Stack-layer dual-element ultrasonic transducer is proposed to cover a broad bandwidth at a coincident position in the acoustic field. First, two active elements, i.e., two piezoelectric layers, were used to cover two spectral ranges, which overcame the intrinsic bandwidth limitations of piezoelectric materials. Second, the high-frequency element was in front of the low-frequency one, which formed a stack-layer arrangement spatially. The stack-layer arrangement of the two elements assures the overlap of the two beams and enables broadband coverage within the beam.

The cross-sectional structure of the dual-element transducer and photograph of the prototype for PAT imaging are shown in Figures 1A,B. The outer diameter of the brass transducer is 12 mm. The piezoelectric elements are made of PZT-5H 1–3 composite, and the aperture is 7 mm × 5 mm and 4 mm × 4 mm for the low and high-frequency elements, respectively. The aperture difference of the two elements was due to the need for electrical impedence matching to the input impedence of the data acquisition system of 50 Ω. Because piezoelectric elements with higher frequencies have smaller electrical impedence. To balance the two elements, the aperture of the high-frequency element needs to be smaller. Gold was sputtered on both surfaces of the piezoelectric layers as electrodes. Tungsten powder mixed with epoxy resin (EPO-TEK 301) was centrifuged and applied to the ceramic as the backing layer. A mixture of alumina powder and epoxy was used for the first matching layer, and pure epoxy was used as the second matching layer. Properties of piezoelectric materials, acoustic matching layers, and backing layer used for the dual-element transducer are shown in Table 2.

The impulse response waveforms of the 3.06 and 11.07 MHz transducer elements were characterized by a needle hydrophone (HGL-0200, Onda Corp, Sunnyvale, CA, United States). In the measurements, electrical pulses from a function generator (UTG2062B, Uni-Trend Technology Co., Guangdong, China) excited the transducer elements and signals from the needle hydrophone were recorded by a digital oscilloscope (MSO54, Tektronix, Inc., Beaverton, OR, United States). To ensure the measurement accuracy, the hydrophone was placed coaxially with the dual-element transducer in a deionized water tank and scanned along the axial direction of the transducer to find the waveform with the maximum peak-to-peak voltage. We compared the time-domain waveforms and their corresponding frequency spectra. The transmission responses of the 3.06 MHz element are shown in Figures 2A,B. The –6 dB bandwidth covers from 1.54 to 4.58 MHz, corresponding to a fractional bandwidth of 99.3%. Figures 2C,D show the transmission responses of the 11.07 MHz element, leading to a –6 dB bandwidth coverage of 6.35–15.78 MHz and a fractional bandwidth of 85.2%.

It’s noticed that there are some alias echos in the high-frequency signal, as noted with an arrow in Figure 2C. The inconsistency of the acoustic impedance between the backing and the piezoelectric material is the main cause of aliasing echo. We glued the backing to the piezoelectric layer with a low-impedance epoxy resin, which also resulted in the reduction of the equivalent acoustic impedance. The reduction of the acoustic impedance of the backing layer results in the reflection of high-frequency echoes at the backing, resulting in some aliasing echoes.

The photoacoustic sensing capability of the dual-element transducer was evaluated on a leaf phantom and a rat brain experimentally. First, to test our transducer, one piece of leaf veins with ∼8 mm dimension was used as a phantom, which was placed on a table with a diameter of 3 cm. Second, the vasculature of the rat brain was used to demonstrate the imaging performance. Both imagings share the same experimental setup with a two-dimensional (2D) scanning. The schematic of the 2D PAT system was shown in Figure 3. The target was illuminated by using a pulsed laser from an optical parametric oscillator (OPO) laser (SpitLight 600 OPO-532 mid band, Innolas). The optical intensity on the top of the phantom was about 5 mJ/cm², and the repetition rate was 20 Hz. The total step number of the 2D scan was 360 with an angular step of 1°. The distance between the rotation center to the transducer detection surface was about 30 mm. The signal was first amplified by a pulser/receiver (DPR500, Ultrasonics), and then digitized with an acquisition card (NI-5124, 12 bit, 100 MHz sampling frequency) in the computer. The whole system was synchronized with the laser. And the data were collected from the hard disk for later processing.

We performed functional PAT of rat brains in vivo to test the capability of our dual-element transducer and assessed the oxygenation level of hemoglobin and [tHb] level simultaneously. Sprague Dawley rats (∼60 g, 3 weeks, Hunan SJA laboratory animal Co., LTD.) were used. The protocol of animal experiments has been approved by the animal ethical committee of the Central South University of China. The hair on the head of the rat was removed using hair remover cream before imaging. The rat was anesthetized with Pentobarbital [120 mg/kg, Intra-peritoneal (IP)] and kept motionless throughout the experiment. A homemade animal holder was used to fix the rat head. And a transparent membrane between water and rat head was used to seal the cylindrical hole.

According to the absorption difference between hemoglobin and tissue, a pulse laser with a wavelength of 760 and 840 nm is used to image the vascular morphology, total hemoglobin, and oxygen saturation distribution. This enables us to achieve multi-parameter photoacoustic tomography.

In order to obtain the photoacoustic signal for each image, the dual-element transducer scanned 360 steps in the horizontal plane around the cerebral cortex at the back of the brain with 1° per step. For signal averaging, four laser pulses were applied at each scanning position. The signal acquisition period of each imaging was about 10 min. In total, we obtained four images corresponding to two frequencies at two wavelengths. After collection of data for imaging, the rat recovered normally and no obvious health problems were observed. In the end, the rat was sacrificed using pentobarbital.

PAT acquisitions with dual-element transducers were conducted to image the phantom. And PAT acquisitions with 760 and 840 nm laser wavelengths were performed to image the rat brain. Then, the Hilbert transform was applied to the signals, and the resulted complex data was employed for the reconstruction of two-dimensional PAT images with a conventional back-projection method (Xu and Wang, 2005). The final reconstructed images were presented by the amplitudes of the pixel values. In this work, we also merged 3.06 MHz element images and 11.07 MHz element images to display the morphology and distribution of the rat brain and its surrounding tissues. For the merged 2D PAT images, the equation is used to calculate pixel values in the hue, saturation and value (HSV) color model (Wang et al., 2021). h = 2 / ( 3 π ) × a n g l e ( A 3.06 + i A 11.07 ) s = 1 v = mod u l u s ( A 3.06 + i A 11.07 ) Here, h, s, and v represent the hue, saturation, value components of the pixel, A 3.06 , A 11.07 are the normalized photoacoustic image acquired by 3.06 MHz frequency element and 11.07 MHz frequency element, i is the imaginary unit.

Photoacoustic images of a piece of leaf veins obtained by the dual-element planar transducer are shown in Figure 4. Figures 4A,B are the reconstruction PAT images employed by the 3.06 MHz element and 11.07 MHz element, respectively. A merged image of Figures 4A,B is shown in Figure 4C. A photograph of the piece of leaf veins shown in Figure 4D was used for comparison.

The color pixel in Figure 4C represents the relative intensities of the 3.06 and 11.07 MHz frequencies photoacoustic signals. It becomes purely red if only 3.06 MHz photoacoustic signals are generated, and a purely green pixel means that only 11.07 MHz photoacoustic signals are received. The overall signal intensity is represented by the brightness of the pixel. As shown in Figure 4A, the 3.06 MHz element has a better imaging effect on the main vein branches but has a poor reconstruction effect on the small branches. On the contrary, the 11.07 MHz element image the small veins clearly with a compromised performance at the main veins. Therefore, with the benefit of our dual-element transducer, the broadband signal can be obtained simultaneously at one scan without changing the rotation phase. This has been indicated with the merged reconstruction image in Figure 4C, which is similar to the photograph of the leaf veins shown in Figure 4D.

By application of our transducer, we obtained the results of photoacoustic reconstruction of the rat cortex. Results given by the 3.06 MHz element, the 11.07 MHz element, and the merged signals at the 760 nm laser are shown in Figures 5A–C, respectively. Similarly, Figures 5D–F show the set of the three images excited by the 840 nm laser. Since tissues have different absorption characteristics under various light excitations, the PAT images of the two wavelengths show different structural characteristics. Images with 760 nm laser excitation mainly show the distribution of deoxyhemoglobin, whereas those with 840 nm laser excitation mainly show the distribution of oxygenated hemoglobin.

The brain images obtained under the 760 and 840 nm laser present the same vascular structure but different magnitudes of optical absorption. With high absorption contrast between the blood and background brain tissue, all of the brain images show the superior sagittal sinus and some branches can be clearly observed and match well with the vascular.

Similar to the leaf veins results, we obtained a good imaging effect on the main vascular branches but a poor effect on the small vascular by the 3.06 MHz element with both 760 and 840 nm of optical excitation (Figures 5A,B). In comparison, the small vascular, but not the main vascular, was observed by the 11.07 MHz element with the two excitations (Figures 5B–E). Therefore, our results further confirmed that the dual-element transducer obtained the 11.07 MHz and the 3.06 MHz information at the same time without changing the rotation phase.

For further investigation, we calculated the [tHb] and [SO₂] distribution of the rat brain [10]. We selected the region around the superior sagittal sinus as the region of interest. Results showed that the levels of [SO₂] and [tHb] in this region were significantly higher than that of other branches in Figure 6.