2175
Anal. Chem. 1002, 64, 2175-2179
CORRESPONDENCE
Fluorescence Imaging of Gas-Phase Molecules Produced by Matrix-Assisted Laser Desorption Theodore W. Heise and Edward S. Yeung* Department of Chemistry and Ames Laboratory-USDOE, Iowa State University, Ames, Iowa 50011 In the analysis of large biomolecules,one important aspect is the determination of molecular weight. An excellent tool for determining molecular weights is mass spectrometry (MS). One requirement for MS analysis is that the sample either be itselfin the gas phase or can be converted to the gas phase. However, biomolecules, by nature large and typically containing both polar and nonpolar regions, are not easily converted to the vapor phase. When vaporization does occur, a common problem for molecular weight determination is that the biomolecules fall apart into fragments. Several approaches that have been effective in producing intact biomolecules in the gas phase1 include field desorption,Z 252Cf plasma desorption,3 SIMS,4 thermo-5 and electrospray,6 and laser desorption (LD).' LD of intact molecules requires that the laser pulse be short. In other words, the heating rate of the sample must be rapid enough to favor vaporization over fragmentation. The energy from fast pulsed lasers has been particularly effective in this regard. However, even with the use of a 10-ns pulse from a frequency-doubled Nd-YAG laser, significant sample destruction has been observed.8 A recent innovation9is matrix-assisted LD (MALD). This approach involves combining a large excess of a matrix material with the sample species. Typical matrices are fairly volatile or sublime easily and are chosen to have significant absorption at the irradiation wavelength. An overview of the technique1 has recently been published. Current understanding of the process10 is that direct absorption by the sample molecules is minimal-rather, they are swept along with the vaporized matrix material. It is also possible that expansive cooling by the matrix minimizes fragmentation." The detailed dynamics of the process are not well understood, however. One interesting feature of MALD is that efficient generation of parent ions depends critically on the laser energy and the matrix-to-sample ratios1 Conditions that fall outside a very (1) Hillenkamp, F.; Karaa, M.; Beavis, R.C.; Chait, B. T.Anal. Chem. 1991,63, 1193A-1203A. (2) Beckey, H. D. Principles of Field Ionization and Field Desorption Mass Spectrometry; Pergamon Press: Oxford, 1977. (3) Macfarlane, R. D.; Torgerson, D. F. Science 1976, 191, 920-925. (4) Barber. M.: Bordoli. R. S.: Sedewick. R. D.: f i l e r . A. N. J . Chem. Soc.,.Chem. Commun. 1981,325-327: (5) Blakely, C. R.; Carmody, J. J.; Vestal, M. L. Anal. Chem. 1980,52, 1636-1641. (6) Whitehouae,C.M.;Dreyer,R.N.;Yamashita,M.;Fenn, J.B.Anal. Chem. 1986.57.675-679. (7) Denoyer,'E.; Van Grieken, R.; Adams, F.; Natusch, D. F. S. Anal. Chem. 1982,54, 26A-41A. (8) Kimbrell, S. M.; Yeung, E. S. Appl. Spectrosc. 1991,45 (3), 442'
447. .. .
(9) Karaa, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Processes 1987, 78, 53-68. (10) Vertes, A.; Levine, R. D. Chem. Phys. Lett. 1990,171,284-290. (11) Vertes, A.; Gijbels, R.; Levine, R. D. Rapid Commun. Mass Spectrom. 1990,4, 228-233. 0003-2700/92/0364-2175$03.00/0
narrow range have been found to be ineffective. A complication that exists in interpreting these MS results is that vaporization and ionization are not separately studied. Conditions that favor one process may not be identical to those required for the other. For example, fragmentation can occur as a result of additional excitation during the ionization step. There has been some indication that the bulk of the vaporized species is in the neutral form.12 If so, secondary ionization after vaporization, appropriately implemented, can conceivably increase the yield of the parent ions. Optical spectroscopic probes based on lasers have been successfully employedfor monitoring laser-generated plumes previously, providing spatial and temporal information on atoms,13 diatomics,14 and large molecules.* These are based on absorption in the plume,8J3J4reflection off the surface,lb and laser-enhanced ionization.16 The distinctive feature is that these probes can follow the neutral aa well as the ionic forms of the vaporized species. If fragmentation involves destruction of the chromophore, one can even measure the extent of fragmentation.8 However, because of the substantially lower concentrations of the species of interest in MALD, the above probes are not suitable for similar studies there. In this article, we report the use of laser-excited fluorescence to provide spatial and temporal maps of the plumes generated by MALD. The information available is similar to that obtained by absorption probes, but with the high sensitivity needed for MALD events.
EXPERIMENTAL SECTION Reagents. Matrices were chosen on the basis of previous utility in MALD-MS.' The matrices used were 2,bdihydroxybenzoic (gentisic) acid, 3,4-dihydroxycinnamic (caffeic) acid, and 4-hydroxy-3-methoxycinnamic(ferulic)acid. All were obtained from Aldrich Chemical (Milwaukee, WI). Dyes were chosen to represent a variety of absorptivities at the desorption wavelength (337 nm) as well as reasonable fluorescence quantum efficiency at the probe wavelength (488 nm). The dyes used included Coumarin 6 and Coumarin334 (EastmanKodak, Rochester, NY) andRhodamine590Chloride(Exciton,Dayton,OH). Allreagents were used as received with the exception of ferulic acid which was recrystallized from boiling water. Dye solutions were prepared in methanol. Approximate concentrations were 10mM Rhodamine 6G (R6G),5 mM Coumarin 334 (C3),and 0.2 mM Coumarin 6 (C6). Matrix material was added to aliquota of dye solutionsto givenearly saturated (-0.5 M) solutions. For ferulic (12) Johnson, R. E.; Sundqvist, B. U. R. Rapid Commun. Mass Spectrom. 1991,5, 574-578. (13) Steenhoek, L. E.; Yeung, E. S. Anal. Chem. 1981,53, 528-532. (14) Huie, C. W.; Yeung, E. S. Spectrochim. Acta 1985, 40B, 12551258. (15) Pang, H. M.; Yeung, E. S. Appl. Spectrosc. 1990,44,1218-1220. (16) Pang, H. M.; Yeung, E. S. Anal. Chem. 1989, 61, 2546-2551. 0 1992 American Chemical Society
2176
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
,Q amplitier 0lens2
no-
beamstop plume
I I
2&o
4Jedm
tlcouslo-optic moduletom
Flgurs 1. Experlmental arrangement for fluorescence Imaging in lasergenerated plumes.
and gentisic acids the approximatematrix/dye molar ratios were 2000,100, and 50 for C6, C3, and R6G respectively. The caffeic acid ratios were approximately half these values. Samples were prepared by depositing 100-pL solution volumes on 5- x 5-cm quartz plates held in an oven at 90 "C. The solvent was allowed to evaporate leaving rather uniform matrix/dye films on the quartz plates. Film thickness was estimated to be approximately 10 pm. Equipment. The experimental geometry is shown in Figure 1. The 337-nmline of a Nz gas laser (Lambda Physik, Acton MA, Model EMG101) was used for desorption. Each run was the result of a single desorption pulse. A 20-cm spherical mirror was used to focusthe desorption laser pulse to a line. The orientation of the line was lengthwise parallel to the probe beam. Measurements from burn spots on photographic film and vaporized samples showed the desorption pulse to be focused to 0.2 x 9 mm. A focused line was used rather than a point to increase the probe path length within the plume. The quartz plate on which the sample was deposited was oriented in the plane of Figure 1 with the desorption pulse focused through the plate onto the sample film from below. This so-called "back-side" desorption is thought to minimize plasma effects due to interaction of the desorption laser with the vaporized sample. Desorption of samples was carried out at atmospheric pressure without any specialenclosures. The pulse energy was monitored by directing a partially reflected portion of the desorption beam to an energy probe (Laser Precision, Utica, NY, Models Rj-7200and Rjp734). Typical pulse energies were 2-3 mJ at the desorption site with pulse widths of -2011s. The pulse energy was varied by adjusting the high-voltage supply of the desorption laser. The laser-generated plume was probed (see Figure 1)with an argon ion laser (Laser Ionics, Orlando, FL, Model 554A) lasing at 488 nm. Two mirrors were used to direct the probe beam into the first acousto-optic deflector (AOD). The first AOD (Isomet, Springfield, VA, Model 1205-C)scanned the probe beam in the horizontal direction. The output from the first AOD was sent through a second AOD (IntraAction,Belwood,IL, Models ADM150 and DE-150) which scanned the probe beam in the vertical direction. The AOD's were driven by asymmetric triangular wave forms at 200 kHz vertically and at 10 kHz horizontally (Wavetek Corp., San Diego, CA, Models 162 and 182A, respectively). The wave form asymmetries produced unidirectional scans with periods of 5 ps vertically and 100 ps horizontally. During the last 5% of each sweep the beam was reset or returned to its starting position. The vertical wave form generator was gated with TTL output pulses from the horizontal wave form generator. The horizontal wave form generator was itself gated by the SYNCH OUT signal from the NZlaser. In order to map the largest area possible, a long focal length (100 cm) lens was used to focus the probe beam. It has been documented" that AOD's can have significantcylindricallensing effects when operated at high frequency. The lower frequency horizontal AOD has an inconsequential focal length of more than 30 m. However, the higher frequency vertical AOD has a focal length of -1.7 m which is significant in comparison to the 100cm lens. Because the AOD lensing characteristics are different in each dimension,the optimum focal lengths for the two scanners (17)Lekavich, J. Lasers Appl. 1986, (41, 59-64.
are not the same. Empirically, the best resolution was obtained with the spherical lens positioned 74 cm from the probe region. Resolution was determined, by measuring the probe beam intensity transmitted through a grid at the focal point. The photodiode measurement demonstrated a resolution of about 20 X 24 spots in a 9-mm vertical X 8-mm horizontal scan. Vertical scans started at a position 1mm away from the sample surface. Naturally, optimization between the resolution and the image size can be changed by proper focusing. An aperture was placed so as to block the zeroth and second orders from the AOD's. A 10-cm-diameterX 10-cm-focallength sphericallens was used to collect the fluorescence signal from the entire region onto a photomultiplier tube (PMT) (Products for Research, Danvers, MA, Model 56 AVP). To minimize scattered radiation the PMT was placed normal to the probe beam. In addition, a 500-nm cutoff filter and a 488-nm holographic Raman edge filter (Physical Optics Corp., Torrance, CA)were usedto discriminate against scattered probe light. The PMT signal (500 s2 termination) was amplified by an oscilloscope (Tektronix,Beaverton, OR, Models 7704 and 7A15) and sent to a transient digitizer(LeCroy,Chestnut Ridge, NY, Model 9410). The digitizer was operated at 10 MHz with 8-bit resolution. For each desorption event l o 4 data points were recorded. This allowed 10 sequential 100-ps two-dimensional images to be collected. Therefore, each plume was monitored for a total of 1ms. Data were dumped to an IBM-AT microcomputer through an IEEE-488 interface (Capital Equipment Corp., Burlington, MA, Model PC-488) for subsequent evaluation. Data acquisition was triggered in the single event mode by the SYNCH OUT signal from the desorption laser and commenced -12 ps after the desorption event. The AOD's efficiency of diffraction into the first order is not necessarilyconstant over the length of a scan, with the efficiency fallingoff at the ends. To allowfor correctionof AOD efficiencies, a portion of the postplume probe beam was reflected to a l-cm X l-cm photodiode (Hamamatsu, Middlesex, NJ, Model 5179001). Normalization of the PMT signal with respect to the photodiode signal showed no detectable change from the raw signal. We therefore conclude that the scanning efficiency was constant over the plume region of the two-dimensional maps.
RESULTS AND DISCUSSION The molar absorptivities (L.mol-l.cm-l) for dye films on quartz plates at 337 nm were determined to be 120,600, and 1600 for C6,C3,and R6G respectively. These values were derived from the observed absorbance over a calibrated area of uniform films of known concentration. These follow the general trend of solution measurements but do not match the latter exactly because of the different environments. These values are, however, all substantially lower than the molar absorptivities of the matrix materials. This, plus the large matrix to analyte ratio, implies that coupling of the laser beam is primarily via the matrix material. The efficiency of sample desorption was checked by visual inspection of the films using a microscope. A marked threshold dependence on the desorption laser power density was observed. This is consistent with previous reporta.18 When the desorption laser irradiance was less than 1MWf cm2, desorption was incomplete or absent. At higher irradiances (i.e.,>1.5MW/cm2)desorption was virtually complete with no observable film residue remaining. The cleaned sample area was found to be a function of laser pulse energy. This can be attributed to the changing mode structure of the beam profile with discharge energy. Since the films are of finite thickness, once the desorption threshold is exceeded, no additional material is removed from the sample surface with increased irradiance. Ferulic acid was found to be the most effective matrix and will be discussed in detail here. Caffeic and gentisic acids gave similar results, but the signals were factors of 2 and 3
-
(18)Ens,W.; Mao, Y.;Mayer, F.; Standing, K. G. Rapid Commun. Mass Spectrom. 1991,5, 117-123.
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
1
2
2177
3
l i o r i z o n t n l p o s i t i o n (mrri)
4
5
6 E E
u
v
0
5
10
0.1
2
4
2
m
-
W
0
k
a
0.05
G:
0
.* 0
L W
-0.05 0
D
liorizontnl position ( m m )
7
8
9 E E
W 0
!i
v
10
0.I
5
2
.L(
0
.*
m
: a
W
0.05
4
4
d 0
[s
-4
0
-0.05
k
u
I,, , , , , , , , ,!, , ,! , , ,,, ,,, , , , , ,, , I , ,
0
I , ,
I , ,
~
1_
0 I i o r i ~ o ~ i t apIo s i t i o n ( m m )
Flgurr 2. Time-lapse sequence of concentration maps of C6 generated by M A D . 0.12 ng desorbed: matrix, ferulic acid In 2000-fold excess; laser energy, 2.2 mJ; laser irradiance, 5.6 MWlcm2;frame rate, every 100 ps: Imaged area, 9 mm (vertical)X 8 mm (horlzontai), 1 mm above the surface. The horizontal scan range Is from 0 to 8 mm and is repeated for each frame. The vertical scan range is from 1 to 10 mm above
the surface.
smaller, respectively. In preliminarywork, sinapinic acid was found to desorb quite well. It was not studied further, however, as ita solubility in methanol was too low to give an acceptable matrix/dye ratio. Figure 2 shows nine consecutive scans (frames) through the plume region starting 1mm away from the surface. To get a sense of the meaning of the plot it may be helpful to visualize the position axes for each frame as forming a plane perpendicular to both the sample surface and the probe laser.
The probe laser starts a scan at the front left corner in Figure 2 and sweeps toward the rear. Upon reaching the back, the scanner resets and rapidly returns the beam to the front. The same process is occurring concurrently from left to right at l/20 the speed. The front axis, then, represents the horizontal dimension of the scan. The vertical axis shows fluorescent intensity. The remaining axis corresponds to vertical distance from the sample surface. The depressions between the 2D images are artifacts which were inserted to emphasize frame
2178
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
3 n
E E
v
0)
3
0.04
8
lo
u u W
.I
n
L
;
c
0
a
0.02
u
.rl
0
b#
! -0.02
h o r i z o n l a l position ( m m )
4
5
6
liorizonlal ponilion ( m m )
I
8
9
-E
-E
PI
,d
0.04
IO
u
n
C 3 .I
4
; iz
01
a
0.02
M
0 .*
0
4
3
-0.02 5
Iiorizonlal p o s i t i o n ("1
Flgur. 9. Tlme-Lapse sequence of concentratlon maps of C3 generated by MALD. Conditions are identical to those In Flgwe 2 except that the amount desorbed Is 1.2 n ~the , matrlx is In 100-fold excess, and the laser energy and iffedlance are 1.7 ml and 5.2 MWlcm2, respectively.
boundaries. The signal at the left edge of the first frame is due to scattered laser light and associated dye fluorescence from the tail end of the desorption pulse. This provides a reference for zero time in the desorption process. One notes that no signal is apparent throughout the f i t 2D image. The full frame represents 100 pa, but the probe has actually cleared the region where the plume will be detected at -60 pa. Adding to this the 12-ps period prior to data acquisition, one can state that the plume had yet to enter the probe region 72 ps after the desorption pulse. From
the first appearance of the plume we can determine the effective velocity of the plume to be about 60m/s. Naturally, the actual velocity changes dramatically as a function of time due to collisions.19 In the second through sixth frames one can see the plume expand and the centroid of fluorescence move away from the sample surface. Fluorescence intensity grows to a maximum at -600 pa. This agrees well with the time of maximum absorbance signal in previous work? (19) Pang,H.M.;Yeung,E.S. Appl. Spectrosc. 1990,44, 871-875.
ANALYTICAL CHEMISTRY, VOL. 64, NO. 18, SEPTEMBER 15, 1992
2179
60-
40-
30-
60 0
amount of dye desorbed (nmol)
Flgure 4. Relative Integrated fluorescence IntensRles as a function of the amount of analyte desorbed for CB. The data labels correspond to the lrradlances (MW/cm? assoclated wlth each plume.
showing that the velocities of the vaporized materials are comparable. We note that these results are representative of plume velocities in atmospheric-pressure gas, and are therefore substantially slower than velocities observed in vacuum.18'20-22 Figure 3 displays results from a similar experiment with Coumarin 334 dye. Again one sees the direct and indirect results of the desorption laser pulse. Also, the plume does not appear until the second sweep through the probe region. In this run the motion of the plume away from the surface is more readily apparent. The signal-to-noise ratio here is lower than in Figure 2, but otherwise the behavior is similar. The maximum signal is once again observed at t 600 ps. Similar results, albeit with smaller signals,were obtainedwith R6G in a ferulic acid matrix. For the three dyes investigated the relative fluorescence signals collected appear to be inversely proportional to the molar absorptivities of the dyes at the desorptionwavelength. This might lead one to conclude the desorptionlaser is causing destruction of the fluorophore. However, this does not take into account any spectral shifts in absorption in the solid phase around 337 nm or the relative absorption strengths and fluorescence yields (488 nm excitation) of the three dyes in the gas phase, which are not known. In particular, it is known that the ionized and the unionized forms of these dyes in solution exhibit dramatically different fluorescence properties. In Figures 4 and 5, we show plots of total fluorescence as a function of the amount of dye desorbed. The latter was determined from the sample area cleaned off by the laser for that particular desorption event and the estimated amount of dye per unit area of the film deposited. These are linear (r = 0.94 and 0.98 for C3 and C6, respectively) but not correlated with laser irradiances which range from 4.4 to 14 MWIcm2. The amount of dye desorbed does not correlate with irradiance because the samplef i b is completely removed with each pulse. If the film was vaporized to only a certain depth, one would expect the correlation to exist. If, in fact, the desorption laser were causing destruction of fluorophore, changes in irradiance would cause nonlinearity in Figures 4 and 5, as was observed in ref 8. The indication here of no sample fragmentation (at least none involvingthe conjugated
-
(20)Hansen, S.G.J. Appl. Phys. 1989,66, 3329-3336. (21)Beavis, R.C.; Chait, B.T. Chem. Phys. Lett. 1991,181,479-484. (22)Pan, Y.;Cotter, R. J. Org. Mass Spectrom. 1992,27, 3-8. (23)Spengler, B.;Kirsch, D.; Kaufmann, R.Rapid Commun. Mass Spectrom. 1991,5, 198-202.
0
0.2
0.4
0.6 0.8 1 1.2 amount of dye desorbed (nmol)
1.4
1.6
Flgure 5. Relathre Integrated fluorescence IntenslUesas a function of the amount of analyte desorbed for C3. The data labels correspond to the lrradlances (MW/cm? assoclated with each plume. part of the fluorophore)agrees with the considerableamount of MALD-Ms work1which hasnot deteded analytafragments under these conditions. In summary, we have developed a transient imaging technique based on laser-excited fluorescence for monitoring laser-generated plumes. Sensitivity is well within the levels required for studying matrix-assistedlaser desorption,where analyte concentrations are about lo00 times lower than those in conventional laser desorption.8 Although the present study involves plumes in a normal atmosphere, adaptation to measurements in vacuum (e.g. inside a mass spectrometer) should be straightforward. A faster acquisition rate and a larger imaged area (fewer points per unit distance) will accommodatethia. By repeatedly imaging a single desorption event, as opposed to single-frame or single-point measurements, one can directly determine frontal and group velocities as well as diffusion rates. The spatial and temporal distributions are important for the design of secondary ionization schemes to enhance ion yields and for the optimization of ion collection in time-of-flight instruments to maximize resolution. Since the integrated concentration is available for each time segment, one can potentially w e a s the rate of desorption and the rate of fragmentation,23 if any. If in the future information about the spectroscopicdifferences between the neutral and the ionic forms becomes available, one can then determine the fraction of the desorbed analytes that are ionized at each point. Such information will provide insight into whether ionization is closely associated with the desorption step or whether it is a result of subsequent collisions with the matrix gas (e.g. proton transfer). Similar conclusions can be drawn about the mechanism of fragmentation. The excitement surrounding MALD is primarily due to the ability to generate intact molecular ions from large proteins and oligonucleotides. However, there is no reason why fluorescence-labeled biomolecules cannot be studied with the present apparatus. As long as the level of tagging is low, inferences can still be made about the inherent mechanisms involved.
ACKNOWLEDGMENT The Ames Laboratory is operated by Iowa State University for the U.S. Department of Energy under contract No. W-7405-Eng-82. This work is supported by the Director of Energy Research, Office of Basic Energy Sciences.
RECEIVED for review April 20, 1992. Accepted June 17, 1992.
