- Email: [email protected]
Secondary ion and electron yield measurements for surfaces bombarded with large molecular ions
Nuclear Instruments and Methods in Physics Research B lOS (1996) 282-289
B•• m Int_lIons withMaterials& Atoms
ELSEVIER
Secondaryion andelectronyield measurementsfor surfaces bombardedwith largemolecularions G. Westmacott, W.Ens',K.G. Standing Department (~( Physics, University ofMan itoba, Winnipeg. CanadaR3T2NZ RPl' (";Vl'rl2S July
1995: revised formreceived 25 August 1995
Abstract Secondary-ion and secondary-electron emission yielJs from surfaces bombardedwi'" large molecular ions have been measured in a tandem timc-of-fhght mass spectrometer. The primary ions were produced by matrix-assisted laser d~:;orptiun/jonization ami ranged in mass from about 6000 to I I oCOO u, and in energy from 5 to 25 keY. The yields were .ueasuredfor surfaces ofstainless steel and CsI inmodest vacuum conditions. Electron and ion emission were observed for incident veloc ities as low as 3.5 km/s (ll .06 eV /Ill. Fora given energy. the yields decrease rapidly with increasingmass in the low mass range, but for large inc identprojectiles at 25 keY (or higher>. the efficiency of secondary i OIl productionis more or less constant near unity . The efficiency of secondary electron production continues to decrease slowly with increasing mass. but remains > - 30% throughoutthe mass range typically encounteredin mass spectrometryof proteins (i .e, molecular weight < - 300000 u corresponding to > 0.1 eV /u). For high velocities, the yield of electrons and ions is significantly higher for a CsI surface compared to stainless steel. but for velocities correspondingto < - 0.4 eV /u, the emission is ratherinsensitive to the surface .
1. Introduction With the steady increase in the acces s-blemass range in mass spectrometry.detection of large molecular ions has become an importantproblem. Detection of molecularions in mass spectrometry is usually accomplished through a collision of the ion of interest with a surface. Either the secondary electrons or secondary ions produced in this colli sion are accelerated to produce a cascade, eventually amplifying the signal to a detectablelevel. It is well establishedthat as the velocity of the incident ion decreases, the secondary electron and ion yields de crease. resulting in lower detection efficiency. For example, the electron emission yield from Cu induced by bombardment with large water clusters (up to 60000 '.1) is reportedto have a "threshold"velocity of about 18 km/s correspondingto an energy per unit mass of about 1.7 eV /u (I). A similar threshold velocity was reported for insulin (5733 u) incident on Cu [2]. For water clusters incidenton an aluminumoxide surface an apparentthreshold of 10 km/s (0.5 eV /u) was reported[3].
: Correspondingauthor.E-mail [email protected].
From these results, a rather low detection efficiency might be expected for molecular ions with energy below about I eV /u, if electron emission facilitates detection. In spite of this, large molecular ions have been routinely detected at considerably lower velocities than this using secondary emission detectors. In particular, in matrix-assisted laser desorption/ionization(MALDI) the accessible molecular weight range is sypically regarded as at least - 300000 u, with an accelerating voltage of 30 kV {41. This corresponds to an energy per unit mass of - 0.1 eV /u. five times lower than the lowest thresholdenergy mentioned above. The nature and the yield of the secondary panicles emitted by large molecular ions are of central importance in time-of-flight mass spectrometry, the most common techniquefor examining ions producedby MALDl. Detectors designed to use secondary electron emission have bettertime resolution than detectorsthat rely on secondary ion emission [5-7] because of the multiple ion species and the large transit times of ions compared to electrons, and this is an importantconsiderationin time-of-flight (TOF) measurements.From this point of view it is preferableto use secondary electrons for detection. On the other hand, the sensitivity and the mass range depend on the emission
0168·583X/96/$1 5.00 tl 1996 Elsevier Science B.V. All rights reserved SSDl 0168-583X(95)OI060-2
G. Wesrmoc(>!f 1'1 al./ Nucl.
III.IIr.
and Meth. in Phys. Res. B 108(19961282-289
yield of the secondary'particles so it seems reasonable to use the secondary particles with the highest yield. It has been clearly demonstratedthat for mcrdcnt ions below - 2 eV lu the secondary ion emission yield is larger than theelectron yield [5-9]. suggesting an advantage in using ionemission for detection. In fact, secondary ion emission is mainly responsible for the detection of large molecular IOns (mentioned above) in some detector geometries [5]. However. if the »lectron emission yield is still greater than or near unity over the mass range of interest, then alower electron yield can be compensatedby a higher multiplieror amplifiergain and sensitivity is not compromised. It is therefore importantto determine the absolute electron yield for large molecules incident on detectorsurfaces.but much less work has beendc"e in this area. especially at low velocities. Many detailed measurementshav-: :Je\ n mil,"" of electron emission induced by atr 11;ci~ .... or relatrvcl- small polyatomic ions incident on clean. well-characterizedsurfaces [10.11], including measurementsof kinetic emission for projectile velocities as low as 12 kmls [12]. There is also a body of work onelectron emission from hot spikes caused by the impact of cosmic dust at velocities lower than thoseconsideredhere [13]. However. it is not obvious how the results of either regime apply to the situation examined here: detection of very high molecularmass ions (- 100 ku) at low velocities (- 4 kmys) in vacuum conditions typical of commercial mass spectrometers. The electronemission yield for singly-chargedmolecular ions incident on conversion surfaces that have been exposed to atmosphereis usually interpretedin the context of kinetic electron emission [13], where the energy needed to eject electrons comes from the kinetic energy of the incident ion. Conditionsfor potentialelectron emission are not generally met for adsorbed molecules; that is. the projectile'spotential energy released iii the Auger process does not normally exceed twice the binding energy of the target electrons. In kinetic emission. a minimum energy must be transferred to a target electron to cause emission, and this energy corresponds to a threshold velocity for a given projectile-targetcombination. For incident atomic ions two extreme values can be identified for the threshold velocity depending on the mechanism assumed for the energy transfer [13]. For collisions between the incident ion and the fastest free electrons, velocities higher than 150 km/s are requiredfor emission from metals, consistent with extrapolationof the experimentalresults forlight projectiles[14]. However, if some of the momentumof the projectile is transferredto a target atom or the lattice, much lower thresholds are possible. The lowest possible threshold. determined by conservation of energy. corresponds to all the kineticenergy being transferredto the target electron. For an oxygen projectile(the largest abundant atom inorganic molecules) incident on a low workfunrtionsurface, this correspondsto about5lml/s or 0.13
Z83
eV lu [13]. The situation becomes more complicated with polyatomic projectiles, where t:l'lltiple collisions could clearly reduce the threshold further; obviously. conservation of energy is not a limit if the totalenergy of the projectile is - 30 keY. Because the energy transferredto a targetelectronmust exceed its bindingenergy. it is reasonablethat most empirical or theoretical characterizationsof kinetic electron emission have threshold velocity as a oarameter. Such descriptions lead to extraction of threshold values from data acquired at higher velocities, without direct evidence of a true threshold. However. if a different mechanism dominates for velocities near or below theextrapolated "threshold".the term is inappropriate.As early as 1958, measurementsof kinetic electron emission for atomic ions on metals have been reported forvelocities down to 15 km Zs, an order of magnitudelower than the extrapolated threshold [15]. Similar results have been reported more recently [12,16] for velocities as low as 12 kmys [12]. The emission probability in these experiments was very low (:s: 10 - 4) and the measurementswere limited by experimental sensitivity and I1(A an obvious threshold velocity [15,16]. The measurementsof Beuhler and Friedman[1,3] using large water clusters are in the same- velocity range, but because of the large polyatomic projectiles. the emission efficiency was greatly enhanced. Indeed, values of electron yield less than one electron per incident cluster could not be measured by their method.The interpretation of the lowest velocity for which a yield higher than I is observed as a threshold velocity is therefore ratherarbitrary. There is a clear need to determinethe behaviourof the electron yield at lower velocities typically encountered in detecting MALDI ions. The expectation from previous work was that the electron yield would vanish at lowvelocities [1-3}, and that seems to be consistent with some of the measurementsof the ratio of the electron-to-ion yields [5]. However, our previous measurementsof the electron-to-ionratio indicate clearly that theelectron signal does not vanish, at least for energy per mass unit as low as 7km/s (0.25 eV lu) [6,9]. For this velocity, the ratio is - 0.10 and only weakly dependent on mass. These earlier measurementsgive no informationon absolute yield, but measurementsmade at Orsay [7] for 18 keV albumin (0.3 eV lu) incident on CsI give an absolute secondary ion yield > 1 and an electronto-ion ratio ofabout0.06. In this experiment, using single-ion counting methods and a coincidence technique similar in principle to the method used by the Orsay group[7]. we measure the absolute emission efficiency of secondary electrons and secondary ions as a function of the velocity and mass of the molecularions incidenton differentsurfacesfor velocities down to 3.5 kmls (0.06 eV lu). Our primary interes: is to determine the feasibility of using the secondary electrons for detection of high mass, low velocity molecular ions. In particular,for a given accelerating voltage,
G.
Wcsl",UCO/I e
'
al, / Nucl . lnstr. and Mcth. ill Pliy» Res. B 108 (/996 282 - 289
how ( ·)~S the efficiency of electron emission depend on mass, .md what lossof efficiency (if any) must be tolerated tc tal.e advantage of the better time resolution t.ccessible with lctectors that use electron emission"
::, t ..
'.t'rimcntlll
TIle experimentalarrangementshown in Fig. I i<: similar to the arrangementdescribed previously (6.9] with a few modifications. The molecular ions produced at the target by MALDI are acceleratedinto a primaryflight tube (- 83 em long), pass through ;;l ,rnall adjustableiris (- 55 em from the target) and strike a conversion plate (converter) at 45°. Secondary electrons and ior-s produced at the converter are accelerated into a second flight tube (- 18 cm long) and arc thendetected in a 4 cm diameter microchannel plate detector. Amplified signals from the microchannclplates are fed into a25S-stop time-to-digital converter (TDO (model CfN-M2, Institut de Physique Nucleaire, Orsay, France)connected by a custom interface to an Atari Tro30 computerfor daILI storage and analysis. Although single-ion counting methods have limitations, it is easier to interpretthe data quantitativelythan is thecase when a transientrecorderis used. The primary ions were produced by laser pulses of width 10-20 ns from a high repetition fate excimer laser (HE-460-HR-B: LurnonicsInc., Kanata, ON. Canada)incident on the target at 70°. Triggered by a digital delay generator (9650 EG&G PARC), the laser was typically operated at 300 Hz with XeCl (30X nrn). The power density was between 106 and 107 W /cm 2 controlled by a circulargradientneutraldensity disk. The beam focus was manually rastered across the target with a displacement plate. Primary ions were produced from standardsamples of insulin (57~B :1), trypsin (- 23S4D u), human transferrin (- 79500 u) and fi-galuctosidase (- 113600 u). These
I·
\ X
-=-
v."-
=- I.
~-.
- pnma'YIOOs
\
.:
.
.:
Lnsor
-Ll
Irts
G k - .
Targel
\
Convorter
> ~'
/.
".
"~". ~~~~;::,~~' f~·':-=L
cd-a~:' -=
cgo:n ~YTDC}--1Dll
~l
Fig. I. Schematic diagram of the experiment,TIle relevant dimcnsions and voltages are given in thc text,
protein~ were prepared uccontn.g to the usual procedure for r iatrix-assistedlaser desorption (61 using s;'laplOic acid as a matrix. "Primary" ions produced by the laser were acc iler.ited across a potential difference \J~ (tvpiCl!') t\::twen 5 and 25 kV) applied between the l8i" [ and a gr .iunded 909r transmission grid. The accelerated primary ions passed through a small iris with adjustable size, nositioned about .'is cm from tile target. The aperturesize could be varied from zero (completely closed) to -.5 mm'. allowing control cf the primary ion transn.rv.ionto the converter: usually it \I, as set so that the probabilityof transmittingone ion per laser pulse was much less than unity. Secondary ions and electrons w-re produced at the converter from surfaces of stainless steel and CsI; a few measurements were also r.:.lJe with CuBe. The stainless steel surface was cleanec in an ultrasonic bath in acetone, and rinsed with methanol. CsI surfaces were prepared by evaporation onto a clean stainless steel disk under a vacuum of - 100 mTorr.The surfaces were then transported (at atmosphere) into the spectrometer which is subsequzntly pumped to - I X 10 6 TOIT. Secondary ions were accelerated between the converter at - 100 V and a grounded grid (95% transmission). A high transmission grid and low voltage were used to minimize the production of spurious electrons [6,7,9]. To avoid detector saturationfrom the intense secondary electron signal created by the low-mass, primaryions from the matrix. the extraction voltage was pulsed from a small positive voltage to the - 100 V accelerating voltage at some appropriatedelay time after the laser pulse. The front plate of the detector was held at + 4 kV to provide post-accelerationfor the low energy (100 eV) ions and electrons A grounded 95% transmissiongrill about I ern III front of the detector maintained zero field in the flight tube. A magnetic field produced by a coil wound around the detectorwas used to improve secondary electron transmission which is rathersensitive to stray magnetic fields, e.g, the earth'smagnetic field. This applied field, which causes electrons to spiral into the detector, was increased empirically until the transmissionsaturatedat a value about 4D% higher than with zero applied field. The measured field strength for the optimum (6 A through 100 turns) was - lOG near the detector and - 3 G near the converter.
2./, The coincidence method When a large numberof molecularions No are incident on a solid surface(the converter), the numberof secondary electrons ejected N; is given by No = Yo No, where Yo is the secondary electron emission coefficient. Similarly for the secondary ions N, = Y, No. Ideally. our aim is to determine the secondary electron and ion coefficients as a function of mass and velocity for differentsurfaces.
Ii. Wesrmacott et al./ Nllet. Instr. ami Meth. ill Phys. Res. B lOR (!(j96) 282--28)
Ooerall efficiency Although the'emission coefficients are the fundamental quantities,the detection and countingmethod used in these experiments does 'let measure the values of N; and N, directly. When a primary ion strikes the converter, a single-ion counting system will count at most one pulse, even if many electrons are produced. Therefore, with 51: h a system, if No ions are incident on a conversion surface, the number of electron pulses actually counted with a single-ion counting system is =
185
<
N;
E;
<
E;
E;
(I)
(2)
Fig. 2. (a) Time-of-tlight spectrum of secondary ions and secondaryelectrons producedat a slam less steel surface by bombardment with singly-and doubly-charged molecular ions of human Iransferrin(79500 u) at 10 and 20 keV, respectively. The indicated flight limes include the flight lime of theincidentprojectile. (b) Coincidence spectrum recorded with a time window around the secondary electron peak for incident [M·t- Hl". The inset shows the integratedcounts for the indicated peaks tc indicate the degree of random coincidences from different incident projectiles (see text).
Taking account of the probability for detecting an emitted electron Ed' which depends on the secondary electron transmissionand the detectorefficiency, the overall efficiency is
(4) Relation to emission coefficient The emission coefficients 'Ye and Yi can be derived from the measured efficiencies. Assuming the electron multiplicity follows a Poisson distribution[17], the probability for ejecting exactly 11 electrons from a single primary ion impact is
'The validity of Poisson statistics has not been ngcr.csly demonstratedfor low velocity projectiles where 'Y < I, but the assumption is reasonable based on results at higher velocities where 'Y> J [17]. The electron emission efficiency €e correspondsto the probabilityfor ejecting one or more electrons for a single primaryion impact, Thus,
(3)
which enables conversion of electron efficiency to electron coefficient by 'Ye
--In(1 =
<)
~------
.
(5)
Ed
For experimentaldata with appreciableassociated uncertainties, this conversion is only reliable for efficiencies considerably less than unity. As the efficiency approaches 100%, the correspondinguncertaintyin 'Ye becomes more and more magnified. For this reason, although the emission coefficients are more fundamental, the present data are first presentedas efficiencies. This is, in any case, the relevant figure-or-merit in determining the molecular weight range of thetechnique,particularlywhen single-ion counting is used. Only the probability that an ion will produce one or more secondary electrons (or ions) is importantin determiningwhetherit will be detectedornot. The actual number of electrons that are produced only
G. W".,r'll
286
influences the necessary gain of the amplificationelectronic". (The emission coefficient is of COUf!'e relevant to the relative intensity of different peaks when ;\ transient recorderis used. and is therefore related to mass discrimination.)
Random correlations The equationsabove are ..alid underthe condition that only one or zero primary ions strike the surface of the
converterfor every successful laser desorptionevent. Random correlationsOCcurwhen two primary ions strike the converter.distorting the results. To experimentallycontrol the numberof primary ions hitting the converter, an iris (mentioned above) is placed between the MALDI ion source and the converter.TIle apertureis set small enough so that the probabilityfer transmissionof a single primary ion is much smaller than unity to minimize the number of randomcorrelations. To determine the extent of the random contributions. correlationswith a second primary ion projectile (usually the doubly-charged molecular ion) arc also monitored. If the number of events where both the singly-charged (,A) anti doubly-charged\.8) primary ions pass through the iris (AB or BA) is kept very low, then the number of events where two singly-charged (AA) or two doubly-charged (BB) primaryions pass throughthe iris is lower by a factor of 2 assuming all four possibilities have equal probability, Typically, the probabilityfor a singly-and doubly-charged ion to be detected in the same event is less than 50/[. Thus reasonableaccuracy Gin be expected for efficiencies > IOt;~.
The measured time intervals from each laser shot are analysed ami sorted by the data acquisition software and two spectra are recorded. The first spectrum is the full spectrum;it is simply a histogramof all the recordedtime intervals between the laser pulse and the detection of secondary ions and electrons. An example of such a full spectrum is shown in Fig. 2a. The second spectrum (Fig. 2b) is a coincidence spectrum; it is a histogram of the recorded time intervals from selected events in which at least one measuredtime interval falls within a certain lime window. In this experiment the time window is placed aroundthe electron peak from the primary ion of interest. The numberof secondary ions in this coincidence spectrum then represent"the numberof events in which both an electron and ion were detected, i • The efficiencies and (for any single secondary ion species) can then be calculated from the two spectra using Eqs, (t) and (2); the values of N: and N,' are obtained from the full spectrum. To improve the statistics and because individual ion species are not resolved, it was necessary to integrate the intensity for ion species ranging in mass from about I to 125 u for the quantities of and N..:i' However, the development of Eqs. and (2) assumes that multiple secondary ions fromthe same primaryion are counted as a single pulse. This is the case for secondary ions of the
N:
E;
(n
Nt
<
same species hut not usually for secondary ions of differan ent species. To maintainthe validity of Eqs. ( I) and additional constraintwas applied using software 5(1 that at most nne ion within the selected range was registeredfer a given event: in effect the deadtime W;\S increased iii a selected time range. Thus N,' and N:, represent the total number of eticnts in which at least one secondary ion is detected in the selected rangeof I to 125 u.
en.
3. Results and discussion The determination of the secondary electron and ion emission efficiency using the coincidence technique is illustratedfor incident molecular ions of human transferrin (79500 u) at 10 keV (0.13 eV/u). Fig.. 2a shows the full spectrum of the secondary ions and electrons ejected by collisions of the [M + H]+ and [M + 2Hp· ions with the converter. Fig. 2b is the coincidence spectrum recorded with a time window around the secondary electron peak for the [M + t-i]", The numberof counts frorr. the[,\-1 + 2HF· in the coincidence spectrum compared to the full spectrum indicates rhe degree of random correlations. in this case less than 3"!c. On the other hand, the percentage of the ions from the [M + H] + ion in correlationwith the
III
""I:
l:!g
.~~ _CD
~~ c" 0'0 "ii~
.!! ¥ 0.1
e", w ... ~
0.1 1
SS
~~
(
~~Y
.
• tnsultf\ - .... Trypsl" • - Transferrin.+ __e_. Transferrin. • Ogalactosid.se 'i
Ql
1
Energy per mass unit (eVlu)
Fig. 3. Efficiency of secondary electron emissionEo for various incident projectiles on a Csi surface and II stainless steel (55) surface. The efficiency was corrected for known losses due 10 grids. The errorbars represent statistical uncertainty. The curves are fits of the derived electron coefficient (using Eq. (5» to a velocity dependencegiven t-y y = Yo exp( - lie I II).
G, Westmacotl et al./ Nuct. Instr, am ] Melli, in Phys. Res. B 108 (1996) 282,·289
electrons from the same primary ion is about 16%. Thus 16% of the events in which an ion is observed also = N;JN;' = 169c. The data also produce an electron so indicate that in 64% of the events in which an electron b observed, an ion (between 87 and 102 ms) is alsoobserved = N;jN: = 64%. so Fig. 3 shows the secondary electron efficiency as a function of energy per unit mass for insulin, trypsin. transferrinand ji-gn!actosidase incident on (a) a Csf surface and ib) a stainless steel surface. The data have been correctedfor known losses due to grids using Eqs . (3) and (S) in which the geometrical transmissionwas taken as Ed: otherpossible losses .rave not been accountedfor. For both surfaces, the electron efficiency decreases with decreasing velocity for a given mass, and for a given velocity increases for larger masses. Fig. Sb also shows that the secondary electron yields are very similar fer both the singly-charged and doubly-charged molecular ions of transferrinat the same velocity. The efficiency for ion emission for the same projectiles and surfaces is shown in Fig. 4. As has been reported before [7], the efficiency is close to unity even for low velocities, where it is considerablyhigher than the electron efficiency. It is useful to separatethe influence of the mass and the velocity in the electron emission data. If the a..e rage
287
E;
E;
CST]
I
0.1 1~---
+
++
,+ + •
i
.. ~ •
jOlg. 4. Efficiency of secondary ion emission (for ions between I and 125 u) for various incident projecti les on a Csl surface and a stainless steel (SS) surface. TIleefficiency was corrected for known losses due to grids. The error bars represent stadstical uncertainty.
Fig. 5. Reduced electron yield for various projectiles on a CsI surface and a stainless steel (SS) surface. The measured efficiendes were converted to yields eY with Eq. (5) and divided by the mass of \he projectile. TIle lines through the data are fits to Y = {a' ": the exponent is 2.1 for the stainless steel target and 3.3 for the Csi surface. The fit from the 5S data is reproduced as a dashed line on the Csl plot for comparison.
numberof ejected electrons, 1';', is simply proportionalto the primary ion mass, i.e, to the numberof atoms in the molecule, then it is reasonable to normalize the data by dividing 1'0 by the mass, or by the numberof components in the projectile. Such an additivity rule. which assumes the atoms interact independently with the surface, is not universally applicable. and depends on the velocity regime and the size Of the projectiles [3,13}. It appears to hold for water clusters incident on Cu at low velocities [I}, and for peptides incidenton microchannelplates at velocities down to 15 kmy's [18]. Deviations from the additivity rule were reported for water clusters on AI20 3 and for peptides at relatively high velocities (> 100 kmys) on CsI and Al 20, surfaces (J9]. Recent measurements for large multiplychargedmolecularion projectilesat velocities above ~ 30 kmy's show a linear dependence on mass for :1 graphite surface but a sublinear dependence(proportionalto MO,73) for Al 2 0 3 [20]. '01e present data for electron emission (Fig. 3) fall more or less on a single curve if the derived emission coefficient (I'e)' calculated with Eq. (5), IS divided by the mass of the projectile, as shown in Fig. 5. '111e degree of fluctuation in these reduced coefficients is rather large, but Fig. S indicates the data are at least consistent with a linear mass dependence.
288
G. W.·stma cott et a/ . ; Nucl. lnsrr . lind Ml'Ilr. in Ph ys. R.,.<. B 108 ( / 99612 82-289
A non-linear yield enhancement has not been report ed o mbardmentin this for e lectron em iss ionby polyatorn ir b veloci ty range . The deviati ons from linearity menti oned above all corre spond to yields that arc lower than the SLm of the e xpected yields for the constituent atoms. Howe ver. for dust particles at velociti es as low as 0 . 1 krny s, kine tic emiss ion by the ind ividual atoms is not pos sible and the electron emission is attributedto a heat spikeproduced by the impact of the clu ster. clearl y acollective effect [13] . The detection of electrons in this e xperim ent at veloc ities down to 3.5 km/s (0.06 eV lu) similarly sugges ts a cooperative effect may be present even ifcomparisons between various large polyat ornicprojectiles do not indicate an enhancement.The decrease of the electron yield w ith decreasing velocity is character istic of kinetic electron em ission. but there is no indication of threshold.anu a at the lowest velocity the indiv idual atoms do not have suffic ien t energy to induce kineti c em issionatoms ; smaller than 0 have energy less than I cV . The e lectron emi ssio n data for the individual ojectiles pr (F ig. 3 ) fit an exponenti al dep e ndence on the inverse veloci ty, y = Yo ex p{ - /lei v) [13]. but colle ctivel y the reduced data (Fig. 5 ) are better descr ibed by a simple power law. Th e dependence on velocity in this ran ge appears 10 be approximatelyquadratic for the stainles s 'leeI surface, and cubic for the Csl surface. f lf:. 5a and Fig. 5b indicate the difference between the seco ndary electron yield from a non-characterized surface of stai nless steel and a CsIsurface . As previousl yreporte.; [9,19.21] Csl gives a significant enh ancem ent in the ion and ele ctron yieldcompared to stainless I eV/u. Th is is true even afte r both surfaces have beenexposed to atmosphere and are examined at 10 6 Torr. although the enhancement decreases with prolonged exposure to atmosphere [9]. The present results alsosho« that the enhancementis reduced at lower velocities. and below abou t 0.4 eVlu there appears to be little dependence on the typo: of conversion surface , although asomewhat highereffic iency is observed for 'he stainles s steel surface The influence of the velocity on the surfacedependence j.; illustrated clearly in Fig.6, which shows a comparison between the spectra obtained from the tNO surfaces usin g ~if~e re n t primary projectiles with an ene-gy of abou t:'0 ~·:e V. For cytochrome C at 1.5 eVlu there is a large enhanement c (by abouta factorof 6) in the intensity of tile secondary electron peak. from a Csl converter. and there is no notice able enh a ncement for albumin [M + H) +
S51
I I 100000 FI ~~I
Time (ns)
Fig. 6. Timc-ot-tl ight spectre of secondary electrons and secondary ions fromthrc- projectiles inciucnt on a Csl surface and a stainless steel(SS ) surface. The indicated flight time includes the: !light time of the incident r rojecule. The electron peaks from the three projectiles(w hich ha:e different velocities) are denoted.TIle clear enhancement observed with a Csi target for high velocity projectiles IS absent at lower velocities (higher times). the energy and primary ion ma ss were varied. However. for a given mass spectrum. all theprojectiles have the same energy and only the mass var ies.T hus, a more pract icalcurve should ind icate the effic iency as a function of mass for a givenenergy. We have used thedata of Fig. S to derive such a plot for severaldifferent accelerating voltages in the range typicalof MALDI; the results are shown in Fig. 7. FOI accelerating voltage at or above 30 kV (the most common value) Fig. 4 shows that detection based on ion emission gives efficiency near unity for the entire mass range. asmentioned above. Fig. 7 shows that detection based on electron emi ssion for this voltage is
11- l ' "
.
...,
0 ."
"
" '~ ' , ,~,> -- 20 keV
"
<,
-
· 3QkeV 40keV SOkeV
. . . . .- -----.....+
0.1 +--~-~10' 10' Mass (u)
10·
Fig. 7. Efficiency for electron emission from a Csl surface as a function of projectile mass for various accelerating voltages. The Sa. curvesare derived from the fit to the nata of Fig.
G. W",lfmucottet uf./Nllc/. lnstr. and Mnh. ill Pbvs. Res. B J08 (/9961282-289
somewhat lower (about SOIk for molecular weight of 100000 u), but still clearly feasible.
4. Conclusion Electron and ion emission was observed fromstainless steel and CsI for incidentprojectilevelocities as low as 3.5 km/s <0.06 eV /u). Using secondary ion emission, the detection efficiency in mass spectrometryof large molecular ions up to at least300000 u can be near 100% if the ions are acceleratedto 30 keY. The efficiency of a detector that relies on secondary electron emission is somewhat lower but still near30'70 for this extreme case. For large molecular ions with energy per unit mass less than0.4 eV /u, the detection efficiency is insensitive to the nature of the surface in a modest vacuum(- 10 6 Torr).
Acknowledgements This work was supported by grants from NSERC (Canada) and from the US NationalInstitutes of Health (GM 30605).
References (I] RJ. Beuhler and L. Friedman,Nucl. lnstr. and Meth. B 170 (1980) ~09. (2] A. Hedin. P. H~kansson and B.U.R. Sundqvist, Int. J. Mass Spcctrorn.Ion Process. 75 (1987) 275. [3] R.J. Beuhler, J. Appl. Phys. 54(1983) 4118. (4) F. Hillenkarnp,M. Karas, R.C.Beavis and B.T. Chait, Anal. Chern. 63 (1991) 119~A.
289
[5] B. Spengler. D. Kirsch. ~. Kaufmann, M. Karas, F. Hillcnkamp and I.'. Gicssm.mn. Rapid Commun. Mass Spectrom. 4 (1990) 301. [6J J. Martens, W. Ens, ~ .G. Standing and A. Vercntchikov, Rapid Commun. Mass :'pcctrom.6 \1992) 147. [7] A. Brunelle, P. Chaur,.d. S. Della-Negra, Y. l.ebeyec and G.B. Baptista, Int. J. Mass Spcctrom. Ion Process. 126 (199.') 65. [8] H. Haberlandand M. Winterer, Rev. Sci. Insi.. 54 (1983)
764[9] A. Verentchikov, W. Ens. J. Marten,and K.G. Standmg, Int. J. Mass Spectrom.Ton Process. 126(1993) 75. (10) D. Hasselkamp,H. Rothard,K. -0. Groeneveld, J. Kemmler, P. Varga and H.Winter, Particle Induced Electron Emission \I (Springer,Heidelberg, 1992) and references therein. [ll] R.A. Baragiola. in: Low Energy Ion-Surface Interactions, Ed. J.W, Rabalais (Wiley, Chichester, 1994) r, 187. and references therein. [12] u. Lakits and H.Winter, Nucl. Instr. and Meth. B 48 (:9 90) 597. [13] R.A. Baragiola, Nucl. lnstr. and Mcth B 88(1994) 35, and references therein. i:~] RA. Baragiola. E.V. Alonso and A, Oliva, Phys, Rev. B 19 ([979) 121. [15J P.M. Waters, Phys. Rev. III (1958) 1053. [16} E.V. Alonso, M. Alurralde and R.A. Baragiola, Surf. Sci. 166 (1986) L155. [17J K. Toglhofer, F. Aumayr, H. Kurz, H.Wimer,P. Scheier and T.D. Mark,Nucl. Instr. andMeth. R 88 (1994) 44. [18] P.W. Geno and R.D. Macfarlane,Int. J. Mass Spectrom, Ion Process. 92 (1989) 195. [19] R. Mosshammerand R. Matthaus,J. Phys. (Paris) C 2 (1989)
III. [20} J. Axelss» '. ( 1 I,. ..Ihl II 1 R Sundqvist, Nucl. Instr. and Meth.13 ~,1'- ( ' I ' •. /l I l i. [21] S. Della-Negra, C. Deprunand Y. Lelseyec, Rapid Commun. Mass Spectrom. 1 (1987) 10.
B•• m Int_lIons withMaterials& Atoms
ELSEVIER
Secondaryion andelectronyield measurementsfor surfaces bombardedwith largemolecularions G. Westmacott, W.Ens',K.G. Standing Department (~( Physics, University ofMan itoba, Winnipeg. CanadaR3T2NZ RPl' (";Vl'rl2S July
1995: revised formreceived 25 August 1995
Abstract Secondary-ion and secondary-electron emission yielJs from surfaces bombardedwi'" large molecular ions have been measured in a tandem timc-of-fhght mass spectrometer. The primary ions were produced by matrix-assisted laser d~:;orptiun/jonization ami ranged in mass from about 6000 to I I oCOO u, and in energy from 5 to 25 keY. The yields were .ueasuredfor surfaces ofstainless steel and CsI inmodest vacuum conditions. Electron and ion emission were observed for incident veloc ities as low as 3.5 km/s (ll .06 eV /Ill. Fora given energy. the yields decrease rapidly with increasingmass in the low mass range, but for large inc identprojectiles at 25 keY (or higher>. the efficiency of secondary i OIl productionis more or less constant near unity . The efficiency of secondary electron production continues to decrease slowly with increasing mass. but remains > - 30% throughoutthe mass range typically encounteredin mass spectrometryof proteins (i .e, molecular weight < - 300000 u corresponding to > 0.1 eV /u). For high velocities, the yield of electrons and ions is significantly higher for a CsI surface compared to stainless steel. but for velocities correspondingto < - 0.4 eV /u, the emission is ratherinsensitive to the surface .
1. Introduction With the steady increase in the acces s-blemass range in mass spectrometry.detection of large molecular ions has become an importantproblem. Detection of molecularions in mass spectrometry is usually accomplished through a collision of the ion of interest with a surface. Either the secondary electrons or secondary ions produced in this colli sion are accelerated to produce a cascade, eventually amplifying the signal to a detectablelevel. It is well establishedthat as the velocity of the incident ion decreases, the secondary electron and ion yields de crease. resulting in lower detection efficiency. For example, the electron emission yield from Cu induced by bombardment with large water clusters (up to 60000 '.1) is reportedto have a "threshold"velocity of about 18 km/s correspondingto an energy per unit mass of about 1.7 eV /u (I). A similar threshold velocity was reported for insulin (5733 u) incident on Cu [2]. For water clusters incidenton an aluminumoxide surface an apparentthreshold of 10 km/s (0.5 eV /u) was reported[3].
: Correspondingauthor.E-mail [email protected].
From these results, a rather low detection efficiency might be expected for molecular ions with energy below about I eV /u, if electron emission facilitates detection. In spite of this, large molecular ions have been routinely detected at considerably lower velocities than this using secondary emission detectors. In particular, in matrix-assisted laser desorption/ionization(MALDI) the accessible molecular weight range is sypically regarded as at least - 300000 u, with an accelerating voltage of 30 kV {41. This corresponds to an energy per unit mass of - 0.1 eV /u. five times lower than the lowest thresholdenergy mentioned above. The nature and the yield of the secondary panicles emitted by large molecular ions are of central importance in time-of-flight mass spectrometry, the most common techniquefor examining ions producedby MALDl. Detectors designed to use secondary electron emission have bettertime resolution than detectorsthat rely on secondary ion emission [5-7] because of the multiple ion species and the large transit times of ions compared to electrons, and this is an importantconsiderationin time-of-flight (TOF) measurements.From this point of view it is preferableto use secondary electrons for detection. On the other hand, the sensitivity and the mass range depend on the emission
0168·583X/96/$1 5.00 tl 1996 Elsevier Science B.V. All rights reserved SSDl 0168-583X(95)OI060-2
G. Wesrmoc(>!f 1'1 al./ Nucl.
III.IIr.
and Meth. in Phys. Res. B 108(19961282-289
yield of the secondary'particles so it seems reasonable to use the secondary particles with the highest yield. It has been clearly demonstratedthat for mcrdcnt ions below - 2 eV lu the secondary ion emission yield is larger than theelectron yield [5-9]. suggesting an advantage in using ionemission for detection. In fact, secondary ion emission is mainly responsible for the detection of large molecular IOns (mentioned above) in some detector geometries [5]. However. if the »lectron emission yield is still greater than or near unity over the mass range of interest, then alower electron yield can be compensatedby a higher multiplieror amplifiergain and sensitivity is not compromised. It is therefore importantto determine the absolute electron yield for large molecules incident on detectorsurfaces.but much less work has beendc"e in this area. especially at low velocities. Many detailed measurementshav-: :Je\ n mil,"" of electron emission induced by atr 11;ci~ .... or relatrvcl- small polyatomic ions incident on clean. well-characterizedsurfaces [10.11], including measurementsof kinetic emission for projectile velocities as low as 12 kmls [12]. There is also a body of work onelectron emission from hot spikes caused by the impact of cosmic dust at velocities lower than thoseconsideredhere [13]. However. it is not obvious how the results of either regime apply to the situation examined here: detection of very high molecularmass ions (- 100 ku) at low velocities (- 4 kmys) in vacuum conditions typical of commercial mass spectrometers. The electronemission yield for singly-chargedmolecular ions incident on conversion surfaces that have been exposed to atmosphereis usually interpretedin the context of kinetic electron emission [13], where the energy needed to eject electrons comes from the kinetic energy of the incident ion. Conditionsfor potentialelectron emission are not generally met for adsorbed molecules; that is. the projectile'spotential energy released iii the Auger process does not normally exceed twice the binding energy of the target electrons. In kinetic emission. a minimum energy must be transferred to a target electron to cause emission, and this energy corresponds to a threshold velocity for a given projectile-targetcombination. For incident atomic ions two extreme values can be identified for the threshold velocity depending on the mechanism assumed for the energy transfer [13]. For collisions between the incident ion and the fastest free electrons, velocities higher than 150 km/s are requiredfor emission from metals, consistent with extrapolationof the experimentalresults forlight projectiles[14]. However, if some of the momentumof the projectile is transferredto a target atom or the lattice, much lower thresholds are possible. The lowest possible threshold. determined by conservation of energy. corresponds to all the kineticenergy being transferredto the target electron. For an oxygen projectile(the largest abundant atom inorganic molecules) incident on a low workfunrtionsurface, this correspondsto about5lml/s or 0.13
Z83
eV lu [13]. The situation becomes more complicated with polyatomic projectiles, where t:l'lltiple collisions could clearly reduce the threshold further; obviously. conservation of energy is not a limit if the totalenergy of the projectile is - 30 keY. Because the energy transferredto a targetelectronmust exceed its bindingenergy. it is reasonablethat most empirical or theoretical characterizationsof kinetic electron emission have threshold velocity as a oarameter. Such descriptions lead to extraction of threshold values from data acquired at higher velocities, without direct evidence of a true threshold. However. if a different mechanism dominates for velocities near or below theextrapolated "threshold".the term is inappropriate.As early as 1958, measurementsof kinetic electron emission for atomic ions on metals have been reported forvelocities down to 15 km Zs, an order of magnitudelower than the extrapolated threshold [15]. Similar results have been reported more recently [12,16] for velocities as low as 12 kmys [12]. The emission probability in these experiments was very low (:s: 10 - 4) and the measurementswere limited by experimental sensitivity and I1(A an obvious threshold velocity [15,16]. The measurementsof Beuhler and Friedman[1,3] using large water clusters are in the same- velocity range, but because of the large polyatomic projectiles. the emission efficiency was greatly enhanced. Indeed, values of electron yield less than one electron per incident cluster could not be measured by their method.The interpretation of the lowest velocity for which a yield higher than I is observed as a threshold velocity is therefore ratherarbitrary. There is a clear need to determinethe behaviourof the electron yield at lower velocities typically encountered in detecting MALDI ions. The expectation from previous work was that the electron yield would vanish at lowvelocities [1-3}, and that seems to be consistent with some of the measurementsof the ratio of the electron-to-ion yields [5]. However, our previous measurementsof the electron-to-ionratio indicate clearly that theelectron signal does not vanish, at least for energy per mass unit as low as 7km/s (0.25 eV lu) [6,9]. For this velocity, the ratio is - 0.10 and only weakly dependent on mass. These earlier measurementsgive no informationon absolute yield, but measurementsmade at Orsay [7] for 18 keV albumin (0.3 eV lu) incident on CsI give an absolute secondary ion yield > 1 and an electronto-ion ratio ofabout0.06. In this experiment, using single-ion counting methods and a coincidence technique similar in principle to the method used by the Orsay group[7]. we measure the absolute emission efficiency of secondary electrons and secondary ions as a function of the velocity and mass of the molecularions incidenton differentsurfacesfor velocities down to 3.5 kmls (0.06 eV lu). Our primary interes: is to determine the feasibility of using the secondary electrons for detection of high mass, low velocity molecular ions. In particular,for a given accelerating voltage,
G.
Wcsl",UCO/I e
'
al, / Nucl . lnstr. and Mcth. ill Pliy» Res. B 108 (/996 282 - 289
how ( ·)~S the efficiency of electron emission depend on mass, .md what lossof efficiency (if any) must be tolerated tc tal.e advantage of the better time resolution t.ccessible with lctectors that use electron emission"
::, t ..
'.t'rimcntlll
TIle experimentalarrangementshown in Fig. I i<: similar to the arrangementdescribed previously (6.9] with a few modifications. The molecular ions produced at the target by MALDI are acceleratedinto a primaryflight tube (- 83 em long), pass through ;;l ,rnall adjustableiris (- 55 em from the target) and strike a conversion plate (converter) at 45°. Secondary electrons and ior-s produced at the converter are accelerated into a second flight tube (- 18 cm long) and arc thendetected in a 4 cm diameter microchannel plate detector. Amplified signals from the microchannclplates are fed into a25S-stop time-to-digital converter (TDO (model CfN-M2, Institut de Physique Nucleaire, Orsay, France)connected by a custom interface to an Atari Tro30 computerfor daILI storage and analysis. Although single-ion counting methods have limitations, it is easier to interpretthe data quantitativelythan is thecase when a transientrecorderis used. The primary ions were produced by laser pulses of width 10-20 ns from a high repetition fate excimer laser (HE-460-HR-B: LurnonicsInc., Kanata, ON. Canada)incident on the target at 70°. Triggered by a digital delay generator (9650 EG&G PARC), the laser was typically operated at 300 Hz with XeCl (30X nrn). The power density was between 106 and 107 W /cm 2 controlled by a circulargradientneutraldensity disk. The beam focus was manually rastered across the target with a displacement plate. Primary ions were produced from standardsamples of insulin (57~B :1), trypsin (- 23S4D u), human transferrin (- 79500 u) and fi-galuctosidase (- 113600 u). These
I·
\ X
-=-
v."-
=- I.
~-.
- pnma'YIOOs
\
.:
.
.:
Lnsor
-Ll
Irts
G k - .
Targel
\
Convorter
> ~'
/.
".
"~". ~~~~;::,~~' f~·':-=L
cd-a~:' -=
cgo:n ~YTDC}--1Dll
~l
Fig. I. Schematic diagram of the experiment,TIle relevant dimcnsions and voltages are given in thc text,
protein~ were prepared uccontn.g to the usual procedure for r iatrix-assistedlaser desorption (61 using s;'laplOic acid as a matrix. "Primary" ions produced by the laser were acc iler.ited across a potential difference \J~ (tvpiCl!') t\::twen 5 and 25 kV) applied between the l8i" [ and a gr .iunded 909r transmission grid. The accelerated primary ions passed through a small iris with adjustable size, nositioned about .'is cm from tile target. The aperturesize could be varied from zero (completely closed) to -.5 mm'. allowing control cf the primary ion transn.rv.ionto the converter: usually it \I, as set so that the probabilityof transmittingone ion per laser pulse was much less than unity. Secondary ions and electrons w-re produced at the converter from surfaces of stainless steel and CsI; a few measurements were also r.:.lJe with CuBe. The stainless steel surface was cleanec in an ultrasonic bath in acetone, and rinsed with methanol. CsI surfaces were prepared by evaporation onto a clean stainless steel disk under a vacuum of - 100 mTorr.The surfaces were then transported (at atmosphere) into the spectrometer which is subsequzntly pumped to - I X 10 6 TOIT. Secondary ions were accelerated between the converter at - 100 V and a grounded grid (95% transmission). A high transmission grid and low voltage were used to minimize the production of spurious electrons [6,7,9]. To avoid detector saturationfrom the intense secondary electron signal created by the low-mass, primaryions from the matrix. the extraction voltage was pulsed from a small positive voltage to the - 100 V accelerating voltage at some appropriatedelay time after the laser pulse. The front plate of the detector was held at + 4 kV to provide post-accelerationfor the low energy (100 eV) ions and electrons A grounded 95% transmissiongrill about I ern III front of the detector maintained zero field in the flight tube. A magnetic field produced by a coil wound around the detectorwas used to improve secondary electron transmission which is rathersensitive to stray magnetic fields, e.g, the earth'smagnetic field. This applied field, which causes electrons to spiral into the detector, was increased empirically until the transmissionsaturatedat a value about 4D% higher than with zero applied field. The measured field strength for the optimum (6 A through 100 turns) was - lOG near the detector and - 3 G near the converter.
2./, The coincidence method When a large numberof molecularions No are incident on a solid surface(the converter), the numberof secondary electrons ejected N; is given by No = Yo No, where Yo is the secondary electron emission coefficient. Similarly for the secondary ions N, = Y, No. Ideally. our aim is to determine the secondary electron and ion coefficients as a function of mass and velocity for differentsurfaces.
Ii. Wesrmacott et al./ Nllet. Instr. ami Meth. ill Phys. Res. B lOR (!(j96) 282--28)
Ooerall efficiency Although the'emission coefficients are the fundamental quantities,the detection and countingmethod used in these experiments does 'let measure the values of N; and N, directly. When a primary ion strikes the converter, a single-ion counting system will count at most one pulse, even if many electrons are produced. Therefore, with 51: h a system, if No ions are incident on a conversion surface, the number of electron pulses actually counted with a single-ion counting system is =
185
<
N;
E;
<
E;
E;
(I)
(2)
Fig. 2. (a) Time-of-tlight spectrum of secondary ions and secondaryelectrons producedat a slam less steel surface by bombardment with singly-and doubly-charged molecular ions of human Iransferrin(79500 u) at 10 and 20 keV, respectively. The indicated flight limes include the flight lime of theincidentprojectile. (b) Coincidence spectrum recorded with a time window around the secondary electron peak for incident [M·t- Hl". The inset shows the integratedcounts for the indicated peaks tc indicate the degree of random coincidences from different incident projectiles (see text).
Taking account of the probability for detecting an emitted electron Ed' which depends on the secondary electron transmissionand the detectorefficiency, the overall efficiency is
(4) Relation to emission coefficient The emission coefficients 'Ye and Yi can be derived from the measured efficiencies. Assuming the electron multiplicity follows a Poisson distribution[17], the probability for ejecting exactly 11 electrons from a single primary ion impact is
'The validity of Poisson statistics has not been ngcr.csly demonstratedfor low velocity projectiles where 'Y < I, but the assumption is reasonable based on results at higher velocities where 'Y> J [17]. The electron emission efficiency €e correspondsto the probabilityfor ejecting one or more electrons for a single primaryion impact, Thus,
(3)
which enables conversion of electron efficiency to electron coefficient by 'Ye
--In(1 =
<)
~------
.
(5)
Ed
For experimentaldata with appreciableassociated uncertainties, this conversion is only reliable for efficiencies considerably less than unity. As the efficiency approaches 100%, the correspondinguncertaintyin 'Ye becomes more and more magnified. For this reason, although the emission coefficients are more fundamental, the present data are first presentedas efficiencies. This is, in any case, the relevant figure-or-merit in determining the molecular weight range of thetechnique,particularlywhen single-ion counting is used. Only the probability that an ion will produce one or more secondary electrons (or ions) is importantin determiningwhetherit will be detectedornot. The actual number of electrons that are produced only
G. W".,r'll
286
influences the necessary gain of the amplificationelectronic". (The emission coefficient is of COUf!'e relevant to the relative intensity of different peaks when ;\ transient recorderis used. and is therefore related to mass discrimination.)
Random correlations The equationsabove are ..alid underthe condition that only one or zero primary ions strike the surface of the
converterfor every successful laser desorptionevent. Random correlationsOCcurwhen two primary ions strike the converter.distorting the results. To experimentallycontrol the numberof primary ions hitting the converter, an iris (mentioned above) is placed between the MALDI ion source and the converter.TIle apertureis set small enough so that the probabilityfer transmissionof a single primary ion is much smaller than unity to minimize the number of randomcorrelations. To determine the extent of the random contributions. correlationswith a second primary ion projectile (usually the doubly-charged molecular ion) arc also monitored. If the number of events where both the singly-charged (,A) anti doubly-charged\.8) primary ions pass through the iris (AB or BA) is kept very low, then the number of events where two singly-charged (AA) or two doubly-charged (BB) primaryions pass throughthe iris is lower by a factor of 2 assuming all four possibilities have equal probability, Typically, the probabilityfor a singly-and doubly-charged ion to be detected in the same event is less than 50/[. Thus reasonableaccuracy Gin be expected for efficiencies > IOt;~.
The measured time intervals from each laser shot are analysed ami sorted by the data acquisition software and two spectra are recorded. The first spectrum is the full spectrum;it is simply a histogramof all the recordedtime intervals between the laser pulse and the detection of secondary ions and electrons. An example of such a full spectrum is shown in Fig. 2a. The second spectrum (Fig. 2b) is a coincidence spectrum; it is a histogram of the recorded time intervals from selected events in which at least one measuredtime interval falls within a certain lime window. In this experiment the time window is placed aroundthe electron peak from the primary ion of interest. The numberof secondary ions in this coincidence spectrum then represent"the numberof events in which both an electron and ion were detected, i • The efficiencies and (for any single secondary ion species) can then be calculated from the two spectra using Eqs, (t) and (2); the values of N: and N,' are obtained from the full spectrum. To improve the statistics and because individual ion species are not resolved, it was necessary to integrate the intensity for ion species ranging in mass from about I to 125 u for the quantities of and N..:i' However, the development of Eqs. and (2) assumes that multiple secondary ions fromthe same primaryion are counted as a single pulse. This is the case for secondary ions of the
N:
E;
(n
Nt
<
same species hut not usually for secondary ions of differan ent species. To maintainthe validity of Eqs. ( I) and additional constraintwas applied using software 5(1 that at most nne ion within the selected range was registeredfer a given event: in effect the deadtime W;\S increased iii a selected time range. Thus N,' and N:, represent the total number of eticnts in which at least one secondary ion is detected in the selected rangeof I to 125 u.
en.
3. Results and discussion The determination of the secondary electron and ion emission efficiency using the coincidence technique is illustratedfor incident molecular ions of human transferrin (79500 u) at 10 keV (0.13 eV/u). Fig.. 2a shows the full spectrum of the secondary ions and electrons ejected by collisions of the [M + H]+ and [M + 2Hp· ions with the converter. Fig. 2b is the coincidence spectrum recorded with a time window around the secondary electron peak for the [M + t-i]", The numberof counts frorr. the[,\-1 + 2HF· in the coincidence spectrum compared to the full spectrum indicates rhe degree of random correlations. in this case less than 3"!c. On the other hand, the percentage of the ions from the [M + H] + ion in correlationwith the
III
""I:
l:!g
.~~ _CD
~~ c" 0'0 "ii~
.!! ¥ 0.1
e", w ... ~
0.1 1
SS
~~
(
~~Y
.
• tnsultf\ - .... Trypsl" • - Transferrin.+ __e_. Transferrin. • Ogalactosid.se 'i
Ql
1
Energy per mass unit (eVlu)
Fig. 3. Efficiency of secondary electron emissionEo for various incident projectiles on a Csi surface and II stainless steel (55) surface. The efficiency was corrected for known losses due 10 grids. The errorbars represent statistical uncertainty. The curves are fits of the derived electron coefficient (using Eq. (5» to a velocity dependencegiven t-y y = Yo exp( - lie I II).
G, Westmacotl et al./ Nuct. Instr, am ] Melli, in Phys. Res. B 108 (1996) 282,·289
electrons from the same primary ion is about 16%. Thus 16% of the events in which an ion is observed also = N;JN;' = 169c. The data also produce an electron so indicate that in 64% of the events in which an electron b observed, an ion (between 87 and 102 ms) is alsoobserved = N;jN: = 64%. so Fig. 3 shows the secondary electron efficiency as a function of energy per unit mass for insulin, trypsin. transferrinand ji-gn!actosidase incident on (a) a Csf surface and ib) a stainless steel surface. The data have been correctedfor known losses due to grids using Eqs . (3) and (S) in which the geometrical transmissionwas taken as Ed: otherpossible losses .rave not been accountedfor. For both surfaces, the electron efficiency decreases with decreasing velocity for a given mass, and for a given velocity increases for larger masses. Fig. Sb also shows that the secondary electron yields are very similar fer both the singly-charged and doubly-charged molecular ions of transferrinat the same velocity. The efficiency for ion emission for the same projectiles and surfaces is shown in Fig. 4. As has been reported before [7], the efficiency is close to unity even for low velocities, where it is considerablyhigher than the electron efficiency. It is useful to separatethe influence of the mass and the velocity in the electron emission data. If the a..e rage
287
E;
E;
CST]
I
0.1 1~---
+
++
,+ + •
i
.. ~ •
jOlg. 4. Efficiency of secondary ion emission (for ions between I and 125 u) for various incident projecti les on a Csl surface and a stainless steel (SS) surface. TIleefficiency was corrected for known losses due to grids. The error bars represent stadstical uncertainty.
Fig. 5. Reduced electron yield for various projectiles on a CsI surface and a stainless steel (SS) surface. The measured efficiendes were converted to yields eY with Eq. (5) and divided by the mass of \he projectile. TIle lines through the data are fits to Y = {a' ": the exponent is 2.1 for the stainless steel target and 3.3 for the Csi surface. The fit from the 5S data is reproduced as a dashed line on the Csl plot for comparison.
numberof ejected electrons, 1';', is simply proportionalto the primary ion mass, i.e, to the numberof atoms in the molecule, then it is reasonable to normalize the data by dividing 1'0 by the mass, or by the numberof components in the projectile. Such an additivity rule. which assumes the atoms interact independently with the surface, is not universally applicable. and depends on the velocity regime and the size Of the projectiles [3,13}. It appears to hold for water clusters incident on Cu at low velocities [I}, and for peptides incidenton microchannelplates at velocities down to 15 kmy's [18]. Deviations from the additivity rule were reported for water clusters on AI20 3 and for peptides at relatively high velocities (> 100 kmys) on CsI and Al 20, surfaces (J9]. Recent measurements for large multiplychargedmolecularion projectilesat velocities above ~ 30 kmy's show a linear dependence on mass for :1 graphite surface but a sublinear dependence(proportionalto MO,73) for Al 2 0 3 [20]. '01e present data for electron emission (Fig. 3) fall more or less on a single curve if the derived emission coefficient (I'e)' calculated with Eq. (5), IS divided by the mass of the projectile, as shown in Fig. 5. '111e degree of fluctuation in these reduced coefficients is rather large, but Fig. S indicates the data are at least consistent with a linear mass dependence.
288
G. W.·stma cott et a/ . ; Nucl. lnsrr . lind Ml'Ilr. in Ph ys. R.,.<. B 108 ( / 99612 82-289
A non-linear yield enhancement has not been report ed o mbardmentin this for e lectron em iss ionby polyatorn ir b veloci ty range . The deviati ons from linearity menti oned above all corre spond to yields that arc lower than the SLm of the e xpected yields for the constituent atoms. Howe ver. for dust particles at velociti es as low as 0 . 1 krny s, kine tic emiss ion by the ind ividual atoms is not pos sible and the electron emission is attributedto a heat spikeproduced by the impact of the clu ster. clearl y acollective effect [13] . The detection of electrons in this e xperim ent at veloc ities down to 3.5 km/s (0.06 eV lu) similarly sugges ts a cooperative effect may be present even ifcomparisons between various large polyat ornicprojectiles do not indicate an enhancement.The decrease of the electron yield w ith decreasing velocity is character istic of kinetic electron em ission. but there is no indication of threshold.anu a at the lowest velocity the indiv idual atoms do not have suffic ien t energy to induce kineti c em issionatoms ; smaller than 0 have energy less than I cV . The e lectron emi ssio n data for the individual ojectiles pr (F ig. 3 ) fit an exponenti al dep e ndence on the inverse veloci ty, y = Yo ex p{ - /lei v) [13]. but colle ctivel y the reduced data (Fig. 5 ) are better descr ibed by a simple power law. Th e dependence on velocity in this ran ge appears 10 be approximatelyquadratic for the stainles s 'leeI surface, and cubic for the Csl surface. f lf:. 5a and Fig. 5b indicate the difference between the seco ndary electron yield from a non-characterized surface of stai nless steel and a CsIsurface . As previousl yreporte.; [9,19.21] Csl gives a significant enh ancem ent in the ion and ele ctron yieldcompared to stainless
S51
I I 100000 FI ~~I
Time (ns)
Fig. 6. Timc-ot-tl ight spectre of secondary electrons and secondary ions fromthrc- projectiles inciucnt on a Csl surface and a stainless steel(SS ) surface. The indicated flight time includes the: !light time of the incident r rojecule. The electron peaks from the three projectiles(w hich ha:e different velocities) are denoted.TIle clear enhancement observed with a Csi target for high velocity projectiles IS absent at lower velocities (higher times). the energy and primary ion ma ss were varied. However. for a given mass spectrum. all theprojectiles have the same energy and only the mass var ies.T hus, a more pract icalcurve should ind icate the effic iency as a function of mass for a givenenergy. We have used thedata of Fig. S to derive such a plot for severaldifferent accelerating voltages in the range typicalof MALDI; the results are shown in Fig. 7. FOI accelerating voltage at or above 30 kV (the most common value) Fig. 4 shows that detection based on ion emission gives efficiency near unity for the entire mass range. asmentioned above. Fig. 7 shows that detection based on electron emi ssion for this voltage is
11- l ' "
.
...,
0 ."
"
" '~ ' , ,~,> -- 20 keV
"
<,
-
· 3QkeV 40keV SOkeV
. . . . .- -----.....+
0.1 +--~-~10' 10' Mass (u)
10·
Fig. 7. Efficiency for electron emission from a Csl surface as a function of projectile mass for various accelerating voltages. The Sa. curvesare derived from the fit to the nata of Fig.
G. W",lfmucottet uf./Nllc/. lnstr. and Mnh. ill Pbvs. Res. B J08 (/9961282-289
somewhat lower (about SOIk for molecular weight of 100000 u), but still clearly feasible.
4. Conclusion Electron and ion emission was observed fromstainless steel and CsI for incidentprojectilevelocities as low as 3.5 km/s <0.06 eV /u). Using secondary ion emission, the detection efficiency in mass spectrometryof large molecular ions up to at least300000 u can be near 100% if the ions are acceleratedto 30 keY. The efficiency of a detector that relies on secondary electron emission is somewhat lower but still near30'70 for this extreme case. For large molecular ions with energy per unit mass less than0.4 eV /u, the detection efficiency is insensitive to the nature of the surface in a modest vacuum(- 10 6 Torr).
Acknowledgements This work was supported by grants from NSERC (Canada) and from the US NationalInstitutes of Health (GM 30605).
References (I] RJ. Beuhler and L. Friedman,Nucl. lnstr. and Meth. B 170 (1980) ~09. (2] A. Hedin. P. H~kansson and B.U.R. Sundqvist, Int. J. Mass Spcctrorn.Ion Process. 75 (1987) 275. [3] R.J. Beuhler, J. Appl. Phys. 54(1983) 4118. (4) F. Hillenkarnp,M. Karas, R.C.Beavis and B.T. Chait, Anal. Chern. 63 (1991) 119~A.
289
[5] B. Spengler. D. Kirsch. ~. Kaufmann, M. Karas, F. Hillcnkamp and I.'. Gicssm.mn. Rapid Commun. Mass Spectrom. 4 (1990) 301. [6J J. Martens, W. Ens, ~ .G. Standing and A. Vercntchikov, Rapid Commun. Mass :'pcctrom.6 \1992) 147. [7] A. Brunelle, P. Chaur,.d. S. Della-Negra, Y. l.ebeyec and G.B. Baptista, Int. J. Mass Spcctrom. Ion Process. 126 (199.') 65. [8] H. Haberlandand M. Winterer, Rev. Sci. Insi.. 54 (1983)
764[9] A. Verentchikov, W. Ens. J. Marten,and K.G. Standmg, Int. J. Mass Spectrom.Ton Process. 126(1993) 75. (10) D. Hasselkamp,H. Rothard,K. -0. Groeneveld, J. Kemmler, P. Varga and H.Winter, Particle Induced Electron Emission \I (Springer,Heidelberg, 1992) and references therein. [ll] R.A. Baragiola. in: Low Energy Ion-Surface Interactions, Ed. J.W, Rabalais (Wiley, Chichester, 1994) r, 187. and references therein. [12] u. Lakits and H.Winter, Nucl. Instr. and Meth. B 48 (:9 90) 597. [13] R.A. Baragiola, Nucl. lnstr. and Mcth B 88(1994) 35, and references therein. i:~] RA. Baragiola. E.V. Alonso and A, Oliva, Phys, Rev. B 19 ([979) 121. [15J P.M. Waters, Phys. Rev. III (1958) 1053. [16} E.V. Alonso, M. Alurralde and R.A. Baragiola, Surf. Sci. 166 (1986) L155. [17J K. Toglhofer, F. Aumayr, H. Kurz, H.Wimer,P. Scheier and T.D. Mark,Nucl. Instr. andMeth. R 88 (1994) 44. [18] P.W. Geno and R.D. Macfarlane,Int. J. Mass Spectrom, Ion Process. 92 (1989) 195. [19] R. Mosshammerand R. Matthaus,J. Phys. (Paris) C 2 (1989)
III. [20} J. Axelss» '. ( 1 I,. ..Ihl II 1 R Sundqvist, Nucl. Instr. and Meth.13 ~,1'- ( ' I ' •. /l I l i. [21] S. Della-Negra, C. Deprunand Y. Lelseyec, Rapid Commun. Mass Spectrom. 1 (1987) 10.