Laser-Driven Acoustic Desorption of Organic ... - ACS Publications

various organic dyes (rhodamine B, fluorescein, an- thracene, coumarin, BBQ), have demonstrated detection of intact parent molecules of the analyte ev...
0 downloads 0 Views 191KB Size
Anal. Chem. 2007, 79, 8232-8241

Laser-Driven Acoustic Desorption of Organic Molecules from Back-Irradiated Solid Foils Alexander V. Zinovev,*,† Igor V. Veryovkin,† Jerry F. Moore,‡ and Michael J. Pellin†

Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439, and MassThink, 2308 Hartford Court, Naperville, Illinois 60565

Laser-induced acoustic desorption (LIAD) from thin metal foils is a promising technique for gentle and efficient volatilization of intact organic molecules from surfaces of solid substrates. Using the single-photon ionization method combined with time-of-flight mass spectrometry, we have examined the neutral component of the desorbed flux in LIAD and compared it to that from direct laser desorption. These basic studies of LIAD, conducted for molecules of various organic dyes (rhodamine B, fluorescein, anthracene, coumarin, BBQ), have demonstrated detection of intact parent molecules of the analyte even at its surface concentrations corresponding to a submonolayer coating. In some cases (rhodamine B, fluorescein, BBQ), the parent molecular ion peak was accompanied by a few fragmentation peaks of comparable intensity, whereas for others, only peaks corresponding to intact parent molecules were detected. At all measured desorbing laser intensities (from 100 to 500 MW/cm2), the total amount of desorbed parent molecules depended exponentially on the laser intensity. Translational velocities of the desorbed intact molecules, determined for the first time in this work, were of the order of hundreds of meters per second, less than what has been observed in our experiments for direct laser desorption, but substantially greater than the possible perpendicular velocity of the substrate foil surface due to laser-generated acoustic waves. Moreover, these velocities did not depend on the desorbing laser intensity, which implies the presence of a more sophisticated mechanism of energy transfer than direct mechanical or thermal coupling between the laser pulse and the adsorbed molecules. Also, the total flux of desorbed intact molecules as a function of the total number of desorbing laser pulses, striking the same point on the target, decayed following a power law rather than an exponential function, as would have been predicted by the shake-off model. To summarize, the results of our experiments indicate that the LIAD phenomenon cannot be described in terms of simple mechanical shake-off or direct laser desorption. Rather, they suggest that multistep energytransfer processes are involved. Two possible (and not mutually exclusive) qualitative mechanisms of LIAD that are based on formation of nonequilibrium energy states * To whom correspondence should be addressed. E-mail: [email protected]. † Argonne National Laboratory. ‡ MassThink.

8232 Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

in the adsorbate-substrate system are proposed and discussed. Identifying efficient methods for the nondestructive production of ionized macromolecules in the gas phase for analysis by mass spectrometry (MS) is recognized as a crucial problem.1,2 There are also applications such as the growth of supramolecular nanostructures that can be based on vaporized molecules or ions.3 The widespread use of MALDI and electrospray ionization demonstrates the importance of sample ionization techniques in MS analysis and explains why new methods for the generation of vibrationally cold molecular ions (to allow controlled fragmentation) in the gas phase remain a topic of substantial interest in the mass spectrometry community.4 Laser-induced acoustic desorption (LIAD) of molecules is a molecular volatilization method that does not require direct exposure of the sample to intense laser light5-7 and could be useful in cases where the sample is light sensitive, in contrast to other laser desorption methods such as MALDI. The typical preparation for LIAD begins by depositing the analyte on one side of a 5-25µm metal foil (e.g., Cu, Ti, Au, and Fe). The analyte is then volatilized by subjecting the opposite side of the foil to a driving laser pulse with the intensities of 108-109 W/cm2. This process introduces the analyte into the gas phase without subjecting the sample either to light or a significant temperature rise. The first experimental observation of LIAD was demonstrated by Lindner and Seydel,5 who suggested that the observed ion emission resulted from a laser-generated acoustic wave. Renewed interest in this problem arose after the work of Chen et al.,6 which demonstrated the emission of negative hydrogen ions and electrons. The direct use of LIAD in practical mass spectrometry has been demonstrated in a series of recent studies. The results of these and earlier studies7-13 revealed that most deposited (1) Nelson, R. W.; Rainbow, M. J.; Lohr, D. E.; Williams, P. Science 1989, 246 (4937), 1585-1587. (2) Berry, J. I.; Sun, S. X.; Dou, Y. S.; Wucher, A.; Winograd, N. Anal. Chem. 2003, 75 (19), 5146-5151. (3) Alivisatos, A. P.; Barbara, P. F.; Castleman, A. W.; Chang, J.; Dixon, D. A.; Klein, M. L.; McLendon, G. L.; Miller, J. S.; Ratner, M. A.; Rossky, P. J.; Stupp, S. I.; Thompson, M. E. Adv. Mater. 1998, 10 (16), 1297-1336. (4) Merchant, M.; Weinberger, S. R. Electrophoresis 2000, 21 (6), 1164-1177. (5) Lindner, B.; Seydel, U. Anal. Chem. 1985, 57 (4), 895-899. (6) Golovlev, V. V.; Allman, S. L.; Garrett, W. R.; Chen, C. H. Appl. Phys. Lett. 1997, 71 (6), 852-854. (7) Perez, J.; Petzold, C. J.; Watkins, M. A.; Vaughn, W. E.; Kenttamaa, H. I. J. Am. Soc. Mass Spectrom. 1999, 10 (11), 1105-110. (8) Perez, J.; Ramirez-Arizmendi, L. E.; Petzold, C. J.; Guler, L. P.; Nelson, E. D.; Kenttamaa, H. I. Int. J. Mass Spectrom. 2000, 198 (3), 173-188. 10.1021/ac070584o CCC: $37.00

© 2007 American Chemical Society Published on Web 10/04/2007

organic or biological molecules, as well as bioparticles,14 are efficiently desorbed from the foil surface in the form of neutral, nonfragmented molecules. The velocities of the desorbed molecules, estimated in the work of Kenttamaa et al.,7,8 were much lower than those for direct laser desorption (i.e., when the driving laser hits the same side of the substrate where molecules are deposited). To date, the accepted explanation of the LIAD phenomenon was based on a phenomenological shake-off model, which presumed the following: (1) the generation of acoustic waves in the laser irradiated foil and (2) the subsequent impulse transfer from the generated acoustic waves to the adsorbed molecules.12,15 We note that these studies did not identify a mechanism for the impulse transfer from the vibrating foil surface to the adsorbed molecules. We believe this LIAD mechanism is unlikely since it implies that the molecules are subjected to the same acceleration as the rest of the foil material and that, as the foil motion changes direction during oscillation, the molecules subjected to inertial forces will continue moving in the initial direction resulting in their escape (desorption). In this case, one can easily estimate the minimal surface velocity required for the molecule to overcome the adsorption forces

v g x2Eb/m

(1)

where Eb is the molecule binding energy and m is its mass. Taking into account that typical binding energies of the physically adsorbed molecules are ∼0.05-0.5 eV (see, for example, ref 16), one can calculate from eq 1 that minimal foil surface velocities needed are in the range of (0.1-0.5) × 103 m/s for the molecules with masses of 200 - 500 Da to be desorbed. These values approach the speed of sound, for the solid, which was used in ref 6 as a proof of the validity of the LIAD model mentioned above. In fact, the velocities of acoustic waves propagating in solids are always much higher than the corresponding velocities of the mass motion (see ref 17). Thus, the realization of condition 1 in the vibrating foils is very unlikely. Unfortunately, very few measurements of the surface velocities under conditions similar to those in LIAD experiments have been published. Most of the experimental work studying ultrasound generation in metals due to laser irradiation18 were performed on thick samples, making a direct comparison to LIAD conditions (9) Reid, G. E.; Tichy, S. E.; Perez, J.; O’Hair, R. A. J.; Simpson, R. J.; Kenttamaa, H. I. J. Am. Chem. Soc. 2001, 123 (6), 1184-1192. (10) Campbell, J. L.; Crawford, K. E.; Kenttamaa, H. I. Anal. Chem. 2004, 76 (4), 959-963. (11) Liu, J. A.; Petzold, C. J.; Ramirez-Arizmendi, L. E.; Perez, J.; Kenttamaa, H. J. Am. Chem. Soc. 2005, 127 (37), 12758-12759. (12) Crawford, K. E.; Campbell, J. L.; Fiddler, M. N.; Duan, P.; Qian, K.; Gorbaty, M. L.; Kenttamaa, H. I. Anal. Chem. 2005, 77 (24), 7916-7923. (13) Shea, R. C.; Petzold, C. J.; Campbell, J. L.; Li, S.; Aaserud, D. J.; Kenttamaa, H. I. Anal. Chem. 2006, 78 (17), 6133-6139. (14) Peng, W. P.; Yang, Y. C.; Kang, M. W.; Tzeng, Y. K.; Nie, Z. X.; Chang, H. C.; Chang, W.; Chen, C. H. Angew. Chem., Int. Ed. 2006, 45 (9), 14231426. (15) Lindner, B. Int. J. Mass Spectrom. 1991, 103 (2-3), 203-218. (16) Adamson, A. W.; Gast, A. P. Physical Chemistry of Surface; John Wiley &Sons, Inc.: New York, 1997. (17) Landau, L. D.; Lifshits, E. M. Fluid mechanics; Pergamon: New York, 1987. (18) Hutchins, D. A. Can. J. Phys. 1985, 64, 1247-1263.

difficult. In the very few recent publications reporting experiments with foils,19 the measured velocities of the foil surface motion did not exceed a few meters per seconde. This serious discrepancy between the measured surface velocity and estimates of the required shake-off velocity derived from eq 1 means that the observed molecular desorption must be driven by another mechanism. Moreover, in some experiments on LIAD, the emission of charged particles was observed: positive ions in refs 5 and 15 and electrons and negative ions in ref 6. Any complete explanation of the LIAD mechanism must account for electronic excitations leading to charged particle formation. In summary, despite the growing interest in the LIAD phenomenon and its potential to provide a basis for development of new analytical methods and technologies for materials synthesis, there is no satisfactory mechanism explaining the underlying phenomenon. This lack of understanding limits further development and application of this promising gentle molecular volatilization method. In the present work, we report the results of our studies of the fundamental characteristics of LIAD. EXPERIMENTS Mass Spectrometer and Laser System. Because the dominant fraction of the desorbed flux in LIAD is neutral molecules,7-12 it is very important to select an appropriate ionization method for the molecules as well as the type of mass analysis technique.20 Single-photon ionization (SPI) is well suited for characterization of this phenomenon21 because of its ability to efficiently ionize the desorbing flux with minimal fragmentation. SPI occurs following absorption of a single photon whose energy exceeds the ionization potential (IP) of the molecule of interest, creating a cation. For many molecules, particularly those with aromatic rings to stabilize the cation, fragmentation from photoionization is minimized, and thus, the state of the initial LIAD flux can in principle be revealed.22 Currently, the shortest wavelength of commercially available energetic lasers suitable for SPI is 157 nm (F2 laser), which corresponds to the photon energy of 7.9 eV. This energy limits the range of species that can be ionized by SPI to atoms and molecules with IPs less than 7.9 eV. Therefore, organic dyes were chosen as analytes for this study due to their low IPs, ability to form stable cations, and high photoabsorption coefficients. We selected dyes that have IPs in the range from 5 to 7 eV and can be easily ionized by the F2 laser radiation. A time-of-flight mass spectrometer (TOF MS), with a combined LIAD/SPI ion source,23 was employed in our studies of the LIAD phenomenon. The experiments were conducted under ultrahighvacuum conditions, with the residual gas pressure in the sample chamber less than 3 × 10-7 Pa. The schematic drawing of the target assembly in our instrument is shown in Figure 1. The sample was mounted on one of six sample holders that were supported by a hexagonal carousel. This carousel was driven by (19) Hebert, H.; Vidal, F.; Martin, F.; Kieffer, J. C.; Nadeau, A.; Johnston, T. W.; Blouin, A.; Moreau, A.; Monchalin, J. P. J. Appl. Phys. 2005, 98 (3). (20) Levis, R. J. Annu. Rev. Phys. Chem. 1994, 45, 483-518. (21) Pellin, M. J.; Calaway, W. F.; Veryovkin, I. V. In ToF-SIMS: Surface Analysis by Mass Spectrometry; Vickerman, J., Briggs, D., Eds.; Surface Spectra Ltd. and IM Publications: 2001; p 375. (22) Lipson, R. H.; Shi, Y. J. Ultraviolet Spectroscopy and UV lasers; Marcel Deccer, Inc.: New York, 2002; p 570. (23) Veryovkin, I. V.; Calaway, W. F.; Moore, J. F.; Pellin, M. J.; Burnett, D. S. Nucl. Instrum. Methods Phys. Res., Sect. B 2004, 219-220, 473.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8233

Figure 1. Schematic drawing of the experimental setup.

an ultrahigh-vacuum compatible motion stage (ALIO Industries) equipped with four piezoelectric motors (Nanomotion Inc.) that provided four independent motions (translations in three axes and a rotation) with closed-loop precision of better than 50 nm. The sample holders were secured on the carousel via three 30-mmlong alumina ceramic insulators and connected (using vacuum feedthrough) to a high-voltage pulser unit, which provided voltages necessary for the operating of TOF MS instrument. Each new sample was inserted into the UHV chamber through a vacuum load-lock and was fixed on the sample holder using a sample-transfer system with clips (Thermionics STLC-style). Using the motion stage, the sample was then positioned in the focal plane of the TOF MS source optics. The ion optics and operational principles of our instrument are described in more detail elsewhere.23 A dielectric mirror with 98% reflection at 248 nm was mounted in the center of the carousel, in order to deliver the laser beam to the back side of the sample. Note that we will use the convention that the front of the sample is the side facing the ion source and TOF, while the back side is the opposite. For desorption, an excimer KrF laser with wavelength 248 nm (EX10/ 300 GAM Laser Inc.) was used. The output energy of the laser pulse could be varied between 0.5 and 6 mJ by adjusting the laser discharge voltage and by an additional attenuation with a set of neutral optical filters. The energy of laser beam was measured by an Rm-3700 universal radiometer (Laser Probe Inc.) using 8% of the beam as picked off by a thin quartz plate. The driving laser beam was focused on the target back surface into a spot of rectangular shape ∼200 × 800 µm2 by the fused-silica lens with focusing distance of 500 mm. The laser pulse duration was 7 ns, producing a peak power density on the irradiated surface ranging from 50 to 500 MW/cm2. These laser intensities are close to those used in recent LIAD experiments8-10,12,13 taking into account that the reflection coefficient in the UV is normally less than it is at visible wavelengths. Postionization of the desorbed molecules was performed with an F2 laser (EX100F/1000, GAM Laser Inc.), with an output energy of 2 mJ/pulse and a pulse duration of 10 ns. The F2 laser beam was focused just above the front target surface, with a waist of 400 × 2000 µm2, using of a combination of MgF2 spherical and cylindrical lenses. The F2 laser radiation power density in the focal plane was ∼10 MW/cm2, which assured the saturation of the photoionization process for the investigated molecules (as verified by a laser power study). 8234

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

For comparison with LIAD, direct laser desorption (LD) mass spectra were measured for the same samples, also using the F2 laser for postionization. To this end, an N2 laser (337-nm wavelength, 100 µJ/pulse energy, and 7-ns pulse duration) was focused onto the target front surface using an in-vacuum Schwarzschild optical microscope.23 The beam spot size on the surface was ∼50 µm in diameter. The delay between the driving KrF (or N2) laser pulses and the ionizing F2 laser could be precisely controlled and varied from 0 to 1000 µs. The desorbed molecules that move away from the surface could therefore be ionized at a precisely defined moment in time and volume in space above the target after the desorption event, with the photoions then analyzed by the TOF MS. This approach allowed us to measure mass spectra for the (postionized) desorbed neutral molecules and determine their velocity distribution. Each mass spectrum was the sum of 128 individual acquired spectra. To prevent the rise of the average foil temperature due to adsorption of laser power, the repetition rate of the laser pulses was maintained at 8 Hz. Samples and Sample Preparation Procedure. Foils from different materials with different thicknesses were used in the experiments. Metal foils (Mo, Ta, Au, Ni) were used. All materials were commercially available foils (Alfa Aesar, 99.95% purity). Foil thicknesses ranged from 12.5 µm up to 50 µm. The foils were glued with UHV silver epoxy (Dynaloy Inc.) to the top of a hollow fused-silica cylinder with an outside diameter of 8 mm, a 0.5-mm wall thickness, and a 10-mm height. The epoxy was cured for 2 h at a temperature of 100° C in air. Due to the higher thermal expansion coefficients for metals, compared to quartz, the foil experienced radial tension (∼1 N) once the assembly was cooled to room temperature. Conductive contact between the foil sample and the target bias potential was assisted with a track of silver epoxy on the side of the quartz cylinder. Before applying the analyte to the top surface of the foil, each was cleaned in methanol-acetone solution (1:1) in an ultrasonic bath (10 min). Organic dyes rhodamine B, fluorescein, methylanthracene (MA), coumarin-522 (N-methyl-4-trifluoromethylpiperidino-[3,2-g]coumarin), and BBQ (4,4′′-bisbutyloctyloxy-p-quaterphenyl) were used as received (Eastman Kodak). The dyes were dissolved in methanol (for MA and BB, mixed xylenes were also used as solvent), and then the resulting solution (∼10-3 M) was used for sample preparation. A 1-µL sample of the analyte solution was pipetted onto the foil surface, and then the quartz cylinder-foil assembly was spun at 4500 rpm for 30 s to coat the analyte uniformly over the surface. During spin-coating, a significant part of the solution (90% or more) was taken off the surface, and surface concentrations of the analyte can be estimated to be less than 0.5 nM/cm2. After sample preparation, the foil was introduced into the instrument via the load-lock for analysis. RESULTS In good agreement with the previously published results on LIAD,7-12 we detected strong and stable desorption signals from foils with a thickness of 12.5 µm. Thicker foils (25 µm) produced relatively weak signals for the range of acoustic wave-driven laser intensities used in our experiments. For comprehensive experiments, a Ta foil with 12.5-µm thickness was chosen as optimal, not only because high desorption signals were detected from it

Figure 3. Plot of the LIAD maximal signal for different analytes versus desorbing laser intensity. Dotted lines represent the optimal power law fit of the exerimental data.

Figure 2. LIAD mass spectra for different samples. Spectra were obtained using Ta foil 12.5 mm thick as a substrate at medium laser intensities of ∼200 MW/cm2.

but also for its good mechanical strength, high melting point, and durability under powerful laser irradiation. The TOF mass spectra of different organic dye molecules desorbed from the front of the back-irradiated Ta foil surface and ionized by the 157-nm laser radiation are shown in Figure 2. This figure shows three major features: (1) all analytes display large parent molecular ion signals, (2) all spectra display a small number of peaks, with a few or none in the mass range below 100 Da, and (3) the number of fragment ion peaks is specific to each molecular analyte. In order to characterize the desorption process in terms of the corresponding molecular fragmentation, a parameter σ, can be defined as a ratio of the sum of intensities of the fragment peaks Af to the parent molecular peak intensity Ap, σ ) ∑ Af /Ap. As will be shown below, this parameter depends on laser intensities that drive the acoustic waves. For SPI, the parameter σ characterizes not only the peculiarities of the desorption process but it also generally depends on the photoionization cross section of the parent molecule and specifics of its photofragmentation such as possible decay channels and their activation energies.

To clarify the influence of the ionizing F2 laser intensity on fragmentation, a special series of control experiments were carried out. Keeping the intensity of the driving LIAD KrF laser at a constant level (250 MW/cm2), the parameter σ was measured for a wide range of F2 laser pulse energies (20-280 µJ/pulse). The signal power dependence was linear at low F2 laser energies, and saturated at ∼100 mJ/pulse. Interestingly, however, the value of the σ parameter did not vary beyond the experimental error within the entire range of the examined F2 laser energies. For example, in the case of rhodamine B, its value was 0.113 ( 0.02. Previous photoionization studies of rhodamine B have not detected ionization-induced fragmentation,24 which suggests that the observed fragmentation patterns in the LIAD mass spectra most likely originate from the initial desorption process rather than as a result of photofragmentation. An important characteristic of any desorption phenomenon is its yield. This is why in order to understand the basic processes driving the phenomenon, one has to identify external parameters that have the strongest effect on the desorption yield and then measure a dependence of the yield for each parameter. In the case of LIAD, the dependence of key peak intensities in the mass spectra and the σ parameter on the driving laser intensity appears to be the most important for understanding this phenomenon. The overall desorption yield for all studied analytes strongly increased with desorption laser intensity (within our experimental range), displaying for most peaks approximately exponential dependency. Figure 3 demonstrates this dependence clearly. Plotted on a semilogarithmic scale, these dependences appear linear, with different slopes for each analyte. The variation of the desorption laser power results in some redistribution of peak intensities in the mass spectra, as follows. When this laser power increased (within the range of our experiments outlined above), no additional fragmentation of analyte molecules was observed. Instead, relative abundances of different constituents of the mass spectrum changed noticeably. This is illustrated in Figure 4 where mass spectra are shown that have been measured at different driving laser powers for rhodamine B sample deposited on Ta foil. In some cases, chemicals corresponding to the fragment peaks can be readily identified. For (24) Lee, I.; Callcott, T. A.; Arakawa, E. T. Anal. Chem. 1992, 64 (5), 476-478.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8235

Figure 5. Dependence of parameter σ for rhodamine B desorption from Ta 12.5-µm-thick foil on desorbing laser intensity.

Figure 4. Mass spectra of rhodamine B at different desorbing laser intensities I. (a) 170, (b) 195, (c), and (d) 320 MW/cm2.

example, the peak at 399 Da in rhodamine B mass spectra corresponds to the decarbonylated form of rhodamine (RhB CO2)+. This particular peak is known as a dominant fragment in rhodamine B mass spectra obtained by electron impact ionization.25 In contrast to that, there was a publication in the literature, ref 26 that supposed this peak to be a native impurity in the rhodamine B chemical. In order to check what is the origin of the 399-Da peak, we conducted separate analyses of our rhodamine B solution in methanol using an HPLC (SCL-10A, Shimazdu) and MS/MS instruments (API 3000, Sciex). These control measurements have shown no compounds with masses equal to those of the fragment peaks we identified for LIAD (including the 399-Da peak), which suggests these peaks were not the native impurities present in the initial solution. On the other hand, one can see from Figure 4 that a change in the mutual relationship between peak intensities in the mass spectra is accompanied by a strong increase in signal intensities: one can see that the 399-Da peak grows faster than 443 Da when the driving laser power increases, and at the highest powers, the intensity of the 399-Da peak becomes comparable to that of the parent ion. Altogether, this indicates that the 399-Da peaks correspond to a chemical, which was not initially present in our rhodamine B but is formed due to fragmentation induced by the desorption process. For rhodamine B, the parameter σ grew linearly with KrF laser intensity (Figure 5), and a similar behavior was also detected for (25) NIST Chemistry WebBook. In http://webbook.nist.gov/chemistry. (26) Dunn, J. D.; Siegel, J. A.; Allison, J. J. Forensic Sci. 2003, 48 (3), JFS20022359_483.

8236 Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

Figure 6. Dependence of ζ parameter for BBQ desorption from Ta foil on desorbing laser energy.

fluorescein, although values of σ were different (σ ) 0.8 for fluorescein at 300 MW/cm2 KrF laser intensity). On the contrary, the same experiments conducted for BBQ surprisingly revealed that the parameter σ decreased with increasing desorption laser power, as shown in Figure 6. This discrepancy in σ parameter dependency (Figures 5 and 6) compared to the similarity observed for desorption yield dependencies (Figure 4) indicates that the relationship between these two observations is not a trivial one. This raises the question of whether the desorption and fragmentation phenomena are driven by the same fundamental process. We note that fragmentation is not intrinsic to LIAD. In our experiments, we observed the fragmentation for only three analytes out of five. The analysis of existing data from the literature shows that in some cases fragmentation was observed even with soft ionization of the desorbed molecules8,13,27 whereas in other cases fragmentation was very small or completely absent. A detailed study of fragmentation in LIAD has not been done yet and the current results, presented in Figures 4-6 are the first attempt to quantify this characteristic of LIAD. We have therefore observed that the LIAD signal increased exponentially with the driving laser power, while no additional fragments appeared in the mass spectra. In our opinion, this does indicate that molecular desorption is a much more significant and efficient channel for dissipation of laser energy compared to (27) Shea, R. C.; Habicht, S. C.; Vaughn, W. E.; Kenttamaa, H. I. Anal. Chem. 2007, 79 (7), 2688-2694.

Figure 7. Time-of-flight spectra of the rhodamine B peak (443 Da) obtained in LIAD and LD processes.

internal excitation of these analyte molecules. In this context, it becomes particularly interesting to determine what translational kinetic energies these desorbed molecules have, and if there is any correlation between this energy (as determined by measuring the velocity) and the desorbing laser power. To this end, we have measured and compared the signals of the analytes as a function of delay time between shots from the desorption laser and the photoionization laser. These delay curves obtained for both direct LD and LIAD processes are shown in Figure 7 for the parent ion of rhodamine B. It is immediately apparent that LIAD desorbed molecules have a longer delay time over a much broader range, which agrees with previous LIAD measurements.7,8 The immediate conclusion is that translational velocities of the desorbed molecules in LIAD are extremely small. From the data in Figure 7, it can be estimated that molecular velocities are distributed in the interval 2 - 40 m/s with a maximum of ∼10 m/s, corresponding to an average translational kinetic energy of 0.2-0.3 meV. This conclusion is valid only if the desorption pulse is very short compared to the time-of-flight of the molecules to and through the ionizing laser volume. This condition will be formalized below. Because the width of the ionizing laser beam in the normal direction to the target surface was 400 µm, only that portion of the molecular flux which was in the laser beam volume when the laser was fired could be ionized. This means that only those molecules desorbed from the surface at t0 with velocities in the interval vmin < v < vmax will be ionized, where the values of vmin and vmax can be found from the relations

vmax ) (d + δ)/(t0 + δt)

vmin ) d/(t0 + δt)

(2)

Here δ corresponds to the width of the ionizing laser beam, δt is the delay time between firing the ionizing laser and the desorption laser (the time interval between pulses that trigger F2 and KrF/ N2 lasers, respectively), and d is the distance between the surface and the closest edge of the F2 laser beam. This method is therefore a direct measurement of velocity and does not require any simplifying assumptions,28,29 with the important exception that, as mentioned above, these estimates (28) Gotz, T.; Bergt, M.; Hoheisel, W.; Trager, F.; Stuke, M. Appl. Phys. A 1996, 63, 315-320. (29) Balzer, F.; Gerlach, R.; Manson, J. R.; Rubahn, H.-G. J. Chem. Phys. 1997, 106 (19), 7995-8012.

Figure 8. Group delay time versus distance dependence for rhodamine B peak (443 Da) at different laser intensities. (2) -210 and (b) -370 MW/cm2. Table 1 sample rhodamine B coumarin 522 methylanthracene BBQ

peaks (Da)

velocity (m/s)

standard deviation (m/s)

399 443 283 192 675

157 150 344 360 175

(35 (30 (56 (78 (35

are valid only if the desorption process is momentary, with all molecules desorbing exactly at the t0. In order to test this key assumption, we conducted a series of experiments to experimentally determine the velocities of molecules desorbed by LIAD. This procedure was based on measurements of time-of-flight dependencies as above, but with varying distance rather than delay time. This was accomplished by moving the cylindrical lens along the normal to the sample surface so that the F2 laser beam focus position moved to different distances from this surface. In this way, a series of time-of-flight spectra were measured, and for each of them, the mean value of the group delay of the molecular bunch td was calculated using the formula

∑ At

i i

td )

i



(3) Ai

i

where Ai is the signal intensity corresponding to the moment ti. In Figure 8, the relation between the distance d and the values of the measured group delay td is demonstrated. The obtained dependence can be fitted by a straight line, and the slope of this line corresponds to the average value of the molecular group translational velocities. The estimated values of the group velocities for different molecules using this procedure are listed in Table 1. It is apparent that these values are much higher (more than 1 order of magnitude) than those estimated above under the assumption of a momentary desorption source. This discrepancy implies that the desorption source in the LIAD process is distributed in time rather than being momentary, so the desorption event lasts much longer that the initiating laser pulse. Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8237

of the desorption signal was measured for different analytesubstrate combinations, and the essential results of these measurements are presented on Figure 9. The depletion of rhodamine B signal was measured at a laser intensity of 350 MW/cm2 and a surface concentration of ∼0.5 nM/cm2. It is interesting that, in contrast to the data of ref 13, the signal remained measurable even after 104 laser shots (Figure 9). If these data are plotted in a double-logarithmic scale (see Figure 9 inset), it becomes evident that this dependence (signal vs number of desorbing laser shots) can be well approximated by a power law:

A ) A0N -1 Figure 9. Total rhodamine B LIAD signal intensity (399 and 443 Da) as a function of the cumulative number of laser shots. The inset plot represents the same dependence redrawn in the doublelogarithmic scale and fitted by the strait line.

In the shake-off mechanism, molecular desorption can only occur when the velocity of the moving foil surface is close to its maximum. As was shown in ref 30 this corresponds to the very first foil vibration, which lasts a fraction of a microsecond. On the contrary, we have shown evidence here that the desorption event lasts much longer, possibly tens of microseconds. This contradiction casts doubt on the applicability of the shake-off mechanism to LIAD. We have additionally measured the group velocities of the desorbed molecules at different desorption laser intensities. The shake-off model would describe laser intensity as leading to an increase of the transverse surface motion velocity, as the result of increased acoustic wave amplitude and a corresponding increase in translational velocities of the desorbed molecules. The effect of desorption laser intensity was measured for a rhodamine B sample (deposited onto Ta foil at an estimated concentration of 0.4 nM/cm2) and at two levels of laser intensities, 210 and 370 MW/cm2 (Figure 8). Despite the fact that the photoion signal values for these two intensities differed by ∼2 orders of magnitude, the measured translational velocities were essentially identical: 156 ( 25 and 150 ( 35 m/s for the laser intensities of 370 and 210 MW/cm2, respectively. This apparent independence of the translational group velocities of the desorbed molecules from the intensity of the desorption laser is strong evidence against models of LIAD that assume any direct mechanism of energy transfer from acoustic waves to the analyte molecule. This observation also contradicts the hypothesis of a thermal basis for LIAD, because the increased surface temperature also should cause increased kinetic energy for the desorbed molecules. The independence of the measured velocities of desorbed molecules on different desorption laser intensities was observed for other species in our experiments. This is in a good agreement with recent observations27 and is one of the important features of LIAD. Repeated laser irradiation of the same spot on the foil backsurface caused the signal to continuously decrease, due to the depletion of the analyte molecules on the surface as has been observed previously.13 Figure 9 displays the first quantitative measurements of this behavior. In our experiments, the depletion (30) Zinovev, A. V.; Moore, J. F.; Calaway, W. F.; Pellin, M. J.; Veryovkin, I. V. In Laser Beam Control and Applications; Kudryashov, A. V., Paxton, A. H., Eds. SPIE Proc.: San Jose, CA, 2006; Vol. 6101, p 61011U-2.

8238

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

(4)

where A is the current value of the signal, A0 is the initial signal value, and N is the number of laser shots. If we consider a simple desorption model, with n molecules bound to the surface, and suppose that the desorption probability after each laser shot is p, then we will inevitably come to the exponential signal decay law

A ) A0 exp(- RN)

(5)

(where R ) ln(p)), which is in contradiction with the obtained experimental results. We will discuss this contradiction in detail below. DISCUSSION It is clear at this point that a more sophisticated desorption model is required to explain all of the results described above. The following discussion aims at forming just such a consistent hypothesis. First we shall examine whether thermal desorption or mechanical shake-off mechanisms are more consistent with our observations. The exponential increase of the desorption yield with desorption laser intensity seems to be an argument supporting the thermal desorption nature of LIAD. Also, the measured translational velocities of the desorbed molecules (see Table 1) correspond to temperatures of 500-700 K that (1) are specific for the thermal desorption of the physically adsorbed molecules and (2) seem to be a realistic estimate for the surface temperatures of the front surface of the back-irradiated foil.31 However, the fact that translational velocities of the desorbed molecules are independent of the laser intensities (and, thus, of the surface temperature) casts strong doubts regarding the validity of the thermal mechanism of LIAD. Moreover, the dependences of the σ parameter (which characterizes molecular fragmentation) on the desorption laser intensity (Figures 5 and 6) do not match the exponential dependences observed for the desorption yield. This indicates the lack of thermal equilibrium in the desorbed molecules. The hypothesis of direct impulse transfer from the vibrating foil surface to the adsorbed molecules (i.e., the mechanical shakeoff mechanism) also looks unlikely due to the tremendous difference (∼2 orders of magnitude) between velocities measured for desorbed molecules and those for the surface vibration. The change in the fragmentation parameter σ with the change of (31) Prokhorov, A. M.; Konov, V. I.; Ursu, I.; Mihailescu, I. N. Laser Heating of Metals; Adam Higler: Bristol, 1990; p 240.

desorbing laser intensity should also not be observed if the shakeoff model is correct. This is not surprising because, for acoustic wave generation, the surface vibration velocities are very small, whereas shock wave formation in metals requires the use of much higher laser intensities of 1013-1015 W/cm2 (see, for example, ref 32) that are not achieved in LIAD experiments. So, both the thermal desorption and shake-off mechanisms reveal significant disagreements with our experiments. Let us also consider the signal depletion problem described by Figure 9 and eq 4. We propose for these data the following qualitative model. It can be supposed that a considerable number of small independent desorption centers exist on the sample surface, so that the desorption probability p has independent and different values at each center and can vary from 0 to 1. Then, to calculate the total signal from the desorption area that includes many such desorption centers, one should integrate the signal over all these centers and take into account eq 5. By integration, many independent exponential functions, a power law such as the eq 4, will be obtained. The assumption of the presence of independent desorption centers on the surface seems to be very realistic. This is because of the following: (1) it is very unlikely that the sample preparation method used here (as well as in the works refs 7-12) produces uniform coverage with submonolayer thickness; (2) thin foils are polycrystalline and have a native oxide on their top, which means their surface is very complex and consists of many defects and multiple crystal facets, each with different surface free energies. As a result of the coalescence during drying of the solvent on such a nonuniform surface with varying surface energy, many independent small islands of the analyte can form there. For example, the formation of fluorescein and rhodamine B microcrystals, mostly submicrometer in size, on tin oxide surfaces has been observed.33 The size and the abundance of these islands will be defined by the analyte solution, its concentration, and the surface free energy. Thus, each island can act as an independent desorption source and has its own desorption probability. One can posit next that the acoustic vibration of the metal foil surface in combination with some rise in its temperature is the trigger for starting the desorption process. It can be proposed that isolated islands of analyte, formed on the metal surface after solution drying, are not in minimal energy states. It is well-known that film growth on surfaces is accompanied with stress in the growing film.34 For LIAD phenomena, the important point is that each island has some effective energy excess due to the stress. A previous example was the formation of nonstable fractal islands of silver clusters on a graphite surface with subsequent decomposition due to internal instabilities.35 For LIAD, the acoustic and thermal waves can cause surface stress and, consequently, surface elastic deformation, which can initiate a similar island decomposition process. As a result of this effect, molecules can be desorbed from islands. The amount of energy stored in the residual stress (32) Swift, D. C.; Tierney, T. E.; Kopp, R. A.; Gammel, J. T. Phys. Rev. E 2004, 69 (3), 036406. (33) Griffith, O. H.; Houle, W. A.; Kongslie, K. F.; Sukow, W. W. Ultramicroscopy 1984, 12 (4), 299-308. (34) Freund, L. B.; Surech, S. Thin film materials:stress, defect formation, and surface evolution; Cambridge University Press: Cambridge, 2006; p 750. (35) Brechignac, C.; Cahuzac, P.; Carlier, F.; Colliex, C.; Leroux, J.; Masson, A.; Yoon, B.; Landman, U. Phys. Rev. Lett. 2002, 88 (19), -.

depends on many unknown factors and cannot not be calculated directly for our samples. Even for more simple substrate-film systems, the value of residual stress could be estimated only experimentally, with the use of rather sophisticated methods.34 But it is very important to note that the amount of energy transferred to the desorbed molecule during this process depends only on the specific characteristics of a given island, but does not depend on the total energy of the acoustic or thermal wave. This can readily explain the independence of the desorbed molecules’ velocities from the intensity of desorbing laser radiation, which was also observed in other recent work.27 On the other hand, the increase of the desorption laser intensity and the consequent increase of the surface elastic deformation come to involve more of the stressed analyte islands into the transformation process and lead to an exponential rise in the number of desorbed molecules. Then the desorption probability does not depend on the number of molecules remaining on the surface for a given laser shot. This model therefore fits the observed signal depletion pattern. The mechanism outlined above resembles the one proposed by Vertes22,36,37 for MALDI, which is also based on a thermal stress generation in the layer of organic film deposited on solid substrate. One should notice, however, clear differences in the physical conditions between the desorption methods. For MALDI, the absorption of the laser pulse energy occurs in an optically and thermally dense film, which experiences thermal stress due to its nonuniform and fast heating. In the case of LIAD, laser radiation is absorbed by the back side of the metal foil substrate, opposite from where the sample was deposited. The amount of energy transmitted through the metal foil to this analyte layer should be so significantly attenuated, compared to the (direct) front irradiation, that it cannot directly be sufficient for desorbing molecules with the velocities observed in our experiments. On the other hand, a typical average thickness of the analyte film in LIAD is on the order of several molecular layers. Because of this, the specific density of energy stored in each analyte island due to intrinsic stress can be sufficiently high, and during laser irradiation of the back side of the foil, the heating can simply trigger release of this energy and a molecular desorption event. It is possible to make some numerical estimation of this stress energy, as it is related to the thermal mismatch of the substrate foil and analyte island on the top. Thermal stress energy due to foil heating by laser irradiation G may be expressed in the following equation38

G)

E(1 + ν)2 ‚(Rs - Rf)2∆T22 πr0hf 2 2(1 - ν )

(6)

where E is the elasticity modulus, v is the Poisson ratio, Rs and Rf are the thermal expansion coefficients of the substrate and film correspondingly, ∆T is the temperature rise, and hf and r0 are the island thickness and radius. The average energy per analyte molecule can be calculated using eq 6 (36) Vertes, A.; Levine, R. D. Chem. Phys. Lett. 1990, 171 (4), 284-290. (37) Vertes, A. In Methods and Mechanisms for Producing Ions from Large Moleculres; Standing, K. G., Ens, W., Eds.; Plenum Press: New York, 1991. (38) Boley, B. A.; Weiner, J. H. Theory of Thermal Stress; John Wiley &Sons, Inc.: New YorK, 1960; p 586.

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8239

ga )

2 2 E(1 + ν)2 (Rs - Rf) ∆T M NA 2(1 - ν2)F

(7)

Here M is the molar mass, F is the specific gravity, and NA is the Avogadro number. It is interesting to note that ga does not depend on the analyte island size but strongly depends on the thermal and mechanical parameters. There is little information in the literature on these parameters for molecular crystals, but based on existing data for anthracene,39 we can estimate the value of ga. Suppose that Rf ) 2.8 × 10-4 K-1, Rs ) 6.3 × 10-6 K-1, E ) 13 GPa, and F ≈ 106 g/m3, for ∆T ) 100 K, we will get ga ) 0.025 eV. This is not enough to break intermolecular bonds, but when thermally induced stress exceeds a critical value, the film can start to fracture and the stored energy is released in a small volume in the vicinity of the stress cracks. Some energy excess is taken in the increased surface free energy, but because this process is obviously nonadiabatic, the crack formation is also accompanied by breaking intermolecular bonds and by the formation of new desorption sites. Due to the strong spatial nonuniformity of thermomechanical properties of molecular crystals,39 the physical mechanisms involved in this process are very complicated in nature; therefore, we can give only a qualitative picture of this phenomenon. Presumably, the cracks are formed along grain boundaries, defects, and interfaces. The increase of the desorption laser intensity that causes a rise in ∆T and, in accordance with eq 6, an increase in energy G, results in the formation of additional cracks. This will increase the number of desorption sites and finally the total number of desorbed molecules as observed in the nonlinear laser intensity dependence. But because the formation of any individual cracks is defined only by the intermolecular bonding forces in the vicinity of the crack, the translational kinetic energy of desorbed molecules should still remain independent of driving laser intensity, also matching our observations and other recent LIAD publications. The crack formation that causes the rupture of intermolecular bonds should also result in fragmentation of the analyte, presumably by breaking the weaker bonds (for example, CO2 loss from rhodamine B). However, because the binding energy between the atoms in molecules is higher than the intermolecular bonding, the probability of fragmentation is small, and observed mass spectra should have only a few fragments. It will be true in the conditions of constant analyte temperature, but due to the thermal conductivity process, the heat transfers from a hot substrate surface to deposited species, and their temperatures also increase. This will result in enhanced fragmentation at elevated laser intensities as was demonstrated in Figure 4. Nevertheless, in many experiments (anthracene and coumarin in our work and also in some others9,10,12), a complete absence of fragmentation was observed. This is likely due to (i) the use of more thermally stable molecules and (ii) the different experimental conditions. If factor i is obviously applicable to our experiments, then factor ii may play an important role in other experiments cited above. The surface analyte concentrations used in these experiments are 10100 times greater than the concentrations in our current experiments. Therefore, we propose that, in the previous experiments, (39) Bondi, A. Physical properties of molecular crystals, liquids and gases; john Wiley & Sons: New York, 1968; p 502.

8240

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

the analyte did not form isolated islands but instead created a uniform film of molecular crystals with thickness of hundreds of molecular layers. Such a film will change both heat-transfer characteristics and crack formation processes relative to islands. This factor may play a significant role in LIAD and should be considered when further analytical protocols are developed using LIAD as a molecular volatilization method. Another possible mechanism of molecular desorption in LIAD is related to substrate surface deformation by the generated acoustic wave and the subsequent formation of nonequilibrium surface electron states due to migration and annihilation of solid dislocations.40 These electron states vary with the material, and range from formation of localized holes near the top of the d-band for d-metals41 to generation of strong electric fields within the surface cracks for dielectrics.42 In any case, such states have an energy excess and could serve as an energy source for desorbing molecules. The specific mechanism of energy transfer may be different depending on the substrate-analyte combination and experimental conditions, but the influence of these nonequilibrium states on desorption processes should not be ignored. This mechanism should play a major role at higher desorpion laser intensities (>1010 W/cm2) and at higher acoustic wave energies that may result from confinement conditions of laser plasma ignition.43 Moreover, under some experimental conditions, the emissions of charged particles (both positive and negative) have been observed due to the surface deformation.42 This observation is in good agreement with less recent LIAD experimental results where the emissions of charged ions as well as electrons were observed.5,6,15 Finally, it should be noted that both mechanisms outlined above are not mutually contradicting and may both affect the desorption process, which makes optimization of LIAD into an intricate problem. While we cannot entirely exclude the influence of surface temperature on the desorption process, we consider pure thermal desorption processes, based on our experimental results, to be nondominant and playing only a secondary role in the LIAD phenomenon under commonly used experimental conditions. However, at very small foil thickness, as was pointed out in ref 44 the role of thermal desorption processes may be significant or even dominant. CONCLUSIONS We have shown that LIAD, coupled with SPI and TOF MS, is well suited for the investigation of samples containing organic molecules with relatively small IP. The use of two lasers to separate the desorption and ionization processes made it possible to observe, for the first time, some unusual characteristics of LIAD. Our entire data set cannot be explained in terms of simple intuitive models, and this suggests that a much more complex phenomenon is at play. The results of this work show that a direct energy (40) Abramova, K. B.; Shcherbakov, I. P.; Rusakov, A. I.; Semenov, A. A. Phys. Solid State 1999, 41 (5), 761-762. (41) Abramova, K. B.; Rusakov, A. I.; Semenov, A. A.; Shcherbakov, I. P. J. Appl. Phys. 2000, 87 (6), 3132-3136. (42) Nakayama, K.; Suzuki, N.; Hashimoto, H. J. Phys. D: Appl. Phys. 1992, 25 (2), 303-308. (43) Fabbro, R.; Fournier, J.; Ballard, P.; Devaux, D.; Virmont, J. J. Appl. Phys. 1990, 68 (2), 775-784. (44) Ehring, H.; Costa, C.; Demirev, P. A.; Sundqvist, B. U. R. Rapid Commun. Mass Spectrom. 1996, 10 (7), 821-824.

transfer from the acoustic wave to molecules adsorbed on the substrate surface is very unlikely, and the explanation of the observed LIAD regularities requires a model based on surface stresses caused by thermal and acoustic wave generation due to laser irradiation. These stresses trigger the process of surface reorganization, and thus, the desorption phenomena may be the result of two (not mutually exclusive) mechanisms: (i) crack development in the micro- and nanometer-sized islands of analyte molecular crystals on the metal surface and the consequent formation of desorption sites in the vicinity of the crack boundaries and (ii) formation of nonequilibrium electronic states on the substrate surface, which are in a repulsive state in relation to the adsorbed molecules and therefore become desorption sites. We believe that both mechanisms are relevant for the LIAD process, with a possible domination of one of the two depending on conditions of the experiment. Based on these results, we cannot quantitatively identify the specific channels of energy transfer to the outgoing molecules. The above picture is essentially qualitative and will need to be

refined by further experimental and theoretical studies. Still, the LIAD technique seems to be a broadly applicable and very promising method for volatilization of organic molecules, which is primarily useful when applied to mass spectrometry. Further LIAD studies will allow researchers to optimize the experimental conditions and should help to fully realize the potential of LIAD, in combination with different ionization techniques, as a powerful method for chemical analysis. ACKNOWLEDGMENT The authors are grateful to Dr. Praneeth Edirisinghe (UIC) for his kind help in HPLC/MS/MS analysis of rhodamine B solution. This work is supported by the U.S. Department of Energy, BES-Materials Sciences, under Contract DE-AC0206CH11357. Received for review March 23, 2007. Accepted August 24, 2007. AC070584O

Analytical Chemistry, Vol. 79, No. 21, November 1, 2007

8241