IEEE International Circuits and Systems Symposium(ICSyS)
Comparison on Ti02 thin film deposition method for fluidic based glucose memristor sensor Nor Shahanim Mohamad Hadis, Asrulnizam Abd Manaf Advanced Integrated System Device (A[SDe), School of Electrical and Electronic Engineering, Universiti Sains Malaysia, Engineering Campus, 14300 Nibong Tebal, Penang, Malaysia norsh7 [email protected] [email protected]
Sukreen Hana Herman NANO-ElecTronic Centre (NET), Faculty of Electrical Engineering, Universiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia [email protected]
nano-well memristor sensor structure. The well is across Ti02 thin film with a thickness of 30 nm for the PVD-RF sputtering method and 200nm for the sol-gel spin coating method. Each well has a diameter ofO.5mm.
Abstract-This paper studies the behavior of fluidic based
memristor sensor in detecting glucose liquid. Two types of memristor
deposition methods which were physical vapor deposition (PVD) method and sol-gel spin coating method. The equipment used in doing the PVD was radio frequency (RF) sputtering
machine, and normal spin coating machine was employed for the sol-gel deposition. The memristor sensors were then tested with three glucose concentrations in a fluidic based platform.
spin coating memristor sensor was higher than the PVD also
25 . mm
memristor sensor. The increment in the concentration of the liquid
An observation showed that the ROFF/RoN ratio of the sol gel
05 . mm
25 . mm
methods. The sol-gel spin coating deposition recorded an exponential relationship between its ROFF/RoN ratio to the
glucose concentration while a linear relationship was observed
in PVD method for the same parameters. As conclusion, the
sol-gel spin coating method has more advantages than the PVD
� 150nm , "-'"-�__ , -. 3
method due to its high sensing capability, shorter deposition time and lower cost.
concentrations; sol-gel spin coating; physical vapor deposition;
sputtering; ROFF/RoN ratio
Fig. 1. Fluidic based memristor sensor with nano-well structure
Ti02 is one of the most suitable materials used to develop memristor for various application fields including photocatalysis , memory , computing  and bio sensors [4, 5 ]. [t is deposited in a form of thin film to suit those applications. The Ti02 thin film can be deposited using various deposition methods such as PVD, chemical vapor deposition, sol-gel deposition and in liquid immersion . Different deposition method will produce different behavior according to the parameter used during the deposition such as temperature, time and power. In this paper, two deposition methods are selected and applied to produce a fluidic based glucose memristor sensor. A comparison between these two methods is observed. The selected methods are PVD method using sputtering machine and sol gel deposition method using spin coating technique. The memristor sensor is fabricated in nano-well structure which is obtained using a lift-off technique. Fig. 1 shows the
Fig. 2 shows the SEM image of the nano-well structure fluidic based memristor sensor for the PVD-RF sputtering method.
Fig. 2. SEM image of fluidic based memristor sensor using the PVD-RF sputtering method
The measurement of the well diameter is done on the SEM image recorded. It is noticed that there are some differences are present compared with the well diameter of
The authors would like to thank Universiti Sains Malaysia Research University (RU) grant no. 203/PELECT16071289.
978-1-4799-1731-0/15/$31.00 ©2015 IEEE 36
the planned memristor sensor as being shown in Fig. 1. The difference may be caused by inappropriate labeling during the measurement. Fig. 3 shows the PDMS mold that is used to cover the memristor sensor to suit the fluidic based application.
passed through the memristor sensor (produce I-V curve). There was only positive voltage sweep analyzed due to unipolar behavior of this memristor device. The sweeping voltage is from 0 V to 5 V in dual sweep mode. Based on the obtained [-V curve, the ROFF/RON ratio is determined and observed. The ratio was selected for this analysis instead of the [-V curve due to its stability. A normal [-V curve in the form of pinched hysteresis consists of two points at every reading voltage, while the ROFFIRON ratio presents only one point at every reading voltage. Thus, it is much easier to interpret based only on the ROFF/RON ratio than the I-V curve which lead to an accurate measurement of the loop area of the pinched hysteresis loop.
Fig. 3. PD MS mold to cover the fluidic based memristor sensor
Fluidic based memristor basedfabrication
Table 1 summarizes the fabrication process of the deposition methods. First step is the same for both methods where lithography process was done in order to pattern the photoresist to create area for the Ti02 deposition. This is a compulsory process to produce patterned Ti02. Step 2 is the sol-gel preparation which is crucial for the sol-gel spin coating only. For the PVD, high purity of 99.99% TiOz target is usually used. Step 3 is the TiOz deposition with the PVD. [t required a longer deposition time than the sol-gel spin coating due to the equipment preparation. The sol-gel spin coating required shorter time for 20 seconds for each sample. Step 4 is the baking process. This process is only valid for sol-gel spin coating as it was already being done under high temperature environment for the sputtering deposition in step 4). Steps 5 and 6 are the same for both methods.
Fig. 4 shows the [-V curve of the PVD-sputtering memristor sensor in detecting three glucose liquid concentrations of 10, 20 and 40 mM.Fig.5 displays the [-V curve of the sol-gel spin coating memristor sensor in detecting the same concentrations of the glucose liquid. The hysteresis shape of these two deposition methods are observed to be different in such a way that the sol-gel spin coating deposition produced a larger loop area than the PVD method. The OFF current of the sol-gel spin coating method is examined to be very small compared with the PVD method. A small amount of current contributes to a large ROFF/RoN ratio. The anal sis is shown in Fi . 6.
- l OmM Glucose
TABLE 1. DOl THIN FILM DEPOSITION COMPARISON PVD Step 1
Lithography process - to pattern photoresist
- 20mM Glucos.e
Sol�el �in coating Ti02 sol-gel preparation
Spin coat Ti02 using the
99.99% high purity Ti02
prepared sol-gel with the
target with the following
0.12 0.10 0.08
Bake at 200 °C for minutes
Remove photoresist using chemical immersion - to get
Aluminium (AI) contact deposition
Process vacuum pressure:
Fig. 4. The I-V curve for the fluidic based memristor sensor deposited using sputtering method in detecting glucose liquids
Argon gas: 15 sscm Vacuum
following parameters; Time: 20 seconds/sample
RF Power: 150 W
Speed: 3000 rpm
Fluidic based memristor sensor characterization
.flJ /'V / � � 1
0. 0 -() .02
The memristor sensor was characterized using Semiconductor Characterization System Keithley 4200SCS. The system swept voltage and measured current that
- l OmM
Glucose - 20mM Glucos.e
- 40 m M Glucose
Fig. 5. The I-V curve for the fluidic based memristor sensor deposited using spin coating method in detecting glucose liquids
TABLE 3. THE SPUTTERING DEPOSITION: THE RANGE VOLTAGE OF 0. 5 V TO 2. 5 V
The graph in Fig. 6 shows the RoFFIRON ratio of the fluidic based memristor sensor in detecting glucose liquid. It is classified into two regions, in which the lower region represents the PVD-sputtering region while the upper region represents the sol-gel spin coating region. It is clearly observed that the ROFF/RoN ratio of the memristor sensor produced using sol-gel spin coating method is more than \00 times larger than the PVD-sputtering method. 1.00E+06
Sol·gel spin coating
Z 0 "
mM Glucose -Sputtering -20
l.00 1. 50
-Spin coat -40 mM Glucose
� 0 0:
Fig. 7. Sol-gel spin coating method: The ROFFIRON ratio for the detection of three different glucose concentrations at the reading voltage of l. 5 V
y = 0.0342x + 1.6664 R' = 0.9917
ROFFIRON ratio 20mM Glucose
The glucose concentrationis also one of the factor contributing to the changes occur in the RoFF/RoN ratio. High concentration of glucose records high RoFF/RoN ratio and otherwise.This is due to the reaction between Ti02 surfaces with hydroxide ions inside the glucose liquid (chemical formula for glucose is C6H 1206). The graph in Fig. 6 records a uniform ratio for the reading voltages from 0.5 V to 2.5 V and their values of ROFF/RoN ratio are tabulated in Table 2 and Table 3. TABLE 2. THE SOL-GEL SPIN COATING DEPOSITION: THE FOR THE RANGE VOLTAGE OF 0.5 V TO 2. 5 V
16.70geO.206x = R'= 0.9427
z 0 1.00E+03 0:
Fig. 6. Logarithmic value of ROFF/RoN ratio of the sputtering and spin coating deposition method in detecting glucose liquid
-Spin coat -20
Reading vo�age (V)
-Spin coat -10
The RoFFiRoN ratio of these two memristor sensors at a reading voltage of 1.5 V is plotted against three glucose concentrations as shown in Fig. 7 and Fig. 8.
RATIO FOR THE
z 0 a:
� u.. 0 a:
Gluco.e concentration. (mM)
Fig. 8. PVD-sputtering method: The ROFF/RoN ratio for the detection of three different glucose concentrations at the reading voltage of l. 5 V
The sol-gel spin coating method presented an exponential relationship between the RoFFiRoN ratio and glucose concentration as stated in equation (\), while the PVD-sputtering method showed a linear relationship
operate. On the contrary, the only disadvantage of the sol gel spin coating is the quality of Ti02 pattern produced as it exists with some shadow, therefore it is lower in quality than the pattern produced by the PVD-sputtering which IS superb with sharp edge pattern.
between the RoFF/RoN ratio and glucose concentration as written in equation (2).
The memristor sensctP-) detected various glucose concentrations where it produced high RoFFIRoN ratio at high glucose concentration and low ROFF/RoN ratio at low glucose concentration. The sol-gel spin coating memristor sensor resulted in a much higher ROFF/RoN ratio change than the PVD-sputtering as the glucose concentration varied. The ROFF/RoN ratio of the sol-gel spin coating was examined to have an exponential relationship with the glucose concentration while the RoFFIRoN ratio of the PVD-sputtering responded linearly to the glucose concentration. The sol-gel spin coating method also offers more advantages than the PVD-sputtering method in terms of its high sensing capability, short deposition time and low cost. Even though thePVD-sputtering method has more disadvantages than the sol-gel coating method, it still holds the most important advantage among the rest which is its very high quality Ti02 pattern.
The sol-gel spin coating memristor recorded an obvious increment in its ROFF/RoN ratio at high glucose concentration while the PVD-sputtering memristor has only managed to record about 0.342 uniform changes in its ROFF/RoN ratio for each 10 mM increment in the glucose . a. i:: rrr· = � = 0 I O �4;. concentration '
The differences arose in the ROFF/RoN ratio was expected due to the presence of different chemical composition inside the Ti02 thin film itself. Different number of Titanium and Oxygen atoms will result in a different reaction with glucose liquid (C6HI206). Table 4 summarizes the advantages and disadvantages of the PVD-sputtering and sol-gel spin coating methods. The sol-gel spin coating has three advantages which are more than the PVD-sputtering.
Author would like to thank Universiti Sains Malaysia and Universiti Teknologi MARA for all the research facilities provided in conducting the research.
TABLE 4. COMPARISON BETWEEN THE PVD-SPUTTERING AND SOL-GEL SPIN COATING METHODS PVD-sputtering
ratio) Deposition time Deposition
Cost Quality of
The sol-gel spin coating method produced a high quality memristor sensor with high ratio of RoFF/RoN. It recorded more than 100 times larger Ropp/RoN ratio than the PVD-sputtering method. In addition, its deposition time is shorter than the PVD-sputtering coating technique. The PVD-sputtering took about 2 hours for one run (fit up to 15 samples) while the sol-gel spin coating only consumed about 1 minute per sample (including time to fix and remove sample from the spin coater machine, only 20 seconds for the spin coating process). Therefore, it only took a total of 15 minutes for all 15 samples followed by approximately another 30 minutes to complete step 4 and 5 for the same number of sample. Apart from these, the main interest of implementing this sol-gel spin coating is its low cost in material and equipment. A spin coater is lower in cost and maintenance than the RF sputtering machine which operates in a vacuum chamber and uses Argon gas to
L. 1. Guo, "Nanoimprint Lithography: Methods and Material Requirements," Wiley InterScience, Advanced Materials, vol. 19, p. 19,2007. Y. Ho, G. M. Huang, and P. Li, "Nonvolatile Memristor Memory: Device Characteristics and Design Implications," presented at the IEE/ACM International Conference on Computer Aided-Design Digest of Technical Papers,2009. J. Joshua, D. B. Strukov, and D. R. Stewart, "Memristive Devices for Computing," Nature Nanotechnology, vol. 8,p. 12,2012. S. Carrara, D. Sacchetto, M. A. Doucey, C. B. Rossi, G. D. Micheli, and Y. Leblebici, "Memristive-biosensors: A new detection method by using nanofabricated memristors," Elsevier: Sensors and Actuators B: Chemical, vol. 171-172, p. 9,2012. F. Puppo, M. A. Doucey, M. D. Ventra, G. D. Micheli, and S. Carrara, "Memristor-Based Devices for Sensing," in 2014 IEEE International
Melbourne VIC,2014,pp. 2257 - 2260. N. S. M. Hadis, A. A. Manaf, and S. H. Herman, "Trends of deposition and patterning techniques of Ti02 for memristor based bio-sensing applications," Microsystem technologies, vol. 19, pp. 1889-1896,2013.