The Two Options for Sample Evaporation in Hot GC

Anal. Chem. 2002, 74, 10-16
The Two Options for Sample Evaporation in Hot GC
Injectors: Thermospray and Band Formation.
Optimization of Conditions and Injector Design
Koni Grob* and Maurus Biedermann
Official Food Control Authority of the Canton of Zurich, P.O. Box, CH-8030,Zurich, Switzerland
Although classical split and splitless injection is more than
30 years old, we only start to understand the vaporization
process in the injector. Solvent evaporation determines
much of the process and is the first obstacle to overcome.
Videos recorded on devices imitating injectors showed
that sample (solvent) evaporation is often a violent process
which is poorly controlled and might well explain many
of the puzzling quantitative results often obtained. We do
not adequately take into account that two vaporization
techniques are in use. Partial solvent evaporation inside
the syringe needle (optimized as “hot needle injection”)
produces thermospray: the sample liquid is nebulized
upon leaving the needle. The resulting fog is rapidly
slowed and moves with the gas. Solute evaporation largely
occurs from microparticles suspended in the gas phase.
Empty liners are most suitable. Fast autosamplers suppress vaporization in the needle, i.e., nebulization, and
shoot a band of liquid into the chamber that must be
stopped by a packing or obstacles suitable to hold the
liquid in place during the 0.2-5 s required for solvent
evaporation. Solute evaporation largely occurs from the
surfaces onto which the sample is deposited. Insights into
these mechanisms help optimize conditions in a more
rational manner. Methods should specify whether they
were optimized and validated for injection with thermospray or band formation. The insights should also enable
a significant improvement of the injector design, particularly for splitless injection.
Split and splitless injection are the most widely used techniques
of sample introduction in capillary GC, despite numerous inherent
problems and rather frequent disappointing quantitative performance.1 The sample, usually a liquid, is introduced into a hot
chamber in order to convert it to a vapor. The gas flow carries
the vapor to the column entrance positioned at the bottom of the
chamber. When introduced in split mode, a usually small proportion is driven into the column, whereas the rest is discharged
through the split outlet. In splitless injection, the split exit is closed
during a time that is long enough to transfer almost all sample
vapor into the column and then reopened to flush the injector.
* Corresponding author:-(e-mail) Koni@grob.org; (fax) +41 1 262 47 53.
(1) Grob, K. Split and Splitless Injection in Capillary GC; Wiley/VCH: Weinheim,
2001.
10 Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
Classical split and splitless injection still struggle with basic
problems, such as poor control of the vaporization process (e.g.,
uncertainty on when and how to pack the liner), overfilling of the
injector chamber, uncontrolled expansion of the vapor resulting
from the pressure increase during solvent evaporation, and
insufficient understanding of matrix effects.2 There seems to be
a lack of coherence, i.e., of convincing concepts and rules on
optimizing conditions. These problems may result in high standard
deviations, systematic errors, and puzzling quantitative performance and are the main reason for the reputation of capillary GC
to be excellent in separation but poor in quantitation. Recent
progress in understanding of the vaporization process should
enable a major step forward and help to get the techniques out of
the deadlock in which they have been trapped for more than a
decade.
SAMPLE EVAPORATION
Solvent Evaporation as First Obstacle. A rough estimation
of the amount of energy consumed by sample evaporation is
sufficient to conclude that vaporization cannot be the instant and
smooth process we tend to assume.3 The processes taking place
in classical split/splitless injectors were videotaped, using devices
imitating vaporizing chambers and perylene, a fluorescent compound as long as in liquid phase (solution), as a marker for
nonevaporated sample.4 This provided a wealth of new insights.5,6
Many observations were alarmingsin the end it seemed rather
surprising that quantitative results are, nevertheless, as reproducible and accurate as commonly observed. Particularly the role of
the solvent in the evaporation process was probably underestimated.
(1) Solvent evaporation is the first obstacle to overcome since
solute evaporation only starts afterward. Droplets of sample remain
at a temperature near the solvent boiling point until solvent is
fully vaporized.
(2) Solvent evaporation is time-consuming (around 0.2-5 s)
because of the large heat consumption involved. The liquid must
be held in an appropriate position during this time (see below).
(2) Grob, K. Anal. Chem. 1994, 66, 1009A-10019A.
(3) Grob, K. J. High Resolut. Chromatogr. 1992, 15, 190-194.
(4) Biedermann, M. Visualization of the Evaporation Process during Classical
Split and Splitless Injection in GC, CD-ROM; Restek Corp.: Bellefonte, PA,
2000.
(5) Grob, K.; Biedermann, M. J. Chromatogr., A 2000, 897, 237-246.
(6) Grob, K.; Biedermann, M. J. Chromatogr., A 2000, 897, 247-258.
10.1021/ac0107554 CCC: $22.00
© 2002 American Chemical Society
Published on Web 12/01/2001
(3) Partial solvent evaporation in the syringe needle nebulizes
the sample liquid at the needle exit and provides a totally different
starting point for solute evaporation than fast autosampler injection
shooting a band of liquid through the chamber.
(4) Solvent evaporation in the injector is responsible for
sometimes violent movements of the liquid and exploding droplets
and thus determines the distribution of the solute material in the
vaporizing chamber.
(5) The solvent determines whether the solute material is
deposited on to surfaces, which largely determines matrix effects.
(6) Usually over 99.9% of the vapor is from the solvent. The
large volume of solvent vapor causes the problems regarding
injector overloadingsor, seen from another angle, compels us to
use large vaporizing chambers.
In conclusion, often solvent evaporation should be improved
when solute evaporation causes problems.
The Two Principal Ways of Sample Evaporation. For
understanding of the solvent evaporation process in the injector,
it is essential to take into account the behavior of the sample liquid
in the syringe needle when the latter is inserted into the hot
injector chamber. With classical (in particular, manual) injection,
volatile solvents start evaporating inside the needle before the
plunger of the syringe is depressed. This was recognized as an
important source of discriminative effects and inaccuracy of sample
volumes.7,8 Improved heating of the injector head, improving
elution from the rear of the needle, was one response to it;9
introduction of the fast autosampler by Hewlett-Packard10 was the
other. Fast injection combined with a cool injector head suppresses
vaporization in the needle. It was not properly recognized,
however, that this also changes the basic conditions for sample
evaporation in the injector. Ever since there are the two modes
of sample introduction schematically shown in Figure 1.
Thermospray Injection. Thermospray results from partial solvent
evaporation inside the needle. Without really knowing, thermospray injection was the technique always used by manual and
classical autosampler injection of solutions in volatile solvents.
Passage through the hot syringe needle causes some solvent to
evaporate along the wall. Since the mechanical displacement is
fast, the vapor builds up pressure. The liquid is overheated and
expelled through the center of the needle. Violent evaporation
upon depressurization causes the exiting liquid to explode and
be nebulized near the needle exit. The resulting small droplets
initially move at a high velocity (exceeding 10 m/s), but friction
rapidly slows them to the speed of the gas. The latter is about 2
mm/s in splitless injection and 80 mm/s when the split flow rate
is 60 mL/min (4-mm-i.d. liner). If the distance to the column
entrance is 40 mm, there is 500 ms to 40 s of time for evaporation.
As shown in ref 3, 200-500 ms is sufficient for transferring the
heat for solvent evaporation and heating of the sample if cooling
of the chamber by several tens of degrees is accepted and a gas
of high thermal conductivity is used (such as hydrogen). In fact,
fluorescence mostly disappeared some 5-20 mm below the needle
tip, suggesting complete solvent evaporation within this distance.
(7) Grob, K.; Neukom, H. P. J. High Resolut. Chromatogr., Chromatogr. Commun.
1979, 2, 15-21.
(8) Grob, K.; Rennhard, S. J. High Resolut. Chromatogr., Chromatogr. Commun.
1980, 3, 627-634.
(9) Grob, K.; Neukom, H. P. J. Chromatogr. 1980, 198, 64-69.
(10) Snyder, W. D. Technical Paper 108, Hewlett-Packard, Palo Alto, CA, 1985.
Figure 1. Sample liquid leaving the syringe needle: thermospray
after partial solvent evaporation inside the syringe needle versus band
formation when evaporation in the needle is suppressed, e.g., by fast
autosampler injection.
If the old term “flash evaporation” suits a vaporizing technique, it
is appropriate here.
Thermospray injection results in vaporization from droplets
and microparticles suspended in the gas phase. In fact, the dense
liquid phase (fluorescence) turned into a fine fog some 0.5 mm
from the needle exit. After 80 ms (two frames of the video), the
fog was homogeneously spread over the chamber and had lost
its impulse of movement. Then the gas slowly discharged it,
seemingly more by dilution than by pluglike transfer.4 No
fluorescence (perylene) remained on the glass wall, confirming
that evaporation occurred from droplets and particles in the gas
phase. Contact with the liner wall presupposes diffusion, which
is particularly slow for the droplets or particles from which highboiling components are vaporized. Evaporation from particles rules
out adsorption or degradation on active sites of the liner, that is,
it is the most gentle process we can think of.
From this description, thermospray seems poorly suited for
samples containing large amounts of nonevaporating byproducts: if these byproducts are suspended in the gas as particles,
they enter the column together with the solutes and severely
contaminate its inlet. Practical experience taught us, however, that
most of the contaminants are, nevertheless, deposited onto the
liner wall, visible as a brownish dark ring. This is, in fact, the
reason splitless injection (traditionally performed with thermospray) was always considered as the method of choice for matrixloaded samples. The visual experiments also provided some hints
to resolve this contradiction to what was stated above for solutes:
after injection of a solution containing 5% edible oil into a 5-mmi.d. tube, most of the microdroplets were transferred to the wall
in less than 50 ms. It was speculated that electric charges formed
during thermospray generated the necessary strong forces. It
remains a key point to be explored which droplets or particles
are attracted to surfaces and which others are not.
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
11
Figure 2. Injection with band formation into a gooseneck liner (a 6-cm section being shown) heated at 200 °C in an oil bath: selected frames
of the video representing 40 ms (starting time specified at bottom); no carrier gas flow, i.e., simulating splitless injection (from ref 4).
Injection with Band Formation. Fast autosamplers inject with
a needle dwell time in the injector of some 200 ms. This prevents
solvent evaporation inside the syringe needle provided the head
of the injector is at a far lower temperature than the center.
Lacking solvent vapor acting as a propellant causes the sample
liquid to leave the needle as a band, as shown by Qian et al.11
This band moves through long heated tubes, passes bends, and
even performs slalom around baffles, being repelled from the hot
surfaces by a cushion of solvent vapor.12 Evaporation from the
band is negligible since the velocity is so high (around 10 m/s)
that a distance of, for example, 4 cm between the needle and the
column is covered in a few milliseconds, which is too short for
transferring a significant amount of heat.
When a perylene solution was injected through a cool needle
into an empty liner, the band hit the hot bottom surface of the
chamber. Most liquid was rejected, followed by violent movements. The video pictures in Figure 2 show the course of events
after injecting 5 µL of a hexane solution into a 4-mm-i.d. liner with
a restriction at the bottom (1 in frame A; gooseneck liner, Restek,
Bellefonte, PA). A piece of fused silica (2) marked the column
inlet, classically positioned some 5 mm above the bottom of the
chamber to keep it above a possible deposit of septum particles
and other debris. The band of liquid (3 in frame B) hit the bottom
and was whirled upward (4) some 5 cm high (frames B and C,
frames 0-40 and 40-80 ms after injection). Some liquid was
evaporated while flying through the chamber, but most dropped
back to the bottom, often to be rejected again (frame D). Some
150 ms after injection, most of the liquid settled at the bottom,
forming a ball nervously dancing above the hot surface. It seemed
to glide down to the funnel-shaped bottom (frame E) and finally
evaporated from the outer wall of the inserted capillary inlet (5 in
frame F, 400 ms after injection). Perylene deposited on to the
polyimide has no chance of entering the column. When the
capillary was mounted lower, with its entrance below the orifice
of the restriction, sometimes liquid was driven directly into it or
past it toward the split outlet.
(11) Qian, J.; Polymeropulos, C. E.; Ulisse, R. J. Chromatogr. 1992, 609, 269276.
(12) Grob, K.; De Martin, M. J. High Resolut. Chromatogr. 1992, 15, 399-403.
12
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
The band of liquid must be stopped above the column entrance
to avoid contact with metallic surfaces and prevent that solute
material ends up being deposited below the column entrance from
where it can no longer enter the column. The liquid must be held
in this position during the 0.2-5 s typically required for solvent
evaporation. Hence, no empty liners should be used for fast
autosampler injection.
Sucking the liquid into a packing of low thermal mass is the
first option, deactivated wool of glass, quartz, or fused silica most
widely being used. The liquid arriving first cools the nearest fibers
to the solvent boiling point and thus prepares the site for receiving
the main portion of the sample. The liquid ends up hanging in
the wool as a droplet. The region is cooled to the (pressurecorrected) solvent boiling point until solvent evaporation is
completed. It is then heated during several seconds and releases
the sample components fractionated according to volatility. The
process resembles that of programmed temperature vaporizing
(PTV) injection, with the difference that the temperature of the
injector block is constant and just that of the evaporation zone
varies.
The second option is trapping the liquid between obstacles.
Liners, such as that with the Jennings cup,13 were designed to
improve the mixing of sample vapor with carrier gassnot for
stopping a band of liquid. It is not surprising, therefore, that many
of them do not respond to the requirements of injection with band
formation: the liquid may pass through the obstacle or perform
violent, uncontrolled movements above it. The ideal liner guides
the liquid into a small chamber from where it can no longer
escape. The laminar liner (Restek) approaches this design. The
liquid passes along the liner wall to the bottom, invited to do so
by a funnel-like entrance. It is trapped there because rapid vapor
formation in the vicinity of hot surfaces hinders it entering the
narrow spaces on the ways backward or forward. Solvent evaporation then occurs from one or a few droplets dancing on a hot
surface. When solvent evaporation is complete, the solute material
is deposited onto the surface and evaporates from there.
(13) Jennings, W. G. J. Chromatogr. Sci. 1975, 13, 185-187.
Figure 3. Strategies for injection and sample evaporation in hot
injectors.
OPTIMIZATION OF CONDITIONS
Having these two evaporation processes in mind, optimization
of conditions becomes more rational. The analyst must first decide
whether to inject through a cool or a hot needle, hence to give
preference to band formation or thermospray (Figure 3), taking
into account the injection system available.
Thermospray. The possibility of using empty liners is an
advantage of thermospray injection, since this reduces the risks
of adsorption, degradation, or retention on active or retaining
(contaminated) surfaces. Thermospray enables the injection of
extremely high boiling components at gentle conditions because
the material may enter the column as aerosol rather than vapor.
Packings and obstacles do not have a noticeable positive effect
on evaporation, because nebulized material passes through them
like fog or smoke is blown through the forest.
Thermospray requires a hot needle wall. In fact, the optimum
injector temperature might often primarily be determined by the
needs of thermospray, i.e., solvent evaporation, rather than solute
evaporation. The needle temperature should be at least some 100
°C above the solvent boiling point. A well-heated rear part of the
needle is particularly important: (i) a high solvent vapor pressure
built up there is most effective in expelling the liquid, (ii) it better
overcomes the cooling by sample introduction, and (iii) it
minimizes the losses of high-boiling solute material.9 The maximum temperature of the needle is determined by the injector parts
surrounding it. The temperature regulated and given on the
readout as “injector temperature” does not correspond to that of
the upper part of the injector. Injectors for fast autosamplers have
a particularly cool top. Hence, even when the inserted needle has
reached the temperature of its environment, it may be far cooler
than indicated and may not be sufficiently hot for thermospray
(particularly when solutions are in higher boiling solvents). In
fact, injectors designed for thermospray (e.g., that from CE
Instruments/ThermoFinnigan) keep temperature nearly constant
up to the septum.14 On the other hand, even very fast injection
cannot suppress evaporation inside the needle if the top of the
injector is heated near to the injector temperature. Hence, the
injector is optimized either for nebulization (and hot needle
injection) or for use with a fast autosampler.
It is unreasonable to care for a high injector temperature in
order to obtain efficient thermospray but inject too rapidly for the
(14) Grob, K. Split and Splitless Injection in Capillary GC, 3rd ed.; Hu
¨ thig:
Heidelberg, 1993; p 444.
needle to reach this temperature. It takes 3-5 s to heat the
inserted needle, which leads to the “hot needle” technique:7 the
sample is withdrawn from the needle into the barrel of the syringe,
the needle is allowed to heat up in the injector during 3-5 s, and
then the plunger is depressed as rapidly as possible. Hot needle
injection originates from minimization of the problems resulting
from evaporation inside the needle. Today we know that the
advantages also have to do with optimized nebulization.
Experience tells us that hot needle injection with long (e.g., 3
in., 71 mm) syringe needles provides more reproducible results.
Now the explanation is at hand: it improves thermospray, because
long needles reach further into the well-heated zone of the injector,
the thermal capacity of the needle is higher (thick walls are
preferable for the same reason), and the vapor pressure built up
at the rear of the needle is higher, resulting in stronger overheating of the liquid. For splitless injection, the center of sample
evaporation should be slightly above the column entrance, for 80mm liners anyway calling for a 71-mm needle. For split injection,
a shorter needle would leave more room for mixing the vapor
with carrier gas, but nevertheless, the longer needle often provided
the better results.15
To act as an efficient propellant, the solvent should be volatile,
consume a modest amount of heat for evaporation, and, perhaps,
have a low surface tension. Using a 51-mm 26S-gauge needle
heated to 180 °C, all solvents with boiling points up to 100 °C
were nebulized (sample volumes tested, 5 µL). It should be
reminded that the upper part of some injectors reaches this
temperature only with difficulty, if at all, even when settings are
extremely high.
Band Formation. The sample liquid leaves the needle as a
band in the following three situations: (1) injection by a fast
autosampler through a cool septum cap; (2) when the needle is
short (10-20 mm) and passes through a cool injector head, also
slower (including manual) injection largely results in band
formation;5 (3) samples in a high-boiling matrix (undiluted mixture
or solution in a high-boiling solvent) always form bands.
A 5-mm plug of rather loose deactivated wool is sufficient to
reliably stop the liquid. Any additional wool merely enhances the
risk of adsorption or degradation. An alternative is column packing
material supported (but not covered) by wool. The packing
material is easily stirred 5 mm deep by the thrust of the liquid
and should, therefore, be some 10 mm deep. Frits are not suitable
because the high thermal mass of the sintered particles causes
the liquid to be rejected.6
Of the liners with obstacles tested, only the laminar and the
minilaminar liner reliably stopped the liquid.6 The gooseneck liner
prevents contact with the metal surface at the bottom of the
injector, but it does not provide satisfactory control of evaporation
as it does not hold the liquid at a defined site.
If the sample is released as a band, a rather short needle is
sufficient. The band travels to the stopper even over long
distances; i.e., the center of evaporation is determined by the
position of the packing or obstacle rather than by the needle
length. In split injection, the stopper should be located just below
the tip of the inserted needle in order to provide room for
homogenizing the vapor across the liner. For splitless injection,
(15) Grob, K.; Neukom, H. P.; Hilling, P. J. High Resolut. Chromatogr., Chromatogr.
Commun. 1981, 4, 203-208.
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
13
Figure 4. Reducing matrix effect. Nebulization of the calibration
mixture produces a fog that is efficiently carried into the column (left),
while the droplets formed from the matrix of the sample transfer the
solutes to the liner wall, from where they evaporate with more difficulty
(right).
the vaporizing chamber should be filled from the bottom to the
top and the stopper positioned only slightly above the column.
MATRIX EFFECTS
“Matrix effects” may play an important role in the analysis of
matrix-loaded (contaminated) samples and mean that the sample
influences the analytical result. In particular, absolute or relative
peak areas for a given amount of a compound depend on whether
the latter is injected in pure solvent or a sample extract. Matrix
effects easily cause systematic errors. They tend to be insidious
since they are detectable only through specific testing.
“Enhancing” and “reducing” matrix effects are distinguished.
The enhancing effect produces a larger signal for the compound
in the sample extract than in the standard solution (i.e., results
tend to be excessively high), while with the reducing effect the
peak area is reduced in the sample. The two effects have different
backgrounds.
Thermospray Injection. Reducing matrix effects are frequent
in thermospray injection. Visual observations suggest the following
mechanism. Clean solutions produce a stable fog that also carries
high-boiling solutes into the column (Figure 4, left). When the
same solution is loaded with a high concentration of high-boiling
or involatile matrix material, the resulting droplets are pulled to
the liner wall (right). Being cool as long as they contain solvent,
the droplets carry the higher boiling solutes along and “glue” them
to the wall. In splitless injection, peaks turn out too small if the
solutes are not evaporated from the layer of contaminants and
transferred into the column during the splitless period.
The reducing matrix effects were investigated using a silicone
oil to imitate a nonvolatile matrix.16 With 5% oil in the sample, the
peak of a component as volatile as n-decane was 15% too small
and the areas of n-octadecane and higher alkanes were diminished
(16) Grob K.; Bossard, M. J. Chromatogr. 1984, 294, 65-75.
14
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
by 30-40%. Experiments with edible oil as matrix material and
sterols as solutes showed even stronger deviations. These matrix
effects disappeared when PTV splitless injection was used,
presumably because cold injection avoids nebulization.17
For solutes of up to fairly high boiling points, reducing matrix
effects can be avoided by injection directly into a dense plug of
glass wool.18 In fact, in visual experiments, little fog left the plug
at the bottom,5 indicating that most of the material was transferred
to the wool and the sample evaporated from there (samples
nebulized above the wool pass through it). The vicinity of the wool
to the needle tip seems to suppress nebulization. Unfortunately,
with this transfer to surfaces also the advantages of thermospray
injection are lost; i.e., the problem is solved from the wrong end:
the calibration with a clean mixture now performs as poorly as
the analysis of the sample, rather than that the performance for
the sample is enhanced to that of the calibration.
Band Formation. The term “matrix-induced chromatographic
response enhancement” was coined by Erney et al.19 and refers
to a phenomenon well described for the analysis of pesticide
residues in foods: peak areas per amount injected are larger in
extracts of samples than in the matrix-free calibration mixture.
Since peaks cannot represent more than 100% of the material
injected, the smaller areas obtained for the calibration must be
the problem (the opposite of the reducing matrix effects). Mu¨ller
and Stan20 showed how peak areas developed during a series of
100 injections, a first 15 of a clean mixture of standards followed
by 85 spiked extracts of spinach. The matrix material introduced
by the first injections of the spinach extract caused the areas of
critical components to increase. Then a plateau was reached, and
later areas shrunk again, probably because of an excessive
accumulation of contaminants.
Enhancing matrix effects were primarily observed with packed
liners, often (but not exclusively) in conjunction with band
formation, i.e., deposition of the sample liquid onto surfaces. They
are the result of a temporary deactivation of these surfaces by
sample material: contaminants (which may be as volatile as water)
deactivate the surface for some time and improve the conditions
for the evaporation of the solutes, but largely disappear (evaporate
or degrade) before the next sample is injected. Their source is
insufficient deactivation: sample components obviously deactivate
more efficiently than the treatment applied by the manufacturers.
“Priming” of packed columns was based on the same principle:
a sample was injected several times in short intervals before the
analyses were started. Enhancing matrix effects are also observed
in PTV and (to a smaller extent) on-column injection or for
columns. They presuppose intense contact with surfaces, as
obtained after injection with band formation and deposition of the
liquid onto cooled surfaces.
EVALUATION OF THE TWO APPROACHES
The reader probably long started comparing the two ways of
sample evaporation in order to find the preferable technique for
his application. He should keep in mind that present instrumenta(17) Grob, K.; La¨ubli, T.; Brechbu
¨ hler, B. J. High Resolut. Chromatogr., Chromatogr. Commun. 1988, 11, 462-470.
(18) Grob K.; Neukom, H. P. Chromatographia 1984, 18, 517-519.
(19) Erney, D. R.; Gillespie, A. M.; Gilvydis, D. M.; Poole, C. F. J. Chromatogr.,
A 1993, 638, 57-63.
(20) Mu
¨ ller, H.-M.; Stan, H.-J. J. High Resolut. Chromatogr. 1990, 13, 697-701.
tion is optimized either for thermospray or band formation (or
neither of both). Hence, the choice is limited once he has bought
the instrument. He should consider the following aspects:
(1) Evaporation in the needle (thermospray) causes more
sample to be injected than read on the barrel of the syringe as
well as selective loss (discrimination) of high boilers. This was
the reason for introducing fast autosampler injection, although
with well-heated injectors discrimination can also be kept low.
(2) Nebulized, fairly clean samples evaporate from droplets
or particles suspended in the gas, thus from their own matrix.
This is why hot needle injection tends to perform better than
injection depositing the sample onto deactivated wool, even though
it has classically been performed with raw glass liners (no specific
data available). Thermospray is, therefore, preferable for the
analysis of adsorptive, labile, and high-boiling solutes. This no
longer holds true when the samples contain high concentrations
of high-boiling or involatile components (large particles are
attracted to the liner wall).
(3) Both techniques may suffer from matrix effects. Enhancing
effects are observed after deposition of the solute material onto
surfaces and are primarily a problem for adsorptive and labile
components. Thermospray injection may suffer from matrix
material gluing solutes to the liner wall, which affects all higher
boiling solutes. Solute evaporation is not more hindered than after
deposition onto a surface by injection with band formation, but in
comparison with the better performance of the clean solution, the
losses are more obvious.
(4) In principle, evaporation from a surface after injection with
band formation produces clean vapor, leaving behind in the
injector all the nonevaporating material, while thermospray also
nebulizes involatile material. However, experience taught us that
differences are not as large: column contamination after injection
resulting in band formation is not always negligible, whereas that
by thermospray injection is not a severe problem.
(5) The accuracy obtained with split injection depends on the
stability of the split ratio and on whether all solutes are split by
the same ratio.21 Discrimination from a fluctuating split ratio
results when different components arrive at the split point at
different times (different split ratios). Evaporation from an initially
cooled surface, following injection with band formation, is probably
the most important fractionation mechanism (although under
some conditions the split ratio may have restabilized before the
solutes evaporate). Thermospray injection produces a vapor cloud
with a homogeneous sample composition; i.e., a fluctuating split
ratio does not result in discrimination. Another question is whether
vapor and aerosol particles are split by the same ratio, which could
be a different source of discrimination effects.
(6) If fast autosamplers are used, sample evaporation in the
injector fundamentally differs from that following manual injection.
Hence, a method validated for fast autosampler injection cannot
be carried out with manual injection or a slower autosampler. If
thermospray is applied, manual and automatic hot needle injection
perform equally.
CONSEQUENCES FOR THE INJECTOR DESIGN
With the concepts of sample evaporation in mind, we should
be able to improve the design of the split/splitless injector. The
(21) Grob, K. Split and Splitless Injection in Capillary GC, 3rd ed.; Hu
¨ thig:
Heidelberg, 1993; pp 126-159.
basic design of the injectors presently in use is from the 1970s,22
i.e., from a time long before the fast autosampler was introduced.
The thermospray mechanism was not understood at that time,
but something similar was assumed since the sample was thought
to evaporate just below the needle exit. Hence, the geometry of
our injectors largely reflects the philosophy of thermosprayseven
when used with fast autosamplers. The determining factors were
as follows:
(1) If the injector is designed according to the needs of splitless
injection, it is also suitable for split injection, but not vice versa.
(2) The volume of the vaporizing chamber must be sufficiently
large for the intermediate storage of the sample vapors (splitless
injection and split injection with low split flow rates). Two
microliters of solutions in hexane and water (two extremes)
produces some 300 and 3000 µL of vapor, respectively.
(3) The internal diameter of the chamber cannot exceed 4-5
mm since the gas velocity during splitless sample transfer
becomes too low otherwise.23 Furthermore, a larger bore increases
the capacity less than proportionally to the volume because of
increased mixing (dilution) with carrier gas (the vapors form
something like smoke trails through the gas in the injector instead
of displacing the gas similar to a piston).
(4) To fill the chamber from the bottom (column entrance) to
the top, the syringe needle must be sufficiently long to position
the center of sample evaporation slightly above the column
entrance. Since syringe needles longer than 3 in. (71 mm) are
awkward to use, chambers of 80-mm length became standard.
(5) The accessible volumes in the split outlet line must be small
in order to prevent that pressure increase during solvent evaporation pushes sample vapor into this outlet.
The following elements were introduced after 1978:
(1) Improved heating of the injector head decreased discrimination against high-boiling solutes for hot needle injection.9
(2) The fast autosampler avoided sample evaporation inside
the syringe needle.10
(3) The programmable inlet pressure (Wylie et al.24) enabled
the increase of pressure during splitless injection (“pressure
pulse”), which compresses the vapor cloud and thus enhances
the capacity of the vaporizing chamber. At the same time, it
accelerates the sample transfer into the column, allowing the use
of wider bore chambers.
It has been known for a long time that the internal volume of
a 80 × 4 mm i.d. chamber (1 mL) is often too small to reliably
hold the sample vapor.25,26 At modest inlet pressures it does not
even house the vapor of 1 µL of a solution in methanol.
Overloading is frequent and serious effects on quantitative analysis
have been reported (e.g., refs 27 and 28). Elongation of the liner
is the only efficient means of enhancing the injector capacity. For
instance, a 16 × 4 mm i.d. liner would offer 2 mL of room. Reduced
(22) Grob, K.; Grob, K. J. High Resolut. Chromatogr., Chromatogr. Commun.
1978, 1, 57-63.
(23) Grob, K. Split and Splitless Injection in Capillary GC, 3rd ed.; Hu
¨ thig:
Heidelberg, 1993; pp 273-287.
(24) Wylie, Ph. L.; Phillips, R. J.; Klein, K. J.; Thompson, M. Q.; Hermann, B. W.
J. High Resolut. Chromatogr. 1991, 14, 649-654.
(25) Grob, K.; Biedermann, M. J. High Resolut. Chromatogr. 1989, 12, 89-96.
(26) Hinshaw, J. V. J. High Resolut. Chromatogr. 1993, 16, 247-153.
(27) Lee, H. B.; Szawiola, R.; Chau, R. S. Y. J. Assoc. Off. Anal. Chem. 1987, 70,
929-935.
(28) Grob, K. J. Assoc. Off. Anal. Chem. 1988, 71, 76A-77A.
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
15
far backward toward the top, while the higher boiling solutes
evaporate last and form a small vapor cloud staying near the
column entrance. Their transfer into the column is correspondingly fast. This corresponds to the process used for large-volume
splitless injection by the overflow technique.29 Since the upper
part of the chamber primarily serves for intermediate storage of
solvent and volatile compounds, its temperature may be rather
low, just sufficient to prevent the band touching the liner surface.
For split injection, the band stopper is placed higher in order to
leave enough room for spreading the vapor across the liner.
Figure 5. Ideas for an improved design of an injector for split/
splitless injection with thermospray or band formation.
mixing with carrier gas increases capacity more than proportionally to the volume and should enable injection of about 2 µL of
methanol (1.5 mL of vapor at 30 kPa and 250 °C), 1 µL of water
(1500 µL), 3.5 µL of dichloromethane (1600 µL), or 6 µL of hexane
(1550 µL). With an inlet pressure of 160 kPa, the capacity is
approximately doubled. Such elongation of the chamber presupposes, however, a new concept on how to bring the center of
evaporation so far down the chamber to be situated just slightly
above the column inlet. The options depend on whether injection
is performed with thermospray or band formation.
Optimized Thermospray Injector. An optimized thermospray
injector should comprise a nebulizer of small internal volume
releasing the sample near the column entrance. It could consist
of the syringe needle or a separate device. The thermal mass and
thermal conductivity of the material should be such that there is
no strong cooling by solvent evaporation. The nubulizer could be
positioned laterally near the bottom of the chamber (Figure 5),
which probably presupposes a liner made of deactivated steel, or
enter at the bottom parallel to the column.
Injection with Band Formation. Solutions in volatile solvents
injected with band formation can be shot through long hot tubes
without significant evaporation, since vapor cushions prevent the
transfer of the liquid to the liner wall. A short needle, merely
entering the liner by a few millimeters (totally 15-20 mm long),
is sufficient. If the top of the injector is cool, there might not even
be a need for a particularly fast injection. The needle tip must be
such as to rule out mechanical spray.
For splitless injection, the band is stopped at the bottom of
the chamber by means of a packed bed or a trap. As the solvent
and the highly volatile components evaporate first, they expand
16
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
OUTLOOK
It seems odd that a key technique of capillary GC, such as hot
split and splitless injection, still needs basic development work
decades after its introduction and with maybe 200 000 analysts
using it for daily routine. Improvement is needed, because the
processes are not under a control satisfying the requirements for
a professional tool, and all too often poor quantitative results are
obtained for reasons that cannot be put aside blaming the analyst.
Often methods do not tell what conditions to select (fast
autosampler/manual injection? needle length? side port hole in
the needle? hot needle injection? liner dimension? packing?
position of the packing? etc.), leaving this “up to the responsibility
of the analyst”, as it may be put to avoid saying that nobody could
tell him. Many methods describe in great detail how the round
flask should be rinsed (rather an insult to experienced analysts),
but then provide hardly any advice on the often most important
source of error.
The distinction between thermospray and band formation
should help understand phenomena and optimize conditions in a
rational manner. Methods should specify whether they were
validated for thermospray or band formation. Basically they should
be validated for both, since the two techniques run on different
types of instruments and users commonly just have access to one
of them. In the long run, the question needs to be answered of
whether vaporizing injection could be standardized on a single
principle of sample evaporation. This would simplify the specification of methods but also deprive us of the possibility to choose
the other technique because of better performance for a particular
sample. In fact, it is not obvious which technique to prefer.
Insights resulting from the videos on the evaporation process
make us believe that substantially improved injectors could be
designed, improving the quality of the results, enabling us to inject
somewhat larger volumes, and helping us to better master the
critical conditions.
Received for review July 6, 2001. Accepted October 12,
2001.
AC0107554
(29) Grob, K.; Brem, S.; Fro
¨hlich, D. J. High Resolut. Chromatogr. 1992, 15,
659-664.