Radio Frequency CMOS Chem-bio Viscosity Sensors based on

Dielectric Spectroscopy

Subhajit Guha and Christian Wenger

IHP, Leibniz Insititute for Innovative Microelectronics, Frankfurt (Oder), Germany

Keywords: CMOS Viscosity Sensor, Dielectric Sensor, RF Viscosity Sensor, High Frequency Viscosity Sensor.

Abstract: This paper presents a CMOS Radio frequency dielectric sensor platform for the detection of relative viscosity

changes in a fluid sample. The operating frequency of the sensor is 12.28 GHz. This frequency range has been

chosen for high signal to noise ratio and also to avoid other low frequency dispersion mechanisms for future

lab on chip applications. The sensor chip has been fabricated in 250 nm BiCMOS technology of IHP. The

measurements conducted to show the relative viscosity variation detection capability of the sensor chip, were

based on mixtures of glycerol and water as well as glycerol and organic alcohol. The detection limit of

viscosity is dependent on the permittivity contrast of the sample constituent. Therefore, it is also shown the

choice of frequency inherently aids in the permittivity contrast of the sample constituents.

1 INTRODUCTION

Viscosity sensors are widely developed and utilised

primarily for sensing and analysis of oil or fuel

(Agoston et al., 2005, Perez and Hadfield, 2011).

There is a significant use of such sensors in the

automotive industry where detection of

contamination or breakdown of oil is of extremely

high relevance. Therefore, there is a wide range of

commercially available robust viscosity sensors. The

working of such sensors are mostly based on

mechanical or piezoelectrical sensing mechanisms

(Agoston et al., 2005, Thalhammer et al., 1998, Shih

et al., 2001, Brouwer et al., 2011). The use and

success of viscosity sensors have inspired researchers

to use the same principles for various other

applications.

In the recent past, there have been efforts to use

the principles of viscosity sensors in biosensing

applications. Various research groups have designed

and developed sensors to estimate glucose

concentration in blood (Boss et al., 2012, Zhao et al.,

2007) or analyse blood coagulation using the concept

of viscosity change (Cakmak et al., 2014). The

sensing principle for these developed viscosity based

biosensors are also mechanical or piezoelectrical,

influenced directly from the established commercial

viscosity sensor. The mechanical principle is based

on the deflation of cantilevers or beams when a fluid

comes in contact with the same. The viscosity of the

fluid determines the degree of deflation or

deformation of the beam. A significant drawback

suffered by this kind of sensors is the relaxation of the

beam or the cantilever back to the original position.

Therefore, there can be considerable calibration

issues for such devices. Piezoelectric based sensors

require a piezomaterial, where a variation of

mechanical stress on the piezo material leads to an

electrical output, such as voltage or current,

depending on the read out mechanism. The adhesion

or the influence of the biological sample on the

piezomaterial pose a serious challenge for such

sensors. Recently, MEMS based CMOS compatible

viscosity sensor was shown by (Cakmak et al, 2014).

working on the same principle of micro-cantilever

deformation. Although, reduction in sample volume

was achieved, the cantilever approach could still

suffer from calibration failures. Impedance

spectroscopy method was discussed by (Perez and

Hadfield, 2011), where the authors described a

viscosity sensor based on the detection of variation of

permittivity. The frequency range of operation is in

the order of few 1 MHz. Application of this sensor in

biological applications will be critical, as there are

other dispersion mechanisms of biological samples in

this frequency range which could influence the signal

to noise ratio for viscosity characterization.

In this work, we propose CMOS on chip dielectric

sensor for viscosity characterization. The frequency

range of operation is in the order of 12 GHz. In

142

Guha S. and Wenger C.

Radio Frequency CMOS Chem-bio Viscosity Sensors based on Dielectric Spectroscopy.

DOI: 10.5220/0006151901420148

In Proceedings of the 10th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2017), pages 142-148

ISBN: 978-989-758-216-5

previous works, it was shown that such sensors can

be used for the detection of concentration of cells and

particles in a suspension (Guha et al., 2015), detection

of biomarkers like creatinine and more (Guha et al.

2015). The sensing principle is based on the detection

of dielectric change using a capacitive sensor

embedded in a resonant CMOS oscillator. In this

work, it is proposed that the viscosity variation of

fluid samples can be measured by the detection of the

change in the dielectric constant of the fluid. Such a

sensor can be envisaged for future lab on chip

application like characterization of sputum for

detection of lung diseases. In this paper we will

describe the working principle of the sensor and show

the operation of the sensor with mixtures of liquids

with different viscosities. The operating frequency of

the sensor is chosen to be 12 GHz; this high frequency

leads to small chip size and hence small sample

volume. Additionally, this frequency range enables a

better signal to noise ratio for sensing due to strong

contrast in dielectric permttivities.

The paper is organised in the following manner:

Section 2 describes the sensor and the high frequency

operation. Section 3 shows the results. The

concluding remarks and the application of such a

sensor is described in section 4.

2 RF SENSOR OPERATION

The sensor operation is based on the detection of

dielectric constant of the sample using a capacitive

sensor embedded in the CMOS oscillator. The

variation of the capacitance due to the permittivity of

the samples leads to the tuning of the resonant

frequency of the oscillator. Oscillator based dielectric

sensors have been researched upon considerably for

the past years and the working principle can be found

extensively in literature (Guha et al., 2015, Jamal et

al., 2014, Wang et al., 2010). In this work, we extend

this sensor platform for viscosity detection.

Although, there are considerable amount of

viscosity sensors, CMOS based viscosity sensors are

still not well established. Such sensors could be

extremely useful for lab on chip applications. For

example, sputum samples from different patients

have different viscosities and the detection of the

same can enable us to detect lung diseases at a very

early stage. Viscosity change of a liquid sample can

be detected from the permittivity. This is due to the

relationship between the permittivity of the overall

liquid and the fraction of its various constituents. For

example, a glycerol water mixture sample has

different viscosities depending on the amount of

water in the mixture. At the same time, different water

content in the mixture leads to different permittivity

of the sample. Therefore, there exists a correlation

between the changes in viscosity with the change in

the effective dielectric constant of the sample. It

should be also noted here, that the knowledge of the

fluid constituents of the sample is needed for this kind

of indirect viscosity sensing. In this context, the

question arises if high frequency sensor is of

relevance for the detection of viscosity change. We

have seen in literature that high frequency dielectric

sensors are extremely useful for the detection of

concentration of particles in the suspension (Guha et

al., 2015). In the case of viscosity sensor based on

dielectric measurements, a high signal to noise ratio

can be obtained if the contrast between permittivities

of the constituents are considerably high. Considering

the glycerol water mixture example again, we find

from the literature that the low frequency permittivity

of water is around 80 and that of glycerol is 42.5. This

provides a considerable amount of contrast in the

dielectric permittivities. Therefore, a low frequency

detection could be useful. However, if the sensor

system has to be a unique platform applied to lab on

chip applications, it can be extremely critical at low

frequency to obtain a high signal to noise ratio. This

is because, the biological samples show dispersion

mechanisms at low frequencies and can lead to

considerable variation in the results. Therefore, a

chosen operating frequency of 12 GHz can be useful.

At this frequency considering the glycerol water

mixture as an example again, the permittivity contrast

is still of the same order.

Figure 1: Permittivity and viscosity variation of glycerol

and water mixture with respect to water content at 12 GHz.

The permittivity of glycerol is approximately 17 and

that of water is 40 at this frequency range. Therefore,

operating the sensor in this frequency range provides

the advantage of obtaining the permittivity contrast

and at the same time avoids the low frequency

Radio Frequency CMOS Chem-bio Viscosity Sensors based on Dielectric Spectroscopy

143

dispersion mechanisms. Fig 1 shows the variation of

permittivity and viscosity as a function of water

content in the solution at 12 GHz. Mixture theory was

used to calculate the effective permittivity of the

mixture and the effective viscosity.

Figure 2: Permittivity and viscosity variation of glycerol

and water mixture with respect to water content at 60 GHz.

In case of higher frequency of operation of the sensor,

there is a loss on the dielectric contrast. This is shown

in Fig. 1 and Fig. 2. At 12 GHz, the variation in

permittivity of the glucose water mixture due to water

content variation is considerably higher when

compared to the variation of permittivity of the same

at 60 GHz. However, the variation in the viscosity of

the mixture due to its water content is uniform at both

the frequencies, as viscosity is a frequency

independent parameter.

Figure 3: Chip Photograph of sensor chip.

Additionally, the inherent advantage of the high

frequency sensor is the sensor chip size and in turn

the sample volume required. The sensor chip photo is

shown in Fig. 3.

The sensors as seen from a closer look are

interdigitated capacitors (IDC) with over-all size of

50 µm x 50 µm. Therefore, sample volume of the

order of µl is required for detection. The planar IDC

is a fringing field sensor; the fringing electric fields

between the adjacent fingers are utilised to detect the

dielectric permittivity of the sample placed on top of

it. The penetration depth of the fringing fields is the

function of the geometry of the IDC structure. For

homogeneous solutions, where, the permittivity is

uniform all over the solution the penetration depth is

not of significance. Therefore, it can be understood

that the IDC structures can be used for accurate near

field sensing approaches. The sensor chip was

fabricated in the 250 nm technology of IHP , with

max

of 180 GHz/250 GHz. The overall chip size 1.8

mm x 3 mm. The larger dimension of the chip in spite

of smaller sensor area will be explained in the

subsequent experimental and results section.

The sensor architecture details can be found in

literature. In Fig. 4, the schematic of the sensor

architecture is shown.

Figure 4: Sensor architecture: Sensor IDC embedded in

CMOS oscillator.

As mentioned above the sensor capacitor is embedded

in the oscillator and the read out parameter of the

sensor is the oscillating frequency of the sensor.

3 EXPERIMENTS AND RESULTS

RF boards were prepared for measurements with

viscous liquids. The sensor chip was wire bonded to

the board. For the first measurements with the

viscous liquid samples no mircrofluidic integration

was done. Therefore, one major challenge was

encountered while the measurement of liquids, where

the bond wires used or the bonding of the chip could

be broken due to interaction with the fluid. In order to

avoid this scenario, measures were taken to protect

the bond wires. This was done by sealing the chip pad

area with a dielectric material. This is shown in Fig.

BIODEVICES 2017 - 10th International Conference on Biomedical Electronics and Devices

144

Figure 5: Sensor chip with the dielectric material protecting

the bond pads and bond wires.

The dielectric medium prevents the bond wires

coming in direct contact with the liquid medium and

therefore, no breaking of bond wires can take place.

This kind of an approach adds an additional degree of

robustness to the chip, wherein, the sensor chip can

be dipped in the liquid for accurate calibration of the

chip. This method also aids in the cleaning of the chip

using ethanol. Therefore, after conduction of every

measurement the sensor area can be cleaned for the

next set of measurements. This is a significant issue

when such kinds of sensors are used for accurate long

term usage.

The influence of dielectric packaging of the chip

can be seen on the RF output. It is known, that the RF

measurement devices are calibrated to 50 Ohms,

Therefore, the output lines for the RF signal is

designed to have a characteristic impedance of 50

Ohms. The same is done for the RF lines on the board

in order to incur minimal losses to the RF signal due

to reflection. The bond wires account for some losses.

However, encapsulating the bond pad area with the

dielectric medium results in additional losses to the

RF signal which can reduce the RF output power.

However, the RF output power is considerably higher

than the noise floor of the measurement device and

can be accurately measured.

Fig. 6 shows the measurement setup of the sensor

chip with the viscous liquid. A Rhode and Schwarz

X-band spectrum analyser has been used for the

measurement of the RF spectrum.

It can be seen from the inset view of the spectrum

analyser screen (Fig. 7), that the measured RF output

power (-27 dBm) is considerably higher than the

noise floor of the device. The measurements were

conducted in 2 steps. Initially a calibration of the

sensor chip was performed with known alcohols with

known values of permittivity and viscosity. In the

second set of measurements the mixture of glycerol

and water was used to measure the oscillator output

with varying viscosity. The last set of measurements

were performed with the mixture of glycerol and

ethanol to show the significance of knowing the

constituents of the sample for indirect viscosity

measurement.

Figure 6: Measurement setup showing Spectrum analyser,

power supply and chip holder.

Figure 7: Inset of spectrum analyser. The oscillating

frequency with air is 12.28 GHz.

3.1 Calibration

The sensor chip needs to be calibrated with known

alcohols of different permittivities and viscosities.

With no material on top of the sensor, the output

frequency of the sensor is measured to be 12.28 GHz.

This can be seen from the inset view of the spectrum

analyser in Fig. 7. For the next calibration step the

sensor chip was measured with organic alcohols. Four

different alcohols were used for the measurement,

namely, ethanol, methanol, acetone and isopropanol.

Fig. 8 shows the variation of the oscillating frequency

with respect to permittivity of the alcohol.

Radio Frequency CMOS Chem-bio Viscosity Sensors based on Dielectric Spectroscopy

145

Figure 8: Calibration using different types of organic

alcohols.

It is straight forward to understand the variation of the

output oscillation frequency with respect to

permittivity. As the permittivity increases the

capacitance of the sensor capacitor increases as well.

The output oscillation frequency which has an inverse

relation to the capacitance value of the sensor

capacitor decreases with the increase of the

capacitance. The relative permittivity values of the

alcohols at 12 GHz and at room temperature are given

in Table 1.

The corresponding viscosity and permittivity

values of the alcohols are also shown in Table 1.

Table 1.

Alcohol Permittivity Viscosity(mPa.s)

Isopropanol 4 2.1

Ethanol 4.5 0.983

Methanol 8.2 0.507

Acetone 18 0.30

From Table 1 it is seen that the viscosity values

decrease with the increase in permittivity of the acid.

However, on the other hand, it is seen water has a

permittivity of 42 at 12 GHz and a viscosity of 0.81

mPa.s. Therefore, it can be understood that there is no

relation between the absolute values of the relative

permittivity and the viscosity. The sensor being

ideally a dielectric sensor, can detect variation in the

viscosity of a solution based on the permittivity

variation. Therefore, it is needed to know the

permittivity values of the constituents of a liquid

sample in order to detect the viscosity variation. This

will be shown in more detail with two different

sample mixtures with opposite permittivity trends but

same viscosity trends.

3.2 Viscosity Measurement

The operation of the sensor to detect viscosity

variation is done using the mixture of water and

glycerol. Water being a non-viscous liquid reduces

the viscosity of glycerol considerably with its

increasing concentration. On the other hand, water

has a high permittivity as mentioned in previous

section when compared to glycerol. Therefore,

increasing the concentration of water in the mixture

increases the permittivity of the overall solution. This

was also shown in the previous section. The variation

of the oscillation frequency with respect to viscosity

of the glycerol water mixture is shown in Fig.9.

Figure 9: Variation of oscillating frequency of the oscillator

with respect to viscosity variation of glycerol water

solution.

3.3.1 Discussion

It should be noted here, that the sensor architecture

described here is not used to detect an absolute value

of viscosity of a given sample. Detection of absolute

value of viscosity based on an electrical method is not

feasible, unless, there is a direct mathematical

correlation between viscosity and an electrical

quantity. The sensor architecture in this work is used

to detect a relative variation in the viscosity of a fluid

sample based on its constituents. This argument

becomes clear when a similar experiment was

performed with a mixture of glycerol and ethanol.

Prior to the that, it is important to understand the

exclusiveness of viscosity and permittivity. Fig. 10

shows the variation of permittivity with respect to the

percentage of ethanol in ethanol glycerol mixture.

BIODEVICES 2017 - 10th International Conference on Biomedical Electronics and Devices

146

Figure 10: Variation of permittivity and viscosity with

ethanol concentration in ethanol glycerol mixture.

It can be seen that as the concentration of ethanol

increases in the ethanol glycerol mixture, the

permittivity and the viscosity decrease for the overall

mixture. This is in contrast to the glycerol water

mixture where the increase in water content in the

glycerol water mixture increased the permittivity and

reduced the viscosity. Therefore, it can be very well

understood that the permittivity for a given viscous

sample is dependent on the constituents of the sample

The sensor detects only the permittivity and not

the absolute value of viscosity of the solution. This

can be seen with the experiment on the sensor chip

with the mixture of ethanol and glycerol. The results

of the same are shown in Fig. 11.

It is observed, that the oscillating frequency

reduce with increasing viscosity (reduction in ethanol

content in the solution). This is because ethanol has a

lower permittivity as compared to glycerol.

Therefore, increasing the ethanol concentration

reduces the permittivity of the solution. Ethanol also

has a substantially low viscosity as compared to

glycerol. Therefore, increasing the concentration of

ethanol reduces the viscosity of the solution as well.

This was not the case for water glycerol mixture,

where permittivity of the overall solution increased

with increasing water content due to high permittivity

of water and viscosity decreased with increasing

water content.

Now the question arises about the use of the

sensor in lab on chip application like characterization

of sputum as was mentioned in the introductory

section. From the literature it is known that the

viscosity of sputum sample varies with the constituent

it for different patients with lung disease (M. Lopez-

Vidriero et al., 1973). Especially the water content of

the sputum varies for different patients. Knowing the

constituents of the sputum, the variation of water

content will vary the relative viscosity of the sputum

and along with it its permittivity. This sensor

architecture can thus ideally be used for the detection

of the sputum viscosity variation.

Figure 11: Variation of oscillating frequency of oscillator

with respect to viscosity of ethanol glycerol mixture.

4 CONCLUSIONS

In this work we have shown a high frequency sensor

operating in the frequency range of 12 GHz to detect

the viscosity change in a fluid. It was shown that the

choice of the frequency range was based on high

signal to noise ratio due to high contrast in dielectric

permittivity. At lower frequencies, the contrast

between the dielectric permittivities of the liquids are

still high, however, other dispersion effects might

influence the use of the sensor in future lab on chip

applications. On the other hand, very high frequency

operation reduces the contrast between the dielectric

permittivities of the samples, and was shown in

analytically for glycerol water mixture at 12 GHz and

60 GHz. Solutions with different viscosities were

measured and a shift in the resonant frequency of the

sensor oscillator was shown in this work. Two

solution mixture was characterized: glycerol water

mixture and glycerol ethanol mixture. For glycerol

water mixture, the increase in water content increased

the mixture permittivity and reduced the mixture

viscosity. On the other hand, for the ethanol glycerol

mixture, the permittivity and viscosity decrease with

increasing ethanol content. The sensor architecture is

suitable to detect the permittivity change based on

which the solution viscosity of known constituents

can be obtained. It was highlighted that this method

is an indirect way of detection of viscosity. This is due

to the fact that viscosity being a mechanical quantity,

cannot be measured directly with an electrical

quantity. Therefore, knowing the characteristics of

the constituents is extremely important. This kind of

Radio Frequency CMOS Chem-bio Viscosity Sensors based on Dielectric Spectroscopy

147

sensor architecture is suitable for future lab on chip

applications like sputum characterization based on

viscosity variation of the sample.

ACKNOWLEDGEMENTS

The authors would like to thank BMBF, Germany for

funding the project under the grant no: 13U13862.

The authors would also like to thank the technology

department of IHP for fabrication of the chip.

REFERENCES

Agoston, A., Otsch, C., Jakoby, B., 2005. Viscosity sensors

for engine oil condition monitoring-Application and

interpretation of results. Sensors and Actuators A:

Physical. 121 (2), 327-332.

Torres Perez, A., Hadfield, M., 2011. Low-cost oil quality

sensor based on changes in complex permittivity.

Sensors. 11, 10675-10690.

Thalhammer, R., Braun, S., Devicic-Kuhar, B., Groschl,

M., Trampler, F., Benes, E., Nowotny, H., Kostal, M.

1998. Viscosity sensor utilizing a piezoelectric

thickness shear sandwich resonator. IEEE Transactions

on Ultrasonics, Ferroelectrics and Frequency Control.

45 (5), 1331-1340.

Shih, W., Li, X., Gu, H., Shih, W-H., Aksay, I. 2001.

Simultaneous liquid viscosity and density

determination with piezoelectric unimorph cantilevers.

Journal of Applied Physics. 89(2), 1497-1505.

Brouwer, M., Gupta, L., Sadegi, F., Peroulis, D., Adams, D.

2011. High temperature dynamic viscosity sensor for

engine oil applications. Sensors and Actuator A:

Physical. 173, 102-107.

Boss, C., Meurville, E., Sallese, J.M., Ryser, P. 2011. A

viscosity-dependent affinity sensor for continuous

monitoring of glucose in biological fluids. Biosensors

and Bioelectronics, 30 (1), 223-228.

Zhao, Y., Li, S., Davidson, A., Yang, B., Wang, Q., Lin, Q.

2007. A MEMS viscometric sensor for continuous

glucose monitoring. Journal of Micromechanics and

Microengineering. 17, 2528-2537.

Cakmak, O., Ermek, E., Kilinc, N., Baris, I., Kavakli, I.J.,

Yaralioglu, G.G., Urey, H. 2014. Microcantilever based

LOC system for coagulation measurements. In

Proceedings, 18

International Conference on

Miniaturized systems for chemistry and Life Sciences,

Texas, USA.

Guha, S, Wolf, A., Lisker, M., Trusch, A., Meliani, C.,

Wenger, Ch. 2015. 12 GHz CMOS MEMS lab-on-chip

system for detection of concentration of suspended

particles in bio-suspensions. In Proceedings,

Biodevices, Lisbon, Portugal.

Guha, S., Warsinke, A., Tientcheu, Ch.M., Schamlz, K.,

Meliani, C., Wenger, Ch. 2015. Label free sensing of

creatinine using 6 GHz CMOS near-field dielectric

immunosensor. Analyst 140, 3019-3027.

Jamal, F.I., Guha, S., Eissa, M.H., Borngraeber, J., Meliani,

C., Kissinger, D., Wessel, J. 2016. Comparison of

microstrip stub resonators for dielectric sensing in low

power K-Band VCO. In Proceedings IEEE Topical

Conference on Biomedical Wireless Technologies,

Networks and Sensing, Texas, USA.

Wang, H., Sideris, C., Hajimiri, A. 2010. A frequency shift

based CMOS magnetic biosensor with spatially

uniform sensor transducer gain., IEEE; Custom

Integrated Circuit Conference, USA.

Lopez-Vidriero, M., Charman, J., Keal, E., De Silva, D.,

and L Reid, L. 1973. Sputum viscosity: correlation with

chemical and clinical features in chronic bronchitis.

Thorax, 28, 401- 408.

BIODEVICES 2017 - 10th International Conference on Biomedical Electronics and Devices

148