Anal. Chem. 1990, 62,2506-2509
2508
DRYING NMP BY MOLECULAR SIEVE
the reagent strength' is diminishing with increasing sample volume and sample water content. To maximize the number of analyses for each reagent charge, one should choose the smallest sample loop size applicable to the sample water content, the desired accuracy, and the K-F titrator used.
ACKNOWLEDGMENT I acknowledge Reid Willis for preliminary work and Nathan u
Haese, Tim Stevens, and Nile Frawley for helpful discwions.
140 120
LITERATURE CITED
100 60 40
20 0 -10
10
2b
3b
4b
T I M E , HOUR Figure 3. On-line monitoring of the water content in N-methyl-2pyrrolidone. The molecular sieves were added at time 0. A 103-pL sample was titrated for water every 30 min.
capacity with K-F reagent. Two reasons make this practice possible: one the active reagent, iodine, is generated coulometrically in situ and does not have to be added, two, for every sample added to the cell, an equal volume of the liquid in the cell is flushed away. By the conventional method, the sample capacity or the number of analyses is limited by the head space of the cell. For example, the sample capacity cannot exceed 50 mL if 150 mL of K-F reagent is charged into a 2WmL cell; thus, one can expect a maximum of 50 determinations for a 1.0-mL sample size. In comparison,by the loop sampling, 200 mL of the K-F reagent can be charged into the 200-mL cell and more than 50 mL of sample can be added. Clearly the number of analyses also depends on the reagent strength, and
(1) Scholr, E. Karl Fischer Titration; Springer-Verlag: Berlin, M b e r g , Germany, New York, 1984. (2) HYDfANALRManual-Eugen schab R e e w t s fw Karl F k d w l l h tion; Riedekle Haen Laboratory Chemicals or Crescent Chemicals Co., Inc.: Hauppauge, New York, 1987. (3) Kagevall; Aastrom, 0.; Cedergren, A. Anal. Chim. Acta 1980, 114, 199-208. (4) Escott, R. E. A.; Taylor, A. F. Analyst 1985, 110, 847-9. (5) Liang. C.; Vacha, P.; Van der Llnden, W. E. Talanta 1388, 35 (l), 59-61. (8) Nordin-Andersson. I.; Cedergn, A. Anal. Chem. 1985, 57 (13), 2571-5. (7) Nordin-Andersson, 1.; Aastroem, 0.; Cedergren, A. Anal. Chlm. Acta 1984. 162, 9-18. (8) Spohn, U.; Hahn, M.; Ruettinger, H. H.; Mltschlner, H. Fresenhs' Z .
Anal. Chem. 1989, 333(1),39-41; Chem. Abstr. 1989, 110, 79-2. (9) SchneMer, W.; Schalch, E.:Walther, R. Am. Lab. 1988, 20 (2),136, 138,140-1. (10) Mottola, H. A. Anal. Chem. 1981, 53(12), 1312-6A. (11) Mottola, H. A.; Hanna, A. Anal. Chim. Acta 1978, 100, 167-80.
RECEIVED for review April 30,1990. Accepted August 2,1990. Thanks are given to The Dow Chemical Co. for support and permission to publish this work. A portion of the work was presented on January 4, 1990, at the second annual winter conference on flow injection analysis in Orlando, FL. A patent application (U.S. 317879, March 2,1989) has been submitted for this work.
Laser-Enhanced Ionization Spectroscopy in an Extended Inductively Coupled Plasma Kin C. Ng,' Martin J. Angebranndt, and James D. Winefordner* Department of Chemistry, University of Florida, Gainesuille, Florida 32611 INTRODUCTION Laser-enhanced ionization (LEI) spectroscopy has been proven to be a very sensitive trace element technique. Excellent reviews on LEI are provided by Travis and co-workers (1, 21, by Green (31, and recently by Axner and Rubinsztein-Dunlop (4). The technique involves selective photoexcitation of analyte atoms by laser radiation, followed by collisional ionization of the excited atoms in the atom reservoir. The ion increase is detected by measuring the current change at an electrode and is proportional to the analyte concentration. The electrode designs and arrangements in atom reservoirs have been reviewed (1-4). Since LEI detection is nonoptical, problems associated with laser radiation scattering and light collection efficiency are not important, and the possibility of nearly 100%ion collection efficiency allows the detection of very low analyte concentrations. The detection limits of LEI in flames are comparable to those obtained with graphite furnace atomic absorption spectroscopy (4). Flames are inexpensive, simple, reproducible, and readily available in most laboratories and provide an excellent collisional thermal medium; flames are most commonly used as
* To whom corres
ondence should be sent.
On leave from gepartment o f Chemistry, California State University at Fresno, Fresno, CA 93740-0070. 0003-2700/90/0362-2506$02.50/0
atom reservoirs in LEI experiments. Standard burners used in analytical flame atomic spectroscopyare employed. These include the premixed 5-cm slot burner (51,the premixed 10-cm slot burner (6),the premixed capillary burner (3,and the total consumption burner (8). These burners are generally equipped with a pneumatic nebulizer, which transports a sample solution into the flame in the form of small droplets. The ability for a flame to produce free sample atoms depends on the fuel and oxidant used. The acetylene-air flame is the choice for most elements (4). The cooler hydrogen-air flame is useful for elements with low ionization potentials, such as rubidium. The hot acetylene-nitrous oxide flame is effective for refractory elements such as vanadium and titanium. The combustion products of flames, however, may hinder some of the spectral regions for successful implementation of LEI. Nonflame atomizers such as the graphite furnace also have been used (9).
The inductively coupled plasma (ICP) has long enjoyed popularity as an efficient atom reservoir for atomic emission (10) and atomic fluorescence (AF) (11) spectroscopy. Its effectiveness is partially attributed to the high temperature inert argon atmosphere. The ICP also is used as an ion source for mass spectrometry (12). The low detection limits (11) obtained with ICP-AFS excited by hollow cathode lamps, lasers and ICP have led one to believe there is a high popu0 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 22, NOVEMBER 15, 1990
2507
Table I. Experimental Components and Manufacturers components/model lasers argon ion (2040) dye lasers Coherent (590) SP (375) inductively coupled plasma rf generator (2500D) sample introduction system Concentric nebulizer Heating chamber, conical (Pyrex 3.5 cm 0.d. x 35 cm) heating supply (Powerstat) condenser (32cm long) electrodes voltage supply (PAR 280) anode/cathode Stainless steel tubing (0.635cm 0.d.) detection electronics lock-in amplifier (186A) chopper motor (frequency generator) current amplifier (427) chart recorder (5000) oscilloscope (V-152B)
AMPLIFIER
manufacturer Spectra Physics, Mountain View, CA
CURRENT AMPLIFIER
Coherent Radiation, Palo Alto, CA Spectra Physics, Mountain View, CA Plasma-Therm,Kresson, NJ J. E. Meinhard Associates, Santa Ana, CA
lab constructed
Fisher Scientific, Pittsburgh, PA lab constructed EG&G Princeton Applied Research, Princeton, N J lab constructed EG&G Princeton Applied Research, Princeton, N J Photon Technology, Princeton, N J Keithley Instruments, Inc., Cleveland, OH Fisher Scientific, Pittsburgh, PA Hitachi Denshi, Ltd., Woodbury, NY
NEBULIZER
HEATER
OMDENSER
Flgure 1. Block diagram of the LEI-ICP experimental setup.
Table 11. Operation Conditions Lasers dye laser output: 0.4 W at 589 nm (Na);0.3 W at 610 nm (Li); 0.13 W at 670 nm (Li); 0.15 W at 423 nm (Ca); 0.15 W at 461 nm (Sr); 0.2 W at 451 nm (In);0.12 W at 417 nm (Ga) inductively coupled plasma coolant argon, 13 L/min injector argon, 3.75 L/min nebulization (auxilialry/intermediate) argon, 0.45 L/min at 15 psi forward power, 1.2 kW reflected power, 3-20 W electrode applied voltage, -0.8 to -1.1 kV
lation of low-energy-level atoms available in the ICP, which may be amenable for LEI. The spectral cleanliness of argon ICP may expand the useful spectral region for LEI, compared to using a flame as the atom reservoir. The only previous attempt to use ICP for LEI was by Turk and Watters (13). Interaction between the radio frequency plasma and the detection circuitry required that an extended ICP torch be used, and the electrodes had to be placed 19 cm above the load coil. Detection limits a t the low micrograms per milliliter (parts per million) level were obtained. Long and Winefordner (14) and Demers (15) have used an extended torch to produce a “pencil” ICP for AFS. This pencil ICP was generated with the injector tube placed between coils and with a tangential argon flow of 10 L/min and an injedor argon flow of -3 L/min. We have modified the torch-coil arrangement and operation argon flow rates for generating an extended ICP. We report here the results of using this plasma for LEI. ‘Samples in the form of dried aerosols are introduced into the plasma through the intermediate gas inlet of the torch. A continuous wave (CW) argon ion laser pumped dye laser is used in the investigation. The detection limits obtained are improved over those of Turk and Watters, who used a Nd:YAG laser pumped dye laser, for their ICP-LEI system (13).
-
EXPERIMENTAL SECTION Instrumentation. The instrumental setup is shown in Figure 1 and the component manufacturers are listed in Table I. The outer tube of the ICP torch extended 6 cm from the tip of the injector tube. A three-turn copper coil was used with a coil spacing equal to the tubing thickness. The torch was arranged such that the inner tubes were 1-2 mm below the load-coil. The stainless steel (0.635 cm 0.d. X 5 cm) electrodeswere slightly flattened and bent, separated by 7 mm, tap-water cooled, and placed on either side of the plasma. The laser beam ( - 3 mm2)was directed to pass parallel, 1 mm away from the collection anode. Ice-chilled
-
detection electronics chopped frequency, 1.8 kHz lock-in amplifier, 3-stime constant, 10-kHz filter water was used to cool the condenser and the ICP coil. The desolvation chamber was heated electrically with heating tape. The plasma torch was enclosed in an aluminum box. The entire plasma enclosure and the impedance matching box were wrapped with copper cloth which was subsequently grounded. The anode cable was also wrapped with grounding wire. The shielding and grounding greatly reduced the radio-frequencyand environmental noise picked up by the detection electronics. It should be noted that care must be exercised not to allow the electrode or the plasma to touch the metal shield box or copper cloth, otherwise electrical arching can occur and high voltage (current) can conduct to optical table, resulting in electrical shock and/or possibly damage to instrumentation. Proper grounding and shielding of the plasma and the electrode will reduce the rf leakage. Procedure. The sample heater-condenserunit was turned on and the nebulization gas was allowed to flow through the unit and into the ICP torch for at least 5 min. This procedure allowed removal of air trapped in the apparatus and establishment of heating and gas flow rate steady state. The plasma was ignited with 21 L/min coolant flow, 3.75 L/min injector flow, and 1.25 kW forward power. The plasma was then adjusted to the operating condition (Table 11) by slowly reducing the tangential argon flow rate and increasing the nebulization argon pressure. The detection electronics operating condition is also listed in Table 11. The dye laser radiation was wavelength tuned by scattering laser radiation to a monochromator;it was fine tuned by observing and optimizing the ionization signal on an oscilloscope and a digital voltmeter. The signal output from the lock-in amplifier was monitored with a strip-chart recorder and was averaged for approximately 1 min. Chemicals. Standard solutionswere prepared by serial dilution of the stock standard (Venture Inorganics, Inc., Lakewood, NJ) with deionized distilled water. The laser dyes (Exciton Chemical Co., Inc., Dayton, OH) used were Rhodamine 6 G for Na and Li,
2508
ANALYTICAL CHEMISTRY, VOL. 62, NO. 22, NOVEMBER 15, 1990
DCM for Li, and Stilbene 420 for Ca, Sr, In, and Ga. RESULTS AND DISCUSSION Operation Conditions. Plasma. When the extended torch was arranged conventionally with the load-coil (i.e. the inner tubes located approximately 1-2 mm below the lowest coil), an extended plasma could easily be produced by reducing the tangential argon flow and increasing the injector flow. The length of the plasma depended on the injector argon flow rate. The higher the injector argon flow rate, the longer the plasma. With the plasma argon flows (13 L/min coolant, 3.75 L/min injector) selected, the plasma extended approximately 27-30 cm above the load-coil. A t low heights (0-12 cm above the load-coil), the plasma had channels characterized by the different gas inlets. In the plasma tail region, however, the plasma appeared homogeneous with complete mixing of argon and air entrained from the atmosphere. Only the "homogeneous" tail region was used in this investigation. The optimum signal was found at 17 cm above the load-coil. The plasma was extinguished with an observation height less than 15 cm due to interaction between the load-coil and the electrodes. Sample Introduction. Initially, a pencil plasma was produced according to the condition of Long and Winefordner (14),i.e. by placing the torch such that the inner tubes were between load-coils. However, the high injector argon flow (3-4 L/min) required use of a compatible nebulizer, which was unavailable, and a drain system that could withstand the high back-pressure due to the small injector orifice of the torch. Long and Winefordner (14) used a 3.9 L/min Meinhard nebulizer to create a pencil plasma, and found a poorer analytical sensitivity compared with a convential ICP/nebulizer system in ICP excited ICP for AFS studies. They attributed this to a lower sample transport efficiency of the 3.9 L/min nebulizer. The high flow rate (3.9 L/min) occurred because of a large nebulizer orifice, which produced large-sized droplets (14); these large droplets resulted in reduced efficiency of atomization, producing decreased analytical sensitivity. We decided to introduce the sample through the intermediate tube (auxiliary gas inlet of the torch) to avoid the need for a high gas flow nebulizer and the difficulty with drain pressure. Other considerations were also involved in this decision: (1) the LEI technique is not based on optical detection and that the analytes are not required to be confined in the central portion of the plasma; (2) the tail plasma region (>13 cm above the load-coil) appeared as a homogeneous mix of argon (from the different gas inlets into the torch/plasma) and entrained air. This mixing suggests it might be of little difference if the sample is introduced into the injector tube or the intermediate tube, since the analytes in the plasma are probed >13 cm above the load-coil. By introducing samples into the intermediate tube and with the inner tubes arranged between load-coils (14) (i.e. a pencil plasma), we observed memory effects. The memory was a result of deposition of sample particles on the hot end of the inner tubes of the ICP torch. The injector tube was also found to degrade progressively. Demers (15) used a replaceable central tube tip to minimize this problem. We decided to lower the torch such that the inner tubes of the torch were 1-2 mm below the lowest coil; the memory effects were removed. This plasma/torch arrangement was used for all subsequent experiments. Applied Voltage. The highest voltage applicable before arcing between the electrodes occurred was dependent on the plasma operation power, height of the electrodes in the plasma, and the position of the plasma relative to the electrodes. The voltage applied was lower for higher plasma powers, for shorter distances between the electrodes and the load-coil, and for closer positioning of the plasma to the high-voltage eledrode. A typical voltage of -0.8 to -1.1 kV was used for the optimized
Table 111. Detection Limits (3u of Bfank) for the Laser-Enhanced Ionization in the ICP
element
wavelength, nm
detection limit, pg/mL
Na Li Li Ca
589.0 67018 610.3 422.7 460.7 451.1 417.2
0.03 0.18 0.92 0.03 0.81 0.22 0.20
Sr In Ga 4e+0
1
d
m l
Na 589nm
Z 3e+0
Oe+O, -2
'
I
-1
'
I
0
'
1
1
//
I
Li 671 nm
I
I
2
3
Log Conc. PPH Flgure 2. Calibration curves of the LEI-ICP technique.
operation condition employed (Table 11). It was found that when the plasma was positioned beneath the "cathode" electrode (-0.8 to -1.0 kV), arcing between the electrodes occurred. The best plasma position was between the electrodes, with the plasma positioned closer to the signal electrode (anode). The present electrode design did not allow change of the distance between electrodes, and this parameter should be optimizable. Analytical Figures of Merit. Table I11 shows the detection limits of this system. The sodium value (Na is the only common element investigated; using the same line) is approximately 2 orders of magnitude superior to that of Turk and Watters (13),who used a Nd-YAG pumped laser induced ionization-ICP system. In the Turk and Watters system, detection limits for Cu, Fe, Mn, and Na were a t the low parts-per-million level. It should be mentioned here that the electrodes (0.635 cm 0.d. X 5 cm long) used in this system are dimensionally much smaller than those (2.3 cm X 15 cm) used by Turk and Watters; the smaller electrode dimensions might minimize noise pick-up, improving the signal-to-noise ratio. The background noise in this ICP-LEI system is at the subnanovolt level. Furthermore, the electrodes in this system were 0.7 cm apart, compared with those of Turk and Watters (131, which were separated by 2.1 cm; the smaller separation might be more efficient for analyte probing in the plasma. The higher dye laser power produced a higher ionization signal. For example, with a 90-mW dye laser, the detection limit for Ca was 0.06 ppm; with a 150-mW dye laser, the detection limit was improved to 0.03 ppm, because the laser irradiance was closer to saturation of the atomic transition. Use of a higher irradiance dye laser should improve the detection power even more. Figure 2 shows the linearities for Na and Li, indicating an approximately 3 orders of magnitude linear dynamic range. The slopes ranged between 0.92 and 1.1. Concentrationsabove 100 ppm produced a positive deviation of the calibration lines; the reason for this is not known at this time. In the case for Na, the lo00 ppm solution extinguished the plasma, probably due to ion (sodium) loading in the plasma which caused an increased plasma-electrode interaction (arching). Figure 3 shows a representative signal (Ca) of the ICP-LEI system. The relative standard deviation (RSD) for the 10 ppm solution of Ca was 2.8%; the RSD for the 250 ppm solution
Anal. Chem. 1990, 62, 2509-2512
E
2509
mechanisms involved are subject to further investigation. CONCLUSIONS We have reevaluated the analytical applicability of ICP-LEI. The results have been improved over the previous investigation (13), although the figures of merit are rather poor compared with other LEI systems (1, 3, 4). Additional progress is still needed to justify the cost of the ICP-LEI system and the complexity of the measurement. To improve the analytical performance of the present system, considerations should include the following: (a) use a higher peak irradiance pulsed laser to saturate the atomic transition, allowing a maximum number density of atoms available for ionization, improving the signal intensity and the signal precision; (b) shield the long plasma from moving air and ground and filter more effectively to minimize picking up the rf and environmental noise, thus reducing the background noise and improving the signal precision; and (c) construct specialized ICP torches and electrodes and use for optimizing the LEI approach.
A Flguro 3. Representative signal of the LEI-ICP system. The x axis is relative signal intensity and the y axis is tlme with the scale indicated as 1 min. A is for deionized dlstilled water; B is for 10 ppm Ca; and C is for 5 ppm Ca. E is the result of sample surge during sample
changing.
was 4.6%. The poorer precision obtained with the higher concentration may have been a result of the increased instability of the plasma. A higher number density of (sample) ions can increase the electron conductivity of the plasma, which will lead to a higher tendency to arc (interactions between electrodes and between the cathode and the load-coil), reducing the plasma stability; the higher dc background also adds shot noise to the detection system. Ionization Processes. The temperature of the pencil plasma (0.75 kW) measured by Long and Bolton (16)6 cm above the load-coil was about 3290 K, whereas Barnes and Schleicher (17) suggested a temperature of 2000 and 3000 K in the region well above the plasma. In the tail region of the extended plasma, a significant fraction of the analytes exists as ground-state atoms. Since the CW dye laser radiation used in this experiment does not supply sufficient energy to saturate atomic transitions, it is less likely the laser radiation is able to ionize the laser-excited atoms. The ionization of the laser-excited atoms should be dominantly a result of collisional processes. For LEI in flames, the ionization of excited atoms is usually the result of collisions between excited atoms and thermally excited molecules such as nitrogen and free oxygen. An enhanced transfer rate of excited Na atoms to the continuum state with O2was discussed by van Dijk (18). In our case, air has been entrained into the extended argon plasma; therefore metastable argon, metastable nitrogen, and free oxygen may participate in the ionization process. The detailed
ACKNOWLEDGMENT The authors thank Tom Manning for technical assistance. LITERATURE CITED (1) Travis, J. C.; Turk, G. C.; Green, R. 6. Anal. Chem. 1982, 54, 1006A. (2) Travis, J. C.; Turk, G. C.; DeVoe; Schenck, P. K.; van Dijk, C. A. Rog. Anal. At. Spectrosc. 1984, 7 , 199. (3) Green, R. 6. Analytical Applications of Lasers; Chemical Ana&& 87; Piepmeier, E. H., Ed.; Wiley & Sons: New York, 1986; Chapter 3. Act8 1989. 448,835. (4) Axner, 0.; RubinszteinDuniop, H. S p e d ” . (5) Havrilla, G. J.; Choi, K. J. Anal. Chem. 1988, 58. 3095. (6) Nlppolt, M. A.; Green, R. 6. Anal. Chem. 1983, 5 5 , 554. (7) Smith, 6. W.; Hart, L. P.; Omenetto, N. Anal. Chem. 1988, 58. 2147. (8) Hall, J. E.; Green, R. 6. Ana/. Chem. 1983, 55, 1811. (9) Magnusson, I.;Axner, 0.; Lindgren, I.; Rublnsztein-Dunlop, H. Appl. Spectrosc. 1988, 4 0 , 968. (10) Inductively Coupled Plasma Emission Spectroscopy, Parts I & 11. Chemkxl Analysis 90; Boumans, P. W. J. M., Ed.; John Wiley & Sons: New York, 1987. (11) Omenetto, N.; Winefordner, J. D. Inductivery Coupled “ a s k, Ana&ticalAtomic Spectroscopy; Montaser, A., &lightly, D. W., Eds.; VCH Publishers: Inc. 1987; Chapter 9. (12) Koppenall, D. W. Anal. Chem. 1988, 6 0 , 113R. (13) Turk, G. C.; Watters, R. L. Anal. Chem. 1985, 57. 1979. (14) Long, G. L.; Winefordner, J. D. Appl. Spectrosc. 1984, 38, 563. (15) Demers, D. R. Spectrmhim. Acta 1985, 408, 93. (16) Long, 0. L.; Bolton, J. S. Spectrochim. Acta 1987, 428, 581. (17) Barnes, R. M.; Schleicher, R. G. Spectrmhlm. Acta 1975. 308, 109. (18) Van Dijk, C. A. Two-photon Excitation of Higher Sodium Levels and Population Transfer in a Flame. Doctoral Thesis, Rijksunlverslteit te Utrecht, The Netherlands, 1978.
RECEIVED for review June 5,1990. Accepted August 13,1990. Research supported by NIH-5-R01-GM38434-03.
Selective Detection of Carbon-I3-Labeled Compounds by Gas Chromatography/Emision Spectroscopy Bruce D. Quimby,* P a u l C. Dryden, a n d James J. Sullivan
Hewlett-Packard Company, Route 41 and Starr Road, Avondale, Pennsylvania 19311 INTRODUCTION The stable isotope i3c is useful as a tracer in determining the fate of labeled compounds in reactive systems. Compounds to be studied are synthesized with 13Cincorporated into the stmctureat leveh higher than the natural abundance. The reaction producta are then analyzed for compounds with
* To whom correspondence should be addressed. 0003-2700/90/0362-2509$02.50/0
elevated 13C/12Cratios to find the parent compound and its reaction products. One area where this type of experiment is important is drug Gas chromatography/mass spectrometry (GC/MS) is often used to analyze for ‘%-labeled compounds. AS discussed by Chace and Abra”n (11, mass spedral techniques currently used in screening for labeled products suffer several shortcomings. Chromatographic overlap with unlabeled peaks can mask the presence of elevated 13C content. A more specific 0 1990 American Chemical Society
