3-photon microscopy uses longer wavelength excitation and briefer pulses than 2-photon excitation, necessitating a different laser system, optics coated to maximize transmission at 1300 or 1700 nm, and a compressor to compensate for dispersion in the microscope optics.
Several early studies of 3-photon excitation were performed with custom-built soliton fiber lasers that could supply illumination at ~1700 but not ~1300 nm (Horton et al., 2013; Cheng et al., 2014; Sinefeld et al., 2015; Liu et al., 2019). These early laser systems have been largely superseded by commercially manufactured laser sources that are tunable and supply illumination at 1300 or 1700 nm.
Most lasers suitable for biological 3-photon excitation are two-box systems consisting of a pump laser and an Optical Parametric Amplifier (OPA). Commercial pump-OPA pairs include the Amplitude Systems Mango, Coherent Monaco/Opera-F, Spectra-Physics Spirit-NOPA and Thorlabs Y-Fi-OPA (formerly manufactured by KM labs). In addition, APE manufacture the AVUS, an OPA that can be combined with a suitable pump laser. Specs, including repetition rate, pulse energy and pulse duration, differ across these pump-OPA systems.
Of these laser systems, the most extensively used for 3-photon excitation have been the Spirit-NOPA (Chen et al., 2018; Ouzounov et al., 2019; Yildirim et al., 2019; Chow et al., 2020; Farinella et al., 2020; Liu et al., 2020b; Wang et al., 2020a) and Monaco/Opera-F (Rodriguez et al., 2018; Takasaki et al., 2019; Wang et al., 2019; Takasaki et al., 2020). Published studies used sources with pump power ≤40 W and ≤1 MHz repetition rates. Spirit-NOPA and Monaco/Opera-F systems are now available with 60-75 W pumps, permitting faster repetition rates, ≤4 MHz. The repetition rates of these newer systems are factory configured, but can be changed in the field by the manufacturer. We are not aware of any publications with 60-75W pump, 2-4 MHz repetition rate sources and their long-term stability remains untested.
A suitable one-box source for 3-photon excitation is available. The Class 5 White Dwarf is an Optical Parametric Chirped-Pulse Amplifier (OPCPA) system containing a Coherent Monaco pump laser. The White Dwarf offers greater time-averaged power and higher maximum repetition rates than pump-OPA systems and has been used for 3-photon excitation at 1 MHz (Weisenberger et al., 2019).
A recent development is an adaptive laser source that enables 3-photon excitation at 32 MHz, with a final stage based on the soliton fiber laser (Li et al., 2019). Repetition rate in biological 3-photon imaging is typically limited by the thermal capacities of the tissue and of the final stage of the laser system. Li et al. (2019) used fast modulation to control illumination intensity pixel-wise across the image. Modulation was performed after the pump laser, thereby reducing time-averaged power entering the soliton fiber and the tissue. The result was 30 Hz imaging of hundreds of cortical neurons 750 μm below the brain surface across a 6-700 μm field of view at 512 x 512 pixel resolution using only ~30-40 mW time-averaged power, far below the threshold for tissue damage.
Pockels cells are in common use, but cause substantial dispersion, often an additional 5,000 fs^2. A half-wave plate and polarizing beamsplitter is a viable alternative, particularly with the former in a motorized rotation mount.
Efficient 3-photon excitation requires brief pulses, with excitation per incident photon declining at pulse durations >50 fs at 1300 nm (Takasaki et al., 2019) and at 1700 nm (Horton & Xu, 2015). ~50 fs pulses are available from laser systems suitable for 3-photon excitation, but pulse dispersion in the microscope and sample is almost inevitable, making dispersion compensation a necessity.
At 1300 nm most microscope glasses introduce (positive) Group Velocity Dispersion (GVD). There's often a need to compensate >5,000 fs^2 Group Velocity Dispersion (GVD), more if using a Pockels cell. Compensation is generally performed with a one or two glass prism compressor (Yildirim et al., 2019; Takasaki et al., 2019).
At 1700 nm, GVD is near zero or anomalous (negative) for many glasses. As a result, anomalous GVD is dominant in many microscopes and glass prism compressors are ineffective compensators. The GVD of silicon remains positive at 1700 nm and a silicon wafer at the Brewster angle has proven a simple and effective solution to correct for pulse dispersion at 1700 nm (Horton & Xu, 2015).
Options include a coated achromat (e.g. Thorlabs AC254-030-C-ML; Horton et al., 2013) or a telecentric scan lens (e.g. Thorlabs SL50-3P; Takasaki et al., 2019; Takasaki et al., 2020) with the telecentric lens likely offering better performance at high scan angles.
A Plössl tube lens (paired AC254-400-C, Thorlabs) is sufficient, but a larger field of view may be accessible with a dedicated tube lens. For microscope designs with widefield fluorescence capability, the tube lens needs to be corrected across the visible spectrum as well as at the 3P illumination wavelength.
Several commercial objectives have been used for 3-photon excitation.
The most commonly used objective is the Olympus x25 multiphoton objective, NA 1.1, WD 2 mm (XLPLN25XWMP2; Ouzounov et al., 2017; Tao et al., 2017; Rodriguez et al., 2018; Li et al., 2019; Liu et al., 2019; Ouzounov et al., 2019; Takasaki et al., 2019; Yildirim et al., 2019; Takasaki et al., 2020; Chow et al., 2020; Liu et al., 2020b; Wang et al., 2020a) which has a correction collar and 70-75% transmission at 1300 nm (Yildirim et al., 2019; Takasaki et al., 2020).
The Olympus x25 Scaleview objective, NA 1.0, WD 4 mm (XLPLN25XSVMP2) offers similar performance and a longer working distance.
The Nikon 16x CFI LWD Plan Fluorite Objective, NA 0.80, WD 3 mm (Weisenberger et al., 2019; Takasaki et al., 2020) has a steep tip profile and long working distance. For many applications, the resulting improved access more than compensates for the relatively low ~50% transmission at 1300 nm (Takasaki et al., 2020) and lack of a correction collar.
Other objectives have been used with 3-photon excitation, such as the Olympus x40/0.8 long working distance water immersion objective (LUMPLFLN40XW; Liu et al., 2020b).
Primary dichroic mirror
Several Semrock and Chroma mirrors have been used as primary dichroic, all likely offering similar performance:
Semrock FF705-Di01(Ouzounov et al., 2017; Wang et al., 2018b; Li et al., 2019; Ouzounov et al., 2019; Yildirim et al., 2019; Chow et al., 2020).
Semrock FF735-Di02 (Takasaki et al., 2019; Takasaki et al., 2020).
Chroma t770spxrxt (Farinella et al., 2020).
Semrock FF875-Di01 (Sinefeld et al., 2015).
APE PulseCheck is suitable for autocorrelation of the free-space beam. The PulseCheck with IR short pulse crystal has been used to estimate pulse widths with ~50 fs pulses at 1300 nm (Takasaki et al., 2019).