- Email: [email protected]
Thermal aspects of laser desorption mass spectrometry
International
Elsevier
Journal
Scientific
THERMAL
G.J.Q. F03f (The
(First
VAN
of
Mass
Publishing
ASPECTS
DER
Institute
PEYL,
Spectrometry
Company,
and
OF LASER
Atomic
10
September
and
Physics,
-
Molecular
and
P-G.
Physics.
42
Printed
DESORPTION
J. HAVERKAMP
for
Zon
Amsterdam
(1962)
125-141
125
in the Netherlands
MASS
SPECTROMETRY
KISTEMAKER
h’ruisiaan
407,
1098
SJAmsterdam
Netherlands)
received
1981;
in final
form
29 October
19Sl)
_4BSTR_4CT QuasimoIecular and fragment-ion formation of sucrose in laser desorption mass spectrometry (LDMS) have been investigated as functions of laser intensity and substrate material_ The ion currents generated from stainless-steel and quartz substrates show differing dependences on radiated laser energy_ However, the ion currents show highly congruent behavior as a function of the calculated surface temperature resulting from the applied laser energy. This encourages a thermal interpretation of LDMS results. Thermal desorption of quasimolecuiar ions from the substrate seems to be unlikely and a gas-phase ion-
ization model is proposed_
INTRODUCTION
During the last decade several ‘soft’ ionization techniques have been introduced in organic mass spectrometry for the analysis of so-called thermally labile, low-vapor-pressure compounds_ The most widely applied techniques nowadays are field desorption [l] and direct chemical-ionization mass spectrometry [2]_ Also laser desorption mass spectrometry (LDMS) [ 31 and ion-induced desorption mass spectrometry [4] have been promoted as ionization techniques with promising potential in analytical mass spectrometry. Although these techniques are quite different in their experimental arrangements they yield highly similar spectra for various classes of general feature in the biologically important samples_ The most interesting mass spectra is that instead of radical molecular ions, quasimolecular ions are observed that consist of molecule--alkali-ion complexes. This points to an ionization process based on ion-molecule reactions rather than on electronic excitation of the molecules. To date, the sequence of ionization and desorption and the energetics of the desorption process have remained unclear, and it is especially unclear whether the processes are themal or nonthermalThe most consistent description of the ionization and desorption processes has been given for field desorption [5]_ Here ion formation and desorption take place via field-enhanced desolvation_ It is assumed that dur0020-7SS1/S2/0000-0000/$02_75
@
1982
Pkevier
Scientific
Publishing
Company
ing this process the sample is in thermal equilibrium; the major influence of temperature is to enhance the mobility of the sample_ In direct chemical ionization the sequence of ionization and desorption has not been fully elucidated_ Complex ions can be formed on the surface as well as in the reagent gas after evaporation_ It seems, however, that the desorption process is of a thermal nature_ A more obscure situation is found for the more violent techniques where the samples are irradiated by beams of energetic particles: ions, neutral species or photons. Under ion bombardment spike temperatures ranging from a few thousand up to several tens of thousands of degrees have been calculated at the impact regions from the kinetic energy distributions of the desorbed ions 161. The highly fragmented ionic species in the low mass range of these spectra can be explained by these high temperatures_ At the same time, however, large abundances of quasimolecular ions are found in the higher mass range of the spectra_ It is highly improbable that both kinds of ions can have been formed under the same conditions. Therefore the concept of temperature was abandoned in the interpretation of results obtained by ion and photon bombardment_ Nonthermal processes were introduced, such as plasmon- 1’71 and shockwave-stimulated ES] desorption. In LDMS also a thermal nature for the ionization and desorption processes has been viewed with great scepticism for some time. Nevertheless, in early work on faser evaporation and ionization of metals a purely thermal concept was found sufficient to describe the experimental results [9] _ Evaporation rates and degrees of ionization could be related to calculated temperatures of the sample material. Laser desorption of some relatively stable organic compounds, such as quaternary ammonium and phosphonium salts, has already been explained in terms of thermal desorption [lo]_ Also the thermal desorption of quasimolecular ions from glucose, adenosine and a crown ether has been confirmed experimentally [Xl], while other results [X2--15] have pointed to possible thermal desorption of sucrose (-complexes). This encouraged a more detailed study of the correlation between qua&molecular ion currents and surface temperatures in LDMS, The present evaluation is performed on the basis of new esperimental data concerning the variation of ion currents as a function of laser intensity_ SURFACE-TEMPERATURE
CALCULATIONS
In this Section calculations are made concerning the place- and tirnedependent temperature distribution on a substrate surface subjected to pulsed laser irradiation_ Our first assumption is that the laser beam has a Gaussian intensity distribution. The laser intensity flux I(r, t) as a function of place and time is then given by l(r, Q = I,,
p(t) expf--r;“ld*)
(1)
where 1,,X is the maximum laser intensity flux, p(t) represents the pulse form of the laser pulse and eq-trdls one at maximum and d is the Gaussian
127
radius of the spot_ Because of the symmetry the place dependency is given only by the distancer from the center of the spot. If it is assumecithatthe reflection coefficient R is independent oftemperature,the flux Fin the surfacewillbe F(r, t)= (1 --R)I,,,p(t)exp(--r'/d')
(2)
Furthermore, it is assumed that the absorption of the material is very high, i.e., thatmostoftheheatisproduced at the surface. Also, thermal constants such as the thermal conductivity X and the thermal diffusivityK are taken to be independent of temperature. Using the model described by Ready [ZS] the temperature rise AZ'on the surface can be calculated as a function ofthelaserfhzxintensity I,,, as AT&,,,,
r, f) = [(1--R)~,,,d2/h](h'/71-)~2 (3)
Actual values of the parameters which were used in the calculations for stainless-steelsurfacesare[17] X = 17 WR-lm-*, K = 4X10e6 m2 s-l,and d = 1.2 X 10m4 m. The time variation of the laser intensity, p(t), cannot be approximated by a simple analytical expression for our type of laser, and therefore the experimentally determined form as shown in Fig. 1 has to be used. For short times t<< d'/4~ (in our case d"/4K= 9 X 10w4s), eqn. (3) reduces to
ATKiM*
r, f) = [((l --R)/X)(K/T)“*~
I,,,
exp(-r2/d’l
with
Fig. 1. Experimental
laser pulse form p(t)
as a function
of time.
P(t)
(4)
~?‘)/t’“~] dt' (5) 0 This is essentially the one-dimensional case treated by Ready [lS]_ In the framework of the assumptions given before, the factor in brackets is acon&ant, independent of the variables Imax, r-and t_ Note that AT increases linearlywithImax, themaximum fluxoftheincomingIaserbeam,underconstant focusing conditions and using a constant pulse form_ The placedependence of the temperature pulseis Gaussian andthetimedependenceisgiven by P(t) (Fig-Z), which was obtained by numerical evaluation of eqn_(5)_ The maximum surface temperature I', in the cenfxe of the laser spot 1s reached at the time t, when P(t) hasits maximum value. Because ofuncertaintiesin the actual values of A, K and x2 the calculated absolute values for AT can be inaccurate_ Therefore the temperature scale was calibrated using the following method_ Experimentally the laser pulse energy emelt is determined for which surface melting has started, i-e_T, = Tmelt, the melting temperature of the surface. It will be cIearthatthelaserfluxI,, islinearly proportional to the laser energy E using a constant pulse form p(t) and constant focusing conditions: Imax = cre with Q a constant. Using this reiation,eqn_(4)can berewrittenasafunction ofe: theconstanta! maythenbe incorporatedinthe factorin brackets.For E = E,,~* we cannow write
P(t)
= _i- [p(t-
T melt = To f ATl~rn,~t>
0, G-n)
(6)
this calibration point the effective value of [oL((~- R)/h)(~/r)“~] be determined and accordingto eqn(4)temperaturescannow be calculatedasafunction ofe, rand tprovidedthate< E,,.~ and t<< d2/4K. Applic:ation of these surface-temperature calculations to LDMS results requires a numberofassumptionsFirst,itisassumed that the presence of an From
P(t,)
can
1.d
I
_
Cl 0
( 2
1
t
3
4
t (ws)
Fig. 2_ Time eqn_ (5).
dependence
P(t)
of
the
temperature
rise
AT.
calculated
according
to
129
organic sample layer does not influence the temperature distribution on the substrate_ This means that most of the laser energy must be absorbed in the substrate material and not in the organic layer. Furthermore, the heat flux into the organic layer should be small compared to the flux in the substrate. It is nonetheless assumed that this flux is sufficient to heat the sample to the momentary substrate temperatureAdopting a simple thermionic process, as can be assumed for alkali ion emissions, the number dN of ions leaving a surface area d0 with a temperature T during time dt is equal to d.N = Cn’ exp(-AE/kT)
d0 dt
(‘7)
In this Polanyi-Wigner model [18] n is the number of particles per unit area and I the reaction order. The next assumption is that n is almost constant; that is, there is always a layer of particles left on the surface. To find the total number N of desorbed particles an integration has to be made over the surface and over time N = 27r Cn’ 7 0
fw exp [-AE/kT(r, 0
f)] r dr d t
(8)
A numerical integration using the temperature distribution from eqn. (4) and the K and X values for stainless steel revealed that the total number of particles desorbed per laser pulse can be approximated by N = exp(-AE*/kT,)
(9)
where the effective desorption ener,qy AE* deviates by less than 20% from the real desorption energy AE. The accuracy of this approximation is
Fig. 3_ Calculated number of ions desorbed by a laser surface temperature T,. The activation energies used (a) &Y = 0.69 eV and (b) &Z = 0.26 eV_
pulse as a function of the maximum in the thermal desorption model are
130
demonstrated by Fig_ 3,where calculated values ofN are plotted as afunction of L/Z', overthetemperature range700K< T,<1800Kfortwovalues of AE. This relationship between the integralnumberofdesorbed particles andthe maximumsnrfacetemperatureisnotcriticallydependentonthe laser pulse form,and the incorporation of a linear temperature dependence in the pre-exponential factorC in eqn_(7) checked numerically_ Itwillbe evident that interpretation ofexperimental data on the basis ofthisrelationship results in an approximate activation energy for thermionic desorption which underthe zsumptions madecan differ by -20% fromthetruevalue. EXPERIMENTAL
The experimental arrangement is shown in Fig.4 and has been described previously [lS]_Theion sourceis oftheconvent&xxxlNiertype.I-Iowever,for LD work the electron beam is switched off. The laser beam and sample can be admitted to the source via separate entrance ports_ Focusing of part of the ion spectrum onto the simultaneous ion detector is performed by the combined action of the electricaland magnetic quadrupoles and the sector magnet_ The detector consists of a Chevron CEMAdetectorand aphosphor screen. Light signals produced after arrivalof ions at the detectoraretransported out of the vacuum via a glass-fibersystem followedbytwolensesto focus the image onto a Vidicon camera_ Using an optical multichannel
Sector
magnet -‘.
g=Z$?T Vidicon
----EJ
camera
I Fig_ 4. Esperimental
arrangement
of laser
desorption
mass
spectrometer
system_
131 analyzer the image could be stored for further processing_ Time-resolved measurements could be made by allowing ions to enter the sector magnet only during a certain time interval (2100 /-IS). The electronic circuit for this is triggered by the visible light produced in the pre-discharge in the laser_ The laser used was a Lumonics K-101-2 TEA CO, laser (wavelength ns and a maximum energy of 10.6 pm) with a nominal pulse length of -200 1 -J_ Only a small fraction of the laser energy (O-6-21 MJ) was focused onto a spot with an estimated Gaussian diameter of 0.24 mm using a Ge lens. This results in energy densities in the range 1.3-46 J CM-‘_ Mean power densities were produced in this way. The time depenof S-7 X 106--3.1 X lo* W cm-= dence (Fig. 1) of the laser pulse intensity was Measured using a photon drag monitor combined with a storage oscilloscope. Energies of the laser pulse were measured using a Joule meter. Laser energies could be varied by using a set of CaF2 absorber plates and a continuous-polarization attenuator. was used as a sample material and as substrate Sucrose (Merck art. 7651) materials we chose stainless steel and quartz_ Best results were obtained if the surfaces were ground and chemicahy cleaned before applying the sample. An amount of 10 lug of sucrose dissolved in 10 ,~l of methanol was deposited onto the substrate (surface area 20 mm’) using a syringe. Ten drops each of 1~1 were deposited to obtain a more homogeneous distribution of the sample on the surface_ No sodium in the form of NaCl was added to the sample because addition of NaCl up to a relative concentration NaClr sucrose = 100 I 1 did not result in an appreciably increased ion current.
RESULTS
Qua&molecular ions could be measured only for a relatively smd range of laser pulse energies_ Lower laser energy limits were posed by correspondingly Pow ion signals, whereas higher limits were posed by ion-source instabilities laser desorption mass resulting from high ion currents_ In Fig. 5 a typical
Sucrose
LM-NCLl-
Cl%Kl-
181
-
l400
Fig_ 5_ Typical
laser
desorption
mass
spectrum
of sucrose
(MW
342).
132 TABLE1 Relative changes ainionintensity Laserenergy (mJ)
Mean m/z
with increasinglaserenergy
relativechange 23
m/z
3.85
m/z
203
m/z
Stainlesssteel 2.29+ 3-29 3-29' 4-98 4_9S-+ 6.36 6.369.76 9.76 +14-O 14.0 -21.2
20 4-l 2.6 2.1 O-55 l-4
14 2.2 1.9 1.6 1.0 -b
14 l-4 1.6 l-3 1.1 -
l-4 2.3. 2.1 2.0 l-3 O-66
Quartz 0.60+ 0.86-t l.OO-+
0.86 1.31 1.31
25 3.8 2.2
12 -
12 -
l-8 2.6 -
1.31+ 1.31'
1.67 2.00
2.2 -
2.8
3.2
1.1 -
X.672.00+
2.56 2.56
1-s -
0.91
1.1
O-93 -
2.56 -+ 3.67-t
3.67 5.56
2.0 2.0
0.64 0.93
o-59 0.67
0.93 -
365
a Average precision 15%. b Not determined.
spectrum of sucrose ispresented.Isotopicpeaksarenotgiven becauseofthe limitedresolution (300)oftheinstrument_The spectrum is dominated in the molecularmassrangeby[M+Na]'~d[M+K]'peaksatm/z365and381, respectively_The main fragmentation observed is acleavage at the glycosidic linkage, whilenolessofH,O fiomthemolec~~arion is observed_ Na*and K" complexes ofthesefragmentsarepresentatrsz/z185,203 and m/z 201,219, respectively-Very intense signals due to Na* and K'are observedinthelow massrangeatm/z 23 and 39,respectively. Before presenting the results for the ion intensitiesof particular mass peaks asa function oflaserenergy,themethodofdataacquisitionwillfirst be explained_ At a fixed laser energy the ion intensitiesvary considerably froLm pulse to pulse because of, for example, inhomogeneities in the sample layer and substratesurface_BycomparingonIy intensities-fromtwo adjacent sample sp.otsand using a ground surfacethesevariationswereminimized. In two consecutive laser pulses two different laser energies were used and the relativechange oftheioncurrentsataparticularmassvaIuewasdetermined_ The difference in laser energies could not be made too large because ofthe limited dynamic range of the ion detector This procedure was repeated about35timesandthemeanrelativeincreaseinion~tensitywascalculated. ThesemeanvaluesarepresentedinTablel.
m/z
365
io- i
10 loserenergy
15 ( m J1
!O laserenergy
(mJ
5
0
m
.!
ZO‘
laserenergy
Fig_ 6_ Relative ion intensities less-steel substrate (0).
(m J )
as a function
of
laser
-I.
0 energy
5
for
a quartz
i5
(X
)
20
1
and
a stain-
The factors of Table 1 can now be multiplied to produce Fig_ 6, where the relative ion intensities at m/z 23, 185, 203 and 365 are plotted as function of laser energy_ The curves are normalized to the maximum ion intensities_ The absolute ion intensities of the quasimolecular ions are of the same order of magnitude as those of the fragment ions (m/z 185, 203) and are about two orders of magnitude smaller than the Nd ion intensities_ The number of Na’ ions arriving at the detector is estimated to be IO6 for a laser pulse of 5 mJ incident on a stainless-steel substrate_ With this value and an estimated extraction and transmission efficiency of IO-’ for the mass spectrometer system, we calculated the number of alkali ions produced to be 1O1r per pulse. The maxim um ion intensities are of the same order of magnitude for the two substrates. It will be clear from Fig. 6 that the ion intensity curves for stainless steel and quartz as a function of incident laser energy show different behavior, indicating that the substrate plays a major role in the process Perhaps these differences can be attributed partly to different temperatures being reached by the two substrates during laser pulses with the same incident laser energy. An indication of this is that although the melting temperatures of 1883 K for quartz and 1808 K for stainless steel [17] are very close,
134
the start of surface melting of the two substrates was observed laser energies: 2.4 mJ (%15%) for quartz and 5.6 mJ (215%) steel_
at different for stainless
DISCUSSION From the experimental results presented in this paper the following chara&e&tics of laser-induced ion currents from a stainless-steel substrate covered with a sucrose layer can be extracted_ (i) A minimum laser energy of -2 MJ is required to produce measurable currents of alkali ions. From this experimental ‘threshold’ the ion intensities increase with laser pulse energy up to a maximum value. A further increase of the laser energy does not lead to higher ion yields and sometimes even a slight decrease in ion intensity is observed(ii) Quasimolecular and fragment-ion yields are of the same order of magnitude, although a stronger increase of the fragment ion relative to the quasimolecular ion can be observed with increasing laser energy_ (iii) The Nd intensity exceeds the organic ion intensities by two orders of magnitude_ Nevertheless, the dependence on laser pulse energy is similar to that for the organic ions. These characteristics are also observed using a quartz substrate instead of a stainless-steel substrate; however, the onset of and the maximum in the ion currents occur at lower laser pulse energies and the increase of the ion currents with increasing laser energy is strongerComparison
with other
data
Only a few results have been published concerning the role of laser energy in LDMS of organic samples [20,21]. In some cases a minimum laser energy to produce quasimolecular ions can be extracted from the published results [21,23] _ However, the minimum laser energy appears to be highly dependent on the laser and sample systems used. A definite increase in ion currents with laser power density has also been reported_ In particular, Heresch et al. [Zl] observed non-linear behavior in the range of nominal power densities 2 X 103-lo6 W cm-‘. It is not easy to compare in more detail results obtained by using different lasing systems and sample preparations. The lack of knowledge concerning the underlying physico-chemical processes makes it difficult to select the “characteristic” laser param eter for comparing resultsMany LDMS data have, almost intuitively, been presented as a function of the applied laser power density, In this terminology it has been stated that a laser power density of -lo6 W cm-’ is needed to produce quasimolecular ions [24]. Indeed, many similar spectra have been recorded at power density values within two orders of magnitude of this level using quite different laser systems_ Although this might be taken to indicate that laser power density is
-
co2
Ruby Ruby cw co2 Nd-glass cw co2 .-__
G94 nm 1064 nm 10G411111 265 nni 347 nm 347 11111 265 nm lOG4~111 lOG4nm 1O.G,um 694 nm 694 nm 10,Gj.lm 1064 nm 1O.G/.~m l.O,GI’m
Wavelength
;o-K”l.O-” 8 x lo-” 4 x 10-s 3 x lo-” 3 x lo-4 5 10”” 10-l 1.5 x 1o-7
lo-” lo-” IO-” 1.5 x lo-” 3 x 10-s 3 x 10-s
Pulsetime (s)
5
x 10” 25 x 10” 2,s x 10’ >107 10s 10” 10s 10s >2,5 x 10” >3 x 10” 10” 10” 20 5 x 10” 10” 8,7 x 1O-7
Power density (Wcm-2)
25 25 0,15 3 3 Q8 1 0,2 0,12 300 300 >lOO 500 100 >1,3
50
Energydensity (J CC-~)
____..-
..-
--_
- _^
.
,-
..
._
-..
* Wherepossible,minimumenergydensitiesfor quasimolecularion formalion we presentedas extracted from the literature.
Znkett et nl, [25] Hereschet al. [ 211 cotter [31] Vastolaand Pirone [32] Mummaand Vastola[33] Stall and Rtillgcn[ 231 Kistemnkcret al, [ 221 co2
Ruby Nd-Y AG Nd-YAG Nd-YAG Ruby Ruby Nd-YAG Nd-YAG Nd-YAG
Dnveset al, [ 271
Hcinenet al, [28] Kupka et al. [29] Kruegerand Schwcler[ 301
Lasertype
Rcf,
Laserparametersu used in severalLDMSexperimentsusingorganicsamples
TABLE2
136
the characteristicparameter in LDMSthreerecently obtained results do not support this_ Firstly, LDMS spectra are also obtained at power densities in the range of 20-lo3 W cm-?- [22,23]_ Undertheselow-levelirradianceconditions a time delay with respect to the appliedlaserpulse was observed for ion current generation_ Secondly, in some experiments ion generation did not stop immediately after the laser pulse; effectively, ion generation occurred at zero laser power densi@ 125,26-J_ Thirdly, in several previous papers [21,23] and in particular in the present paper, a strong influence of the substrate material isreported onthelaserpowerdensity required for ion generation_ These observations lead to the conclusion that laser power density aloneis notsufficientto characterize the differentlasersused in LDMS_ The above-mentioned time behavior oftheion emissionandthesubstrate influence can betentativelyinterpreted onthebasis of the assumption that a minimum substrate temperature is required for ion emission. Therefore it seems worthwhile to take the energy depositioninthe sample/substrate system as the basic parameter_ However,there are many obstaclesinthe evaluation because of the fact that reflection and absorption coefficients for the samples at the various wavelengths and intensitiesare not available.Consequently we could calculate only the radiated laser energy per unit surface area used in the various experiments_ The quoted laser parameters, from whichtheradiated energy density was calculated,hadto be treated with caution because generally these values are estimated rather than measured. In Table 2 the calculated energy density values from the literaturearesummarized; they show a variation of four orders of magnitude_ This variation is small compared to the variation of seven orders of magnitude in the laser power density values used in these experiments_ The absolute variation is rather large but is acceptable in view oftheinaccuracy in thelaserparameters and different sampling systems used. It is striking that the three lasing systems used by Daves et al. [2'i] in one experiment delivered about the same energy density despite the quite different intensities, focusing conditions and pulse times. The high energy densities needed when using lowpower cw CC, lasers canbeexplainedbythelargepartoftheabsorbedlaser energy which is conducted away during these long pulses_ Otherhighvalues as used by Vastola et al. 132,331 and Kistemaker et al. (Nd-Glass laser) [22] cannot be rationalized at present. In conclusion, we statethatcomparison of different laser systems used in LDMS on the basis of the irradiated energy density seemsto bemore appropriate than onthebasis oflaserpower density. Surface temperatures The next step in the interpretation ofthe data on the basis of a thermal model is to determine the sample and substrate temperatures obtained in LDMS, However,because of the unknown reflection and absorption coefficients, thermal conductivities and heat capacities of the irradiated sample/
137
substrate systems, these temperatures cannot be calculated directly from the applied laser energies. Under the assumptions made in the surface-temperature calculations it has been shown that an approximately linear relationship exists between the maximum surface temperature and the incident laser energy. It has to be noted that this relation only holds for relatively short pulses (<10F3 s) and in a limited temperature range from ambient temperature to the melting point of the substrate material_ With the help of a calibration point, namely the laser energy required for heating the substrate to the melting point, we could approximate surface temperatures as a function of applied laser energy_ Of course, the temperature varies rapidly in time and place on the substrate and therefore we characterized the surface temperature by the maximum temperature reached in the centre of the irradiated spot. Conversion of the laser energy scale to this temperature scale leads to plots as shown in Fig. 7_ Note that only a small part of the experimental data is shown, namely data corresponding to calculated surface temperatures below the melting points of the substrates. Highly congruent ion intensity curves are obtained for stainless steel and quartz, so that it might be concluded that the surface temperature of the substrate is the relevant parameter in the desorption process. The onsets of the alkali and organic ion cur-
Fig_ 7. Relative ion intensities as a function of the maximum surface temperature T, of an irradiated quartz (X) and stainless-steel substrate (g)_ The solid lines result from fitting the experimental points to a thermal desorption model (see test).
138
rents occur at surface temperatures in the range 600-800 are
Gas-phase
quite
reasonable
with
respect
to a thermionic
K. These tempera-
of the alkali ions_ For this reason we adopted a thermionic emission model to describe the alkali ion current. As outlined in the surface-temperature calculations, for thermionic emission can be obapproximate activation energies A??’ tained from a plot of ion current versus maximum surface temperature_ A least-squares fit of AI? to the experimental data in Fig. 7 leads to AE’ = 0-8 and O-6 eV for Na’ emission from the stainless-steel and quartz substrates, respectively_ Interpretation of these values on the basis of ionization potentials and work functions is tentative because of the undefined surface conditions during the emission process_ If this thermal emission model is also used to interpret the results for quasimolecular ion emission, it is necessary to assume that these ion-molecule complexes can survive temperatures in the rage 600-800 K. This does not seem to be a reasonable assumption unless the quasimolecular ions are much more stable than the neutral molecules. Nevertheless, the esperimental results for the organic ions in Fig_ 7 can be described reasonably by a similar relationship (eqn_ (9)) between maximum surface temperature and integrated ion currents as used for alkali ions_ The resulting values for AEf , 0.2 eV for the quasimolecular ions and -0-6 eV for the fragment ions, are used only for quantification of the increase of ion current with temperature. This facilitates comparison of the various ion currents_ Concluding this Section, it will be clear that differences in ion current intensities using different laser energies on stainless-steel and quartz substrates can be qualitatively attributed to the different temperature responses of the substrates_ tures
origin
conzplesation
The concept of a simple thermal desorption process for ion-molecule compIesation had to be abandoned because of the fact that quasimolecular ions are observed only at high surface temperaturesThe presence of a strong alkali ion current which always accompanies the quasimolecular ions may point to the prerequisite of an intense Nd ion current for cationization reactions_ This leads to the postulation of a gas-phase complexation reaction for the formation of quasimolecular ions. In this hypothesis sucrose particles are evaporated at relatively low surface temperatures during the early stage of the laser pulse from the centre of the laser spot and later from the edges of the irradiated spot_ Thermionic Nd ions are emitted only from the hot central part of the spot_ This model supposes the evaporation of intact sucrose molecules_ Although for some time this assumption was considered to be unrealistic, various esperimental data obtained subsequently support this idea [ 34,15-J_ This qualitative picture of evaporation and gas-phase complesation describes adequately the experimental data reported in this paper_
139
However, quantitative interpretation of the data using this model has not to date been successful. The main obstacles are the unknown reaction kinetics of the desorpi;ion and fragmentation processes, which determine the gasphase concentration of sucrose. A simple molecular evaporation and ionization model does not suffice to explain the results quantitatively, as is shown in the following paragraph. Suppose that the sucrose sample at the irradiated spot is evaporated completely as intact molecules and that the sucrose cloud is overhauled by the much faster (more energetic) alkali ions emitted somewhat later. The probability PC that an alkali ion forms a complex with a sucrose molecule is then given by P, = I-
exp(--n50)
(10)
where n, is the number of sucrose molecules vaporized per unit surface area and 0 is the cross-section for complexation. Approximate values for the parameters under the present experimental conditions are n, = 10” molecules cm-’ and (T = 150 A2 (Langevin cross-section) 1353 _ These values lead to P, -r 1; thus every alkali ion can form a complex, provided that a sufficient number of gas-phase molecules are available: rx, > nxa+ (>zxa+ is the number of sodium ions emitted per unit surface -Lea). It will be clear that this condition can be fulfilled at relatively low substrate temperatures and consequently the behavior of the quasimolecular ion current should be linearly proportional to the number of alkali ions emitted at low laser energies_ At higher laser energies the number of alkali ions may esceed the number of sucrose molecules and the quasimolecular ion current will be proportiondl to the sucrose gas-phase density. From Figs. 6 and 7 it is observed that the quasimolecular ion current is not linearly proportional to the alkali ion current, indicating that rzNa+ > IZ,. This conclusion contradicts the value of rzxa+ 2: 10’” ions cm-’ at 5 mJ, as deduced from the measured Na+ current. Obviously, in the contest of this model, the number of gas-phase sucrose molecules is highly overestimated; vaporization of the sample does not proceed uniquely via molecular evaporation_ This point of view is supported by a consideration of the heat transport from the heated substrate into the sample under :he assumption that direct heating of the sucrose sample by laser irradiation is negligible. At the high heating rates in the centre of the laser spot (>5 X 10” K s-l), the ener-gy transport via thermal conduction is not sufficient to heat the sample layer uniformly to the momentary substrate temperature. This leads to high temperature-gradients in the sample layer. First-crder calculations support the possibility of sample vaporization and fragmentation of the layer in close contact with the substrate while the top layers are still in the solid state. The resulting explosive removal of sample material proceeds via cluster ejection rather than molecular evaporat.ion. Molecular evaporation can be expected from adjacent low-temperature areas_ The fact that in our esperiments no cluster ions could be detected may be explained by a mass distribution such
140
that most of the particles fall beyond the detection window of our instrument_ Another indication for the supply of sucrose from the outer regions of the irradiated spot can be found in the work of Heresch et al. [21]. A much improved quasimolecular ion current was obtained by focusing the laser beam onto surface irregularities with dimensions comparable to tne laser spot_ These irregularities were introduced by mechanically scratching the sample surface, apparently leaving ‘bare’ substrate at the centre of the laser spot_
In the foregoing discussion it has been shown that all of the present experimental results can, at least qualitatively, be explained by a thermal model where a minimum temperature has to be reached before quasimolecular ion formation can take place_ Minimum temperatures extracted from the experimental data are too high for thermal desorption of intact quasimolecular ions and favor a model where intact sucrose molecules and alkali ions combine in the gas phase. Furthermore, in the framework of such a model it is suggested that sucrose molecules involved in the complexation reaction _ _ ongmate mostly from the outermost areas of the laser spot. _4CKNOWLEDGMENTS
The
authors
are indebted
to Prof_ Dr_ J_ Kistemaker
for stimulating
discus-
sions_ This work is part of the research program of the Stichting voor Fundamenteel Onderzoek der Materie (Foundation for Fundamental Research on Matter) and was made possible by financial support from- the Nederlandse Organisatie voor Zuiverwetenschappelijk Onderzoek (Netherlands Organization for the Advancement of Pure Research)_ REFERENCES 1 2 3
H-R. Schulten, Int_ J_ Mass Spectrom. Ion Phys., 32 (1979) 97. D.F. Hunt, J. Shabanowitz, F-K. Botz and D.A. Brent, Anal. Chem., 49 (1977) 1160. P.G. Kistemaker, G.J.Q. van der Peyl and J_ Haverkamp, Adv_ Mass Spectrom-, (1981) to be published_ 4 (a) R.D. Macfarlane and D-F_ Torgerson, Int_ J. Mass Spectrom. Ion Phys., 21 (1976) 81; (b) R.J. Day, SE. Unger and R-G. Cooks, Anal. Chem., 52 (1980) 557A. 5 U. Gieswnan and F-W. Riillgen, Int. J. Mass Spectrom. Ion Phys., 38 (1981) 267. 6 R.D. Macfarlane and D-F. Torgerson, Phys. Rev. Lett., 36 (1976) 486. 7 (a) F-R. Krueger, 2. Naturforsch., 32a (1977) 1084; (b) F-R. Krueger, Surf. Sci., 86 (1979) 246. 8 R-D_ Marfarhme, Proc- 26’h Ann_ Conf. on Mass Spectrometry and Allied Topics, St. Louis, MO, 1978, p. 490. 9 J-F. Ready, Effects of High Power Laser Radiation, Academic Press, New York, 1971, p_ 140_
141 10 11 12 13 14 15 16 17
R. St011 and F.W. Riillgen, J. Chem. Sot., Chem. Commun., (1980) 789. R. St011 and F.W. RSllgen, Org. Mass Spectrom., 16 (1981) 72. U. Giessmann and F.W. RSIlgen, Org. Mass Spectrom., 11 (1976) 1094. W-R_ Anderson, Jr., W_ Frick, G-D_ Daves, Jr., D-F_ Barofsky, D. Chang kers, J. Am. Chem. Sot., 100 (1978) 1974. R-J_ Cotter and C. Fenselau, Biomed. Mass Spectrom., 6 (1979) 287. R-J_ Cotter, Anal. Chem., 52 (1980) 1589A.
J.F. Ready, pp_ 74.
Effects
of High Power Laser Radiation,
Academic
and
K.
Press, New York,
Fol-
1971,
29 30 31 32
R.C_ Weast (Ed.), CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, 1980_ D.A. King, Surf. Sci., 47 (1975) 384. M.A. Posthumus, P.G. Kistemaker, H.L. Meuzelaar and M.C. Ten Noever de Brauw, Anal. Chem., 50 (1978) 985_ F. UnsSld, G. Renner, F. Hillenkamp and R. Nitsche, Adv. Mass Spectrom., 7B (1978) 1423. F. Heresch, E.R. Schmid and J.F.K. Huber, Anal. Chem., 52 (1980) 1803. P.G. Kistemaker, M.M.J. Lens, G.J.Q. van der Peyl and A.J.H. Boerboom, Adv. Mass Spectrom., 8A (1980) 928. R. Stoil and F-W_ Rbllgen, Org. Mass Spectrom-, 14 (1979) 642. R-J. Cotter, Anal_ Chem., 53 (1981) 719. D_ Zakett, A-E. Schoen, R-G. Cooks and P-H. Hemberger, J_ Am_ Chem. Sot., 103 (1981) 1295_ G.J.Q. van der Peyl, K_ Isa, J. Haverkamp and P.G. Kistemaker, Nucl. Instrum. Methods, accepted for publication_ G-D. Daves, Jr., T-D. Lee, W-R. Anderson, Jr., D.F. Barofsky, G-A. Massey, J-C. Johnson and P-A_ Pincus, Adv. Mass Spectrom., 8A (1980) 1012. H.J. Heinen, S. Meier, H. Vogt and R_ Wechsung, Adv. Mass Spectrom., S-4 (1980) 942. K.-D. Kupta, F. Hillenkamp and Ch. Schiller, Adv. Mass Spectrom., 8A (1980) 935. F-R_ Krueger and B. Schueler, Adv. Mass Spectrom., 8A (1980) 918. R-J_ Cotter, Anal. Chem., 52 (1980) 1767. F.J. Vastola and A.J. Pirone, Adv. Mass Spectrom., 4 (1968) 107.
33
R-O_ Mumma
34
G-J-Q_ van der Peyl. K. Isa, J_ Haverkamp and P-G. 16 (1981) 416. M. Henchman, in J.L. Franklin (Ed.), Ion-Molecule York, 1972, Ch. 5.
18 19 20 21 22 23 24 25 26 27 28
35
and F-J. Vastola,
Org. Mass Spech-om.,
6 (1972) Kistemaker,
1373. Org. Mass Spectrom..
Reactions,
Plenum
Press,
New
Elsevier
Journal
Scientific
THERMAL
G.J.Q. F03f (The
(First
VAN
of
Mass
Publishing
ASPECTS
DER
Institute
PEYL,
Spectrometry
Company,
and
OF LASER
Atomic
10
September
and
Physics,
-
Molecular
and
P-G.
Physics.
42
Printed
DESORPTION
J. HAVERKAMP
for
Zon
Amsterdam
(1962)
125-141
125
in the Netherlands
MASS
SPECTROMETRY
KISTEMAKER
h’ruisiaan
407,
1098
SJAmsterdam
Netherlands)
received
1981;
in final
form
29 October
19Sl)
_4BSTR_4CT QuasimoIecular and fragment-ion formation of sucrose in laser desorption mass spectrometry (LDMS) have been investigated as functions of laser intensity and substrate material_ The ion currents generated from stainless-steel and quartz substrates show differing dependences on radiated laser energy_ However, the ion currents show highly congruent behavior as a function of the calculated surface temperature resulting from the applied laser energy. This encourages a thermal interpretation of LDMS results. Thermal desorption of quasimolecuiar ions from the substrate seems to be unlikely and a gas-phase ion-
ization model is proposed_
INTRODUCTION
During the last decade several ‘soft’ ionization techniques have been introduced in organic mass spectrometry for the analysis of so-called thermally labile, low-vapor-pressure compounds_ The most widely applied techniques nowadays are field desorption [l] and direct chemical-ionization mass spectrometry [2]_ Also laser desorption mass spectrometry (LDMS) [ 31 and ion-induced desorption mass spectrometry [4] have been promoted as ionization techniques with promising potential in analytical mass spectrometry. Although these techniques are quite different in their experimental arrangements they yield highly similar spectra for various classes of general feature in the biologically important samples_ The most interesting mass spectra is that instead of radical molecular ions, quasimolecular ions are observed that consist of molecule--alkali-ion complexes. This points to an ionization process based on ion-molecule reactions rather than on electronic excitation of the molecules. To date, the sequence of ionization and desorption and the energetics of the desorption process have remained unclear, and it is especially unclear whether the processes are themal or nonthermalThe most consistent description of the ionization and desorption processes has been given for field desorption [5]_ Here ion formation and desorption take place via field-enhanced desolvation_ It is assumed that dur0020-7SS1/S2/0000-0000/$02_75
@
1982
Pkevier
Scientific
Publishing
Company
ing this process the sample is in thermal equilibrium; the major influence of temperature is to enhance the mobility of the sample_ In direct chemical ionization the sequence of ionization and desorption has not been fully elucidated_ Complex ions can be formed on the surface as well as in the reagent gas after evaporation_ It seems, however, that the desorption process is of a thermal nature_ A more obscure situation is found for the more violent techniques where the samples are irradiated by beams of energetic particles: ions, neutral species or photons. Under ion bombardment spike temperatures ranging from a few thousand up to several tens of thousands of degrees have been calculated at the impact regions from the kinetic energy distributions of the desorbed ions 161. The highly fragmented ionic species in the low mass range of these spectra can be explained by these high temperatures_ At the same time, however, large abundances of quasimolecular ions are found in the higher mass range of the spectra_ It is highly improbable that both kinds of ions can have been formed under the same conditions. Therefore the concept of temperature was abandoned in the interpretation of results obtained by ion and photon bombardment_ Nonthermal processes were introduced, such as plasmon- 1’71 and shockwave-stimulated ES] desorption. In LDMS also a thermal nature for the ionization and desorption processes has been viewed with great scepticism for some time. Nevertheless, in early work on faser evaporation and ionization of metals a purely thermal concept was found sufficient to describe the experimental results [9] _ Evaporation rates and degrees of ionization could be related to calculated temperatures of the sample material. Laser desorption of some relatively stable organic compounds, such as quaternary ammonium and phosphonium salts, has already been explained in terms of thermal desorption [lo]_ Also the thermal desorption of quasimolecular ions from glucose, adenosine and a crown ether has been confirmed experimentally [Xl], while other results [X2--15] have pointed to possible thermal desorption of sucrose (-complexes). This encouraged a more detailed study of the correlation between qua&molecular ion currents and surface temperatures in LDMS, The present evaluation is performed on the basis of new esperimental data concerning the variation of ion currents as a function of laser intensity_ SURFACE-TEMPERATURE
CALCULATIONS
In this Section calculations are made concerning the place- and tirnedependent temperature distribution on a substrate surface subjected to pulsed laser irradiation_ Our first assumption is that the laser beam has a Gaussian intensity distribution. The laser intensity flux I(r, t) as a function of place and time is then given by l(r, Q = I,,
p(t) expf--r;“ld*)
(1)
where 1,,X is the maximum laser intensity flux, p(t) represents the pulse form of the laser pulse and eq-trdls one at maximum and d is the Gaussian
127
radius of the spot_ Because of the symmetry the place dependency is given only by the distancer from the center of the spot. If it is assumecithatthe reflection coefficient R is independent oftemperature,the flux Fin the surfacewillbe F(r, t)= (1 --R)I,,,p(t)exp(--r'/d')
(2)
Furthermore, it is assumed that the absorption of the material is very high, i.e., thatmostoftheheatisproduced at the surface. Also, thermal constants such as the thermal conductivity X and the thermal diffusivityK are taken to be independent of temperature. Using the model described by Ready [ZS] the temperature rise AZ'on the surface can be calculated as a function ofthelaserfhzxintensity I,,, as AT&,,,,
r, f) = [(1--R)~,,,d2/h](h'/71-)~2 (3)
Actual values of the parameters which were used in the calculations for stainless-steelsurfacesare[17] X = 17 WR-lm-*, K = 4X10e6 m2 s-l,and d = 1.2 X 10m4 m. The time variation of the laser intensity, p(t), cannot be approximated by a simple analytical expression for our type of laser, and therefore the experimentally determined form as shown in Fig. 1 has to be used. For short times t<< d'/4~ (in our case d"/4K= 9 X 10w4s), eqn. (3) reduces to
ATKiM*
r, f) = [((l --R)/X)(K/T)“*~
I,,,
exp(-r2/d’l
with
Fig. 1. Experimental
laser pulse form p(t)
as a function
of time.
P(t)
(4)
~?‘)/t’“~] dt' (5) 0 This is essentially the one-dimensional case treated by Ready [lS]_ In the framework of the assumptions given before, the factor in brackets is acon&ant, independent of the variables Imax, r-and t_ Note that AT increases linearlywithImax, themaximum fluxoftheincomingIaserbeam,underconstant focusing conditions and using a constant pulse form_ The placedependence of the temperature pulseis Gaussian andthetimedependenceisgiven by P(t) (Fig-Z), which was obtained by numerical evaluation of eqn_(5)_ The maximum surface temperature I', in the cenfxe of the laser spot 1s reached at the time t, when P(t) hasits maximum value. Because ofuncertaintiesin the actual values of A, K and x2 the calculated absolute values for AT can be inaccurate_ Therefore the temperature scale was calibrated using the following method_ Experimentally the laser pulse energy emelt is determined for which surface melting has started, i-e_T, = Tmelt, the melting temperature of the surface. It will be cIearthatthelaserfluxI,, islinearly proportional to the laser energy E using a constant pulse form p(t) and constant focusing conditions: Imax = cre with Q a constant. Using this reiation,eqn_(4)can berewrittenasafunction ofe: theconstanta! maythenbe incorporatedinthe factorin brackets.For E = E,,~* we cannow write
P(t)
= _i- [p(t-
T melt = To f ATl~rn,~t>
0, G-n)
(6)
this calibration point the effective value of [oL((~- R)/h)(~/r)“~] be determined and accordingto eqn(4)temperaturescannow be calculatedasafunction ofe, rand tprovidedthate< E,,.~ and t<< d2/4K. Applic:ation of these surface-temperature calculations to LDMS results requires a numberofassumptionsFirst,itisassumed that the presence of an From
P(t,)
can
1.d
I
_
Cl 0
( 2
1
t
3
4
t (ws)
Fig. 2_ Time eqn_ (5).
dependence
P(t)
of
the
temperature
rise
AT.
calculated
according
to
129
organic sample layer does not influence the temperature distribution on the substrate_ This means that most of the laser energy must be absorbed in the substrate material and not in the organic layer. Furthermore, the heat flux into the organic layer should be small compared to the flux in the substrate. It is nonetheless assumed that this flux is sufficient to heat the sample to the momentary substrate temperatureAdopting a simple thermionic process, as can be assumed for alkali ion emissions, the number dN of ions leaving a surface area d0 with a temperature T during time dt is equal to d.N = Cn’ exp(-AE/kT)
d0 dt
(‘7)
In this Polanyi-Wigner model [18] n is the number of particles per unit area and I the reaction order. The next assumption is that n is almost constant; that is, there is always a layer of particles left on the surface. To find the total number N of desorbed particles an integration has to be made over the surface and over time N = 27r Cn’ 7 0
fw exp [-AE/kT(r, 0
f)] r dr d t
(8)
A numerical integration using the temperature distribution from eqn. (4) and the K and X values for stainless steel revealed that the total number of particles desorbed per laser pulse can be approximated by N = exp(-AE*/kT,)
(9)
where the effective desorption ener,qy AE* deviates by less than 20% from the real desorption energy AE. The accuracy of this approximation is
Fig. 3_ Calculated number of ions desorbed by a laser surface temperature T,. The activation energies used (a) &Y = 0.69 eV and (b) &Z = 0.26 eV_
pulse as a function of the maximum in the thermal desorption model are
130
demonstrated by Fig_ 3,where calculated values ofN are plotted as afunction of L/Z', overthetemperature range700K< T,<1800Kfortwovalues of AE. This relationship between the integralnumberofdesorbed particles andthe maximumsnrfacetemperatureisnotcriticallydependentonthe laser pulse form,and the incorporation of a linear temperature dependence in the pre-exponential factorC in eqn_(7) checked numerically_ Itwillbe evident that interpretation ofexperimental data on the basis ofthisrelationship results in an approximate activation energy for thermionic desorption which underthe zsumptions madecan differ by -20% fromthetruevalue. EXPERIMENTAL
The experimental arrangement is shown in Fig.4 and has been described previously [lS]_Theion sourceis oftheconvent&xxxlNiertype.I-Iowever,for LD work the electron beam is switched off. The laser beam and sample can be admitted to the source via separate entrance ports_ Focusing of part of the ion spectrum onto the simultaneous ion detector is performed by the combined action of the electricaland magnetic quadrupoles and the sector magnet_ The detector consists of a Chevron CEMAdetectorand aphosphor screen. Light signals produced after arrivalof ions at the detectoraretransported out of the vacuum via a glass-fibersystem followedbytwolensesto focus the image onto a Vidicon camera_ Using an optical multichannel
Sector
magnet -‘.
g=Z$?T Vidicon
----EJ
camera
I Fig_ 4. Esperimental
arrangement
of laser
desorption
mass
spectrometer
system_
131 analyzer the image could be stored for further processing_ Time-resolved measurements could be made by allowing ions to enter the sector magnet only during a certain time interval (2100 /-IS). The electronic circuit for this is triggered by the visible light produced in the pre-discharge in the laser_ The laser used was a Lumonics K-101-2 TEA CO, laser (wavelength ns and a maximum energy of 10.6 pm) with a nominal pulse length of -200 1 -J_ Only a small fraction of the laser energy (O-6-21 MJ) was focused onto a spot with an estimated Gaussian diameter of 0.24 mm using a Ge lens. This results in energy densities in the range 1.3-46 J CM-‘_ Mean power densities were produced in this way. The time depenof S-7 X 106--3.1 X lo* W cm-= dence (Fig. 1) of the laser pulse intensity was Measured using a photon drag monitor combined with a storage oscilloscope. Energies of the laser pulse were measured using a Joule meter. Laser energies could be varied by using a set of CaF2 absorber plates and a continuous-polarization attenuator. was used as a sample material and as substrate Sucrose (Merck art. 7651) materials we chose stainless steel and quartz_ Best results were obtained if the surfaces were ground and chemicahy cleaned before applying the sample. An amount of 10 lug of sucrose dissolved in 10 ,~l of methanol was deposited onto the substrate (surface area 20 mm’) using a syringe. Ten drops each of 1~1 were deposited to obtain a more homogeneous distribution of the sample on the surface_ No sodium in the form of NaCl was added to the sample because addition of NaCl up to a relative concentration NaClr sucrose = 100 I 1 did not result in an appreciably increased ion current.
RESULTS
Qua&molecular ions could be measured only for a relatively smd range of laser pulse energies_ Lower laser energy limits were posed by correspondingly Pow ion signals, whereas higher limits were posed by ion-source instabilities laser desorption mass resulting from high ion currents_ In Fig. 5 a typical
Sucrose
LM-NCLl-
Cl%Kl-
181
-
l400
Fig_ 5_ Typical
laser
desorption
mass
spectrum
of sucrose
(MW
342).
132 TABLE1 Relative changes ainionintensity Laserenergy (mJ)
Mean m/z
with increasinglaserenergy
relativechange 23
m/z
3.85
m/z
203
m/z
Stainlesssteel 2.29+ 3-29 3-29' 4-98 4_9S-+ 6.36 6.369.76 9.76 +14-O 14.0 -21.2
20 4-l 2.6 2.1 O-55 l-4
14 2.2 1.9 1.6 1.0 -b
14 l-4 1.6 l-3 1.1 -
l-4 2.3. 2.1 2.0 l-3 O-66
Quartz 0.60+ 0.86-t l.OO-+
0.86 1.31 1.31
25 3.8 2.2
12 -
12 -
l-8 2.6 -
1.31+ 1.31'
1.67 2.00
2.2 -
2.8
3.2
1.1 -
X.672.00+
2.56 2.56
1-s -
0.91
1.1
O-93 -
2.56 -+ 3.67-t
3.67 5.56
2.0 2.0
0.64 0.93
o-59 0.67
0.93 -
365
a Average precision 15%. b Not determined.
spectrum of sucrose ispresented.Isotopicpeaksarenotgiven becauseofthe limitedresolution (300)oftheinstrument_The spectrum is dominated in the molecularmassrangeby[M+Na]'~d[M+K]'peaksatm/z365and381, respectively_The main fragmentation observed is acleavage at the glycosidic linkage, whilenolessofH,O fiomthemolec~~arion is observed_ Na*and K" complexes ofthesefragmentsarepresentatrsz/z185,203 and m/z 201,219, respectively-Very intense signals due to Na* and K'are observedinthelow massrangeatm/z 23 and 39,respectively. Before presenting the results for the ion intensitiesof particular mass peaks asa function oflaserenergy,themethodofdataacquisitionwillfirst be explained_ At a fixed laser energy the ion intensitiesvary considerably froLm pulse to pulse because of, for example, inhomogeneities in the sample layer and substratesurface_BycomparingonIy intensities-fromtwo adjacent sample sp.otsand using a ground surfacethesevariationswereminimized. In two consecutive laser pulses two different laser energies were used and the relativechange oftheioncurrentsataparticularmassvaIuewasdetermined_ The difference in laser energies could not be made too large because ofthe limited dynamic range of the ion detector This procedure was repeated about35timesandthemeanrelativeincreaseinion~tensitywascalculated. ThesemeanvaluesarepresentedinTablel.
m/z
365
io- i
10 loserenergy
15 ( m J1
!O laserenergy
(mJ
5
0
m
.!
ZO‘
laserenergy
Fig_ 6_ Relative ion intensities less-steel substrate (0).
(m J )
as a function
of
laser
-I.
0 energy
5
for
a quartz
i5
(X
)
20
1
and
a stain-
The factors of Table 1 can now be multiplied to produce Fig_ 6, where the relative ion intensities at m/z 23, 185, 203 and 365 are plotted as function of laser energy_ The curves are normalized to the maximum ion intensities_ The absolute ion intensities of the quasimolecular ions are of the same order of magnitude as those of the fragment ions (m/z 185, 203) and are about two orders of magnitude smaller than the Nd ion intensities_ The number of Na’ ions arriving at the detector is estimated to be IO6 for a laser pulse of 5 mJ incident on a stainless-steel substrate_ With this value and an estimated extraction and transmission efficiency of IO-’ for the mass spectrometer system, we calculated the number of alkali ions produced to be 1O1r per pulse. The maxim um ion intensities are of the same order of magnitude for the two substrates. It will be clear from Fig. 6 that the ion intensity curves for stainless steel and quartz as a function of incident laser energy show different behavior, indicating that the substrate plays a major role in the process Perhaps these differences can be attributed partly to different temperatures being reached by the two substrates during laser pulses with the same incident laser energy. An indication of this is that although the melting temperatures of 1883 K for quartz and 1808 K for stainless steel [17] are very close,
134
the start of surface melting of the two substrates was observed laser energies: 2.4 mJ (%15%) for quartz and 5.6 mJ (215%) steel_
at different for stainless
DISCUSSION From the experimental results presented in this paper the following chara&e&tics of laser-induced ion currents from a stainless-steel substrate covered with a sucrose layer can be extracted_ (i) A minimum laser energy of -2 MJ is required to produce measurable currents of alkali ions. From this experimental ‘threshold’ the ion intensities increase with laser pulse energy up to a maximum value. A further increase of the laser energy does not lead to higher ion yields and sometimes even a slight decrease in ion intensity is observed(ii) Quasimolecular and fragment-ion yields are of the same order of magnitude, although a stronger increase of the fragment ion relative to the quasimolecular ion can be observed with increasing laser energy_ (iii) The Nd intensity exceeds the organic ion intensities by two orders of magnitude_ Nevertheless, the dependence on laser pulse energy is similar to that for the organic ions. These characteristics are also observed using a quartz substrate instead of a stainless-steel substrate; however, the onset of and the maximum in the ion currents occur at lower laser pulse energies and the increase of the ion currents with increasing laser energy is strongerComparison
with other
data
Only a few results have been published concerning the role of laser energy in LDMS of organic samples [20,21]. In some cases a minimum laser energy to produce quasimolecular ions can be extracted from the published results [21,23] _ However, the minimum laser energy appears to be highly dependent on the laser and sample systems used. A definite increase in ion currents with laser power density has also been reported_ In particular, Heresch et al. [Zl] observed non-linear behavior in the range of nominal power densities 2 X 103-lo6 W cm-‘. It is not easy to compare in more detail results obtained by using different lasing systems and sample preparations. The lack of knowledge concerning the underlying physico-chemical processes makes it difficult to select the “characteristic” laser param eter for comparing resultsMany LDMS data have, almost intuitively, been presented as a function of the applied laser power density, In this terminology it has been stated that a laser power density of -lo6 W cm-’ is needed to produce quasimolecular ions [24]. Indeed, many similar spectra have been recorded at power density values within two orders of magnitude of this level using quite different laser systems_ Although this might be taken to indicate that laser power density is
-
co2
Ruby Ruby cw co2 Nd-glass cw co2 .-__
G94 nm 1064 nm 10G411111 265 nni 347 nm 347 11111 265 nm lOG4~111 lOG4nm 1O.G,um 694 nm 694 nm 10,Gj.lm 1064 nm 1O.G/.~m l.O,GI’m
Wavelength
;o-K”l.O-” 8 x lo-” 4 x 10-s 3 x lo-” 3 x lo-4 5 10”” 10-l 1.5 x 1o-7
lo-” lo-” IO-” 1.5 x lo-” 3 x 10-s 3 x 10-s
Pulsetime (s)
5
x 10” 25 x 10” 2,s x 10’ >107 10s 10” 10s 10s >2,5 x 10” >3 x 10” 10” 10” 20 5 x 10” 10” 8,7 x 1O-7
Power density (Wcm-2)
25 25 0,15 3 3 Q8 1 0,2 0,12 300 300 >lOO 500 100 >1,3
50
Energydensity (J CC-~)
____..-
..-
--_
- _^
.
,-
..
._
-..
* Wherepossible,minimumenergydensitiesfor quasimolecularion formalion we presentedas extracted from the literature.
Znkett et nl, [25] Hereschet al. [ 211 cotter [31] Vastolaand Pirone [32] Mummaand Vastola[33] Stall and Rtillgcn[ 231 Kistemnkcret al, [ 221 co2
Ruby Nd-Y AG Nd-YAG Nd-YAG Ruby Ruby Nd-YAG Nd-YAG Nd-YAG
Dnveset al, [ 271
Hcinenet al, [28] Kupka et al. [29] Kruegerand Schwcler[ 301
Lasertype
Rcf,
Laserparametersu used in severalLDMSexperimentsusingorganicsamples
TABLE2
136
the characteristicparameter in LDMSthreerecently obtained results do not support this_ Firstly, LDMS spectra are also obtained at power densities in the range of 20-lo3 W cm-?- [22,23]_ Undertheselow-levelirradianceconditions a time delay with respect to the appliedlaserpulse was observed for ion current generation_ Secondly, in some experiments ion generation did not stop immediately after the laser pulse; effectively, ion generation occurred at zero laser power densi@ 125,26-J_ Thirdly, in several previous papers [21,23] and in particular in the present paper, a strong influence of the substrate material isreported onthelaserpowerdensity required for ion generation_ These observations lead to the conclusion that laser power density aloneis notsufficientto characterize the differentlasersused in LDMS_ The above-mentioned time behavior oftheion emissionandthesubstrate influence can betentativelyinterpreted onthebasis of the assumption that a minimum substrate temperature is required for ion emission. Therefore it seems worthwhile to take the energy depositioninthe sample/substrate system as the basic parameter_ However,there are many obstaclesinthe evaluation because of the fact that reflection and absorption coefficients for the samples at the various wavelengths and intensitiesare not available.Consequently we could calculate only the radiated laser energy per unit surface area used in the various experiments_ The quoted laser parameters, from whichtheradiated energy density was calculated,hadto be treated with caution because generally these values are estimated rather than measured. In Table 2 the calculated energy density values from the literaturearesummarized; they show a variation of four orders of magnitude_ This variation is small compared to the variation of seven orders of magnitude in the laser power density values used in these experiments_ The absolute variation is rather large but is acceptable in view oftheinaccuracy in thelaserparameters and different sampling systems used. It is striking that the three lasing systems used by Daves et al. [2'i] in one experiment delivered about the same energy density despite the quite different intensities, focusing conditions and pulse times. The high energy densities needed when using lowpower cw CC, lasers canbeexplainedbythelargepartoftheabsorbedlaser energy which is conducted away during these long pulses_ Otherhighvalues as used by Vastola et al. 132,331 and Kistemaker et al. (Nd-Glass laser) [22] cannot be rationalized at present. In conclusion, we statethatcomparison of different laser systems used in LDMS on the basis of the irradiated energy density seemsto bemore appropriate than onthebasis oflaserpower density. Surface temperatures The next step in the interpretation ofthe data on the basis of a thermal model is to determine the sample and substrate temperatures obtained in LDMS, However,because of the unknown reflection and absorption coefficients, thermal conductivities and heat capacities of the irradiated sample/
137
substrate systems, these temperatures cannot be calculated directly from the applied laser energies. Under the assumptions made in the surface-temperature calculations it has been shown that an approximately linear relationship exists between the maximum surface temperature and the incident laser energy. It has to be noted that this relation only holds for relatively short pulses (<10F3 s) and in a limited temperature range from ambient temperature to the melting point of the substrate material_ With the help of a calibration point, namely the laser energy required for heating the substrate to the melting point, we could approximate surface temperatures as a function of applied laser energy_ Of course, the temperature varies rapidly in time and place on the substrate and therefore we characterized the surface temperature by the maximum temperature reached in the centre of the irradiated spot. Conversion of the laser energy scale to this temperature scale leads to plots as shown in Fig. 7_ Note that only a small part of the experimental data is shown, namely data corresponding to calculated surface temperatures below the melting points of the substrates. Highly congruent ion intensity curves are obtained for stainless steel and quartz, so that it might be concluded that the surface temperature of the substrate is the relevant parameter in the desorption process. The onsets of the alkali and organic ion cur-
Fig_ 7. Relative ion intensities as a function of the maximum surface temperature T, of an irradiated quartz (X) and stainless-steel substrate (g)_ The solid lines result from fitting the experimental points to a thermal desorption model (see test).
138
rents occur at surface temperatures in the range 600-800 are
Gas-phase
quite
reasonable
with
respect
to a thermionic
K. These tempera-
of the alkali ions_ For this reason we adopted a thermionic emission model to describe the alkali ion current. As outlined in the surface-temperature calculations, for thermionic emission can be obapproximate activation energies A??’ tained from a plot of ion current versus maximum surface temperature_ A least-squares fit of AI? to the experimental data in Fig. 7 leads to AE’ = 0-8 and O-6 eV for Na’ emission from the stainless-steel and quartz substrates, respectively_ Interpretation of these values on the basis of ionization potentials and work functions is tentative because of the undefined surface conditions during the emission process_ If this thermal emission model is also used to interpret the results for quasimolecular ion emission, it is necessary to assume that these ion-molecule complexes can survive temperatures in the rage 600-800 K. This does not seem to be a reasonable assumption unless the quasimolecular ions are much more stable than the neutral molecules. Nevertheless, the esperimental results for the organic ions in Fig_ 7 can be described reasonably by a similar relationship (eqn_ (9)) between maximum surface temperature and integrated ion currents as used for alkali ions_ The resulting values for AEf , 0.2 eV for the quasimolecular ions and -0-6 eV for the fragment ions, are used only for quantification of the increase of ion current with temperature. This facilitates comparison of the various ion currents_ Concluding this Section, it will be clear that differences in ion current intensities using different laser energies on stainless-steel and quartz substrates can be qualitatively attributed to the different temperature responses of the substrates_ tures
origin
conzplesation
The concept of a simple thermal desorption process for ion-molecule compIesation had to be abandoned because of the fact that quasimolecular ions are observed only at high surface temperaturesThe presence of a strong alkali ion current which always accompanies the quasimolecular ions may point to the prerequisite of an intense Nd ion current for cationization reactions_ This leads to the postulation of a gas-phase complexation reaction for the formation of quasimolecular ions. In this hypothesis sucrose particles are evaporated at relatively low surface temperatures during the early stage of the laser pulse from the centre of the laser spot and later from the edges of the irradiated spot_ Thermionic Nd ions are emitted only from the hot central part of the spot_ This model supposes the evaporation of intact sucrose molecules_ Although for some time this assumption was considered to be unrealistic, various esperimental data obtained subsequently support this idea [ 34,15-J_ This qualitative picture of evaporation and gas-phase complesation describes adequately the experimental data reported in this paper_
139
However, quantitative interpretation of the data using this model has not to date been successful. The main obstacles are the unknown reaction kinetics of the desorpi;ion and fragmentation processes, which determine the gasphase concentration of sucrose. A simple molecular evaporation and ionization model does not suffice to explain the results quantitatively, as is shown in the following paragraph. Suppose that the sucrose sample at the irradiated spot is evaporated completely as intact molecules and that the sucrose cloud is overhauled by the much faster (more energetic) alkali ions emitted somewhat later. The probability PC that an alkali ion forms a complex with a sucrose molecule is then given by P, = I-
exp(--n50)
(10)
where n, is the number of sucrose molecules vaporized per unit surface area and 0 is the cross-section for complexation. Approximate values for the parameters under the present experimental conditions are n, = 10” molecules cm-’ and (T = 150 A2 (Langevin cross-section) 1353 _ These values lead to P, -r 1; thus every alkali ion can form a complex, provided that a sufficient number of gas-phase molecules are available: rx, > nxa+ (>zxa+ is the number of sodium ions emitted per unit surface -Lea). It will be clear that this condition can be fulfilled at relatively low substrate temperatures and consequently the behavior of the quasimolecular ion current should be linearly proportional to the number of alkali ions emitted at low laser energies_ At higher laser energies the number of alkali ions may esceed the number of sucrose molecules and the quasimolecular ion current will be proportiondl to the sucrose gas-phase density. From Figs. 6 and 7 it is observed that the quasimolecular ion current is not linearly proportional to the alkali ion current, indicating that rzNa+ > IZ,. This conclusion contradicts the value of rzxa+ 2: 10’” ions cm-’ at 5 mJ, as deduced from the measured Na+ current. Obviously, in the contest of this model, the number of gas-phase sucrose molecules is highly overestimated; vaporization of the sample does not proceed uniquely via molecular evaporation_ This point of view is supported by a consideration of the heat transport from the heated substrate into the sample under :he assumption that direct heating of the sucrose sample by laser irradiation is negligible. At the high heating rates in the centre of the laser spot (>5 X 10” K s-l), the ener-gy transport via thermal conduction is not sufficient to heat the sample layer uniformly to the momentary substrate temperature. This leads to high temperature-gradients in the sample layer. First-crder calculations support the possibility of sample vaporization and fragmentation of the layer in close contact with the substrate while the top layers are still in the solid state. The resulting explosive removal of sample material proceeds via cluster ejection rather than molecular evaporat.ion. Molecular evaporation can be expected from adjacent low-temperature areas_ The fact that in our esperiments no cluster ions could be detected may be explained by a mass distribution such
140
that most of the particles fall beyond the detection window of our instrument_ Another indication for the supply of sucrose from the outer regions of the irradiated spot can be found in the work of Heresch et al. [21]. A much improved quasimolecular ion current was obtained by focusing the laser beam onto surface irregularities with dimensions comparable to tne laser spot_ These irregularities were introduced by mechanically scratching the sample surface, apparently leaving ‘bare’ substrate at the centre of the laser spot_
In the foregoing discussion it has been shown that all of the present experimental results can, at least qualitatively, be explained by a thermal model where a minimum temperature has to be reached before quasimolecular ion formation can take place_ Minimum temperatures extracted from the experimental data are too high for thermal desorption of intact quasimolecular ions and favor a model where intact sucrose molecules and alkali ions combine in the gas phase. Furthermore, in the framework of such a model it is suggested that sucrose molecules involved in the complexation reaction _ _ ongmate mostly from the outermost areas of the laser spot. _4CKNOWLEDGMENTS
The
authors
are indebted
to Prof_ Dr_ J_ Kistemaker
for stimulating
discus-
sions_ This work is part of the research program of the Stichting voor Fundamenteel Onderzoek der Materie (Foundation for Fundamental Research on Matter) and was made possible by financial support from- the Nederlandse Organisatie voor Zuiverwetenschappelijk Onderzoek (Netherlands Organization for the Advancement of Pure Research)_ REFERENCES 1 2 3
H-R. Schulten, Int_ J_ Mass Spectrom. Ion Phys., 32 (1979) 97. D.F. Hunt, J. Shabanowitz, F-K. Botz and D.A. Brent, Anal. Chem., 49 (1977) 1160. P.G. Kistemaker, G.J.Q. van der Peyl and J_ Haverkamp, Adv_ Mass Spectrom-, (1981) to be published_ 4 (a) R.D. Macfarlane and D-F_ Torgerson, Int_ J. Mass Spectrom. Ion Phys., 21 (1976) 81; (b) R.J. Day, SE. Unger and R-G. Cooks, Anal. Chem., 52 (1980) 557A. 5 U. Gieswnan and F-W. Riillgen, Int. J. Mass Spectrom. Ion Phys., 38 (1981) 267. 6 R.D. Macfarlane and D-F. Torgerson, Phys. Rev. Lett., 36 (1976) 486. 7 (a) F-R. Krueger, 2. Naturforsch., 32a (1977) 1084; (b) F-R. Krueger, Surf. Sci., 86 (1979) 246. 8 R-D_ Marfarhme, Proc- 26’h Ann_ Conf. on Mass Spectrometry and Allied Topics, St. Louis, MO, 1978, p. 490. 9 J-F. Ready, Effects of High Power Laser Radiation, Academic Press, New York, 1971, p_ 140_
141 10 11 12 13 14 15 16 17
R. St011 and F.W. Riillgen, J. Chem. Sot., Chem. Commun., (1980) 789. R. St011 and F.W. RSllgen, Org. Mass Spectrom., 16 (1981) 72. U. Giessmann and F.W. RSIlgen, Org. Mass Spectrom., 11 (1976) 1094. W-R_ Anderson, Jr., W_ Frick, G-D_ Daves, Jr., D-F_ Barofsky, D. Chang kers, J. Am. Chem. Sot., 100 (1978) 1974. R-J_ Cotter and C. Fenselau, Biomed. Mass Spectrom., 6 (1979) 287. R-J_ Cotter, Anal. Chem., 52 (1980) 1589A.
J.F. Ready, pp_ 74.
Effects
of High Power Laser Radiation,
Academic
and
K.
Press, New York,
Fol-
1971,
29 30 31 32
R.C_ Weast (Ed.), CRC Handbook of Chemistry and Physics, CRC Press, Boca Raton, 1980_ D.A. King, Surf. Sci., 47 (1975) 384. M.A. Posthumus, P.G. Kistemaker, H.L. Meuzelaar and M.C. Ten Noever de Brauw, Anal. Chem., 50 (1978) 985_ F. UnsSld, G. Renner, F. Hillenkamp and R. Nitsche, Adv. Mass Spectrom., 7B (1978) 1423. F. Heresch, E.R. Schmid and J.F.K. Huber, Anal. Chem., 52 (1980) 1803. P.G. Kistemaker, M.M.J. Lens, G.J.Q. van der Peyl and A.J.H. Boerboom, Adv. Mass Spectrom., 8A (1980) 928. R. Stoil and F-W_ Rbllgen, Org. Mass Spectrom-, 14 (1979) 642. R-J. Cotter, Anal_ Chem., 53 (1981) 719. D_ Zakett, A-E. Schoen, R-G. Cooks and P-H. Hemberger, J_ Am_ Chem. Sot., 103 (1981) 1295_ G.J.Q. van der Peyl, K_ Isa, J. Haverkamp and P.G. Kistemaker, Nucl. Instrum. Methods, accepted for publication_ G-D. Daves, Jr., T-D. Lee, W-R. Anderson, Jr., D.F. Barofsky, G-A. Massey, J-C. Johnson and P-A_ Pincus, Adv. Mass Spectrom., 8A (1980) 1012. H.J. Heinen, S. Meier, H. Vogt and R_ Wechsung, Adv. Mass Spectrom., S-4 (1980) 942. K.-D. Kupta, F. Hillenkamp and Ch. Schiller, Adv. Mass Spectrom., 8A (1980) 935. F-R_ Krueger and B. Schueler, Adv. Mass Spectrom., 8A (1980) 918. R-J_ Cotter, Anal. Chem., 52 (1980) 1767. F.J. Vastola and A.J. Pirone, Adv. Mass Spectrom., 4 (1968) 107.
33
R-O_ Mumma
34
G-J-Q_ van der Peyl. K. Isa, J_ Haverkamp and P-G. 16 (1981) 416. M. Henchman, in J.L. Franklin (Ed.), Ion-Molecule York, 1972, Ch. 5.
18 19 20 21 22 23 24 25 26 27 28
35
and F-J. Vastola,
Org. Mass Spech-om.,
6 (1972) Kistemaker,
1373. Org. Mass Spectrom..
Reactions,
Plenum
Press,
New