422 Brazilian Journal of Physics, vol. 29, no. 3, Septemb er, 1999
An Intro duction to the Time-of-FlightTechnique
Per Hakansson
The Angstrom Laboratory,
Div. of Ion Physics, Box 534, S-751 21 Uppsala, Sweden
Received May, 1999
In the last two decades, several new typ es of ion sources have b een intro duced in organic mass
sp ectrometry likebombardment with heavy ions, PDMS, and laser light, MALDI. Hand in hand
with the new ionization techniques followed also a renaissance for the 50 year old time-of- ight
technique for the mass analyses. In this pap er the basics of the ToF technique is describ ed in a
tutorial non-theoretical way for the b eginner together with some practical hints. The electrostatic
mirror, the delayed extraction technique as well as some recent technical developments are also
included.
ments like the Q-ToF: a quadrup ole is used to select a I Intro duction
certain mass, the parent ions. By collisions with gas
A mass sp ectrometer is build up by three ma jor parts:
molecules in a small cell the parent ions will break up
an ion source to create ions of the sample to b e inves-
into fragments or daughter ions. The masses of the
tigated, a mass analyzer to determine the mass distri-
daughter ions are then analyzed with a time-of- ight
bution of the ions from the source and a detector to
analyzer. This technique for obtaining structure infor-
detect the ions that have b een selected by the mass an-
mation ab out molecules is called MS/MS.
alyzer. A common mass analyzer is a combination of
A drawback with the ToF technique has b een the
magnetic and electric elds in a so called sector instru-
relatively p o or mass resolving p ower due to the spread
ment. Other p ossibilities are to use quadrup oles or ion
in the initial energies of the ions as well as the spa-
traps.
tial distribution of them. However, with the use of the
The idea to measure the time that ions, with a
delayed extraction technique [7] and electrostatic mir-
known energy, need to travel a certain distance and
rors [8] the mass resolving p ower b ecomes sucient for
then calculate their mass was rst exploited by Ham-
most applications to day [9]. In this short pap er the
mer in 1911. The use of ToF in mass sp ectrometry was
basic ToF concepts will b e describ ed in a p opular way
consolidated by W. E. Stephans [1] in 1946, Later in
together with references for further reading. A recent
1948, A. E. Cameron and D. F. Eggers [2] at Clinton
review article ab out ToF has b een written by Guilhaus
engineer works, Tennessee, intro duced an instrument
[10] and a nice b o ok by Cotter [11].
containing the basic building blo cks of a mo dern ToF
instrument: an ion source with an acceleration region
followed by a eld free region and with a stop detector
II The straight sp ectrometer
at the end.
However, the resolving p ower of a ToF instrument
Consider a straight sp ectrometer as in Fig 1. The sam-
is p o or compared to a sector instrument and it was
ple molecules to b e studied are dep osited on a metallic
not commonly used until Macfarlane [3] et. al. intro-
target backing kept at the acceleration p otential U in
duced the PDMS technique, plasma desorption mass
front of a grounded grid. The target is b ombarded with
sp ectrometry, in 1974. Then the advantages with the
for e.g. fast heavy ions from an accelerator. The start
ToF technique b ecame clear: no scanning is needed like
signal could b e generated from the accelerator itself or
in a sector instrument, high transmission, high sensitiv-
from a start detector that the b eam pass just b efore
ity, fast, cheep, simple and in principle unlimited mass
the target. The start detector contains a thin foil that
range.
pro duces secondary electrons when the fast heavy ion
passes. The burst of electrons is ampli ed with channel Today the ToF technique is well established and
plates. The secondary ions from the target are accel- used in combination with quite di erent ion sources like
erated into the eld free region with length L where in SIMS [4], secondary ion mass sp ectrometry, MALDI
they drift with constantvelo cityuntil they reach the [5], matrix assisted desorption/ionization and ESI [6],
stop detector. Dep ending on how the stop detector is electrospray ionization. There are also hybrid instru-
Per Hakansson 423
coupled the secondary ions will b e p ost accelerated or slightly retarded.
Figure 1. Working principle of a straight time-of- ight mass sp ectrometer. Molecules are desorb ed and ionized when fast
heavy ions from an accelerator hit the target at high voltage. The secondary ions are accelerated through a grid and enter
a eld free drift tub e. At the end, the ions are stopp ed in micro channel plate detector which gives a stop pulse to the
digital clo ck. The start pulse comes from a burst of electrons that are generated when the b eam passes a thin foil. The time
di erence b etween start and stop pulses are prop ortional to the square ro ot of the ion mass. The main principle is the same
252
for b ombardment with ssion fragments from Cf source PDMS or keV ions SIMS or laser light MALDI.
424 Brazilian Journal of Physics, vol. 29, no. 3, Septemb er, 1999
The energy of the ions when entering the accel-
eration region is qU: This is equal to kinetic energy
2
E =1=2mv , where the velo city v = L=t. The basic
k
equation for the time- of- ightisthus
p
t = L m=q U
Even if the o set time b etween the start pulse and
the time when the primary ion hits the target and the
time sp ent in the stop detector are taken into account
the same simple relation holds:
p
ToF = A m + B
where A and B are constants. To mass calibrate an
unknown sp ectrum it is thus enough to determine the
ToF for two known masses and then calculate A and B .
The start and stop signals are fed into a TDC, time
to digital converter. This is a digital clo ck that will give
as output a numb er that is prop ortional to the time dif-
ference b etween the start and stop pulses. This number
will b e the channel address in a sp ectrum. The content
of that channel will incrementby 1. In this way a his-
togram sp ectrum will b e built up step by step. This
way of recording data is called event-by-eventmodeor
single ion counting: all stop signals followed one start
signal are registered. To a given start signal, can follow
stop signals from light ions as well as from the molecular
ion itself. Every event can contain information ab out
the whole mass range. This is in contrast to a mag-
netic analyzer where the magnetic eld is scanned and
the intensity of the corresp onding masses are recorded.
III Detectors
The standard detector used to day is an assembly of
two micro channel plates coupled together as in Fig
2. Across the plate is a p otential di erence of ab out
Figure 2. The typical detector in a time-of- ight sp ectrom-
1000 V. The plates are built up of small channels that
eter consists of two micro channel plates coupled in tandem.
6
are tilted at an angle with resp ect to the surface nor-
The total gain is ab out 10 . The plates can b e coupled ei-
ther with the ano de plate grounded a or at p ositive high
mal. When an ion hits the channel it will pro duce some
voltage b.
electrons that are accelerated into the channel where it
will hit the wall and pro duce more electrons and so on.
Each plate has a gain of ab out 1000. After two plates To trigger a micro channel plate detector it is in
so many electrons are pro duced that it is p ossible to most cases enough to just let the ions hit the front plate
directly.However, when the pro duction of secondary
detect a pulse from the ano de. The channel plate sig-
electrons is to o low, the ion is forced to hit a thin foil
nals are fast, whichisamust for go o d timing. The rise
in front of the detector. The secondary electrons from
time of the ano de pulse is less than a ns.
the foil will then b e accelerated into the plates. To
The plates can b e coupled in two di erentways.
enhance the electron yield and to make it p ossible to
The ano de can b e at ground p otential, which means
go to high p ostacceleration voltages, it is also p ossible
that the front end of the rst plate must b e negatively
to use a converter plate b eside the detector. The sec-
biased. This has the consequence that p ositive ions will
ondary electrons from the plate are then steered into
b e p ostaccelerated but negative ions will b e retarded.
the detector by a magnetic eld.
The ano de can also b e oated at high p ositivevoltage.
In that case, the front plate is at ground p otential and A drawback of the channel plates is that they have
the secondary ions will not b e in uenced in the stop a recovery time in the s time range. This means that
detector region. ifachannel is hit by an ion it will take a certain time
Per Hakansson 425
the output pulse from the detector will b e twice the b efore it is ready for detecting the next ion. If the
pulse for one ion, but the TDC will still count itasone stop rate is low like in PDMS or SIMS this e ect has
ion. In applications with a high probability that many no in uence, but if many ions come p er start likein
molecular ions with the same mass are desorb ed at the MALDI the high low-mass count rate will shadow the
same time, like in MALDI, it is therefore b etter to use a interesting high-mass ions.
transient recorder or a digital oscilloscop e in order not This problem can b e resolved in many di erent
to lo ose statistics in the p eak. ways. One of the channel plates can b e pulsed. When
This problem can also b e circumvented by using a the low-mass ions come the detector, it is not active but
segmented ano de [13] and add up contributions from after a certain time the detector gets full high voltage
each part. At IPN the electronic department has also and can record the high-mass ions. Another wayisto
develop ed a charge recorder that integrates the stop use a set of de ection plates or anytyp e of ion gate
detector p eak and give an output pulse prop ortional to and de ect awayunwanted low-mass ions from the sec-
the area. ondary ion b eam. The gate is then switched o when
the interesting molecules comes.
V The electrostatic mirror
IV Electronics
Apowerful way of comp ensating for the spread in the
initial energy distribution of the secondary ions is to To record a ToF sp ectrum it is, in most cases, enough
to use a timing ampli er for the detector signal followed re ect the b eam in an electrostatic eld, see Fig 3.
by a constant faction discriminator, CFD. This mo dule Consider a package of ions having the same mass and
gives a well-de ned timing signal indep endently of the approaching the mirror after b een accelerated to a cer-
amplitude of the stop pulse from the detector, the so tain energy. Due to the initial energy distribution some
called \amplitude walk" phenomenon. ions will havelower energy i.e. moving with to o low
In applications with few stops/start the time can b e velo city and some higher energy i.e. moving to o fast
measured with an analog time to pulse height converter, than the correct one. Ions with to o high energy will
TPHC, or b etter a digital time converter, TDC. A go o d p enetrate deep er into the mirror b efore they are re-
converter, like the CTN-M2 built by the IPN at Orsay ected back, compared to the ones with correct energy.
[12], has a resolution of 0.5 ns, a time range of 256 s Thereby they will lo ose time. Ions with to o low energy
and can handle 256 stops p er start. The dead time is 20 will not p enetrate so deep into the mirror b efore they
ns. It sounds more than enough the quantity 256 stops are re ected back, compared to the ones with correct
p er start, but these stops must b e of di erent masses. energy. Thereby they will gain time. The net e ect is a
If two ions arrive at the same time in the stop-detector time fo cusing of the ion package at the detector plane.
Figure 3. A time-of- ight sp ectrometer equipp ed with an electrostatic re ector or ion mirror for improved mass resolving
power. The mirror is built up by high precision stainless steel rings separated with high precision ceramic balls. A resistor
chain distribute the mirror voltage over the rings uniformly. A stop detector at the rear of the mirror gives the p ossibili ty
to study metastable decays by coincidence measurements b etween that detector and the detector for the re ected ions. The
sp ectrometer has also an einzel lens to fo cus the secondary ion b eam and a set of de ection plates for steering the b eam
correctly into the mirror.
For a single stage mirror, Fig. 3, the length of the two stage mirror that comp ensates for higher order ef-
mirror should b e one quarter of the total eld free path fects in the time-of- ight equation for the system. This
which means that the total dimensions of an instru- is obtained by inserting one more grid in the mirror and
ment can b e quite large. A more compact mirror is a set the dimensions and voltages correct.
426 Brazilian Journal of Physics, vol. 29, no. 3, Septemb er, 1999
The comp ensating e ect on time-of- ight instru-
ment with a mirror can easily b e demonstrated by
changing the acceleration voltage and monitor the ToF
for any ion. With a single stage mirror [14] one can typ-
ically change the acceleration voltage +150V and the
p eak p osition is the same within 1 ns. Foratwo stage
mirror the comp ensation is roughly 10 times higher [15].
VI Delayed extraction
Delayed extraction or time lag fo cusing was invented
by Wiley & McLaren already in 1955. Howpowerful
this technique is b ecame however not clear until Brown
[16] and Vestal [17] applied it to MALDI. The working
principle is the following. Consider two identical ions
that are pro duced with di erent initial energies. In the
standard ion source geometry with one grid they b oth
gain the same amount of energy in the acceleration pro-
cess. After the grid, the ion with highest initial energy
will continuously increase the distance to the ion com-
ing b ehind and it will reach the detector rst.
Consider now the pulsed case, with an extra grid
between the sample and grounded grid, Fig. 4. In the
desorption moment, the target and 1st grid are on the
same p otential. The ion with the highest initial energy
will move longer out in the acceleration region than the
Figure 4. In delayed extraction or time lag fo cusing an extra
ion with low initial energy. After a certain delay time,
grid is mounted b etween the target and grounded acceler-
ation grid. In the desorption moment, t=0, the target and
the high voltage is increased on the target to pro duce
rst grid are on the same p otential U2. After a certain de-
a linear acceleration eld. However, in this case the
lay,t the target p otential is quickly switched from U2
delay
slower ion will b e accelerated more than the faster ion.
to U1 and the ions are extracted from the ion source. An
By cho osing correctly the distances and delay, the ions
ion with high initial energy will b e less accelerated than an
can b e made to reach the stop detector at the same
ion with low initial energy.Bycho osing the delay time and
the distances correctly a net fo cusing e ect can b e obtained
time.
at the stop detector plane.
VI I Conclusion
References
From b eing a technique that was hardly accepted 20
[1] W.E. Stephans, Bulletin of the American Physical So-
years ago, ToF is to daywell established and the num-
ciety, 21, 22 1946.
b er of commercial instruments on the market increases
[2] A.E. Cameron and D.F. Eggers, Rev. Sci. Ins., 19, 605
fast. New concepts and ideas are constantly presented.
1948.
Vestal et al have develop ed a ToF MS/MS instrument
[3] D.F. Torgerson, R.P.Skowronski and R.D. Macfarlane
with a collision cell [18]. Ens et.al. has develop ed an
Bio chem. Biophys. Res. Com. 60, 616 1974.
instrument with a pulsed orthogonal extraction ToF an-
[4] W. Ens, K.G. Standing, B.T. Chait and F.H. Field,
alyzer with two ion sources: electrospray, ESI, or laser
Anal. Chem. 53, 1241 1981.
light, MALDI [19]. Gamini et al has develop ed an ion
[5] M. Karas, D. Bachmann, U. Bahr and Hillenkamp, Int.
gate for ToF MS/MS exp eriments with a mass resolv-
J. Mass Sp ectrom. Ion Pro cess, 78, 53 1987.
ing p ower of up to 5 200 [20]. The same group has
[6] J.G. Boyle, C.M. Whitehouse and J.B. Fenn, Rapid
also demonstrated a mass resolving p ower of 55 000 for
Comm. Mass Sp ectrom. 5, 400 1991.
molecular ions of substance-P [21] in a MALDI exp eri-
mentby using multiple re ections in two mirrors facing
[7] W.C. Wiley and I.H. McLaren, Rev. Sci. Ins., 26, 1150
each other. The era of ToF has just started! 1955.
Per Hakansson 427
[16] R.S. Brown and J.J. Lennon, Anal. Chem. 67, 1998 [8] B.A. Mamyrin, V.J. Karata jev, D.V. Smikk and V.A.
1995. Zagulin, Soviet Phys. JEPT 37, 45 1973.
[9] P. Juhasz, M. L. Vestal and S. A. Martin; Optimization
[17] M.L. Vestal, P. Juhasz and S.A. Martin, Rapid Comm.
Strategies in Delayed Extraction MALDI TOF, Pro-
Mass Sp ectrom. 9, 1044 1995.
ceedings of the 45th ASMS Conference on Mass Sp ec-
trometry and Allied Topics, Palm Springs, California,
[18] M. Vestal, P. Juhasz, W, Hines and S. Martin; An
June 1-5, 1997, p 1100.
Improved Delayed Extraction MALDI TOF MS for
[10] M. Guilhaus, J. Mass Sp ectrom. 30, 1519 1995.
PSD and CID Pro ceedings of the 46th ASMS Confer-
ence on Mass Sp ectrometry and Allied Topics, Orlando,
[11] R.J. Cotter, Time-of-Flight Mass Sp ectrometry,ACS
Florida, May 31-June 4, 1998, p 35.
Professional Reference Bo oks, American Chemical So-
ciety, 1997.
19. W. Ens et al, A MALDI Source with Collisional
Co oling for Orthogonal-Injection TOF Presented at
[12] R. Sellem, Service Electronique Physique, Institut de
the Desorption `98 conference, Rio de Janeiro, Brazil,
Physique Nucl eaire, B^at 102, F-914 06 Orsay,France.
Septemb er 21-26, 1998.
[13] S. Della-Negra et al, The Multi-Impact Lo calisation
Detector LAG 1, Presented at the Desorption `98 con-
[19] C. K. G. Piyadasa, P.Hakansson, T. R. Ariyaratne and
ference, Rio de Janeiro, Brazil, Septemb er 21-26, 1998.
D. F. Barofsky Rapid. Commun. Mass. Sp ectrom. 12,
[14] G. Brinkmalm, P Hakansson, J. Kjellb erg, P. Demirev,
1655 1998.
B. U. R. Sundqvist and W. Ens, Int. J. Mass. Sp ectrom
[20] C. K. G. Piyadasa, P.Hakansson, T. R. Ariyaratne,
Ion Pro cesses 114, 183 1992.
Submitted to Rapid. Commun. Mass. Sp ectrom.
[15] A. Brunelle, S. Della-Negra, J. Depauw, H. Joret, Y. Le
Beyec, Rapid Commun Mass Sp ectrom. 5, 40 1991.