Anal. Chem. 2005, 77, 4042-4050
MALDI of Individual Biomolecule-Containing Airborne Particles in an Ion Trap Mass Spectrometer William A. Harris, Peter T. A. Reilly,* and William B. Whitten
Oak Ridge National Laboratory, P.O. Box 2008, MS 6142, Oak Ridge, Tennessee 37831
Individual airborne biomolecule-containing particles were detected and characterized in near real-time by matrixassisted laser desorption/ionization (MALDI) with an ion trap mass spectrometer. Biomolecule-containing particles were laboratory-generated and passed through a heated region containing a solution of matrix in equilibrium with the gas phase. Passage into a cooler region created a supersaturation, resulting in rapid deposition of the matrix vapor onto the biomolecule-containing particles, whereupon they were sampled into the inlet of our spectrometer. The coated particles were collimated and individually sized by light-scattering-based time-of-flight. When the sized particle reached the center of the ion trap, it was irradiated with a focused 266-nm laser, and the resulting ions were mass-analyzed. Mass spectra of leucine enkephalin, bradykinin, substance P, and polylysinecontaining particles were determined with attomole sensitivity. Structural information of the peptides contained in an individual particle was obtained by tandem mass spectrometry. Analysis of the results yields insights into the aerosol laser ablation ionization process that suggests an optically limited mechanism for ion production that has interesting ramifications on the utility of aerosol-based MALDI as an analytical technique. Since the development of matrix-assisted laser desorption ionization (MALDI) by Karas et al.,1 research has been conducted to develop the technique for analytical analysis. The MALDI technique as first practiced by Karas et al.1 was not readily amenable to high throughput. In most MALDI experiments, the analyte and the matrix were mixed and deposited on a probe, allowed to crystalize, and placed into the mass spectrometer to be evacuated, and finally, mass analyzed. Often, the final result depended on the sample preparation technique. Their original work has stimulated many schemes to couple the MALDI technique to chromatography and permit continuous monitoring that have been reviewed.2-4 Additionally, the push for sequencing spurred innovations in MALDI instrumentation. For example, the * Corresponding author. Fax: 865-574-8363. E-mail:
[email protected]. (1) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1987, 78, 53-68. (2) Murray, K. K. Mass Spectrom. Rev. 1997, 16, 283-299. (3) Orsnes, H.; Zenobi, R. Chem. Soc. Rev. 2001, 30, 104-112. (4) Papac, D. I.; Shahrokh, Z. Pharm. Res. 2001, 18, 131-145.
4042 Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
sample production process has been semiautomated with piezoelectric pipets to deposit low-nanoliter aliquots of matrix and analyte into etched wells.5,6 Automation has improved spot-to-spot reproducibility and permitted subfemtomole quantities of analyte to be detected. Murray et al.7 were the first to incorporate aerosols into the MALDI process. The matrix and the analyte were sprayed into the inlet of the mass spectrometer, where the particles were dried in a heated tube. Ions were subsequently formed by randomly ablating and ionizing the particles with a pulsed UV laser. This technique was not prominently used because prohibitively large quantities were consumed due to the inefficient transfer of the sample into the ionization region of the mass spectrometer and the lack of synchronization of the laser pulse to the presence of a particle. Later, Mansoori et al.8 improved the technique by creating monodisperse aerosol droplets containing the analyte, matrix, and solvent that were subsequently dried at atmospheric pressure and admitted into the differentially pumped inlet of their mass spectrometer. The inlet created a well-collimated beam of particles that passed through the center of the ionization region of their time-of-flight mass spectrometer by skimming away the off-axis particles in the initial expansion. Light-scattering was used to detect the presence of an incoming particle and to fire a pulsed laser when the particle was at the focal point to initiate the MALDI process. By systematically varying the analyte and matrix ratio, particle size, and laser intensity, they explored the quantitative relationship between concentration and signal intensity. They showed that aerosols could be used to eliminate sample inhomogeneities and provide consistent and nominally quantifiable MALDI mass spectra in the low analyte concentration limit. Stowers et al.9 furthered this work by condensing a UV absorbing matrix onto airborne particles before admitting the particles into their inlet. They also triggered their ablation/ (5) Little, D. P.; Cornish, T. J.; Odonnell, M. J.; Braun, A.; Cotter, R. J.; Koster, H. Anal. Chem. 1997, 69, 4540-4546. (6) Grimm, R.; Birrell, H.; Ross, G.; Charlwood, J.; Camilleri, P. Anal. Chem. 1998, 70, 3840-3844. (7) Murray, K. K.; Lewis, T. M.; Beeson, M. D.; Russell, D. H. Anal. Chem. 1994, 66, 1601-1609. (8) Mansoori, B. A.; Johnston, M. V.; Wexler, A. S. Anal. Chem. 1996, 68, 3595-3601. (9) Stowers, M. A.; van Wuijckhuijse, A. L.; Marijnissen, J. C. M.; Scarlett, B.; van Baar, B. L. M.; Kientz, C. E. Rapid Commun. Mass Spectrom. 2000, 14, 829-833. 10.1021/ac050187i CCC: $30.25
© 2005 American Chemical Society Published on Web 05/14/2005
Figure 1. Schematic of aerosol generator, matrix applicator, and aerosol ion trap mass spectrometer.
ionization laser by light-scattering-based detection so that they could analyze individual particles. Their technique permitted near real-time identification of biological aerosols. Using their technique, they generated mass spectra of individual bacterial spores. In the work presented here, we have extended the aerosolMALDI technique for use with an ion trap mass spectrometer to enable tandem mass spectrometry. Our technique also includes the use of an aerodynamic lens-based inlet system.10,11 An aerodynamic lens system consists of a flow limiting orifice followed by a series of constrictions and expansions that collimates a large range of particle diameters onto the axis of the aerodynamic lens system. Our lens system was equipped with six lenses before the final expansion through a 3-mm nozzle. The lens configuration used in this work had 5.0-, 4.5-, 4.0-, 3.75-, 3.5-, and 3.25-mm orifices. Using a 100-µm inlet orifice creates a flow rate of ∼83 mL/min, and the resulting pressure in the region between the inlet orifice and the first lens is approximately 3.5 Torr. The particle beam waist of a polydisperse aerosol at the center of the ion trap is typically 0.25 mm. The transmission efficiency of the particles in the design range of the lens system (∼50 to 3500 nm) is essentially unity, provided that the particles are moving at the same speed as the carrier gas when they reach the first lens.10,11 The distance between the first lens and the inlet orifice was 10 cm and should provide enough stopping distance for the particles in this range that are traveling along the axis of the lens system. 12 Unfortunately, a portion of larger particles will collide with the walls of the tube and stick before they are entrained in the gas flow. This is a process that limits the transmission efficiency of (10) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995, 22, 293-313. (11) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995, 22, 314-324. (12) Harris, W. A.; Reilly, P. T. A.; Whitten, W. B.; Ramsey, J. M. Rev. Sci. Instrum., Accepted.
the lens system in the upper limit. However, the probability of a particle entering the inlet orifice passing through the focal point of the ablation/ionization laser is very high, even for the larger particles relative to the a differentially pumped system composed of a nozzle and skimmers.13 The ability that sets our analytical technique apart is the capacity for performing tandem mass spectrometry. Because we use an ion trap as a mass analyzer, the aerosol MALDI-created biomolecular ions can be isolated, fragmented, and scanned for positive identification. We have also suggested a not-so-new concept to explain our results. The optical properties of particles suggest that the creation of the plasma that produces the ions may be self-limiting in terms of absorbing the energy from the laser. The optical properties of particles results in a plasma volume that is relatively invariant with particle size, yielding invariance in the number of ions produced with particle size. The ramifications of this concept are explored in terms of the utility of aerosol-based MALDI as an analytical technique.
EXPERIMENTAL METHODS The experimental setup is shown in Figure 1. Polydisperse biomolecule-containing particles were generated with a Collison nebulizer from a solution of 50/50 (v/v) methanol and water. The nebulized aerosol was passed through heated and cooled tubes consecutively to desorb the solvent from the particles and then condense it on the cooled wall to yield a dry aerosol. Aerosol generation and coating were performed at atmospheric pressure. The emerging aerosol then entered the matrix applicator consisting of a heated saturator and condenser. The saturator was a 15cm-long, 19-mm-diameter, heated glass tube with a felt lining. Two different matrix solutions, 3-nitrobenzyl alcohol and picolinic acid, (13) Reilly, P. T. A.; Gieray, R. A.; Yang, M.; Whitten, W. B.; Ramsey, J. M. Anal. Chem. 1997, 69, 36-39.
Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
4043
were individually added to the bottom of the saturator heated to between 70 and 85 °C (measured on the outer surface of the tube). Picolinic acid (PA) was used near saturation in water, and 3-nitorbenzyl alcohol (3-NBA) was used without dilution. These matrixes were chosen because they work well with 266-nm excitation, and they represent solid and liquid coating materials. Other matrixes, such as ferulic acid and DHB, were also successful, but for consistency, only spectra from 3-NBA and PA are presented. Inside the heated saturator, the aerosol and matrix vapor were in equilibrium so that the rate of deposition was equal to the rate of evaporation onto all surfaces. Upon leaving the saturator, the aerosol and matrix vapor were cooled to 3-10 °C, resulting in a supersaturation that causes the matrix vapor to rapidly condense onto the aerosol particles. The amount of matrix added to the surfaces of the particles was dependent on the linear flow rate and the saturator temperature. Up to a few hundred nanometers of matrix could be added to the radius of the particles. The flow rate through the Collison nebulizer and matrix-coating apparatus ranged between 0.5 and 1.5 L/min. The residence time in the saturator ranged from 5 to 1.7 s, respectively. A diagram of the aerosol generator, matrix applicator, and mass spectrometer is shown in Figure 1. This design is similar to that used by Stowers et al.9 and in condensation nuclei counters in commercial particle sizing instruments. A portion of the coated aerosols was drawn into the instrument and analyzed on an individual particle basis in real-time with our ion trap-based aerosol mass spectrometer. The spectrometer used herein has been described in detail elsewhere.12 It is an updated and transportable version of the instrument used previously in our laboratory.13 Our present instrument was converted from a commercially available ion trap mass spectrometer (Polaris Q, ThermoFinnigan, Austin, TX). The commercial system was minimally modified by replacing the top blank flange with one that contained our inlet, light-scattering detection optics, and laser ablation optics. In addition, four holes were drilled into the ring electrode to pass the particle beam and the ablation laser. The GC inlet and ionization source were not modified, and the instrument can still operate as a detector for a gas chromatograph, its original purpose. With these modifications, the instrument now operates as an aerosol mass spectrometer. First, an aerodynamic lens-based inlet system, based on the design of Liu et al.,10,11 produced an extremely well collimated particle beam from an aerosol admitted into the instrument through a 100-µm diameter orifice. A series of constrictions and expansions forced the particles toward the central axis of the lens system. The trajectory lines of the particles in the lens system are represented in Figure 1. Because the particles were moving along the central axis during the final expansion into vacuum, they experienced minimal radial dispersive force and remained collimated. The particle beam was further separated from the atmospheric carrier gas by passage through a 250-µm-diameter skimmer into the main chamber of the ion trap mass spectrometer. Inside the main chamber, the collimated particle beam passed through two focused 532-nm diode laser beams propagating into and out of the page in Figure 1. The scattered light from each laser was collected and focused onto a separate photomultiplier tube (PMT) through a pinhole to minimize background scattered 4044
Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
light. The PMT signals were inverted and converted to TTL pulses that were used to measure the time-of-flight between the two detection points. The measured flight time provided a direct correlation that determined the aerodynamic size of each measured particle. The scattered light signals also permit the timing of a focused 266-nm ablation/ionization laser (∼1.1 mJ/pulse, 0.1mm-diameter spot, 1.4 GW/cm2) to fire when the particle arrives at the center of the ion trap, regardless of its size or velocity. The timing ciruit is also designed to produce a TTL pulse to fire the flashlamps of the YAG laser at the optimal time (150 µs) before the detected particle reaches the center of the ion trap. Because the minimum flight time between the second detection point and the center of the ion trap is greater than the optimal 150 µs delay, there are no limitations to the size of the particles that may be analyzed other than the ability to detect the scattered light from the 532-nm detection lasers. The nascent ions created by the 266nm laser pulse were trapped and subsequently subjected to standard mass analysis techniques, including tandem mass spectrometry.12 Ions were detected with the stock channeltron detector that comes with the instrument. Under normal operating conditions, a high mass limit of 1000 m/z was obtained. This mass range could be extended through high mass resonance ion ejection via user-based custom interface software based on the Xcalibur Development Kit (ThermoFinnigan).14 For these experiments, the middle electrode of the ion trap was usually held at an rf voltage where low-mass ions are immediately ejected, thus improving sensitivity for high masses. When MS/MS was performed, isolation and collision waveforms were also applied. The instrument can scan at a maximum rate of 14 s-1. Actual scan rates depend on the scan function required. The majority of our work was performed for high resolution at a scan rate of 485 µs per amu scanned over a 1000-Da range, yielding a scan time of ∼600 ms, including ion preparation and data storage. Particle flows were set to achieve a detection rate of one particle/s. This was done to minimize clogging of the 100-µm inlet orifice. For comparison, normal MALDI experiments were performed on bulk samples in a Voyager linear time-of-flight mass spectrometer with a 337-nm desorption/ionization laser. These samples were prepared at a 1:9 (v/v) analyte/matrix ratio (10 mg/mL R-cyano-4-hydroxycinnamic acid in 2:1 acetonitrile/0.1% TFA). Samples used in these experiments were obtained from the following suppliers and used without additional purification: R-cyano-4-hydroxycinnamic acid and leucine enkephalin, Sigma, St Louis MO; bradykinin and substance P, Global Peptides, Ft. Collins, CO; 3-nitrobenzyl alcohol and picolinic acid, Aldrich, Milwaukee, WI.
RESULTS The size distributions (obtained by light-scattering-based timeof-flight in the instrument) of the bradykinin-containing particles before and after the matrix coating with 3-NBA and picolinic acid is shown in Figure 2. The distribution in panel (a) depicts the polydisperse nature of a bradykinin aerosol after it had been nebulized and dried by passage through the heated and cooled tubes (see Figure 1). The distribution peaked at ∼630 nm and (14) Kaiser, R. E.; Cooks, R. G.; Stafford, G. C.; Syka, J. E. P.; Hemberger, P. H. Int. J. Mass Spectrom. Ion Processes 1991, 106, 79-115.
Figure 2. Size distribution of bradykinin bioaerosols (a) before and (b) after coating with 3-nitrobenzyl alcohol matrix coating (400-nm layer). Size distribution of bradykinin (c) before and (d) after coating with picolinic acid (200-nm layer).
ranged over more than 500 nm. The graph in panel (b) depicts the size distribution after the aerosol has been passed through the matrix-coating apparatus containing neat 3-NBA at 70 °C. The new distribution has shifted ∼800 nm and become noticeably narrower. In a separate run, the bradykinin aerosol yielded a slightly different distribution (shown in panel c) peaking early in the distribution at 530 nm and ranging ∼1000 nm. Passage of this distribution through the matrix coater containing vaporized picolinic acid yielded a broader distribution. The changes in the distributions observed in Figure 2 are interesting because they illustrate the range of processes that can occur during matrix deposition. In the continuum regime where the radius of the particle is much greater than the mean free path of the vapor, matrix deposition occurs though a diffusion-limited process so that the matrix deposits directly from the gas phase molecularly and the particles themselves do not interact. In this limit, the rate of change of the particle radius is proportional to the gas-phase concentration divided by the particle radius.15 Consequently, smaller particles grow faster than larger particles, resulting in a narrowing of the size distribution exiting the coating apparatus. This is precisely what we see when we coat the particles with the liquid matrix, 3-NBA (see Figure 2a and b). Especially for the solid matrixes, the type of distribution changes observed markedly depend on saturator temperatures and flow rates. Coating the particles with a solid matrix can be accomplished to yield results similar to those observed for the liquid matrix, 3-NBA, by appropriately adjusting the condenser temperature and subliming the solid matrix. This has been accomplished with picolinic acid and yields distribution changes similar to those shown in Figure 2a and b. (The narrowing picolinic acid deposition distributions are not shown for the sake of brevity.) On the other hand, if the temperature difference between the saturator and condenser is too great, then broadening of the size distribution can result. Figure 2c and d depicts the broadening of the distribution observed in the case of picolinic (15) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics; John Wiley & Sons: New York, 1998.
Figure 3. Averaged positive ion mass spectra of MALDI of bradykinin particles coated with a 210-nm layer of picolinic acid for (a) 600-800 nm (av ) 53), (b) 800-1000 nm (av ) 55), (c) 10001200 nm (av ) 79), (d) 1200-1400 nm (av ) 72), (e) 1400-1600 nm (av ) 73), and (f) 1600-1800 nm (av ) 66).
acid deposition. It indicates that other processes have occurred during matrix deposition besides diffusion-limited deposition onto particle surfaces. Picolinic acid is a solid at room temperature and at the temperature of the saturator. It is deposited in the saturator chamber as a solution that is wicked up into felt. Matrix exists in three phases in the saturator: in solution, as crystals on the felt, and in the gas phase with interaction between each. Additionally, condensation in the condenser does not necessarily only occur heterogeneously on the biomolecule containing aerosol. Homogeneous nucleation can produce particles of pure matrix that then can accumulate on the analyte particles. These phenomena can rapidly complicate the coating process. We suggest that the markedly broadened coated particle distribution results from these processes. Obviously, the matrix-to-analyte ratio changes within a given distribution; therefore, it is a quantity that is difficult to precisely define. However, lack of quantitative deposition is not critical because MALDI mass spectra do not vary much within a given size distribution, regardless of changes in experimental conditions. Figure 3 shows the averaged positive ion MALDI mass spectra of PA-coated bradykinin particles averaged as a function of particle size. The spectra shown were all derived from the same aerosol whose size distribution is shown in Figure 2d. Remarkably, the biomolecular ion signal from the polydisperse aerosol particles did not change as a function of size or thickness of coating on average. The particle mass spectra averaged over the 600-800nm range were similar to the average over the 1600-1800-nm range in terms of both intensity and fragmentation pattern. The lack of change in the total ion signal is surprising given the order Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
4045
Figure 5. (a) Single-particle MALDI of a 860-nm particle of bradykinin coated with 3-nitrobenzyl alcohol. (b) Single-particle MALDI of a 1130-nm particle of bradykinin coated with picolinic acid. (c) An average of 256 shots of 94 pmol of bradykinin in a MALDI-TOF-MS when the sample was deposited on a probe.
Figure 4. Mass spectra from MALDI of leucine enkephalin with 3-nitrobenzyl alcohol with 266 nm of (a) 0.20, (b) 0.37, (c) 0.76, (d) 2.15, and (e) 3.32 GW/cm2.
of magnitude difference in the mass of the particles. Another characteristic of the aerosol MALDI technique is that the parent ion intensities were relatively invariant to laser intensity. Figure 4 presents the averaged mass spectra of polydisperse leucine enkephalin particles coated with 3-NBA as a function of laser intensity. Our results corroborate the same invariance with laser intensity observed by Mansoori et al.8 For completeness, a comparison of the single-particle MALDI spectra of bradykinin from 3-NBA and PA are shown in Figure 5, along with an averaged MALDI mass spectrum of bradykinin from PA taken on a commercial MALDI TOF instrument. Approximately 110 amol of bradykinin coated with 1400 amol of 3-NBA (the top spectrum) yields a spectrum that is very similar to 340 amol of bradykinin coated with 3400 amol of PA (middle spectrum). The small variations in the relative intensities of the protonated and sodiated parent ions and the b8 + H2O fragments in the aerosol mass spectra resulted from shot-to-shot fluctuations. We observed that there is noticeably less b8 + H2O fragment in the conventional MALDI measurement of 94 pmol of bradykinin in 47 nmol of R-cyano-4-hydroxycinnamic acid deposited on the probe. We suggest that the aerosol MALDI process may be kinetically hotter because a 266-nm laser was used instead of the 337-nm laser in the conventional MALDI instrument, and there is no surface to act as a heat sink. There was also a large unresolved distribution of ion intensity on the low-mass side of the conventional MALDI mass spectrum that resulted from the use of the “low-mass gate” in the commercial instrument used to diminish the matrix ion signal that may reduce the sensitivity of the channel plate detector. A large concentration of matrix ions can also decrease the sensitivity of ion trap-based MALDI by destabilizing the heavier analyte ions. Matrix ion interference was 4046 Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
not an issue in our ion trap-based work because the rf voltage was set so that the MALDI matrix ions were not trapped to keep them from interfering with the analyte ions. We have also provided crude estimates of the quantities of analyte and matrix to provide a ball park comparison of the sensitivities of the techniques. Estimates of the analyte and coating compositions were obtained by assuming a uniform thickness of matrix over the entire distribution defined by the shift in the average of the distributions between the coated and uncoated particles. A direct comparison of sensitivity of the aerosol MALDI and normal MALDI techniques is not possible, since each spot placed on a probe provides a number of shots, and the sensitivities we are demonstrating for aerosols are from a single laser shot hitting an individual particle. We provide these quantities to demonstrate the development potential of the aerosol MALDI technique for delivering and analyzing minute quantities of analyte. Time-of-flight-based aerosol MALDI in the literature has demonstrated the same range of sensitivity; however, there is an issue of resolution and mass accuracy for these instruments.16 The lack of resolution and mass accuracy for this method results from the kinetic energy imparted to the analyte ions during the ablation process. On the other hand, the resolution of the analyzer is not limited by the ablation/ionization process in ion trap-based aerosol MALDI because the ions are trapped and kinetically cooled by collisions with a buffer gas before the mass analysis is performed. Collision-induced cooling also eliminates postsource decay effects that reduce resolution. Consequently, the resolution and mass accuracy for the aerosol MALDI technique with an ion trap is the same as with any other ion trap-based technique. To illustrate this point, the top panel in Figure 6a depicts the negative ion MALDI of a 730-nm particle containing ∼40 amol of substance P. The inset showing the mass scale expansion around the parent M - H- ion clearly shows each of the isotope peaks baselineresolved. This type of resolution is routine for ion trap analysis. (16) Van Wuijckhuijse, A. L.; Stowers, M. A.; Kleefsman, W. A.; Van Baar, B. L. M.; Kientz, C. E.; Marijnissen, J. C. M. J. Aerosol Sci. 2004, In press.
Figure 7. MALDI of a 780-nm particle of polylysine coated with 3-nitrobenzyl alcohol. M ) Lysine.
Figure 8. Log of TIC of tryptophan as a function of particle size. Figure 6. (a) Negative ion MALDI and (b) MS/MS of substance P from coated 730-nm particles. Particles were coated with a 125-nm layer of picolinic acid; and the original amount of substance P was 40 amol. (c) Positive ion MALDI and (d) MS/MS of substance P from 1.4-µm particles coated with 3-NBA. A 450-nm layer was applied, resulting in an original amount of 50 amol of substance P. (e) MS/ MS of 110 amol of bradykinin from a 1.4-µm particle containing a 400-nm layer of 3-NBA.
Perhaps the greatest advantage of an ion trap-based aerosol MALDI system is the ability to perform tandem mass spectrometry. Figure 6b shows the MS/MS of another particle (same size and composition) that readily reveals the identifying fragmentation pattern from only 40 amol of analyte. This technique works equally well for positive ion MALDI. Figure 6c and d show the positive ion MALDI and the corresponding MS/MS of substance P. These results agree well with the tandem mass spectrometry of Jainhuknan and Cassady of substance P by post source decay time-of-flight mass spectrometry.17 MS/MS works equally well for any peptide with the aerosol MALDI technique. The tandem mass spectrum of bradykinin is shown in Figure 6e. The best representation of the technique’s potential for sensitive analysis is represented in Figure 7. It presents the single particle MALDI mass spectra from 3-NBA-coated polylysine. The analyte-to-matrix mole ratio was ∼1:10 for a 780-nm coated particle. The polylysine analyte was a mixture of polymer molecules of different lengths with a known distribution. The ion intensity at M7H+ was from an estimated 25 amol of M7 polymer, and the M11H+ was from ∼4 amol of M11 polymer. (17) Jainhuknan, J.; Cassady, C. J. Rapid Commun. Mass Spectrom. 1996, 10, 1678-1682.
DISCUSSION Ion Signal Versus Particle Size. From more than a decade of our experience and in the literature, 8,18 one of the most intriguing observations in laser ablation-based aerosol mass spectrometry is that the total ion intensity does not directly correlate with particle mass. To study this point, we nebulized a concentrated solution of tryptophan and dried the aerosol to produce a broad distribution of sizes of pure tryptophan particles and sampled them directly into our mass spectrometer (without matrix). Figure 8 presents the log of the integrated ion intensity as a function of particle size. Any mass spectrum with a total integrated ion intensity below 105 was deemed a miss and did not produce a recognizable tryptophan spectrum. This plot clearly shows that the capacity of our commercial ion trap system is between 108 and 109 on our scale. More ions can be created by the aerosol MALDI technique, but they cannot all be stored in the trap. Any excess ions will not be trapped, leaving the trap filled to its capacity. The other thing that is clearly represented is that the ion signal intensity for each size spans the full range of represented intensities. In other words, 180-nm particles yield essentially the same signals that 1800-nm particles do, yet they are a factor of 1000 different in mass. Signal intensity seems to be completely stochastic and independent of particle size. From our experience, this phenomenon is observed for any particle material, including MALDI matrixes. During the ablation process, the MALDI matrix transfers charge to the analyte. The lack of dependence of the MALDI matrix ionization on the particle size likely translates to analyte ionization as revealed in Figure 3. It is logical to suggest that this phenomenon and the relative independence of the ion signal with respect to laser intensity seen in Figures 3 and 8 and elsewhere for larger particles (>2µm) in the literature8 have the same root cause. We believe that the Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
4047
Figure 9. Interaction of light with an individual particle, diameter D, with a focal distance, F.
relative independence of ion signal with respect to particle size and laser intensity can be attributed to optical properties of the particles. Consider the simplest case, a spherical particle of diameter, D, interacting with a coherent light source depicted in Figure 9. In the regime where the circumference of the particle (πD) is much greater than the wavelength of the coherent light source (λ), the path of the light can be described by tracing rays. If the particle is transparent to the light and the refractive index, n, is 2, the light focuses inside the sphere according to the following equation,
F ) ND/[4(n - 1)]
where F is the focal distance (see Figure 9). Some portion of the light passing into the sphere will be reflected back into the sphere (dashed line) at the surface, while most exits. Typically, between 4 and 10% of the light gets reflected at any given boundary between one transparent medium and another, depending on the relative refractive indexes. The light that gets reflected at the back surface (dashed line) collects at another focal point where the rays cross just inside the back surface of the sphere. Another fraction of this reflected light will undergo a second reflection (the dotted line) and collect at yet another focal point just inside the front face of the sphere. This behavior has been described and calculated in detail in the literature.19 The optical behavior of droplets has been studied experimentally.19-21 An image of a levitated droplet containing a laser dye can be found on the cover of the August 20, 1990, issue of Applied Optics.20 One thing that is evident from the levitated droplet images in the literature is that the secondary reflectioninduced focal spots inside the droplet are relatively small, as compared to the size of the droplet. This observation has been corroborated by calculations performed by Benincasa et al.,19 in which they determined the maximum intensities at the internal and external near-field peaks 279 and 2356, respectively, relative to the incident intensity for a 35-µm sphere at n ) 1.3611. This represents very tight focusing of the incident light. If one assumes that between 5 and 10% of the light is reflected at the droplet surface, then their calculations suggest the focal spot inside the (18) Kane, D. B.; Johnston, M. V. Environ. Sci. Technol. 2000, 34, 4887-4893. (19) Benincasa, D. S.; Barber, P. W.; Zhang, J. Z.; Hsieh, W. F.; Chang, R. K. Appl. Optics 1987, 26, 1348-1356. (20) Arnold, S.; Spock, D. E.; Folan, L. M. Opt. Lett. 1990, 15, 1111-1113. (21) Dusel, P. W.; Kerker, M.; Cooke, D. D. J. Opt. Soc. Am. 1979, 69, 55-59.
4048
Analytical Chemistry, Vol. 77, No. 13, July 1, 2005
back surface and outside the droplet have roughly the same beam waist. The light is equally well focused inside and outside of the droplet because the same surface or boundary produces both focal points from the same rays. The ability to focus light inside a droplet extends to materials that absorb. The only criterion is that the material cannot be too reflective (e.g., a metal sphere) or significantly attenuate the light by absorption. The absorption path length inside an aerosol particle is very short so that even materials with relatively strong absorption cross sections will not significantly attenuate the incident light before it gets focused. Consequently, this type of focusing is applicable to the laser ablation mechanisms.22,23 Laser ablation/ionization is initiated at the first inner focal point for “transparent” spherical particles where the intensity is the highest. This process begins by formation of a plasma that destroys the sphere’s ability to redirect the energy toward the front.24 The plasma acts as an energy sink until it reaches some critical electron density, whereupon the plasma becomes reflective.22 The plasma is the source of the ions that are observed in aerosol mass spectrometry. Once the plasma becomes reflective, its growth is limited and so, too, is the number of ions it produces. The buildup to critical electron density happens rapidly with increasing incident light intensity so that the plasma saturates. It is our contention that it is the optical creation of the plasma in the relatively small volume inside the back surface of the sphere and the dynamics of the plasma evolution that limits the size of the plasma created. In the case in which the sphere circumference is much greater than the wavelength of the incident light, the size of the focal spot does not vary much with changing sphere size. The beam waist (w) generated by a spherical optic is given by
w ) CλF/D ) Cλn/[4(n - 1)]
where C is a constant. Hence, the focal volume does not depend on the size of the optic and the plasma volume, and the number of ions produced does not change appreciably with changing spherical particle size in this regime. As the particle circumference approaches the wavelength of the incident light, the intensity distribution inside the sphere becomes isotropic (the Rayleigh limit). Essentially, the volume of the particle is approaching the focal volume. When this happens, a plasma is still created, but it can be created throughout the volume of the sphere. We would then expect the ion signal to be proportional to the particle mass; however, it has been noted that the ion signals from nanoparticles, where the circumference of the particles is much smaller than the wavelength, still drops slowly with decreasing size.25 In the case of nanoparticles, the plasma can be created throughout the volume of the entire particle because it is believed that the entire nanoparticle is vaporized during the ablation/ionization process. To be measured, the ions must escape the plasma without being neutralized. Typically, for (22) Alexander, D. R.; Schaub, S. A.; Zhang, J.; Poulain, D. E.; Barton, J. P. Opt. Lett. 1989, 14, 548-550. (23) Lee, J.; Becker, M. F.; Keto, J. W. J. Appl. Phys. 2001, 89, 8146-8152. (24) Alexander, D. R.; Barton, J. P.; Schaub, S. A.; Holtmeier, G. M. Appl. Opt. 1991, 30, 1455-1460. (25) Kane, D. B.; Oktem, B.; Johnston, M. V. Aerosol Sci. Technol. 2001, 34, 520-527.
laser ablation/ionization, only 1 part in 1000 ablated species is ionized. As the particle volume and, hence, the plasma volume decreases for nanoparticles, the probability of an ion in the laserinduced particle plasma escaping actually increases with decreasing particle size, thereby offsetting the losses due to reduction in particle mass or plasma volume. We have made the above arguments for the simplest case of spherical transparent and absorbing particles; however, the above reasoning applies to nonspherical particles, as well. For this class of particles, light will still collect at specific points inside the particle volume and begin the creation of the plasma whose growth and volume are still limited by the same optical properties. The plasma can be created in more than one spot in the volume and saturates rapidly by becoming reflective. Variance in the number of points for plasma initiation may be a key factor that contributes to the vast range of total ion intensities observed. It also follows that the above discussion relates to the aerosol MALDI mechanism in that creation of the plasma is part of the MALDI desorption/ionization process, and hence, the MALDI ions are produced in the same way. This suggests that MALDI signals can be created with much smaller particles than those studied here to further improve the sensitivity of the technique. The Aerosol MALDI Technique. The aerosol MALDI technique has tremendous potential for becoming a sensitive method for biochemical analysis. With the use of an aerodynamic lensbased inlet, aerosols have become one of the most efficient methods of transporting analyte into vacuum for mass analysis. The transfer efficiency of particles from the inlet to a spot
