IOP PUBLISHING
JOURNAL OF MICROMECHANICS AND MICROENGINEERING
doi:10.1088/0960-1317/19/2/025022
J. Micromech. Microeng. 19 (2009) 025022 (6pp)
A cobalt-coated needle-type microelectrode array sensor for in situ monitoring of phosphate Jin-Hwan Lee1, Woo Hyoung Lee2, Paul L Bishop2 and Ian Papautsky1,3 1
Department of Electrical and Computer Engineering, University of Cincinnati, Cincinnati, OH 45221, USA 2 Department of Civil and Environmental Engineering, University of Cincinnati, Cincinnati, OH 45221, USA E-mail:
[email protected]
Received 31 August 2008, in final form 10 December 2008 Published 26 January 2009 Online at stacks.iop.org/JMM/19/025022 Abstract In this paper, a new cobalt-coated needle-type microelectrode array sensor for in situ measurements of phosphate has successfully been designed, fabricated and characterized. MEMS technologies were used to fabricate microelectrode arrays with a small tip size. An HF-based etching technique was used to taper and sharpen three-dimensional glass probes. A cobalt thin film was electroplated on a gold conductive layer using a cobalt sulfate electrolyte and then was oxidized to form CoO on the surface. The microelectrode array (MEA) was packaged on a designed printed circuit board (PCB) for electrical connections. The MEA sensors were fully characterized with potassium sulfate solution in the concentration range of 1 × 10−5.1−1 × 10−3 M at pH 7.5. The repeatable phosphate-selective potentials with a sensitivity of ∼96 mV per decade were exhibited with less than 30 s response times and good signal stability. This sensitivity is the highest value among the reported cobalt-based phosphate sensors to date. Ultimately, in the long term, we envision extension of this MEA sensor to include additional sensors for multi-analyte, rapid, accurate and reliable in situ sensing in biological applications. (Some figures in this article are in colour only in the electronic version)
The sensing mechanism involves dissolution of cobalt on the electrode surface and the formation of oxide film [1]:
1. Introduction Measurement of phosphate has been of tremendous interest for the past three decades [1]. Phosphate is an essential nutrient for plants, and its measurement has been used to control fertilizers applied to maximize crop yield and quality in hydroponics and agriculture, or to control undesired grown of the algae and other aquatic vegetation to prevent the eutrophication of natural water bodies. Thus, due to the importance of phosphate and the protection of the global environment, there has been accelerating progress in the development of simple and compact phosphate sensors [2]. One approach to measuring phosphate directly is based on the formation of Co3(PO4)2 precipitate on Co wire [1, 3–6]. 3
2Co + 2H2 O ↔ 2CoO + 4H+ + 4e−
(1a) (1b)
2Co + O2 ↔ 2CoO.
(2)
When phosphate is present in solution, cobalt phosphate is formed on the electrode surface, depending on the solution pH [1]:
Author to whom correspondence should be addressed.
0960-1317/09/025022+06$30.00
−
O2 + 4H + 4e ↔ 2H2 O +
1
+ 3CoO + 2H2 PO− 4 + 2H ↔ Co3 (PO4 )2 + 3H2 O (at pH 4.0)
(3)
− 3CoO + 2HPO2− 4 + H2 O ↔ Co3 (PO4 )2 + 4OH (at pH 8.0)
(4)
© 2009 IOP Publishing Ltd Printed in the UK
J-H Lee et al
J. Micromech. Microeng. 19 (2009) 025022 − 3CoO + 2PO3− 4 + 3H2 O ↔ Co3 (PO4 )2 + 6OH
(at pH 11.0).
(5)
These coupled reactions show the shift in the equilibrium potential that is dependent upon oxidation of cobalt, reduction of oxygen and the Co3(PO4)2 precipitate forming on the electrode surface. This leads to a shift of the mixed potential to more negative, while keeping other factors constant. The shift will be related to the phosphate concentration since equilibrium potentials are governed by the Nernst equation [7]. Thus, a linear potential response may be expected with exponential change in phosphate ion concentration (at constant levels of dissolved oxygen) [8]. Such cobalt metal electrodes can detect both the inorganic and organic phosphates [1, 3, 6], although they have been shown to be sensitive to interference from dissolved oxygen. The detection limit of these sensors is typically around 10 μM [3]. Alternatively, biosensors have also been used to measure phosphate levels [2, 9–11]. However, these sensors do not measure phosphate directly, but rather use enzymes immobilized on substrates that catalyze reactions to ultimately produce oxygen which is measured. Phosphate sensor electrodes based on liquid membranes [12–14] and solidstate membranes [15–17] have also been demonstrated. The detection limits of these sensors are also in the 10 μM range. Overall, the large size of these sensors has caused difficulty in applying them to small volume samples such as in biofilm and tissue research. In this paper, MEMS technologies were used to demonstrate needle-type phosphate microelectrode array (MEA) sensors using a cobalt ion-selective solid-state membrane. The phosphate MEA was fabricated using meniscus etching process that we previously applied to the development of oxidation-reduction potential (ORP) [18, 19] and dissolved oxygen (DO) [20, 21] sensors. To overcome oxygen sensitivity of the cobalt-based phosphate-sensing scheme, the MEA sensor integrates a potentiometric phosphate sensor with a DO sensor. The micrometer tip size and the needle nature of the sensor permit in situ analysis by penetrating biological or environmental samples, e.g. tissues and biofilms. This sensor system will permit the development of multi-analyte sensor in situ for applications by integrating sensors with a signal conditioning system. With further development, additional sensors (e.g., ammonia, nitrate, pH, etc) may be developed by applying ion-selective membranes at the sensor tip.
(c)
(a)
(b) (d )
(e)
Figure 1. Phosphate MEA fabrication sequence: (a) form glass probes by removing excess material with a dicing saw; (b) taper probes by slowly withdrawing from an HF-based etchant and use meniscus etching to sharpen probe tips; (c) deposit an Au conductive layer and use silver epoxy to establish electrical connections between the MEA and the carrier; (d) coat microelectrode shafts with polypyrrole for insulation. Close-up shows the structure of the cobalt-coated phosphate MEA sensor tip.
Briefly, MEAs were fabricated from 175 μm thick 50 mm × 45 mm borosilicate glass wafers by first making 2 cm long cuts with a dicing saw to form four probe arrays of 175 μm wide glass probes at 900 μm center-to-center spacing. Longer 2.5 cm cuts were made every four probes to define edges of individual MEAs. The cut wafers were then annealed at 550 ◦ C for 10 min in a programmable box furnace (Lindberg/Blue M, Thermo Scientific, Norwood, MA) to relieve stress from the dicing process. The formed glass probes were sharpened into needle-type microelectrodes using meniscus etching process that relies on the organic layer to modify the contact angle at the glass–etchant interface, as we described previously [18, 20]. The probes were etched statically for 10 min to smooth the diced surface with a 10:7:33 (v/v/v) mixture of HF, HNO3 and H2O; then the tapered structure was formed by gradual withdrawal (∼1.8 mm min−1), and sharpened tips with meniscus etching yielding ∼200 nm tips [18, 20]. This final etch step was self-terminating, permitting consistent and reliable fabrication of microelectrode sensor tips. The probes were metalized on both sides by thermal evaporation of a 200 nm thick layer of Au as a conductive layer on top of a 20 nm thick layer of a Ti adhesion layer. A shadow mask was used to protect the microelectrode array base on both sides to prevent metal deposition and to electrically isolate individual microelectrodes. MEAs were packaged on a printed circuit board (PCB) laminate using UV-cured epoxy (3301, Loctite, Rocky Hill, CT). The copper-clad laminate was 790 μm thick (D&L
2. Methods Instead of using an insulated wire cathode typical in conventional sensors, the MEA uses a sharpened solid glass core with a thin-film metal cathode and a thin-film insulating layer. The basic fabrication process of the needle-type phosphate MEA is similar to the ORP and DO sensors described previously [18, 20] and involves five major steps: dicing, etching, metallization, packaging and cobalt electroplating. The process is schematically shown in figure 1. 2
J-H Lee et al
J. Micromech. Microeng. 19 (2009) 025022
(a)
Figure 2. Schematic diagram of the PCB carrier, indicating copper tracers. Shaded area indicates the position of the microelectrode array.
Products, Inc.) with a 35 μm thick layer of copper and a 33 μm thick layer of a dry film negative photoresist. The copper layer was photolithographically patterned and etched in ferric chloride to define electrical traces on the carrier surface. Carriers were cut to size from the patterned board by circuit milling (Quick Circuit 5000, T-Tech) and manually filed down to the exact size. The carrier design is schematically shown in figure 2. Microelectrodes were fixed to carriers using UV-cured epoxy (3301, Loctite, Rocky Hill, CT) and electrical connections to individual sensors were established with conductive silver epoxy (Ablebond 8700E, Emerson & Cuming, Billerica, MA). These interconnects were then covered with UV-cured epoxy for protection. Once electrical connections to each microelectrode were established, a 2 μm thick layer of polypyrrole was electrodeposited on the microelectrodes for insulation (figure 1(d)). The gold layer at the tip was dip coated with low-melting temperature paraffin at 62 ◦ C to shield it from polypyrrole deposition. Polypyrrole was electrodeposited at 6 mA cm−2 or ∼200 nm min−1. The polypyrrole electrodeposition cell contained two stainless steel plates (3 cm × 5 cm) used as counter cathodes, which permitted electrodeposition of polypyrrole on all sides of microelectrodes simultaneously. The gold layer was subsequently exposed by dissolving paraffin in Opticlear. For phosphate sensors, cobalt was electrodeposited on the exposed gold tips. Two cobalt plates 3 cm × 5 cm were used as anodes. The plates were cleaned in 20% HCl for 15 min to remove the cobalt oxide layer. An electrolyte solution was prepared by dissolving 33 g CoSO4 and 3 g H3BO3 in 100 mL of water. The MEA was placed between the cobalt plates at 1 cm distance from each. Cobalt was electrodeposited at 10 mA cm−2 for 2 min, yielding a ∼200 nm thick film. The phosphate MEA was characterized with six different concentrations of standard solution ranging from 10−5.1 to 10−3 M KH2PO4 at ambient temperature. The pH of each standard solution was adjusted to pH 7.5 by adding potassium hydroxide. At this nearly neutral pH, the
(b)
Figure 3. (a) Photograph of an etched glass wafer. (b) Close-up shows probe tips at 900 μm center-to-center spacing.
orthophosphate ions are predominantly HPO42− and H2PO4−. The phosphate MEA was oxidized by immersing in DI water along with a commercial Ag/AgCl reference electrode (MI401, Microelectrodes Inc.) for 30 min, followed by 30 min in 10−4 M KH2PO4 solution. The data acquisition system included the Denver Instruments pH/mV meter (model 225) and BalanceTalk Software (Labtronics Inc.).
3. Results and discussion 3.1. Microfabricated phosphate microelectrode arrays The batch fabrication process was used to fabricate the phosphate MEA sensor. Twelve microelectrode arrays were fabricated from a single 45 mm × 50 mm borosilicate glass wafer. This batch processing substantially simplified fabrication, permitting increased uniformity and consistency of microelectrodes and savings in time and cost. The entire wafer was processed in a single step, yielding 12 arrays. Ten arrays at the center had microelectrodes of uniform shape and length; two arrays near the edge guides exhibited a gradient of microelectrodes due to modified menisci. Although disadvantageous in batch fabrication, this phenomenon may be used fortuitously to fabricate sensors for measuring analyte stratification in a sample. Figure 3 illustrates electrically isolated 12 arrays containing 48 microelectrodes. Figure 4 illustrates a packaged phosphate MEA consisting of four 1 cm long probes at 900 μm center-to-center spacing. At the end of the etch process and prior to cobalt electroplating, dimensions of the resulting microelectrode 3
J-H Lee et al
J. Micromech. Microeng. 19 (2009) 025022
Figure 6. Oxygen sensitivity of the phosphate MEA sensor in 10−3.9 M KH2PO4 at pH 7.5. Figure 4. The fabricated phosphate MEA sensor packaged on a PCB carrier. Inset: SEM of the cobalt-coated phosphate MEA tip illustrating that the tip remains sharp following the electro deposition. Close-up: electrical connections using silver epoxy between the individual microelectrodes and copper traces on PCB.
response with 32 mV–59 mV per decade change of phosphate concentration in the same range [1, 3–6]. De Marco et al [7] calculated that the calibration curve slope of approximately is predicted by the −59 mV/decade change in H2 PO− 4 Nernst equation. The increased sensitivity of the phosphate MEA indicates that the sensor does not exactly exhibit the Nernstian behavior. This is most likely due to the sensitivity to all three orthophosphate ions, the 3D thin-film structure of the sensor and the simple electrical interface. The response time (t90) of the MEA ranged from ∼1 s to ∼30 s as the KH2PO4 concentration was decreased from 10−3 M to 10−5.1 M at pH 7.5. 3.3. Oxygen sensitivity Dissolved oxygen can be expected and has been reported to affect activity of phosphate ions [1, 3]. The sensor response to H2PO4− ions should decrease with increasing oxygen concentration. The effect of dissolved oxygen was evaluated using 0% and 21% DO concentrations in a 10−3.9 M KH2PO4 solution. Nitrogen gas and air were bubbled for 20 min to produce 0% and 21% DO solutions using methods given in [20, 21]). A commercial oxygen milli-electrode (MI-730, Microelectrodes Inc.) was used to verify concentration of oxygen in the test solution and during calibration. Figure 6 shows the ∼69 mV offset between 0% and 21% DO in the same 10−3.9 M KH2PO4 solution. This is a substantial change, significantly higher than the measurement variations (standard deviations were
