Laser desorption mass spectrometry with thermospray sample

Stan. Analytical Chemistry 1985 57 (12), 2280-2282. Abstract | PDF | PDF w/ Links .... Donald B. Borders , Guy T. Carter , Ronald T. Hargreaves , Mars...
1 downloads 0 Views 714KB Size
2

Anal. Chem. 1984, 56, 2-7

Laser Desorption Mass Spectrometry with Thermospray Sample Deposition for Determination of Nonvolatile Biornolecules E. D. Hardin, T. P. Fan, C. R. Blakley, and M. L. Vestal*

Department of Chemistry, University of Houston, Houston, Texas 77004

A new laser desorptlon mass spectrometer has been interfaced to a liquid chromatographusing a movlng stalnless steel belt. Samples are sprayed on-line onto the belt under partial vacuum wlth a thermospray vaporizer. These samples are transported through a dlfferentlally pumped vacuum lock and ionlzed in the source of the mass spectrometer with 45 ns, lo8 W/cm2 laser pulses from a Q-switched Nd:YAG laser. Data on the performance of thls new LCILDMS are presented for several classes of nonvolatlle, thermally lablle blomolecules.

An area of great analytical importance to organic and biomedical research is the development of mass spectrometric techniques for determination of high molecular weight (>lo00 daltons), nonvolatile,thermally labile biomolecules (see reviews in ref 1-3). Some of the most successful approaches to this problem involve the use of desorption-ionization (DI) techniques where energetic beams of fission fragments, ions, atoms, or photons are used to desorb molecular ions from solid samples present in the source of a mass spectrometer. These DI techniques include plasma desorption mass spectrometry (PDMS) (4), secondary ion mass spectrometry (SIMS) (5,6), fast atom bombardment (FAB) (7,8),and laser desorption mass spectrometry (LDMS) (9-19). In 1978 Kistemaker and co-workers showed that LDMS could be used to analyze representative samples from many major classes of nonvolatile biomolecules such as amino acids, nucleosides, peptides, nucleotides, and saccharides. Much of the current LDMS research has focused on elucidating the ionization mechanisms involved in the LD process. In LDMS submicrosecond laser pulses with power densities on the order of 106-108W/cm2 are used to desorb molecular ions. Proton transfer, alkali attachment, and desorption of “preformed ions”, e.g., organic salts, have been studied by LDMS. Cationization by silver ion attachment has been studied in laser desorption experiments by Cooks and co-workers (13),and attachment of many other metal cations was also observed. Cluster formation with metastable dissociation has been observed (12). With magnetic instruments large cluster ions formed in semifluid chemical ionization type reactions of sucrose have been observed (14). Cotter has determined the temporal distribution of laser desorbed neutrals and ions (15, 16). Laser desorbed neutrals have been ionized by CI (17), EI, and alkali ion attachment from a crossed alkali ion beam (18, 19). Cationized molecular ions have been shown to be formed in gas-phase ion/molecule reactions between neutrals desorbed around the periphery of the laser hot spot and codesorbed alkali ions thermionically desorbed from the center of the beam (19). Intense, long-lasting molecular ion beams have been laser desorbed from samples diluted in an ammonium chloride matrix (13). Cationized molecular ions and “preformed ions” have been desorbed from bulk samples and salt mixtures upon irradiation with CW lasers (20). These 0003-2700/84/0356-0002$01.50/0

experiments have helped to isolate and identify ionization mechanisms involved in LDMS, some of which may be shared by all of the energetic beam DI techniques. Our research involves both understanding the LD process and developing its analytical potential. In our view, a practical technique should provide both molecular weight and structural information on a wide range of nonvolatile and/or thermally labile compounds, it should be compatible with conventional rapid scanning mass analyzers, both magnetic and quadrupole, and it should be suitable for combination with the techniques commonly used for separating and purifying mixtures of involatile compounds such as liquid chromatography (LC) and thin-layer chromatography (TLC). With these criteria in mind, we have coupled a liquid chromatograph to a LD mass spectrometer using a moving stainless steel belt sample introduction system for continuously supplying sample to the source of a LD quadrupole mass spectrometer. SIMS has previously been used as an on-line LC detector utilizing the moving belt interface (21,22) and PDMS has also been used as a detector in combined LCMS (23) with an on-line rotating disk sample collection device. We are using a new approach for sample deposition in which a thermospray vaporizer (24) is used to spray the LC effluent onto the moving belt under partial vacuum. The advantage of this approach is that most of the solvent (-95%) is removed before sample deposition onto the belt. This prevents backmixing on the belt and also eliminates the need for strip heaters and extra differential pumping stages.

INSTRUMENTAL METHODS The mass spectrometer used in this work (see Figure 1) consists of a hyperbolic rod quadrupole mass filter, an Extranuclear Model QPS quadrupole controller, an rf-only quadrupole filter, an off-axis electron multiplier, a PAR CW-1 boxcar integrator, and a Finnigan INCOS data system. Ions are focused into the quadrupole entrance aperture by using two ion extraction cones and an einzel lens. The source lenses and quadrupoles are usually operated at a potential of 10-15 V with respect to the grounded belt. Differential pumping is supplied by a 1000 L/min mechanical pump on the thermospray region, a %in. diffusion pump on the intermediate region, and a 4-in. diffusion pump on the analyzer region. A source pressure of 2 x IO4 torr and a thermospray region pressure of approximately 400 mtorr is maintained when an LC flow rate of 1 mL/min of water is used. Samples are introduced into the system with a Valco LC injector valve backed by a Waters M-45 HPLC solvent delivery system. The belt (Ebtec Corp.) is 304 alloy stainless steel, 120 cm X 3.19 cm X 0.05 nm, and is normally operated from 0.20 to 1.0 cm/s. We have found that belts darkened by mild oxidation at 450 O C in a furnace give better results, presumably due to more efficient laser power absorption. The thermosprayer (see Figure 2) consists of a 0.15 mm i.d. stainless steel capillary tube brazed into a copper block heated by two cartridge heaters. The thermosprayer has been operated in two modes, with and without the transfer tube. In 0 1983 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 56, NO. 1, JANUARY 1984

ANALYZER QUADRUPOLE

jifEM LASER BEAM

.

'1'

1000 I/min

ROUGH PUMP

f

150 l h e c 2" DIFFUSION PUMP

1200 I h e c

4"DlFFUSlON PUMP

Figure 1. Schematic diagram of the laser desorption mass spectrometer showing the thermospray deposition apparatus and sample transport system. The laser beam enters through the window at the right and intersects the belt at a 45" angle. 0.006"I D SS CAPILLARY TUBE

CARTRIDGE HEATERS

-

CAPILLARY SWAGELOK FITTING

I/8 I.D. S S TRANSFER TUBE 118 WIDE

3

DEPOSITION REGION

.

DOWNSTREAM THERMOCOUPLE

OOlO" ssB* SLITS

i l

1/16 DIA GAS EXIT APERTURE

Flgure 2. Schematic diagram of the enclosed thermospray deposition assembly. The internal pressure is optimized by varying the gas exit aperture diameter.

the open arrangement (without the transfer tube) the thermosprayer is positioned about 1 cm above the belt and approximately 220-240 "C is necessary to produce a fine, visible mist covering the belt when using 1mL/min of water. The thermospray temperature is used to adjust the degree of solvent vaporization, and the temperature at which a visible mist is produced without liquid buildup on the belt gives the best results. With the transfer tube in place (see Figure 2) slightly lower temperatures can be used. In this mode the thermosprayer is adjusted to a temperature just high enough to prevent ice formation on the belt or on the downstream thermocouple. With this approach there is a more gradual desolvation and less thermal energy loss than when spraying directly into the vacuum. This arrangement has only recently been installed but preliminary results show better sample deposition efficiency than the open arrangement. We are currently optimizing the pressure of the enclosed thermospray assembly to further increase the sample deposition efficiency. Some samples were prepared by the electrospray deposition method (25). This is accomplished by removing the entire belt assembly (which conveniently slides out in one piece)

followed by electrospraying at ambient pressure. Replacement of the belt assembly and evacuation of the mass spectrometer take approximately 20 min. Energy deposition into the sample is provided by a Quantronix Model 210 Q-switched NdYAG laser which can operate from 100 to 50 000 Hz. The unfocused output of the 1.06 pm, 0.5 mm diameter beam has a power density of 2 X lo8 W/cm2 during a 45-ns pulse. This beam is expanded to 5 mm diameter and then refocused with a cylindrical lens (f = 10 cm) and a spherical lens (f = 50 cm) to produce a line image (3.1 mm x 0.11 mm) across the full width of the belt. From the lens equations the power density delivered to the belt during normal operation is calculated to be on the order of 108W/cm2 at the center and lo6 W/cm2 at the ends. The lenses are mounted on a translation stage (fba= 10.4 cm) which allows the central power density to be varied from lo5 W/cm2 to above lo9 W/cm2. It should be noted that these are calculated power densities; the actual power density absorbed by the sample and belt is the important parameter and cannot be easily measured directly. When operating in the on-line LC/MS mode several parameters, quadrupole scan rate, laser pulse rate, and belt speed are coordinated and some compromises must be made. We have found experimentally that over the normal range of belt speeds used, 0.2-1.0 cm/s, a laser pulse rate of 100-200 Hz gives the best molecular ion yields, presumably because there is less overlap on the belt between consecutive laser pulses causing less local heating. At 100 Hz this limits our quadrupole scan rates to 100 daltons/s, which allows for 1 pulse/dalton. At a belt speed of 0.25 cm/s a flat thermocouple placed under the belt measures a belt temperature of 31 "C at 100 Hz, but at lo00 Hz the temperature of the belt is raised to 125 "C. More pulses per dalton give more reproducible mass spectra but this requires either higher pulse rates or slower mass scans. Local heating produced by the higher pulse rates may cause some sample pyrolysis or premature demption, while slower scan rates limit the mass range available for rapidly eluting chromatographic peaks.

RESULTS AND DISCUSSION All of our laser desorption studies have been carried out by using a stainless steel belt as the substrate and source pressures in the lo4 torr range. Under these conditions there are many contaminants adsorbed on the belt. Figure 3 shows two summed background spectra from a well-used belt obtained at different power densities but at otherwise normal operating conditions. In the lo7 W/cm2 range there is little ionization except for Na+, Kt (not shown), and a few clusters such as KCl-K+and CsCl.K+. There are also unidentified mass peaks, presumably from previously pyrolyzed samples or vacuum pump oil contaminates. Above approximately 1.4 x lo8 W/cm2 there is a sharp rise in total ion current as atomic ion emission, Cr+ and Fe+, from the stainless steel itself now dominates the spectrum. There are also more cluster ions and more low mass unidentified ions at this power density. This high power density range has been called the laser desorption plasma mode (11,261 and its use results in formation of visible craters on the surface. We generally operate at an experimentally determined power density just above that necessary for atomic ion emission from the stainless steel; this gives the best trade off between background interference, sample fragmentation, and molecular ion intensity. One must keep in mind that only the power density at the center of the beam image is lo8 W/cm2;the outer ends of the line image have a power density 100 times less than the center. Whatever the power density used there is no discernible background above 375 daltons, the K(CsC1)2+ ion. In the positive ion mode cluster ions are practically unavoidable but generally not a problem except in the low mass range (below 150 daltons)

4

ANALYTICAL CHEMISTRY, VOL. 50, NO. 1, JANUARY 1984

BACKGROUND

GUANOSINE m.w 283

13

r

C1.K'

>

t

NEGATIVE ION SPECTRUM

m z .

E-

14

52

lNaC

POSITIVE ION SPECTRUM

1.4X 10' W / c d 133 C$

r5.0 X

190 BK+

.

MK+ Fe* (B-H)NaK+

212

W

a

160

m/z Flgure 3. Summed background spectra from the "clean" belt at 8 X IO' W/cm2 and 1.4 X 10' W/cm2. Not shown is the K+ Ion peak which is at least 2 orders of magnltude larger than the peaks shown. These spectra were taken at the following operating conditions: belt speed, 0.25 cm/s; laser repetition rate, 100 Hz; and quadrupole scanning rate, 100 daltons/s.

where they are quite intense. The negative ion background spectrum is substantially less intense. In the negative ion background intense peaks at 63 and 79 are observed. The nucleoside guanosine is a thermally labile, nonvolatile biomolecule that has frequently been used as a test sample for soft ionization techniques. Figure 4 shows the LD positive and negative ion spectra of guanosine taken at normal operating conditions using an LC flow rate of 1mL/min of water. The positive ion spectrum shows alkali-attached molecular ions and double alkali-attached deprotonated molecular ions as well as corresponding guanine base fragment ions. The degree of fragmentation is dependent on the power density as has been noted by others (11). The negative ion spectrum shows only (A4- H)- and B-ions. These results are typical of our laser desorption mass spectra from nucleosides and amino acids; MH+ ions are usually weak or absent. Some samples tested always yield MH" ions, however, most notably the basic amino acids arginine and histidine, but alkali-attached positive molecular ions are by far the most frequently observed species in the LD process. Also, samples with the highest positive molecular ion yields give the lowest negative molecular ion yields and vice versa. It would seem that alkali ion attachment would be enhanced and variability suppressed by using an alkali salt matrix containing the sample, and in fact this does yield an increase in cationized molecular ion production, but unfortunately this also caused a high increase in background from many different salt clusters and adducts. There is also a long lasting memory effect on the belt when using pure alkali halide salts. For these reasons we are presently not using a salt matrix with our samples. AI1 of the common nucleosides and amino acids we have tested thus far yield strong alkali-attached molecular ion signals. We have also been testing samples of higher molecular

IO

228 (B-H)K:

MNa+ 306

(M-HINaK'

344'

200 220 240 260 280 300 320 340

m/z Flgure 4. Positive and negative ion spectra of guanosine taken at normal operating condltions. Not shown is the (M - H)K2+ ion peak in the positive ion spectrum.

weights with similar results. Figure 5a shows the LD mass spectrum obtained from a 1 pg injection of the antibiotic erythromycin into the LC sample loop. This sample yields intense MNa+ and MK+ ions as well as structurally significant fragments at 583 and 599 daltons. The peaks at 739 and 755 daltons may be due to loss of water from the molecular ion. Figure 5b shows the LD mass spectrum obtained from 1 pg of the cyclic peptide antibiotic, gramicidine S hydrochloride. This spectrum also shows intense MNa+ and MK+ ions. The excessive peak broadening at the base of the molecular ion peaks is partly due to incomplete resolution by the quadrupole mass filter, but there appears to be an additional contribution, particularly at higher masses, which is not yet understood. This result demonstrates the high mass potential of LDMS, but it should be noted that desorption of alkali-attached molecular ions from cyclic compounds has been shown to occur by purely thermal methods (27) even at temperatures below that necessary for alkali ion emission. Figure 5c shows the LD mass spectrum obtained from 1pg of the peptide trp-met-asp-phe-amide HCl. The spectrum shows a strong MH+ ion signal but uncharacteristically little signal from alkali-attached molecular ions. This may be due to the fact that the sample is a hydrochloride salt. Again there appears to be loss of water or ammonia from the protonated molecular ion. All of the spectra shown here were taken by use of on-line thermospray sample deposition LC/LDMS. We have also prepared many samples by using off-line electrospray sample deposition onto the moving belt. The electrospray method can be used to provide relatively smooth sample coverages of known surface concentration. It would not make a suitable sprayer in an on-line moving belt interface, however, in that it requires high voltage, atmospheric pressure, 80% alcohol solutions,and low flow rates. We use the electrospray method as a comparison tool for studying relative thermospray sample deposition efficiency and homogeneity. Figure 6a shows the reconstructed ion chromatogram (RIC) from an electrosprayed

ANALYTICAL CHEMISTRY, VOL. 56, NO. 1, JANUARY 1984

ERYTHROMYCI N m.w. 733.9

ELECTROSPRAY

a

CYTl DINE

1001

757

a

5

100

:

SPEED RATE RATE

I cm/sec O.9KHz 230 dalton/sec

.4pg/crn2

>

t

9

266 M N b 282 MK+

F

253

W

-> tJ

r

3

K W

W

80ng/crn2

a

500

800

700

600

m /z b

GRAMlClDlNE S HCL

'0°1

1164 MM*

1

I > k z

MK'

1

504

v)

'r"

110 (M-45f

W

! i

156 MH 178 MNa' 194 MK'

- 5

w

2

5a W

730

6:OO

900

1000

m/z

1100

1200

12:OO

lool

HISTIDINE

6 3 4 MH'

w

t i z

10:30

C

TRP-MET-ASP-PHE-AMIDE HCI r n w 633

9:OO T I M E ( rnin)

9

IO

2

P 50

615

W

2

5 1.0 W

a

0.01

500

600

700

SURFACE 0.I COVERAGE 1.0 ( pg/cm2 )

IO

800

m/z Figure 5. (a) Positive ion LD mass spectrum taken from a 1-pg injection of erythromycin into the LC sample loop. (b) Positive ion LD mass spectrum taken from a 1-pg inlection of gramicidlne S hydrochloride. (c) Posltive ion LD mass spectrum of the peptide trp-metasp-phe-amide HCI. With this sample the MH+ ion is the major molecular ion produced.

series of different surface concentrations. This RIC of the summed intensities of the MNa+ and MK+ ions of cytidine, a particularly stable nucleoside, was obtained at a 900 Hz laser repetition rate, a 230 daltons/s scan rate, and a 1 cm/s belt speed. It characterizes the relative ion yields from above and below nominal monolayer (-1 pg/cm2) surface coverages. Below a monolayer, the response is approximately proportional to the surface coverage, but above a monolayer, the response increases only slightly with increasing surface concentration. This may be due to attenuation of the laser intensity by the

Figure 6. (a) Reconstructed ion chromatogram of the summed MNa+ and MK+ ion intensities from a series of electrosprayed surface coverages of approximately 80 ng/cm2, 1 pg/cm2, and 3.4 pg/cm2. I n this case the operating conditions are as follows: belt speed, 1 cm/s; laser repetition rate, 900 Hz; and scan rate, 230 daltons/s. (b) Reconstructed Ion chromatogram from thermosprayed histidine obtained by using 250-pL sample injections and the concentrations shown. A 150 dalton mass range was scanned In 3 s which allows for 2 puises/datton. (c) log-log plot of peak area response vs. sample surface coverage obtained from four replicates of the experiment illustrated in Figure 6b. A least-squares plot Is shown for data up to and below a nominal monolayer. The bars indicate the upper and lower limits of the mean standard deviation at each data point.

*

sample preventing good energy deposition, or it may be an artifact of bulk vs. surface ionization reactions. This RIC also shows the scan to scan fluctuations involved in the laser desorption measurement. We are not sure if these fluctuations are primarily from sample coverage inhomogeneity induced

6

ANALYTICAL CHEMISTRY, VOL. 56, NO. 1, JANUARY 1984

in the electrospray process or from differences in local substrate condition. Darker areas on the belt have a higher capacity for laser power absorption usually resulting in higher molecular ion yields. Also, at this pulse rate and belt speed there is overlap of the consecutive laser pulse images on the belt surface which may induce localized heating and sample pyrolysis. In experiments a t different belt speeds, we have determined that sample ionization efficiency is better when every laser pulse impinges on a fresh surface area. Figure 6b shows the RIC obtained from on-line thermosprayed histidine (mol wt 155) in various concentrations to show the response from above and below nominal monolayer surface coverages. The RIC is a summation of all the major molecular ions that are observed from histidine, (M COOH)', MH', MNa+, and MK+. Figure 6c shows the data from Figure 6b along with three other replicates in terms of relative peak areas vs. sample surface concentration. Surface coverages are estimated assuming a 40% thermospray sample deposition efficiency and average peak widths of 30 s. Mean and standard deviations were calculated for each data point and are shown on the graph. As can be seen, the response is linearly proportional to surface coverage up to slightly over a monolayer (-0.5 pg/cm2 for histidine), but at higher surface concentrations the response increases only slightly. This agrees with results obtained by electrospray deposition. Although all experimental conditions were carefully controlled to be identical in each repeated experiment, the absolute response still may vary by as much as f30% between experiments. This variation can be attributed to several factors. Liquid flow modulation inherent in the LC pump operation is one factor; while thermospray deposition is observed, the mist spot formed under the spray on the belt can be seen to grow and shrink in phase with the LC pump piston operation. This effect can be effectively eliminated, however, with a column in place. Another difficulty is in regenerating a reproducibly clean belt surface between experiments. Nonhomogeneous pump oil accumulations and local surface discolorations almost certainly contribute to variations in response. In spite of these uncontrolled factors, the relationship between relative response and surface coverage is still apparently linear. The relative thermospray sample deposition efficiency was obtained from detailed comparisons with electrospray sample deposition. In thdse experiments it was assumed that electrospray deposition provided 100% transfer efficiency. Comparisons between the open and closed thermosprayer arrangements have shown the closed arrangement to have a transfer efficiency 40% that of electrospray and approximately twice the transfer efficiency of the open arrangement. These comparisons were based on the intensity of the LD response; it is assumed that differences in response are due to actual surface concentration and that the response is not strongely influenced by the size of aggregates which may be formed on the belt. Also, in our comparisons using the two sample deposition techniques we have seen no evidence for sample decomposition induced by the thermospray process. For example, the apparent fragmentation of a thermally labile sample such as guanosine is the same with thermospray deposition as with electrospray deposition. Figure 7 shows a series of 12 injections of cytidine ranging from 2 to 100 ng using a 20-pL injection loop. Only the MNa+ peak at 266 daltons is shown in the reconstructed ion chromatograph. As can be seen, the chromatographic peak shapes are good and the response is approximately proportional to sample size. The peaks are approximately 3 s wide at the base and there is no apparent peak broadening induced by the thermospray deposition process. To catch these rapidly eluting peaks, 1 s mass scans were taken covering a range of

THERMOSPRAY DEPOSITION I Oml/mln

>

.!i W z

I-

z

266 MNa'

W

2

+

4W

a

IO0 150 TIME(ssc)

50

200

Flgure 7. Reconstructed ion chromatogram from a series of 12 injections of cytidine ranging from 100 ng to 2 ng each. A 20-pL sample loop was used with a flow rate of 1 mL/min. The RIC of the MNa ion of cytidine is shown. The noise peak is from a sample deposked during a previous run. 100-

UV CHROMATOGRAM COLUMN: CIS 5, REVERSE PHASE MOBILE PHASE: 0.02% HOAC, 4%MOOH, H 00

W

z u v)

m a CYTIDINE

Ioo1

Em -

1

ADENOSINE

LD T-SPRAYER R'C

TEMP: 230' BELT SPEED: 0.2cm/sec

I

z

z

IO0

200

400 TIME ( m i d

'

e00

Figure 8. UV trace and LDMS reconstructed ion chromatogram from an on-line separation of 1 pg each of four nucleosides. The RIC is a summation of all the molecular and fragment Ions that these nucleosides have been observed to yield.

100 daltons each. A representative mass spectrum covering 100 daltons can be obtained with a 10-ng sample in this way. A small number of LC separations have been carried out to test the capability of the LD mass spectrometer as an LC detector. For these tests we chose a mixture of four common nucleosides, cytidine, guanosine, uridine, and adenosine. Figure 8 shows the UV trace and the LD reconstructed ion chromatogram taken from an on-line separation of a mixture containing 1 pg each of these four nucleosides. The RIC is a summation of the molecular ions and fragments these molecules are known to yield. As can be seen, the RIC follows the UV trace very closely. There is little or no peak broadening induced by the thermosprayer because there is no liquid or ice build up on the belt. The RIC also shows the relative

ANALYTICAL CHEMISTRY, VOL. 56, NO. 1, JANUARY 1984

molecular ion yields of the different nucleosides; at the power density used in this experiment guanosine and cytidine have a somewhat higher ionization efficiency than uridine and adenosine. Compared to W detection at 254 nm the LD mass spectrometer is about 100 times less sensitive for nucleosides. The mass spectrometer does provide molecular weight and, in some cases, structural information where the UV detector does not. For samples that do not have strong UV absorptions, the LD mass spectrometer may be more sensitive. A timeof-flight (TOF) mass spectrometer with simultaneous ion detection would greatly improve our sensitivity; with the scanning quadrupole mass spectrometer we are sampling only a small portion of the total ions produced. A major limitation of our present moving belt system is that the LC separations must be accomplished in less than 9 min in order to prevent eluting sample components from overlapping each other on the belt. With the belt operating at 0.25 cm/s small sample amounts (less than a monolayer) are over 95% removed after one pass under the laser but larger sample amounts or faster belt speeds leave more residue. We do not, as yet, have a good method for on-line belt cleanup. Our approach thus far has been to clean off the belt after each revolution. This is accomplished by allowing the belt to rotate under the laser while operating at high repetition rate and high power density. Cleanup by conventional heaters has not proven satisfactory because their use causes nonvolatile samples to pyrolyze leading to localized discoloration of the belt and uneven laser power absorption. It appears that either a solvent scrubber combined with heating or using a single-pass disposable belt may be better approaches. Preliminary results with the new LC/LDMS are promising. Currently, we are attempting to optimize the enclosed thermospray deposition apparatus, experimenting with darkened belt surfaces, studying the effects of different sample matrices on molecular ion yields, and measuring the effects of different belt speeds. An ion gun has recently been installed for direct comparisons between SIMS and LDMS from the moving stainless steel belt. More detailed studies will be necessary before the analytical utility of these techniques can be established. Such studies are presently in progress.

7

Registry No. Guanosine, 118-00-3;erythromycin, 114-07-8; gramicidin S hydrochloride, 57572-76-6; Trp-Met-Asp-Pheamide.HC1,5609-49-4;cytidine, 65-46-3;histidine, 71-00-1;uridine, 58-96-8; adenosine, 58-61-7.

LITERATURE CITED (1) Vestal, M. L. Mass Spectrom. Rev., In press. (2) Bennlnghoven, A., Ed. "Ion Formation from Organlc Sollds"; SprlngerVerlag: New York, 1983. (3) Busch, K. L.; Cooks, R. 0.Science 1982, 278, 247. (4) Macfarlane, R. D.; Torgerson, D. F. Science 1978, 797, 920. (5) Benninghoven, A.; Slchtermann, W. K. Anal. Chem. 1978, 50, 1180. (6) Unger, S. E.; Day, R. J.; Cooks, R. G.,Int. J . Mass Spectrom. Ion Phys. 1981, 3 9 , 231. (7) Barber, M.; Bordoli, R. S.; Sedgwlck, R. D.; Tyler, A. N. Nature (London) 1981, 293, 270. (8) Rlnehart, K. L., Jr. Science 1982, 218, 254. (9) Posthumus, M. A.; Klstemaker, P. G.; Meuzelaar, H. L. C.; Ten Noever de Brauw, M. C. Anal. Chem. W78, 50, 985. (IO) Kaufmann. R.; Hlllenkamp, F.; Nltsche, R.; Schurmann, M.; Wechsung, R. Microsc. Acta 1978, 2 , 297. (11) Helnen, H. J. Int. J . Mass Spectrom. Ion Phys. 1981, 3 8 , 309. (12) Hardin, E. D.; Vestal, M. L. Anal. Cbem. 1981, 53, 1492. (13) Zakett, D.;Schoen, A. E.; Cooks, R. 0.J . Am. Chem. SOC. 1981, 103, 1295. (14) Heresch, F. Int. J . Mass Spectrom. Ion Phys. 1983. 4 7 , 27. (15) Van Breeman, R. B.; Snow, M.; Cotter, R. J. Int. J . Mass Spectrom. Ion Phys. 1983, 4g9 35. (16) Cotter, R. J.; Tabet, J. C. Int. J . Mass Spectrom. Ion Phys., In press. (17) Cotter, R. J. Anal. Chem. 1980, 52, 1767. (18) Stoll, R.; Rollgen, F. W. Z . Naturforsch, A 1982, 37A, 9. (19) Van der Peyl, G. J. 0.; Isa, K.; Haverkamp, J.; Kistemaker, P. G. Org. Mass Spectrom. 1981, 76, 416. (20) Stoll, R.; Rollgen, F. W. Org. Mass Spectrom. 1979, 74, 642. (21) Smith, R. D.; Burger, J. E.; Johnson, A. L. Anal. Chem. 1981, 53, 1603. (22) Bennlnghoven, A.; Elcke, A.; Junack, M.; Sichtermann, W.; Krlzek, J.; Peter, H. Org. Mass Spectrom. 1980, 75, 459. (23) Jungclas, Hartmut; Danigel, Harald; Schmidt. Lothar; Dellbrugge, Jorg Org. Mass Spectrom. 198P9 17, 499. (24) Blakley, C. R.; Carmody, J. J.; Vestal, M. L. Anal. Chem. 1980, 5 2 , 1636. (25) McNeal, C. J.; Macfarlane, R. D.; Thurston, E. L. Anal. Chem. 1979, 5 7 , 2036. (26) Schueler, B.; Krueger, F. R. Org. Mass Spectrom. 1979, 74, 439. (27) Stoll, R.; Rollgen, F. W. Org. Mass Spectrom. 1981, 76, 72.

RECEIVED June 30,1983. Accepted October 3,1983. This work was supported by the Institute of the General Medical Sciences (NIH)under Grant GM 29451 and the Robert A. Welch Foundation.