Advancement in Micromachining Techniques of MEMS Piezoresistive Pressure Sensors to Minimize Offset Drift Due to Humidity and Temperature

Abstract

The present chapter focuses on the micromachining technology advancement to enhance the silicon micromachined piezoresistive pressure sensor’s output characteristics. The differential pressure sensors with a glass bond or a passivation layer of Cr/Au metals at the device’s rear side are fabricated using MEMS techniques. The sensors with modifications can minimize and almost eliminate the absorption/adsorption of moisture content on the sensor surface for a long time, indicating improved device performance. The effect of atmospheric temperature and humidity on the four piezo resistors and the sensor’s drift is investigated in this chapter, highlighting the challenges involved in glass micromachining techniques, including wet etching, sandblasting, and electrochemical discharge machining (ECDM). Ultimately, the fabricated sensor’s pressure calibration and offset drift values are studied due to atmospheric effects being reported.

1. Introduction

Pressure sensors based on Micro Electro Mechanical Systems (MEMS) technology are devices that combine micromechanical structures and electronic components integrated into silicon chips. Most pressure transducers have a diaphragm as the sensing element, which deflects under pressure. This mechanical deflection ultimately gives the electrical output according to the transduction mechanism viz, capacitive, piezoresistive, or piezoelectric methods [1]. The piezoresistive property, which was discovered back in 1920, gained significant application in silicon-based transducers in the 1950s [2, 3]. Piezoresistive-based transducers find vast applications over other MEMS devices due to their linear output over a wide range of pressure and batch fabrication processes. Since the invention of MEMS pressure transducers, many studies have focused on addressing issues with drift, accuracy, sensitivity, nonlinearity, and hysteresis. As a result, we have a matured MEMS technology today. The applications of piezoresistive-based pressure sensors are receiving close attention due to their wide variety of applications in biomedical, defense, industrial control, automotive, and aerospace applications.

According to several publications [4], sensors can experience drift due to humidity, temperature (thermal), and residual stresses on the diaphragm caused by fabrication and packaging. Many efforts have been made to improve the pressure sensor’s characteristic properties. Coating an n-type shield layer over p-type piezo resistors to counteract drift caused by electrical fields and temperature [5]. SOI (Silicon On Insulator) wafer-based pressure sensors came into existence to get a very low temperature coefficient of gauge factor [6, 7]. Piezoresistive pressure sensors fabricated from wideband-gap semiconductors such as GaN, diamond, and SiC-based pressure sensors were used for extreme temperature environments [8, 9, 10]. Polycrystalline silicon-based pressure sensors were developed to overcome thermal drift issues [11]. Via etch technology and surface micromachining technologies were explored in depth for improved pressure sensor characteristics [12, 13].

The atmospheric humidity and temperature adversely affect device performance, which leads to the misinterpretation of the real sensor’s output/data. Even though the necessary electronic compensation can effectively (or somewhat) reduce this effect, an advancement in the sensor microfabrication process is investigated here.

Since the invention of transistors, glasses have been used as sealings which protect the device against external moisture and other contamination [14, 15]. Researchers have extensively used glass substrate which is bonded anodically to the MEMS absolute pressure sensors using an evacuated bonding system [16]. Whereas this step is overlooked due to the fabrication complications for low pressure MEMS differential pressure sensors. This chapter focuses on addressing those fabrication difficulties and highlights the importance of glass bonding in sensor’s drift. Besides, an alternate method of Cr/Au passivation layer on the rear side of the sensor is introduced to counteract the drift issues of low pressure (0–140 mbar) differential MEMS pressure sensors.

1.1 Working principle of piezoresistive pressure sensors

Piezoresistive pressure sensors consisting of a silicon diaphragm with four resistors connected in a Wheatstone bridge configuration. When the pressure is applied, the diaphragm undergoes mechanical deformation as shown in Figure 1a. As a result, the piezo resistors (R₁, R₂, R₃ and R₄) on the diaphragm also change their resistance values accordingly. The proportional change in resistance is being converted into electrical output through a Wheatstone bridge configuration, as shown in Figure 1b. Figures 2 and 3 represents the fabricated MEMS pressure sensor die.

Figure 1.

(a) Optical profilometric image showing the mechanical deformation of the diaphragm upon applied pressure and (b) piezo resistors in Wheatstone bridge configuration.

Figure 2.

MEMS differential pressure sensor die (a) front view, (b) rear view and (c) schematic view.

Figure 3.

Schematic view of double BOSS structured differential MEMS pressure sensor.

2. COMSOL analysis and design considerations

Generally, piezoresistive pressure sensors are constituted by a silicon diaphragm on which four resistors are located at high-stress regions. The measuring method followed in such sensors is a Wheatstone bridge. Before the sensor fabrication process, finding the diaphragm dimension and the placement of the four piezo resistors on the diaphragm to get a high-sensitivity region is a prerequisite. COMSOL simulations analyze the piezo-resistors’ position and the diaphragm’s dimensions. The detailed analysis and design considerations are well reported in [1]. The fabrication of a differential pressure sensor in this study, wherein the diaphragm (2 × 2 mm) with a double BOSS structure (0.8 × 0.8 mm) called a sculptured diaphragm was considered to get a linear output at a small pressure range of 0–140 mbar. The diaphragm is anchored all around its edges, and the piezo-resistors are placed at the high-stress region as shown in Figure 3. The rigorous COMSOL analysis (Figure 4b) of the micromachined double BOSS diaphragm estimates the position of the transverse stress distribution along XX’ (marked in Figure 4a). The location of the maximum stress is usually localized at thin areas of the diaphragm.

Figure 4.

COMSOL simulation indicating (a) stress distribution along XX’ for 140 mbar applied pressure and (b) the stress distribution on the top surface of the diaphragm (inset) for varying magnitude of applied pressure P (in mbar) = 0, 28, 56, 84, 112, and 140.

It can be seen from Figures 3 and 4a and b that all four resistors placed at regions I and II which undergo transvers stress upon the application of input pressure. When the pressure is applied onto the diaphragm, the resistors R1 and R3 placed at region-I undergo identical transverse tensile stress and are connected to the opposite arms of the Wheatstone bridge. Conversely, the resistors R2 and R4 placed at region-II undergo transverse compressive stress and are connected to the other opposite arms of the bridge. From the simulation, the average stress values were found to be 0.016 GPa for transverse tensile stress and − 0.016 GPa for transverse compressive stress. The negative sign indicates the compressive stress.

3. Designing of pressure sensors by Clewin software

Piezoresistive MEMS pressure sensor fabrication involves seven photo lithographic patterning steps. As per the simulation, the photomask layout for the sensor fabrication was drawn using Clewin software, as shown in Figure 5, indicated by different colors for the respective layers. The fabrication involves five mask-layer processes for SOI wafer and two mask-layer processes for glass wafer.

Figure 5.

Clewin layout for the fabrication process on (a) 4-inch SOI wafer and (b) 4-inch glass wafer.

4. Fabrication of MEMS pressure sensors

The pressure sensor dice were fabricated on 4-inch SOI substrates at Centre for Nano Science and Engineering (CeNSE), Indian Institute of Science (IISc), Bangalore, India, as shown in Figure 6. Four lithographic steps on frontside of the SOI wafer were performed defining the P+ region, piezo resistors, metal contacts for the open Wheatstone bridge configuration, and the passivation layer. The fifth lithographic step was carried out to release the diaphragm from the wafer backside by dry etching. The detailed process steps are described as follows.

Figure 6.

(A) Steps for the fabrication of differential MEMS pressure sensors and (B) glass wafer fabrication process flow.

4.1 Fabrication process on SOI wafer

The fabrication process was commenced by thermally oxidizing the SOI (100) wafer with an n-type device layer of 15 μm, a buried oxide layer of 1 μm, and p-type handle layer of 400 μm thickness. Photomask-1 was used to define the P+ region by etching the oxide layer using buffered hydrofluoric acid (BHF) solution. Boron doping was carried out to get 5–6 Ω/square sheet resistance at 1100°C for 1 h. The four resistors were patterned using photomask-2, and BHF solution was used to etch the oxide layer. B-doping was performed to obtain sheet resistance of ~300 Ω/square at 1100°C for 10 min. Photomask-3 was used to define the metal contacts followed by the SiN passivation layer deposition by PECVD. The SiN film on the metal pads was etched off to define the area for wire bonding (photomask-4). Finally, front-to-back alignment lithography and DRIE were performed to release the diaphragm. Figure 6A shows the fabricated SOI wafer steps.

4.2 Glass bond and metal passivation on rear side of SOI wafer

From the fabrication flow chart (Figure 6A), two approaches were considered to study the pressure sensor’s drift issues. Approach 1 is the glass bonding, and approach 2 is the Cr/Au deposition at the rear side of the SOI wafer as follows.

4.2.1 Glass etch and anodic bonding (approach 1)

Figure 6B describes the fabrication steps for glass etch and anodic bonding. Before anodic bonding, the glass (SD32 HOYA glass) was etched to a depth of 50 μm (photomask-6) on one side by the wet etch method. 32% hydrofluoric acid and 3.2% hydrochloric acid as an etchant, and 1 μm polycrystalline Si (poly-Si) deposited by LPCVD was used as an etch mask. This etch is necessary for the safe deflection of the BOSS structured diaphragm upon applied pressure range. Through hole is made on glass from the other side by ECDM technique (photomask-7). The different glass micromachining techniques are discussed in detail in Section 5. Processed SOI and glass wafers were bonded successfully using an anodic bonding method at 1100 V DC, 400°C for 10 min and 10⁻⁴ mbar vacuum. The fabricated wafer was diced (Figure 7) using the dicing tool (DISCO).

Figure 7.

Photograph of the fabricated 4-inch SOI wafer after dicing.

4.2.2 Metal passivation layer (approach 2)

Cr/Au thin film of 40/200 nm was deposited on fabricated SOI wafer rear side using thermal evaporation at a base pressure of 10⁻⁵ mbar as passivation layer.

5. Glass machining techniques

Glass and quartz are extensively used in MEMS, biochip fabrication, and micro-fluidic devices due to their excellent material properties such as mechanical robustness, transparency, hardness, biocompatibility, chemical and heat resistant nature. The hermetic sealing of glass is required for the absolute pressure sensors, which is usually bonded anodically in a high vacuum. Though the glass bonding in the case of differential pressure sensors is not necessary, it is an added advantage if included. Glass bond provides mechanical strength, enables high electrical insulation for the sensor from the Kovar header (on which the sensor is mounted using compatible epoxy material), protects the die from the external environment, and enhances the lifetime. Importantly, the temperature effects are minimized as the thermal expansion coefficient of glass matches with Kovar material. Unfortunately, every additional fabrication step is accompanied by some process complexities. Glass machining has many challenges that must be carefully addressed. Various conventional glass etch techniques include mechanical methods such as ultrasonic drilling, powder/sand blasting, and electrochemical micromachining etc. As mechanical machining has limitations below 100 μm, microfabrication techniques such as dry plasma etch, and wet chemical etch are preferred. A careful selection of glass micromachining techniques according to the applications is a crucial step.

This work considers three glass patterning techniques for differential MEMS pressure sensor fabrication. Before the anodic bonding, the glass machining of ~50 μm depth from one side and a through hole from the other side must be carried out using photomask 6 and 7 (Figure 6b). The glass was bonded anodically to silicon with an excellent hermeticity. However, the surface roughness and cleanliness of the wafer surface to be bonded are highly important factors for the fabrication yield. The prerequisite for the best glass etching is a compatible mask and the lowest possible effects on the etched surface. Higher surface roughness (> 20 nm) results in anodic bonding failures. The challenges faced in each of the machining techniques are discussed as follows.

5.1 Glass pattern by wet etching and sand blasting

The growing electronic industry demands cost-effective and efficient fabrication technology. The advantages and disadvantages of glass machining by wet etch, sandblasting and ECDM are compared in Table 1. Irrespective of isotropic etch nature, the wet etch received much attention compared to others due to its process simplicity, high mask selectivity, low surface roughness, and cost-effectiveness. As far as the through glass etch is concerned, the wet method is accompanied by process complexities such as isotropic etch profile and failure in etch mask for longer etch duration. Table 2 summarizes the etch trials for glass using 32% HF and 3.2% HCl as an etchant and different etch masks. HF was used to etch glass, while the small quantity of HCl increases the etch byproduct dissolution and enhances surface quality [17].

Parameter	Wet etching	Sandblasting	ECDM
Etch depth	350 μm	500 μm	500 μm
Lateral etch	Yes (~1:1)	No	No
Aspect ratio	Low	High	High
Process time	Less (2 h)	Less (30 min)	Moderate (6 h)
Mask	Poly silicon	Stainless steel shadow mask	No
Etched surface for post bonding process	Very good	Poor due to particles	Very good

Table 1.

Advantage and disadvantage of the micromachining techniques in view of 500 μm through glass machining process and post anodic bonding of SOI wafer for differential pressure sensors.

Mask numbers	Masking layer	Etching depth (μm)	Etch selectivity
M1	Az4562	20	Poor
M2	a-Si/Cr/Au	100	Good
M3	0.5 μm Molybdenum	150	Good
M4	Polycrystalline silicon	350	Excellent

Table 2.

Summary of glass machining using different etch mask in 32% hydrofluoric acid etchant.

The investigation results show that M1 (Az4562 photoresist) can etch glass up to 20 μm. However, this etch mask is more prone to failure due to its poor adhesion and damaged by high HF concentration. Further, we have tried three different etch masks such as M2 (stack of a-Si/Cr/Au, 300/20/100 nm thickness), M3 (0.5 μm thick molybdenum) and M4 (1 μm thick poly-Si) as etch masks. Glass etched depths of 100, 150 and 350 μm were achieved successfully for M2, M3 and M4 respectively as tabulated in Table 2. As per this table, beyond the mentioned etch depths, the glass machining process was unsatisfactory due to etch mask failure. It was observed that the continued etching beyond a point resulted in pinhole generation on the etch mask due to high HF concentration. Though HCl was used, the etch byproducts redeposited on the etched surface and inhibited the glass etching, resulting in high surface roughness due to longer etch time. Figure 8 shows the high surface roughness due to pinhole generation on poly-Si after 2 h of etching in HF. Table 2 shows masks M2, M3, and M4 which can be used to etch glass for a depth of 50 μm (Photomask 6). It must be noted that while performing glass etch on one side, the other side has to be protected using the etch mask. In this aspect, two-sided deposition of poly-Si in LPCVD is advantageous over other etch masks such as M2 and M3. This eliminates the additional etch mask deposition step on the rear side of the glass wafer.

Figure 8.

Optical image of glass through etch by wet etch using HF solution.

For glass etching up to 350 μm depth, wet etch is preferred, however for the glass through etch sand blasting method is considered. Sand blasting is a physical process technique where a high-speed particle jet colloids with the glass surface resulting in mechanical erosion through a shadow mask of 0.8 mm thick stainless steel sheet. This shadow mask was aligned to the glass wafer so that circular pattern aligns at the centre of the square pattern. During the sand blasting process, abrasive sand particles of 50 μm size from the nozzle with a high-velocity stream impinges onto the glass substrate, causing glass etching. The process is simple, easy and drills 500 μm thick glass within 30 min. The biggest challenge involved in sand blasting is the glass wafer cleaning for the anodic bonding process. Also, as soon as the 500 μm etch is completed, the sand particles exit through the hole and etch the glass on the other side of the wafer up to 500 μm circumference (Figure 9). Thus, creating high surface roughness and particle contamination across the wafer.

Figure 9.

Optical image of glass through etch by sand blasting.

Though the sandblasting process is successful in through glass etch, we have explored the ECDM technique in view of surface quality for anodic bonding.

5.2 Glass pattern by electro chemical discharge machining (ECDM)

Electrochemical discharge machining (ECDM) also called electrochemical arc machining (ECAM) is a combination of electric discharge machining (EDM) and electrochemical machining (ECM) techniques. Both EDM and ECM methods are limited only to conducting materials. This drawback was overcome by hybrid.

ECDM process, which can remove materials from both conductive (process is called as ECAM) and non-conductive (process is called as ECDM) materials [18]. This technique is undergoing lot of research and has much scope in commercializing it. The experimental setup is as shown in Figures 10 and 11. KOH solution was used as an electrolyte. The glass machining by ECDM undergoes thermal and chemical etching. The following reactions occur at the anode Eq. (1) and cathode Eq. (2) upon the applied voltage.

Figure 10.

Electro chemical discharge machining (ECDM) tool- schematic.

Figure 11.

Electro chemical discharge machining (ECDM) tools at Meukron technologies private limited, Bangalore, India.

At the anode,

$${\text{4OH}}^{-}\to {\text{2H}}_2\text{O}+{\text{O}}_2+{\text{4e}}^{-}$$E1

At the cathode,

$${\text{2H}}_2\text{O}+{\text{2e}}^{-}\to {\text{2OH}}^{-}+{\text{H}}_2$$E2

At a critical voltage, the spark is produced by the electrical discharge in the generated hydrogen gas layer. Due to the thermal energy of the spark and chemical reaction the material in contact will be eroded.

Generally, the glass etch profile by ECDM machining was improved by varying the experimental parameters such as type of electrolyte and its concentration, voltage, pulse on/off-time ratio, and feed rate [19]. In the current study, ECDM process was optimized by varying the experimental parameters to get a crack free etch profile with a slow etch rate of 1.4 μm/min. Importantly, the surface roughness after ECDM etching of the glass wafer which was exposed to the electrolytic solution was measured using atomic force microscopy. It was found that the surface roughness was <10 nm, favorable to anodic bonding process. The micromachined glass through etch is shown in Figure 12.

Figure 12.

Images of glass through etch by electro chemical discharge machining (a) Optical and (b) SEM images.

Finally, in comparison to wet etch and sand blasting process, the ECDM resulted in best of etch profile and smooth surface. Hence, the ECDM process was used to through etching of the glass wafer using photomask 7 for MEMS differential pressure sensors fabrication. The resulted glass substrate was bonded anodically to the SOI wafer (Section 4.1) Figure 13 represents a cross sectional view of sensor die after bonding.

Figure 13.

Cross section SEM image of the SOI-glass bonding interface.

The fabricated wafer was diced, the individual sensor die (Figure 7) was separated, mounted on the header (also called glass to metal seal for hermeticity and high electrical insulation), proceeded further for wire bonding and testing. The detailed post fabrication steps are briefed in sections 6 and 7.

6. Packaging steps

Figure 14 shows the sequence of packaging steps. Fabricated headers (Figure 15a and b) are visually inspected using the stereo microscope for defect if any. Further the electrical insulation resistance test was carried out by applying 50 Volts between the body and the individual pins, the insulation should be greater than 100 MΩ. To test the hermeticity, helium leak test was conducted to verify the leak rate between the header and pins, which must be minimum of 10⁻⁸ scc/s. The cleanliness of the headers was confirmed by visual inspection.

Figure 14.

Packaging process step flow chart.

Figure 15.

(a) Header top view and (b) header side view.

Identified MEMS pressure sensor die (Figure 16b) was cleaned using plasma cleaner for 10 min. Using DC probe station (Figure 16a), individual piezo resistances R1, R2, R3 and R4 were measured. Figure 17a and b shows the sensor die mounted on header with suitable epoxy and cured at room temperature for 24 h.

Figure 16.

(a) DC probe station setup and (b) probing on MEMS pressure sensor die.

Figure 17.

(a) Die mounted on the header and (b) closure view of MEMS pressure sensor die attached on the header.

Figure 18 shows the wire bonding (33 μm Al) in open Wheatstone bridge configuration (Figure 1b). Pull strength of the bonded wires were measured which was 9 ± 1 grams (Figure 18b). Measurement of individual resistances was carried out using a 6 1/2 digit voltmeter. Imbalance (offset voltage) of the closed Wheatstone bridge was measured using 5 V DC power supply. To protect the device from environmental dust, dirt, and humidity, parylene C was coated using the Parylene Coater, further individual resistance was measured using 6 1/2 digit voltmeter. The offset of the Wheatstone bridge was measured with 5 Volts DC excitation under ambient conditions. For stress relieving process, thermal cycling of the wired die was carried out using the hot and cold chamber (H &CC). To measure the output of the sensor with pressure (0–140 mbar), a pneumatic calibrator was used (Fluke make) with the 5Volts excitation (Figure 19).

Figure 18.

(a) Wire bonding tool setup. (b) Pull strength testing of bonded wires.

Figure 19.

Pneumatic pressure calibrator setup.

Initial pressure calibration in five ascending and five descending steps of equal pressure intervals was carried out to find out the zero-offset value, full scale output, sensitivity, nonlinearity, and hysteresis (Table 3). Further, varying humidity tests at constant temperature (22°C) and direct environmental exposure were carried out to ascertain drift or shift of the sensor output due to humidity effect (Section 7) using Data Acquisition System (DAS).

Sensor dice	With glass bond	With Cr/Au
Zero offset voltage (mV)	35.99	36.01
Sensitivity (mV/mbar)	0.73	0.78
Span (mV)	102.47	108.91
MAX NL + HYS (% F.S.O)	0.42	0.52

Table 3.

Consolidated data on pressure calibration for dice with glass bond and Cr/Au layer.

7. Pressure calibration and effect of humidity on piezo resistors

Humidity exposure experiments were conducted using bare die and two more sensor dice one with glass bonding and another with Cr/Au passivation on the rear side. Pressure calibration (Figure 19) was conducted and measured zero offset values, and output voltages for five ascending and descending pressure of 28, 56, 84, 112 and 140 mbar with an excitation voltage of 5 V DC at 25°C (Table 3). After recording the pressure response data, the sensors were subjected to study the humidity and temperature effect on piezo resistance values in a controlled environment as well as direct exposure to the atmosphere. The humidity effect in a controlled environment (Figure 20) was performed in a closed chamber with pre-defined humidity values of 30, 60 and 90% Rh with 2 h of exposure time. From the experimental results, the percentage change in piezo resistance values for the bare die, glass bonded die and die having Cr/Au passivation was found to be 0.106, 0.035 and 0.034% respectively (Figure 21).

Figure 20.

Humidity experiment setup: bare die, dice with glass bonding and Cr/Au passivation placed inside the chamber.

Figure 21.

Effect of humidity on piezo resistance for the sensors placed inside a humidity chamber (a) bare die (0.106% resistance change) (b) with glass bond (0.035% resistance change) and (c) with Cr/Au passivation layer (0.034% resistance change).

Further, a similar experiment was conducted by exposing the sensor dice directly to the atmosphere by placing the sensors outside the window for 2 days (Figure 22). The resistance values change due to the environmental conditions were recorded using DAS. Figure 23 shows the percentage change in piezo resistance value for the sensors without glass bond, with glass bond & Cr/Au layer which were found to be 0.092, 0.026 and 0.022% respectively.

Figure 22.

Humidity experiment set up: dice with & without glass bonding and Cr/Au passivation exposed to direct environment by placing them near the window.

Figure 23.

Effect of humidity on piezo resistance for the sensors exposed directly to the atmosphere (a) without glass bond (0.092%resistance change) and (b) with glass bond (0.026% resistance change) and (c) with Cr/Au passivation layer (0.022% resistance change).

As compared to the bare die, piezo resistance change due to humidity is nearly four times better for other two dice. Cr/Au layered die found to be slightly better than anodic bonded die. The glass bonding to the sensor die provides thermal stability and protection from the environmental dust particles. The Cr/Au passivation layer protects the die from the absorption/adsorption of moisture on the sensors surface. This was confirmed by measuring the sensor’s zero offset values and output response to the applied pressure after subjecting it to extreme humidity tests (Table 3).

Table 3 shows zero offset values, sensitivity, span, and maximum non linearity and hysteresis. The zero offset values for Cr/Au and glass bonded matches closely (0.05%) and nonlinearity + hysteresis 0.1%. Whereas sensitivity 0.05 mVolt/mbar. Bare die found to be having humidity effect and hence zero offset voltage variation.

8. Conclusions

The present chapter reports the improved MEMS differential pressure sensor output characteristics by incorporating new modifications in the fabrication stage. The environmental humidity and temperature variation directly affect the offset of the sensor. The addition of glass bonding at the rear side of the pressure sensor die enhances thermal stability and reduces the sensor’s offset drift. The Cr/Au passivation layer minimizes and almost eliminates the absorption/adsorption of moisture content on the sensor surface, which results in reduced offset drift. The critical challenges involved in glass machining by wet etching, sandblasting, and electrochemical discharge machining (ECDM) were discussed in view of glass bonding process. It was found that the fabrication of differential MEMS pressure sensor complexities were minimized in case of ECDM technique in view of glass machining and anodic bonding processes.

Acknowledgments

We acknowledge funding support from MHRD (Ministry of Human Resource Development), MeitY and DST (Department of Science and Technology) Nano Mission, India through NNetRA. The authors would also like to acknowledge the National Nano Fabrication Centre (NNFC), the Micro and Nano Characterization Facility (MNCF) and Advanced systems and packaging lab of the Centre for Nano Science and Engineering, Indian Institute of Science for providing access to the fabrication, characterization and testing facilities. The authors also acknowledge the Meukron Technologies Private Limited, Bangalore for ECDM of glass.

Conflict of interest

The authors declare no conflict of interest.