Modeling an Enzyme Based Electrochemical Blood Glucose

Sensor
Stephen A. Mackintosh*1, James I. Rodgers1, Stephen P. Blythe1
1
Lifescan Scotland Ltd, Inverness, IV2 3ED, Scotland.
*Corresponding author email: [email protected]

Abstract: This paper describes the modelling of 2. Model Development

a blood glucose sensor using COMSOL
Multiphysics. Chemical species interaction and 2.1 Geometry and Reagents
diffusion, coupled with electrochemical
oxidation of multiple blood species produced a The simplified geometry comprises two
powerful working model used in developing and electrodes, a working and counter, over-coated
refining a range of blood glucose sensors for the with a thin deposited reagent layer composed of
commercial market. an oxidoreductase enzyme and a redox active
electrochemical mediator, all located within a
Keywords: diabetes, electrochemistry, diffusion, sample capillary volume, which is filled with a
blood, biosensor blood sample.

1. Introduction

Self monitoring blood glucose (SMBG) test

systems are essential in the management and
control of diabetes[1]. This paper describes the
modeling of the glucose dependent, signal-
generating, reactions that occur within the blood-
filled reaction chamber of an example SMBG
test strip during a blood glucose concentration
Figure 2a. Test strip with capillary volume
measurement.
highlighted in red.

Figure 2b. 2D cross section of interior of capillary

volume, comprising electrodes (working and counter),
reagent layer and sample void.

2.2 Amperometric Reactions

Figure 1. Blood Glucose Measurement using SMBG
strip and meter The glucose analyte within blood is
enzymatically oxidised to gluconolactone, with
This model has built upon legacy work at oxidised electrochemical mediator (potassium
Lifescan Scotland initially based around ferricyanide) being consumed, and reduced
analytical solutions to partial differential mediator (potassium ferrocyanide) being
equations on simplified domains and boundary generated. This reduced ferrocyanide
conditions. Later work extended these models to subsequently diffuses toward the polarized
include multiple species, chemical interactions working electrode where it is electrochemically
and full electrochemistry. This COMSOL model re-oxidised to ferricyanide, while an equivalent
has allowed additional flexibility in rapid model amount of the excess ferricyanide is
based prototyping, involving alternative chamber simultaneously reduced at the counter electrode,
geometries and/or reagent compositions.
completing the circuit and producing an assumptions that might to be made in order to
electrochemical current signal[2,3]. simplify domains and boundary conditions so
that a solution can easily be computed.

When finite and/or complex geometries are

required to describe the system, numerical
models become more desirable.

2.4 Theory

Figure 3. Mediated enzymatic electrochemical

The physics used to build the more complete
reaction. numerical models included Fick’s diffusion
law[4] coupled with Michaelis-Menten based
chemical interactions[5]. A Bultler-Volhmer
2.3 Legacy work with analytical models expression was used to provide concentration
dependent changes to the base driving potential
Early modeling work on this problem involved of the system[6]. Conservation of mass was
finding solutions (where possible) to partial implicit in the system with ferrocyanide and
differential equations with simplified domains ferricyanide ions exchanging electrons and
and boundary conditions. Detailed chemical driving the base amperometric response.
interactions and electrochemistry were ignored.

An example of this was the modeling of a gel

layer assumed thin in comparison to the cavity [4]
height. The resulting PDE with simplified
boundary conditions could be solved analytically Here S represents the concentration of one of
via the application of Laplace Transforms. Here potentially many diffusing species in the
is the resulting current-time profile for such a simulation (for example glucose), Ds is the
solution: diffusion coefficient of this species (which may
vary across the spatial domain according to the
materials employed), x is the generic space
D1  ∞
 m 2 l 2 
I (t ) = nFAC0 1 + 2∑ α exp−
m
 vector and t is time.
πt  m =1  D1t 

Here D1 is the diffusivity in the pad, l is the [5]

reagent thickness, A is the electrode area, F is
Faraday’s constant, n is the number of
participating electrons, m is the summation index Here P is the concentration of a generated
and product, Vmax is the characteristic limiting rate of
the enzyme-substrate reaction, Km is the
Michaelis-Menten constant for the enzyme-
1 − D1 substrate combination and S is the concentration
D2
α= of substrate.
1 + D1
D2

Where D2 is the diffusivity in the cell cavity. [6]

The strength of this approach is the power and Here io is the electrode-redox couple exchange
utility of a general solution that encapsulates the current density, CR is the concentration of
main variables of the system. The weakness is reduced species, Co is the concentration of
the rarity of simple solutions and the oxidized species, R is the universal gas constant,
F is Faraday’s constant, T is the temperature in 500mg/dL glucose. Higher concentrations show
kelvin, α is the transfer coefficient, η is the over- greater current response in the plot.
potential.

3. Use of COMSOL Multiphysics

Concentration of diluted species, secondary

current distribution (battery and fuel cells),
chemical reactions, and other boundary
conditions using the custom mathematical
equation editor were employed. Mesh
generation was automated via Comsol’s physics
controlled mesh setting, with extra refinement
nodes added near each electrode-electrolyte
boundary.
Figure 5. Linearity plot showing sensor response
against glucose.
1, 2 and 3D versions of the models were
developed allowing rapid solutions for more
simplified symmetric geometries. 2 and 3D
versions were employed where specific detail
was required to capture the impact of geometric
anomalies on the system.

Solutions were always time dependent due to the

finite test times required from commercial
biosensors (typically 5 seconds or thereabouts). Figure 6. Concentration profile showing the
distribution of ferrocyanide in the cell at a given time
4. Results point. Local deficiency of ferrocyanide over the
working electrode and its regeneration next to the
The model has been used to characterise existing counter electrode is clearly evident.
and new biosensor geometries. Outputs such as
the correlation between analyte concentration 5. Conclusions
and current signal at different time periods, and
using alternate layouts, has aided with cost Results achieved in this simulation were in
effective optimisation of device chemistry and general agreement with experimental work using
geometry. The detailed concentration gradients existing systems and prototypes. Whilst such
produced have increased understanding of signal models can never be used to facilitate decisions
features and the impact of design changes on the safety or efficacy of medical devices
thereupon. released to the public, they are expected to prove
a useful tool in the design and optimisation of
such devices in the future.

6. References

1. American Diabetes Association, Standards of

Medical Care in Diabetes 2012, Section 5(c).

2. Asha Chaubey *, B.D. Malhotra, Mediated

Biosensors, Biosensors & Bioelectronics 17
(2002) 441–456
Figure 4. Model chronoamperometric response
3. Mediated Electrochemistry – A Practical
current at concentrations of 50, 150, 250 and
Approach to Biosensing, Cardosi M F & Turner
APF in Advances in Biosensors Volume 1
(1991) JAI Press Ltd

4. A. J. Bard & L. R. Faulkner, Electrochemical

Methods 2nd Ed, Page 148-155

5. Menten, L.; Michaelis, M.I. (1913), "Die

Kinetik der Invertinwirkung", Biochem Z 49:
333–369

6. A. J. Bard & L. R. Faulkner, Electrochemical

Methods 2nd Ed, Page 92-107

7. Acknowledgements

The authors would like to thank the following

people working for the next-generation strip
platform division of R&D at Lifescan Scotland:

Dr. Marco F. Cardosi, Research Fellow

Gavin Macfie, Staff Scientist
Adam Craggs, Senior Engineer (Algorithm)
James Moffat, Senior Engineer (Mechanical)