502
Anal. Chem. 1986, 58,502-504
Dual Phase Gas Diffusion Flow Injection Analysis/Hydride Generation Atomic Absorption Spectrometry Sir: The utilization of gas diffusion flow injection analysis (GD-FIA) has been demonstrated as an effective technique to increase the selectivity and sensitivity of an analytical method (1-4). In conventional GD-FIA a liquid phase donor stream that contains the sample is passed under another liquid phase, the acceptor stream, which either contains reagent or will merge with the reagent downstream. The phases are separated by a microporous membrane. In this configuration and under the conditions used, only gases diffuse through the membrane of the gas diffusion cell. Even the rates that the different gases diffuse are dependent on the membrane and the physical properties of the gases. These differences can be exploited to increase selectivity (1-3). There are chemistries where the room-temperature gasphase reactions are more suited for analytical measurements than liquid-phase reactions. For example the chemiluminescent gas-phase reaction of ozone with nitric oxide has better sensitivity than the luminol reaction with ozone in the liquid phase. In order to take advantage of the increases in sensitivities and retain the enhanced selectivity character observed for gas diffusion systems, a dual phase gas diffusion system was developed. The technique of dual phase gas diffusion, like conventional GD-FIA, utilizes a liquid phase donor stream that contains the sample and passes under the membrane. However, in dual phase GD-FIA, a stream of gas flows on the other side of the membrane. The gaseous analyte of interest passes into the gas phase side of the membrane and is transported to the chemistry and/or detector. The concept of a gas phase carrier in FIA has been demonstrated (5); however, no FIA system has been designed to permit the separation of a gaseous analyte from solution to a gas phase carrier for subsequent detection in the gas phase. The chemical system to be tested in this paper is the generation of hydrides for their subsequent determination by atomic absorption spectrometry. Hydride generation as a sampling technique for atomic spectrometry has become a well-established method for the determination of volatile hydride forming elements: selenium, tellerium, arsenic, antimony, tin, germanium, lead, and bismuth. Hydride generation provides a significant increase in sensitivity over traditional direct solution nebulization methods. However, the hydride generation technique is not without interferences, and recent studies have focused on increasing sensitivity by improving selectivity. Hydride generation interferences are divided into two categories, kinetic interferences in the liquid phase and atomizer interferences in the gaseous phase. Significant reductions in atomization interferences have been demonstrated by Dedina ( 6 ) through the use of a hydrogen/oxygen “flame in a tube” quartz atomizer as compared to an electrically heated quartz atomization cell. Welz and Melcher (7-9)have investigated liquid-phase transition-metal interferences in hydride generation and developed a number of chemical approaches for decreasing these interferences. However, little work has been done with instrumental approaches for reducing liquid-phase interference. Because the hydride forming reaction is quite fast, continuous flow techniques have been used to rapidly generate hydrides and remove them from the liquid phase, thereby kinetically discriminating the analyte from the slower interfering reactions (10-13). Typically the hydride gas removal has been achieved with a conventional reaction flask gaslliquid separator. The purpose of this communication is to demonstrate the 0003-2700/86/0358-0502$0 1.50/0
dual phase gas diffusion FIA cell and to evaluate its enhancement of selectivities for the hydride-forming elements being tested.
EXPERIMENTAL SECTION Apparatus. The basic FIA manifold is presented in Figure 1. Reagents were pumped through 0.5 mm i.d. tubing made of Teflon using a Tecator, Inc., Model 5020 flow injection analyzer equipped with poly(viny1 chloride) peristaltic pump tubes. The dual phase gas diffusion cell was constructed from a standard gas diffusion cell (Tecator AB, Hoganas, Sweden) with a diffusion area 70 mm long by 1 mm wide fitted with a 0.45 pm pore size Teflon membrane (W. L. Gore & Associates, Inc., Elkton, MD) and nylon mesh membrane support backing. Measurements were carried out with a Perkin-Elmer 560 atomic absorption spectrophotometer fitted with a specially constructed Dedina quartz “flame in a tube” atomization cell (5). The flow of oxygen, which entered the bottom of the T-shaped cell through 2 mm i.d. Tygon tubing, and the flow of hdrogen/hydride, which entered the side of the cell through 0.5 mm i.d. Teflon tubing, were controlled with two Matheson Series 7630 flowmeters. A Perkin-Elmer arsenic hollow cathode lamp was the atomic absorption light source. Operating parameters for the atomic absorption spectrophotometer included a wavelength of 193.7 nm, with a 0.7 nm spectral bandwidth and a 20-mA lamp current. Samples were quantitated by peak heights as recorded on a Fisher Recordall strip chart recorder. A second FIA manifold, which was used to produce matched As(II1) and As(V) sensitivities, is presented in Figure 2. Reagents. Arsenic(II1) stock solution, 1000 mg/L, was prepared by dissolving primary standard arsenic trioxide, Asz03 (Matheson, Coleman & Bell, Norwood, OH), in 25 mL of 20% (w/v) potassium hydroxide solution. This solution was then neutralized to a phenolphthalein end point with 20% (v/v) sulfuric acid and diluted to 1L with 1%(v/v) sulfuric acid. Arsenic(II1) standard solutions were prepared through appropriate dilutions of stock solution with 1%(v/v) sulfuric acid. Arsenic(V) stock solution, 1000 mg/L, was prepared by dissolving sodium arsenate, NazHAs04.7Hz0 (Matheson, Coleman & Bell), in deionized, doubly distilled Barnstead-Nanopure water and diluting to 1L with 1%(v/v) sulfuric acid solution. Arsenic(V) standard solutions were prepared through appropriate dilutions of stock solution with distilled water. Sodium tetrahydroborate(II1)solution, 2% (w/v),was prepared daily by dissolving sodium tetrahydroborate(III), NaBH4 (Matheson, Coleman & Bell), in 1%(w/v) potassium hydroxide solution. The 50% (w/w) potassiwq iodide solution was prepared by dissolving KI in distilled water. Fe(II), Fe(III), Cu(II), Ni(II), and Co(I1) stock solutions, 2000 mg/L, were prepared for interference studies by dissolving in distilled water Fe(NH4),(S04)z.6H20 (Matheson, Coleman & Bell), FeNH4(S04)z.7Hz0 (Matheson, Coleman & Bell), CuSO4.5HZO(Fisher Scientific Co., Fairlawn, NJ), NiCl2.6Hz0 (Matheson, Coleman & Bell), and CoS04.7Hz0 (Matheson, Coleman & Bell), respectively. Sb(III), Bi(III), Te(IV),and Se(IV), 1000 mg/L, and Sn(II), 2000 mg/L, stock solutions were prepared for interference studies by dissolving in 2 M hydrochloric acid SbC1, (Alfa Products, Danvers, MA), Bi(N03),.5H20 (Fisher Scientific Co.), TeOz (Aldrich Chemical Co., Milwaukee, WI), NazSe03 (Alfa Products), and SnClz (Matheson, Coleman & Bell), respectively.
RESULTS AND DISCUSSION The first question to be answered was whether the dual phase gas diffusion cell would perform with good membrane stability. Since the membrane acts as a barrier between the liquid flow channel and the gas flow channel, it is subjected to the pressure difference in the two phases. The result was that unbacked or nonsupported membranes stretched into a flow channel, which in this case significantly reduced or 0 1986 Amerlcan Chemical Soclety
ANALYTICAL CHEMISTRY, VOL. 58, NO. 2, FEBRUhRY 1986
503
Table I. Comparison of Permissible Levels" of Interfering Ions in Arsenic Hydride Generation by Other Flow Techniques vs. Dual Phase Gas Diffusion Cell
continuous continuous/segmented flow method KI added
AA
Waste
Figure 1. FIA manifold for the determination of arsenic(II1) using dual phase gas diffusion.
Dual Pnase Gas D I I I U S I O ~
NaBH, 11
"
mllmin
AA
25
Waste
zoo P L
Figure 2. FIA manifold for the determination of arsenic(II1) and -(V)
using dual phase gas diffusion.
blocked the hydrogen gas flow. It was obvious that some type of membrane support was necessary to increase stability and lifetime. Several types of supported membranes and support backings placed on the gas stream side of the membrane were tested in the dual phase gas diffusion cell. Teflon membranes supported by nonwoven polypropylene scrims were not tested due to previous problems with solution leakage. The addition of other backings such as polyethylene films or wax-impregnated cellulose did not significantly increase membrane life. Perforated aluminum foil provided good support for the membrane and extended the membrane lifetimes; however, the Teflon membranes were eventually torn by the rough edges of the holes in the aluminum foil. The best support was a nylon mesh (289 squares/in), which gave the most reliable maintenance of the gas flow and exhibited good membrane lifetimes. Optimization of several parameters was carried out. For some of these parameters the conditions selected are compromises to allow the atomizer and the FIA system to be interfaced. The gas flow rates were optimized for Hz flow at 70 mL/min and O2flow at 10 mL/min. An HC1 concentration of 9 M pumped at 2.5 mL/min provided the best signal for the hydride. The order of mixing and coil lengths in Figures 1 and 2 produced the largest signals. A sample size of 200 pL assured that the dual phase gas diffusion cell was completely filled with sample. This system is quite safe despite the fact that 9 M HCl and poisonous arsenic compounds are involved, since, as in any typical FIA system, the reagents, samples, carriers, and reaction products are handled in a closed system. The optimized system exhibited a detection limit ( S I N = 3) of 10 pg/L with a sensitivity of 0.0015 A/(pg/L). A linear working range was found to be 10-160 pg/L. Based on the time required for a single analysis, a sampling rate of 180 injections/h was achieved with a relative standard deviation of 0.8% for replicate injections. At this point of development, the observed detection limit is about a factor of 25 above the most recent segmented flow system (IO) and a factor of 100 above the most recent continuous flow system (11). However, the actual amount of arsenic detected was 200 pg for the segmented flow system (IO) and 24 pg for the continuous flow system (11) as compared to 170 pg for the dual phase gas diffusion system. The efficiency of the transfer in the dual
ion Fe(I1) Cu(I1) Ni(I1) Co(I1) Sn(I1) Bi(II1) Sb(II1) Te(1V) Se(1V)
(11)
(10)
dual phase gas diffusion
1000
>10 000 >10000
>10000 >10000
>10000 >10000
>10000
3000 20 100
100
200 100
5
100
300 500
50 10
10
2 500
1250 25 5
a Permissible amounts (weight ratio) correspond to the concentrations that give 10% or less negative error.
phase gas diffusion system is not optimized. Increases in the size of the channel in order to maximize membrane surface area and to allow for larger sample sizes (the current channel is 70 pL) should greatly enhance signal. Additional decreases in detection limit may be realized by utilizing stopped flow. The real asset exhibited by the dual phase gas diffusion system was the decreased effect of interferences. Table I compares the results for the two most recent continuous flow systems vs. the dual phase gas diffusion system ( 1 0 , I I ) . It is evident that when compared to the continuous flow system, the dual phase gas diffusion system exhibits significantly decreased transition-metal interferences. The selectivities achieved with the continuous flow system when both KI and ",OH were added to suppress interference were comparable to those exhibited by the dual phase GD-FIA manifold. The segmented system, which included KI, exhibited transitionmetal interferences comparable to those demonstrated by the dual phase system. Previous work has established that the effect of the addition of KI or KI/NHzOH to the arsenic method results in signficant reductinn in the transition-metal interferences, typically by a factor of 10. Given that the selectivities reported for the dual phase system in Table I were determined without KI or KI/NH,OH, it is reasonable to expect that the dual phase system is capable of further improvement in selectivity over the other two systems. In addition the dual phase gas diffusion system is chemically and operationally simpler. The arsenic determination had additional interferences from the other hydride forming elements. Table I shows that dual phase gas diffusion when compared to the segmented flow system exhibited improved selectivity except for selenium(IV). The comparison between the continuous flow system and the dual phase gas diffusion system is a curious mixture of decreases and increases in interference levels. The explanation for this behavior is unknown. However, it is interesting to point out that the interference from other hydride forming ions decreased in the dual phase gas diffusion system as the samples approached the detection limit. Increased selectivities for arsenic over the other hydride forming elements is expected for the dual phase system through the use of KI or KI/ NHZOH, as was the case for selectivity over transition metals. However, unlike the transition-metal case, the expected increase in selectivity over other hydride-forming elements is typically only as much as a factor of 4. Despite this smaller increase in selectivity, it is evident that the potential exists for significant improvement of the dual phase system over the other systems. The primary reason for the addition of KI was to produce equal signals for the same metal in different oxidation states ' (i.e., As(II1) and As(V)). The manifold in Figure 2 incorporating KI produced equal signals for the same concentration
504
Anal. Chem. 1986, 58,504-505
of As(II1) and As(V). Both manifolds produced the same signal for a given As(II1) concentration. The selectivity increases expected, as described above. for the addition of KI in the second dual phase manifold were observed. Further increases in selectivity through the use of KI/NH20H remain to be investigated.
(9) Welz, B.;Melcher, M. Analyst (London) 1984, 109, 577. (10) Narasaki, H.; Ikeda, M. Anal. Chem. 1984, 56, 2059. (11) Yamamoto, M.; Yasuda, M.; Yamamoto, Y. Anal. Chem. 1985, 5 7 , IRR9
ikeda,
(12) M. Anal. Chim. Acta 1985, 167,289. (13) Astrom, 0. Anal. Chem. 1082, 5 4 , 190.
G. E. Pacey* M. R. Straka J. R. Gord
LITERATURE CITED Straka, M. R.; Pacey, G. E.; Gordon, G. Anal. Chem. 1984, 56, 1973. Straka, M. R.; Gordon, G.; Pacey, G. E. Anal. Chem. 1985, 57, 1799. Hollowell, D. A,; Pacey, G. E.; Gordon, G. Anal. Chem. 1985, 57, 285 1. Van der Linden, W. E. Anal. Chlm. Acta 1983, 151,359. Attlyat, A. S.; Chrlstlan, G. D. Talanta 1985, 3 1 , 463. Dedina, J. Anal. Chem. 1982, 54,2097. Welz, B.;Melcher. M. Ana/yst (London) 1984, 109, 569. Welz, B.; Melcher, M. Analyst (London) 1984, 109, 575.
Department of Chemistry Miami University Oxford, Ohio 45056
RECEIVED for review September 19,1985. Accepted November 21, 1985.
AIDS FOR ANALYTICAL CHEMISTS Slmple Nanoliter Refractive Index Detector Darryl J. Bornhop and Norman J. Dovichi*
Department of Chemistry, University of Wyoming, Laramie, Wyoming 82071 Lasers are convenient light sources for optical methods of analysis. I s particular, the spatial coherence of the laser is useful when probing small volume samples. For example, probe volumes in the nanoliter range have been reported for fluorescence (1-41, absorbance (5-9), and light scatter ( 1 0 , l l ) measurements. The push for small-volume detectors comes from the development of narrow-bore liquid chromatography (12-16), where submicroliter samples must be detected to achieve the full separation power of liquid chromatography. Further interest in small-volume optical techniques arises from biological analysis of small amounts of rare substances (17). The refractive index detector is one of the few universal optical detectors available for liquid samples. Unfortunately, the sensitivity of refractive index measurements is inherently low; the refractive indexes of most solutes differ only in the second or third decimal place. It is not suprising that state of the art refractive index detectors tend to be rather complex. For example, detection limits of An = 1.5 X have been reported using an interferometric measurement of optical path length, proportional to the refractive index of the solute (18). Unfortunately, the high-quality Fabry-Perot interferometer used in this refractive index detector is quite expensive. Furthermore, the interferometric refractive index design produces a signal that is proportional to path length. To achieve the excellent detection limits, a 10-cm path length and 2OO-pL probed volume was required. Scaling to submicroliter volume would require a significant decrease in path length, producing a proportional decrease in sensitivity. In this paper, we wish to describe a simple, inexpensive, and low-volume refractive index detector. This detector is based on the propagation of a beam of light through a liquid-filled tube as shown in Figure 1. The fluid-filled tube may be considered by analogy to a lens; an off-axis ray passing through a lens is both defocused and deflected by the lens. The deflection and defocusing is quantitated as a change in the laser beam intensity measured by a small-area photodiode located near the inflection point of the beam profile. These measurements are taken relatively far from the cuvette. Of course, the small diameter of the cuvette suggests that a simple model based upon a lenslike behavior is a gross 0003-2700/86/0358-0504$01.50/0
simplification. Diffraction and reflection from the edges of the tube and aberrations induced by the tube will act to distort the laser beam profile. Several models exist for the propagation of a light beam through a pair of concentric tubes with differing refractive index. These models arise in the determination of the refractive index profile of a clad optical fiber or single mode preform (1S22). We note that Jorgenson has proposed an on-column detector for capillary liquid chromatography that is based upon the scattering of light by the fluid-filled capillary (14).
EXPERIMENTAL SECTION Optical System. A schematic of the optical system is shown in Figure 2. The system is constructure on an optical table, NRC Model KST-48. A 5-mW helium neon laser, Melles Griot Model 05LHP151, provides a linearly polarized beam at 632.8 nm. A microscope slide is used to split a small portion of the beam to a reference detector, described below. The main beam is focused by a 16-mm focal-length microscope objective, Melles Griot, into the round Pyrex capillary sample cell, 0.5 mm i.d., 0.7 mm o.d., Wale Apparatus. The tube is mounted on two stacked translation stages that provided sample positioning both along and perpendicular to the laser beam. Retroreflections from the tube can produce unwanted feedback into the laser resonator. We tilt the tube by about 20° to eliminate these retroreflections. This laser produces very low drift in intensity. However, detecting very small reflective index changes requires very stable intensity measurements. To reduce long-term, low-frequencydrift, we utilize a double beam in space optical design (23). The reference and signal detectors are identical 1-mm2silicon photodiodes, Silicon Detector Model SD 041-11-11-011. The output of the photodiodes is conditioned with matched current-to-voltage converters, consisting of a JFET operational amplifier, Linear Devices LF 351, wired with a 1-Ma feedback resistor in parallel with a 47-pF capacitor. The signal and the reference photodiode outputs are subtracted with an instrumentation amplifier, Analog Devices Model AD 524. This circuit allowed for the removal of common-mode intensity fluctuation in the beam, primarily line frequency noise, in addition to providing 10 times signal gain. A piece of frosted glass is located before the reference detector. Translation of the frosted glass is a convenient way to adjust the intensity of light reaching the reference detector, which facilitates nulling of the dc signal component. The signal is next sent to 0 1986 American Chemlcal Society