1072
Anal. Chem. 1990, 62, 1872-1876
Capillary Electrophoresis in 2 and 5 pm Diameter Capillaries: Application to Cytoplasmic Analysis Teresa M. Olefirowicz and Andrew G . Ewing*
Department of Chemistry, 152 Davey Laboratory, Penn State Uniuersity, University Park, Pennsylvania 16802
Caplllary electrophoresls wtth electrochemlcal detection in 2and 5-pm caplllaries has been developed to study ultrasmall blologkal environments. Sample volumes as low as 270 fL have been Injected into the electrophoresls capillary with subattomole detection llmlts for easlly oxldlzed species. We have applied this method to the analysis of slngle cell cytoplasm. Sampllng of the cytoplasm is accomplished by inserting one end of the electrophoresls capillary directly Into a single nerve cell. The high-voltage end of the electrophoreds caplllary has been etched wtth hydrofluoric acid to form a mlcrolnjector. thls lnjectlon scheme represents an improvement over those previously used for simHar appllcatlm. The excellent selectivity of thls method is demonstrated for catechoiamlne and lndolamlne neurotransmitters and their metabolltes found In the Invertebrate system, the pond snail
Planorbis corneus.
INTRODUCTION The major goal in developing capillary electrophoresis (CE) in very small capillaries is in its compatibility with the direct analysis of biological microenvironments, specifically at the single cell level (1-3). We as well as several other researchers have studied neurotransmitters in single cells by using indirect sampling methods and whole cell analysis (1-9). One method has involved the extrapolation of neurotransmitter concentrations, based on enzyme kinetics (5). Barber and Kempter (6) have removed and homogenized whole cells and subsequently injected them onto an high-performance liquid chromatography column. Jorgenson and co-workers have separated and detected several components in aliquots of homogenized single nerve cells by using open tubular liquid chromatography (7,8) and CE (9).Previous single cell work in our laboratory used a glass microinjector which was placed over the end of an electrophoresis capillary and allowed us to crudely inject fairly large (approximately 200 pL) volumes of cytoplasm onto to the CE system (1-3, 10). In this paper, we present a means to directly sample single cell cytoplasm without extensive sample handling. A microinjector is used for transferring the cytoplasmic samples into an electrophoresis capillary. This technique represents an improvement over other microinjection techniques used with CE (1-3,9) with respect to minimizing band broadening effects due to large dead volumes (10) and eliminates laminar flow during microsyringe injection as well as extensive sample preparation (9). Additionally, no excess volume is needed for mechanical sample injection. Our microinjector tip is constructed at one end of the fused silica electrophoresis capillary. Tips are etched to 8-10 pm 0.d. in HF ( I I ) , which allows direct insertion of the electrophoresis capillary into the cell body of a neuron with minimal cell membrane destruction. The membrane appears to reseal around the injector tip. Use of this microinjector and the * Author to whom correspondence should be addressed.
techniques described allows us to obtain information concerning the identity and concentration of easily oxidized neurotransmitters and metabolites in the cytoplasm of single, intact neurons. In addition, these methods should allow determination of the spatial distribution of neurotransmitters within cells. EXPERIMENTAL SECTION CE Apparatus. The system used for electrochemicaldetection with CE has been described elsewhere (1,12,13). Briefly, a small (approximately 1 cm) length of capillary is coupled to the end of the longer electrophoresis capillary by use of a porous glass capillary (Corning Glass, Corning, NY)and a carbon fiber electrode is inserted into this assembly. In this work, 5 pm diameter carbon fibers ( h o c 0 Performance Products, Greenville,SC) were etched to smaller diameters for use in 2 and 5 pm i.d. capillaries. Carbon fiber electrodes used in 5 pm i.d. capillaries were cut to the desired length, placed into a 3 M KOH solution, and etched electrochemically(14). A waveform generator and a two-electrode potentiostat were used to apply a square waveform with the potential limits of h1.75 V for etching. A Pt wire served as the reference electrode. The waveform was applied for approximately 2 min at a frequency of 1 Hz until the required electrode size was obtained, typically 2.0-2.5 pm diameter. Exposed lengths of 100-200 pm were employed as working electrodes for electrochemical detection. Carbon fiber electrodes used in 2 wm i.d. capillaries were etched to 1 pm diameter by placing them in a methane flame for a few seconds. With practice, submicrometer tips can be obtained by this procedure. Detection was performed in the amperometric mode with a two-electrode configuration. Electrochemical detection was at 0.7 V versus a sodium-saturated calomel electrode (SSCE) reference electrode. The detection end of the system was housed in a Faraday cage in order to minimize the effects of external noise sources. Fused silica capillaries having the dimensions of 2 pm i.d./150 pm 0.d. and 5 pm i.d./140 pm o.d. were obtained from Polymicro Technologies (Phoenix, AZ) and used as received. Injections were performed by electromigration. Injection volumes were calculated based on electroosmotic flow measured with a neutral marker. All data were collected by an IBM PS-2 computer with “System Gold” software (Beckman Instruments, Palo Alto, CA). Etched Microinjector. Microinjectors were prepared by removing about 5 mm of the polyimide polymer coating from one end of an electrophoresiscapillary to expose the fused silica. The capillary was then connected to a He tank and purged with 100 psi of He during the etching process, which eliminated the diffusion of HF into the capillary. The capillary tip was lowered, using a micromanipulator, into a 40% aqueous solution of HF. The exposed region of fused silica was etched to an outer diameter of 8-10 pm after about 35 min, while the inner diameter remained at 5 pm. After etching, the tip was placed into a concentrated solution of calcium carbonate to neutralize the acid and then washed with doubly distilled water. Injectors were cut with a scalpel to give a length of approximately 300 pm. Capillaries were etched before coupling to the electrochemical detector. Snail Preparation. The snail was pinned to a wax-filled petri dish and dissected, under an optical microscope, to reveal the brain. The microinjector tip was inserted directly into the cell body of the neurons of interest using a micromanipulator. A Pt wire, placed in contact with the snail ringer solution covering the snail preparation, served as the electrophoresis injection anode. Injection of cell cytoplasm into the electrophoresis capillary was carried out by electromigration using a 10-kV injection potential.
0003-2700/90/0362-1872$02.50/00 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 17, SEPTEMBER 1, 1990
1873
I
0.1 pA
I
-0.2
1
0.0
0.2
0.4
0.6
0.8
E [ V vs SSCE]
Figure 1. Oxidation of 1 X lo4 M DA,L-DOPA,and DOPAC in 25 mM MES buffer (pH = 5.6) at a 2 pm diameter electrochemically etched cylindrical carbon fiber electrode. Scan rate was 100 mV s-’.
Currents passed during the injection were typically 5-10 nA. Following injection, the microinjector end of the capillary was removed from the cell and placed into a buffer reservoir for a normal CE run. Detection of the easily oxidized species from the cytoplasm was carried out with off-column electrochemical detection (12, 13). Reagents. 2-(N-Morpholino)ethanesulfonicacid (MES), dopamine (DA), L-dihydroxyphenylalanine (L-DOPA),dihydroxyphenylacetic acid (DOPAC), serotonin (5-HT), 5-hydroxytryptophan (5-HTP), 5-hydroxyindoleacetic acid (bHIAA), ascorbic acid (AA), homovanillic acid (HVA), and catechol (CAT) were obtained from Sigma. All chemicals were used as received. The electrolysis buffer was 25 mM MES adjusted to pH 5.65 with NaOH. Solutes were prepared as 0.01 M stock solutions in 0.1 M perchloric acid. Calibration standards were diluted to the desired concentration in operating buffer. Snail ringer solution was composed of 39 mM NaC1,1.3 mM KCl, 4.5 mM CaC12, 1.5 mM MgC12, and 6.9 mM NaHC03, pH 7.4 in doubly distilled water. Hydrofluoric acid was obtained as a 40% aqueous solution from Aldrich. Safety Considerations. Care must be taken when working with high voltage. Accidental contact with high voltage can be prevented by use of a Plexiglas box interlock system surrounding the high-voltage electrode. Appropriate precautions must also be taken when working with HF since it can cause serious burns. HF should be neutralized with sodium carbonate prior to disposal. RESULTS AND DISCUSSION Electrophoresis in 5 pm i.d. Capillaries. The smallest inner diameter electrophoresis capillary that can be used with electrochemical detection is limited by the size of the detection electrode. Sufficiently small electrodes used for detection in 5-pm capillaries have been made by electrochemically etching (14) commercially available carbon fibers to approximately 2 pm diameter. The electrochemical etching procedure can be used to etch the electrodes to about half their original outer diameter but was only marginally successful in reducing the diameter further. Figure 1shows cyclic voltammograms for oxidation of DA, L-DOPA, and DOPAC a t an electrochemically etched carbon fiber electrode. Electrochemically etched carbon fiber electrodes exhibit a decreased reversibility to anions relative to cations, and hence, anions display a decreased sensitivity at electrode potentials below the limiting current. This may be due to a buildup of surface oxides which might form an electrostatic barrier to anions (15). Quantitation of cations and anions must be carried out using calibration standards for each compound. Figure 2 shows an electropherogram obtained in a 5 pm i.d. capillary. The buffer was modified with 10% IPA and the injection volume was 36 pL (based on electroosmotic flow). The total column volume for an 85 cm, 5 pm i.d. capillary is
1 9
I 24
19
14
29
34
time lmlnl
Electropherogramobtained in a 5 pm 1.d. capillary using an electrochemically etched carbon fiber electrode: capillary length, 85 cm; buffer, 25 mM MES (pH = 5.65)modified with 10% (v/v) 2propanol; injection, 5 s at 25 kV; separation potential 25 kV (i = 13 nA). The injection volume, based on electroosmotic flow was 36 pL: A, serotonin: B, norepinephrine; C, epinephrine; D, ldihydroxyphenylalanine;E, 5hydroxyindoleaceticacid; F, homovanillic acld; G, dihydroxyphenylaceticacid; H, ascorbic acid. Flgure 2.
Table I. Linear Range of Selected Compounds DA y intercept
2.47
slope 3.813-2 corr coeff 0.989 max amt inj, fmol 0.979 min amt inj, amol 10.0
5-HT
L-DOPA
CAT
1.78 5.793-2 0.998
0.829 3.483-2 0.995
1.32 3.333-2 0.992 1.04 3.30
1.12
1.27
5.10
6.90
17 nL. The 36-pL injection represents 0.2% of the total column volume. Assuming a diffusion coefficient of 5 X lo4 cm2s-l, this injection volume limits the maximum efficiency that can be obtained to 960000 theoretical plates (16). The signal in amperometric detection is proportional to the efficiency of the oxidation reaction. In our detection scheme, as the capillary diameter is decreased with a constant electrode diameter, an increase in the coulometric efficiency is observed since the electrode to capillary wall distance decreases. This leads to increased sensitivity, and lower mass detection limits are observed. A 2 pm 0.d. carbon fiber inserted into a 5 pm i.d. capillary produces an annular flow region thickness of about 1.5 pm. This thin-layer flow cell geometry allows more complete oxidation of the analyte zones a t the electrode surface and, thus, increases the observed signal over detectors having a geometry which limits oxidation efficiency. In the work presented in this paper, oxidation efficiencies ranged from 30 to 50%. This range represents variation in electrode sensitivity and detection capillaries. In one set of experiments, detection of four substances of interest has been examined for total injection amounts ranging from 3.3 to 1270 amol to evaluate the linear range of our detection system. Linear regression analysis provides correlation coefficients of 0.99 or better for DA, 5-HT, L-DOPA, and CAT over this range (Table I). The detection limit routinely obtained for CAT is 3.3 amol (SIN = 2, peak-to-peak noise evaluated over 10 peak widths). This detection limit corresponds to a concentration detection limit of 6 x lo4 M for a 50-pL injection. Injection of Ultrasmall Volumes Samples. The ultrasmall sample capabilities of small bore capillaries are demonstrated in Figure 3 for the separation of 5-HT from catechol. In this experiment, only 270 fL of sample volume has been injected. Amounts of 2 amol and 1amol of 5-HT
1874
ANALYTICAL CMMISTRY; VOL. 62. NO. 17, SEPTEMBER 1. 1990 A
55
4.5
5.5
time lminl
F b r e 3. Caplllary elechophwellc separation 01 two easiiy oxidized Substances in an injection volume 01 270 n. calculated based on elecboomnlk M w : capillary length. 56 cm: capwary imide diimetef. 5 pm: buHer, 25 mM MES: injecaon. 2 sat 100 V; separation potential. 25 kV ( i = 28 nA): (A) 2 amol of serotonin. (B) 1 amol 01 catechol.
System used lor removal. separation. and detectbn 01 cytoplasmic samples Ifom single newe ceiis wim an expanded view of the len and right pedai ganglia 01 h s brain 01 Planwbs conwus.
Fbue 5
6A
1
PA
SEROTONIN NEURON
1
c, 68
P
12
8
I€
TIME (MINI
Flgure 4. Scanninng electron micrograph of an etched microinjector la acquiring and injecting cytoplasmic samples. constructed at G wh e voltage end of a 5 prn i.d. electrophoresis capillary. Bar represents 100 p m .
and catechol were injected, respectively. The detection limits calculated for this particular experiment are 0.5 and 0.6 amol, respectively, and represent the best detection limits obtained for this system. The ability to inject ultralow volume samples is extremely important in the analysis of biological microenvironments. We have been developing methods to monitor substances in the cytoplasm of single nerve cell bodies (1-4). One such system is the giant dopamine neuron of Planorbis comeus (pond snail) which is approximately 200 pm in diameter. Assuming that this cell is spherical, the total cell volume is approximately 4 nL. Hence, a 36-pL injection volume represents 0.9% of the total cell volume. This suggests that sampling cytoplasm directly onto the capillary system is possible. In fact, the subpicoliter volume injections demonstrated above might soon allow us to use capillary electrophoresis to acquire and separate samples from single mammalian neurons. Since a 56-em
Figwe 8. Comparison 01 -ants 01 a cytoplasmic JamW obtained from a large serotonin neuron 01 Planorbis comeus(part A) to an electropherogram 01 serotonin and catechol obtained from a standard sduiion (part E): separaw capillary le@. 77 cm: capWlaw inside diameter. 5 pm: buffer. 25 mM MES (pH = 5.65); in)action. 5 s a t 10 kV (i= 9 nA); Separation potentail 25 kV (/ = 21 nA). The electrophwetic mobilities 01 peaks A and C in part A cwrespond to the calculated electrophoretic mobilities 01 serotonin (basis for identification)and an unidentified anion, respectively.
capillary with an inside diameter of 5 pm was used here, a 270-fL injection corresponds to only 0.002% of the total column volume and would be approximately 2% of the volume of a 30 pm diameter mammalian cell. Injection of Cytoplasm from Single Cells. Figure 4 shows a scanning electron micrograph of the microinjector developed for acquiring and injecting samples from single cell cytoplasm. The injector was constructed by etching the positive electrode end of the capillary in HF as described in the Experimental Section. A schematic of the CE system used for acquiring, separating, and detecting cytoplasmic samples directly obtained from single, intact neurons is shown in Figure 5. The etched electrophoresis capillary has been inserted directly into the cell bodies of single large dopamine- and serotonin-containing nerve cells. The neurons chosen for these studies are relatively large and easily identified on the left and right pedal ganglia of the snail brain (17, 16). Figure 6 compares an electropherogram of a cytoplasmic sample obtained from a large serotonin (5-HT) neuron of Planorbis corneus (Figure 6A) to an electropherogram of 5-HT
ANALYTICAL CHEMISTRY, VOL. 62, NO. 17, SEPTEMBER 1, 1990
Table 11. Determination of Serotonin and Dopamine in the Cytoplasm of Single Nerve Cells
A
10.2pA
serotonin neuron dopamine neuron [5-HTIo am01 [DA]" amol
N
3 3.1 pM SEM 0.57 pM apparent volumes* 105-134 pL
x
358 88
5 2.2 pM 0.52 pM 73-103 pL
1875
197 48
T
a Quantitation was carried out with peak area, terminating the area measurement prior to the asymmetric portion of the peak. *Apparent volumes were calculated with the equation V = [pe + peo]EAtid. Volumes for injection by electroosmotic flow were 67-83 pL for the serotonin experimenta and 5 W 6 pL for the dopamine experiments.
and CAT obtained from a standard solution (Figure 6B). The peaks in the electropherogram of the cytoplasmic sample were identified by calculating the electrophoretic mobilities of each peak and comparing them to those of authentic standards. The electrophoretic mobility of peak A from the cytoplasmic sample corresponds to that of the 5-HT standard. The neutral peak (peak B) cannot be identified, even tentatively, by zone electrophoresis; however, it might contain a neutral precursor of 5-HT and possibly more than one component. Quantitation of the cytoplasmic levels of neurotransmitters, using the capillary electrophoresis microinjection and detection techniques described, is shown in Table 11. These measurements test the reproducibility of the entire assay, including brain sampling inhomogeneities. Data are shown for multiple analyses on both a large serotonin cell and the giant dopamine cell where the neurotransmitter amounts were calculated by using the average calibrations from both the pre- and postcalibration electropherograms. One analysis was performed on each snail. Multiple analyses from one cell have not yet been attempted due to the possibility of cell membrane destruction. Further reduction in microinjector diameter, through the use of 2 pm i.d. capillaries, may be necessary for multiple analyses from one cell. Work is in progress to test the feasibility of this approach. An additional point of interest in these data is that the peaks that appear to represent neurotransmitters removed from single cells (both serotonin and dopamine) are highly asymmetric. One explanation for this observation might be the sampling of vesicles from within the cell cytoplasm. This is currently under investigation in our laboratory. Injection volumes for the cytoplasmic analysis, based on electroosmotic flow, ranged from 50 to 83 pL. However, since the injection of cations and anions involves both electrophoresis and electroosmotic flaw, the apparent injection volume for these ions is different than the liquid volume injected. Thus, an apparent volume must be determined when using calibration standards that are based in total moles in order to calculate the concentration of the injected sample. This apparent volume of injection, V, can be calculated by the equation where pe is the electrophoretic mobility of the substance, pw is the coefficient of electroosmotic flow, E is the field strength, A is the cross-sectional area of the capillary bore, and tinjis the injection time. In these experiments, apparent volumes of injected cation ranged from 73 to 134 pL. On the basis of the apparent volumes, these results indicate that the cytoplasmic level of serotonin in the serotonin cell examined is 3.1 f 0.57 pM (SEM, standard error of the mean). The cytoplasmic level of dopamine in the giant dopamine cell is 2.2 f 0.52 pM. One anionic peak removed from the dopamine cell has a similar electrophoretic mobility to that of DOPAC
1 5
15
10
I
20
time lminl Flgure 7. Capillary electrophoretic separation of several amines in a 2 pm 1.d. capillary: separation capillary length, 81 cm; buffer, 25 mM MES (pH = 5.65); injection, 5 s at 25 kV; separation potential, 25 kV (i = 1.4 nA); injection volume, 4 pL calculated based on electroosmotic flow; (A) dopamine, (B) 3-methoxytyramine, (C) ldihydroxyphenylalanine, (D) homovanillic acid, (E) dihydroxyphenylacetic acid.
and appears to be this major metabolite. A fourth peak that has been observed in two experiments with the dopamine cell and the peak that appears at about 15 min (peak C in Figure 6A) in the serotonin cells have mobilities that appear to result from migrating anions but do not match major metabolites examined to date. Neither of these peaks have electrophoretic mobilities corresponding to ascorbic acid, homovanillic acid, or 5-hydroxyindoleacetic acid. Electrophoresis in 2 M r n i.d. Capillaries. As the total column volume becomes increasingly small, the capability of handling ultrasmall (subpicoliter) sample volumes becomes possible. Until now, this powerful aspect of CE has not been investigated due to difficulties with detection. We present some preliminary data demonstrating the potential of reducing the total operating and sampling volumes of this technique that should permit the analysis of extremely small cytoplasmic samples. An electrophoretic separation of cationic, neutral, and anionic species in a 2 pm i.d. capillary is shown in Figure 7. Detection has been accomplished with carbon fiber electrodes that have been etched to approximately 1 pm 0.d. in a methane flame. The total amount of each compound injected ranges from 0.6 to 0.9 fmol. The electroosmotic flow based injection volume is calculated to be 4 pL. The development of CE with 2 and 5 pm i.d. capillaries appears to be very important in the area of sampling from discrete biological environments. The work demonstrated in this paper has shown the applications of very small bore capillaries to the analysis of cytoplasmic samples from single neurons in invertebrate systems. This technique should not be limited to the study of neurochemistry but could also be applied to the investigation of hormone levels, intracellular messenger concentrations, and general metabolism as well as the study of the immune system at the single cell level. The key developments for the application of micro-CE to these areas are going to be in new detection capabilities. The use of 2 pm i.d. capillaries should permit still smaller microinjectors and move the technique a step closer toward an ability to acquire and separate solutes from single mammalian cells in addition to invertebrate cells.
ACKNOWLEDGMENT The authors thank R. A. Saracen0 and S. J. Stranick for technical assistance and helpful discussions. In addition, the
Anal. Chem. 1990,62,1876-1883
1876
aid of R. Walsh in obtaining the scanning electron micrograph is gratefully acknowledged. We also thank Beckman Instruments for their gift of System Gold hardware and software as well as Polymicro Technologies for their gift of 2 pm i.d. capillary. LITERATURE CITED (1) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1988, 6 0 , 1972-1975. (2) Ewing, A. G.;Wailingford, R . A.; Olefirowicz. T. M. Anal. Chem. 1989, 61, 292A-303A. (3) Chien, J. B.: Wallingford, R. A.; Ewing, A. G. J. Neurochem. 1990, 5 4 , 633-638. (4) Chien, J. B.; Saraceno, R. A.; Ewing, A. G. Redox Chemistty and Interfacial Behavfor of Biological Molecuk; Plenum Press: New York, 1988; pp 417-424. (5) Periman, R. L.; Sheard, B. E. Biochim. Biophys. Acta 1982, 779, 334-340. (6) Barber, A.; Kempter, 6. Comp. Biochem. Physiol. 1986, 64C, 17 1- 174. (7) Kennedy, R. T.; St. Claire, R. L., 111; White, J. G.:Jorgenson, J. W. Mkrochlm. Acta (Weln) 1987, I! 37-45. (8) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 67, 436-441. (9) Kennedy, R. T.; Oates, M. D.; Cooper, B.; Nickerson, B.; Jorgenson, J. W. Science 1989, 246, 57-63. I
(10) Wallingford, R. A.; Ewlng. A. 0.Anal. Chem. 1987. 59. 678-681. (11) Budavarl, S., O'Neii, M. J.. Smith, A., Eds. Merdc Index; Merck 8. Co., Inc.: Rahway. NJ, 1969; p 758. (12) Waillngford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762-1766. (13) Waiiingford, R. A.; Ewing, A. G. Anal. Chem. 1988, 6 0 , 258-263. (14) Abe, T.; Itaya, K.; Uchida, I.Chem. Lett. 1988, 399-402. (15) Kepley, L. J.; Bard, A. J. Anal. Chem. 1988, 60, 1459-1467. (16) Xiaohua, H.; Coleman, W. F.; Zare, R. N. J. Chromatogr. 1989, 480, 95-1 10. (17) Marsden, C. A.; Kerkut, G. A. Comp. Gen. Pharmacol. 1970, 7 , 101-1 16. (18) Pentreath, V. W.; Berry, M. S.; Cottrell, G. A. Cell Tissue Res. 1974, 757, 369-384.
RECEIVEDfor review February 15, 1990. Accepted June 4, 1990. This research was supported, in part, by grants from the National Institutes of Health and the National Science Foundation. Financial support of research related to the project has been provided by Beckman Instruments and Shell Development. A.G.E. is a National Science Foundation Presidential Young Investigator, an Alfred P. Sloan Fellow, and a Camille and Henry Dreyfus Teacher Scholar.
Gas Chromatographic Determination of Acetonitrile in Air Using a Thermionic Detector Stephan Hamm and Peter Warneck*
Max-Planck-Institut fur Chemie (Otto-Hahn-Institut), 6500 Mainz, Federal Republic of Germany
A gas chromatographic procedure for the determination of acetonitrile (CH,CN) in air is described, based a thermionic nitrogen+eiective detector. The bwer llmH of detection is 15 pg of CH,CN for a blank value of 5.0 f 3.2 pg. The method involves the collection of air In 2 - d d vessels (glass or stainless steel) and preconcentrationof acetonitrile by cryogenic trapping. Preconcentration by sorption on modHied Chromosorb 102 was found unsuitable due to undesirable memory effects. Air samples stored In containers for periods longer than 3 days generally show a rise In CH,CN concentration, so that alr samples must be processed within a few days. Precautions agalnst contamination in the gas handling system or by intrusion of laboratory air also are important. Calibration tests based on gaseous and liquid dilution series lead to comparable accuracies of about 4 % in each case but a better reproducibility for liquid calibration mixtures. Air samples taken over the Atlantic Ocean, In the Bay of Helgoland, during akcrafl ascents over Europe, and in the urban air of Mainz indicate the range of CH,CN mixing ratios In the troposphere from 52.6 f 13.1 pptr in background air to 731 f 82 pptr in polluted city air.
INTRODUCTION Acetonitrile, CH3CN, is emitted into the atmosphere as a byproduct of combustion processes. The main sources according to recent estimates (I,2) are automobile exhaust gases and the burning of biomass (living and dead plants). Thus, acetonitrile will be a useful indicator for emissions from such processes. The presence of CH3CN in the atmosphere was originally inferred from the composition of positive water-cluster ions, studied in the stratosphere by means of air-borne mass 0003-2700/90/0362-1676$02.50/0
spectrometers (3,4). Since then, two mass spectrometric (3-6) and two gas chromatographic (7, 8) techniques have been utilized to determine the CH3CN content of air and its distribution in the atmosphere. In this paper we describe a third gas chromatographic technique, which we consider particularly suitable for measurements of CH3CN in tropospheric air, and we shall present results from several applications. Mass spectrometry requires either a passive sampling of ambient ions or the artificial ionization of an air sample. Ionization is then followed by a series of ion-neutral reactions leading to the formation of protonated acetonitrile hydrates. Both methods currently are restricted to altitudes above 10 km. A knowledge of the ion chemistry is necessary to interpret the data (5, 6 ) . Gas chromatography (GC) requires a preconcentration of acetonitrile from ambient air samples, because detectors currently in use are not sufficiently sensitive for a direct analysis of air containing less than about 50 ppb of CH&N (1 ppb = parts by volume). The mixing ratio of acetonitrile in the atmosphere is 2 to 3 orders of magnitude lower. Becker and Ionescu (7) collected air in stainless steel containers. They preconcentrated acetonitrile from air by passing it through a Chromosorb trap kept at -90 "C. The trap was then heated to 270 "C to transfer the material into the GC sampling loop, which was cooled with liquid nitrogen. Finally, the sample was flushed onto the GC column by heating the loop. A packed column was used, and the detector was either a mass spectrometer operated in the single ion mode at e l m = 38 daltons or an electron capture detector (ECD). The detection limit was given as 0.01 ppb. Snider and Dawson (8,9) used a vertical metal plate cooled to a temperature below the ambient dew point to collect acetonitrile from the air by co-condensation with water vapor. The aqueous condensate drained from the collector plate was further concentrated by microdistillation and adsorption in 0 1990 American Chemical Society
