ANALYTICAL CHEMISTRY, VOL. 50,
the temperature range 325 K < T < 440 K: T = 462.3 - 100.1 116D,~
(4)
with an uncertainty of ca. 1.5 K, including the uncertainty of the standard ( 2 ) . Possible systematic temperature errors arising from lock detector phase errors must be considered. If both lock-channel NMR lines in a thermometric liquid had the same widths and intensities, then any phase error or dc bias error in the lock detector would cause exactly the same deviation of the lock point from the line center in both lines; the lock-point separation would exactly equal the line separation, causing no net error. The deuteroxyl lines are weaker and broader than the aliphatic deuterium lines in both methanol-d, and ethylene glycol-d6, so both lock detector nonidealities are potential sources of error. We have found, for example, that a phase shift of a / 6 radian can cause lock-point separation changes corresponding to a few "C. In our spectrometer, we monitor the signal input to the autoshim circuit on an oscilloscope, and adjust lock detector phase to eliminate dispersion components from this signal. Since the lock signal and autoshim input are obtained from separate phase-sensitive detectors whose reference inputs might not be shifted by the ideal a/2 radians, our phase adjustment procedure guarantees reproducibility, but not necessarily accuracy. We can only conclude that any nonorthogonality and dc bias in the lock-channel detectors must be relatively small, since large errors would preclude proper operation of the autoshim circuit which is observed to function satisfactorily.
NO. 12,
OCTOBER 1978
1719
We have used this method for over a year to calibrate our temperature controller for variable temperature studies of carbon and phosphorous NMR. The controller calibration varies with probe insert, proton-decoupler, transmitter power, and cooling-gas flow rate. The incorrect assumption that the sample temperature was equal to the controller setting would have introduced systematic errors as large as 15% in a kinetic activation energies calculated from the measured temperature dependence of NMR line shapes.
ACKNOWLEDGMENT We thank F. A. L. Anet for bringing to our attention the work of Slonim et al. (8). LITERATURE CITED (1) (2) (3) (4) (5)
A. L. Van Geet, Anal. Chem., 42, 679 (1970). M.L. Kaphn, F. A. Bovey, and H.N. Cheng, Anal. Chem.,47, 1703 (1975). R. A. Newmark and R. E. Graves, J . Phys. Chem., 72, 4299 (1968). D. W. Vidrine and P. E. Peterson, Anal. Chem., 48. 1301 (1976). C. J. Jameson, A. K . Jameson, and S. M . Cohen, J . Magn. Reson., 19, 385 (1975). (6) J. Bornais and S. Brownstein, J . Magn. Reson., 29, 207 (1978). (7) D. W. Vidrine and P. E . Peterson, Anal. Chem., 5 0 , 298 (1978). (8) I. Ya. S h i m , B. M. Arshava, and V. N. Klyuchnikov, Russ. J. phys. Chem. (Engl. Trans/.), 50, 165 (1976).
RECEIVED for review May 22, 1978. Accepted ,June 26,1978. A preliminary account of this work was given as Poster Presentation A5 at the 19th Experimental NMR Conference, Blacksburg, Va., April 16-20, 1978.
Microsampling Nebulizer Technique for Premixed Flame Atomic Spectrometry R. C. Fry,' S. J. Northway, and M. B. Denton" Department of Chemistry and University Analytical Center, University of Arizona, Tucson, Arizona 8572 1
Although the development of thermal (Delves' cup, boat, etc. ( I , 2 ) ) and electrothermal (graphite furnace, carbon rod, tantalum strip, etc. (3-5)) atomization, ultrasonic nebulizers (6, 3,high solids nebulizers (a),and on-line pre-concentration (hydride generation, etc. (9, 10)) techniques may provide greatly improved performance in a variety of specialized applications, the combined premixed flame and capillary pneumatic nebulizer will probably continue for some time to be the most "purchased" and most "used" sample introduction technique for general analytical atomic spectrometry of most samples and elements. The improved operational convenience, reduced analysis time, level of precision, and low cost of the capillary nebulizer generally make it the method of choice unless the samples are too high in salt and solids content, too high in viscosity, too low in analyte concentration, or too small in volume. These extreme conditions will then dictate the trade-off of one or more desirable characteristics of the pneumatic nebulizer in favor of a different method. The atomic absorption studies of Sebastiani et al. ( I I ) , Manning (121, and Berndt and Jackwerth (13) demonstrate that samples in the range of 15-200 pL may be pipetted into nonwetted plastic sampling cones mounted on the capillary tube of a premix capillary pneumatic nebulizer yielding transient recorder responses of precision and sensitivity similar 'Present address, Department of Chemistry, Kansas State University, Manhattan, Kansas 66506.
to that achievable with normal larger (1-5 mL) samples. This is a result of the zero dead volume of such cones and the hydrophobic cone material that enable the entire sample to enter the nebulizer capillary in a single plug. The method is reported to be rapid, convenient, and less susceptible to system clogging (11) than normal steady-state aspiration techniques since less total amount of matrix is aspirated per analysis. Regardless of the sample introduction technique used for analysis, it is generally desirable to analyze digested complex materials by the method of standard additions to avoid the deleterious results of calibration curve alteration caused by a variety of matrix effects. It should be apparent that standard volumetric glassware used to make dilutions, standards, or standard additions prior to analysis, tend to require larger samples and defeat the purpose of a microsample introduction technique. Semiquantitative estimation of the analyte concentration is also desired before a standard addition determination can proceed on a sample of totally unknown analyte concentration. Further difficulties arise in that a large number of samples of totally unknown content may vary as much as four orders of magnitude in individual analyte concentration when the instrument is adjusted for a given range covering only one or two orders of magnitude simultaneously. A rather inconvenient logistics problem thereby results if large numbers of unknown small volume samples are to be analyzed. This may
0003-2700/78/0350-1719$01,00/0 (E 1978 American Chemical Society
1720
ANALYTICAL CHEMISTRY, VOL. 50, NO. 12, OCTOBER 1978 100 pL S A M P L E ,
TO
NERULIZER
p/Jr-
I O O p L SAMPLE B
wood samples in 30% hydrofluoric acid. Samples are delivered into cone A using Eppendorf pipets. A transient recorder signal results immediately. Peak height was the parameter measured. The combined instrument recorder full scale response time was -0.5 s. The nebulizer aspiration rate was -5 mL/min. Samples pipetted into cone B may be diluted or further standard additions made by pipet on-the-spot. The composite droplet is then introduced by manually inserting the small bore plastic nebulizer tubing into the cone bottom. The conical shape guides the tube directly and reproducibly to the sample droplet, thereby eliminating wavering of the hand or manual bad aim that might otherwise result in partial or interrupted aspiration of the drop. The drop stands up in easily accessible form in the cone owing to the shape of the vessel and its hydrophobic construction material. The tubing end is cut at approximately 10°-~200to the normal to ensure that the resultant oval end will riot seal against the side of the cone causing a sudden deleterious shift in the aspiration rate of the capillary nebulizer. The entire drop is readily aspirated because of its surface tension which makes it hold together as a head even though the tubing does not seal against the side of the cone. Both cones are rinsed between samples with a portion of distilled water, although it was found that droplet aspiration is normally so complete that six or more samples of similar concentration can frequently be run in succession without rinsing before a 1% cone contamination accumulates. Rinse water is conveniently removed by aspiration using the nebulizer capillary tube.
Figure 1. Microsampling cones for premixed capillary nebulizer. (a) mounted cone (Teflon) 10 pL-2 mL capacity, 45' taper; (b) manual cone (Teflon) 5 pL-2 mL capacity, 45' taper
require that a separate set of analyses be made (requiring storage of microsamples between times). Alternatively, a scheme may be used requiring that instrumental operation be continually interrupted for wet bench standard addition and/or dilution operations following each semiquantitative estimate. The net result of either undesirable approach is that many high work load laboratories avoid the method of additions altogether. Attempts t o compensate by making u p working standards in artificially concocted matrices (designed to approximate the composition of the samples) also leave much to he desired. An on-the-spot microaddition and microdilution procedure is therefore clearly desirable from a logistics and sample conservation standpoint. This paper deals with further evaluation of the characteristics of the microsampling nebulizer technique as well as several pertinent design and procedural innovations. A new method is presented for convenient (on-the-spot) microstandard addit,ions and microdilutions to be performed by the analyst without interrupting instrument operation. Several common burner memory effects are evaluated and found to play a different role in the transient microsampling approach than would normally occur with steady-state sampling. The microsampling premix nebulizer technique is found to be applicable to flame emission and atomic fluorescence spectrometry as well as the atomic absorption applications originally reported (11-13). EXPERIMENTAL Apparatus. A Varian Techtron (Springvale, Vic., Australia) AA-5 atomic absorption spectrometer equipped with standard nebulizer, 10-cm and 6-cm slot burner heads, circular burner head. hollow cathode lamps, and Hewlett Packard (Palo Alto, Calif.) model 17501A strip chart recorder was used for these investigations. Cd, Zn, and TI Osram lamps, Gates (Long Island, N.Y.) power supply, and 25-mm focal length, 25-mm diameter fused silica condensing lens are provided for atomic fluorescence excitation. Normal hollow cathode and flame excitation were used for atomic absorption and flame emission analyses, respectively. Two Teflon sampling cones were constructed as diagrammed in Figure 1. Cone A is similar to that described elsewhere (12) except that connection t o the instrument is via the small bore vertically oriented plastic nebulizer tubing rather than direct mounting on the horizontal nebulizer capillary. The tubing is press-fit into a hole drilled vertically through the apex of the cone. Cone B was developed in these studies fnr smaller sample sizes as well as on-the-spot microstandard addition and microdilution procedures. Procedure. Fuel rich air- acetylene (2.2 L/min C,H,) flames and a nitrous oxide-acetylene (-6.6 L/min C2H,) flame were maintained at oxidant flows of 1 2 L/min. Samples were digested according to standard dry ashing procedures for Cu. Fe, and Zn (14). Silicon analysis was made following digestion of petrified
RESULTS AND DISCUSSION Volume Response. Samples, 100 pL, (e.g. of 1 ppm Cd) give approximately t h e same concentration sensitivity achieved with larger samples (e.g. 1mL) for atomic absorption spectrometry. T h e absolute sensitivity (weight basis) is however -lox improved for a 100-pL sample. Samples smaller than 100 pL may he employed a t constant absoliite sensitivity and correspondingly reduced concentration sensitivity. Once a sample volume is chosen, conventional linear calibration curves are established a t constant volume, or the method of standard additions may be employed. These results for the manual cone (B) are in good agreement with the atomic absorption results of Sebastiani e t al. ( I I ) for mounted cones. The present studies indicate that the microsamplin,0 cone also works for premixed flame emission (N20/C2H2)and atomic fluorescence (air/C2H2). Aliquots, 100 pL, of 1 ppm Ca and 15 ppm Cd, respectively, produced transient emission and fluorescent signals (peak height) equivalent to -95% of the corresponding value attained using continuous (steady state) nebulization of larger sample volumes under otherwise identical conditions for each technique. These premixed flame emission and atomic fluorescence results indicate similar volume behavior to the atomic absorption. results of Sebastiani et al. ( 1 1 ) . Precision. The precision of this method is generally independent of the mode of excitation employed and was measured in the present studies for atomic absorption (1 ppm Cd) to be -2% RSD (relative standard deviation) a t a 100-pL sample volume. The microsampling precision of both cones (Figure 1) in the range 50-200 pL was found to be essentially no different from that of steady-state aspiration of larger (LmL) samples. The precision worsens below 50 pL but does not exceed 10% RSD for samples as small as 10 pL for cone A and 5 pL for Cone B. The fact that this occurs for transient sampling suggests that it is totally unnecessary from a p r r cision standpoint to aspirate large amounts of sample (21mL) over extended periods of time for the purpose of averaging noise in the steady-state signal. A similar conclusion was reached in earlier atomic absorption work (11, 1 3 ) . In the present studies, the improvement of cone E3 a t volumes G O gL is undoubtedly due to the fact that the sample pipettirig is completed before any sample uptake occurs. The mounted
ANALYTICAL CHEMISTRY, VOL. 50, NO. 12, OCTOBER 1978
1721
T
I
W
0
-
Ln
m
a
a.
b.
d.
C.
F--+ 60 sec
T I M E Figure 2. Example of on-the-spot microdilution and microstandard addition for Orchard Leaf digest (Cu). (a) Blank; (b) initial trial; (c) on-the-spot microdilution (1: 10); (d) subsequent on-the-spot standard addition (+ 0.909 p p m duplicates): chart left running continuously
cone (cone A). however, permits sample uptake to begin before pipetting is completed, allowing interrupted uptake to produce undesirable multiple transients when the sample size drops below 10 pL. Cone B therefore covers the microsampling range from 5-200 pL and cone A operates well from 10-200 pL. The low cost and ease of construction should make it reasonable to obtain both cones as a set for an individual instrument. Samples larger than 200 pL are accommodated by either cone, but these and earlier (11-13) studies should make it clear that larger samples are generally not necessary on a peak height, analog readout basis. New Micro-Calibration Technique (Cone B Only); Dilution and Calibration, A semiquantitative estimate on the first sample of a large group is performed using a 100-pL sample plus 10 pL of water. (The 10 pL of mater is added to compensate a small percentage volume dilution by a subsequent 10-pL standard addition in order to avoid uncorrected variation in the 70matrix effect.) If the resultant level is not suitable for the instrument scale factor, the factor is changed. If the level is altogether too high for atomic spectral quantitation, it is diluted on-the-spot by a known factor. This is accomplished without returning to the wet bench by using the same cone (B) and set of micropipets used to introduce the sample into the spectrometer. If the sample was approximately 10 or 100 times too high, 100 pL or 1 mL respectively of water are quickly delivered to the empty (but rinsed) cone by Eppendorf pipet followed by a 10-pL aliquot of sample. The composite is immediately aspirated completely, resulting in a rapid dilution by a precisely known factor. Adequate solution mixing appears to be induced on such a small scale by the pipetting action. Standard addition is then carried out following a quick semiquantitaiive estimation a t the instrument (using a small electronic calculator if necessary) of the resultant concentration. Standard addition is accomplished by pipetting the same volume (equalling that (100 pL or 1 mL) of the earlier used diluent water) of an appropriately selected aqueous standard into the empty (but rinsed) cone followed by 10 pL of sample. The composite is again aspirated completely and the signal increase noted a t the recorder. Data are reduced later. Direct Calibration. If the original undiluted 100-pL s a m p l e trial (+ 10 pL water) fell within the selected scale range, then that value is taken (instead of discarding it as an
-+
Figure 3. Type of memory encountered in steady-state (continuous) nebulizer sampling (Fe). (a) Normal steady state (0.2 ppm Fe in deionized water); (b) transient memory signal accompanying sudden acidity increases (dilute acid blank): (c) undetermined steady-state (continually increasing signal) memory encountered with dilute acid (blank) for some elements; (d) large continually increasing undetermined steady-state memory encountered with concd acid (blank) for some elements and burner systems. This is not due to reagent contamination
“estimate only”). T o make a standard addition following semiquantitative estimation of the initial reading, a 100-pL sample is again pipetted into the cone followed by quick pipet addition of 10 pL of an appropriately selected aqueous standard. The second pipetting a h o n appears to mix the composite well. The composite is completely aspirated and the signal increase noted on the recorder. Data are reduced later. The cone is rinsed once or twice in preparation for the next sample. These studies have resulted in the ability to sit at the instrument with a large number of a variety of small volume digested complex samples of totally unknown concentrations that vary over many orders of magnitude and efficiently analyze them all by the method of additions without leaving the instrument for wet bench operations. Sample is conserved since the standard addition is made directlv following the initial estimation trial (before the calibratioll drifts) rather than discarding the signal used for estimation (as is normally done in favor of a later rerun when the analyst is ready to come back to the instrument following wet bench operations for the final calibration by standard addition). An example of results using this procedure is shown in Figure 2. The claimed convenience of this procedure was borne out in the rapid determination of Pb, Cu, Zn, Fe, and Ca in digested samples including sea water, tomato sauce, blood, urine, orange juice, pineapple syrup, Grecian Formula 16, pickled beet juice, evaporated milk concentrate, air filters, orchard leaves, bovine liver, etc. The earlier report ( 1 1 ) of superior tolerance t o complex materials (less nebulizer clogging) was also confirmed in the analyses of these high salt content digests. Memory Effects. Two types of memory effects occasionally encountered with continuous (steady-state)aspiration using burner-nebulizer systems are illustrated in Figure 3. It should be apparent that any transient memory signals such as those in part b (Figure 3) will directly interfere with the transient microsampling technique. These transients tend to accompany sudden increases in sample acidity. Since the steady-state (continuous) sampling method is not seriously affected by such transient memory (one simply ignores the transient and uses the ensuing steadystate reading), this will
1722
ANALYTICAL CHEMISTRY, VOL. 50, NO. 12, OCTOBER 1978
T I M E
Figure 4. Memory effects. (a) 0.2 ppm Fe in diluted mineral acid, steady state undetermined, analysis fails; (b) 0.4 ppm Fe in diluted acid, steady
state undetermined, analysis fails again; (c) norreproducible acid blank (erroneous signal equivalent to that of = I ppm Fe), steady state undetermined; (d) 0.2 pprn Fe in deionized water, steady state attained, analysis succeeds: (e) 0.4 ppm Fe in deionized water, steady state attained, analysis succeeds again; (f) microsampling memory transients (acid blank) accompanying sudden acidity increase; interference decays rapidly with trial number until it reaches the small, reproducibly constant value seen in (9). This small value is near or below the limit of detection imposed by system noise; (9) greatly reduced acid blank (signal equivalent to or less than that of -0.1 ppm Fe) encountered when using microsampling; (h) 0.4 ppm Fe microsampling replicates in acid, analysis succeeds (when the small memory signal (9) is subtracted). Similar memory improvements were observed for the microsampling determination of silicon in petrified wood digest (in HF) by atomic absorption spectrometry represent a n interference disadvantage heretofore not mentioned in the literature for the microsampling approach. The opposite is true of the second type of memory (Figure 3, parts c and d showing large, steadily increasing, nonreproducible memory signals) which presents a more serious problem with steady-state sampling. This problem is dramatically reduced (see Figure 4) by the microsampling approach. The steady-state nebulizer determination of silicon and iron in digested complex materials is frequently plagued by the second memory type (not due to reagent contamination) and can be seen to be greatly improved by the small sample approach even if large amounts of sample are available. Figure 4 also shows that the first memory type (transients) generally occurs only when the sample acidity is suddenly raised (parts d, e, and f). The figure also shows that this memory becomes very small and reproducibly constant a t a
value near or below the limit of detection for microsampling (parts f and 9). This dramatic reduction in memory occurs even at high acidity for microsampling if the sample acidity is maintained relatively constant thereafter. No attempt was made to identify the source of this acid memory; it was simply observed to be heavily pronounced for Fe and Si when steady-state nebulizer sampling was employed with the Varian Techtron AA-5. This normally high acid memory is subsequently observed in the present studies to be very aptly reduced and controlled using the microsampling nebulizer technique.
C 0NCLUS IO NS The new microsampling cone for capillary pneumatic nebulization represents a technique of similar concentration sensitivity, improved absolute sensitivity, similar precision, improved sample size requirements, improved analysis time, reduced memory problems, and improved tolerance to high salt content materials in comparison to normal steady-state aspiration approaches. Microsampling cones are found to be very desirable for premixed burner nebulizer systems used in flame emission and atomic fluorescence as well as atomic absorption spectrometry.
LITERATURE CITED (1) H. T. Delves, Ana/yst(London), 95, 431 (1970). (2) H. L. Kahn, G.E. Peterson, and J. E. Schallis, At. Absorpt. Newsl., 7, 35 (1968). (3) H. Massmann, Spectrochim. Acta. far7 8.23, 215 (1968). (4) T. S. West and X . K. Williams, Anal. Chim. Acta, 45, 26 (1969). (5) J. Y. Hwang, P. A. Ullucci, and S. B. Smith, Am. Lab., August (1971). (6) M. B. Denton and H. V. Malmstadt, Anal. Chem., 44, 241 (1972). (7) K. W. Olson, W. J. Haas. Jr., and V. A. Fassei, Anal. Chem.,49, 632 (1977). (8) R. C. Fry and M. 8.Denton, Anal. Chem., 49, 1413 (1977). (9) F. J. Schmidt, and J. L. Royer, Anal. Lett., 6 , 17 (1973). (IO) K . G.Brodie, Am. Lab.. March, 73 (1977). (1 1) E. Sebastiani, K. Ohls, and G. Riemer, Fresenius' Z. Anal. Chem., 264, 105 (1973). (12) D. C. Manning, At. Absorpt. Newsl., 14, 99 (1975). (13) H. Berndt and E. Jackwerth, Spectrochim.Acta, Par7 8,30, 169 (1975). (14) T. T. Gcrsuch, Analyst (London),84, 135 (1959).
RECEIVEDfor review October 3,1977. Accepted June 19, 1978. This research was supported in part by the Office of Naval Research and by a n A. P. Sloane Foundation Research Fellowship (to M.B.D.).
Flow-Through Electrochemical Cell with Open Liquid Junction W. J. Blaedel" and
Z.Yim
Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin 53 706
We report the construction and characterization of a flow-through electrochemical cell with open liquid junction, which permits measurements in the steady-state mode. The cell is of compact design, simple to use, and has good wash characteristics. Most flow-through type cells employ porous plugs (I),microcracks (2), membranes (3),or agar-impregnated frits ( 4 ) t o effect contact between sample and reference solutions, and hence are prone to clog and are difficult to clean, especially when surfactants or proteins are involved. Cells with open liquid junctions have been used to alleviate such difficulties (5). Our design is of the open liquid junction type with the reference and working electrodes situated at channels that form the arms of a Y, and with the flowing reference and sample solutions merging at a junction downstream from both electrodes (see Figure 1). Cell performance is characterized by examining the cathodic reduction of K3Fe(CN)6.
EXPERIMENTAL Figure 1 is a schematic diagram of the cell, which is machined from Plexiglas. The main body of the cell is 1.5 in. thick and 1.5 in. in diameter, with 2-mm diameter flow channels drilled as shown. There are two wells to accommodate the working electrode (platinum) and reference electrode (silver/silver chloride electrode, SSCE). Inlet and outlet ports are all threaded to accommodate Cheminert fittings. The removable working electrode consists of two platinum tubular electrodes (each 2 mm in diameter, 2 mm long) separated by a thin layer of nonconducting epoxy. Such a design can have various applications, anodic stripping voltammetry with collection (ASVU'C) (6), for example. When the two electrode leads are connected together as is the case here, it works as a single electrode with an effective length of 4 mm. The potting with nonconducting epoxy (Epotek 349, Epoxy Technology, Watertown, Mass.) of the platinum disk inside the Plexiglas rod, the drilling of the flow
0003-2700/78/0350-1722$01.0010 1978 American Chemical Society
