Microfabricated Electrochemical Detector for High-Performance Liquid

ARTICLE pubs.acs.org/ac

Microfabricated Electrochemical Detector for High-Performance Liquid Chromatography Evan T. Ogburn,† Michael Dziewatkoski,‡ Don Moles,§ Jay M. Johnson,‡ and William R. Heineman*,† †

Department of Chemistry, University of Cincinnati, Cincinnati, Ohio 45221-0172, United States YSI Incorporated, 1700/1725 Brannum Lane, Yellow Springs, Ohio 45387, United States § Micro Molar Systems, LLC, Cedarville, Ohio 45314, United States ‡

ABSTRACT: A microfabricated electrochemical cell has been developed as a disposable detector for flow injection analysis (FIA) and high-performance liquid chromatography (HPLC). The simplicity of the fabrication process allows this detector to be used as a low-cost, disposable device that can be replaced easily if its performance degrades rather than disassembling the detector and polishing the electrode surface, which is the usual procedure. The detector consists of thin film-metal electrodes—platinum working electrode, platinum auxiliary electrode, and silver/silver chloride coated on Pt reference electrode—deposited on a polyimide substrate with a locking layer of chromium in between. A microfluidic cover made of polyimide directs the solution flow across the electrodes. The detector was evaluated with FIA of ferrocyanide and HPLC of ascorbic acid and acetaminophen and a mixture of two pharmaceutical compounds—dextrorphan and levallorphan—with acetaminophen internal standard. The HPLC calibration curves showed good linearity (R2 > 0.99). Limits of detection were 1 nM for acetaminophen, 1 nM for ascorbic acid, 50 nM for dextrorphan, and 80 nM for levallorphan. When detected with a commercial detector dextrorphan and levallorphan had lower limits of detection, 3 and 5 nM, respectively. Chromatograms of the mixture were comparable to those obtained with a commercial detector. The detector could be used continuously for about 48 h with FIA and about 1020 h with HPLC after which performance gradually degraded as the AgCl on the reference electrode dissolved causing loss of potential control.

H

igh-performance liquid chromatography coupled with electrochemical detection (HPLC-ED) is commonly used for the determination of electroactive compounds, especially those with spectral properties unsuitable for the more commonly used UVvis absorption and fluorescence detectors.1 Detection limits as low as 0.1 pmol have been achieved for oxidizable compounds and 1 pmol for reducible compounds. Electrochemical detectors have rapid response time, wide dynamic range, and low active dead volume (10%, which is greater than what is typically required for an analytical system. The length of time for optimal performance of the microfabricated electrochemical detector 6966

dx.doi.org/10.1021/ac200476j |Anal. Chem. 2011, 83, 6963–6970

Analytical Chemistry

ARTICLE

Figure 4. Peak current (μA) vs injection time (h), over the course of 40 h to monitor the performance of the microfabricated electrochemical detector in an HPLC-ED system using two different analytes, acetaminophen (0.25 μM, blue) and ascorbic acid (1.0 μM, black). The detector performance was also monitored in an FIA-ED system throughout a 48 h period using potassium ferricyanide (2.0 mM, red) as the model analyte; the peak current for this system has the units of nanoamps (nA).

Figure 3. HPLC-ED of acetaminophen (1 μM) with the microfabricated electrochemical detector (A) and a commercial detector (B). The separation was performed using a mobile phase that consisted of 50% 10 mM KH2PO4 (with 1.0 mM EDTA and 10 mM KCl)/50% methanol (v/v) with a flow rate of 0.5 mL/min and a detection potential of þ1000 mV.

Table 2. Stability of HPLC-ED Response of Acetaminophen (0.25 μM) and Ascorbic Acid (1.0 μM) for the Microfabricated Electrochemical Detector average time period

SD of

signal (μA) signal (μA)

percent variability of signal

010 h (acetaminophen)

185

1.1

0.59


184

2.4

1.3


133

15.0

11.3

010 h (ascorbic acid)

292

2.2

0.75


279

3.8

1.4


149

19.0

12.8

depended on the quality of the plating of the reference electrode. When the reference electrode was composed of a uniform layer of Ag/AgCl, the electrode maintained performance in an HPLCED system for 1224 h of continuous flow of mobile phase. The microfabricated electrochemical detector’s lifetime was not as long as in the FIA-ED system, which can be attributed to the greater flow rate that was used in HPLC-ED. The silver chloride that is plated on the surface of the reference electrode slowly dissolved during the continuous flow of mobile phase because of its slight solubility. Increasing flow rate is one way to increase the sensitivity of a hydrodynamic flow electrochemical system such as HPLC-ED as the mass transport of analyte to the surface of the electrode is increased.6 This results in a higher current response in HPLC-ED when compared to FIA-ED but also leads to a more rapid degradation of the detector’s electrochemical performance, primarily because of the faster dissolution

of AgCl from the reference electrode surface. The comparison of performance of the microfabricated electrochemical detector between the FIA-ED and HPLC-ED systems is depicted in Figure 4. The intraday and interday variability of the HPLC-ED system, containing the microfabricated electrochemical detector, was monitored by analyzing acetaminophen and ascorbic acid across their respective linear ranges (see Table 3). The intraday variability of this system was determined by making triplicate injections of acetaminophen at concentrations ranging from 0.001 to 10 μM and of ascorbic acid at concentrations ranging from 0.01 to 10 μM. The intraday variability of system is good, for both compounds, as the percent relative standard deviation at each concentration tested is e3%. The interday variability was determined by repeating the intraday variability test on three consecutive days. The interday variability for both compounds, at each concentration, is e9%. Comparison of the Microfabricated Detector with a Commercial Detector. We compared the performance of the microfabricated electrochemical detector to a commonly used commercial electrochemical detector. A Hewlett-Packard model HP1049A programmable electrochemical detector was chosen for comparison because it was readily available. Comparison chromatograms of acetaminophen for the two detectors are shown in Figure 3. Both peaks are well-defined, but the peak for the microfabricated electrochemical detector is less symmetrical than that for the commercial system, The peak obtained with the microfabricated system has a slightly wider half-peak width (w1/2 = 0.42) compared to that of the commercial system (w1/2 = 0.33). The volume of the flow cell within the microfabricated detector is approximately 20 nL, whereas the commercial detector has a larger flow cell volume of approximately 0.5 μL. Thus, the basic design of the microfluidics electrode chamber provides adequate hydrodynamic performance for this application. A significant difference between the two detectors is the area of the working electrodes: the microfabricated electrochemical detector is 17.0 μm2, whereas the HP1049A is 5020 μm2. This 295-fold difference in area results in the approximately 300 times greater peak height for the commercial detector. Both chromatograms show comparable noise at this current scale, but expanding the current showed the 6967



ARTICLE

Table 3. Intraday and Interday Variability of Acetaminophen and Ascorbic Acid Detection Using the Microfabricated Electrochemical System acetaminophen

ascorbic acid

izntraday variability analyte concn (μM)

average peak height (μA)

intraday variability % RSD

analyte concn (μM)


% RSD

0.001

1.2Eþ01

0.036

0.01

3.2Eþ01

2.6

0.01

6.4Eþ01

0.13

0.05

8.6Eþ01

2.9

0.05 0.1

2.9Eþ02 5.6Eþ02

0.45 0.062

0.1 0.5

5.8Eþ02 1.7Eþ03

0.18 0.33

0.5

2.5Eþ03

0.060

1

2.9Eþ03

0.058

1

5.8Eþ03

0.00053

5

1.5Eþ04

0.23

5

2.6Eþ04

0.064

10

3.1Eþ04

0.015

10

4.9Eþ04

0.027

interday variability analyte concn (μM)


interday variability % RSD

analyte concn (μM)


% RSD

0.001

1.1Eþ01

5.2

0.01

3.0Eþ01

8.8

0.01 0.05

6.0Eþ01 2.8Eþ02

9.0 7.3

0.05 0.1

8.1Eþ01 5.5Eþ02

8.6 6.8

0.1

5.4Eþ02

4.9

0.5

1.7Eþ03

7.3

0.5

2.4Eþ03

5.3

1

2.8Eþ03

6.5

1

5.6Eþ03

6.5

5

1.5Eþ04

5.4

5

2.4Eþ04

8.5

10

3.0Eþ04

5.1

10

4.7Eþ04

7.1

Table 4. Line Statistics and Correlation Coefficients for Calibration Curves of Ascorbic Acid and Acetaminophen Using the Commercial and Microfabricated Electrochemical Systems for Detection compound

electrochemical system used for detection

equation of line

correlation coefficient

acetaminophen acetaminophen ascorbic acid ascorbic acid

commercial microfabricated commercial microfabricated

y = (2.23 106)x þ 9.26 104 y = (4.94 103)x þ 2.20 102 y = (1.17 106)x þ 8.72 104 y = (3.12 103)x þ 2.16 101

R2 = 0.9996 R2 = 0.9995 R2 = 0.9994 R2 = 0.9998

microfabricated electrode to have a noisier baseline due to the smaller currents being measured. The calibration graphs constructed using both the microfabricated and commercial HPLC-ED systems both demonstrate the expected good linear relationship (R2 = 0.999) between analyte concentration and peak current over a wide range (Table 4). The percent relative standard deviations for both compounds at each concentration tested within the linear range was less than 2% for triplicate injections. The linear range for ascorbic acid is equal in both systems, but for acetaminophen the linear range is lower for the commercial detector than the microfabricated electrochemical detector. The noticeable difference when comparing the data in Table 4 is the greater slopes (see equations for plots) for the commercial system, which is due to the difference in working electrode areas. The peak-to-peak noise levels for the microfabricated system ranged from 0.001 to 0.90 nA depending on the mobile phase and the flow rate used during analysis. For the commercial detector the noise levels were of a similar magnitude ranging from 0.005 to 0.80 nA, also depending on the flow rate and mobile phase. Separation of Pharmaceutical Compounds in Multiple Matrixes. The ability to accurately detect chemical compounds

and their metabolites is essential in understanding the metabolic pathways taken by drugs in the human body. Dextromethorphan (DXM), a synthetic analogue of codeine belonging to the morphinan family of drugs, is an antitussive used to treat various neurological disorders.7 Dextromethorphan is known to metabolize to dextrorphan (DXO) through O-demethylation which is mediated by polymorphic CYP2D6.8 The ability to detect dextrorphan allows for dextromethorphan to be used as a probe for human CYP2D6 enzymatic activity both in vitro and in vivo. Levallorphan also belongs to the morphinan family. Compounds such as these that are similar in structure present a difficult challenge when trying to adequately separate and quantify multiple compounds simultaneously. However, compounds with similar structures can sometimes be detected at similar working electrode potentials, as is the case for dextrorphan and levallorphan. Dextrorphan and levallorphan in two different sample matrixes, mobile phase and human plasma, were chosen to further evaluate the performance of the microfabricated detector. The human plasma samples were prepared differently from samples run in mobile phase, as described in the Experimental Section. Levallorphan and dextrorphan are known to be detected at 6968



ARTICLE

potentials of þ1.0 and þ1.2 V.9,10 We found the highest signals at þ1.0 V for both compounds via hydrodynamic voltammetry for this mobile phase. Acetaminophen had an optimum detection potential between þ0.9 and þ1.2 mV. Chromatograms of dextrorphan and levallorphan extracted and reconstituted with mobile phase with acetaminophen as an internal standard for the microfabricated detector (Figure 5A) and the commercial detector (Figure 5B) are very clean with comparable peak shapes and resolution for all three compounds across the linear range (500 nM to 100 μM). The major difference is the higher peak currents for all compounds at the commercial detector than the microfabricated detector because of the difference in the electrode surface area. Calibration curves were constructed for samples in mobile phase and drug-free human plasma that were spiked with known concentrations of dextrorphan and levallorphan. These curves were compared within both systems and matrixes based on reproducibility and accuracy. The standard curves for dextrorphan

in mobile phase (y = 0.013x, R2 = 0.998) and human plasma (y = 0.014x, R2 = 0.996) and levallorphan in mobile phase (y = 0.018x, R2 = 0.998) and human plasma (y = 0.010x, R2 = 0.997) both showed linearity and reproducibility when the microfabricated electrochemical detector was used (Table 5). By comparison, the commercial system showed good linear response for both dextrorphan and levallorphan when diluted in mobile phase but less linear response when the drugs were spiked in human plasma (Table 5). When samples were prepared in mobile phase the response for both systems was clean and precise and reproducible across the entire linear range for dextrorphan and levallorphan (500 nM to 100 μM). The limit of detection for dextrorphan (3 nM) and levallorphan (5 nM) in mobile phase was different from that in human plasma, 10 nM for dextrorphan and 20 nM for levallorphan, when analyzed with the commercial detector. By comparison, the limit of detection was 50 nM for dextrorphan and 80 nM for levallorphan in both mobile phase and human plasma for the microfabricated detector. A major difference in the calibration plots in Table 5, for the two detectors, is the linearity of response for dextrorphan between the two systems, when spiked in human plasma. The linearity and slope of the calibration curve for dextrorphan, when diluted in mobile phase and detected with the commercial detector, (m = 0.027; R2 = 0.994), are comparable to those determined by the microfabricated electrochemical detector (m = 0.013; R2 = 0.998). However, when dextrorphan was spiked in human plasma the linearity of the response decreased drastically for the commercial system (m = 0.0083; R2 = 0.818) and the curve seems as if the signal has reached its maximum for the detector once dextrorphan concentrations become g50 μM. This is not the case for the microfabricated detector as the slope and linearity of the curve are not affected (m = 0.014; R2 = 0.996). The difference in linearity for dextrorphan between the two matrixes, when detected using the commercial detector, could be a result of contaminants found within the spiked plasma samples. These contaminants could have resulted in electrode fouling or the interference of peak signal due to coelution of a contaminant with either the analyte or internal standard. Even though the signal was quite linear and consistent for the microfabricated detector for both dextrorphan and levallorphan, in both matrixes, the ability to restore the system after problems, such as electrode fouling, is more convenient with these detectors since they are made to be disposable and can be readily replaced without any cleaning or further preparation.

Figure 5. Separation of acetaminophen (IS, 50 μM), dextrorphan (0, 0.5, 1, 5, 10, 25, 50, 75, 100 μM), and levallorphan (0, 0.5, 1, 5, 10, 25, 50, 75, 100 μM) using the microfabricated detector (A) and the commercial detector (B). The separation was performed using a mobile phase that consisted of 70% 10 mM KH2PO4 (with 1.0 mM EDTA and 10 mM KCl)/30% acetonitrile (v/v) with a flow rate of 0.5 mL/min and a detection potential of þ1.0 m.

’ CONCLUSIONS The microfabricated detector performed well in many respects compared to a commercially available detector. Good peak

Table 5. Line Statistics and Correlation Coefficients for Calibration Curves of Dextrorphan and Levallorphan Extracted from Mobile Phase and Human Plasma Using the Commercial and Microfabricated Electrochemical Systems for Detection compound

extracted matrix

levallorphan

mobile phase

levallorphan

electrochemical system used for detection

equation of line

correlation coefficient

commercial

y = 0.027x þ 0.058

R2 = 0.995

human plasma

commercial

y = 0.017x þ 0.0075

R2 = 0.958

levallorphan

mobile phase

microfabricated

y = 0.018x þ 0.073

R2 = 0.998

levallorphan dextrorphan

human plasma mobile phase

microfabricated commercial

y = 0.010x þ 0.035 y = 0.027x þ 0.051

R2 = 0.997 R2 = 0.994

dextrorphan

human plasma

commercial

y = 0.0083x þ 0.17

R2 = 0.818

dextrorphan

mobile phase

microfabricated

y = 0.013x þ 0.038

R2 = 0.998

dextrorphan

human plasma

microfabricated

y = 0.014x þ 0.040

R2 = 0.996

6969


Analytical Chemistry shapes and widths with little tailing indicate that the detector volume and hydrodynamic flow are adequate for use with the injection volumes for both FIA and HPLC. Wide linear ranges were obtained for the microfabricated detector. However, the limits of detection were about 10 times higher than for the commercial detector. We attribute this to the much larger working electrode surface area of the commercial detector (295) and the proportionally larger peak currents. This resulted in a larger S/N for the commercial electrode. The S/ N for the microfabricated detector could be improved by optimizing the detection electronics so as to reduce the noise level. Another major difference was the exposed Ag/AgCl reference electrode, which poses two problems. First, the halfcell potential is set by the chloride concentration in the mobile phase. Thus, the potential needed for optimum performance can change if one shifts to another mobile phase with a different chloride concentration. Second, the lifetime of the microfabricated detector was limited by the longevity of the Ag/AgCl reference electrode in the constantly flowing mobile phase. Even though the solubility of AgCl is low, it eventually dissolves, resulting in loss of control of the half-cell potential, which in turn changes the effective potential applied to the working electrode. The lifetime for excellent performance varied from 48 h for FIA to 1020 h for HPLC, after which precision was gradually lost. If this microfabricated system were to be commercialized, then the AgCl plating process of the reference electrode would need to be optimized to provide for a greater reference electrode lifetime. This would eliminate the stability of the reference electrode being the limiting factor, which was the case for the presented work. From a practical standpoint, the electrode system is designed to be disposable, and so only minimal increases in the stability of the reference electrode would be needed. However, the device could be easily redesigned to incorporate a separate, more conventional, reference electrode in the vicinity of the working and auxiliary electrodes on the chip. This should decrease the maintenance requirements of the device by requiring the electrode chip to be replaced only when the working electrode is fouled. A major consideration in adopting the philosophy of a disposable working electrode instead of polishing would be the cost. For this to be attractive, sufficient quantities would have to be manufactured so the cost of each one would approach or be less than the time cost of periodically polishing. This depends on the compounds being analyzed. Some are very prone to fouling, making electrode polishing frequent, whereas others do not exhibit this problem to a great extent. It is the former case where the disposable electrode might be cost-effective. It is estimated that the manufacturing cost of the complete electrochemical detector chip/microfluidic combination would be approximately $25 for a manufacturing volume of 100 000 units per year. Given typical profit margins, the unit might sell for between $40 and $50. This is relatively inexpensive given the labor costs and opportunity costs associated with frequent disassembly and cleaning of a conventional electrochemical detector. Furthermore, from a cGMP standpoint, the cost of the device is minimal for the high level of quality assurance it can provide. The control hardware associated with the electrode chip/microfluidic system would be reusable and hence would represent a one-time cost. Further development of this technology could easily decrease the cost stated above. For example, a future embodiment of this device could involve having only the electrode detector chip be replaceable while reusing the fluidic portion. This would result in a consumable cost of approximately $15 or less.

ARTICLE

’ ACKNOWLEDGMENT We thank Dr. Zereusenay Desta (Indiana University Department of Clinical Pharmacology, Indianapolis, IN) for allowing the use of various analytical instruments and laboratory space that was crucial to the completion of this work. ’ REFERENCES (1) Kissinger, P. T.; Heineman, W. R. Laboratory Techniques in Electroanalytical Chemistry, 2nd ed.; Dekker: New York, 1996. (2) Madaras, M. B.; Buck, R. P. Anal. Chem. 1996, 68, 3832–3839. (3) Iwase, H.; Ono, I. J. Agric. Food Chem. 1997, 45, 4664–4667. (4) Washko, P. W.; Hartzell, W. O.; Levine, M. Anal. Biochem. 1989, 181, 276–282. (5) Courade, J.-P.; Besse, D.; Delchambre, C.; Hanoun, N.; Hamon, M.; Eshalier, A.; Caussade, F.; Cloarec, A. Life Sci. 2001, 69, 1455–1464. (6) Toth, K.; Stulik, K.; Kutner, W.; Feher, Z.; Linder, E. Pure Appl. Chem. 2004, 76, 1119–1138. (7) Pechnick, R.; Poland, R. J. Pharmacol. Exp. Ther. 2004, 309, 515– 522. (8) Yu, A.; Haining, R. Am. Soc. Pharm. Exp. Ther. 2001, 29, 1514– 1520. (9) Jane, I.; McKinnon, A.; Flanagan, R. J. J. Chromatogr., A 1985, 323, 191–225. (10) Shibanoki, S.; Imamura, Y.; Itoh, T.; Ishikawa, K.; Noda, Y.; Yazaki, S.; Suzuki, H. J. Chromatogr., B 1987, 421, 425–429.

6970


Microfabricated Electrochemical Detector for High-Performance Liquid

Recommend Documents