Photoacoustic (PA) imaging has gradually become a powerful tool in the biomedical imaging field [1,2]. In this hybrid imaging modality, a pulsed laser is usually employed to irradiate a sample. The photon energy will be absorbed and converted into heat, which will induce a transient temperature rise and local initial pressure change. Finally, acoustic waves, referred to as PA signals [1], propagate out from the sample and are detected by an ultrasound transducer. Therefore, the conventional PA technique requires physical contact of the sample with the transducer itself or acoustic coupling media to access PA signals, which, otherwise, attenuate rapidly in air. This requirement of physical contact of the sample is less ideal in some aspects for PA imaging applications. Infection is likely to be a concern due to the contact operation, especially for applications such as ophthalmic imaging, wound healing inspection, burn diagnostics, and brain imaging [3]. Meanwhile, the requirement of acoustic coupling also restricts the development of PA endoscopy (PAE). Due to the finite size and opaque nature of piezoelectric transducers commonly used in PAE, the design and fabrication of a PAE imaging catheter are complicated, which eventually deteriorates the sensitivity and field of view of the PAE system [3,4]. These issues motivate further development of non-contact PA techniques.

PA remote sensing (PARS) microscopy, as a non-contact imaging approach, was first introduced by Haji Reza et al. in 2017 [5]. Different from conventional PA imaging, an interrogation light beam is used in this technique to detect the local initial pressure change following pulsed light excitation. Briefly, the initial pressure change induces refractive index modulation via an elasto-optic effect inside the absorber, which changes the reflectivity of the backreflected interrogation light beam. Therefore, PA signals can be detected, and the PA signal amplitude is proportional to the optical absorption of the excitation wavelength [5]. Compared with other non-contact methods, PARS acquires the reflectivity changes of the interrogation light beam, which is phase independent. Therefore, PARS is insensitive to phase noise and artifacts, which can be caused by the interrogation light source, the absorber, and the medium for acoustic propagation [3,6]. Currently, PARS microscopy has been successfully employed in pathology assessment, intraoperative histology, and ophthalmic imaging, to name a few [7–17].

Until now, all PARS microscopy systems have been constructed mainly based on free-space light, which imposes several limitations. First, the imaging head is overall bulky, making it inaccessible to internal tissues/organs such as the oral cavity and gastrointestinal tract. Secondly, systems based on free-space light have relatively high requirements for construction, e.g., demanding alignment of the optical path especially for confocal alignment of the excitation light beam and interrogation light beam. The optical path of such systems is also complicated, which is not desired for further development (e.g., integrating another imaging modality [14]) and is not conducive for maintenance. These limitations restrict widespread applications of PARS microscopy. Efforts have been made to simplify the PARS microscopy system by employing fiber-optic components such as an optical circulator [18]. However, the scan head is still based on free-space light [18], including components such as dichroic mirrors, scanning mirrors, etc., which still suffers from the limitations mentioned above. Therefore, it needs further development to build a fully fiber-optic PARS microscopy system and to miniaturize the imaging head.

In this Letter, we report an elegant approach to realize a miniature probe (capable of PARS microscopy) and a fiber-optic PARS microscopy system. Specifically, the imaging head (i.e., the miniature probe for PARS microscopy), mainly consists of a single-mode fiber (SMF) and a gradient-index (GRIN) lens, is highly miniaturized, and has a diameter of only ${\sim}{3.0}\;{\rm{mm}}$. In addition, the fiber-optic PARS microscopy system is implemented mainly using an optical circulator as well as a wavelength division multiplexer (WDM). The miniature probe based on fiber-optic PARS microscopy has several advantages including: (i) facilitating scanning of the probe, which is convenient for in vivo imaging applications; (ii) easing the difficulty to build a PARS microscopy system because free-space confocal alignment can be circumvented; (iii) paving the way for further development of PARS microscopy to realize PAE and handheld operations. To show the imaging capability of the miniature probe, imaging of a zebrafish larva in vivo and lithium (Li) metal batteries is demonstrated. Compared with existing PARS microscopy, to our knowledge, our work reports the first miniature imaging probe and the first fiber-optic PARS microscopy system. Our work opens up new opportunities for PARS microscopy applications.

The experiment setup is shown in Fig. 1. A pulsed laser (AO-S-532, cnilaser, China) with 1 kHz repetition rate was used to provide the light source for PA excitation. Note that the laser provides 532 and 1064 nm laser pulses, and 1064 nm was used here. The excitation light beam emitted from the laser head was split into two paths using a 10:90 beam splitter. The reflected light beam with 10% power was detected by a photodetector (PD1 in Fig. 1) (DET10A, Thorlabs) to provide trigger signals. The transmitted light beam with 90% power passed through a beam expanding and shaping module (consisting of L1, L2, and iris in Fig. 1) and was then coupled into port $a$ of a WDM (WDM-1X2-1064/1310-0-A40, Shconnet). On the other hand, the interrogation light beam was provided by a 1310 nm continuous laser (IPSDM1302C, INPHENIX) with bandwidth of 45 nm to detect PA signals. This light was delivered into port 1 of an optical circulator (CIR-1310-3-P-900-1-FA, Shconnet), and was guided to port 2 of the optical circulator and then into port $b$ of the WDM. So far, both the excitation and interrogation light beams were combined into port $c$ of the WDM. For the miniature probe, it mainly consists of a SMF extended from port $c$ of the WDM and a GRIN lens (GRIN2313A, Thorlabs). Note that the end-face of the SMF was cut with a tilt angle of 8° to eliminate the influence of the reflected light from the end-face of the SMF. The GRIN lens with a tilt angle of 8° is elaborated in Section 1 of Supplement 1. Both the excitation and interrogation light beams emitted from the SMF were focused by the GRIN lens on a sample. PA signals were excited by the excitation light beam and detected by the interrogation light beam. For PA signal detection, the interrogation beam was returned to the WDM (port $c$ and then port $b$) and then the circulator (port 2 and then port 3), which was subsequently recorded by another photodetector (PD2 in Fig. 1) (1811-FS, New Focus) and amplified by a preamplifier (5073PR, Olympus). Finally, the PA signals were digitized by a data acquisition card (CSE1422, GaGe) and then stored in a personal computer for further image formation. In our imaging platform, the miniature probe, as a scan head, was mounted in a two-dimensional servo motorized stage (M-404, Physik Instrumente [PI], Karlsruhe, Germany) for scanning.

 

Fig. 1. Schematic of the imaging system and the miniature probe. BS, beam splitter; NDF, neutral density filter; PD, photodetector; Amp, preamplifier; PC, personal computer; DAC, data acquisition card; L, lens. Inset (upper-right): picture of the miniature probe, showing its diameter of ${\sim}3.0\; {\rm mm}$.

Download Full Size | PDF

The focused beam spot of the excitation light was imaged by a beam profiler (LBP2-HR-VIS2, Newport) and a ${{10}} \times$ beam expander. As shown in Fig. 2(a), the focused beam spot size [full width at half maximum (FWHM)] was measured as ${\sim}{{6}}\;\unicode{x00B5}{\rm m}$. Lateral resolution was measured to be ${\rm{2 {-} 5}}\;\unicode{x00B5}{\rm m}$ by scanning the sharp edge of a United States Air Force resolution test chart. The results of measured lateral resolution are shown in Section 2 of Supplement 1. Note that the wide range of the measured lateral resolution was obtained by scanning the miniature probe with different orientations of the probe. The inconsistency between the measured lateral resolution and the focused beam spot size may be explained as follows. Lateral resolution of the PARS microscopy system is determined by the overlapped region between the excitation and interrogation light beams. First, due to the chromatic aberration of the GRIN lens together with the tilt angle at both the SMF (end-face) and the GRIN lens (as shown in Fig. 1), confocal alignment of the excitation and interrogation light beams in the lateral direction (and the axial direction) cannot be perfectly achieved. The beam profiles of the light with two different wavelengths are shown and described in Section 3 of Supplement 1. This may explain why the measured lateral resolution is smaller than the focused beam spot size. Second, the overlapped region between the excitation and interrogation light beams is likely not a circle. This may account for the variation of the measured lateral resolution when scanning the miniature probe with different orientations. As for measuring axial resolution, a method similar to our previous study was used [19]. A 6 µm carbon fiber was placed at different axial positions, and a one-dimensional (1D) profile of PA signal amplitudes as a function of depth was plotted. As shown in Fig. 2(b), axial resolution is determined by the FWHM of the 1D profile, which is ${\sim}{{123}}\;\unicode{x00B5}{\rm m}$. The discrepancy between the measured axial resolution and the calculated depth of focus of the excitation light beam is described in Section 4 of Supplement 1. Figure 2(c) is a time-domain PA signal of PARS microscopy by imaging the 6 µm carbon fiber, and Fig. 2(d) shows the spectrum of Fig. 2(c). As can be seen in Fig. 2(d), the bandwidth at ${-}{{6}}\;{\rm{dB}}$ of the time-domain signal is below 4.3 MHz. The possible reason for the narrow bandwidth in Fig. 2(d) is elaborated in Section 5 of Supplement 1. Therefore, in the following experiments, a digital low-pass filter was applied to enhance the image quality. To show the imaging capability of the miniature probe, randomly distributed carbon fibers were imaged. The excitation light energy of ${{16}}\;{\rm{nJ/pulse}}$ was used. A thin layer of water was applied to cover the carbon fibers to improve the stability of the PA signals. The reason to add the thin water layer [Fig. 2(e)] is detailed in Section 6 of Supplement 1. Note that water did not contact the miniature probe. As shown in Fig. 2(e), the fine features of the carbon fibers can be clearly revealed, showing the high resolution and satisfactory sensitivity of our miniature probe. We found that the sensitivity of our miniature probe is roughly comparable to that of our previous work based on free-space PARS microscopy [19].

 

Fig. 2. (a) Focused beam spot of the excitation light measured by a beam profiler. (b) 1D profile of PA signal amplitudes as a function of depth by imaging a 6 µm carbon fiber placed at different axial positions. (c) Time-domain PA signal by imaging a 6 µm carbon fiber. (d) Spectrum of (c). (e) PA image of randomly distributed carbon fibers. Scale bar: 100 µm.

Download Full Size | PDF

Figure 3(a) shows the PA image of bodhi leaf veins stained with black ink. The excitation light energy of ${{110}}\;{\rm{nJ/pulse}}$ was used. As can be seen, leaf networks are clearly visualized. It is worth noting that a large field of view of ${\sim}{8.2}\;{\rm{mm}} \times {8.2}\;{\rm{mm}}$ was scanned by our miniature probe with a scanning step size of 8 µm. The capability of scanning the miniature probe, rather than the sample, would greatly facilitate in vivo imaging applications because the animals can be kept stationary. To further demonstrate in vivo imaging application by our miniature probe, non-contact PA microscopy (PAM) imaging of pigments in a three-day-old post-fertilization zebrafish larva was conducted. More information about PAM imaging of pigments in the zebrafish larva is described in Section 7 of Supplement 1. During image acquisition, the zebrafish larva was fixed using ultrasound gel smeared in the bottom of a homemade sample holder. Nutrient solution with 0.03% Tricaine (Sigma) was used to anesthetize the zebrafish larva and keep it alive. Figure 3(b) shows the images acquired by an optical microscope and our imaging system. For the PA image, the excitation light energy of ${{340}}\;{\rm{nJ/pulse}}$ was used. By comparing the two images in Fig. 3(b), excellent agreement of the pigments in the two images (e.g., in the brain and yolk sac regions) can be observed.

 

Fig. 3. Phantom and in vivo imaging by the miniature probe. (a) PA image of leaf veins. Scale bar: 1 mm. (b) Bright-field image (left) and PA image (right) of the zebrafish larva. Scale bar: 500 µm.

Download Full Size | PDF

Figure 4 demonstrates in vivo PARS images of pigment distribution in a three-day-old post-fertilization zebrafish larva. During the imaging procedure, the zebrafish larva was fixed using ultrasound gel smeared in the bottom of homemade sample holder. Nutrient solution containing 0.03% Tricaine (Sigma) was used to anesthetize zebrafish larva and keep it alive.

 

Fig. 4. Li metal battery imaging by the miniature probe. (a) Schematic of the cross-sectional sidewall surface of the Li/Li cells. The dashed box indicates the imaged region. (b) PA image of the Li/Li cell without charging. (c) PA image of the Li/Li cell charged under current density of ${{1}}\;{\rm{mA/c}}{{\rm{m}}^2}$ for 15 h. Scale bar: 200 µm.

Download Full Size | PDF

Our miniature probe was also used to image Li metal batteries. It is known that Li ion batteries play a vital role in present-day technological applications and have been widely used in electric vehicles, mobile phones, and other portable devices. To improve the energy density of Li ion batteries, Li metal batteries have been studied in recent years. More information about PAM imaging of Li metal batteries is described in Section 8 of Supplement 1. The unique characteristic of the non-contact operation of our miniature probe for PARS microscopy may facilitate PAM imaging of Li metal batteries. Therefore, it would be of interest to also image Li metal batteries using the miniature probe. Similar to our previous work [20], the cross-sectional sidewall surface of Li/Li liquid electrolyte symmetric cells [Fig. 4(a)] was imaged by our miniature probe. The cathode and anode of the Li/Li cell were all fabricated using Li metal foils with thickness of 240 µm. The liquid electrolyte layer was made using a 2-mm-thick glass fiber separator (GFS) soaked in electrolyte solution and was sandwiched between the two Li metal films of the Li/Li cell. Two Li/Li cells were prepared. One was without charging, and the other was charged under current density of ${{1}}\;{\rm{mA/c}}{{\rm{m}}^2}$ for 15 h. Before imaging, each Li/Li cell was sealed in a plastic bag. More details about the sample preparation can be found in [20]. Similarly, for image acquisition, the whole sample did not contact the miniature probe.

The cross-sectional sidewall surface of the two Li/Li cells around the region of a Li metal electrode was imaged, as illustrated in Fig. 4(a). The excitation light energy of ${\sim}{{100}}\;{\rm{nJ/pulse}}$ was used. Strong PA signals acquired by PARS microscopy using a 1064 nm pulsed laser for excitation were experimentally verified. As can be seen, for the Li/Li cell without charging, no obvious Li protrusions inside the GFS are observed [Fig. 4(b)]. In contrast, for the Li/Li cell with charging, apparent Li protrusions toward the GFS due to Li deposition can be easily identified [Fig. 4(c)]. In both Figs. 4(b) and 4(c), the thickness of the Li metal film visualized by the miniature probe is consistent with the actual thickness. Inhomogeneous deposition (different thickness of Li protrusions toward the GFS along the Li/GFS boundary) is observed in Fig. 4(c), which is consistent with our previous study [20].

Different excitation light energy used in imaging demonstration is explained in Section 9 of Supplement 1. Interrogation light power used in imaging demonstration is described in Section 10 of Supplement 1. Laser damage of the fiber-optic PARS microscopy system is described in Section 11 of Supplement 1. The imaging speed is described in Section 12 of Supplement 1.

In summary, we demonstrated a miniature probe for non-contact PAM based on PARS microscopy. Since the system adopted fiber-optic components (mainly the WDM and optical circulator), the design and implementation of the PARS microscopy system can be greatly simplified. Specifically, compared with existing PARS microscopy systems, in this work, the imaging head is highly miniaturized, and the system implementation becomes easier. With further development, our work has potential to conduct PARS microscopy for PAE and handheld imaging applications. Currently, the excitation light at 1064 nm is designed in our miniature probe. This is because the corresponding fiber-optic components are commercially available and cost effective, which is benefited by the mature technology of optical fiber telecommunications, and no custom-made parts are needed. In the future, the excitation light at the visible wavelength (e.g., 532 nm, the most commonly used wavelength in PAM) can be designed in the miniature probe for non-contact PAM, which would facilitate vascular imaging applications. In this case, some custom-made fiber-optic components are required. Although there is still some free-space light in our current system (i.e., the region inside the box named “beam expanding and shaping subsystem” in Fig. 1), the laser head that can deliver the excitation light directly through an optical fiber can be adopted in the future, which further simplifies the implementation of the imaging system. We demonstrated the feasibility and potential of PARS microscopy for imaging liquid-based Li/Li symmetric cells. Taking advantage of the non-contact operation of our miniature probe, PAM imaging of solid-state batteries (without liquid electrolyte for acoustic coupling) is technically feasible and worth further investigation.