A. Fiber-optic ultrasound transmitter

Distinguished from the ultrasound generation based on the PA mechanism, the thermo-cavitation employed a CW laser to trigger the nucleation of a microbubble in the liquid. To study the dynamics of the bubble growth and collapse, we injected the probe light at a wavelength of 1550 nm into the bubble and continuously monitored the intensity of the reflected light. The light intensity was converted into electrical signals by a high-speed photodetector (PD, New Focus, 2053-FC) with a built-in transimpedance amplifier. The bubble diameter was then acquired from the curve of the reflected light intensity by counting the interference fringes. As the bubble can be regarded as a low-finesse F–P cavity, one period of the fringes represents the change in the diameter of one-half the optical wavelength. This optical readout technique can replace the conventional CCD camera with a limited frame rate to continuously monitor the diameter variation for the bubble dynamics study. As shown in Fig. 2(a), the first series of fringes denotes the bubble growth period, and the growth rate gradually decreases as the fringes become sparse. The corresponding variation of the diameter as a function of time has also been plotted. The bubble begins to collapse at the time of 22 µs after reaching a maximum diameter of 90 µm. When the bubble collapses, a strong ultrasound pulse is released into the surrounding.The characteristics of the ultrasound emission from the fiber transmitter were examined by a 0.5 in.-diameter commercial piezoelectric transducer (PZT) transducer (Olympus, Panametrics-NDT-A311S) with a central frequency of 10 MHz and a bandwidth of 7.20–12.67 MHz, which was placed at a distance of 5 mm from the polymer window of the transmitter. The sensitivity of the PZT transducer at 10 MHz was calibrated beforehand to obtain the absolute pressure. A peak amplitude of ∼300 kPa and a repetition rate of 5 kHz were achieved at the heating power of 50 mW, as shown in Fig. 2(b). As the heating power continued to increase, the repetition rate of the ultrasound pulses reached up to 18 kHz and the amplitude was reduced to ∼100 kPa at a heating power of 80 mW. The explanation for the reduced pressure at higher heating power is caused by the reduced volume of the superheated volume, or the bubble diameter.2626. J. C. Ramirez-San-Juan, E. Rodriguez-Aboytes, A. E. Martinez-Canton, O. Baldovino-Pantaleon, A. Robledo-Martinez, N. Korneev, and R. Ramos-Garcia, “Time-resolved analysis of cavitation induced by CW lasers in absorbing liquids,” Opt. Express 18, 8735–8742 (2010). https://doi.org/10.1364/oe.18.008735 For the cavitation bubble in the collapse phase, the energy EB of the bubble can be estimated byEB=4πRmax33Pstat−Pv,(1)where Rmax denotes the maximum bubble radius before the collapse, and Pstat and Pv are the static pressure (∼101 kPa) and vapor pressure (∼2.33 kPa), respectively.2727. A. Vogel and W. Lauterborn, “Acoustic transient generation by laser‐produced cavitation bubbles near solid boundaries,” J. Acoust. Soc. Am. 84, 719–731 (1988). https://doi.org/10.1121/1.396852 When the heating power increases, the thermo-cavitation process accelerates and increases the repetition rate of the emitted ultrasound pulses. The shorter heating time reduces the average input energy from the heating light, thereby decreasing the maximum bubble diameter Rmax. The reduced bubble energy, as indicated by Eq. (1), caused the decrease in the amplitude of the ultrasound pulses as observed in Fig. 2(b).We also prepared several solutions with different Cu(NO3)2 concentrations (1.3, 0.65, and 0.325 g/ml) and measured the amplitude and repetition rate of the ultrasound pulses for each solution. Three ultrasound pulses at each heating power were acquired and processed. As shown in Figs. 2(c) and 2(d), the threshold power for the thermo-cavitation process increases with the reduction in the Cu(NO3)2 concentration, and the repetition rate for the different solutions follows a similar trend as that for the increasing heating power. The increase in the repetition rate with the reduction in the Cu(NO3)2 concentration can be explained by the formulaρC∂T∂t+∇⋅−k∇T=Q,(2)where ρ is the liquid density, C is the specific heat capacity, and k is the thermal conductivity of the water. Q is the heat source term given by Q = αI, where α is the absorption coefficient of the solution and I is the intensity of heating laser. From Eq. (2), it can be found that the liquid can be heated to the spinodal temperature faster for the solution with a larger Cu(NO3)2 concentration or optical absorption α,28,2928. M. Blander, D. Hengstenberg, and J. L. Katz, “Bubble nucleation in n-pentane, n-hexane and n-pentane + hexadecane mixtures and water,” J. Phys. Chem. 75, 3613–3619 (1971). https://doi.org/10.1021/j100692a02229. R. E. Apfel, “Water superheated to 279.5° C at atmospheric pressure,” Nature 238, 63–64 (1972). https://doi.org/10.1038/physci238063a0 which results in a higher repetition rate. For the relationship between the threshold power and the Cu(NO3)2 concentration, the temperature distribution nearby the optical fiber tip was simulated with the heat transfer module of the finite element analysis software (Comsol 6.0). As shown in Fig. 3(a), for the solution with a Cu(NO3)2 concentration of 1.3 g/ml, a heating power of ∼32 mW can raise the liquid temperature up to the spinodal line (300 °C) in 6.5 ms and trigger the thermo-cavitation. Once the thermo-cavitation was triggered, the heating induced thermal energy vaporized the water to form the bubble, diffused during bubble expansion/collapse, and was finally converted to acoustic energy, accompanied by the reduction of liquid temperature below the spinodal point. The temperature distribution in the Cu(NO3)2 solutions of different concentrations was also calculated and compared with the experimental data, as shown in Fig. 3(b). The threshold power for triggering the thermo-cavitation increased with a lower Cu(NO3)2 concentration due to the reduced absorption coefficient. For a heating power lower than the threshold, the solution temperature is below the spinodal point, and no thermo-cavitation happens, such as in the situation with 20 mW of heating power for the solution with a Cu(NO3)2 concentration of 1.3 g/ml as shown in Fig. 2(c). Here, the spinodal point of the water at room temperature (∼300 °C) was used in the calculation to predict the necessary heating powers for triggering the cavitation, i.e., the values denoted by the three asterisks, for the solutions with different Cu(NO3)2 concentrations, as shown in Fig. 3(b). The deviation between the theoretical and experimental results might result from the heat transfer caused by the liquid convention, which was neglected in the simulation for simplicity. To reduce the heating power required for the ultrasound generation and avoid the possible damage to the fiber tip facet caused by the high heating power, the saturated Cu(NO3)2 solution with a concentration of 1.3 g/ml and a heating power of 50 mW was selected for the following imaging experiment.It needs to be mentioned that despite the considerable temperature rise during the cavitation process, the high-temperature region is confined to a small area nearby the fiber tip and has a much smaller size than the Teflon tube with an inner diameter of 0.8 mm, as shown in the simulation results in Fig. 3(a). During the experiment, no obvious temperature rise at the tube wall was observed, indicating no potential risk of thermal damage to biological tissues for patients.To characterize the property of the emitted ultrasound, we recorded a series of ultrasound pulses with a repetition rate of ∼5 kHz, as shown in Fig. 4(a). A slight variation of the peak amplitude of the ultrasound pulses was observed due to the stochastic nature of the thermo-cavitation, which results in slightly different bubble sizes before the collapse. The large bubble generates stronger ultrasound as more energy is stored before the cavitation. By performing the Fourier transform of a single ultrasound pulse, as shown in Fig. 4(b), the frequency of the emitted ultrasound ranges from 5 to 12 MHz, which can meet the need for biological tissue imaging. We also measured the ultrasound pulses generated at different heating powers and summarized the variation of the time interval or jitter between the adjacent pulses as plotted in Fig. 4(c). To determine the jitter between consecutive US pulses, we first acquired seven consecutive pulses each time, and then six intervals of time between the peaks of two adjacent ultrasound pulses were estimated. The standard deviation of the time intervals was selected as the jitter. The pulse jitter resulted from the random thermo-cavitation occurring in the superheated volume of the solution2626. J. C. Ramirez-San-Juan, E. Rodriguez-Aboytes, A. E. Martinez-Canton, O. Baldovino-Pantaleon, A. Robledo-Martinez, N. Korneev, and R. Ramos-Garcia, “Time-resolved analysis of cavitation induced by CW lasers in absorbing liquids,” Opt. Express 18, 8735–8742 (2010). https://doi.org/10.1364/oe.18.008735 and decreased from 118.4 to 4.3 µs as the heating power increased from 40 to 80 mW. The decrease in the jitter was attributed to the smaller superheated volume created by the higher heating power, as discussed previously, and thus the smaller randomness of the cavitation. The jitter may prevent the synchronization of the ultrasound emission and receiving for phase steering if multiple fiber transmitters are assembled into an array. Fortunately, to construct a single fiber-optic ultrasound imaging probe working in the pulse-echo mode, the ultrasound wave before reaching the imaging object will first arrive at the ultrasound detector located close to the transmitter. The converted electrical signals can then be used as the trigger signal for synchronal detection of the ultrasound echo signals. This trigger signal can also be used to compensate for the fluctuation in the cavitation-generated ultrasound amplitude, as mentioned previously. The directivity of three fiber transmitters with the same preparation parameters was also measured by adjusting the incident angle of the ultrasonic waves and is shown in Fig. 4(d). The ripples at different angles as observed might be caused by the interference of the ultrasound waves reflected between the probe and the flat PZT transducer. As a result of the spherical shape of the generated microbubble in the cavitation, the emitted ultrasound behaves as a spherical wave and exhibits a broad emission angle from −60° to 60°. As observed from the inset of Fig. 4(d), which shows the ultrasound field distribution measured by scanning a focused 10 MHz PZT transducer (Olympus, V312-SM), the angle range is mainly limited by the window size (0.8 × 1.2 mm2) of the Teflon tubing and the distance between the fiber tip and the window. The low directivity can broaden the view angle for nondestructive testing and ultrasound imaging.

B. Fiber-tip ultrasound detector

For ultrasound imaging, a fiber-tip polymer F–P ultrasound detector was developed to integrate with the optical fiber ultrasound transmitter. The thin polymer film sandwiched between a pair of Au reflective mirrors worked as a high-finesse F–P etalon and converted the ultrasound wave into a change in the intensity of the reflected light. Figure 5(a) shows the measured reflection spectrum of the detector by delivering light from a broadband amplified spontaneous emission (ASE) source (Golight, ASE-C+L) to the fiber tip. The reflected light traveling through the three-port C+L band optical circulator (MChlight, MCCIR-CL-00-C-S2-10-L-FA) was recorded via an optical spectrum analyzer (OSA, Yokogawa, AQ6370B) with the spectral range of 600–1700 nm and the resolution of 0.5 nm. The spectrum of the ASE source was also measured by the OSA, as plotted in Fig. 5(a). The fringe period of the reflection spectrum is ∼32.7 nm, corresponding to a cavity length of 14 µm, assuming the polymer refractive index of 1.48. This value agrees well with that estimated from the microscopic image of the polymer cavity at the fiber tip [see inset of Fig. 5(a)]. To characterize the sensitivity of the detector, a narrowband laser source (Santec, TSL-570) with a power of 4.8 mW was used, and its wavelength was tuned to the maximum slope point of the etalon resonant dip for optimum sensitivity. The sensitivity of the F–P ultrasound detector was evaluated by measuring its response to an ultrasound pulse with a peak amplitude of 500 kPa and a central frequency of 10 MHz. The signal-to-noise ratio (SNR) of 260 corresponded to a noise equivalent pressure (NEP) level of 1.9 kPa, as shown in Fig. 5(b). As the optimum sensitivity could be obtained only when the wavelength of the probe light was located at the maximum slope point of the etalon resonant dip, a PID-algorithm-based servo-control program was employed to lock the operation point, which might drift due to environmental temperature fluctuation as well as mechanical vibration. Instead of using an expensive continuously-tunable laser to adjust its output wavelength according to the feedback signal, a part of the CW 980 nm laser light was split out with its power dynamically controlled to photothermally tune the reflection spectrum and lock the operation point. As shown in Fig. 5(c), the reflection spectrum shifts with the increase of the 980 nm light power, as indicated by the dashed line, and the wavelength shift reaches ∼3.6 nm for the light power of ∼22.3 mW, as a result of the thermal expansion of the polymer etalon after the light absorption by the Au film. During the stabilization process, the reflected light was detected by the PD with the AC component of the output voltage used for retrieving the ultrasound signals and the DC component used as the feedback signal to monitor the spectrum shift, as illustrated in the inset of Fig. 5(d). Once the servo-control process started, the maximum slope point of the spectrum could be well stabilized to the wavelength of the probe light, as indicated by the small fluctuations of the output voltage in Fig. 5(d), by dynamically adjusting the 980 nm light power via an electronic variable optical attenuator (EVOA).

C. Imaging experiments

The experimental setup for the ultrasound imaging is illustrated in Fig. 6. The light from the CW 980 nm laser was split into two portions by the optical coupler. One part of the light was delivered to the fiber ultrasound transmitter to generate ultrasound waves, and its power was controlled by a mechanically-tunable VOA (MChlight, MCVOA-980-00-H1-10-L-FA) with a central wavelength of 980 nm. The other part of the light was combined with the narrow band probe light by a taper-type 980/1550 nm wavelength division multiplexer (WDM, MChlight, MCWDM-9815-00-H1-10-L-FA) and delivered to the fiber ultrasound detector. This portion of the 980 nm light was used for photothermal stabilization of the detector and its power was servo-controlled by the EVOA (Thorlabs, EVOA800A) , which could adjust its optical attenuation by varying the control voltage with a response time of less than 1 ms. As mentioned previously, the reflected light was detected by a PD, and the direct current (DC) voltage of the output signal was used as the feedback signal. The alternating current (AC) voltage was first amplified by 100 times with the built-in electrical transimpedance amplifier of the PD and then acquired by a 14 bit and 500 MSa/s data acquisition board (DAQ, AlazarTech, ATS9350) with the sampling rate set to 250 MSa/s. The all-fiber ultrasound imaging probe integrating the fiber-tip ultrasound transmitter and detector was immersed in a water tank mounted onto a translation stage. During the imaging process, the fiber-optic ultrasound transmitter emitted an ultrasound pulse toward the imaging object. The ultrasound pulse, before reaching the object, would first pass through the polymer window and be received by the fiber ultrasound detector. This output voltage signal from the detector was used as the trigger signal for the DAQ board to acquire the following pulse-echo signal from the imaging object. Once the signal acquisition was completed, the stepper-motor-driven mechanical translation stage (BeiJing Optical Century Instrument, MTS204) moved forward with a step of 10 µm for a distance of 15 mm to acquire the next A-line. No average was performed on all the received signals during the imaging. After the scanning process, the object image was reconstructed using the acquired A-lines using the delay-and-sum algorithm with the sound speed assumed to be 1.48 mm/μs.30,3130. D. Feng, Y. Xu, G. Ku, and L. V. Wang, “Microwave-induced thermoacoustic tomography: Reconstruction by synthetic aperture,” Med. Phys. 28, 2427–2431 (2001). https://doi.org/10.1118/1.141801531. V. A. Del Grosso and C. W. Mader, “Speed of sound in pure water,” J. Acoust. Soc. Am. 52, 1442–1446 (1972). https://doi.org/10.1121/1.1913258Figure 7(a) shows the A-lines at different positions as indicated by the arrow in the reconstructed B-scan image in Fig. 7(b), acquired by scanning a stainless steel cylindrical rod of a diameter of 6 mm with a duration of ∼2.4 s. As shown in Fig. 7(b), the view angle of the probe is ∼120°, which is beneficial to improve the spatial resolution of the reconstructed image.3232. J. Ma, Y. He, X. Bai, L.-P. Sun, K. Chen, K. Oh, and B.-O. Guan, “Flexible microbubble-based Fabry–Pérot cavity for sensitive ultrasound detection and wide-view photoacoustic imaging,” Photonics Res. 8, 1558–1565 (2020). https://doi.org/10.1364/prj.394941 The single amplitude at the boundary of the view angle was reduced to one-half of the maximum value. The relatively large view angle and weak acoustic shadowing enabled the imaging of hair with no obvious “X”-shape artifacts, as shown in Fig. 7(c). The lateral and axial resolutions of the probe, as estimated from a human hair image [Fig. 7(c)], are 86 and 91 μm, respectively, which are close to the half wavelength of the ultrasound wave at the frequency of 10 MHz. As the full size of the probe is ∼1.2 mm, the miniature size is attractive for endoscopic imaging. As a proof-of-concept demonstration, the probe was inserted into a phantom made of a plastic tube with six iron wires bonded to the inner surface. The wires with a diameter of ∼0.8 mm were evenly fixed on the inner surface of the plastic tube with a diameter of ∼1.5 cm by using ultraviolet curable epoxy. By acquiring the A-lines at each step of the mechanical rotation, the cross-section image of the tube was obtained, as shown in Fig. 7(d). The clearly observed six wires locate at the inner surface of the tube proving the potential of this all-fiber ultrasound probe for endoscopic biomedical imaging and industrial nondestructive testing. Due to the omni-directional nature of the ultrasound emitted during the cavitation, the Teflon tube can cause ultrasonic ringing and reduce the imaging dynamic range, which may be reduced by coating its inner surface with sound-absorptive materials such as the microparticle-dispersed polymer layer.3333. M. Nagli, J. Koch, Y. Hazan, O. Volodarsky, R. Ravi Kumar, A. Levi, E. Hahamovich, O. Ternyak, L. Overmeyer, and A. Rosenthal, “Silicon-photonics focused ultrasound detector for minimally invasive optoacoustic imaging,” Biomed. Opt. Express 13, 6229–6244 (2022). https://doi.org/10.1364/boe.470295 For imaging applications desiring ultrasound emission with different directivities, adjusting the open window of the tube for ultrasound transmission, coupled with coating the tube inner surface with sound-absorptive materials to reduce the ringing, may allow to tailor the directivity of the ultrasound emission.