Anal. Chem. 1998, 70, 83-88
Comparison of Potential-Time Waveforms for the Detection of Biogenic Amines in Complex Mixtures following Their Separation by Liquid Chromatography John C. Hoekstra and Dennis C. Johnson*
Department of Chemistry, Iowa State University, Ames, Iowa 50011-3111
Pulsed amperometric detection (PAD), integrated voltammetric detection (IVD), and integrated square-wave detection (ISWD) at gold electrodes are compared for the flow injection detection of 1,3-diaminopropane in a liquid chromatograph (LC). These detection methods are especially significant for alkylamines and amino acids because (i) the majority of these compounds do not naturally possess chromophoric or fluorophoric functional groups and (ii) their amperometric detection at constant potential at Au electrodes fails because the electrode activity is severely attenuated by the formation of an inert surface oxide (AuO). The anodic response mechanisms for detection of amines require concomitant formation of AuO; therefore, a large background signal is observed with conventional PAD. In comparison, the background is much smaller for IVD and ISWD because the anodic charge for oxide formation (positive scan/step) is compensated by the cathodic charge for oxide reduction (negative scan/step). The limits of detection (S/N ) 3) for 1,3-diaminopropane by LC-PAD, LC-IVD and LCISWD are about 3 × 102 pg (4 pmol), 5 × 101 pg (0.7 pmol), and 3 × 101 pg (0.5 pmol), respectively, for 25µL injections. Results are also shown to demonstrate ISWD for detection of nine biogenic amines following their chromatographic separation. Numerous biogenic amines are produced by microbial decomposition of proteins and amino acids. Therefore, quantification of biogenic amines in complex mixtures can find significance in a variety of disciplines. Biogenic diamines, e.g., 1,3-diaminopropane, 1,4-diaminobutane (putrescine), and 1,5-diaminopentane (cadavarine), are found in many food products,1-5 fermented beverages,6,7 and plant tissues.8,9 These same diamines are found (1) Kirschbaum, J.; Luckas, B.; Beinert, W. D. Am. Lab. 1994, 28C-28H. (2) Ingles, D. L.; Back, J. F.; Gallimore, D.; Tindale, R.; Shaw, K. J. J. Sci. Food Agric. 1985, 36, 402-406. (3) Eerola, S.; Hinkkanen, R.; Lindfors, E.; Hirvi, T. J. AOAC Int. 1993, 76, 575-577. (4) Stratton, J. E.; Hutkins, R. W.; Taylor, S. L. J. Food Protein 1991, 54, 460470. (5) Draisci, R.; Cavalli, S.; Lucentini, L.; Stacchini, A. Chromatographia 1993, 35, 584-590. (6) Izquierdo-Pulido, M. L.; Vidal-Carou, M. C.; Marine-Font, A. J. AOAC Int. 1993, 76, 1027-1032. S0003-2700(97)00806-8 CCC: $14.00 Published on Web 01/01/1998
© 1997 American Chemical Society
in tumor cells,10,11 and their detection in organ transplant recipients has been the basis for monitoring the extent of tissue rejection.12-14 The majority of alkylamines do not naturally possess chromophoric or fluorophoric moieties, and current LC analyses require pre- or postcolumn derivatization to facilitate photometric detection. Derivatization procedures increase analysis time and risk for indeterminate errors. Therefore, we are convinced that direct detection of amines, when available with adequate sensitivity, will be preferred by the majority of analysts. For more than a decade, research in this laboratory has been focused on the development of direct electrochemical methods for the anodic detection of polar aliphatic compounds in complex samples following their chromatographic separations. A majority of this work has involved pulsed amperometric detection (PAD) to achieve a high and reproducible response at Pt and Au electrodes.15-21 The conventional three-step PAD waveform is shown in Figure 1A. The discrete potential steps achieve the sequential operations of (i) amperometric detection (EDET), (ii) oxidative surface cleaning (EOXD), and (iii) reductive surface reactivation (ERED). This waveform ensures the regeneration of a pristine electrode surface having maximum activity following completion of each cycle of the waveform. The significance of PAD applied to polar aliphatic compounds is especially significant in view of the common observation that dc amperometry fails for these compounds at these same electrodes. The failure of dc amperometry is especially apparent for compounds that are (7) Yen, G.-C.; Chandra, T. J. Sci. Food Agric. 1988, 44, 273-280. (8) Flores, H. E.; Galston, A. W. Plant Phys. 1982, 69, 901-706. (9) Slocum, R. D.; Flores, H. E.; Galston, A. W.; Weinstein, L. H. Plant Phys. 1989, 89, 512-517. (10) Danner, R. M.; Reddy, T. V.; Guion, C. W. LC-GC 1994, 12, 244-248. (11) Zhang, R.; Cooper, C. L.; Ma, Y. Anal. Chem. 1993, 65, 704-706. (12) Russel, D. H.; Durie, B. G. Polyamines as Biochemical Markers of Normal and Malignant Growth; Raven Press: New York, 1978. (13) Janne, J.; Poso, H.; Raina, A. Biochim. Biophys. Acta 1978, 473, 241-293. (14) Tabor, C. W.; Tabor, H. Annu. Rev. Biochem. 1976, 45, 285-306. (15) Austin-Harrison, D. S.; Johnson, D. C. Electroanalysis 1989, 1, 189-197. (16) Johnson, D. C.; Polta, T. Z. Chromatogr. Forum 1986, 1, 37-43. (17) Johnson, D. C.; Polta, J. A.; Polta, T. Z.; Neuburger, G. G.; Johnson, J.; Tang, A. P.-C.; Yeo, I.-H.; Baur, J. J. Chem. Soc., Faraday Trans. 1986, 82, 10811098. (18) Austin, D. S.; Polta, J. A.; Polta, T. Z.; Tang, A. P.-C.; Cabelka, T. D.; Johnson, D. C. J. Electroanal. Chem. 1984, 168, 227-248. (19) Johnson, D. C.; LaCourse, W. R. Anal. Chem. 1990, 62, 589A-597A. (20) Johnson, D. C.; LaCourse, W. R. Electroanalysis 1992, 4, 367-380. (21) Polta, J. A.; Johnson, D. C. Anal. Chem. 1985, 57, 1373-1376.
Analytical Chemistry, Vol. 70, No. 1, January 1, 1998 83
Figure 1. Potential-time (E-t) waveforms for PAD, IVD, and ISWD. PAD parameters: EDET ) detection potential, TDET ) detection period, TDEL ) delay period, TINT ) integration period (TDET ) TDEL + TINT), EOXD ) oxidation potential, TOXD ) oxidation period, ERED ) reduction potential, and TRED ) reduction period. IVD and ISWD parameters: EMAX ) maximum detection potential, EMIN ) minimum detection potential, and TINT ) integration period.
electroactive only at potentials where adsorption of electroinactive species and/or formation of inert surface oxides can result in deactivation of electrode surfaces. PAD is successful for detection of numerous classes of polar aliphatic compounds including simple alcohols, glycols, and polyalcohols; monosaccharides, disaccharides, and oligosaccharides; amines, alkanolamines, and amino acids; and organosulfur compounds.22-31 The PAD waveform in Figure 1A is well established for the sensitive detection of alcohols, polyalcohols, and carbohydrates at Au electrodes in alkaline media. These detections are characterized by low background signals because the EDET values that produce the highest sensitivity do not result in formation of AuO. However, performance of the PAD waveform is not as admirable when applied for amines whose anodic response requires concomitant formation of AuO on the electrode surface. The anodic current from oxide formation is the source of a large background signal. To minimize the baseline signal in LC-PAD, integration of the electrode current at EDET can be delayed for a substantial period following the step from ERED to EDET to permit the oxide formation current to decay. However, during this delay time (TDEL), the inert oxide formed has the effect of attenuating electrode response for further detection of amines within that cycle of the waveform. Waveforms have been described that were designed to automatically reject the background signal from oxide forma(22) Dobberpuhl, D. A.; Hoekstra, J. C.; Johnson, D. C. Anal. Chim. Acta 1996, 322, 55-62. (23) Johnson, D. C.; Dobberpuhl, D. A.; Roberts, R. E.; Vandeberg, P. J. J. Chromatogr. 1993, 640, 79-96. (24) Johnson, D. C.; LaCourse, W. R. In Carbohydrate Analysis: High Performance Liquid Chromatography and Capillary Electrophoreses; Rassi, Z. E., Ed.; Elsevier: Amsterdam, 1995; pp 391-429. (25) LaCourse, W. R.; Jackson, W. A.; Johnson, D. C. Anal. Chem. 1989, 61, 2466-2471. (26) Neuburger, G. G.; Johnson, D. C. Anal. Chim. Acta 1987, 192, 205-213. (27) Ngoviwatchai, A.; Johnson, D. C. Anal. Chim. Acta 1988, 215, 1-12. (28) Polta, J. A.; Johnson, D. C. J. Liq. Chromatogr. 1983, 6, 1727-1743. (29) Polta, T. Z.; Johnson, D. C. J. Electroanal. Chem. 1986, 209, 159-169. (30) Vandeberg, P. J.; Kawagoe, J. L.; Johnson, D. C. Anal. Chim. Acta 1992, 260, 1-11. (31) Neuburger, G. G.; Johnson, D. C. Anal. Chem. 1988, 60, 2288-2293.
84 Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
tion.31,32 These waveforms have been characterized by various names, including potential sweep/step pulsed coulometric detection (PS-PCD). Based on the principles described for PS-PCD, Vandeberg and Johnson33 described integrated voltammetric detection (IVD). The IVD waveform, shown in Figure 1B, effectively minimizes any contributions to the net analytical signal from the formation (positive scan) and reduction (negative scan) of surface oxide because these faradaic signals are of opposite polarity. Whereas potential steps can be added following the triangular potential scan in the IVD waveform to achieve oxidative cleaning and reductive reactivation of the electrode surface, these steps generally are not necessary because the formation/reduction of surface oxide within the triangular scan is adequate for maintaining uniformly high electrode activity. Although IVD successfully increases the signal-to-background ratio (S/B), the frequency of the waveform is constrained by the kinetics of the oxide formation/reduction reactions. More specifically, it is observed that, at fast scan rates, the formation and reduction waves are shifted to more positive and negative potentials, respectively. If the reduction wave shifts into a region where the O2(aq) is reduced simultaneously with surface oxide, then EMIN in the IVD waveform must be shifted to more negative values to ensure complete reduction of the oxide layer. The result is that slight variations in O2(aq) concentration in the mobile phase can produce undesirable shifts in the LC-IVD baseline signal. To address the problems that inherently plague PAD and IVD, we propose a new waveform for detection of biogenic amines. This waveform, described as integrated square-wave detection (ISWD), is designed to combine the best aspects of PS-PCD and IVD using the two-step waveform shown in Figure 1C. The first potential step is made to a value where the analyte is oxidized concomitantly with AuO formation. The second step is made back to a value where surface oxide is reduced. Current integration continues throughout both potential steps; therefore, the net current integral has minimal contribution from formation of surface oxide. Furthermore, because the potential is stepped, rather than scanned, the negative effects from slow oxide formation/stripping kinetics are minimized by eliminating the wasted time from scanning the potential to its maximum and minimum values. In effect, ISWD corresponds to IVD with very fast scan rates and holding periods at EMAX and EMIN. Consequently, the frequency of the ISWD waveform can be made sufficiently fast to accurately define chromatographic peaks without the risk of detecting O2(aq) at EMIN. Just as in IVD, ISWD samples all of the current produced by analyte oxidation, so sensitivity is better than that for PAD. EXPERIMENTAL SECTION Reagents. All chemicals were used as received. Ethylenediamine, 1,3-diaminopropane, 1,4-diaminobutane (putrescine), 1,5diaminopentane (cadavarine), spermine, and spermidine were reagent grade or better from Aldrich. Agmatine (H2SO4 salt), β-phenylalanine (HCl salt), histamine (HCl salt), tyramine (HCl salt), and tryptamine (HCl salt) were from Sigma. All other chemicals were from Fisher Scientific. Tap water was treated in a serial fashion by passage through ion-exchange cartridges (Culligan) and a Milli-Q water purification system (Millipore). (32) Welch, L. E.; LaCourse, W. R.; Mead, D. A., Jr.; Johnson, D. C.; Hu, T. Anal. Chem. 1989, 61, 555-559. (33) Vandeberg, P. J.; Johnson, D. C. Anal. Chim. Acta 1994, 290, 317-327.
Chromatographic eluents were aspirated through a 0.22-µm filter (Micron Separations) to remove particulate matter. Potassium hydroxide solutions were prepared from KOH pellets (Fisher Scientific) and then aspirated through a 0.22-µm filter (Micron Separations). Voltammetric Apparatus and Procedures. Cyclic and pulsed voltammetry were performed at the gold disk of an AMFT28AUAU rotated ring-disk electrode (RRDE, Pine Instruments) or a Au minidisk (1.0-mm diameter, Pine Instruments). Electrode rotation was provided by an AFMSRX rotator (Pine Instruments). The electrochemical cell was made of Pyrex, with the chambers for the reference and auxiliary electrodes separated from the chamber for the working electrode by fritted glass disks (medium porosity). The auxiliary electrode was a coiled Pt wire (∼1 cm2), and all potentials were measured versus a saturated calomel electrode (SCE, Fisher Scientific). Voltammetric data were obtained using an AFRDE5 bipotentiostat (Pine Instruments) interfaced to an Apex 486DX2 personal computer via a Lab PC+ data acquisition board (National Instruments). For cyclic voltammetry at low scan rates (1000 mV/s) was performed with an ED40 electrochemical detector operated in the amperometry mode (Dionex). For pulsed voltammetry, the bipotentiostat was placed under computer control using Labview 3.1 software (National Instruments). Programs were written to generate the desired waveform for pulsed voltammetry and to collect the data for both cyclic and pulsed voltammetry. In PAD and ISWD waveforms, rapid potential changes were executed as fast scans having a 10ms duration rather than as step functions to prevent the operational amplifiers in the potentiostats from being driven to potential/ current saturation. Unless otherwise stated, all solutions used in voltammetric experiments were deaerated by dispersed N2(g) (99.99%, Air Products). HPLC System and Procedures. Biogenic amines were separated on OmniPac PCX-500 guard (4 mm × 50 mm) and PCX500 (4 mm × 250 mm) columns in series (Dionex) or an IonPac CS14 analytical column (4 mm × 250 mm). Sample injection was achieved by a pneumatically controlled injector with a 25-µL sample loop. A GPM gradient pump and an ED40 electrochemical detector (Dionex) were interfaced to a personal computer through a DX LAN network connection and controlled with PeakNet chromatography software (Dionex). The flow-through electrochemical cell (Dionex) consisted of a 1-mm-diameter Au working electrode and a pH-Ag/AgCl combination reference electrode. The counter electrode was provided by the titanium half of the cell. The active volume of the cell was 0.2 µL. A CSRS-I (4-mm i.d.) cation suppressor (Dionex) was applied postcolumn to elevate the mobile phase pH to 13. The cation suppressor was regenerated by 0.5 M KOH flowing at 1 mL/min. RESULTS AND DISCUSSION Voltammetric Response of Diamines. Selection of waveform potentials used in the PAD, IVD, and ISWD waveforms was based on current-potential (i-E) curves obtained by pulsed and cyclic voltammetry. The use of pulsed voltammetry for optimization of PAD waveforms has been described.34 The dashed curve in Figure 2 shows the residual response of the Au RDE in 0.10 M
Figure 2. Pulsed voltammetry for 1,3-diaminopropane at a Au RDE in 0.10 M KOH. Rotation rate: 1000 rpm. PAD waveform: EDET, stepped by 10-mV increments; TDEL ) 250 ms; TINT ) 50 ms; EOXD ) 0.80 V; TOXD ) 100 ms; ERED ) -0.40 V; and TRED ) 360 ms. DAP concentration (µM): (- - -) 0 and (s) 70.
KOH using the PAD waveform in which the detection potential (EDET) was incremented by 10-mV steps for each cycle of the waveform. The oxidation (EOXD) and reduction (ERED) potentials were held constant at 0.80 and 0.40 V, respectively, for 100 and 360 ms. The values of charge shown on the y-axis correspond to the faradaic charge determined by digital integration of electrode current passed at EDET during a 50-ms integration period following a 250-ms delay period. A large anodic wave is observed to grow, starting at EDET ≈ 0.1 V, corresponding to the formation of surface oxide (Au f AuOH f AuO). The anodic response for 70 µM DAP, represented by the solid curve in Figure 2, is produced concomitantly with formation of AuO in the region EDET > ∼0.15 V. Based on these results, it is apparent that the maximum value for the signal-to-background ratio (S/B) is obtained in the region EDET ) 0.3-0.4 V. The maximum value of S/B is pertinent because the noise (N) on the baseline is proportional to B and, therefore, S/N is optimized when S/B is at its maximum value. Pulsed voltammetry is recommended for determination of optimal waveform parameters in PAD;32 however, cyclic voltammetry is recommended for optimization of waveforms to be applied for IVD and ISWD. The residual cyclic voltammetric response of the Au RDE in 0.10 M KOH is given by the dashed curve in Figure 3A. The large anodic wave observed during the positive scan (E > ∼0.15 V) corresponds to formation of AuO, and the cathodic peak observed during the negative scan (E ) 0.1 to -0.2 V) corresponds to subsequent reduction of AuO. The response for 10 µM DAP is given by the solid curve in Figure 3A. During the positive scan, oxidation of DAP occurs for E > ∼0.1 V and continues until the scan direction is reversed at E ) 0.60 V. Following scan reversal, anodic response from DAP quickly terminates. Furthermore, during the negative scan, the cathodic (34) LaCourse, W. R.; Johnson, D. C. Anal. Chem. 1993, 65, 50-55.
Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
85
Figure 4. Residual cyclic voltammetric response at a Au RDE in 0.10 M KOH. Rotation speed, 1000 rpm. Scan rate (mV/s): (-‚‚-) 375, (- - -) 750, and (s) 1500.
Figure 3. Cyclic voltammetric (A) and integrated cyclic voltammetric (B) response for 1,3-diaminopropane at a Au RDE in 0.10 M KOH. Scan rate, 750 mV/s; rotation rate, 1000 rpm. DAP concentration (µM): (- - -) 0 and (s) 10.
peak for AuO reduction is attenuated slightly by in the presence of DAP. This slight attenuation is attributed to inhibition of AuO formation (positive scan) caused by DAP that is adsorbed on the oxide-free surface in the region E > -0.1 V (negative scan). The inhibition of AuO formation by adsorbed diamines can decrease the sensitivity of LC-PAD; i.e., the slope of the calibration line is decreased. Typically, the baseline underneath a chromatographic peak is assumed to be equal to the baseline observed for the absence of analyte. But this is not a valid assumption when PAD is applied for detection of strongly adsorbed amines, i.e., diamines. Because of strong adsorption, the baseline signal corresponding to the detection peak is less than that observed in the absence of the diamines. Therefore, the apparent peak height, measured with respect to the baseline observed before the peak, is less than the true peak signal. IVD and ISWD are not affected by this problem because these waveforms make the background correction regardless of the amount of AuO formed. The integrated voltammetric response is shown by the dashed curve in Figure 3B. During the positive scan, anodic charge accumulates as a result of the formation of AuO (E > ∼0.2 V). When the positive scan is reversed (E ) 0.6 V), oxide formation ceases, and a plateau is observed (E ≈ 0.4-0.2 V). As the potential is scanned to more negative values (E < 0.2 V), the total 86 Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
accumulated charge decreases because the cathodic charge for oxide reduction is effectively subtracted from the total anodic charge acquired during the previous positive scan. Therefore, although a large anodic charge (∼225 µC) is obtained during the positive scan, the net residual charge is only ∼20 µC at the completion of the negative scan. It is speculated that the majority of the remaining net charge is the results of some O2 evolution during the positive scan in the region E > ∼0.4 V. The presence of dissolved O2 in the chromatographic eluent is a concern when discussing electrochemical detection at Au electrodes. Under alkaline conditions, i.e., pH g 13, O2(aq) is cathodically reduced at E < -0.2 V. PAD is not affected by the presence of O2(aq) in applications for amine detections because EDET > -0.2 V. However, IVD and ISWD require integration of the electrode current at potentials near -0.2 V to ensure the complete reduction of AuO. IVD in thin-layer flow-through cells is especially susceptible to this problem because a rapid potential scan rate must be applied to achieve a waveform frequency that is sufficiently high to accurately describe very sharp chromatographic peaks. As a consequence of the large current values resulting from anodic/cathodic formation/reduction of surface oxide, the large cell resistance (Rcell) in the thin-layer cells produce a large iR drop, which effectively shifts the anodic and cathodic processes to higher values of applied potential. This phenomenon is illustrated in Figure 4, where the residual voltammetric response is shown for scan rates of IVD. Figure 4 contains residual cyclic voltammetric data for three scan rates at a Au electrode in the thin-layer flow-through cell: 375 (-‚‚-), 750 (- - -), and 1500 mV/s (s). The fact that the anodic and cathodic waves obtained at 750 mV/s are shifted significantly in comparison to the corresponding waves in Figure 3A obtained using the same scan rate is a consequence of the large Rcell. Therefore, for high scan rates in IVD, the negative scan limit must be shifted negative to values E . -0.2 V to ensure that all AuO is reduced; as a consequence, reduction of O2(aq) will contribute to the IVD signal. The rationale for recommending ISWD over IVD is that ISWD does not waste time by scanning the potential over the desired intervals. Instead, the potential is stepped between the maximum
Table 1. Description of Waveforms for Pulsed Amperometric Detection (PAD), Integrated Voltammetric Detection (IVD), and Integrated Square-Wave Detection (ISWD) PAD
IVD
ISWD
time (ms)
potential (V)
time (ms)
potential (V)
0 300 310 410 420 780
0.45 0.45 0.80 0.80 -0.40 -0.40
0 1000 2000
-0.10 0.60 -0.10
EDET ) 0.45 V TDEL ) 250 - 0 ms ) 250 ms TINT ) 300 - 250 ms ) 50 ms TUP ) 310 - 300 ms ) 10 ms
EMIN ) -0.10 V TUP ) 1000 - 0 ms ) 1000 ms
time (ms)
potential (V)
0 40 50 450 460 1000
-0.10 -0.10 0.60 0.60 -0.10 -0.10
EMIN ) -0.10 V TUP ) 50 - 40 ms ) 10 ms
EOXD ) 0.80 V EMAX ) 0.60 V EMAX ) 0.60 V TOXD ) 410 - 310 ms TDN ) 2000 - 1000 ms TDN ) 460 - 450 ms ) 100 ms ) 1000 ms ) 10 ms TDN ) 420 - 410 ms ) 10 ms ERED ) -0.40 V
TINT ) 850 - 150 ms ) 600 ms
TINT ) 860 - 40 ms ) 820 ms TRED ) 780 - 420 ms ) 350 ms
frequency ) 0.50 Hz
frequency ) 1.00 Hz
TRED ) 780 - 420 ms ) 350 ms frequency ) 1.28 Hz
Table 2. Statistical Summary for Calibration Data Shown in Figure 5
waveform frequency (Hz) sensitivity (nC/ng) r2 standard deviation of baseline (nC, N ) 61) limit of detection (pg, S/N ) 3)
PAD
IVD
ISWD
1.16 0.41 0.9981 0.03
0.50 20.3 0.9994 0.33
1.00 13.0 0.9997 0.16
3 × 102
5 × 101
3 × 101
potential (EMAX) and minimum potential (EMIN) with allowance of ample time periods at each potential value to achieve anodic detection, with simultaneous oxidative surface cleaning (EMAX) and AuO reduction (EMIN). With appropriate choice of EMAX and EMIN values in ISWD, application of separate anodic cleaning and cathodic reactivation steps is not necessary. Therefore, ISWD can be applied at higher waveform frequencies than IVD to provide better definition of sharp chromatographic peaks. Comparison of PAD, IVD, and ISWD Responses. Calibration curves are used to compare the chromatographic response of PAD, IVD, and ISWD applied for detection of DAP (2.0-18.5 ng/injection) using the waveforms specified in Table 1. The PAD waveform was optimized by Dobberpuhl et al.,22 and that for IVD was optimized in this study. The statistics for the calibration curves are summarized in Table 2. The low sensitivity of PAD, in comparison to that for IVD and ISWD, is explained by the incorporation of a long delay period (TDEL ) 250 ms) prior to current integration, following the step to the detection potential (EDET ) 0.40 V), to minimize contribution from AuO formation to the PAD signal. The sensitivity of IVD is greater than that of ISWD because AuO is not formed as fast in the IVD waveform,
Figure 5. ISWD signal vs concentration of 1,3-diaminopropane at the Au minidisk electrode in 0.10 M KOH. Rotation speed: 1000 rpm. Waveform: see Table 1.
allowing greater contribution to the analytical response from DAP transported to the electrode when the surface is only partially covered by the inert oxide (AuO). However, the standard deviation of the IVD baseline was approximately twice that of ISWD in this study; therefore, the lowest detection limit was obtained for ISWD. It is worthy of note that the IVD waveform used to obtain these data had a frequency only one-half that of the ISWD waveform. For reasons already mentioned, IVD frequencies larger than the value applied to obtain the data in Table 2 can create problems either because AuO is not totally reduced for EMIN ≈ -0.2 V or because of the cathodic detection of O2(aq) for EMIN < -0.2 V. Based on the statistics in Table 2, the limits of detection (S/N ) 3) for PAD, IVD, and ISWD are estimated to be about 3 × 102 pg (4 pmol), 5 × 101 pg (0.7 pmol), and 3 × 101 pg (0.5 pmol), respectively, for the 25-µL injections of 1,3-diaminopropane in this LC system. Although the detection limits for ISWD are only a factor of 2 better (lower) than those for IVD, ISWD allows the additional advantage of higher waveform frequencies. This becomes significant when trying to define sharp chromatographic peaks. ISWD Response. Responses of IVD and ISWD for amines can be expected to deviate from linearity at high concentrations of amines. This is explained on the basis of simultaneous contributions to total signal from the oxidation of preadsorbed amines and from oxidation of amines transported to the electrode surface during the oxide formation process.35 At low concentrations, the fractional surface coverage of adsorbed amine is low and is expected to be a linear function of amine concentration. Therefore, in this region, the total anodic response is a linear function of concentration. However, at higher concentrations, the fractional surface coverage is not a linear function of amine concentration; therefore, the total response is not a linear function of concentration. Figure 5 illustrates deviation from linear (35) Dobberpuhl D. A. Ph.D. Dissertation, Iowa State University, 1994, 39-74.
Analytical Chemistry, Vol. 70, No. 1, January 1, 1998
87
Figure 6. Reproducibility of LC-ISWD response for 1,3-diaminopropane. DAP concentrations are given in the figure. Column: Dionex CS-14 (4 mm × 250 mm). Mobile phase: 460 mM KNO3/17.5% MeCN/30 mM HOAc at 1.0 mL/min. Cation suppression: Dionex CSRS-I. Regenerant: 0.5 M KOH at 1.0 mL/min. Waveforms: see Table 1. Table 3. Statistical Summary for Consecutive LC-ISWD Peaks Obtained for Seven 25-µL Injections Containing 185 and 18.5 ng of 1,3-Diaminopropane (DAP), As Represented in Figure 7 DAP content
mean peak height (nC) standard deviation (nC, N ) 7) relative standard deviation (%)
185 ng
18.5 ng
2274 5.6 0.24
227.0 1.1 0.48
response for ISWD applied over the concentration range 0-80 µM. Linearity of response is satisfactory in the range 20-80 µM (r2 ) 0.9993). The reproducibility of LC-ISWD was determined on the basis of peak signals for seven 25-µL injections of 25 and 0.25 nmol of DAP. Representative detection peaks are shown in Figure 6. Responses at both concentrations were very reproducible, with relative standard deviations ∼0.6 V) and cathodic reduction of O2(aq) (E < -0.2 V) in 0.10 M KOH. The response of PAD for 1,3diaminopropane (∼0.4 nC/ng) was negligible compared to those of IVD (∼20 nC/ng) and ISWD (∼13 nC/ng). Although IVD was more sensitive than ISWD, ISWD produced lower detection limits because the standard deviation of the ISWD baseline (0.16 nC, N ) 61) was smaller than that for IVD (0.33 nC, N ) 61). Detection limits (S/N ) 3) for PAD, IVD, and ISWD were about 2 × 102, 5 × 101, and 3 × 101 pg, respectively. The main advantage of ISWD is the ability to work at higher waveform frequencies than IVD while maintaining a larger S/B than that of PAD. ISWD response was very reproducible, with RSD values for peak height