Article pubs.acs.org/ac
Miniature Amperometric Self-Powered Continuous Glucose Sensor with Linear Response Zenghe Liu, Brian Cho, Tianmei Ouyang, and Ben Feldman* Abbott Diabetes Care, 1380 South Loop Road, Alameda, California 94502, United States ABSTRACT: Continuous glucose measurement has improved the treatment of type 1 diabetes and is typically provided by externally powered transcutaneous amperometric sensors. Selfpowered glucose sensors (SPGSs) could provide an improvement over these conventionally powered devices, especially for fully implanted long-term applications where implanted power sources are problematic. Toward this end, we describe a robust SPGS that may be built from four simple components: (1) a lowpotential, wired glucose oxidase anode; (2) a Pt/C cathode; (3) an overlying glucose flux−limiting membrane; and (4) a resistor bridging the anode and cathode. In vitro evaluation showed that the sensor output is linear over physiologic glucose concentrations (2−30 mM), even at low O2 concentrations. Output was independent of the connecting resistor values over the range from 0 to 10 MΩ. The output was also stable over 60 days of continuous in vitro operation at 37 °C in 30 mM glucose. A 5day trial in a volunteer demonstrated that the performance of the device was virtually identical to that of a conventional amperometric sensor. Thus, this SPGS is an attractive alternative to conventionally powered devices, especially for fully implanted long-term applications.
T
miniaturized, fully implanted glucose sensors has mainly been limited by the substantial sizes of coimplanted electronic parts and the batteries powering them. Therefore, it is desirable to develop a sensor that derives energy from the body’s internal glucose supply and does not require an additional power source to function. Self-powered glucose sensors (SPGSs) have been described before in which glucose is oxidized at the anode and O2 is reduced at the cathode. This thermodynamically favorable reaction may be made to proceed at an appreciable rate, in the absence of any applied potential or any active electronics, through the judicious choice of electrode catalysts. A first example was provided by Willner and co-workers.5,6 They described a self-powered glucose sensor with an anode consisting of GOx reconstituted on an flavin adenine dinucleotide (FAD)-modified monolayer and a layered crosslinked cytochrome c/cytochrome oxidase cathode. The open circuit voltage (Voc) developed between these two electrodes increased with glucose concentration, in the absence of any external power source. This first example of an enzyme-based self-powered glucose sensor was a notable accomplishment. However, this approach has a number of practical disadvantages. Monolayer electrodes are difficult to fabricate and tend to have poor operational stabilityin this case 5 h at 30 °C. More
he current standard of care for treatment of type 1 diabetes mellitus includes measurement of blood glucose levels at least four times daily, coupled with insulin injections to stabilize glucose levels in the normal range. However, a number of studies have shown that continuous glucose measurement, with updated blood glucose concentrations available every 1−5 min, can improve blood glucose control.1−3 Commercially available continuous glucose sensors have recently been reviewed.4 Without exception, these devices are transcutaneous amperometric sensors, with the glucose sensing element residing a few millimeters below the skin surface in the subcutaneous fat. Glucose-derived electrons are first captured by glucose oxidase (GOx) and then passed to either an Oscontaining redox polymer (the “Wired Enzyme” approach) or, more commonly, to O2, with subsequent oxidative detection of H2O2. Unfortunately, transcutaneous amperometric sensors have a limited lifetime; at present, the maximum is 7 days for a commercially available product. This lifetime is limited not only by the inherent instability of the sensor but also by the risk of infection from the skin-breaching sensor and the difficulty in keeping the sensor adhesively affixed to the skin for longer than 1 week. Fully implanted sensors in which the entire apparatus is beneath the skin are an attractive alternative to the transcutaneous approach, with potentially much longer lifetimes, since failure modes associated with infection and poor adhesion are reduced. Unfortunately, the development of long-lived, © 2012 American Chemical Society
Received: January 20, 2012 Accepted: February 27, 2012 Published: March 14, 2012 3403
dx.doi.org/10.1021/ac300217p | Anal. Chem. 2012, 84, 3403−3409
Analytical Chemistry
Article
Figure 1. Structures of polymers employed. (A) GOx-wiring polymer; (B) glucose flux−restricting polymer.
importantly, the reported sensor has a response that is both small and decidedly nonlinear over the analytical range of interest, with the generated Voc varying from about 5 mV at 1 mM glucose to about 50 mV at 25 mM glucose. Similarly, Sode and co-workers7 described an SPGS consisting of (a) an anode constructed from several subunits of glucose dehydrogenase (GDH) from Burkholderia cepacia adsorbed onto carbon paste and (b) a Pt cathode. Also using Voc as the output signal, this SPGS developed a nonlinear response that was saturated at 6 mM glucose. An ingenious elaboration8 of this design produces sufficient current to power an attached transmitter. Both6,7 of the above-mentioned SPGSs were designed to operate under open circuit conditions and thus did not generate (or consume) any power. For the purposes of power generation, there is precedent for an enzyme-based biofuel cell that does generate power and also develops a much more substantial potential. Heller and co-workers9−11 described a series of glucose-O2 biofuel cells based on redox polymer-wired GOx anodes. These fuel cells operated under physiologic conditions at potentials and current densities of about 0.5 V and 1 mA/cm2, respectively. From this initial work, it became apparent that, in a miniature glucose-O2 biofuel cell, the current becomes glucose flux controlled when the glucose flux to the anode is much smaller than the O2 flux to the cathode. We describe here an SPGS consisting of a multilayer Os redox polymer-wired GOx anode and a Pt/C cathode, separated by a load resistor, and overlaid with a derivatized poly(vinylpyridine) glucose flux−reducing membrane. In contrast to the open circuit sensing described by Willner and co-workers and Sode and co-workers, this system operates in an amperometric mode by measuring the current flowing through a load resistor that connects the anode and the cathode. The current is independent of this resistance between ca. 1 and 10 MΩ and it varies in a linear fashion with glucose concentration over a range of 1−30 mM. The current response is in fact identical to that observed when the above Os redox polymer-
wired GOx anode is poised, as the working electrode in a conventional three-electrode cell, at +40 mV vs Ag/AgCl. The system has exceptional operational stability, with less than a 10% drift in response to 30 mM glucose in a phosphatebuffered saline (PBS) buffer solution at 37 °C, over 60 days of continuous operation.
■
EXPERIMENTAL SECTION Materials. Two types of sensors, gold and carbon, were used in this study. Gold sensors on polyimide substrates were fabricated lithographically. A dielectric insulator layer over the gold layer formed two wells that defined the exposed electrode areas for both the anode (about 0.1 mm2) and cathode (about 0.15 mm2). Carbon sensors on polyester substrates were prepared through screen printing and obtained from Steven Label, Inc. (Santa Fe Springs, CA). The exposed anode and cathode areas (both carbon) were about 0.5 and 1.5 mm2, respectively. Active electrode areas were determined by the controlled deposition of glucose-oxidizing or oxygen-reducing catalysts (see below). Two proprietary polymers for GOx wiring (A) and for the glucose flux−restricting membrane (B) were synthesized according to published procedures (Figure 1).12,13 Another membrane polymer, poly(4-vinylpyridine) (Mw = 160 000; catalog no. 472352), and the carbon nanopowder (
