Characterization of a Glow Discharge Ion

Source for the Mass Spectrometric Analysis of

Organic Compounds

D. Carazzato and M. J. Bertrand

RegionalCenter for Mass Spech’amehy, Department of Chemistry, University of Mont&l, Montr6a1, Canada

A glow discharge ion source has been constructed for the mass spectrometric analysis of
organic compounds. Characterization of the source has been made by studying the effect of
pressure and discharge current on ionic distributions by anodic ion sampling along the
discharge axis. Ion and electron densities and electronic temperatures have been calculated
by using the single Langmuir probe technique to correlate the extraction efficiency with
measured ion distributions and gain some insight into the ionization of organic molecules.
The spectra obtained for several classes of organic compounds show that formation of
parent-molecular ions by proton transfer, resulting partly from the background water
molecules, is a major low energy process while charge transfer, Penning ionization, and
electron ionization ace probably responsible for the fragmentation observed. The spectra
result from the simultaneous occurrence of high and low energy reactions, and their
structural information content is very high, yielding both molecular and extensive fragment
ion information. The glow discharge ion source has proved to be essentially maintenance-free,
easy to operate, stable, and can be used at reasonable mass resolution (up to 70001. The
source also provides picogram range detection limits and has a linear response range of
about six orders of magnitude, which makes it an interesting ion source for routine analysis.
Preliminary work conducted with chromatographic interfaces indicates that its use can be
easily extended to both gas and liquid chromatography. (1 Am Sot Mass Spectrom 1994, 5,
305-325)

G
low discharge devices have been mainly used and operating conditions, it may or may not leave
as hollow cathode lamps for producing charac- enough space for a positive column to develop. In
teristic emission lines in spectroscopic tech- most applications, however, the discharge is resticted
niques, and since their introduction in mass spectrom- to the cathode fall and the negative glow. The tech-
etry in the early 1970s 11-101, they have been used nique uses positive ions formed and accelerated
extensively in inorganic analysis 111-131. Some of these through the cathode fall to bombard the cathode to
applications involve the analysis of thin metallic films eject atoms and particles. This process, called sputter-
[14], bulk materials, and semiconductors [15-M]. Al- ing [20], is followed by diffusion of the neutrals to-
though their design can differ from one application to ward the negative glow. Ionization mostly takes place
another, they may all be represented by a classical gas in this region by Penning ionization [Zl, 221 when
tube in which one of the electrodes (cathode) is made metastable atoms in the discharge collide with sput-
of the sample material and the other (anode), of lesser tered particles, producing positive ions and electrons
importance, floats at the source block potential or is with a range of energies that depends on the nature of
simply grounded. The tube is usually filled with a rare the discharge gas. Because the electrons in the glow
gas at a pressure in the range of 0.01 to 1 torr. In the discharge have a kinetic energy distribution ranging
case of a direct current discharge, a potential of a few from slow to fast (0.3-25 eV) 1231, electron ionization
hundred volts, negative to anode, is applied to the (EI) may also occur but its contribution to ionic pro-
cathode to initiate and maintain the discharge, which duction is usually much less than the Penning process.
usually involves a current density at the cathode rang- Except for its two boundaries, where sharp potential
ing from 0.1 to 10 mA/cm’. The discharge appears barriers are observed, the glow is a relatively field-free
nonhomogenous, showing glowing regions and dark region [241 and, thus, ion diffusion may lead to ion
spaces 1191, and, depending on the source dimensions losses by recombination with slow-moving electrons in
the glow or at the walls or changes in ion identity by
ion-molecule reactions such as proton exchange and
Address reprint requests to M. J. Bertrand, Regional Center for Mass
Spectrometry, Department of Chemistry, University of Mont&l, P.O.
clustering. The ions observed in the mass spectrum are
Box 6128, Station A, Mm&al, Canada H3C 3J7. those arriving at the anode sheath in the vicinity of the

0 1994 American !%xiety for Mass Spectrometty Received June 18,1993

1044-0305/94/$7.00 Revised December 8,1993
Accepted December 8,1993
306 CARAZZATO AND BERTRAND J Am Sot Mass Spectrom 1994,5,305-315

electrode aperture. They are accelerated across the suitable for GC or LC operation. It also discusses the
sheath and then mass analyzed. Ions formed by colli- effects of the different glow parameters mentioned
sions with high energy electrons in the cathode fall above on analyte signal and rationalizes the probable
region will not be observed in the mass spectrum as ionization processes occurring in the source.
they will not be extracted from the discharge. Al-
though the use of glow discharge sources mainly con-
Experimental
cerns inorganic analysis, in which they are used to
both sputter and ionize the analytes, they have also All experiments were performed using an upgraded
been used for the analysis of organic compounds. Kratos MS-9 mass spectrometer (Kratos Analytical,
Tanabe and Wiiefordner [25] have used the afterglow Ramsey, NJ) connected to a VG ZAB console (VG
of an electrical discharge under reduced pressure (l-10 Analytical, Manchester, UK). Some modifications have
torr) to generate metastable argon, helium, and nitro- been made to the original instrument because it had
gen to study the excitation and fluorescence of benzene not been designed to accept large amounts of gas. Two
and anthracene, while Anderson et al. [26] have used diffusion pumps (total of 225 L/s) were added to the
the same technique as an ionizmg medium. Unusual source housing in addition to an E2-M2 roughing pump
reagent gases for chemical ionization known to rapidly (Edwards, West Sussex, UK) connected to the glow
deteriorate the source filament, such as nitrogen oxides discharge source itself. Despite these additions, the
[27] or methanol [28], performed successfully in a glow pressure in the housing sometimes reached 2 X 10m4
discharge device, providing mass spectra similar to torrs during operation and the accelerating voltage
those obtained by the filament method but with the was kept at 4 kV to prevent source arcing. A vacuum
advantage that the glow discharge source was essen- lock (from a VG 12-12 mass spectrometer) oriented in
tially maintenance-free. Townsend [29, 301 and corona the beam axis has also been added to the source flange
[31] discharges have also been used to generate cherni- to introduce the probe in the high vacuum system. The
cal ionization (CI) plasmas in both negative and posi- direct current discharge was obtained by using a
tive ion mode with good detection limits (low homemade 1450 VDC variable power supply floating
picogram) and reduced fragmentation at pressures at the accelerating voltage. Beam centering and “z”
ranging from 1 torr to atmospheric pressure. Introduc- focusing plate controls, required by the use of a small
tion systems for glow discharge analysis have also circular aperture in the source, have been added in the
been investigated quite extensively. Air sampling via a source region with no other modifications to the VG
thin aperture has been used for the analysis of TNT in ZAB console. Calibration of the instrument was
the negative ion mode with high sensitivity (0.4 pg) at achieved with a very small quantity of perfluoro-
a sampling rate of 5 mL/s [32]. Samples can also be kerosene (Pfaltz and Bauer, Waterbury, CT) injected in
deposited on a probe, although in this case thermal the argon line. During the optimization experiments,
desorption rather than sputtering or fast-atom bom- the sample (butylbenzene) was placed in a tempera-
bardment-type ejection of ihe sample from the probe ture controlled vial to maintain a constant vapor pres-
surface controls the sample concentration in the plasma sure, and hence, a constant flow rate of the analyte
[33]. Nebulized liquid chromatographic effluents, us- through a 150 pm id. x 360 pm o.d. quartz capillary
ing helium as drift gas followed by atmospheric pres- (Polymicro Tech., Phoenix, AZ) inserted in the move-
sure ionization [34, 351, have also been analyzed and able cathode. The discharge gases used in this work
have been used to demonstrate the versatility of the were those employed for GC (zero grade, Cryo Gaz,
technique. In general, the mass spectra obtained with Montreal, Canada), and no specific trap was added on
glow discharge ionization are of the CI-type, showing the gas lime. All chemicals were purchased from
adducts, protonation, and some fragment ions. How- Aldrich Chemical Co. (Milwaukee, WI). The pressure
ever, despite the numerous analytical applications cited in the source was measured with a Baratron gauge
above, only a few studies have probed (1) the effect of (MKS Instruments Inc., Andover, MA) and could be
the different source parameters on ionization, (2) the controlled between O-4-1.2 torr with argon flow rates
behavior of the organic analytes in the discharge, or (3) of 0.554 mL/min. During all experiments the source
the relative contribution of the ionization mechanisms. was kept at 150 “C to avoid organic deposits. Sensitiv-
Parameters such as the sampling distance (interelec- ity studies were conducted, using a splitless GC injec-
trode spacing), source pressure, discharge current and tor installed on the argon line, by injecting different
voltage, and the nature of the discharge gas used have gaseous concentrations of the analyte. When a liquid
important effects on the signal intensity and quality. effluent was used, a stainless-steel tubing, 0.72 mm
The use of a rare gas as a discharge gas makes the o.d. X0.43 mm id. (Small Parts Inc, Miami, FL), con-
technique compatible with gas chromatography (GC) taining a 15 pm X 60 cm quartz capillary was resis-
as well as with capillary liquid chromatography (LC) if tively heated by a homemade 10 V (100 watts) power
the liquid effluent is vaporized and used as a dis- supply floated at the cathode potential to provide good
charge gas. The present work focuses on organic analy- vaporization of the sample. A Rheodyne 7125 LC injec-
sis. It describes and discusses the construction and tor (Cotati, CA) connected to a high pressure syringe
operation of a direct current glow discharge ion source pump, Pheonix-20 (Carlo-Erba Str., Milano, Italy),
J Am Six h&s +ctmm 1994,5, 305315 GLOW DISCHARGE FOR ORGANIC COMPOUNDS 307

operated at a flow rate of 3.1 pL/min (70:30, limit its conductance to maintain, under normal oper-
CH,CN/H,O) was used for the Introduction of the ating conditions, the pressure in the source at around 1
samples. In this case, the evaporated mobile phase was torr. The whole source assembly has been adapted to
used as discharge gas. Mass spectra were recorded at a the quartz pillars of the original source. In addition, a
resolution of 2000 (10% valley) and a scan speed of gas inlet, and a roughing pump outlet have been
10 s/decade using a Kratos DS-90 data system. The added to the glass cell to control the source pressure
Langmuir probe measurements were conducted in a when a liquid effluent was introduced into it. The
t-shaped Pyrex* glass discharge tube similar In di- purpose of the glass cell is to provide a linear and
mensions to those of the source cell, in which a thin axially symmetrical discharge between the electrodes
brass cylinder could be inserted as anode. The probe, to sample the different regions of the glow, which are
which was inserted through the side of the glass cylin- also shown in Figure 1. By moving the cathode closer
der, is composed of a copper wire 3 mm long by 0.25 to the anode from 17 mm to 0 mm, it is possible to
mm diameter and was placed in the center of the glow, sample the plasma until the glow extinguishes. The
on the discharge tube axis. The whole arrangement is choice of axial sampling was based on studies of charge
similar to the one of Fang and Marcus [36], except that [9, 19, 371 and metastable [38] radial distributions in
Swagelok@ fittings were used for the connections to glow discharges, showing that molecular species were
the discharge tube. The direct current voltage for the found in higher concentrations at some distance from
Langmuir probe was provided from serial combina- the cathode and from the walls. It also considerably
tions of 22.5 V batteries, and the probe current was simplifies the source construction and alignment and
measured with a Keithley 485 picoammeter. For the the interpretation of the results by eliminating dis-
Langmuir probe measurements, the discharge was ob- charge asymmetry and problems related to inhomoge-
tamed by using the same homemade power supply neous sampling as the discharge conditions are
used for our glow discharge source on the mass spec- changed. With this arrangement, the surface of the cell
trometer. takes a negative potential (wall potential) [20] that
reflects the secondary electrons back into the glow,
which may increase the ionization efficiency.
Results and Discussion
GlowDischarge Source Operating Parameters

The glow discharge source used in this work was Electrical. The effect of the discharge current and
designed to fit a Kratos MS9 mass spectrometer and is voltage on ionic species present in the glow will be
illustrated in Figure 1. It comprises a moveable cath- discussed further on. However, because the current-
ode and a fixed extraction anode placed at both ends voltage relationship depends on factors such as the
of a cylindrical Pyrex glass cell. The cathode is made of nature of the cathode and the discharge gas, (Town-
a 998@ ahunina tube (McDanel Refractory, Beaver Falls, send’s factors), pressure, tube geometry, and distance
PA) into which is inserted a stainless-steel capillary between electrodes [39], it is difficult to directly com-
ended by a copper disk joined to the ceramic by pare one source to another and each design will have
Torr-Seal” epoxy resin (Varian, Lexington, MA). The its own characteristics. A breakdown potential (Paschen
conical sampling anode is made from commercial brass. type) curve obtained at a pressure of 1 torr is shown in
It has a sampling hole (0.4 mm dia.) drilled in the flat Figure 2. It shows a minimum sparking potential that
surface which is 5 mm in diameter with a thickness of depends on the factors mentioned previously and also
0.25 mm. The size of the hole was chosen such as to on small amounts of impurities present which can
drastically change the sparking values, as was ob-
served in these experiments. This minimum indicates
the position of a virtual anode [ZO]and roughly corrc-
sponds to the tailing of the negative glow. However,
the abscissa on Figure 2 cannot illustrate p x d, as is
the case in typical Paschen curves, because the rela-
tively small dimensions of the source restrict the dis-
charge to a certain volume and may not leave enough
room for the negative glow at low pressure. The cur-
rent-voltage dependence is illustrated in Figure 3 for
argon pressures of 1.5, 1.0, and 0.5 torr and an inter-
electrode spacing of 15 mm. As shown on the graph,
the current increases rapidly with the applied voltage.
This current-voltage characteristic indicates that a ca-
thodic saturation occurs at very low applied voltages.
Under these conditions the discharge is called
Figure 1. Schematic of the glow discharge ion source. “abnormal” and is usually employed for inorganic
308 CARAZZATO AND BERTRAND J Am 5oc Mass Spechwm 1994,5,305~315

Figure 4. Axial sampling of the argon discharge at a pressure of

Figure 2. Breakdown potential with distance at a pressure of 1 1.0 torr; 0 Ar;; 0: ArH+; A: AI+; 0: 0;; +: Ar++; X: H,O+;
torr (argon). 0: H,O+; n : discharge current along the axis at constant volt-
age.

analysis in sputtering devices. However, the ranges in are fewer high energy particles. Here the molecular
currents (20-200 PA) and pressures (0.5-1.5 torr) used species are more likely to remain intact (the cathodic
in this work do not favor cathodic sputtering and edge of the glow corresponds approximately to where
heating and, under normal operating conditions, the the argon ions are the most abundant). These results
signal for copper (cathode material) was either absent agree well with those of Knewstubb and Tickner 141,
or extremely weak. 421, which were obtained by using a glass sampling
probe with a 90” discharge-sampling geometry. The
Sampling distance. To evaluate the ionic distributions uneven ionic distributions can only be partially ex-
ln the discharge, a 17 mm longitudinal cut of the plained by differences in ion mobilities in the glow
plasma was made and the results are shown in Figure because Are’, Ar+, and A$, which have different
4. The intensities for a few important plasma ions are mobilities, show quasicoincident maxima. This has also
plotted against the distance between electrodes. In been reported by Knewstubb and Tickner. In fact, the
these experiments the anode acts as a sampling probe high energy electrons that enter the negative glow lose
and its position in the glow does not significantly kinetic energy by successive collisions, with neutrals
affect the glow itself [401. As can be seen on the graph, producing initially doubly and singly charged species
the plasma is not homogenous because singly and and then excited argon atoms. Thus, higher ionization
doubly charged argon ions are found closer to the states of argon should be observed closer to the cath-
cathode. On the other hand, molecular species are ode dark space at the edge of the negative glow, but
found in the tailing of the negative glow where there this sets the cone into the cathode fall, thus, inhibiting
the ion extraction efficiency and reducing the recorded
signal. In this region, the high energy electrons can
also fragment the molecular species by electron im-
pact, which would explain the absence of the latter.
Furthermore, the presence of charge exchange reac-
tions of low cross sections between different atoms
when the energy excess is large is more probable
between atoms and molecules because the latter can
undergo dissociative ionization. It should be noted
that from this position on a further decrease of the
interelectrode distance finally extinguishes the dis-
charge. The discharge current flowing through the cell
at constant voltage along the axis is also shown in
Figure 4. From the graph, it can be seen that the
relation between discharge current and ion formation
is rather complex. However, because the anode can
only collect the slow electrons from the discharge lo-
cated within a volume of a few Debye lengths in
Figure 3. Current-voltage dependence in argon at different radius, the shape of the current curve already gives an
pressures; 0: 0.5 torr; 0: 1.0 ton; * : 1.5 torr. idea of the relative electron density at different sam-
J Am SW Mass Spectwn 1994,5,3@5-315 GLOW DISCHARGE FOR ORGANIC COMPOUNDS 309

pling positions. Also, because the ion intensity curves zene (concentration < 100 ppm) in the discharge gas
do not match the current curve, it can therefore be to evaluate the effect of plasma composition on the
assumed that the mean kinetic energy of electrons behavior of organic compounds. The results obtained
rather than their number controls ionization in the indicate that intense protonation (m/z 135) occurs in
discharge and that the electrons in the negative glow the negative glow tailing, as was observed with water.
tailing must have a mean energy a lot lower than the Higher energy processes such as charge exchange, Pen-
ionization potential of argon, as already reported by ning transfer or EI may also occur and are indicated by
Knewstubb and Tickner [41]. The variation of plasma the presence of the molecular and the fragment ions.
composition with pressure and sampling position is As expected, the signals at m/z 92 and 91 increased
summarized in Table 1. The plasma compositions are when the sampling occurred closer to the cathode. The
given as the percentage of the total ion current col- molecular ion (m/z 134) also follows this pattern,
lected at distances corresponding to the maximum showing that the molecular species can be produced
intensity for Ar+ and HaO+, respectively. Important more efficiently in this region, but with higher internal
variations are observed resulting from the different energy resulting from collisions with electrons of higher
ionization processes along the discharge axis. The pres- velocities leading to fragmentation. In the discharge,
ence of the intense water peaks (its relative concentra- low energy processes are represented by the ions at
tion in the bulk discharge gas being in the order of 10 m/z 135 formed as a result of proton transfer and the
ppm) has already been studied by Lindeger [43]. The ion at m/z 57, which ia produced by the fragmentation
ion at m/z 18 is first produced by charge exchange of m/z 135. Among the possible CI reagents present in
reaction of water with Arc (reaction 0, and m/z 19 is the discharge, ArH+ (concentration up to 25% that of
a result of either proton transfer from ArH+ or Ar+ at 1.5 torr) and H,O* (up to twice that of Ar+)
ion-molecule reactions of the water species. The se- may be noted but relatively high pressure and long
quence of reactions 2 and 3 is thought to be the most residence time in the nearly field-free negative glow
important process leading to the formation of m/z 19 also make self-protonation a probable process. Very
in these systems. Reaction 4 is also a probable process few ions are formed by collision with electrons in the
but, because the concentration of H,O is low, produc- Faraday dark space because the latter have lost most of
tion of H,O’by this pathway is less probable and can their kinetic energy by successive collisions and a large
be neglected. proportion of them (75-95%) [Z, 231 are slow electrons
(0.2-1.0 eV). The production of metastable atoms in
this region of the glow must also decrease, resulting in
AI++ H,O + Ar + H20+ (k,) Chargeexchange (1)
a lower production of molecular ionic species. This is
illustrated in Figures 4 and 5 for distances between 14
AC++ H,O + ArH++ OH’ (k,) (2) and 16 mm. The small increases in intensities observed
on the right-hand side of the graphs coincide with the
ArH++ Hz0 + H,O++ Ar Ik,) (3) beginning of a red positive column or a striation being
separated from the negative glow by a dark space.
H,O++ Hz0 --) H,O++ OH’ (k,) (4) To compare the ion distribution curves obtained by
ion sampling (Figure 4) with another plasma measure-
Ion distribution measurements have been repeated at ment technique,- Langmuir probe experiments have
1.3 torr (Figure 5) in the presence of gaseous butylben-

Table 1. Plasma composition at pressures of 0.5 and 1.5 torr

TIC’ 10.5 tow) TIC’ (1.5 torrJ
Ion 1%) (%)

Ar+b H&l+= Ar+b H30+e

Ar: (m/z 80) 3.2 1.1 3.9 0.7

ArH+ lm/z41) 9.8 6.8 12.5 6.6
Ar+ (m/z 40) 56.2 42.4 48.6 25.0
0: (m/z 32) 1.2 2.4 1.2 6.6
NO+ bn /I 30) 1.8 2.9 1.8 2.0
Art2 (m h 20) 3.2 1.6 1.7 0.7
H,O+ (m/z 19) 6.7 26.5 13.2 47.3
H,O+ (m/z 18) 15.7 14.7 15.4 10.4

“TIC = total ion current.

bSampling distance corresponds to the distance of maximum Figure 5. Axial distribution of ions of butylbenmw in the argon
intensity for Ar’.
CSampling distance corresponds to the distance of maximum
discharge at 1.3 tom and 60 pA; 0: C,,H&; 0: C,,H:i A:
intensity for H,O+. C,H,+; 0: C,H;; X: A$+; +: C,H;.
310 CARAZZATCJ AND BERTRAND J Am Sue MassSptrom 1994,5,305-315

been conducted in the glow for various discharge

conditions and distances from the cathode. Probe cur-
rent-voltage characteristic curves obtained with and
without the glass cell are shown in Figure 6 for a
sampling distance of 5 mm from the cathode at a
discharge current of 75 PA and a pressure of 1.04 torr.
The results obtained in these experiments show the
characteristic shapes similar to those obtained by Ball
1441 and Cox et al. [45] but the curves are shifted to
lower probe potentials probably because of the small
inner diameter of the discharge tube. The experimental
data were processed according to the previous refer-
ences and led to calculated electron temperatures of
2.2 eV and 0.48 eV for the curves in Figure 4 with the
discharge operated with and without the glass cell,
respectively. As in ref 44, second electron temperature
distributions were also present, superimposed on the
first ones, showing electron temperatures of 9.8 eV and
5.2 eV. They were of less importance, carrying 13%
and 4.9% of the total current to the Langmuir probe,
respectively. The differences in the values of 2.2 eV
and 0.48 eV can be explained by the small dimensions
of the discharge tube that makes the Langmuir probe
to be very close to the walls of the glass cell and to the
region where a radial field exists in the discharge (on
Figure 7. Effect of current on plasma parameters along the
the order of lo-15 V for this setup [19, 201). Also, the discharge axis at a pressure of 1 torr. (a) Electron density: m 25
higher voltages required to sustain the discharge in the @A; n : 50 pA; d : 75 FA: (b) Ion density: e 25 PA; m : 50 PA;
setup containing the glass cell (some 100 V have to be h : 75 PA; W Electron temperature: @ 25 PA; W: 50 PA; A: 75
added) may affect the drift velocities of the particles in UA.
the discharge.
A more complete mapping of the plasma is pre-
sented in Figures 7 and 8, which show the effect of
current and pressure on important parameters. The
results in these figures are presented for the discharge
without the glass cell because, in this case, probe
measurements were not always feasible because the
probe suddenly became the anode when approaching
plasma potential. Although the absolute values of elec-

Figure 8. Effect of pressure on plasma parameters along the

discharge axis at 75 PA, (a) Electron density: A : 1.5 ton; l : 1.0
Figure 6. Typical Langmuir probe characteristics obtained for ton; n : 0.5 tom; (b) Ion density: A: 1.5 tart; m 1.0 torr; n : 0.5
the argon discharge at a distance of 5 mm from the cathode, 75 torr; (c) Electron temperature: A: 1.5 torr; Oz 1.0 torr; n : 0.5 torr.
@A, and 1.04 torr. 0: with the glass cell; 0: without the glass
cell.
J Am See Mass Spectmm 1994,5,305-315 GLOW DISCHARGE FOR ORGANIC COMPOUNDS 311

tron temperature would be different from one setup to radicals (IP CH, = 14.5 eV). Radicals have also been
another, it is expected that the ion and electron densi- observed by molecular spectroscopy by Schiiler and
ties would not be greatly affected and the effects of the Lutz [46] when introducing aromatic hydrocarbons in
source current and pressure along the discharge axis the glow discharge as the result of thermal activation.
would be similar. Figure 7 shows the effect of current In the case of butylbenzene, molecular ions and frag-
on electron temperatures (c) and electron (a) and ion ments can be formed by Penning transfer with internal
fb) densities along the sampling axis at a pressure of 1 energies up to 3 eV (IP C,,H,, = 8.69 eV [47] and
torr. The overall effect of current on the density curves E(Ar*) = 11.55 and 11.72 eV [13]). However, the rela-
is similar in all cases. When more power is dissipated tive contribution of the Penning transfer to the total
in the discharge, more charged species are produced, ionization process cannot be quantitatively measured
giving rise to space-charge sheets (normal glow dis- because the population and distribution of metastable
charge) and also to local thermalization of species. In atoms could not be measured in this work.
general, results obtained with our discharge device The effect of pressure on ion and electron densities
agree well, and show similar trends to those of Fang is illustrated in Figure 8. The pressure was varied from
and Marcus [36] who have already discussed the sub- 0.5 to 1.5 torr at a constant discharge current of 75 PA.
ject in depth. In this work, Langmuir probe measure- As expected, the increase in pressure increased the rate
ments were compared to ion distributions measured of recombination and, so, decreased both the ion and
by ion sampling of Figures 4 and 5 to assess the electron current flowing to the probe. It also narrows
importance of EI in the production of positive ions in the dark space adjacent to the cathode as seen by the
the discharge. As stated before, the function of the displacement of the maximum in Figure 8a. Its effect
anode is not very different from that of a probe [40], over the electron temperature is not fully understood,
and this can be seen by comparing the shape of the but the shape of the curves on Figure 8c can be
curves of current along the axis for Figures 4 and 7a explained by the fact that all the regions of the glow
(50 PA). Also, the ion density curve on Figure 7b at are stretched when decreasing the source pressure so
constant discharge current can be compared to the the displacement of the minimum of the curve is
total ion distributions (not shown) of Figures 4 and 5. observed. The elevation of the electron temperature
They both show a decrease when sampling is made observed on the right-hand side of the graph is an
farther from the cathode, suggesting that the anodic indication of the presence of weak potential gradient in
ion sampling is most probably representative of the the Faraday dark space. Although the total ion density
axial ion distribution in the glow. The marked decrease appears to be lower for higher pressures, it will be
of ion densities along the discharge axis can be ex- seen later on that high pressure has a favorable effect
plained by ion diffusion and recombination at the because it promotes the formation of parent-molecular
walls and in the glow but also by underestimation in iOllS.
the calculation of their number. As shown in Table 1
and Figure 4, the plasma composition is not constant Mass Spectrometric Study of the Effects of Source
along the axis and, so, calculations of ion densities
Pressure and Current
should be made with the averaged mass instead of the
mass of argon. As an example of this, the average mass An extensive mapping of ionic distribution is pre-
of ions in the glow is 34.6 Da at 5 mm and 28.6 Da at 7 sented in Figures 9 to 11. Figure 9 shows the effect of
mm from the cathode. Because an ion mass of 40 Da source pressure on the intensity of the (M + H)+ and
was assumed for the calculations,~it would give an Mf ions as a function of the sampling distance. The
underestimation of ion density by 7.5% at 5 mm but as data show that an increase in source pressure, in-
much as 18.2% at 7 mm Nonetheless, even with the creases the protonation efficiency (M + H)+. On the
proper correction, the ion densities would still de- other hand, the M+ ion is not really affected by higher
crease faster than the electron density despite the fact pressures, suggesting that electron ionizaton is not the
that the electron temperature increases with the dis- only process that leads to its production. The increase
tance (Figure 7~). This was also reported by Fang and in pressure also reduces the thickness of the cathode
Marcus [36] and as yet has not been explained. The fall sheath, decreasing the optimum sampling distance
production of argon ions from the ground state by EI is for a selected species. In our experiments the use of
important only at a short distance from the cathode, higher pressures was not desirable because of the poor
while excited argon atoms are produced and ionized at pumping efficiency of the source housing, giving rise
a greater distance by collision with slower electrons to arcing problems. The combined effects of current
[ZO].Stable organic ions are also formed at a greater and voltage on ion intensities are shown in Figure 10
distance (Figure 5) where the electrons have already where the absolute intensities of some selected ions in
lost most of their kinetic energy after going through butylbenzene are reported as a function of the dis-
the cathode fall (ref 41 and Figure 7~). After studies by charge current at a constant sampling distance. The
Hess and co-workers 1211on metastable argon quench- graph shows that a small current of 40 PA (0.5
ing by methane, it is thought that reaction by Penning mA/cu?) is sufficient to ionize and fragment the com-
transfer is an important source of organic ions and pound. The use of a higher current in the discharge
312 CAKAZZATO AND BERTRAND J Am Sot Mass Spectrom 19??4,5,30!-315

narrows the cathode fall and stretches the negative

glow, thus, any change in the appropriate discharge
current requires a modification of the sampling dis-
tance to sample back the right region, as shown in
Figure 11. Even if the effect of the current on ion
intensities is not fully understood, increasing the cur-
rent increases the electron and the plasma density and
also increases the concentration of energetic atomic
species such as Ar2’ and Ar3+, N ’ and O+, thus
fragmenting molecular species which results in a loss
of signal intensity for (M + H)+, M+, and other ions at
a currents over 90 PA (1.1 mA/cm’). The anode sheath
potential is also decreased at higher currents, leading
to poorer ion extraction efficiencies [ZO].
umwg dlslmoe from&“,d. ,mm;
Figure 9, Effect of source pressure on distribution of ions from Nuture of Discharge Gus
butylbemne in argon at 60 PA; +, 0, ,I: C,,H& at 1.3, 0.65,
0.35 torr, respectively; X, 0, 0: C,,H;, at 1.3, 0.65, 0.35 ton, Although the use of discharge gases other than argon
respectively.
or helium would be quite improbable due to their cost,
the effect of the nature of the discharge gas has been
investigated by using phenylheptane rather than
butylbenzene to avoid the interference of xenon at
m/z 134. The mass spectra obtained with different
gases are presented in Figure 12a-d at a pressure of
0.75 torr. It can be observed from Figure 12 that spectra
a, b, and c show both parent-molecular ions (M/Z 1771
and fragments. The parent-molecular (M + H)+ peak
increases when using gases of lower ionization poten-
tials and metastable states as well as the ratio of the
intensities of peaks at m/z 92 and 91, indicating a
lower overall internal energy transfer from the dis-
charge to the phenylheptane. It is interesting to note
that the intensity of the ion at m/z 57, which is a
daughter ion of m/z 177, follows the same trend as
m/z 92. When the discharge is operated with Xe some
interesting features are observed, as can be seen in
Figure 10. Effect of discharge current on the butylbenzene frag- Figure 12d. Figure 12d shows that not only is the
mentation pattern at a sampling distance of 7 mm. 0: C,,)H&
parent-molecular ion at m/z 177 absent with Xe, but
c/3); 0: C,,H;i A: C,H; c/3); +: C,H;; 0: AK+ (/200); X:
CH:. Number in parentheses indicates scaling factors. also that the fragment ions at nr/z 57 and 79 have
been substantially reduced, indicating a much lower
energy deposition in the Xe plasma. This is supported
24

1 by the intensity ratio of ions at m/z 92 and 91, which

is strongly increased with Xe, as can be observed in
Figure 126. The absence of the protonated molecule
and the corresponding fragments can be explained by
the fact that enthalpies associated with reactions 1 and
2 do not favor the production of XeH+ and H,O+,
resulting in a poor H,O’ yield. Furthermore, high
pressure charge exchange experiments with xenon and
argon on some organic molecules that have been con-
ducted in this lab (results not shown) have given mass
spectra in which very few or no molecular ions could
be observed. This had already been reported by Einolf
and Munson [48] and suggest that the molecular ion
present in the spectrum obtained with xenon is formed
only by the contributions of Penning and EL Moreover,
the phenylheptane molecular ions that could be formed
Figure 11. Effect of discharge current on the optimum sampling
distance for C,,H& in argon. 0: 5 mm; A: 7 mm; 0: 9 mm; 0: by collisions with the low mean kinetic energy elec-
11 mm. trons from the negative glow or by Penning transfer
J Am Sot Mass Spectmm 1994,5.305-315 GLOW DISCHARGE FOR ORGANIC COMPOUNDS 313

MIZ Bl
Figure Effect of the nature of the discharge gas on the mam
12.
spectrumof phenylheptane. (a) neon; (b) argon; 6) krypton; cd) Figure 13. Glow discharge mass spectrum of (a) 0.5 pg caffeine,
xenon. (b) 0.5 pg adenosim in 70:30 CH,CN/H,O.

with excited xenon would result in a low internal duced in the gas or liquid phase and that the use of the
energy in the ion that would lead to rearrangements evaporated liquid as a discharge gas does not essen-
rather than fragmentation, as can be seen in the mass tially alter the general operation of the ion source.
spectrum obtained with xenon (Figure 1261. It should However, the use of a liquid-mobile phase is expected
be stated that the results presented in Figure 12 have to increase the background. Additional investigations
been taken at the distance corresponding to the maxi- are currently underway with the liquid introduction
mum for the ion at m/z 177 in each of the spectra to system and they include the operation of the discharge
compensate for the fact that the volume of the plasma with other solvents to reduce the background signal,
is altered as the discharge gas is varied. The use of especially high in the spectrum of adenosme. The re-
different discharge gases can thus be revealing in terms sults of these investigations will be published else-
of the different processes leading to ion formation and where.
fragmentation that can occur in the glow discharge
source.
Spectral and Analytical Features
To evaluate the analytical potential of the glow dis-
Liquid Introduction Mode
charee ion source for the characterization of organic
The glow discharge source described in this work has compounds
” several typical experiments have been con-
also been used to analyze compounds introduced via a ducted. The results obtained indicate that the spectra
liquid mobile phase. Preliminary studies in this mode generated from the gaseous or liquid introduction of
of operation have been conducted by using caffeine the sample do not differ significantly as they exhibit
and adenosine in a mixture of 70/30 CH,CN/H,O. the same features in terms of parent-molecular peaks
The spectra obtained by injecting samples containing and fragment ions. The analysis with gaseous intro-
0.5 pg of analytes are presented-in Figure 13a and b duction of several compounds belonging to different
Protonated molecules are observed in both cases, with chemical groups such as alkanes, ketones, and amines
a few fragment ions being produced for adenosme. In reveals that all of these classes of compounds give
this spectrum, the peak at m/z 136 represents the either (M + H)+ or (M - H)+ as parent peak when
(B + H)+ usually present with the thermospray inter- analyzed with Ar and that many fragments are present
face. However, the peak at m/z 134 has not yet been in the spectra 1491. For all the compounds that have
explained. The experiments conducted in this mode of been studied the information content of the glow dis-
operation indicate that the parameters studied in this charge spectra was higher than those obtained by more
work have similar effects whether the sample is intro conventional ionization techniques such as ET or CL
314 CARAZZATO AND BERTRAND Jh Sot Mass spectrmn1994,5,305-315

To provide examples for the comparison of the

spectral features observed in glow discharge with those
present with other ionization techniques, 2-phenyl-
1,3-dioxolane (MW = 150) was chosen because its
spectrum differs substantially from one technique to
another. The mass spectra obtained for this analyte
using an argon glow discharge and three other com-
mon ionization techniques are reported in Table 2,
which gives a mass/intensity list for selected signifi-
cant ions. As can be seen from Table 2, the glow
discharge mass spectrum exhibits peaks in the molecu-
lar region accompanied by intense fragments at lower
masses. The presence of ions at m/z 151, 150, and 149
indicates that both high and low energy ionization
processes are occurring in the source. This example
illustrates quite well the fact that many ionization
processes leading to different fragmentation paths oc- Ngure 14. Calibration curve in argon for parent-molecular ions
cur simultaneously in the discharge as a result that of: 0: 2.heptanone; 0: propylbenzene [(M + H)+ ions]; A:
nearly all the fragments found in the three other tech- isobutane [(M - H)+ ion].
niques are also found in the glow discharge spectrum.
For example, ions at m/z 151, 107, and 73 present in
width in single ion monitoring of the parent-molecular
the glow discharge spectrum are also observed in the
Cl data, while ions at m/z 105, 78, and 77 are present
ion (M + H)+. It is, however, possible to reduce the
peak width significantly with a proper injection sys-
in both the glow discharge and EI spectra and are
tem, such as those found in GC, and there is no doubt
relatively weak in the CI spectrum. This interesting
that the detection limits can be reduced to lower
feature provides extremely useful structural informa-
picogram levels. To avoid the introduction of a large
tion that must for many types of analytes be otherwise
quantity of solvent in the gas discharge, isobutane was
obtained by the use of different techniques such as
used (Figure 14) to evaluate the linearity of the source
alternate EI/CI (ACE).
signal. It has shown to be linear over at least six orders
Another important analytical feature besides struc-
of magnitude, showing slight curvature, which would
tural information content that has to be considered is
make this source a suitable detector for gas chromate
the sensitivity associated with this type of source.
Despite the low pressure used in this study (1 torr), a graphic techniques.
fair sensitivity was obtained in measurements made
using propylbenzene and 2-heptanone, as shown in
Conclusion
Figure 14. For both compounds a slight deviation from
linearity was observed that is thought to come from The results presented in this work indicate that the
self-protonation and long residence time of the com- prototype glow discharge ion source described pos-
pounds in the source. Detection limits of 100 pg (sig- sesses interesting analytical features and that the pa-
nal-to-noise > 3) were obtained for a 25-second peak rameters leading to ionization and fragmentation can
be rationalized in this type of device. In terms of
Table 2. Relative intensities in the spectrum of 2-phenyl,
organic compound analysis, the source provides excel-
13-dioxalane with several ionization techniques lent structural information that often surpasses that
obtained by EI or CI. Its sensitivity is good and its
Ion i/n /‘z) GDa Clb (CH,) Clb l&H,,) El’
operation allows the introduction of both gaseous and
151 86.8 33.3 100.0 2.3 liquid samples. The ion source has proved to be very
150 15.1 6.6 7.5 27.3 resistant to contamination because it has been working
149 51.9 21.4 14.6 100.0 for severals months without requiring cleaning or
119 8.0 0.9 0.0 5.6 showing significant losses in sensitivity. During this
107 43.4 40.0 5.7 0.0 period it was allowed to work with a resolution that
105 51.9 5.2 1.4 26.4 could be adjusted between 2000 (flat top) and 5000
91 37.7 4.3 1.9 17.9 (10% transmission) with a maximum resolution of
79 66.0 7.6 0.6 2.8 7000 at 2% transmission. It is self-cleaning because
78 53.8 4.3 9.9 25.5 reversing the discharge voltage has the effect of bom-
77 41.0 1.9 0.0 29.3 barding what used to be the anode. Little modification
73 100.0 100.0 18.9 6.9 of the cathode design would allow the use of thermal
desorption, the most commonly employed technique
' GD= glow discharge. for routine analysis. It also gives reasonable detection
b Cl = Chemical ianizatian.
‘El = electron ionization. limits and dynamic range and, thus, could easily fit a
J Am %c Mass Spectrom 1994, 5, 305-315 GLOW DISCHARGE FOR ORGANIC COMPOUNDS 315

gas chromatograph. Moreover, the spectra obtained 23. Howorka, F.; PahI, M. 2. Nuturfirsch. 1972, 27a, 1425.
show intense fragmentation and parent-molecular 24. Pahl, M.; Lindeger, W.; Howorka, F. Z. Naturfmsck.1972,
peaks that are essential for the identification of some 27a, 678.
25. Yu, T.; Tanabe, K.; Wiiefordner, J. D. Anal. Chim. A& 1987,
type of organic compounds during routine analysis.
201, 317.
26. Anderson, D. R.: Bierbaum. V.: Deuuv. C. H.: Grabowskv.,,, 1.1.
I
Acknowledgment ht. J. Ma& Spectmm. Ion k’hyi. l&3:52,65.’
27. Schneider, B.; Brueur, M.; Hatmann, H.; Budzikiewicz, H.
The authorsacknowledge
the financial support of the Natural
Org. Mass Spectrom. 1989, 24, 216.
Sciences and Engineering Research Council of Canada (NSERC)
28. Matsumoto, K.; Kojima, H.; Yasuda, K.; Tsuge, S. Org. Mass
during this project.
Srzctrom. 1985, 20. 243.
29. Hunt, D. F.; k&Ewe”, C. N.; Harvey, T. M. Anal. Chem. 1975,
References 47, 1730.
30. Scheibel, R. C.; Tanner, 0. I’.; Wood, K. V. Anal. Chem. 1981,
1. Coburn, J. W. Rev. Sci. Insfr. 1970, 41, 1219. 53, 552.
2. Cobum, J. W.; Kay, E. Appl. Phys. Letters 1971, 19,350. 31. Dzidic, L.; Carrol, D. I.; Stillwell, R N.; Horning, E. C. Anal.
3. Oechsner, H.; Gerhard, W. Phys. Lett. 1972, &?A,211. Chem. 1976, 48, 1763.
4. Colby, B. N.; Evans, Jr., C. A. Anal. Chem. 1974, 46, 1236. 32. McLuckey, S. A.; Glish, G. L.; Asano, K. G.; Grant, B. C. Anal.
5. Harrison, W. W.; Magee, C. W. Anal. Chem. 1974,46,461. Chem. 1988, 60,222O.
6. Daughtrey, Jr., E. H.; Harrison, W. W. Anal. Chem. 1975, 47, 33. Mason, R.; Milton, D. ht. J. Mass Spectrom. ion Proc. 1983,
1024. 52, 65.
7. Bruhn, C. G.; Bentz, 8. L.; Harrison, W. W. Anal. Chem. 1978, 34. Safer, I.; Zhu, J.; Lee, H. S.; Antes, W.; Lubman, D. M. Appl.
50, 373. Spectrosc. 1990, 44, 1391.
8. Eckstein, E. W.; Coburn, J. W.; Kay, E. Int. 1, Mm Spectrom. 35. Zhu, J.; Zhao, J.; Lubman, D. M. Anal. Chem. 1992, 64, 1426.
ion Phys. 1975, 17, 129. 36. Fang, D.; Marcus, R. K. Spectmchimica Acta 1990, 458, 1053.
9. Howorka, F.; Lindeger, W.; PahI, M. ht. J. Mass Spectrom.
37. Howorka, F.; Scherleitner, A.; Gieseke, V.; Kuen, I. Inf. 1.
Ion Phys. 1973, 12, 67.
Mass Spectrom. Ion Php. 1980, 32, 321.
10. Be&, B. L.; Bruhn, C. G.; Harrison, W. W. 1st. I. Mass
38. Rajotte, R J. Can. J. Spec. 1974, 19, 178.
Spectrom. Ion Phys. 1978, 25, 409.
39. Electrical Brekimun and Discharges in Eases, Kunhardt, E. E.;
11. Shzewer, D. Fresozius. J. Anal. Chem. 1990, 337, 737.
Luessen, L. H., Eds. Plenum: New York, 1983.
12. Hess, K. R; Marcus, R K. Specfroscopy, 1987, 2.
40. How&on, A. M. An rntroduction to Gas Discharge, 2d Ed.
13. Harrison, W. W.; Hess, K. R.; Marcus, R. K.; King, F. L. Annl.
Pergamon: Oxford, 1976.
Chem. 1986, 58,341A.
41. Knewstubb, P. F.; Tickner, A. W. J, Chem. Phys. 1962, 36,684.
14. Hecq, M.; Hecq, A.; Fontigny, M. Anal. Chim. Acta, 1983,155,
191. 42. Knewstubb, P. F.; Tickner, A. W. 1. Churn.Phys. 1962, 36,673.
15. Hall, D. J.; Robinson, I’. K. American Laboratory Aug. 1987. 43. Liideger, W. Phys. Rev. A 1973, 7, 328.
16. Jabukowsky, N.; Stuewer, D.; Vieth, W. Anal. Chem. 1987, 59, 44. Ball, D. J. J. A&. Phys. 1972, 43, 3047.
1825. 45. Cox, T. I.; Deshmukh, V. G. I.; Hope, D. A. 0.; Hydes, A. J.;
17. CantIe, J. E.; HaII, E. P.; Shaw, C. J.; Turner, P. J. ht. J. Mass Braithwaite, N. St. J.; Benjamin, N. M. P. J. Phys. D: AppL
Spectrom. Ion Phys. 1983, 46, 11. Phys~1987, 20, 820.
18. Loving, T. J.; Harrison, W. W. Anal. Chem. 1983, 55,1523. 46. Schiiler, H.; Lutz. E. SpectrochimimActo 1957, 10,61.
19. Llewellyn-Jones, F. The Glow Discharge; Methuen’s Mono- 47. Harrison, A. G.; Lin, M. S. Int. J. Mass Spectmm. Ion Phys.
graphs, Worsnop, B. L., Ed. London, 1966. 1983, 51, 353.
20. Chapman, 6. Glow Discharge Processes; Wiley: New Yob 48. EinoIf, N.; Munson, B. Int. J. Muss Spectrom. ion Phys. 1972,
1980. 9, 141.
21. Smith, R. L.; Serxner, 0.; Hess, K. R. Awf. Chem. 1969, 61, 49. Carazzato, D.; Bertrand, M. J. Proceedings of the 40th Annual
1103. ASMS Conference on Mass Spectrometry and Allied Topics;
22. Hess, K. R; Harrison, W. W. Anal. Chem. 1988, 60, 691. Washington, DC, 1992; p 216.