532 and 355 nm laser are shown in Figure 6. At high laser fluence, Lys-Lys ion signals were also observable from the stainless steel using both 532 and 355 nm lasers (Figure 6c and 6d), but the most intense and homogeneous ion signals were produced from the gold nanorods using 532 nm laser (Figure 6a). Besides Au+, some background ions (
