Liquids play indispensable roles across diverse domains including medical diagnostics, industrial processing, environmental monitoring, food safety, and scientific research. Precise determination and control of liquid concentrations constitute a critical prerequisite for their effective utilization. In clinical medicine, for instance, solution concentration analysis not only enables neurological disorder assessment^[1,2] but also supports noninvasive biofluid profiling through metabolite quantification^[3]. Such capabilities drive the demand for high-precision and miniaturized liquid measurement technology to enable in vivo diagnostics and portable medical devices^[4]. Within industrial sectors, concentration monitoring ensures quality control in chemical concentration ratios^[5] and dye synthesis^[6]. In the environmental sector, high-sensitivity liquid detection technologies serve as a critical safeguard for environmental health and safety, enabling both the screening of trace-level pollutants in industrial waste^[7,8] and real-time water quality monitoring in intensive aquaculture operations^[9,10]. Despite the pivotal importance of high-sensitivity liquid concentration analysis, achieving this objective remains technically challenging. Methods for measuring liquid concentration can be primarily divided into electrical^[11,12], chemical^[13,14], and optical categories^[15–17]. Optical methods, owing to their low destructiveness, high precision, and real-time capabilities, have become essential technologies in this field, particularly suitable for complex environments and trace analysis scenarios^[18,19].

Current optical mainstream methodologies of liquid concentration analysis include high-performance liquid chromatography (HPLC), ultraviolet-visible (UV-Vis) spectrophotometry, and surface-enhanced Raman spectroscopy (SERS). While HPLC offers exceptional separation capability and accuracy, it suffers from sample pretreatment and prolonged analysis durations^[20]. UV-Vis spectrophotometry provides rapid and cost-effective detection but requires optically transparent samples. Turbid or colored liquids demand purification, and substances lacking UV-Vis absorption (e.g.,  inorganic salts and non-conjugated organics) cannot be directly detected^[21]. Although SERS achieves remarkable sensitivity through plasmonic enhancement, its performance relies heavily on engineered substrates (e.g.,  Au@AgNPs), mandating exogenous chemical additives (e.g.,  NaCl) to induce nanoparticle aggregation for signal amplification^[22]. This exogenous intervention compromises true in situ and nondestructive analysis.

To address these limitations in sensitivity, noninvasiveness, and the need for in situ measurement, a way of photoacoustic sensing has emerged as a promising alternative. By converting optical absorption into acoustic waves via thermoelastic expansion, photoacoustic techniques combine molecular specificity, inherent immunity to optical scattering, and noncontact operation, thereby enabling in situ analysis of complex liquids, such as glucose concentration monitoring^[23], dye concentration measurement^[24], and related applications, such as nano analysis^[25,26] and photoacoustic spectroscopy^[27]. In photoacoustic methods, the sensitivity of ultrasound detectors is crucial for effective detection. Nevertheless, conventional ultrasound transducers face intrinsic constraints: their sensitivity plateaus under fixed geometries^[28], and finite resonance bandwidths restrict high-fidelity detection of broadband acoustic signals. Currently, novel kinds of piezoelectric or capacitive ultrasound transducers, such as PMUTs (piezoelectric micromachined ultrasound transducers), are widely used^[29,30]. Unlike traditional large-scale piezoelectric transducers, PMUTs utilize micrometer-scale piezoelectric films, making them ideal for miniaturized, integrated concentration detection applications in biomedical and industrial fields. Numerous approaches now employ PMUTs in photoacoustic systems to measure liquid concentration^[31,32]. The detection limit reported for PMUT-based photoacoustic liquid concentration measurement reaches about 6.397×10−4g/mL of Methylene Blue (MB)^[33]. With the continuous advancement of photoacoustic (PA) technology, more sensitive and highly integrated ultrasound detectors have become an indispensable part of PA research.

In this paper, we propose a novel dual-layer polymer microcavity ultrasound probe, along with a fully integrated mode-locked system, for ultrasensitive solution concentration detection for the first time. This system achieves automatic mode searching and locking for the microcavity probe through temperature tuning of a cost-effective distributed feedback (DFB) laser diode. Matching light sources with different solutions enables the detection of varying solution concentrations. Our whispering-gallery-mode (WGM) microcavity probe system, which creatively integrates the microcavity probe, light source, and its control system, is highly miniaturized and precise, allowing direct use in liquids after packaging. By virtue of its direct liquid immersion capability without requiring sample extraction or pretreatment, the system enables authentically in situ and non-destructive detection, thereby achieving the highest reported accuracy among truly noninvasive PA concentration measurement methodologies to date. Using this probe, we successfully measured the photoacoustic signals generated by crystal violet (CV) solutions of different concentrations. The optical setup is remarkably simple, requiring only a low-cost single-frequency DFB light source to achieve a limit of detection (LOD) for CV as low as 10−7g/mL. This probe system maintains exceptional stability and sensitivity in complex conditions, and thus is highly suitable for environments outside the laboratory.

The structure and detection mechanism of the polymer probe are illustrated in Fig. 1. During operation, the polymer probe is immersed in the test liquid, where photoacoustic signals are generated by a pulsed laser and subsequently detected by the probe. Ultrasound signals are reconstructed by tracking the resonance wavelength shift of the polymer microcavity. In this study, we have innovatively proposed a novel polymer microcavity ultrasound probe that significantly enhances the ultrasound sensitivity of the microcavity. The support structure of the probe consists of a silica microsphere, typically around 40 µm in diameter, which is formed by melting the end of an optical fiber core. The precise diameter of the microsphere can be controlled by the laser power and heating time applied during this melting process, with a typical precision of approximately 1 µm. Polymer droplets are then transferred to the silica microsphere using a tapered optical fiber. The volume of these droplets, estimated from their diameter observed on the tapered fiber, allows for the control of the final polymer coating thickness. Due to the surface tension of the polymer droplets, a uniform thin polymer coating is naturally formed on the surface of the silica microsphere. This coated microsphere is coupled with a U-shaped optical fiber via an evanescent field. The whole structure is encapsulated under critical coupling conditions using a point encapsulation process to form the polymer microcavity probe. The working principle of this ultrasound probe is to utilize ultrasound to modulate the refractive index and geometry of the microcavity, thereby inducing resonance shifts^[34]. The concept of a probe-shaped microcavity ultrasound sensor was also proposed in our previous work^[35].

The experimental characterization reveals that the polymer probe achieves a Q-factor of 1.2×106. The polymer material used for coating has a low Young’s modulus, which provides excellent deformation properties. Compared to traditional silica materials, this design significantly improves the acoustic sensitivity of the probe. Furthermore, the refractive index of the encapsulation material is lower than that of the polymer material, which diminishes optical field losses of the microcavity, thereby preserving a relatively high Q-factor. Notably, the formation of the polymer structure relies entirely on the surface tension of the polymer material, eliminating the need for complex manual operations and enabling easy fabrication. This novel polymer probe, with its compact structure, high sensitivity, and simple fabrication process, offers significant convenience for high-performance ultrasound detection.

For microcavity ultrasound probes, a high Q-factor is not the sole determinant of detection performance. A WGM microcavity usually holds multiple resonant modes; each represents a specific light field structure within the cavity, and each bears a specific resonant wavelength. By adjusting the wavelength of the laser entering the microcavity or modifying the physical parameters of the cavity (e.g.,  temperature or external pressure), different modes can be selectively excited, each with distinct Q-factors. Furthermore, even within the same mode, the acoustic detection sensitivity can vary significantly depending on the probe’s operating position within the mode.

Thus, the measurement sensitivity of the microcavity ultrasound probe is heavily dependent on its operating point. In complex and dynamic measurement environments, external factors such as temperature and pressure fluctuations can disrupt the stability of the probe, making it challenging to maintain a consistent operating point. This highlights the critical importance of locking the operating point.

To achieve precise mode control, a cost-effective and user-friendly tunable laser source is indispensable. We developed a temperature-tuned DFB laser system to realize a tunable light source, as shown in Fig. 2.

Figure 2(a) illustrates the overall circuit structure of the light source. The main control unit regulates the temperature of the DFB laser via a temperature control system, which alters the cavity length of the DFB laser, enabling tunable output wavelength. Additionally, a constant current source is used to control the input current to the DFB laser, ensuring that only the DFB laser’s temperature changes. The system also incorporates functionalities for data sampling and serial communication.

Figure 2(b) provides a detailed depiction of the internal temperature control process for the DFB laser. The core of this process is a continuous loop consisting of four main stages: temperature setting, DFB temperature change, DFB temperature measurement, and temperature feedback. In this setup, the internal temperature of the DFB laser is measured by a thermistor (Rt) in the temperature detection module, while temperature regulation is achieved using a thermoelectric cooler (TEC). The microcontroller determines the desired voltage value Vset corresponding to the target temperature, either through internal computations or external input. This value is compared with the voltage across the thermistor Vtemp, and both are fed into the differential input of the PID control module. The PID module processes the difference between Vset and Vtemp, referred to as the system error, using proportional (P), integral (I), and derivative (D) calculations. The output of the PID module Vout drives the pulse width modulation (PWM) generator, which generates PWM signals to control the opening and closing of the metal-oxide-semiconductor field-effect transistor (MOSFET) gate. This adjusts the drive current to the TEC. The DFB laser chip is mounted on the TEC, with a thermal pad placed beneath the TEC to facilitate efficient heat transfer. This configuration ensures precise temperature control of the DFB laser, enabling fine adjustments to its output wavelength.

Figure 2(c) shows the temperature tuning results of the DFB laser. The temperature of the DFB laser was adjusted from 15°C to 35°C using the temperature control module of the light source, and the output spectrum of the DFB laser was measured with a spectrometer. It can be seen that the central wavelength of the DFB output increases with temperature.

Figure 2(d) presents the variation of the central wavelength of the DFB output with temperature and provides a linear fit. The coefficient of determination is R2=0.99, indicating that the DFB output wavelength varies linearly with temperature. The red line represents the variation in the peak intensity of the DFB output during this process, with a fluctuation of less than 0.07 dBm, demonstrating that the DFB output power remains essentially constant during temperature tuning. This stability originates from the DFB being driven by a constant current source, ensuring its operating current remained essentially constant during this process.

Figure 2(e) tests the response time of the light source to temperature change commands. As shown in the figure, the main control unit adjusted the DFB temperature from 20°C to 25°C and further modulated the temperature up and down in 1°C steps. It can be observed that the time for the temperature to rise from 20°C to 25°C is 15.8 s. During the heating process, the temperature increased at a uniform speed of approximately 0.5°C/s. Due to the asynchronous nature of the DFB heating and temperature measurement, as well as the time required for the TEC to transfer heat, the DFB temperature requires additional time to stabilize after reaching the target temperature. This is reflected in the temperature curve as a temporary overshoot and a fallback phenomenon. However, during microcavity mode locking, only small-scale, high-speed temperature adjustments are typically required. In such cases, the presence of overshoot only affects the beginning or end of the mode scanning curve. Therefore, more attention is generally paid to the response time and temperature stability of the light source during small-scale rapid adjustments.

The top-right inset of Fig. 2(e) shows the response time of the light source for small-scale rapid temperature changes. When the main control unit adjusts the temperature by 0.001°C, the response time of the light source is only 0.2 s. The response duration of the DFB temperature tuning is determined by the combined effect of the hardware circuit’s response time and the software system’s logic processing time. The bottom-right inset shows the fluctuation of the DFB’s actual temperature when the light source is stabilized at a preset temperature. During a 10 s test period, the actual temperature drift of the DFB was less than 0.9 mK, which is 9×10−4°C. These results demonstrate that the light source exhibits fast response speed and high stability during small-scale fine temperature adjustments.

In this study, we design a mode-locked system to perform mode scanning and locking of the microcavity ultrasound probe utilizing a tunable DFB laser. The mode-locked system assists ultrasound detection by automatically selecting the most suitable mode for ultrasound detection through forward and backward wavelength scanning and stabilizing the probe at its optimal working point.

As shown in Fig. 3(a), a packaged polymer microcavity ultrasound probe was used in the experiment. The mode scanning process follows these logical steps: •System initialization: the control unit is activated, powering on each module, then checks the operational status of each module;•Temperature scanning: the system then performs forward (heating) and reverse (cooling) temperature scans. According to the fitted relationship between DFB wavelength and temperature in Fig. 2(d), a linear interpolation method provides a one-to-one correspondence between wavelength and temperature. Meanwhile, the temperature stability of the DFB, as shown in Fig. 2(e), ensures the accuracy and uniformity of the temperature scan.

During the wide temperature scan, thermal nonlinear effects in the microcavity lead to mode broadening. This phenomenon arises from the combined effects of a high Q-factor, small mode volume, miniature cavity size, and temperature-sensitive polymer cavity material. As a result, modes of the probe exhibit bistable behavior, where the high-Q modes broaden or compress during bidirectional scanning.

Thermal broadening essentially reflects the same mode under forward and reverse scans, but it alters the mode’s line shape without affecting its Q-factor. This results in some modes with high Q-factors but flattened resonance slopes, thereby facilitating robust mode locking due to reduced sensitivity to frequency jitter. Figure 3(b) shows the normal mode line shape, while Fig. 3(c) illustrates the changes of the same mode caused by thermal broadening with opposite temperature scanning. In thermally broadened modes, the broadened side exhibits a lower slope, making it easier to lock the probe’s working point onto this side. Therefore, the thermally broadened mode is one of the preferred working modes.

At room temperature, the polymer employed in our study exhibited a positive thermal expansion coefficient and a negative thermo-optic coefficient, with the absolute value of the latter exceeding that of the former. Consequently, temperature variations within the microcavity induced a net negative optical response, leading to blue shifts in WGM spectral lines when the microcavity is heated. Therefore, in our research, thermal broadening emerged during DFB temperature descent. To accommodate different probe materials, the system typically performs bidirectional wavelength scans to identify the optimal mode line shape.

After completing the bidirectional scanning, the system proceeds to mode selection and locking. Figure 3(d) illustrates the entire process, which is designed as follows.•Wide temperature scanning

By adjusting the DFB laser temperature, the system tunes the wavelength over a wide range, enabling a comprehensive characterization of the microcavity modes. This step provides the foundational data for subsequent mode selection. After the scan, the system analyzes the data, storing the transmission data of the microcavity ultrasound probe at each DFB temperature. It also calculates the Q-factors of all modes within the scanning range to select the optimal mode.•Narrow temperature scanning

Due to thermal broadening, the two sides of the same mode exhibit different slopes. In the same mode, the smaller-slope side provides stabler locking, while for different modes, a larger slope indicates a larger Q-factor, thus higher sensitivity. The system ultimately selects the counterpart of the side with the largest slope as the optimal mode.

For example, if thermal broadening causes one side of the mode to narrow (increased slope) and the other side to widen (decreased slope), the system prioritizes the side with the smaller slope as the optimal mode.

The system then adjusts the DFB temperature to the vicinity of the optimal mode and narrows the scanning range. By increasing the sampling density and the tuning speed, the system further pinpoints the exact position of the optimal mode while minimizing potential mode drift during the scanning and selection process. The mode slope k is also recorded as a key reference for subsequent locking. •Mode tuning. After locating the optimal mode, the system sets the mode’s center point as the working point, typically positioned at the midpoint of the mode’s broadened side. Using the linear interpolation relationship between wavelength and temperature, the system calculates the corresponding DFB temperature for this center point. Because of the linear relationship between the DFB laser temperature and its emission wavelength, the DFB temperature is then adjusted stepwise to bring the microcavity ultrasound probe’s working point into the optimal mode.•Mode locking. Once the working point enters the optimal mode, the system records the transmitted light intensity of the microcavity ultrasound probe and corresponding voltage output by the photodetector, setting it as the locking voltage VL.

The system then uses a data acquisition card (DAQ) to record the transmission spectral voltage at fixed intervals as VA, and compares it to the locking voltage. If the voltage error exceeds the allowable range, the system identifies mode drift and initiates the mode drift correction process. The allowable error is typically a fixed proportion of the mode depth. •If error signal error>0:

if VL>VA and k>0, or VL<VA and k<0, the system increases the DFB temperature in 0.001°C steps. As the DFB temperature rises, the DFB wavelength increases, and the working point of the microcavity ultrasound probe shifts upward, correcting the mode drift.

After correction, the system rechecks the actual voltage against the locking voltage. If the error remains above the allowable range, the temperature adjustment process is repeated until the error is within the allowable range. •If error signal error<0:

if VL<VA and k>0, or VL>VA and k<0, the system decreases the DFB temperature in 0.001°C steps. As the DFB temperature decreases, the DFB wavelength decreases, and the working point of the microcavity ultrasound probe shifts downward, correcting the mode drift.

Then the system rechecks again. If the error remains above the allowable range, the temperature adjustment process is repeated until the error is within the allowable range. The mode-locked program operates in a loop until manually terminated. Based on testing, our proposed mode-locked system can operate stably for over 30 min, meeting the measurement requirements for short-term in situ detection.

Using this mode scanning and locking system, we calibrated the performance of the polymer microcavity probe. During the calibration process, the polymer microcavity probe achieved self-locking through the scanning and locking system.

For bandwidth calibration, a 532 nm pulsed laser (Elforlight STOP-10-100-532, pulse width ≤1.8ns) was directed upward onto a 50 nm silver film, generating PA signals. When a single pulse irradiates the silver film, it generates a PA signal with a duration similar to the pulse width. The corresponding frequency-domain bandwidth is much larger than that of the ultrasound sensor. Therefore, the bandwidth of the PA signal measured by the polymer microcavity probe can be approximately regarded as the response bandwidth of the probe^[36]. The PA signal measured by the polymer microcavity probe is shown in Fig. 3(e), and Fig. 3(f) presents the normalized Fourier transform of the PA signal. It can be observed that the −6dB bandwidth of our polymer probe is 42 MHz. This bandwidth is shaped by the cavity material’s acoustic characteristics, including its deformation properties and sound velocity, and the internal reflection of ultrasound waves.

To evaluate the sensitivity of the polymer microcavity ultrasound probe, we designed and conducted a comparative experiment. The experimental setup is shown in Fig. 4(a). Light from a 532 nm pulsed laser was focused onto a 5mm×5mm black tape attached to a glass slide using a lens (Olympus PLN, magnification: 10×) to generate PA signals. The detector was placed approximately 1 cm above the black tape and was coupled to the tape via water. The PA signals collected by the detector were filtered using a band-pass filter (1.2–800 MHz).

First, we measured the PA signal response of a hydrophone (ONDA, HGL-0400). As shown in Fig. 4(b), the peak-to-peak (P-P) value of the PA signal was 10.6 mV, and the signal-to-noise ratio (SNR) was nearly 4.82. Next, we replaced the hydrophone with our mode-locked polymer microcavity ultrasound probe and measured the PA signal under the same conditions. As shown in Fig. 4(c), the P-P value of the PA signal was 1346.9 mV with an SNR of 238.61 and an NEP of 13.5 Pa. For further comparison, we measured the PA signal using a traditional silica microcavity ultrasound probe under the same conditions. As shown in Fig. 4(d), the P-P value of the PA signal was 327.7 mV. Although there are differences in the Q-factors of the WGM microcavity ultrasound probes made of the two different materials, during the measurement process, we selected modes with Q-factors (~10⁶) as similar as possible, and adjusted the laser power so that the output power from these probes was the same, approximately 40 mW. All acquired measurement datasets have undergone identical filters and signal processing procedures. Our polymer microcavity ultrasound probe achieves a US response improvement of two orders of magnitude compared to the hydrophone. Additionally, it exhibits approximately 4.11 times higher sensitivity than the traditional silica microcavity probe. These findings highlight the significant sensitivity advantage of the polymer microcavity ultrasound probe in PA signal detection, indicating its promising potential for broader applications. The electrical system demonstrates excellent long-term stability, while the WGM probe itself maintains its Q-factor and sensitivity without degradation for at least six months. Consequently, the entire system exhibits robust operational stability for a comparable period. The system exhibits broad compatibility with ultrasound-sensitive microcavities of arbitrary materials (e.g.,  silica and polymer) and geometries (e.g., sphere and toroid), since its programmable logic is universally applicable to multi-mode WGM microcavities. By simply replacing the DFB laser wavelength according to the solution’s absorption spectrum, the operational wavelength of any WGM microcavity can be customized, demonstrating a highly versatile and integrated platform.

With our mode scanning and locking system, the microcavity ultrasound probe can operate stably in complex environments and adapt to various application scenarios. A representative case is the detection of crystal violet, a chemical organic compound primarily used as a pH indicator and dye^[37], and a cost-effective antimicrobial agent historically employed in aquaculture^[38,39]. However, CV exhibits documented toxicity and carcinogenicity, necessitating strict regulatory prohibitions against its illicit use in fishery practices^[40,41]. Consequently, developing ultrasensitive detection methods for CV holds substantial public health implications.

We demonstrated the acoustic measurement capability of the device through a PA signal test involving CV concentration. This approach of using a pulsed laser to generate PA signals from opaque liquids is a common method for depicting absorption spectra and measuring concentrations^[42]. With the mode-locked polymer microcavity ultrasound probe, we successfully measured the PA signals of crystal violet at different concentrations, as shown in Fig. 5(a).

Figure 5(b) shows our experimental setup, which uses a 532 nm pulsed laser as the excitation light for PA signals and a 1064 nm DFB laser as the detection light. The 532 nm pulsed light is adjusted by a circular variable attenuator and iris diaphragm, then focused by an objective in the center of the crystal violet solution. The microcavity probe is inserted from above into the test tube containing the CV solution to measure the PA signal. In each measurement, the microcavity probe is locked at the same operating point. Figure 5(c) shows the CV solutions used, with concentrations decreasing by a factor of 10 from 10−3 to 10−7g/mL from left to right. To eliminate background noise, we also measured the PA signal of water. It should be noted that the concentration is determined by first establishing a calibration curve from the PA signals of known concentration solutions, which is then used to inversely derive the concentration of unknown concentration solutions from their measured PA signal intensity. For unknown samples, the solution would first be characterized (e.g.,  using methods such as photoacoustic spectroscopy to confirm the analyte). Figure 5(d) shows the relationship between the peak-to-peak value of the PA signal measured by the polymer microcavity ultrasound probe and the concentration of crystal violet. In the logarithmic coordinate system, the response of the microcavity probe shows a good linear relationship with the solution concentration.

Table 1 compares the performance of our WGM-microcavity-based liquid concentration detection method with existing techniques, using dye solutions (e.g.,  CV and MB) as test targets, where all reported LODs consistently follow the principle of representing the lowest reliably detectable concentration. Our approach achieves comparable or superior performance in truly non-destructive detection, fully in situ operation, and broad applicability to diverse solutions. While UV-Vis and SERS methods offer high sensitivity for pre-sampled solutions, their dependence on extraction or engineered substrates necessitates complex preprocessing, preventing truly noninvasive in situ measurements. In contrast, our microcavity ultrasonic probe attains a detection limit of 10−7g/mL for solutions through direct immersion without sample alteration, setting the highest sensitivity record among reported totally in situ liquid concentration measurements. The optical setup utilizes an ultra-simple configuration with a low-cost single-frequency laser source, enabling cost-effective deployment in resource-limited settings. This enables solution research to have broader applicability and potential for dissemination in developing countries.

In summary, we have developed a novel dual-layer polymer microcavity ultrasound probe (silica inner core and polymer outer layer) optimized for ultrasensitive liquid concentration detection via photoacoustic sensing. This design achieves a remarkable sensitivity of 13.5 Pa and a broad bandwidth of 42 MHz at −6dB, enabling precise quantification of trace analytes in complex liquids. By integrating a temperature-tuned DFB laser and automated mode-locked system, we resolved the instability operating points of traditional WGM sensors, ensuring robust performance in non-laboratory environments. The system’s practicality was validated through CV detection in aqueous solutions, demonstrating a detection limit as low as 10−7g/mL for solution concentration measurements, demonstrating the highest sensitivity reported to date for in situ liquid concentration analysis from PA methods. Unlike HPLC or SERS, our platform eliminates the need for chemical additives or extensive sample pretreatment, offering real-time liquid analysis with laboratory-grade accuracy in field settings.

The entire setup is cost-effective, easy to operate, and robust, providing a new, more stable solution for liquid concentration analysis. This advancement paves the way for WGM microcavities to extend beyond the laboratory, enabling long-term ultrasound detection in challenging external environments with remarkably high sensitivity. Our innovation brings great promise for transforming the field of liquid concentration analysis, providing unprecedented stability and adaptability. We envision microcavity technology becoming an indispensable part of industrial detector applications, offering low-cost and ultra-sensitive liquid analysis methods to more developing regions, revolutionizing the precision and efficiency of liquid detection.