*3.3. Fabrication of Second Matching Layer*

using drainage and pulse insertion

the calculation.

*3.3. Fabrication of Second Matching Layer* The second matching layer needs a low theoretical acoustic impedance and is cost‐effectively machined to the designed thickness. Aerogel has low acoustic imped‐ ance, but its machinability and bonding properties are poor [21]. In this study, the second matching layer is made from polypropylene foam. It has a low density and a high bubble ratio leading to a low acoustic impedance, and can be considered as a composite of the The second matching layer needs a low theoretical acoustic impedance and is costeffectively machined to the designed thickness. Aerogel has low acoustic impedance, but its machinability and bonding properties are poor [21]. In this study, the second matching layer is made from polypropylene foam. It has a low density and a high bubble ratio leading to a low acoustic impedance, and can be considered as a composite of the polypropylene and air microsphere. The foaming rate controls the matching layer acoustic properties.

polypropylene and air microsphere. The foaming rate controls the matching layer acous‐

#### tic properties. *3.4. Transducer Assembly*

*3.4. Transducer Assembly* Figure 2 shows the transducer design in this study. The front and rear surfaces of 1‐3 piezoelectric composite are coated with metal films as the electrodes. On the front surface of 1‐3 piezoelectric composite, the first and second matching layers are bonded to form the core of the transducer. A standard BNC connector is screwed to the housing through Figure 2 shows the transducer design in this study. The front and rear surfaces of 1-3 piezoelectric composite are coated with metal films as the electrodes. On the front surface of 1-3 piezoelectric composite, the first and second matching layers are bonded to form the core of the transducer. A standard BNC connector is screwed to the housing through the front cover. BNC connector links the wire of the 1-3 piezoelectric composite to receive and send the electrical signal from the outside device. *Micromachines* **2022**, *13*, x FOR PEER REVIEW 6 of 18

the front cover. BNC connector links the wire of the 1‐3 piezoelectric composite to receive

**Figure 2.** Structure of air‐coupled transducer. **Figure 2.** Structure of air-coupled transducer.

copper wires, as shown in Figure 3a.

### **4. Experiment Setup 4. Experiment Setup**

The PZT‐4 ceramics (Shandong Weifang Jude Electronics Co. Ltd., Shanghai, China) was used as the piezoelectric composite due to the high electromechanical coupling fac‐ tor, which can generate more ultrasonic waves with a provided energy. A 20×14×8 mm cuboid PZT‐4 was cut by the dicing machine (SYJ‐400, Shenyang Kejing, Shenyang, The PZT-4 ceramics (Shandong Weifang Jude Electronics Co. Ltd., Shanghai, China) was used as the piezoelectric composite due to the high electromechanical coupling factor, which can generate more ultrasonic waves with a provided energy. A 20 × 14 × 8 mm cuboid PZT-4 was cut by the dicing machine (SYJ-400, Shenyang Kejing, Shenyang, China)

China) to create the 0.4 mm width and 7.8 mm depth kerf grid on the surface. The parallel distance of kerfs was 2.4 mm which can create 2 mm × 2 mm PZT‐4 columns on the sur‐

Shanghai, China) was poured on the PZT‐4 to fill the kerfs and connect the columns. The composite was put in a vacuum chamber to remove air from the composite. After 24 h curing, the composite was ground by a surface grinder (M7230H, Hangzhou grinding machine Co. Ltd., Hangzhou, China) to remove the uncut layer on the bottom and epoxy on the front surface. The ultimate thickness of the composite controlled by grinding was 7.6 mm to match the 200 kHz resonance frequency. The density and acoustic propagation velocity of the 1‐3 type piezoelectric composite were measured by the drainage and pulse insertion method, which were 5501.81 kg/m3 and 3521 m/s, respectively. Based on Equa‐ tion (1), the 1‐3 piezoelectric composite acoustic impedance was 19.37 MRayl. Table 1 summarizes the 1‐3 piezoelectric composite properties. The front and bottom surfaces 1‐3 piezoelectric composite were coated with 50 nm silver via chemical vapor deposition to generate the positive and negative electrodes. The electrode surfaces were welded to two

to create the 0.4 mm width and 7.8 mm depth kerf grid on the surface. The parallel distance of kerfs was 2.4 mm which can create 2 mm × 2 mm PZT-4 columns on the surface after dicing. The epoxy (E51, Shanghai Aotun Chemical Technology Co., LTD, Shanghai, China) was poured on the PZT-4 to fill the kerfs and connect the columns. The composite was put in a vacuum chamber to remove air from the composite. After 24 h curing, the composite was ground by a surface grinder (M7230H, Hangzhou grinding machine Co. Ltd., Hangzhou, China) to remove the uncut layer on the bottom and epoxy on the front surface. The ultimate thickness of the composite controlled by grinding was 7.6 mm to match the 200 kHz resonance frequency. The density and acoustic propagation velocity of the 1-3 type piezoelectric composite were measured by the drainage and pulse insertion method, which were 5501.81 kg/m<sup>3</sup> and 3521 m/s, respectively. Based on Equation (1), the 1-3 piezoelectric composite acoustic impedance was 19.37 MRayl. Table 1 summarizes the 1-3 piezoelectric composite properties. The front and bottom surfaces 1-3 piezoelectric composite were coated with 50 nm silver via chemical vapor deposition to generate the positive and negative electrodes. The electrode surfaces were welded to two copper wires, as shown in Figure 3a. *Micromachines* **2022**, *13*, x FOR PEER REVIEW 7 of 18

**Figure 3.** Air‐coupled ultrasonic transducer. (**a**) 1‐3 piezoelectric composite with double matching layer and (**b**) self‐developed transducer. **Figure 3.** Air-coupled ultrasonic transducer. (**a**) 1-3 piezoelectric composite with double matching layer and (**b**) self-developed transducer.

From the matching theory mentioned in Section 2, for the double-layer matching system, the acoustic impedance of the first and second matching layers should be 1.32 and 0.0062 MRayl, respectively. In order to achieve the acoustic impedance needed for the first layer, a series of matching layers were fabricated. The first layer consisted of the hollow glass microspheres and E51 epoxy resin. In order to identify the effects of fabrication parameters on the matching layer's acoustic properties, especially in acoustic impedances, the matching layers with different glass microsphere sizes and weight ratios were made. Table 2 shows the parameters of the hollow glass microspheres used in this study, which have an average diameter of 100, 85, and 70 µm. The glass microsphere weight ratios of 10%, 15%, 20%, and 30% were experimental studied to help the modeling process. In order to facilitate the mixture and curing process, the diluent (butyl glycidyl ether) was added, if necessary, when the bubble could not be expelled completely. After the fabrication process, the matching layer's density and acoustic velocity were measured to calculate the acoustic impedances. The experiment results were used to validate the proposed matching layer modeling. The exact weight ratios of the first and second matching layers based on the calculated theoretical acoustic impedances were provided. After the determination of the recipe of the matching layer, the first and second matching layers were fabricated, and the acoustic parameters were tested.


BR40 85 400 240 1–2 2280

### **Table 1.** List of experiment parameters.

BR20 100 200 120 0.5–1

BR60 70 600 390 1.5–3.5

The influence of the bonding layer on acoustic propagation and transducer performance is also discussed in this study. In order to connect the 1-3 piezoelectric composite, first matching layer, and second matching layer, the bonding layer should be added to the contact surfaces. Table 1 also lists three types of the bonding layer to be used in the experiment to identify the potential effect.

The first matching layer, bonding layer, and second matching layer, after determination, were attached to the surface of 1-3 piezoelectric composite to form the core of the transducer, as seen in Figure 3a. An aluminum housing enclosed the core, and a BNC connector was on the front surface of the transducer to link the 1-3 piezoelectric composite and outside device. Figure 3b. shows the transducer by using the 1-3 piezoelectric composite, first matching layer, bonding layer, and second matching layer fabrication in this study.

In order to validate the proposed matching layer modeling and study the effect of bonding layer properties on the transducer performance, a testing platform was built, as presented in Figure 4. The self-developed 200 kHz transducers and a commercially available one (0.2 K 14 × 20 N-TX, Japan Probe Co. Ltd., Japan) were put on the two ends of a linear stage to emit and receive the acoustic signal. Air and testing plate were in the middle of the transducers to act as the medium and inspected material. Both transducers are connected with ultrasonic instruments and computers to control the testing process. The exciting signals output from the self-developed ultrasonic instruments were 200 kHz Hanning windowed three-cycle sine bursts with Vpp of 138 V. The received signals were filtered and enveloped to obtain the amplitude. The comparisons of amplitude were conducted to evaluate the performance of the transducer. In this study, the plate was not used first to validate the influence of different matching layers and bonding layers on the transducer sensitive for the modeling and theory validation. Two types of plates (CFRP and PVC foam) with and without artificial defects were put in the middle of the transducers to emphasize the practical application of the transducer. The sizes for the CFRP and PVC foam were 540 mm × 500 mm × 5 mm and 200 mm × 200 mm × 30 mm, respectively.

*Micromachines* **2022**, *13*, x FOR PEER REVIEW 9 of 18

The artificial defects were created by milling at the bottom of the tested materials with the diameter and depth of Φ 5 × 3 and Φ 10 × 8 mm for the CFRP and PVC foam, respectively. tively. The artificial defects were created by milling at the bottom of the tested materials with the diameter and depth of Φ 5 × 3 and Φ10 × 8 mm for the CFRP and PVC foam, respectively.

The first matching layer, bonding layer, and second matching layer, after determi‐ nation, were attached to the surface of 1‐3 piezoelectric composite to form the core of the transducer, as seen in Figure 3a. An aluminum housing enclosed the core, and a BNC connector was on the front surface of the transducer to link the 1‐3 piezoelectric compo‐ site and outside device. Figure 3b. shows the transducer by using the 1‐3 piezoelectric composite, first matching layer, bonding layer, and second matching layer fabrication in

In order to validate the proposed matching layer modeling and study the effect of bonding layer properties on the transducer performance, a testing platform was built, as presented in Figure 4. The self‐developed 200 kHz transducers and a commercially available one (0.2 K 14 × 20 N‐TX, Japan Probe Co. Ltd., Japan) were put on the two ends of a linear stage to emit and receive the acoustic signal. Air and testing plate were in the middle of the transducers to act as the medium and inspected material. Both transducers are connected with ultrasonic instruments and computers to control the testing process. The exciting signals output from the self‐developed ultrasonic instruments were 200 kHz Hanning windowed three‐cycle sine bursts with Vpp of 138 V. The received signals were filtered and enveloped to obtain the amplitude. The comparisons of amplitude were conducted to evaluate the performance of the transducer. In this study, the plate was not used first to validate the influence of different matching layers and bonding layers on the transducer sensitive for the modeling and theory validation. Two types of plates (CFRP and PVC foam) with and without artificial defects were put in the middle of the trans‐ ducers to emphasize the practical application of the transducer. The sizes for the CFRP and PVC foam were 540 mm × 500 mm × 5 mm and 200 mm × 200 mm × 30 mm, respec‐

**Figure 4.** Testing platform. **Figure 4.** Testing platform.

this study.

#### **5. Results and Discussions 5. Results and Discussions**

*5.1. Effect of Material Properties on Acoustic Impedance 5.1. Effect of Material Properties on Acoustic Impedance*

Table 3 lists the measured acoustic properties of the first matching layers with dif‐ ferent glass microsphere diameters and weight ratios. The data of the BR20 glass micro‐ sphere in the 30% weight ratio are not reported due to the high viscosity, which pre‐ vented the air bubble extraction even with the diluent and failed the homogeneous acoustic properties of the matching layer. Based on the results in Table 3, the higher the hollow glass microsphere weight ratio is, the lower the density and longitudinal wave Table 3 lists the measured acoustic properties of the first matching layers with different glass microsphere diameters and weight ratios. The data of the BR20 glass microsphere in the 30% weight ratio are not reported due to the high viscosity, which prevented the air bubble extraction even with the diluent and failed the homogeneous acoustic properties of the matching layer. Based on the results in Table 3, the higher the hollow glass microsphere weight ratio is, the lower the density and longitudinal wave velocity are. From the microstructure of the first matching layer, the higher hollow glass microsphere weight ratio means that more space is filled by the air inside of the microsphere and longer time the acoustic wave takes to propagate in the glass microsphere, which decreases the overall density and acoustic velocity. With the decrease in the hollow glass microsphere diameter under the same weight ratio, both the density and velocity increase due to the increase in the glass weight percentage.



\* The high error is due to the bubble and inhomogeneous microstructure in the matching layer.

The first matching layer was cut into the design shape to be pasted on the 1-3 piezoelectric composite. The surface after cutting was observed by the microscopy to check the microstructure. The white microsphere and epoxy matrix can be identified in all figures.

Some white dots, marked by the arrows, are air bubbles. With the increase in the microsphere weight ratio (*R*c), the number of bubbles increased, based on the observation in Figure 5a. Figure 5b,c show a similar trend. The higher *R*<sup>c</sup> increases the viscosity of the mixture, which makes the bubbles hard to escape. The material flowing in the high viscosity environment is also constrained, which makes the matching layer easy to inhomogeneous. The inhomogeneous microstructure of the matching layer will affect the longitudinal wave velocity, which can be another possible reason for the experiments BR20-20-15. The bubble and inhomogeneous material also enlarge the error between the theoretical model and experiments results, indicated in Table 3, which makes the acoustic properties of the matching layer difficult to predict. The diluent can improve the bubble release, as seen in the 15% and 20% *R*<sup>c</sup> in Figure 5a and Table 3. *Micromachines* **2022**, *13*, x FOR PEER REVIEW 11 of 18

**Figure 5.** Microstructure of the epoxy/hollow glass microsphere composites. (**a**) BR20, (**b**) BR40, and (**c**) BR60 microsphere. (The value on the figure is the weight ratio *R*c, the red arrows are the air bubble). **Figure 5.** Microstructure of the epoxy/hollow glass microsphere composites. (**a**) BR20, (**b**) BR40, and (**c**) BR60 microsphere. (The value on the figure is the weight ratio *R*c, the red arrows are the air bubble).

#### *5.2. Validation of the Matching Layer Modeling 5.2. Validation of the Matching Layer Modeling*

Table 3 also provides the modeling results of density, longitudinal wave velocity, and acoustic impedance. The proposed density model can predict all experiment density results with an error of less than 3%, which proves the correctness and accuracy of the modeling. For the longitudinal wave velocity, all experiments are modeled in the error of less than 6.5%, except BR20‐20‐15, which has an error of up to 10.9%. The acoustic im‐ pedance prediction results share a similar trend with the longitudinal wave velocity, which has high accuracy with an error of less than 6.3% if the BR20‐20‐15 is not consid‐ ered. As mentioned above, in BR20, the viscosity of the composite increases tremen‐ dously when the weight ratio is larger than 15%, and the diluent has to be used. When 15% diluent is added in BR‐20‐15‐15, the matching layer can be formed. However, when the weight ratio of hollow glass microspheres increases to 30%, the microsphere cannot Table 3 also provides the modeling results of density, longitudinal wave velocity, and acoustic impedance. The proposed density model can predict all experiment density results with an error of less than 3%, which proves the correctness and accuracy of the modeling. For the longitudinal wave velocity, all experiments are modeled in the error of less than 6.5%, except BR20-20-15, which has an error of up to 10.9%. The acoustic impedance prediction results share a similar trend with the longitudinal wave velocity, which has high accuracy with an error of less than 6.3% if the BR20-20-15 is not considered. As mentioned above, in BR20, the viscosity of the composite increases tremendously when the weight ratio is larger than 15%, and the diluent has to be used. When 15% diluent is added in BR-20-15-15, the matching layer can be formed. However, when the weight ratio of hollow glass microspheres increases to 30%, the microsphere cannot be stirred to mix with the

be stirred to mix with the resin even with 15% diluent. More diluent, up to 50%, had been tried, which still can obtain many visible bubbles when the composite is cured. The rea‐

After verification of the matching layer modeling in this study, the best material components needed to form the theoretical acoustic impedance are predicted. Based on the proposed modeling, the theoretical first matching layer can be made by BR20 with a 19% weight ratio and 15% diluent. The fabricated sample from the modeling has a den‐ sity of 625.35 kg/m3 and 2186 m/s acoustic velocity. The calculated acoustic impedance is 1.36 MRayl, which is a 1.4% error from the theoretical value. The second matching layer, resin even with 15% diluent. More diluent, up to 50%, had been tried, which still can obtain many visible bubbles when the composite is cured. The reason for the relatively poor accuracy in BR20-20-15 maybe result from the inhomogeneous of the material and microbubble inside, which has been confirmed by Figure 5.

After verification of the matching layer modeling in this study, the best material components needed to form the theoretical acoustic impedance are predicted. Based on the proposed modeling, the theoretical first matching layer can be made by BR20 with a 19% weight ratio and 15% diluent. The fabricated sample from the modeling has a density of 625.35 kg/m<sup>3</sup> and 2186 m/s acoustic velocity. The calculated acoustic impedance is 1.36 MRayl, which is a 1.4% error from the theoretical value. The second matching layer, which can be considered as the composition of polypropylene and air bubble, needs an expansion rate of 80 based on the modeling in this study. The density, acoustic velocity, and acoustic impedance of the fabricated second layer are 67.13 kg/m<sup>3</sup> , 738 m/s, and 0.049 MRayl, respectively. The larger acoustic impedance error results from the inaccuracy of the expansion rate, which is hard to control.

To further verify the modeling results on the sensitivity, several matching layers were used to make the transducer to test the signal amplitude in the air by using the testing platform in Figure 4. Due to the small acoustic impedance in the second matching layer, the accurate controlling of acoustic impedance is difficult. Thus, only the first matching layer acoustic impedance changes in the experiments. The second matching layer used the 0.049 MRayl samples. All matching materials were sliced into 20 mm × 14 mm blocks with a quarter wavelength of the acoustic in matching material thickness and pasted on the 1-3 composite with the E51 epoxy to fabricate the transducer. From the modeling result, the first layer in BR20-19-15 has the acoustic impedance closest to the theoretical value. The other two matching layers with BR20-15-15 and BR20-20-15, with higher and lower acoustic impedance, respectively, were also tested. Moreover, another purchased transducer from Japan Probe was also used to compare the performance of the self-developed with the existing industry transducer. The received signal results are shown in Figure 6. The BR20-19- 15 (seen in Figure 6b) has the highest amplitude, up to 961.4 mV, which is about 10% higher than BR 20-15-15 (Figure 6a) and BR 20-20-15 (Figure 6c), which are 848.2 and 886.4 mV, respectively. This phenomenon confirms that the higher or lower acoustic impedance compared to the theoretical acoustic impedance will result in a weaker amplitude and reduce the performance of the air-coupled transducer. Compared to the Japan Probe transducer with the same 200 kHz resonant frequency, the self-developed transducer has a 20% higher amplitude, which further proves the significance of this study.
