June 1, 2007 / Vol. 32, No. 11 / OPTICS LETTERS 1405

Optoelectric biosensor using indium–tin–oxideelectrodes

Chang Kyoung Choi, Kenneth D. Kihm,* and Anthony E. EnglishDepartment of Mechanical, Aerospace, and Biomedical Engineering, University of Tennessee, Knoxville,

Tennessee 37996-2210, USA*Corresponding author: kkihm@utk.edu

Received January 12, 2007; accepted March 9, 2007;posted March 29, 2007 (Doc. ID 78908); published April 26, 2007

The use of an optically thin indium–tin–oxide (ITO) electrode is presented for an optoelectric biosensor si-multaneously recording optical images and microimpedance to examine time-dependent cellular growth.The transmittance of a 100 nm thick ITO electrode layer is approximately the same as the transmittance ofa clean glass substrate, whereas the industry-standard Au �47.5 nm� /Ti �2.5 nm� electrode layer drops thetransmittance to less than 10% of that of the glass substrate. The simultaneous optoelectric measurementspermit determining the correlation of the cell-covered area increase with the microimpedance increase, andthe example results obtained for live porcine pulmonary artery endothelial cells delineate the quantitativeand comprehensive nature of cellular attachment and spreading to the substrate, which has not been clearlyperceived before. © 2007 Optical Society of America

OCIS codes: 180.0180, 170.1530.

Indium–tin–oxide (ITO) is an electrically conductivemixture of indium oxide �In2O3� and tin oxide �SnO2�.It is transparent and colorless in thin layers, while itis yellow to gray in bulk form. ITO thin films aremainly used to make transparent conductive coatingsfor electronic displays (LCD, flat-panel displays, andplasma displays), touch panels, and solar cells [1].Benefiting the combined optical transparency andelectrical conductance of ITO thin films, this studydescribes the development of an optoelectric biosen-sor that allows simultaneous correlation of opticallive cell imaging with cellular microimpedance char-acterization.

In quantifying cellular adhesion and barrier func-tion, one widespread methodology is based on moni-toring microimpedance variations using a gold elec-trode [2]. The underlying principle assumes that thelevel of cell–cell and cell-matrix attachment is a func-tion of the electrical impedance [3,4]. This conjecturehas not been thoroughly examined because of thelack of information on dynamic cellular imaging,mainly attributed to the nontrivial opaqueness of thegold electrode (approximately 50 nm thick for indus-try standard), which transmits only less than 10% ofthe incident light and does not allow simultaneouscellular imaging.

Furthermore, the dynamic imaging of live cells willallow a more comprehensive understanding of cellu-lar physiology, including cell–cell adhesion, cell-substrate adhesion, and cellular membrane proper-ties. Optically thin ITO, which is also stable inelectrolyte cell culture solutions, makes for an attrac-tive alternative to gold as an optoelectric biosensor[5]. While ITO has substantial potential in probingcellular barrier function to both optical and electricaldata correlations, few studies have examined the op-tical functioning and performance of ITO as a cellularoptoelectric sensor [6].

This study describes the optical transmittance andimaging performance of ITO thin films as an opto-electric biosensor. Also presented is the digital image

0146-9592/07/111405-3/$15.00 ©

analysis of evolutionary images of live porcine pulmo-nary artery endothelial cells [7] (PPAECs) on ITOelectrodes in correlation with the simultaneouslymeasured microimpedance profiles.

Figure 1 shows normal transmittance, calculatedfor bare slideglass �I� and Au �47.5 nm� /Ti �2.5 nm�coated slideglass (II) and measured for bare slide-glass (III), 100 nm ITO-coated slideglass (IV), 500 nmITO-coated slideglass �V�, and Au/Ti-coated slide-glass (VI). The transmittances are measured using aCary 5000 UV-Vis-near-infrared (NIR) spectropho-tometer (Version 1.9, Varian) that consists of a doubleLittrow monochromator, two beam path holders (onefor reference and one for sample), PbS/InGaAs detec-tors, and covers from UV �200 nm� to NIR �800 nm�at intervals of 1 nm and a scanning rate of 10 nm/s.

Fig. 1. (Color online) Normal transmittance, calculated forbare slideglass �I� and Au �47.5 nm� /Ti �2.5 nm� coatedslideglass (II), and measured for bare slideglass (III),100 nm ITO-coated slideglass (IV), 500 nm ITO-coatedslideglass �V�, and Au/Ti-coated slideglass (VI). The PCMimages of PPAECs on a bare glass are comparable with thecorresponding images on a 100 nm ITO-coated glass, whileno discernable cellular image is available for Au/Ti-coated

glass under the same illumination intensity.

2007 Optical Society of America

˜

1406 OPTICS LETTERS / Vol. 32, No. 11 / June 1, 2007

Measurements are repeated five times at each wave-length, and the extended bars indicate the resultingstandard deviations.

To validate the experimental data, the transmit-tance of the multilayer thin-film structure is calcu-lated using the characteristic transmission matrixand taking the wave interference effect into consider-ation. The characteristic transmission matrix of themth layer is expressed in terms of s-polarized light asfollows:

Mm�z� = �M11 M12

M21 M22�

m

= � cos�2�

�pmdm� , −

i

pmsin�2�

�pmdm�

− ipm sin�2�

�pmdm� , cos�2�

�pmdm� � ,

�1�

where pm= nm cos �m. The complex angle of incidence�m is given by the generalized Snell’s law, sin �i

= nm sin �m, where i means incident and dm denotesthe thickness of the mth layer. The complex refrac-tive index of the mth layer is defined as nm=nm− i�m, where both refractive index �n� and extinctioncoefficient ��� depend on the wavelength ���.

Assuming all layers are nonmagnetic, the overalltransmission matrix M is defined by multiplication ofthe elementary matrix of Eq. (1) for the total N lay-ers, i.e., m=1,2,3, . . . ,N as follows:

M = m=1

N

Mm�z� = �M11 M12

M21 M22� . �2�

Assigning m=0 for the incident air medium and m=N+1 for the transmitting air medium, the overalltransmittance �T� for multilayers is given as

T =pN+1

p0t2,

where

t =2p0

�M11 + M12pN+1�p0 + �M21 + M22pN+1�. �3�

The calculated transmittance, Eq. (3), is in agree-ment with the measurements for the cases of bareslideglass �I� and gold-coated slideglass (II), for thevisible range in Fig. 1. The transmittance of the ITO-coated sideglass is not calculated because the com-plex refractive index of the present ITO thin film isunavailable. At �=510 nm (green light), the mea-sured transmittance of 88% for 100 nm thick ITOsputtered layer on the slideglass shows a mere 3.2%reduction from the transmittance of 91.2% for thebare slideglass, whereas the transmittance is dra-matically reduced by more than 90% for the case ofthe industry standard Au �47.5 nm� /Ti �2.5 nm�

coated sideglass.

Phase contrast microscopic (PCM) images ofPPAECs inserted in Fig. 1 clearly show that the hightransmittance of an ITO electrode makes it compa-rable with images seen through bare glass, whereasimages seen through the Au/Ti layer, under an iden-tical illumination condition, are imperceptible.

Figure 2 shows the schematic of simultaneous op-toelectric measurement of a dynamic live cellular at-tachment. To effectively use ITO as an optoelectricsensor, the 100 nm thick ITO electrode is laid on aslideglass, sputtering coated with a 300 nm thick sili-con nitride �Si3N4� insulation layer, and chemicallyetched to create a 250 �m diameter opening elec-trode. During the ITO film deposition, a compromisedfabrication process has to be reached, as the opticaltransmittance and electrical conductance are in-versely proportional to each other depending on theprocess parameters.

An Olympus IX-71 inverted platform accommo-dates an incubator, and the dynamic images of livecells are recorded using a 14 bit electron multiplyingcharge-coupled device (EMCCD), which has the com-bined functions of a cooled and intensified CCD. Theac electric impedance is scanned by a lock-in ampli-fier at each of 17 selected frequencies from 10 to105 Hz, and the data acquisition and analysis systemis implemented using LABVIEW. The microimpedancemeasured between the working and the ground ITOelectrode represents three electrical resistance con-tributions associated with cell membrane capacitance

Fig. 2. Schematic of the optoelectric cellular sensor usingan ITO–Si3N4 electrode. The ITO electrode consists of a300 nm thick Si3N4 insulating layer on a 100 nm thick ITOelectrode layer sputtered on a slideglass. An OlympusIX-71 inverted platform accommodates an incubator(WeatherStation, Olympus) for dynamic imaging of livecells using the 14 bit EMCCD. The impedance is dependenton membrane capacitance, cell–cell adhesion, and cell–

substrate adhesion.

June 1, 2007 / Vol. 32, No. 11 / OPTICS LETTERS 1407

(cell physiology), resistance between cells (cell–celladhesion), and resistance under cells (cell-substrateadhesion) as schematically shown in the inset figure.

Figure 3 shows normalized resistance as a functionof the normalized cell-covered areas that have beendetermined based on digital image analysis. Pre-sented are selected 12 data points with time lapses oft=1, 3, 6, 12, 21, 87, 126, 168, 240, 294, 447, and759 min after inoculation. The three inset imagesshow the confluent stages of PPAECs at three differ-ent time lapses of t=6, 126, and 447 min (32�,250 �m�250 �m field of view). The relationship be-tween the measured resistance and the cell-coveredarea delineates the quantitative and comprehensivenature of cellular attachment and spreading to theelectrode substrate. The resistance generally in-creases when gaps between cells (spreading) and/orbetween cells and extracellular matrix substrates(cellular adhesion and/or attachment) are narrowerbut not in a consistent manner as in the aforemen-tioned underlying conjecture of cellular microimped-ance analysis.

The covered area correlation with resistance di-vides into three steps, as shown by the three dividingcircles in Fig. 3 and also schematically illustrated inthe three sketches above the three images: (1) Ini-tially the cell-covered area increases rapidly whilethe electrical resistance does not show any substan-tial increase. This indicates that the majority of cellsare globular without spread or attachment, whiletheir population increases, and the electrical resis-tance increase is very gradual, as the electric currentcan bypass the cells. (2) Next, the electrical resis-

Fig. 3. (Color online) Normalized resistance, �Rc−Rn� /Rn,with Rc being the cell-covered electrode resistance and Rnbeing the naked electrode resistance, as functions of thenormalized cell-covered area �Ac /Ae�, where Ae is the areaof the total examined opening electrode, 0.05 mm2, and Acis the cell-covered area. The total 12 data points at t=1, 3,6, 12, 21, 87, 126, 168, 240, 294, 447, and 759 min andthree representative differential interference contrast mi-croscopy (DICM) images at t=6, 126, and 447 min, includ-ing the schematic of the corresponding cellular attachment,are shown. The DICM images clearly show correlations be-tween the level of cellular confluence and the measured re-

sistance of the ITO electrode.

tance proportionally increases with cell-covered area.This second stage represents the initial globular en-dothelial cells’ transformation to flatter profiles asthey are tightly attached to the electrode over time.(3) Finally, the cell-covered area remains constant atthe more-or-less fully occupied level �Ac /Ae=1.0�, butthe resistance progressively increases with time.This is attributed to the fact that both cell-substrateadhesion and cell–cell adhesion are much tighter,which increases the resistance continually even afterthe whole electrode surface is apparently occupied bycells. Note that cells at the final stage (3) look muchflatter and more tightly packed in comparison withthose of the previous stage (2).

The feasibility of using an optically thinITO–Si3N4 electrode is demonstrated as an optoelec-tric biosensor to simultaneously produce dynamic im-ages of the motion and growth of cells while concur-rently measuring cellular impedance. Compared withthe industry-standard Au electrode, the ITO biosen-sor provides superior optical transmittance that per-mits live cellular imaging while also being suffi-ciently sensitive to function properly as amicroimpedance sensor. Furthermore, the ITO elec-trode patterned by the Si3N4 layer allows for re-peated use since both materials (ITO and Si3N4) areundamaged by ethanol cleaning or de-ionized water-sterilization processes, while standard Au electrodesare significantly deteriorating after repeated clean-ings. Additionally, the ITO electrode is simpler to fab-ricate. ITO can be directly layered on a coverglasssubstrate, while Au requires an additional metal-coated layer (Ti or Cr) to promote its adhesion to theglass surface.

In summary, the ITO silicon-nitride electrode opto-electric biosensor has strong potential for compre-hensively monitoring intricate cellular activities, in-cluding future study of dynamic morphology andadhesion of angiogenic cells, particularly endothelialcells, cancer cells, and stem cells.

This work was supported in part by the NationalScience Foundation under grant BES-023905 (A. E.English and C. K. Choi), by the Korea Institute ofScience and Technology Evaluation Policy (K.D.Kihm), and by the American Heart Association undergrant 0265029B (A.E. English). The authors thank S.Jun and P. Rack for their help with constructing ITOelectrodes.References

1. M. Katayama, Thin Solid Films 341, 140 (1999).2. I. Giaever and C. R. Keese, Proc. Natl. Acad. Sci.

U.S.A.81, 3761 (1984).3. W. Tschugguel, Z. Zhegu, L. Gajdzik, M. Maier, B. R.

Binder, and J. Graf, Pfluegers Arch.430, 145 (1995).4. C. Xiao, B. Lachance, G. Sunahara, and J. H. T. Luong,

Anal. Chem.74, 5748 (2002).5. S. Ray, R. Banerjee, N. Basu, A. K. Batabyal, and A. K.

Barua, J. Appl. Phys.54, 3497 (1983).6. H. Hillebrandt, A. Abdelghani, C. Abdelghani-Jaquin,

M. Aepfelbacher, and E. Sackmann, Appl. Phys. A 73,539 (2001).

7. C. K. Choi, A. E. English, S. Jun, K. D. Kihm, and P. I.

Rack, Biosens. Bioelectron. 22, 2585 (2007).