trends in analytical chemistry, vol. 21, nos.

9+10, 2002

547

History of gas chromatography+

Keith D. Bartle*, Peter Myers School of Chemistry, University of Leeds, Leeds LS2 9JT, UK

Modern gas chromatography (GC) was invented by Martin and James in 1952 [1], and has become one of the most important and widely applied analytical techniques in modern chemistry. Major milestones in the development of GC, especially in column technology, detection and sample introduction are described in this historical review. Many trends in current progress can be seen to originate in the rst two decades of the history of GC, but the invention of fused-silica capillary columns greatly increased the application of high-resolution GC across the eld of organic analysis; the development of low-cost, bench-top mass spectrometers led to further advances. Progress continues to be rapid in comprehensive 2D GC, fast analysis, detection by atomic emission and time-of-ight mass spectrometry, and in applications to process analysis. # 2002 Published by Elsevier Science B.V. All rights reserved.

1. What is gas chromatography?

Our world is one of complex mixtures. Petroleum may well contain over 100,000 components. It is thought that there are of the order of 100,000 different proteins in the human body. And natural products, such as essential oils, are also often highly complex. Separation methods are necessary to analyse these, while even the simpler, but still often difcult, mixtures encountered in, for example, the pharmaceutical industry most often require chromatography or a related technique. Separations are achieved by GC by a series of partitions between a moving gas phase and a
Based, in part, on a lecture in November 2001 at the designation by the Royal Society of Chemistry of a National Historical Landmark on the Leeds, UK, site of the invention of partition chromatography by A.J.P. Martin and R.L.M. Synge. *Corresponding author. Tel.: +44 (0)113 2336490; Fax: +44 (0)113 2336565. E-mail: k.d.bartle@chem.leeds.ac.uk
+

stationary liquid phase held in a small diameter tube (the column) after a mixture is injected as a narrow band. A detector then monitors the composition of the gas stream as it emerges from the column carrying separated components, and the resulting signals provide the input for data acquisition. GC can be applied to the analysis of mixtures, which contain compounds with boiling points from near zero to over 700 K, or which can be heated sufciently without decomposition to give a vapour pressure of a few mmHg. Derivatisation to increase volatility extends this range. The sample size can be as small as pg, but, in preparative, as opposed to analytical, applications, tens of grammes can be handled. GC is now a standard analytical method that underpins research, development, and quality control in many industries, especially petrochemical manufacture, and in environmental, food contaminant, and drug residue and forensic analysis.

2. Origin of GC
The invention of GC is generally attributed to A.T. James and A.J.P. Martin in their 1952 paper [1] which followed a presentation on 20 October 1950 at a meeting of the Biochemical Society, with wide interest generated by a lecture at a Society of Chemical Industry meeting in Oxford in September 1952. They reported the separation of volatile fatty acids by partition chromatography with nitrogen gas as the mobile phase and a stationary phase of silicone oil/ stearic acid supported on diatomaceous earth. In fact, the origin of GC lies in a sentence, overlooked by other researchers, in the 1941 publication [2] in which Martin, with R.L.M. Synge, rst described liquid-phase partition
# 2002 Published by Elsevier Science B.V. All rights reserved.

0165-9936/02/$ - see front matter PII: S0165-9936(02)00806-3

548

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

chromatography: Very rened separations of volatile substances should be possible in a column in which permanent gas is made to ow over gel impregnated with a non-volatile solvent. . .. The novel aspect of Martins work was the employment of partition as the separation principle, although gas adsorption chromatography was also developed in the 1940s and 1950s by a number of researchers (notably Hesse, Cramer, and Phillips [3]). James and Martin followed their rst paper with reports in the same year of the analysis of bases by GC [4,5]. A photograph of Martin demonstrating GC on a version of his rst GC is shown in Fig. 1. One of the greatest analytical challenges of the 1950s was the composition of petroleum, then replacing coal as the predominant source of liquid fuels and chemical feedstocks. The petroleum industry rapidly adopted GC for compositional analysis, with outstanding developments and applications, with researchers at Shell (e.g. Keulemans and Adlard) and British Petroleum (Desty) at the forefront [3]. GC also enjoyed explosively rapid growth in the decade after its discovery in a host of other application areas [6], especially biochemistry [7], including amino-acid analysis [8], and natural product (e.g. steroids [9]), and food and avour [10] studies. Another driving force in putting GC on a rm physico-chemical basis (see, for

example, the monograph by Purnell published in 1962 [11]) was its application in the investigation of reaction kinetics [12]. As many as 200 papers on the analysis of fatty acids and fatty acid esters had appeared by 1960 [6]. Rapid expansion continued, with an exponential growth of capillary column GC (see below) in the 1980s.

3. The gas chromatograph

The early pioneers of GC would have no difculty in recognising the layout of a modern instrument, shown schematically in Fig. 2. The main item is the column, originally a tube packed with a solid support coated with the stationary liquid, but now a ne tube with the liquid coated on the inner surface. The carrier gas, at rst nitrogen, but now helium or hydrogen, passes from a cylinder through a pressureor ow-rate-controlling device to the sample injector at the column inlet. When the separated mixture components of the mixture emerge (are eluted) from the column, they are detected by the measurement of some chemical or physical property. An important factor inuencing column performance is its temperature; for most mixtures, it is necessary to work at higher temperatures, and good temperature control is always important, originally achieved by enclosing the column in a vapour jacket or thermostatted block, but now in a hot-air oven. Resistive heating of metal-clad columns to allow extremely rapid heating and cooling is also being introduced.

3.1. Columns for GC

The column is at the centre of the analytical gas chromatograph; the quality of the separation achieved by the whole system can be that of the column only. Early GC was carried out on packed columns, typically 15 m long and 15 mm i.d., and lled with particles each of which was coated with a liquid or elastomeric stationary phase. Micro-packed columns are similar but have i.d. < 1 mm. The resolution of

Fig. 1. Video photograph of Martin and Jamess GC.

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

549

Fig. 2. Schematic diagram of a basic gas chromatograph. (Reproduced with permission of John Wiley and Sons Limited).

packed columns is limited by their length, itself restricted by the pressure drop consequent on the resistance to gas ow. This restriction was removed by the invention of the capillary column, suggested by Martin [13] at a meeting in 1956, but independently realised in 1957 by Golay [14], who laid out the theory of operation and demonstrated its use in 1958 [15]. In a capillary column, the stationary phase is coated on the inner wall, either as a thin lm (wall-coated open tubular) or impregnated into a porous layer on the inner (porous layer or support coated open tubular), and the differing paths taken by solute molecules as they pass through the (inevitably) non-uniform packing (i.e. a bundle of capillaries) is replaced by a single channel. As well as the advantage over packed columns of greatly increased separation efciency, capillary columns work at a lower temperature, and give much better separation (see Fig. 3) in equal times, or the same separation in shorter time (Table 1). Results are generally generated up to 10 times faster than in a

packed column. Of course, the much smaller amounts of stationary phase in a capillary column mean that their capacity is limited, and special sample-introduction methods (see Section 3.5) and sensitive detectors (see Section 3.3) are required. A number of column materials were employed in the early history of capillary GC copper, nickel, stainless steel, and even nylon tubing but the relative inertness and transparency of glass columns were quickly recognised as advantageous. In 1961, Desty [16] proposed a device that drew out and coiled glass capillary tubing from laboratory glass tube. Glass capillary columns were widely used until 1980, with a voluminous literature devoted to their preparation (see e.g. [1719]) with landmark contributions from the groups of Novotny, Grob Sr., and Schomburg. Progress with glass columns was impeded, however, by their fragility, and activity towards highly polar analytes. These problems were to a large extent solved by the invention of fused-silica columns

Table 1 Comparison of wall-coated capillary, support-coated open tubular, and packed columns Wall-coated capillary Length (m) Internal diameter (mm) Liquid lm thickness (mm) Capacity per peak (ng) Resolution 10100 0.10.8 0.11 < 100 High Support-coated open tubular 1050 0.50.8 0.82 50300 Moderate Packed 15 24 10 10,000 Low

550

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

Fig. 3. Chromatograms of Calmus oil on (A) a 50 m capillary column and (B) a 4 m packed column with the same stationary phase.

in 1979 by Dandeneau and Zerenner [20]. Such columns, manufactured by a process originally based on bre-optic technology, are highly exible, durable and chemically inert. Externally coated with a protective layer of polyimide and with an immobilized lm of one of a wide variety of stationary phases, fused-silica columns provide the means of separation of almost all mixtures to be analysed; the few exceptions, such as permanent gases and low molecular-weight compounds, are separated on porous layer open-tubular columns with, for example, alumina-based stationary phases [21]. Extensive research was carried out in the 1980s, notably by the group of Lee [2224], to determine the physical chemistry principles underlying coating of fused-silica columns with a uniform, homogeneous and stable lm of stationary phase. While the fused-silica surface

is substantially free of the metallic oxides that act as Lewis-acid adsorption sites for polar and aromatic molecules on glass columns, it still has a number of reactive hydroxyl groups attached to the surface silica atoms; these active sites must be deactivated by high-temperature silylation. Clean fused-silica is a high-energy surface and is hence wettable by most organic liquids. But, after the deactivation step, only stationary phases of appropriate surface tension will spread, since the critical surface tension of the surface is greatly reduced, although it can be adjusted by the presence of the necessary functional group in the deactivating reagent. The wettability of a column surface is not inuenced by high temperature, since changes in the surface tension of the phase are matched by changes in the critical surface tension of the surface. However, the stationary phase lm is still

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

551

inherently unstable [25] so-called Rayleigh instability. The rate of rearrangement of the lm depends inversely on its viscosity, and disruption and stationary phase bleed is overcome by in-situ cross-linking of (usually) unsaturated groups in the stationary phase molecule by means of free-radical initiators, such as peroxides, or best, azo compounds to yield elastomers. Column manufacturers now make available a wide range of capillary columns, making use of the principles discussed above. The choice of the appropriate column for a given required separation depends on the chemical nature of the analyte, the sample matrix, and the solvent (for a summary, see [26]), and especially on the nature of the molecular interactions between analyte and stationary phase. Dispersive (nonpolar) interactions give rise to separations based on analyte volatility, of which a simple measure is boiling temperature, while dipole-dipole interactions, and hence increased retention and selectivity, occur between solutes and stationary phases with polar groups (e.g. cyanopropyl and triuoropropyl). A polar stationary phase or analyte may also induce a dipole moment in an electron-rich solute or phase and hence result in selective interactions. Analytes capable of hydrogen bonding may be separated by a stationary phase containing hydroxyl groups, such as polyethylene glycol. Shape-selective separations are also possible with chiral or liquid-crystal stationary phases. The rst stationary phases for GC were a varied collection of hydrocarbon and silicone oils and greases, esters and polymers, but now by far the largest number of GC stationary phases are based on a cross-linked polysiloxane backbone with appropriate pendant groups (e.g. CH3, Ph, CH2CH2CH2CN, CF3) The pendant group may also be tailored for a steric separation. A number of so-called wax polyethylene glycol phases are also popular. A silylarene backbone increases the temperature stability of the phase. Other choices in capillary column selection include column internal diameter (dc), lm thickness (df) and length (L). It follows from

Golays 1958 theoretical treatment [15] that the number of theoretical plates, N, is proportional to 1/d2 c ; the narrower the column, the greater the resolution, but the less the capacity. Values of dc of 100500 mm are standard, with df 0.25 0.5 mm. Column lengths vary from 5 m to > 100 m, but the advantages of increased efciency in long columns are offset by increased analysis times and the fact that resolution depends on N1/2 and hence L1/2; quadrupling L only doubles resolution. There is an increasing trend to shorter, narrower columns. Elution of analytes is brought about by holding the temperature of the column at a high enough value to ensure solubility in the gas phase, a process well characterised by 1962 [11]. However, for mixtures with a wide range of volatilities, the spreading of late-eluting compounds can be overcome by programmed temperature GC (PTGC) i.e. increasing the column temperature during the run, a procedure invented by Grifths et al. as early as 1952 [27]; progress and increase in use of PTGC were so rapid that it was already the subject of a book published in 1966 [28] and has been the predominant technique for many years.

3.2. Pneumatic systems in GC

With the advent of capillary columns, greater precision was required in the pneumatic control systems. Control of the gases required to run a GC has been through a combination of on-off valves, forward and back-pressure regulators, needle valves and mass-ow-control regulators. These have evolved together with the instrumentation, such that today we see total feedback controls to maintain constant ow rates of the carrier gases by monitoring the gas pressures and ow rates that, in turn, control electronic regulators. Much of the ow regime in capillary instrumentation is closely associated with the requirements of sample injection and this is now possible through the automatic computer control of the pneumatics. As the instruments have evolved, we have seen a trend towards the use of keypads, either on the instrument or on a separate keypad, to set the conditions of the

552

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

instrument. With the advent of the modern PC, control tended to move to control and dataacquisition programs on the PC. Now, it is possible with the automatic ow-control modules to function under mass control or pressure control of the carrier gas. This allows a choice of constant pressure, constant ow or even pressure programming.

3.3. Detection in GC
The rst gas chromatograms were generated by an automated titration system. But, in 1954, Ray used the temperature (and hence electrical resistance) change of a lament of a thermal conductivity-measuring device a katharometer as a means of detection [29]. The katharometer remained popular for packed-column work because of its response to most analytes, but the requirement of trace analysis and the development of the capillary column quickly resulted in a new emphasis. Rather than use bulk properties (based e.g. on gas density, owimpedance, and gravimetry - a sensitive version of the latter was proposed by Martin [30] in 1962 as potentially the ideal detector for GC), more sensitive ionization-based detectors were investigated. While a number of ame-based detectors based on ame temperature and emissivity had already been devised, in 1958, Harley et al. [31] and McWilliam and Dewar [32] realized, almost simultaneously, that the thermal energy of a hydrogen/air ame should bring about emission of electrons from organic molecules in the ame to produce ions, so that the measured ion current would depend on the amount of eluent per unit time entering the ame. The ame ionisation detector (FID) was more sensitive than the katharometer by a factor of 103104, with sub-ng detection limits demonstrated [32]. The FID rapidly became, and has remained, extremely popular, overtaking a number of other ionisation detectors proposed at the same time [33]. Its low cost, unsurpassed linearity, linear dynamic range and sensitivity for carboncontaining compounds (Table 2) still make it the universal detector of choice, although the mass

spectrometer (see below) is having an increasing impact, and the helium ionisation detector is a modern ionisation detector with a better response to a number of compounds, for which the FID has limitations. A number of ionization detectors other than the FID originated at about the same time, for selective detection. The electron-capture detector (ECD) of Lovelock and Lipsky [34], based on the ability of a molecule to capture free electrons from a b radiation source, has since found wide acceptance as an extremely sensitive detector (pg to fg range) for halogen-containing compounds, with wide applications to trace pesticide residues. The analysis of atmospheric chlorouorocarbons, destroyers of the stratospheric ozone layer, represents a particular triumph for the ECD. Among a number of nonradioactive ionization detectors, the photo-ionization detector (PID), again due to Lovelock [35], and the alkali ame or thermionic detector (TID) have also found long-standing application. Depending on the discharge gas, which determines photon energy, and the materials of the optical window, the PID allows selective detection for e.g. aromatics over aliphatic compounds. In the TID, selective detection, particularly for nitrogen and phosphorus compounds (thus making the detector particularly useful in trace herbicide and pesticide analysis), is achieved by an alkali-metal salt placed near a hydrogen/air ame. The original TID, invented in 1964 by Kamen and Giuffrida [36], employed a volatile alkali salt attached to the FID ame jet, but new versions [37] used an externally heated alkali source, such as a rubidium silicate bead. Non-ionization ame detectors proposed in early GC included the ame emissivity detector, suggested by Grant in 1956 [38], which had high selectivity for aromatics. But the most commonly used ame emission detector is the ame photometric detector (FPD), in which light-emitting combustion products are generated, especially of sulphur compounds, in a cool hydrogen-rich ame. The dual-ame version of the FPD has advantages over the original 1966 single-ame FPD of Brody and Chaney [39].

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

553

The ultimate element-selective detector for GC (Table 2) is the atomic emission detector (AED). As early as 1965, McCormack used [41] a microwave-induced plasma (MIP) in GC detection, and a variety of other emission sources, both ame and plasma, have been investigated [42]. An important advance came in 1990 when Quimby and Sullivan [43] developed an AED with a water-cooled MIP with a number of reagent gases to enhance performance for elements hitherto exhibiting poor sensitivity, such as N and O. The spectrometer incorporates a photodiode array with a wavelength range of 160800 nm so that elements across the periodic table can be detected - up to four simultaneously. The high sensitivity and selectivity, linearity and compound-independent response have led to numerous applications, to not only the more usual S, N, P, halogens, etc., in a wide variety of analytes, but also Sn (in fungicides), V and Ni (petroleum porphyrins), and Hg and Pb compounds. The on-line use of AED, combined with mass-spectrometric detection for tracing targets as well as unknowns, was another powerful addition [44].

3.4. Mass spectrometric detection

It was realised early that the structural information and selectivity available from mass spectrometry (MS) made the combination of MS with GC the most effective technique for the analysis of complex mixtures. While off-line analyses of GC fractions were made initially, Gohlke described in 1959 [45] the direct introduction of GC efuent into a mass spectro-

meter, and detailed GC-MS analysis of natural products was being reported by 1963. The high cost of MS instrumentation and the incompatibility of high GC ow rates in packed columns with the vacuum requirements of the mass spectrometer both hindered progress of GCMS. However, the much smaller ow rates in capillary columns, along with the enrichment possible with molecular separators made GCMS more attractive. GC-MS was advanced enough in the 1970s for an instrument to form part of the Viking Mars Lander [46]. The availability of less expensive bench-top MS led to the routine use of the GC-MS technique in the 1980s and it is now pre-eminent. Time-of-ight MS was endorsed by Martin in 1962 [30] because of its inherent sensitivity, but only recently has its performance been developed to allow routine coupling to GC. In GC-MS of all types, as in other GC ionisation detectors, ions are produced by electron or chemical ionisation but are now sorted according to molecular weight (or mass-tocharge, m/z, ratio) by one of a range of analysers: magnetic sector; (less expensive) quadrupole or ion-trap in bench-top instruments; or time-of-ight. The resulting mass spectrum is related to the molecular weight and structure of the analyte, and allows identication through comparison with a library, or through a-priori interpretation. The MS detector can be operated in either scanning mode or, in a technique introduced in 1968 [47], with selected-ion monitoring. Thus, a range of m/z values may be scanned (say 50500) in a bench-top instrument in 1 s, typically. A computer is then used to plot

Table 2 Properties of commonly used detectors on capillary GC [40] FID Operation mode Selectivity Detection limit Linearity Destructive Mass ow rate Organic carbon 0.5 pg/s 107 Yes ECD Concentration Electron afnity 1pg/mL 104 No PID Concentration Hydrocarbons (aromatics) 0.5 pg/mL 107 No TID Mass ow rate N/P 25 fg/s (P) 50 pg/s (S) 104(P) 105(N) Yes FPD Mass ow rate S/P 50 fg/s (N) 0.5 pg/s (P) 103(S) 105(P) Yes

554

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

either the total-ion current (analogous to the response of a universal TC detector) or individual ion currents (mass chromatograms), which may be specic to selected compound types; much greater sensitivity is achieved by monitoring only one ion or a few ions. The great advantages of time-of-ight MS lie in the possibilities for accurate mass measurement (in contrast to the unit m/z resolution of bench-top MS), which allow the molecular formulae of ions to be determined, and rapid rates of accumulation of spectra (up to 500 Hz), which allow GC peaks with widths as narrow as 12 ms to be identied [48].

3.5. Sample introduction in GC

Injection of a sample into the gas stream at the column head was already carried out in early work by means of a syringe and a hypodermic needle. At rst, a re-sealable rubber subaseal cap was employed, but this was replaced as early as 1964 by a heat-resistant elastomeric septum compressed in a metal tting, the procedure which has persisted until now. Because injection of a representative part of the sample as a narrow band in a quantity consistent with the capacity of the capillary column is often a limiting factor, a variety of injection methods have been developed. The basic splitter injector, which allows a small fraction of a rapidly volatilised sample to enter the column while the major portion is vented to waste, was introduced by Desty in 1959 [49] and much development of this device followed [50]. The injection of 20100 mL liquid volumes is now routinely possible, rather than the formerly standard 12 mL, by using sample-introduction systems, such as on-column, loop-type, and programmed temperature vaporization (PTV) [51]. A 100 mL injection will easily allow measurement of solute concentrations of 100 ppt. The conventional splitting injection may be used to allow large volumes of a dilute sample solution to be introduced into the column at low temperature by simply closing the split valve. Grob and Grob [52] showed that the more volatile excess sample passes through, and

trace solutes are concentrated in a narrow band in the column a technique routinely used in environmental analysis, analysis of pesticides in foods, and drug screening. The possibility of sample loss during vaporisation of a liquid sample and transfer of vapour to the column can be eliminated by direct injection into the inlet of the column. The use of a syringe for on-column injection into a widebore capillary was rst described by Zlatkis in 1963 [53] and later adapted and developed by the Grobs [54] for smaller diameter columns into a technique widely used today. Other more specialised sample-introduction systems became available for capillary GC and include the PTV injector, and a variety of pyrolyzer systems [22]. PTV injectors can be used in split, splitless or direct mode. Here, the inlet temperature is maintained below the boiling points of solvent and solutes, but is rapidly programmed to vaporize each component in turn. Solutes are then exposed to less thermal stress and large volumes can also be injected. Low levels of volatile organics in environmental matrices may be analysed by headspace, dynamic stripping or purge-and-trap sampling. In purge-and-trap, the sample is purged with helium and volatile analytes collected on a trap of adsorbent material from which they are released by rapid heating.

4. Multi-dimensional GC
Two-dimensional (2D) chromatography was suggested by Martin in 1944 [55]. A considerable increase in peak capacity is achieved if a mixture to be analysed is subjected to two coupled separations with different separation mechanisms [56]. On-line coupling of liquid chromatography (LC) with capillary GC was reported [57] by Majors in 1980, and Grob used concepts developed in his laboratory for largevolume GC injection to achieve considerable improvements in coupled LC-GC [58]. In 1968, Deans showed [59] heart-cutting via the so-called Deans switch uidic valve of a region of a GC chromatogram into a second

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

555

GC (so-called GCxGC) should, in principle, achieve the maximum degree of component separation, and this was rst carried out in 1991 by Phillips [61] by modulated injection of consecutive short-time period fractions from a conventional capillary GC column for rapid analysis in a much shorter second column. Remarkable separations of, especially fossil fuels [56], essential oils and air pollutants [62] have been observed.

5. Conclusion and future of GC

Fig. 4. Photograph of the Stanford GC-on-a-Chip.

column with different polarity, and Schomburg demonstrated [60] how this could be achieved for capillary columns. Fully comprehensive GC-

GC expanded with great rapidity over the two decades following its invention in 1952, and much current practice has its roots in that period. The introduction of robust, efcient and reproducible fused-silica capillary columns and the provision of relatively inexpensive but reliable equipment for GC-MS provided a crucial new impetus in the 1980s. High-speed GC, timeof-ight MS, AED, and comprehensive GCGC now promise further expansion. The versatility of modern GC will expand its application areas. In particular, chromatography has a vital role to play in process control in chemical industry, and on-line and at-line GC methods are replacing analysis in a central laboratory, continuing a trend begun at Shell reneries in 1956 [63]. Fast-GC separations, rst demonstrated by Desty in 1962

Fig. 5. Photograph of a modern portable GC (Reproduced with permission from Varian).

Fig. 6. Dr. Lipsky (right) and Maurice Godet, working on a 1950s GC. Not a small device. (Reproduced with permission from the Quadrex Corporation).

556

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

[64] using miniaturised instruments for process [65] and eld-monitoring [66] applications (a GC system for airborne atmospheric monitoring now ies routinely!), are now available. The rst known reference to a GC-on-a-chip was made verbally by James Lovelock in the early 1970s. It was made on a magnesium oxide chip, less than 50 mm20 mm. It used electrolytic and coulometric methods to produce the gas. The Stanford GC was also reported in the 1970s. Shown in Fig. 4, it was a complete, working GC on a silicon wafer, including column, injector and detector. A number of attempts were made to commercialize this, but all failed. It is only relatively recently that reliable microfabricated components have started to appear in GCs, but we still await a truly handheld, portable, small unit. Today, the closest is the Varian Micro GC. It contains its own carrier gas and power supply in a unit measuring only 15 cm25 cm36 cm and weighing only 7 kg (Fig. 5). It makes an interesting comparison with a GC from the late 1950s (Fig. 6). GC has not only 50 years of history, but an exciting future. Acknowledgements The authors are grateful to a number of chromatographers for helpful discussions during the preparation of this article, especially Ted Adlard, David Grant, Ally Lewis and Luigi Mondello. The views expressed, however, are our own; history can only be subjective, and we apologise both for inevitable omissions and for conicts between our recollections and those of others!

References
[1] A.T. James, A.J.P. Martin, Biochem. J. 50 (1952) 679. [2] A.J.P. Martin, R.L.M. Synge, Biochem. J. 35 (1941) 1358. [3] E. Smolkova -Keulemansova , J. High Resol. Chromatogr. 23 (2000) 497. [4] A.T. James, A.J.P. Martin, G.M. Smith, Biochem. J. 52 (1952) 238. [5] A.T. James, Biochem. J. 52 (1952) 242. [6] S. Dal Nogare, R.S. Juvet Jr., Gas-Liquid Chromatography, Interscience, New York, USA, 1962.

[7] S.R. Lipsky, R.A. Landowne, Ann. Rev. Biochem. 29 (1960) 649. [8] A. Zlatkis, J.F. Oro, A.P. Kimball, Anal. Chem. 32 (1960) 162. [9] G.R., Eglinton, R.J., Hamilton, R., Hodges, R.A., Raphael, Chem. Ind. (London) (1959) 955. [10] W.G. Jennings, S. Leonard, R.M. Pangborn, Food Technol. 14 (1960) 583. [11] H.G. Purnell, Gas Chromatography, John Wiley, New York, USA, 1962. [12] J.H. Knox, , Chem. Ind. (London) (1955) 1631. [13] A.J.P. Martin, D.H. Desty (Editors), Vapour Phase Chromatography, Butterworths, London, 1957, p. 2. [14] M.J.E. Golay, in: V.J. Coates (Editor), Gas Chromatography (1957) (Lansing Symposium), Academic Press, New York, USA, 1958, p. 1. [15] M.J.E. Golay, in: D.H. Desty (Editor), Gas Chromatography, Butterworths, London, 1958, p. 36. [16] D.H. Desty, J.N. Haresnape, B.H.F. Whyman, Anal. Chem. 32 (1960) 302. [17] M. Novotny, K.D. Bartle, Chromatographia 7 (1974) 122. [18] K. Grob, Chromatographia 8 (1975) 423. [19] G. Schomburg, H. Husmann, F. Weeke, J. Chromatogr. 99 (1974) 63. [20] R.D. Dandeneau, E.H. Zerenner, J. High Res. Chromatogr. 2 (1979) 351. [21] L.S. Ettre, J.E. Purcell, Adv. Chromatogr. 10 (1974) 1. [22] M.L. Lee, F.J. Yang, K.D. Bartle, Open Tubular Column Gas Chromatography, Wiley, New York, USA, 1984. [23] C.L. Woolley, K.E. Markides, M.L. Lee, K.D. Bartle, J. High Res. Chromatogr. 9 (1986) 506. [24] B.W. Wright, B.E. Richter, M. Lee, in: M. Novotny (Editor), Recent Advances in Capillary Chromatography, Wiley, New York, USA, 1982. [25] K.D. Bartle, C.L. Woolley, K.E. Markides, M.L. Lee, R.S. Hansen, J. High Res. Chromatogr. 10 (1987) 128. [26] W. Jennings, E. Mittlefehldt, P. Stremple, Analytical Gas Chromatography, 2nd Edition, Academic Press, New York, USA, 1997. [27] J.H. Grifths, D.H. James, C.S.G. Phillips, Analyst 77 (1952) 897. [28] W.E. Harris, H.W. Habgood, Programmed Temperature Gas Chromatography, Wiley, New York, USA, 1966. [29] N. Ray, J. Appl. Chem. 4 (1954) 21. [30] A.J.P. Martin, in: M. van Swaay (Editor), Proceedings 4th International Symposium on GC, Hamburg, 1962, Butterworths, London, 1962, p. xxvii. [31] J. Harley, W. Nel, V. Pretorius, Nature 181 (1958) 177. [32] I.G. McWilliam, R.A. Dewar, Nature 181 (1958) 760. [33] B.J. Gudzinski, in: L.S. Ettre, A. Zlatkis (Editors), The Practice of Gas Chromatography, Interscience, New York, USA, 1967, p. 239. [34] J.E. Lovelock, S.R. Lipsky, J. Amer. Chem. Soc. 82 (1960) 431. [35] J.E. Lovelock, Anal. Chem. 33 (1961) 162. [36] A. Karmen, L. Giuffrida, Nature 201 (1964) 1204. [37] B. Kolb, J. Bischoff, J. Chromatogr. Sci. 12 (1974) 625. [38] D.W. Grant, in: D.H. Desty (Editor), Vapour Phase Chromatography, Butterworths, London, 1957. [39] S.S. Brody, J.E. Chaney, J. Gas Chromatogr. 4 (1966) 42. [40] K. Holden, Ph.D. thesis, University of Leeds, 1998.

trends in analytical chemistry, vol. 21, nos. 9+10, 2002

557

[41] A.J. McCormack, S.C. Tong, W.D. Cooke, Anal. Chem. 37 (1965) 147. [42] L. Ebdon, S. Hill, R.W. Ward, Analyst 111 (1986) 1113. [43] B.D. Quimby, J.D. Sullivan, Anal. Chem. 62 (1990) 1027. [44] H.G.J. Mol, Th. Hankemeier, U.A.Th. Brinkman, LC.GC Intern. 12 (1999) 108. [45] C. Gohlke, Anal. Chem. 31 (1959) 535. [46] M. Novotny, J.M. Hayes, F. Bruner, P.G. Simmonds, Science 189 (1995) 215. [47] G. Hammer, B. Holmstedt, R. Ryhage, Anal. Biochem. 25 (1968) 532. [48] M. van Deursen, J. Beens, H.-G. Janssen, P.A. Leclercq, C.A. Cramers, J. Chromatogr. A 878 (2000) 205. [49] D.H. Desty, A. Goldup, B.A.F. Whyman, J. Inst. Petrol 45 (1959) 287. [50] G. Schomburg, in: R.E. Kaiser (Editor), Proceedings 4th International Symposium on Capillary Chromatography, Huethig, Heidelberg, Germany, 1981, p. 371. [51] P. Sandra (Editor), Sample Introduction in Capillary Gas Chromatography, Huethig, Heidelberg, Germany, 1985. [52] K. Grob, G. Grob Jr., J. Chromatogr. 7 (1969) 584. [53] A. Zlatkis, J.Q. Walker, J. Gas Chromatogr. 1 (1963) 9. [54] K. Grob, K. Grob Jr., J. Chromatogr. 151 (1978) 311.

[55] R. Consden, A.H. Gordon, A.J.P. Martin, Biochem. J. 38 (1944) 244. [56] J. Beens, J. Blomberg, P.J. Schoenmakers, J. High Resol. Chromatogr. 23 (2000) 182. [57] R.E. Majors, J. Chromatogr. Sci. 18 (1980) 571. [58] K. Grob, B. Schilling, J. High Resol. Chromatogr. 8 (1985) 726. [59] D.R. Deans, Chromatographia 1 (1968) 18. [60] G. Schomburg, F. Weeke, in: S.G. Perry E.R. Adlard (Editors), Gas Chromatography 1972, The Institute of Petroleum, London, 1972, p. 285. [61] Z. Liu, J.B. Phillips, J. Chromatogr. Sci. 29 (1991) 227. [62] J. Beens, U.A.Th. Brinkman, Trends Anal. Chem. 19 (2000) 260. [63] J. Hooimeijer, A. Kwantes, F. van de Craats, in: D.H. Desty (Editor), Gas Chromatography 1958, Butterworths, London, 1958, p. 288. [64] D.H. Desty, A. Goldup, W.T. Swanton, in: N. Brenner (Editor), Gas Chromatography, Academic Press, New York, USA, 1962, p. 105. [65] T. Lynch, in: A.J. Handley, E.R. Adlard (Editors), Gas Chromatographic Techniques and Applications, Shefeld Academic Press, Shefeld, UK, 2001, p. 298. [66] J.R. Hopkins, I.D. Jones, A.C. Lewis, J.B. McQuaid, P.W. Seakins, Atmos. Environ. 36 (2002) 3217.