Micron Sized Glucose Sensors for
In-Vitro Monitoring of Glucose
in Interstitial Fluid

William McDonald and Raoul Kopelman
Department of Chemistry
University of Michigan
Ann Arbor, MI 48109-1055
313-647-9992 voice
313-647-4865 fax
williamf@umich.edu or
Kopelman@umich.edu

Introduction

The development of glucose sensors displaying high sensitivity, fast response, reproducibility, and long term stability has been a main target in sensor research during the last three decades. Amperometric glucose sensors were first developed in the 1960’s and continue to receive considerable attention. Significant effort is devoted to the development of a new and improved amperometric sensors that will continuously monitor physiologically important levels or concentrations of glucose over extended periods of time. Interestingly, a new direction of amperometric glucose sensors has been towards needle type sensors for long term implantation in diabetic patients.

More recent sensor research has seen the development of micrometer and submicrometer sized sensors, particularly optical based sensor systems for non- and minimally invasive glucose determinations. Miniaturized glucose sensors are particularly attractive in connection with clinical applications. Specifically, measurements of glucose in very small volumes or monitoring of localized events where good spatial resolution is required. An alternative approach to the ultra-microelectrodes is the fabrication of fiber optic chemical and biochemical sensors.

Fiber optic chemical and biochemical sensors typically utilize a fluorescent probe immobilized in a matrix (e.g. inorganic or organic) which is then quenched by oxygen or interacts with an analyte of interest in a way that provides quantitative information about the analyte or the oxygen levels. The advantages of the fiber optic chemical sensors or optodes over electrochemical sensors include miniaturization, geometric flexibility, and the lack of electrical contact between the sensor and sample. The main problems with optodes are their limited chemical and photochemical stability. Other challenges to the use of optodes includes the ability to “back” detect the signal or detecting the sensor response on the same end of the fiber as the excitation or light source. This requires the use of bifurcated fibers or fibers split into a “Y” configuration and very sensitive detectors and spectrometers.

The University of Michigan, Ann Arbor has and continues to develop sensor technology that will contribute significantly to glucose monitoring technology. This glucose monitoring technology is related to the use of nano-optodes with specialized sensor coatings applied to the very tip of this optode. This special coating interacts with glucose through immobilized glucose oxidase which consumes oxygen through a series of reactions shown in Equations 1 and 2. This oxygen consumption is the heart of the glucose monitor or sensor. A sensitive dye detects these oxygen concentration changes. The use of a bifurcated fiber and a miniaturized spectrometer system for back detection round out our sensor system for, ultimately, physiological glucose determinations.

Ru (Phen) + glucose oxidase + glucose + O₂ + peroxide ®     Equation 1

glucose oxidase + gluconic acid + water ® increased fluorescence    Equation 2

The Glucose Sensor

The glucose sensor is made of several components including the optical fiber, the sensor coating, and the detection system. The optical fiber is a standard silica fiber and can be either single or multimode. The fiber core is clad with doped silica and buffered with an acrylate coating for improved mechanical stability. A small amount of the buffer is stripped from each end of the fiber to prepare it for the pulling process to create the nano-optode. The fiber is placed in a micropipette puller (P87 from Sutter Instrument Co.) where it is simultaneously heated with a CO₂ laser and pulled with a pretensile force of 4 N. Single mode fibers with 3 micron core sizes typically produce tips well below 1 micron in size, while multimode fibers with a core size of 100 microns will typically produce tips sized below 10 microns. These pulled fibers can have the sensor attached through photopolymerization directly or can be metal coated in a vacuum deposition process (10^-6 torr) before sensor attachment. The metallization coats only the tip sides leaving the tip face as a transmissive aperture by rotating the tips that are held at a 30 degree angle relative to the metal source during the metal coating process. If the tips are metallized, they are first coated with chromium to improve the adhesion of the much thicker aluminum coating.

Before the sensor coating is attached to the pulled fiber tip, the tip is silanized with gamma-methacryloxy propyl trimethoxy silane to promote adhesion of the sensor coating to the fiber tip. The sensor coating is attached to the pulled and silanized fiber tip through the photopolymerization of acrlyamide with a small amount of a di-functional monomer (NN' methylene bisacrylamide), the photo-initiator (triethylamine or TEA), the fluorescent dye (1,10-phenanthroline ruthenium (II) chloride), and the enzyme (glucose oxidase). The photopolymerization can be initiated with either the 488 nm or the 514 nm line of the argon ion laser. Care is taken to adjust the laser intensity, photopolymerization time, and photoinitiator concentration to allow for a useful polymerization time and sensor tip size. Only a small amount of sensor coating is required to provide adequate signal when measuring glucose concentrations in-vivo or in-vitro.

One mode of sensor operation is transmissive, using an inverted microscope, a monochromator, and a CCD with the appropriate neutral density and band pass filters generating a signal in a transmissive mode that is very readily detected. This inverted microscope setup can not be used when glucose levels are monitored in turbid environments (yeast and cell cultures) or when the sensor tip is inserted just under the skin to monitor glucose levels in the interstitial fluid. This requires a back detection methodology that utilizes a bifurcated fiber assembly to couple the laser to the sensor and then couple the sensor signal back to the detector system. Specifically, the light is focused onto the monochromator slit. This back detection setup can have a good deal of loss associated with it, but if the bifurcated fiber is coupled appropriately to the laser and sensor, these losses are minimized. To date, back detection has been accomplished with pulled single mode sensors (e.g. sensor tip size is less than 2 microns) and pulled multi-mode sensors (e.g. sensor tip size is less than 10 microns). The use of the fiber coupled miniaturized spectrometer (Ocean Optics S2000) improves the detectable signal levels on the bifurcated fiber significantly allowing for the use of blue light emitting diodes (450 to 470 nm).

The most important aspect of the sensor tip is the immobilized glucose oxidase since it provides the needed reaction or interaction with glucose consuming oxygen along the way. The glucose oxidase is added to the photopolymerization solution and is immobilized during the crosslinking reaction(s) of the polymer. The amount of glucose oxidase used is 100 microliters of 225 unit solution of glucose oxidase to 600 microliters of the polymerization mixture (30 % solids).

As mentioned above, a ruthenium based fluorescent dye is used to detect the oxygen concentration changes caused by the glucose oxidase - glucose reaction shown in Equations 1 and 2. Currently, this dye is excited with the 488 nm line of the argon ion laser and emits at approximately 590 to 650 nm in a relatively broad emission spectrum. Moving to the blue led will allow us to eliminate the laser and make a truly portable glucose monitoring system. Figure 1 provides a diagram of our sensor system including the light source, optics for coupling the laser to the fiber, bifurcated fiber, sensor and sensor coupler, filter for eliminating the source light at the detector, and the spectrometer / detector.

Figure 2 illustrates a glucose level calibration curve with fluorescence intensity in arbitrary units plotted as a function of glucose concentration changes in solution for the pulled multimode sensor (sensor tip size <10 microns). The data were collected by successively changing the glucose concentration from 0 to 20 mM in 5 mM increments. A 5 minute waiting period is employed to allow the solution and new glucose addition to equilibrate before the glucose concentration is measured. Instrumental parameters vary as a function of light source to fiber and fiber to fiber coupling efficiencies and the amount of of fluorescent coating present on the fiber tip, however, typical integration or exposure times are 500 milliseconds or less.

The line connecting the data points is placed simply to aid the reader in observing the linearity of the data and has not been “fit”. If it were fit, we would likely choose to use linear regression since the data is quite linear. This allows for very simple algorithms to provide glucose concentrations from the fluorescence data.

Summary:

Glucose monitoring is an important sensor application for business, technical, and humanitarian reasons. Sensors that are smaller and less invasive, preferably non-invasive, are very attractive since the subject experiences no pain or significantly less pain due to the lowered invasiveness of the optode sensor.

The sensor technology developed by the University of Michigan provides this sensor technology in the form of the optode or nano-optode with glucose oxidase and a ruthenium based fluorescent dye immobilized on the sensor tip. Initial work clearly shows the capability to monitor glucose in-vitro with these sensors. Additional work focuses upon improvements to the dynamic range of the sensors, the immobilization methodology, glucose oxidase activity, and conducting in-vivo tests through a partnership with Microsense of St. Louis, MO and the Endocrinology Department at Washington University also in St. Louis.

References:

1. Rosenzweig, Z.; Kopelman, R.; Sensors and Actuators; B Chem.; 1996, 35-36, p.475-483.

2. Fiber Optic Chemical Sensors and BioSensors; Volume II; ed. Wolfbeis, O.S.; 1991, CRC Press, Boca Raton, FL; Chapter 10 Oxygen Sensors; contributed by Wolfbeis, O.S.

3. Walt, D.R.; Lin, L.; Anal. Chem.; 1995, 67, 3746-3752.

4. Bambot, S.B.; Rao, G.; Romauld, M.; Carter, G. M.; Sipior, J.; Terpetchnig, E.; Lakowicz, J.R.; Biosensors and Bioelectronics; 1995; 10; 6/7; 643-652.

5. Kunzelmann, U.; Bottcher, H.; Sensors and Actuators; B Chem.; 1997, 38/39, 222-228.

6. McDonagh, C.; MacCraith, B.D.; McEvoy, A.K.; Anal. Chem.; 1998, 70, 45-50.

7. Lin, A.L.; Feldman, M.S.; Kopelman, R.; J. Phys. Chem. B; 1997, 101, p. 7881-7884.

8. Service, F. J.; O’Brien, P.C.; Wise, S.D.; Ness, S.; LeBlanc, S.M.; Diabetes Care, 1997, 20, 9, p. 1426 - 1436.

9. Rosenzweig, Z.; Kopelman; Anal. Chem.; 1995, 67, p. 2650 - 2654.

10. Rosenzweig, Z.; Kopelman; Anal. Chem.; 1996, 68, p. 1408 - 1413.

Back to In This Issue . . .