Optical fiber acoustic sensing technology was rapidly developed in the 1970s [1,2], due to its high accuracy of light wave detection combined with the advantages of optical fiber working frequency bandwidth and small transmission loss. Compared with free-space optical microphones [3,4] and traditional electroacoustic sensors [5,6], optical fiber acoustic sensors [7–9] have the advantages of high sensitivity, high signal-to-noise ratio, wideband response, and large dynamic range. Simultaneously, due to its immunity to electromagnetic interference and its ability to be miniaturized, optical fiber acoustic sensing technology can be employed in environments characterized by high temperatures, high pressures, robust corrosion, intense radiation, and other similar conditions where traditional electroacoustic sensors have difficulty working normally. Therefore, optical fiber acoustic sensors have been valued and widely used in national defense security, industrial non-destructive testing, photoacoustic (PA) imaging diagnosis, and other fields [10–12].

Recently, optical microcavities integrated on optical fibers have attracted considerable interest in ultrasound detection and photoacoustic imaging [13–15]. With the enhanced acousto-optic interaction [16–21], microfiber-coupled microcavity devices can further obtain a noticeable improvement for ultrasound sensing [22–25]. In particular, whispering-gallery-mode (WGM) microcavities with ultrahigh-Q factors (Q=λ/Δλ, λ is the central wavelength, and Δλ is the linewidth of the resonance mode) have shown exceptional detection performances in both sensitivity and bandwidth [26–29]. However, the evanescent coupling between microcavities and fragile microfibers, in order to enable signal readout, disrupts the inherent benefits of optical fiber sensing, notably its small dimensions and high stability as a probe. Additionally, achieving the delicate coupling in aqueous media employed for acoustic transmission remains a challenging task.

In this work, we demonstrate an ultrasensitive and broadband ultrasound microprobe using a WGM microcavity integrated on an optical fiber. The miniature ultrasound detector has a probe tip that encapsulates a 30-μm microspherical polystyrene cavity coupled to a U-shaped tapered microfiber. The high-Q factors and miniature sizes of the polymer microcavity probe enable a sensitivity up to 1.07mPa/√Hz and record frequency responses up to 150 MHz. Moreover, the small microprobe size not only allows for a large sensing angle over 180° but also enables approaching targets that are much closer, which can significantly reduce the ultrasound signal loss due to transmission diverging and medium attenuation. The high performance of this microprobe is demonstrated by in vivo PA imaging of zebrafish with high-contrast, whole-body microvascular maps. This work provides a new strategy for developing ultrasound microprobes based on highly sensitive WGM microcavities. The new ultrasound (US) microprobe has the broad potential to be implemented in ultrasound/PA endoscopy and near-field detection.

We demonstrate an application scenario in PA detection with a miniature microprobe detector and further show the detailed composition of the microprobe, as shown in Fig. 1. In PA detection and imaging, objects convert the energy absorption of pulsed light into detectable ultrasound signals. In our later experiments, the high-sensitivity microprobe will serve as an important tool for reading out the ultrasound signal, which can be even used in complex environments, including uneven surfaces and pipes. The detection part of the probe tip, shown in the enlarged inset, consists of a U-shaped fiber coupled to a polystyrene microspherical cavity. Such a microprobe can detect the excited PA signal more closely and effectively from the optical absorber. The working principle of the ultrasound probe relies on the ultrasound-induced resonance shift of the microcavity by modulating its refractive index and geometry [13,30]. When the detection wavelength is fixed at the maximum slope of the cavity mode, the frequency and amplitude information of the ultrasound can be read directly by the change of output laser intensity. Experimentally, the needle-like probe is formed by a 30-μm-diameter polystyrene microsphere (refractive index of 1.6) coupled to a U-shaped microfiber, and the whole microcavity-microfiber coupled system, with a low refractive index of 1.33 for stable detection, is encapsulated by glue. This concept of the miniature ultrasound microprobe was first proposed and implemented by us, and the detailed production process is described in Appendix A.

The Q factor of the encapsulated microprobe is characterized using a 980 nm tunable laser. As shown in Fig. 2(a), multiple sharp cavity modes with high Q factors exceeding 106 and feasible coupling depths are revealed in the laser transmission. The sensitivity of our microprobe sensor is measured by detecting ultrasound waves from a calibrated flat ultrasound transducer of 20 MHz driven by a pulse/echo receiver. The detected signal after a 1.2–800 MHz band-pass filter is shown in Fig. 2(b). The noise equivalent pressure (NEP) is calculated to be as low as 4.8 Pa at an electronic noise level of 4.6 mV without signal averaging, which is converted to 1.07mPa/√Hz. Next, we further measured the directivity of the microprobe by testing the ultrasound response of the microprobe to different receiving angles. As shown in Fig. 2(c), we fixed the microprobe in the water and then rotated the flat ultrasound transducer 180°. We repeatedly measured the angular response of the microprobe to the ultrasound at 5 MHz, 10 MHz, and 20 MHz frequencies. The normalized directivity response is shown in Fig. 2(d). The −6 dB response covers the entire 180° range among those frequencies that will facilitate the detection of the ultrasound, especially PA/ultrasonic tomography.

We then built the PA microscopy (PAM) imaging system, as shown in Fig. 3(a). The PA excitation light is focused upwardly onto the target, and the microprobe detects the PA signal from above. We use a continuously tunable laser at 980 nm as the detection light and a pulsed laser at 532 nm as the PA excitation light. The microprobe sensor and sample are placed in a transparent water tank. The beam width of the 532-nm-pulsed laser is doubled by a 4f lens system and focused on the sample through an objective lens. For PA signal detection, we lock the detection light at the wavelength corresponding to the maximum slope of the mode to obtain the maximum signal response.

To measure the spectral response of our microprobe, we used a 532-nm-pulsed laser (pulse width ∼1.8ns, pulse energy ∼2μJ) to irradiate a gold film with a thickness of 50 nm to generate a broadband PA signal. The detected time-domain PA signal is shown in Fig. 3(b), and its Fourier transform result is shown in Fig. 3(c). The results show that the response is broadband, extending to approximately 150 MHz.

Then, taking advantage of the tiny microprobe tip, we explored the PA signal amplitude attenuation along various distances from a small target. A tungsten wire with a diameter of 20 μm was illuminated by a focused pulsed light of 532 nm to serve as a tiny point PA source. As long as the microprobe tip approaches the point source from 6 mm away to 200 μm away, the measured PA signal increases approximately inversely proportional to the source-detector distance, which is shown in Fig. 4(a). Ultrasound diffuses in the form of spherical waves. Therefore, the energy attenuation of the ultrasound is positively correlated with 1/r2 according to the spherical surface area formula; then the attenuation of ultrasonic amplitude is positively correlated with 1/r [31]. We then use 1/r fitting with an R2 of 0.97. We can see that the 1/r fitting curve fits well at closer distances and starts to deviate with larger distances. We think that the attenuation of the signal at a closer distance is mainly caused by the divergence of spherical waves, and as the distance increases, there is not only the effect of divergence but also the attenuation of high-frequency ultrasound. This result demonstrates that the near-field-like detection ability of this microprobe has a large advantage, especially for small objects with weak signals and high frequencies.

When performing photoacoustic microscopy, we use a 2D scanning stage to continuously move the sample, with the microcavity and focused spot remaining fixed. We then imaged a sharp blade edge to calibrate the lateral resolution in Fig. 4(b). The detected data points are fitted with a black curve which is the so-called line spread function, as shown in Fig. 4(c). The red curve (the edge spread function) is obtained from the derivation of the black curve, and its full-width at half-maximum (FWHM), which equals 3.4 μm, represents the lateral resolution of the PAM system. We can further improve the lateral resolution by expanding the beam and increasing the numerical aperture of the objective.

To further validate the performance of the microprobe for biomedical imaging, we performed in vivo imaging of a zebrafish two days post-fertilization (dpf). Newborn zebrafish, whose red blood cells are just beginning to be produced, are less abundant, and most of the body is transparent. PA imaging of in vivo zebrafish is a challenge. The zebrafish were depigmented during incubation for better PA imaging of blood vessels (see Appendix A). Then, the zebrafish were anesthetized and fixed with 2% gel in a small glass petri dish. Their physiological state and blood flow were completely normal. We add clean water to the petri dish and place the microprobe sensor directly above the zebrafish. Next, we adjust the focus of the light spot to maximize the PA signal and set the scanning interval of the translation stage. We trigger the pulsed light, the translation stage, and the acquisition card synchronously to perform PA scanning and signal acquisition.

The PA microscopy image of the zebrafish is shown in Fig. 4(d), which shows the maximum amplitude projected image with the distribution of blood vessels in the zebrafish body. The illustration in the upper right corner shows a brightfield microscope of the same zebrafish, and we can see its outline and the distribution of the fishbone. The zebrafish heart is highlighted in the PA image compared to the brightfield image. We can also see that the amount of blood on both sides of the yolk is different and many capillaries can be displayed in the PA image. Further, we give profile views of the PA imaging of zebrafish at different depths, as shown in Fig. 4(e). The vascular distribution of zebrafish is significantly different in tomographic images at different depths. The most obvious is around the eyeball, where a circular lens structure can be seen in the first image. This is due to the fact that the lens is nearly transparent and has weak absorption of pulsed light, while the surrounding blood vessels have strong absorption. The other two images show vascular and pigment distributions in the lower layer of the eye. The distribution of blood around the yolk also varies significantly at different depths. We also estimate the axial resolution of the system to be about 20 μm, based on the imaging of capillaries in the zebrafish.

In summary, we propose a novel scheme to make a WGM microcavity-based US microprobe, in which we encapsulated a microsphere coupled with a U-shaped optical fiber. Our microprobe has superior comprehensive performance characteristics, including high sensitivity, wide bandwidth, large detection angle, and miniaturization. The NEP of the microprobe sensor can reach 1.07 mPa/√Hz, and our microprobe has a broad frequency response of approximately 150 MHz, which can satisfy most ultrasound medical imaging and PA imaging applications [32,33]. The comparison of our sensor with other state-of-the-art microresonator-based acoustic sensors used in photoacoustic imaging is presented in Table 1. Our sensors have a decisive advantage in comprehensive performances.Table 1.

Comparison of Our Sensor with Other State-of-the-Art Microresonator-Based Acoustic Sensors Used in Photoacoustic Imaging [7,9,24,27,28]^a

Further, this ultra-large-angle ultrasound response is also beneficial for PA computed tomography, which relies on image reconstruction after the detection of PA signals from different angles. The needle-like microprobe will greatly benefit detecting weak PA signals from small targets since this microprobe can be placed as close as tens of micrometers to the target, substantially reducing the signal loss. We demonstrated the superior performance of this microprobe by in vivo PA microscopy of zebrafish. Our developed microprobe has broad application prospects in the field of biomedicine, such as endoscopy. What is more, this microprobe can be feasibly fabricated in the lab because it does not require special coating like an FP cavity, nor does it require a complicated lithography etching process in micro-nano processing.

Acknowledgment. The authors thank Professors Jingwei Xiong and Shiqing Qiu at Peking University for their help in breeding the zebrafish used in the PA imaging experiment. S.-J.T. is supported by the China Postdoctoral Science Foundation.