A Surface Ionization Detector for Gas Chromatography: Use of a

There are two requirements: use of a supersonic free jet nozzle and the high-work-function surface of rhenium oxide. This detector, which is sensitive...
0 downloads 0 Views 135KB Size
Download PDF
Anal. Chem. 1996, 68, 2776-2781

A Surface Ionization Detector for Gas Chromatography: Use of a Supersonic Free Jet Hiroshi Kishi

Oyama National College of Technology, 771 Nakakuki, Oyama 323, Japan Toshihiro Fujii*

National Institute for Environmental Studies, 16-2 Onogawa, Tsukuba, Ibaraki 305, Japan

A new design of a gas chromatographic surface ionization detector (SID) based on hyperthermal positive surface ionization is described. There are two requirements: use of a supersonic free jet nozzle and the high-work-function surface of rhenium oxide. This detector, which is sensitive in response to all the organics, can be operated as an universal detector with much higher sensitivity toward some species having low ionization energy but with selectivity to a lesser degree than that of a conventional SID. The minimum detectable amount (at S/N ) 3) of pyrene is around 4.4 × 10-13 g/s, with linearity greater than 106, while that of toluene is around 10-12 g/s. Other unique properties are (1) the ability to control the degree of selectivity through molecular kinetic energy, the surface, and its temperature and (2) a very short response time. Surface ionization detectors (SIDs) are used in the gas chromatographic detection of organic compounds, which form their dissociative species at a low ionization energy (IE). It constitutes a very sensitive and selective detection method.1 The advent of a SID in combination with gas chromatography dates back to the early 1960s, when Zandberg and Rasulev2 described the nature of such a detector. In 1984, Fujii and Arimoto3 presented a SID with a hot platinum emitter, which led to the first commercially available SID from Shimadzu Corp. Applications are increasing.4-6 One of the reasons for the increased usage is that this detector provides extremely sensitive and specific responses to organic compounds, which form their dissociative species at a low IE. When an organic compound is supplied to the surface, ionization takes place. This response, which is directly related to the ions formed on the surface, varies with the IE of the sample organics and the work function (φ) of the surface. Therefore, some organics do not show the response. There is no selective response to compounds having high IEs. (1) Fujii, T.; Arimoto, H. In Detectors for Capillary Chromatography; Hill, H. H., McMinn, D. G., Eds.; Wiley Interscience: New York, 1993; pp 169-191. (2) Zandberg, E. Ya.; Rasulev, U. Kh. Russ. Chem. Rev. 1982, 51, 819. (3) Fujii,T.; Arimoto, H. Anal. Chem. 1985, 57, 2625. (4) Rasulev, U. Kh.; Zandberg, E. Ya. Prog. Surf. Sci. 1988, 28, 181. (5) Fujii, T.; Jimba, H.; Ogura, M.; Arimoto, H.; Ozaki, K. Analyst 1988, 113, 789. (6) Hattori, H.; Yamamoto, S.; Iwata, M.; Takashima, E.; Yamada, T.; Suzuki, O. J. Chromatogr. 1992, 581 (2), 213.

2776 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

Recently, a new approach to gas chromatographic SIDs was reported by Amirav.7,8 He presented a new SID based on the finding that, when the organics are supplied with kinetic energy to the surface, surface ionization efficiency is greatly enhanced. The molecular kinetic energy is obtained in a supersonic free expansion of the organic heavy molecule seeded in helium (or possibly hydrogen) carrier gas through a pinhole nozzle. This detector may have been termed a hyperthermal SID. Although the development of the new SID, as described by Amirav, was a breakthrough in the analysis of trace quantities of compounds, the detector has not received the attention that would have been expected in view of its sensitivity, selectivity, and versatility. Part of this is probably due to the limited investigations of this kind of detector, which is not widely available. The purposes of this study are (1) to design and construct a new GC detector based on the hyperthermal surface ionization (the home-built detector in this work is similar to Amirav’s design in that a supersonic free jet technique is used to gain the kinetic energy), (2) to interface to a GC and show that surface ionization is linearly related to the concentration of a sample, (3) to describe the optimal operation parameters for its use in gas chromatography of organic chemicals, (4) to explain its detection mechanism, at least to an extent such that the experimental results and the influence of working parameters can be understood, and (5) to compare its detection characteristics with those of the conventional SID and the well-established flame ionization detector (FID). EXPERIMENTAL SECTION Design of Detector. A new SID was constructed, which is shown schematically in Figure 1. The detector is comprised of (1) a ceramic nozzle assembly, (2) a shield plate disk, (3) a highwork-function surface, and (4) an ion collector, which are installed in the vacuum chamber. The detector’s vacuum chamber (volume, ∼110 cm3) was pumped by a single 1000 mL/min rotary pump (Model EC-602, Ulvac, Tokyo, Japan) to around 2 × 10-3 Torr as an ultimate pressure. A flow meter serving as a flux detector was connected at the outlet of the pump. Vacuum was measured by a pirani gauge (GP-1S, Ulvac), which also serves as a flux detector. The pressure under typical gas flow conditions was around 0.06 Torr. Supersonic Free Jet. The technique of aerodynamic acceleration was used to obtain molecular kinetic energy in the range of 1-10 eV. The organic molecules from the gas chromatographic (7) Danon, A.; Amirav, A. J. Phys. Chem. 1989, 93, 5549. (8) Amirav, A. Org. Mass Spectrom. 1991, 26, 1. S0003-2700(96)00028-5 CCC: $12.00

© 1996 American Chemical Society

Figure 1. Schematic drawing of a new surface ionization detector with the nozzle. The detector chamber is pumped by a 1000 mL/min rotary pump. The typical operating conditions are as follows: stagnation pressure with helium gas used as a seeding gas, 1000 Torr; pressure of the detector chamber, 0.06 Torr. Refer to Table 1.

coupling, which were seeded in a helium (or hydrogen) supersonic beam, enter the vacuum chamber through a ceramic nozzle.9 After the supersonic expansion, the molecules attained the carrier gas velocity while losing energy from their vibrational-rotational degrees of freedom. Aerodynamic equations for seeded supersonic free jet10 used in this work are as follows:

Ek∞ ) 5/2(Mh/M1)kTn g Ek

(1)

Xm ) 0.67(P0/Pb)1/2d

(2)

where Ek is the kinetic energy (eV) of the heavy molecule, Ek∞ is its maximum value, Mh and Ml are the molecular weights (g) of the heavy and light molecules, respectively, k is the Boltzmann constant, Tn is the temperature (K) of the nozzle, P0 (Torr) is the nozzle stagnation pressure, Pb (Torr) is the expansion chamber pressure (background pressure), d (cm) is the nozzle diameter, and Xm (cm) is the distance from the nozzle at which the Mach disk of a free jet shock wave structure would be formed. Nozzle. The ceramic nozzle was an 80 µm thin hole in a small ruby disk mounted on a 2 mm aluminum tube. It has been described in detail elsewhere.11 The nozzle is soldered onto the end of a 1/8 in. o.d. stainless steel tube. The stainless steel tube is 10 cm long and is connected with an ultra-Torr union to the 1/8 in. outlet of a gas chromatograph. The ceramic nozzle mounted on a vacuum chamber could be heated differentially up to ∼700 °C to attain the kinetic energy for organic molecules to be tested. The distance from the top of the nozzle to the surface is 5 mm. Shield. The shield plate disk with the 3 mm i.d. hole at the center is placed 2 mm in front of the ion collector. It can reduce (9) Danon, A.; Amirav, A. Rev. Sci. Instrum. 1987, 58, 1724. (10) Miller, A. P. Free Jet Sources. In Atomic and Molecular Beam Methods, Vol. 1; Scoles, G., Ed.; Oxford Univ. Press: New York, 1988; pp 14-53. (11) Kishi, H.; Fujii, T. J. Phys. Chem. 1995, 99, 11153.

background noise signal by more than 2 orders of magnitude while barely affecting the sample signals. The significant difference in S/N of the sample with the disk placed or not placed enables determination of methanol to a better detection limit. Surface. In the vacuum chamber, the beam collided with a rhenium oxide12 or Pt surface2 for efficient positive ion production. The refractory metal foils were purchased, formed to conical shape (surface area, 1.7 cm2), and fixed to the holder, which can be aligned with the nozzle, heated (up to surface temperature Ts ) 1100 °C), and electrically biased. The rhenium oxide surface was prepared separately in the presence of O2 at a pressure of 1 × 10-5 Torr by electrically heating the Re foil around 1200 K for a few hours. Collector. The ion collector is a concentric geometry of a Faraday cup-type collector surrounding the surface. The surface is always at a positive potential of 200 V against the collector electrode. In terms of collection efficiency, there is a definite advantage in the concentric geometry of the detector design over that of the parallel plate. Procedure. This home-built detector was coupled to a Shimadzu G-23 capillary gas chromatograph. The new detector requires the rotary pump vacuum system and must be placed as a separate unit external to the gas chromatograph. This requires that it be interfaced with a chromatograph via heated transfer tubings. Investigations and changes can be made without affecting the rest of the GC system. The FID and conventional SID used for comparison were commercial ones (Shimadzu Models FID-17 and SID-12, respectively). A 30 m long, 0.32 mm i.d. chemically bonded fused silica capillary column, FFS Ulbon HR-25, was used with helium carrier gas at 1.5 mL/min. This gas flow rate was too small as a seed gas for supersonic jet operation. Therefore, another helium gas was added at the outlet of the column to the level of 150-200 mL/min. The temperature program used for the column oven (12) Fujii,T. J. Phys. Chem. 1984, 88, 5228.

Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

2777

Table 1. Instrumental Parameters and Typical Operational Conditions Ek∞ (kinetic energy) Tn (nozzle temperature) v (He flow rate for seeding) P0 (stagnation pressure) Pb (background pressure) Xm (Mach disk distance) D (nozzle-surface distance) Ts (surface temperature) d (nozzle diameter)

3.5-12.0 eV e1000 °C 10-100 mL/min 800-1600 Torr 1.0 × 10-2-6 × 10-2 Torr 5.4-7.4 mm 5 mm 700-1100 °C 80 µm

was as follows: 40 °C hold in 5 min, then raised to 140 °C at 20 °C/min, up to 280 °C at 40 °C/min, and 280 °C hold for 15 min. The injection port and detector oven were held at 300 °C. Molecular samples for study or calibration were reagent-grade alkyl alcohols, alkylbenzenes, and polycyclic aromatic hydrocarbons (PAHs). They were introduced through a GC. The ionization energies of the test samples fall in the 8.8-10.5 eV range. According to the explanation given above, it is obvious that the parameters of this system are related to each other and varied in a wide range. Thus, the values in Table 1 are typical only for working this system. RESULTS AND DISCUSSION (1) Sensitivity Optimization. There are a number of factors that influence sensitivity. There are, however, only a limited number of factors that can potentially contribute to a significant gain in sensitivity. Of these, the kinetic energy (Ek) of the sample molecules (nozzle temperature, Tn) and surface temperature (Ts) are by far the most important factors. The variations in sample response and background currents with the magnitude of the surface electrode bias voltage were investigated. The results of voltage-signal characteristics led to the conclusion that the ion signals level off at a voltage of more than 200 V. Therefore, all the remaining data reported were obtained at the surface electrode bias voltage of +200 V with respect to the collector electrode. Surface (Emitter) Material. It is hardly possible to choose exactly the best surface that will give the best response at the desired conditions. In developing the present surface formulation, both rhenium oxide and Pt were chosen for the investigation, because the former has been favorably used in surface ionization organic mass spectrometry for a long time,13 while the latter provides good performance as a material for use in a conventional SID.2 The sensitivity study was made using the 10 ng methanol sample. Preliminary experiments indicate that 15 times greater sensitivity is produced by rhenium oxide, and the noise level is about equal for both surfaces, suggesting that the minimum detectable amount is smaller with the rhenium oxide surface, now being used exclusively. Signal versus Ek. Sensitivity varied directly with the kinetic energy (Ek) of the sample compound when all other variables were held constant. Successive injections of a 1 µL acetone solution containing 0.1 µg of toluene were analyzed as the nozzle temperature was increased from 700 to 930 °C. The response results for sample compound in coulombs per gram (C/g), with increasing background and noise-level current in amperes, is given in a (13) Fujii, T. Int. J. Mass. Spectrom. Ion Processes 1984, 57, 63.

2778 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

Figure 2. Kinetic energy dependence of the signal response of toluene molecules in coulombs per gram (on the vertical axis of left side) and the nozzle temperature dependence of the blank background and noise level current in amperes (on the vertical axis of right side). The kinetic energy (eV, on the horizontal axis of lower side), which was controlled by the nozzle temperature (Tn, on the horizontal axis of upper side), was calculated at Ek∞ values. The sample size was 100 ng. The surface temperature was set to 900 °C, and the nozzle stagnation pressure was 1000 Torr. The background current signal is probably from alkali impurities on the surface that were ejected by the accelerated residual molecules.

semilogarithmic plot (Figure 2). This result is to be expected on the basis of the recent finding11,14,15 that the molecular kinetic energy has a dramatic effect in increasing the surface ionization yield. Signal versus Ts. Sensitivity was directly related to the temperature (Ts) of the surface. Figure 3 shows the results as the surface temperature was increased from 760 to 1050 °C. The responses for methanol, benzene, and toluene molecules (in C/g), with increasing background current and noise level in amperes, are given in Figure 3. Figures 2 and 3 show that both the background currents and the noise levels increase proportionally with the kinetic energy, but they increase with the surface temperature in a more complex way. The large background current is probably attributable to the appearance of Na+ and K+ ions from Na and K impure atoms in the surface material.16 This assumption is consistent with the experimental fact that background levels gradually decrease with continuing operation time. Perhaps the decay is associated with the outgassing of alkali impurities from the surface. Besides, the emission of these positive ions from the surfaces is a well-known phenomenon, discussed previously3 for the conventional SID. From the sensitivity point of view, another important factor is the fraction of jet gas flux intercepted by the surface. Therefore, we would like to use a free jet supersonic expansion (without collimation) on a surface. The nozzle-surface distance would (14) Campargue, R. J. Phys. Chem. 1984, 88, 4466. (15) Danon, A.; Amirav, A. Int. J. Mass Spectrom. Ion Processes 1990, 96, 139. (16) Helbing, R. K. B.; Hsieh, Y. F.; Pol, V. J. Chem. Phys. 1975, 63, 5058.

Table 2. Comparison of the New SID with Conventional Flame Ionization Detector FID new SID sensitivity (S) (C/g of Xa) 13.5 MDAb (g/s of X) 4.4 × 10-13 linear dynamic range 106

present workc literatured 3.1 × 10-2 3.9 × 10-12 2 × 105

0.015 3 × 10-11 2 × 106

a X corresponds to pyrene as a test sample in the case of the present work. b Minimum detectable amount at S/N ) 3. c The measurements were made under optimum conditions with the MDA at 3 × 10-12 g/s for diphenyl (see text). d Cited from ref 18; X corresponds to n-propane.

have an effect on the fraction of gas flux intercepted by the surface. When the distance is shorter, there is an apparent increase in the signal response. A very short nozzle-to-surface target distance, however, caused large increases in the noise level as well as the background current. The noise appears to be generated by nozzle assembly, which emits alkali ions, especially when it is heated. However, the above-mentioned shield plate disk can reduce the background current drastically. In the following experiments, the nozzle-surface distance was fixed at a constant 5 mm with the shield disk plate so that the optimum signal-to-noise ratio can be provided. The influence of the gaseous environment surrounding the surface was tested. The preliminary studies demonstrated that the oxygen addition to the detector’s chamber has some influence on the sensitivity of the detector. Note the following explanation: Oxygen addition to the detector may somehow lead to the increase in φ.17 The problem, however, is connected with oxygen supply. A 300 mL/min oxygen supply caused the pressure increase of the detector’s chamber to go up to almost 0.1 Torr. Under these conditions, the supersonic free jet does not work in the proper manner, and hence, sensitivity is suppressed. The overall effect of the oxygen supply on detector response was not definite. Therefore, in the present work, the detector remained operational without any additional gas supply. 2. Performance Characteristics. Performance was assessed for pyrene as a test sample with the criteria of dynamic range, sensitivity, selectivity, minimum detectable amount (MDA), and comparison with a conventional SID. This compound was chosen because of its considerably high molecular mass, from which the high kinetic energy can be derived. Dynamic Range. For pyrene, a dynamic range has been investigated under the optimum conditions for the detection

limit: a surface temperature of 900 °C, which gives a background current of 4 × 10-11 A and a noise level of 2 × 10-12 A, and a kinetic energy of 9.9 eV (nozzle temperature of 930 °C). The response is linear over 6 orders of magnitude in a sample amount. Sensitivity. Sensitivities were calculated as the linear regression slope of the linear calibration plots. The sensitivity (S) of the detector can be expressed as coulombs per gram of sample. For pyrene, S ) 13.5 C/g. This is taken under the conditions that Tn ) 930 °C and Ts ) 900 °C. Selectivity. As already mentioned, the sensitivity of the detector is strongly dependent on the IE of the species as well as on the yield of the species generated through the chemical reaction on the surface. Furthermore, the ionization yield strongly increases with the molecular mass, as the available molecular kinetic energy in supersonic molecular beams linearly increases with the molecular mass. Presumably, the relative sensitivity for different organics, which is easily determined by comparison of the signal currents, varies in a wide range from sample to sample. However, selectivity can be minimized to a lower level at high molecular kinetic energy. Limited control over selectivity is also possible through the surface and its temperature or choice of carrier gas (or a seeding gas). Consequently, selectivity of this detector is tunable. MDA (Noise). As can be seen from Figure 3, at higher surface temperatures, a decrease in the ratio of signal peaks to the detector noise level occurs. This result indicates that an optimum surface temperature has to be found for the detection limit. Generally, favorable operation is achieved with a surface temperature of 900 °C. That generates a noise level of 2 × l0-12 A with the nozzle temperature at 930 °C. Under optimum conditions, MDA at the signal-to-noise ratio of 3 was measured using pyrene. The result is listed in Table 2. This gives a comparison of performance characteristics of the new SID with those of the flame ionization detector (FID). All values of the FID are both from the present work and from literature. This comparison demonstrates that the new SID provides extremely high sensitivity compared with that of the well-established FID. 3. Chromatogram. Figure 4 shows examples of responses obtained when a mixture of alkyl alcohols are chromatographed and measured by new SID and FID detectors. The upper trace shows these compounds as analyzed by the FID at an electrometer setting of 10-12 amperes full scale (AFS). The measurements both in the SID and FID modes were made under optimum conditions for MDA. The optimum conditions of the FID gave the MDA at 2.3 × 10-12 g/s for diphenyl, which is very close to the catalog

(17) Davis, W. E. Environ. Sci. Technol. 1977, 11, 587.

(18) Kolb, B.; Auer, M.; Pospisil, P. J. Chromatogr. Sci. 1977, 15, 53.

Figure 3. Plot of surface temperature (Ts) versus the samples (benzene, toluene, and methanol), signal current (solid line), noise level current (dotted line), and background current (dashed line) under the conditions that Tn ) 930 °C and sample size is 100 ng for each sample molecule in acetone.

Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

2779

Figure 4. Comparison between (A) conventional FID (upper trace) and (B) new SID (lower trace). Both gas chromatograms are of a mixture of 1 ng of alkyl alcohols CnH2n+1OH (n ) 1-5); (1) methanol, (2) ethanol, (3) 2-propanol, (4) n-propanol, (5) 2-butanol, (6) 1-butanol, and (7) n-pentanol. This mixture was injected into a gas chromatograph having a 30 m capillary column (FFS Ulbon HR-25, 0.32 mm i.d.). The new SID conditions are Tn ) 930 °C, Ts ) 900 °C. The FID conditions are H2 flow rate, 30 mL/min; air flow rate, 600 mL/min.

specification of 3 × 10-12 g/s. The lower trace shows the same sample analyzed by the new SID at a setting of 2 × l0-10 AFS. It was necessary to attenuate 200 times to keep the response on scale for all the sample components, whose size is 1 × 10-9 g each. In Figure 5, a new SID chromatogram of a series of PAHs is again compared with the FID. The difference in the two chromatograms is clear. In the case of pyrene, the detection capability of the SID is 8.9 times higher than that obtained by the FID. By comparing the records of Figures 4 and 5 from both detectors, it is found that (1) SID operation provides significantly higher sensitivity than the FID operation, while the noise level for the former is ∼2 orders of magnitude higher than the latter; (2) the new SID sensitivity somehow depends on the molecular (or alkyl chain) mass; and (3) the SID detector provides as good resolution as the FID, demonstrating that the new detector responds rapidly enough to follow sharp capillary peaks. This is consistent with the fact that (1) the SI response time is short, because of its character due to scattering events, and (2) the use 2780 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

Figure 5. Comparison between (A) conventional FID (upper trace) and (B) new-SID (lower trace). In both cases, the detected gas chromatograms were of a mixture of polycyclic aromatic hydrocarbons: (1) naphthalene, (2) anthracene, (3) pyrene, and (4) chrysene mixture in a benzene solution. The different sample sizes were used in the FID and the new SID injections, as indicated. The upper trace is for new SID detection with Tn ) 930 °C and Ts ) 900 °C. In the lower trace, the FID signal is given. The FID conditions are H2 flow rate, 30 mL/min; air flow rate, 600 mL/min. The gas chromatographic conditions are the same as those of Figure 4.

of low-pressure or vacuum outlet techniques for gas chromatography is widely appreciated with greater separation efficiency.19 A new SID chromatogram of a series of n-alkylbenzenes (benzene, toluene, and o-xylene) is compared with that of a conventional SID. Five nanograms benzene is barely detected with the old SID operated at Ts ) 800 °C. With the new SID at Tn ) 930 °C and Ts ) 900 °C, 5 ng of benzene will give almost one-tenth response at 6.4 × 10-10 AFS. This is 980 times greater than the response of the conventional SID. It is demonstrated that the former can be operated as an universal detector. However, it should be noted that, in the case of toluene (o-xylene), the new SID gives 120 times (50 times) higher sensitivity than the conventional SID; the former is less selective than the latter. On occasion, the baseline also exhibits a sharp backward response at the peak tail. No completely satisfactory explanation of this unstable baseline has been formulated as yet. Attempts to find the cause were not successful. The similar SID response appears to be sensitive to the analyte matrix, which could be a difficult problem to solve and could result in poor quantitative analysis. 4. Conclusion. The state of the art in supersonic free jet technique has been incorporated in a novel SID. Universal (19) Hail, M. E.; Yost, R. A. Anal. Chem. 1989, 61, 2402.

response using SID is a new development. Conditions have been found under which sensitivity is greatly increased by the introduction of a supersonic free jet technique. All the response characteristics are consistent with a process of positive surface ionization. Therefore, a high-work-function surface is favorable for the sensitive analysis, even for the new SID. In its present state of development, the sensitivity response of the hyperthermal SID to most organics including aliphatic alcohols is about 100 times greater than the response obtained for the same compounds using a conventional FID. The detector has a minimum detectability for pyrene of 4.4 × 10-13 g/s, and a linear range of operation of 106. The conventional SID acts as a highly selective detector, while the new SID acts as a universal one with an additional sensitivity toward some species having low IE. But the new SID can work as a conventional SID if the nozzle is withdrawn from the gas flow path and the additional flow of helium gas is not supplied.

The main properties of high sensitivity and tunable selectivity, together with the further investigation of its applicability, promise significant applications of the new SID in analytical chemistry. ACKNOWLEDGMENT We thank H. Arimoto of Shimadzu Corp. for his contribution to the measurement of the GC-SID chromatogram. This work was supported in part by the Ministry of Education, Science and Culture of Japan, Grants-in-Aid for General Scientific Research No. 04804033 and No. 07804050. The authors are grateful to Tom McMahon at the Alabama Language Academy for manuscript preparations. Received for review January 11, 1996. Accepted June 10, 1996.X AC960028M X

Abstract published in Advance ACS Abstracts, July 15, 1996.

Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

2781