Detection of insulation degradation by-products in transformer oil using ZnO coated IDC sensor

The IDC sensor structure was generated using the geometry feature of COMSOL Multiphysics software. Initially, the substrate is made as a block with specified dimensions of 30 mm in length, 30 mm in breadth, and 1.35 mm in thickness. The IDC sensor electrode structure comprises two interleaved electrode structures. One set of electrode structures is called working electrodes (WE) and another set is called sensing electrodes (SE). Two different electrode structures are made; (1-1-1) type and (1-3-1) type, with two inter-digital electrodes. The total number of fingers in a 1-f-1 sensor configuration is always 13. In the 1-1-1 type IDC sensor, one SE finger is interleaved between two consecutive WE fingers as shown in Figure 1 (a) whereas in the 1-3-1 type sensor, three SE fingers are interleaved between two consecutive WE fingers as shown in Figure 1 (b). As mentioned earlier, each configuration contains 13 fingers and each is 22.225 mm in length, 1.20492 mm in breadth, and 0.025 mm thick. The distance between neighboring fingers (2s) is 0.60246 mm. The relevant materials (Copper, FR4 Epoxy, Transformer Oil, Air, ZnO) were selected from the “add material section” and then allocated to the blocks of the newly formed structure one at a time. The domain of the physics section is determined, after which the terminals of the WE and SE are given the excitation.

Figure 1:

WE is given an excitation of 3 V and the SE is kept at an excitation level of 0 V. The excitation of electrode produces FEF, which passes through the sensing layer (ZnO).

Figure 2 illustrates the generated electric field distribution for all of the IDC structures that were modeled. A shift in electric field can be noticed between the two sensing patterns representing the presence of contaminants in transformer oil, which in our instance are moisture and 2-FAL, for the structural configurations 1-1-1 and 1-3-1 employing ZnO as a sensing material.

Figure 2:

Over the IDC electrodes, a layer of ZnO was deposited. The capacitance of the IDC sensor varies exclusively with an increase or decrease in the thickness and/or relative permittivity of the sensing layer. The planned IDC sensors feature electrode lengths (p), widths (w), and thicknesses (z), with a separation of 2s between each finger that is significantly less than its width. The two-dimensional electrode field distribution can be easily solved using the conformal mapping technique with the help of inverse-cosine transform. The solution of conformal mapping can be used to derive an approximate equation of sensor capacitance. Generalized expression for capacitance between WE and SE electrodes of the 1-f-1 configuration sensor is given by Igreja and Dias [13], Cennamo et al. [16]. (1) C1.f=2εeff ε0pπs∫f−12Q+fsf−12Q+fs+w1(xs)2−1dx {C_{1.f}} = {{2{\varepsilon _{eff}}\,{\varepsilon _0}p} \over {\pi s}}\int_{{{f - 1} \over 2}Q + fs}^{{{f - 1} \over 2}Q + fs + w} {{1 \over {\sqrt {{{\left( {{x \over s}} \right)}^2} - 1}}}dx} where f is the number of SE fingers in between two adjacent WE fingers. The two configurations used in this paper, (1-1-1) and (1-3-1), have the value of f as 1 and 3, respectively. The width of electrode fingers is represented by Q and 2s is the gap between electrode fingers.

As per equivalent circuits shown in Figures 3(a) and (b), the total equivalent capacitance value of two configurations 1-1-1 and 1-3-1 are given as (5) Ceq,1−1−1=12C1,1 {C_{eq,1 - 1 - 1}} = 12{C_{1,1}} (6) Ceq,1−3−1=3[2(C1,1+C1,2+C1,3)] {C_{eq,1 - 3 - 1}} = 3[2({C_{1,1}} + {C_{1,2}} + {C_{1,3}})]

Figure 3:

The values of capacitance C_1,1, C_1,2, and C_1,3 are measured using the equations (2), (3), and (4) respectively. For the two configurations used here, the total equivalent capacitance is computed by the expression (5) and (6).

The considered sensor structure (1-1-1 and 1-3-1) was simulated using COMSOL Multiphysics. The results and its analysis of the simulations are done in Section IV. Fabrication and characterization of IDC sensor with experimental analysis is described in the following section.

The following materials were utilized in the construction of the IDC sensors. The materials are as follows: (1) Single sided copper Printed Circuit Board (PCB) sheet, (2) Double sided copper PCB sheet, (3) Industrial rated 2-FAL, (4) Iso-propyl alcohol, ethanol, De-ionized (DI) water, iron chloride, (5) Connecting wires, (6) Zinc Oxide (ZnO) as sensing film.

As shown in Figures 4(a), both single-sided and double-sided PCBs were cut to the size, with dimensions of 30mm by 30mm.The inter-digital electrode structure was created using AutoCAD software, and the pattern was printed on photopaper, as illustrated in Figures 4(b). Thermal pressing was used to transfer the printed IDC pattern from photopaper to copper-clad PCB. The PCB was then immersed in an etching solution made by combining ferric chloride and DI water. This was done for 5–10 min in order to eliminate the copper print from the PCB, which was then washed with cleaning solution and DI water. The created printed construction was left to dry in the open air. The resulting inter-digital electrode structure is depicted in Figures 4(c).

Figure 4:

We have performed the deposition of ZnO thin film on a copper electrodes printed substrate using the rf-sputtering technique for the purpose of fabricating IDC thin film sensors. These sensors serve as a sensing layer for the detection of 2-FAL and moisture in oil. The ZnO thin film deposition process was carried out under controlled conditions, specifically in the presence of a vacuum with a pressure of 2 × 10⁻⁶ Torr and utilizing 100 W Ar gas plasma for duration of 10 min. The primary aim of the ZnO deposition was to obtain a uniform and well-deposited thin film on the copper electrodes printed substrate. This is crucial for ensuring the efficacy and reliability of the IDC thin film sensors, as an even and consistent ZnO layer is critical for accurate and sensitive detection of 2-FAL and moisture in oil samples. The rf-sputtering technique employed in this research is a widely recognized and effective method for depositing thin films on various substrates. The controlled vacuum conditions and the use of Ar gas plasma at a specific power level allowed for precise control over the deposition process, enabling the production of a desirable thin film on the copper substrate. The successful fabrication of the ZnO thin film through this process paves the way for further investigations into the performance of the IDC thin film sensors for 2-FAL and moisture detection in oil. Fabricated IDC sensors are shown in Figure 5.

Figure 5:

The test samples with varying ppm are made through the incorporation of 2-FAL and humidity in transformer oil. To begin, 0.25 g of 2-FAL is mixed with 250 ml of transformer oil to generate a stock solution of 100 ppm. This stock solution is further diluted to yield 10 ppm test samples. Similarly, 20 ppm, 30 ppm, 40 ppm, and 50 ppm solutions are prepared and tested. A sample is taken as pure oil, which has 0 ppm of 2-FAL and moisture.

The experimental setup comprises of a beaker with a volume of 250 mm filled with several transformer oil test samples in which the sensor is immersed as shown in Figure 6. For Capacitance measurements, the soldered side of the sensor is kept outside. The wires are attached to a handheld Inductance, Capacitance, Resistance (LCR) meter, which measures the sensor's capacitance.

Figure 6:

Response of different type of 1-1-1 electrode structure sensors shown in Figures 7 (a) and (b) illustrate the adjusted curve for predicted and experimental response characteristics for configuration (1-1-1) for SS, SSWS, and DS structures for 2-FAL and moisture. Table 1 shows the sensitivity of all structures with moisture and 2-FAL.

Figure 7:

Using the 1-3-1 design, the region between the electrodes is filled with a sensor film of ZnO, and the change in capacitance is recorded for changes in 2-FAL and moisture concentrations ranging from 0 ppm to 50 ppm. A similar procedure was followed for the various layouts of IDC sensors stated earlier in this section. Figures 8(a) and (b) illustrate the normalized curves for simulated and experimental response characteristics for configuration (1-3-1) with SS, SSWS, and DS, for 2-FAL and moisture, respectively. Table 2 shows the sensitivities at various concentrations of 2-FAL and moisture.

Figure 8:

The experimentally obtained data is used to establish the calibration equation. Thereupon, the knowledge of the response can be used to find the corresponding stimulus. In this work, coefficients of the calibration equation are generated by the curve fitting technique. The capacitance values obtained experimentally corresponding to the known levels of 2-FAL and moisture, have been employed to calibrate the sensor. The calibration equation is a linear fit for all the cases. Tables 3 and 4 display a line fit to the capacitance values for the 1-1-1 and 1-3-1 structures, respectively of the IDC sensor in relation to 2-FAL. The table also displays the most notable deviations of the real response from the linear regression formulas.

The linear equations for moisture sample capacitance values with the maximum fitting errors are shown in Tables 5 and 6. The maximum non-linearity in sensor's response curves is less than 2.5% for all type of sensors. After acquiring the capacitance value of the measurement instrument, the calibrated response curves can be utilized to determine the humidity/2-FAL level in the oil.

The sensor's repeatability index R_t was calculated using equation (7). The IDC sensor capacitance values for ‘s’ (here, s = 10) identical oil samples are CR1, CR2, CRs; and CP is the mean of these ‘s’ number of readings.

For the 1-1-1 configuration the repeatability index of moisture and 2-FAL sensors is 0.00106% and 0.00127%, respectively, whereas the repeatability index for 1-3-1 configuration, of moisture and 2-FAL is 0.00158% and 0.00151%, respectively. These results demonstrate its reproducibility. One of the factors that affects a sensor's functionality over a long length time span is the drift of its electrical characteristics. Due to this flaw, the majority of commercially available thin-film sensors need to be validated every 6–12 months in order to provide accurate and consistent results. Through experimentation, the drift in sensor response over time was also investigated. Figures 9 (a) and (b) shows the variation in capacitance values for transformer oil at 20 ppm moisture for 1-1-1 and 1-3-1 IDC sensor configuration over the period of 30 days.

Figure 9:

Performance of the proposed IDC sensor is compared with other contemporary sensors is shown in Table 7. Not much work is reported on sensor-based estimation of contaminants in transformer oil. The proposed sensor offers good sensitivity in the measurement of moisture and 2-FAL in transformer to assess condition of paper insulation of an in-service oil immersed transformer.

The performance of the developed IDC sensor is compared in Table 7 with other sensors for detection of moisture and 2-FAL. The sensitivity of the sensor is also high, with simple construction. Another advantage of the developed sensor is its reusability and high accuracy which is evident from the performance table.