Photoacoustic microscopy (PAM) is a label-free imaging technique that combines pulsed optical excitation and acoustic detection with high spatial resolution, deep penetration, and rich optical contrast [1–4]. For most optical imaging techniques, the illumination light plays a dominant role in imaging performance. Usually, four parameters of the light source determine the PAM imaging quality: high pulse repetition rate (PRR, $>10\mathrm{kHz}$), nanosecond pulse width (PW, 1–5 ns), microjoule pulse energy (PE, 1–10 μJ), and multiple wavelengths ($\lambda$, 200 nm–2 μm) [5–7]. Specifically, kHz PRR enables a fast scanning speed while avoiding the PA signals overlap between two sequential laser pulses [7]; ns PW ensures high excitation efficiency, proper ultrasound frequency, and reasonable axial resolution; μJ PE provides sufficient illumination energy, especially for the reflection-detection modality (water absorption) and thick samples (scattering); abundant wavelengths are crucial to distinguish various substances in a label-free manner, as PAM is a light-absorption-based imaging technique.

Current studies of light sources for PAM systems employ the fiber laser, light emitting diode (LED)/laser diode (LD), or solid-state laser (SSL). Fiber lasers have advantages over spectral broadening through fiber nonlinearities, but their direct output energy is usually low because of the requirement of intracavity gain/loss balance. Thus, extra power amplifiers are required [5,8–11], while the fiber-based long cavity is also not beneficial to obtain short nanosecond PW. LEDs/LDs are small in size, but their PW is usually large (30–100 ns) owing to the slow switching speed of drive current [12,13], while the PE is relatively low. In addition, LDs/LEDs are limited by the bandgap of their semiconductor materials, which restricts their wavelength tunability. For the SSLs, the actively Q-switched (AQS) one is the most commonly employed source for PAM imaging [14–17]. The AQS-SSL demonstrates minimal timing jitter in output, enabling direct utilization of their laser signals as a trigger source for PAM. However, its long cavity (usually $>10\mathrm{cm}$) for inserting the electrical modulator hinders the achievements of narrow PW owing to the long pulse buildup time. Thus, usually the AQS-SSL requires high pump power to help decrease the PW, accompanied by a water-cooling system, resulting in a large machine size. Furthermore, the currently available wavelengths from AQS-SSL are few, like 532 nm and 1064 nm. On the other hand, apart from the aforementioned four laser parameters for PAM, the size and cost of the light source are also conducive to system portability and popularization, respectively. Therefore, we asked ourselves if there is a kind of light source that can not only perfectly cover the four capabilities for high-performance PAM, but also has a tiny size and low price.

As a type of SSL, the passively Q-switched (PQS) SSL is found to be a potential choice for PAM [18–20], which can meet the above six requirements: 1) the saturable absorber (SA) inside the laser cavity is easy to modulate the laser with kHz PRR; 2) the millimeter cavity length ($\sim 2$ to 10 mm) directly corresponds to a few nanosecond PW; 3) a single oscillator directly achieves microjoule PE without the need of power amplifications; 4) abundant wavelengths can be realized by various gain or nonlinear crystals. Furthermore, these crystals can be bonded together, with optical coatings to form the cavity, realizing a very tiny structure [e.g. $\sim 3\mathrm{mm}\times 3\mathrm{mm}\times (2–10)\mathrm{mm}$] and avoiding tedious beam alignments. Meanwhile, the exclusion of electrical modulators also largely decreases the system cost. Based on the above considerations, the PQS-SSL is capable to be an ideal light source for PAM imaging. However, the reason why PQS-SSL is not widely applied in PAM is mainly the stability of laser pulses, including the intensity fluctuation and timing jitter, which is induced by the mechanism of passive Q-switching. These disadvantages hinder the PAM imaging system with precise synchronization, high-speed scanning, and stable imaging quality.

In this paper, we propose a novel technique, termed pulsed pump, to realize the controllable output of laser pulses in PQS-SSL, overcoming the conventional free-running drawbacks. The lasing frequency is locked by the pump beam, which is determined by the electrically modulated radio-frequency (RF) signal applied on the pump beam. By modulating the period, duty cycle, and current of the electrical signal, the PRR (0–20 kHz), PE ($\sim 10\mathrm{\mu J}$), and pulse number (in a pump cycle) of the output laser can be flexibly controlled, enabling PQS-SSL with the ability to be triggered in by RF signals, realizing the same function as other PA imaging lasers. More promisingly, an intense and fast pump minimizes the saturating duration of the absorber, highly enhancing the lasing stability ($\sim 15$ times in power and $\sim 16$ times in PRR). The PW remains constant at $\sim 3\mathrm{ns}$. The entire cavity length is $<6\mathrm{mm}$. We also demonstrate a senior technique, the multi-pulse pump, enabling the same lasing performance but merely with $\sim 39\%$ of the original pump power. Finally, we showcase the PAM imaging with the pulse-pumped PQS-SSL on various samples, including the USAF1951, carbon fiber, and zebrafish, while illustrating the wavelength extension capacity of the source for imaging lipid at $\sim 1.2\mathrm{\mu m}$, as a demonstration of functional imaging. The novel triggerable, millimeter-scale, and cost-effective pulse-pumped PQS-SSL is highly promising for functional PAM, especially for hand-held applications and miniature imaging systems.

Four essential parameters (PRR, PW, PE, and $\lambda$) of the illumination light should be considered for establishing a high-performance PAM system [Fig. 1(a)]. PRR is a key factor affecting imaging speed, which is usually around 10–100 kHz. It enables a real-time B-scan rate for general settings when above 30 kHz, facilitating rapid data acquisition and improving temporal resolution for dynamic imaging applications. A long PW results in poorer axial resolution, while a very short PW increases PA signal (high frequency) attenuation in tissues. Therefore, to balance axial resolution and imaging depth, a PW in the range of 1–5 ns is appropriate, corresponding to the axial resolution of 1.4–6.9 μm. In a PAM imaging system, the required PE after the objective lens is determined by the imaging target depth, sensitivity of ultrasound transducer (UT), and water absorption, typically ranging from nJ to μJ. Considering system loss before the objective, a PE of 1–10 μJ is sufficient for imaging most of the samples. Plentiful wavelengths ensure the coverage of optical absorption peaks of various substances [ultraviolet band: DNA, RNA; visible band: deoxy-hemoglobin (HbR), oxy-hemoglobin (${\mathrm{HbO}}_2$); near-infrared band: lipid, water, and glucose], which is beneficial for functional PA imaging. Typically, LD/LED, SSL, and fiber lasers are designed as excitation sources to meet the above requirements, but it is challenging for all four requirements to be fully met. As the most widely used light sources among them, SSLs are classified into actively Q-switched and passively Q-switched types based on different Q-switching methods. As shown in Fig. 1(a), when choosing between AQS-SSL and PQS-SSL as light sources for practical PAM, four key factors must be considered, including the cost, size, energy, and trigger.

1. 1.Cost—AQS-SSL is generally more expensive than PQS-SSL, primarily owing to the complex drive electronics required by AQS-SSL. In contrast, PQS-SSL modulates laser pulses through the intrinsic properties of SA, eliminating the need for additional components and associated costs.
2. 2.Size—Most electro-optic and acousto-optic Q-switches are relatively large, typically measuring up to 10 cm in length with 1–2.5 cm clear apertures. In contrast, the SAs can be constructed in virtually any size imaginable on demand. Usually, the cavity length of PQS-SSL is roughly within 10 mm, and the lateral dimension depends on the focal spot size, which can be shortened to 1 mm. By integrating advanced crystal bonding and precision coating techniques, the laser cavity can be achieved at the fingertip size.
3. 3.Energy—μJ-level PE is sufficient for practical PAMs, considering the imaging depth, tissue scattering, and water absorption, which can be easily achieved by AQS- and PQS-SSLs.
4. 4.Trigger—A significant drawback of PQS is that there is no control over laser pulses in the free-running system. Its PRR is solely dependent on SAs, resulting in uncontrollable lasing rates and pulse-to-pulse variability/jitter. In contrast, AQS can be triggered by RF signals owing to the intracavity modulators. If this problem of PQS-SSL is solved, it can realize the same function as the AQS ones to synchronize the laser to other instruments.

Based on the comparisons, the main reason for limiting PQS-SSL as the light source for the PAM system is the triggering. Therefore, we propose a pulse-pumped PQS-SSL to solve this problem [Fig. 1(a)]. First, the principle to minimize the timing jitter in pulse-pumped PQS-SSL is illustrated in Fig. 1(b), demonstrating the lasing processes of CW or quasi-continuous wave (QCW) pump, and pulsed pump. Essentially, timing jitter in PQS-SSL is inevitable due to the uncertainty of the SA bleaching threshold, which is not a fixed value but rather a region. Compared with the CW/QCW pump, the pulsed pump provides greater pump power in a cycle, resulting in a higher gain slope in the growth of population inversion, which largely reduces the bleaching duration within this uncertain region. In this way, the timing jitter is minimized. Second, the principle to lock the pump and lasing frequency is shown in Fig. 1(c). An excessively long pump pulse duration leads to the generation of two laser pulses within a single pump pulse, disrupting the PRR consistency. By ensuring that each pump pulse energy produces exactly one laser pulse, we reduce the pump pulse duration as much as possible while increasing the pump power to maintain constant pump energy. In this way, the laser PRR is synchronized with the frequency of the pump beam, realizing an RF-controlled lasing modality for triggering in. Simultaneously, the timing jitter is further minimized by an intense and fast pump.

A typical construction of the pulse-pumped PQS-SSL is demonstrated in Fig. 1(d), together with the schematic diagram of the laser applied for a PAM system. The details are illustrated in Section 7. The photo of a typical pulse-pumped PQS-SSL is demonstrated in Fig. 1(e), constructed by 16-mm cage systems, which is small enough to be hand-held. It is noted that the size can be further smaller by designed holders.

The lasing frequency can be locked by the pump beam by modulating the pump power and duration. Figures 2(a)–2(c) show the output laser under three kinds of pump modulations, including increasing the pump power, increasing the pump duration, and keeping the pump pulse energy constant. Throughout the modulation process, the PRR of the pump train was maintained at 20 kHz. Figure 2(a) shows the output laser variation at pump powers of 11.7, 12.2, 12.7, and 13.2 W. It can be observed that at the pump power of 11.7 W, a single pump pulse does not produce a laser pulse. At 12.2 W, the pump pulse produces a corresponding laser pulse, which appears at the edge of the pump pulse. As the pump power increases to 12.7 W, the produced laser pulse shifts to the middle position of the pump pulse. When the pump power reaches 13.2 W, a single pump pulse produces two laser pulses. During the process of varying pump power, the pump duration was maintained at 20 μs. Figure 2(b) illustrates the output laser at the pump durations of 20, 25, 30, and 35 μs, while maintaining the pump power at 12.2 W. The number of laser pulses generated by each pump pulse rises from one to four as the pump duration increases. When adjusting the pump power to maintain constant pump pulse energy at pump durations of 20, 25, 30, and 35 μs, the behavior of the output laser is shown in Fig. 2(c): one laser pulse is produced by each pump pulse, and all of the laser pulses are at the edge position of the pump beam. The laser pulse generated at the pump edge indicates that no extra pump energy is either stored or wasted, which benefits the lasing stability and thermal dissipation. Meanwhile, an intense and short pump pulse is preferred for further minimizing the timing jitter, when the frequency of the pump and lasing is locked.

The control of laser PRR is also exhibited. Figures 2(d)–2(f) illustrate the modulation process and the laser pulse train at the PRRs of 20, 15, 10, and 5 kHz, with each pump pulse generating one laser pulse. Under a constant pump duration of 20 μs, higher pump power is required to achieve the corresponding laser pulse train when the PRR of the pump train decreases. This is because the increased time between adjacent pump pulses leads to a greater loss of population inversion, resulting in the requirement of higher pump power to reach the stable lasing threshold. In addition, to further improve the PRR for fast imaging applications, potential strategies include optimizing the recovery dynamics of the SA (e.g., using materials with faster carrier relaxation) and enhancing the modulation capabilities of the pump source (e.g., via advanced driver circuits and higher-power pumping schemes).

The lasing stability, PE, PW, and beam quality of the pulsed-pump PQS-SSL are examined, and compared with the conventional CW-pumped one. Figure 3(a) shows a typical laser pulse train of the pulse-pumped PQS-SSL at 20 kHz over 10 ms. The intensity fluctuation of the laser pulses was minimal. The intensity fluctuation of the 20-kHz laser train is 0.97% (s.d./mean) based on 1000 pulses [Fig. 3(b)]. The frequency spectrum [Fig. 3(c)] of the pulse-pumped PQS-SSL exhibits a narrow peak with a 3-dB bandwidth of 0.1 kHz, while it is 2.6 kHz for the CW-pumped one. The small bandwidth of the frequency signal indicates much higher stability of the laser PRR. Meanwhile, the signal-to-noise ratio (SNR) under the pulsed pump is higher than that under the CW pump, reaching up to 42 dB. The lasing stability of pulsed-pump PQS-SSL is also confirmed by the ultralow output power fluctuations, compared with the CW pump, as shown in Fig. 3(d). The output power stability was measured over 2 h. The output power in the pulsed pump is almost a straight line over time, in contrast to the large power fluctuations in the CW pump. Under the pulsed and CW pump, the root mean square (RMS) stability (s.d./mean) is less than 0.18% and 2.75%, respectively, while the peak-to-peak (PTP) stability [$(P_{\max }-P_{\min })/\mathrm{mean}$] is less than 0.97% and 18.7%, respectively. Obviously, the pulsed pump offers greater stability in both output power (enhanced by 15 times based on the RMS stability) and PRR [improved by 16 times based on the time jitter, Visualization 1 (CW pump, 8.0%; pulsed pump, 0.5%)]. As shown in Fig. 3(e), the laser output power increases with the PRR, achieving a maximum power of 209 mW at 20 kHz. The PW maintains a stable range of 2.8–3.1 ns, primarily determined by the round-trip time of the light within the cavity. Simultaneously, the PE remains within 8.2–10.5 μJ, governed by the modulation depth of the SA. A typical pulse profile of the laser at 20 kHz is shown in Fig. 3(f), where no satellite pulses exist. Figure 3(g) displays the excellent beam quality ($M_x^2=1.2$, $M_y^2=1.2$) and the Gaussian-shaped beam profile.

A senior pump technique, multi-pulse pump, is also demonstrated to achieve the same lasing performance, but requiring much less pump power. Figure 4(a) illustrates the multi-pulse pump process, where multiple pump pulses with low peak power ($P_4$) are utilized for pumping instead of a single pump pulse with high peak power ($P_3$). This method enables stable laser pulse output while reducing the dependence on high pump powers. As shown in Fig. 4(b), under 20-kHz pump rate and 20-μs pump duration, stable laser PRRs of 20, 10, 6.7, and 5 kHz are achieved by lowering the pump power, corresponding to 20/1, 20/2, 20/3, and 20/4 kHz, respectively. The achievable discrete PRRs, rather than continuous ones, are attributed to the energy accumulation of pump power. Specifically, if the energy provided by the pump pulses is more than enough or less than required to establish a laser pulse, the pulse train of the output laser will not be stable. Furthermore, the number of accumulated pump pulses to form a signal laser cannot be indefinitely increased, as the pump accumulation time should match the lifetime of the upper energy level of the gain crystal. It means there exists a gain/loss balance between the pump power accumulation and the population inversion stored in the upper energy level of the gain medium. For example, the upper energy level lifetime of Nd:YAG crystal is $\sim 230\mathrm{\mu s}$, which indicates that at a PRR of 20 kHz (corresponding to a pump period of 50 μs), the stable accumulation is recommended to be within five pulses. Figures 4(c) and 4(d) compare the pump power [Fig. 4(c)] required to generate similar laser performance [Fig. 4(d)] under the single-pulse pump and multi-pulse pump. For producing 5- and 10-kHz lasers, the pump power required by multi-pulse pump is significantly lower ($\sim 61\%$ and $\sim 38\%$, respectively) than that by a single-pulse pump. Owing to the increased loss of population inversion during the energy accumulation process, the overall operating efficiency is slightly decreased in the multi-pulse-pumped PQS-SSL. In Fig. 4(d), the output power of the multi-pulse pump is slightly higher than that of the single-pulse pump at the same PRR, which is attributed to the reduction of thermal effects in the laser cavity. The decrease in thermal effects also leads to a reduction in timing jitter.

The pulse-pumped PQS-SSL was applied to a typical PAM system, to examine its feasibility in practical imaging applications. The experimental setup of the whole system is shown in Fig. 5(a) and details are illustrated in Section 7. The lateral resolution of the PAM was measured by scanning the edge of a resolution target with a scanning step of 1 μm. The raw data of the PA intensity were derived by calculating the peak-to-peak intensity of the PA signal, which was subsequently fitted using the edge spread function (ESF). The line spread function (LSF) was obtained by taking the first-order derivative of the ESF. The full width at half-maximum (FWHM) of the LSF was then calculated to determine the lateral resolution of the PAM system, which was to be 5.5 μm [Fig. 5(b)]. It is close to the theoretical value ($\lambda /2\mathrm{NA}$, where $\lambda$ is the laser wavelength) of 5.3 μm. The axial resolution was measured by extracting the envelope of a single PA signal with Hilbert transformation, as shown in Fig. 5(c), yielding an axial resolution of 300 μm. Generally, the theoretical axial resolution is given by the formula $0.88v/B$, where $v$ represents the speed of sound and $B$ denotes the center response frequency (CRF) of the UT. With $v=1.56\mathrm{\mu m}/\mathrm{ns}$ and $B=4.7\mathrm{MHz}$, the system’s theoretical axial resolution was calculated to be 292 μm. Each laser pulse generated a corresponding PA signal, resulting in a PA signal sequence of 20 kHz when the laser excited the same location on the black tape at 20 kHz. As shown in Fig. 5(d), the obtained PA signal sequence remained stable over 10 ms, facilitating high-speed photoacoustic imaging. Figure 5(e) shows the intensity statistics of 1000 acquired PA signals, exhibiting a low intensity fluctuation of 1.4% (s.d./mean). To assess the stability of the PAM system at different laser PRRs, we imaged Elements 1–3 in Group 5 of the 1951 United States Air Force (USAF) resolution target at PRRs of 10, 15, and 20 kHz. The PE after the objective lens was adjusted to 66 nJ at different PRRs. The resulting maximum projection images are shown in Fig. 5(g), indicating a constant imaging quality of the laser with varied PRRs. After evaluating the performance of the PAM system, we proceeded to image bio-samples. The whole body of a zebrafish was scanned [20 kHz and 27 nJ, Fig. 5(f)], and detailed structures were well revealed, including the retinal, yolk sac, trunk, etc. The image of carbon fibers [20 kHz and 66 nJ, Fig. 5(h)] also confirms the excellent laser property for PAM.

Figures 6(b)–6(d) are a demonstration of wavelength extension to $\sim 1.2\mathrm{\mu m}$ in a typical pulse-pumped PQS-SSL, where lipid has strong optical absorption. A 2-mm-thick a-cut ${\mathrm{YVO}}_4$ crystal was inserted into the laser cavity. As shown in Fig. 6(b), the output laser is shifted to 1176 nm based on the $890\text{-}{\mathrm{cm}}^{-1}$ Raman shift of the ${\mathrm{YVO}}_4$ crystal, which is within the 1.2-μm region for lipid. Using the 1176-nm laser from the pulse-pumped PQS-SSL, a piece of beef was demonstrated as the sample for PA imaging, as shown in Fig. 6(c). The obtained PA image of lipid distribution in the beef sample is shown in Fig. 6(d). The scanning step size is 20 μm and the imaging region is $2\mathrm{mm}\times 4\mathrm{mm}$. Obviously, the lipid in the beef is well distinguished from the muscle.

We propose a high-performance laser source for versatile and functional PAM imaging, with the merits of trigger-in ability, millimeter scale, low cost, and abundant wavelengths. For the output laser, its kilohertz repetition rate, nanosecond pulse width, microjoule pulse energy, and UV to NIR spectrum are exactly within the requirements of PAM imaging. A pulsed pump is the key technique we propose in PQS-SSL to realize the control of laser PRR and minimization of timing jitter, enabling the frequency lock and trigger-in ability. Based on this, the pulse-pumped PQS-SSL owns the capacity of synchronization with imaging systems, like other light sources commonly used in PAM. Similar to the QCW operation, pulsed pump technology can significantly mitigate the thermal effect, enabling stable laser operation at room temperature. The all-crystal-based laser configuration contributes to the tiny size and low cost of the source, enabling a hand-held modality for general PAM applications. The designed configuration of gain and nonlinear crystals is capable to realize the desired wavelengths for functional PAM imaging. A multi-pulse pump, as a senior technique, is also demonstrated to release the requirement of high pump power, which can further compress the size of the pump unit and system cost. We showcase the PAM performance with the pulse-pumped PQS-SSL on various samples, including the USAF1951, carbon fiber, zebrafish, and lipid (wavelength extension to $\sim 1.2\mathrm{\mu m}$). It is noted that not only the laser characteristics of this kind of light source perfectly match PAM imaging, but also it has portable and cost-effective properties. All of these features make the pulse-pumped PQS-SSL ideal and feasible for high-quality, functional, and miniature PAM systems.

Laser construction. In Fig. 1(d), a typical laser construction is demonstrated. The laser diode (LD) driver modulated the pump train by RF signals. An 808-nm fiber-coupled LD (core diameter, 200 μm; numerical aperture, 0.22) was applied. After passing the collimating (L1) and focusing lens (L2) with focal lengths of 11 mm, the pulsed pump light was incident into a laser resonator. The laser cavity was composed of the gain crystal, SA, nonlinear crystal, and output coupler (OC). The gain crystal was a Nd:YAG crystal [2 mm, 1% (atomic fraction) ${\mathrm{Nd}}^{3+}$ ions], and the SA was a ${\mathrm{Cr}}^{4+}\text{:}\mathrm{YAG}$ crystal (1.7 mm, 80% initial transmission). To serve as the rear cavity mirror (M1), the surface of the Nd:YAG crystal near L2 was coated with high reflection at 1064 nm and antireflection at 808 nm. To serve as the front cavity mirror (M2), the entrance of the OC was coated with 85% reflection at 1064 nm. The total length of the cavity was $\sim 3.7\mathrm{mm}$. In addition, a nonlinear crystal can be added to the cavity to achieve wavelength extension. The same RF signal was utilized for both triggering in of the laser source and triggering out of the PAM, achieving the system synchronization.

PAM imaging system. In Fig. 5(a), the RF signal generated by a function generator was used to trigger the PQS laser and PAM system. The beam size of the laser was expanded through a telescope comprising lenses L1 and L2 (JCOPTIX, China), with focal lengths of 75 mm and 125 mm, respectively, to match the back-aperture size of an objective lens. The sample was placed at the bottom of a water tank, and the $x$- and $y$-axis scans were achieved by controlling motorized translation stages. The expanded laser, focused by an objective with a numerical aperture (NA) of 0.1, was directed into the sample. Owing to the PA effect, the sample rapidly underwent thermal expansion after absorbing the pulse laser, generating ultrasonic waves. The PA signal was detected using a UT, and then passed through a 46-dB RFA and an 18-MHz LPF. The amplified and filtered PA signal was transmitted to an OSC, which used the RF signal from the function generator as a trigger source to enable data acquisition. The data were subsequently transferred to a PC, which was utilized to synchronize the motorized translation stage scanning, data acquisition, and data processing.

Zebrafish preparation. Adult zebrafish (Danio rerio) were raised in an aquaculture system (Haisheng, China) under a light-dark cycle of 14 h of light and 10 h of darkness at a temperature of 28°C. The adult zebrafish were mated to obtain embryos, which were then maintained in a 0.3× Daniel solution. After 6 days of post-fertilization (dpf), the larvae were transferred to 4% paraformaldehyde (PFA) and stored at 4°C overnight. Subsequently, the larvae were flattened under a coverslip for the next phase of observation under a microscope.