1. Introduction
In recent years, there has been an increased demand for laser sources operating within the mid-infrared (mid-IR) range, covering from 2 to 5 μm. These laser sources are highly valuable in wide application fields, such as molecular spectroscopy, the generation of mid-IR supercontinuum, medical diagnostics, materials processing, and free-space communications [
1,
2,
3]. The mid-IR spectral region, often referred to as the “molecular fingerprint” region, occupies a significant position in molecular spectroscopy due to its distinctive vibrational absorption characteristics across various molecules. Therefore, the demand for ultrafast and coherent mid-IR sources encompassing a wide spectral range remains a critical necessity, particularly for the progression of molecular spectroscopy. Mid-IR bulk lasers based on Cr
2+-doped ZnS and ZnSe, characterized by broad emission cross-sections spanning 2.0 to 3.5 μm and often called the “Ti:sapphire of the mid-IR”, encompass a crucial section of the fingerprint region [
1,
3]. Thus, Cr
2+:ZnSe (Cr:ZnSe) and the Cr
2+:ZnS (Cr:ZnS) crystal with wide emission bands make it an attractive choice for developing ultrashort mid-IR solid-state lasers. Compared to Cr:ZnSe, the Cr:ZnS crystal distinguishes itself by exhibiting enhanced chemical and mechanical stabilities, a reduced thermal lensing parameter dn/dT, superior thermal conductivity, and a higher damage threshold [
3]. These various features of the Cr:ZnS crystal establish it as a recommendable choice for the development of a high-power femtosecond mid-IR solid-state laser.
In recent years, Kerr-lens mode-locked Cr:ZnS laser has successfully generated <29 fs pulses [
4]. This approach, however, demands careful optical alignment and precise intracavity dispersion control for achieving ultrashort pulses. Furthermore, the mode-locked operation using polycrystalline gain media presents difficulties due to non-uniform grain distribution, leading to cavity misalignment power ramp-up [
5]. Consequently, significant endeavors are essential to establish stable Kerr-lens mode-locked mid-IR bulk lasers [
1].
As a different strategy, ultrafast Cr
2+-doped solid-state lasers have been demonstrated using passive mode-locking techniques, incorporating passive optical components such as semiconductor saturable absorber mirrors (SESAMs) [
6,
7], single-walled carbon nanotube saturable absorbers (SWCNT-SAs) [
8,
9], and graphene saturable absorbers (graphene-SAs) [
10,
11,
12]. However, developing SESAMs and SWCNT-SAs for mid-IR bulk lasers involves obvious technical hurdles. These include the demand for complex epitaxial growth methods and the limited accessibility of commercial SWCNT materials with a sufficiently wide diameter enough to enable nonlinear absorption around 2.3 μm.
However, passive optical devices based on graphene, which are applicable to various photonics and optoelectronics devices [
13,
14,
15,
16,
17], offer a range of benefits, such as exceptionally wide bandwidth, an economically efficient production process, and consistent saturable absorption properties when graphene is used as a saturable absorber [
10,
11]. Until now, several femtosecond Cr
2+-doped bulk lasers with graphene saturable absorbers (graphene-SAs) have been demonstrated successfully [
2,
3]. To date, an ultrafast Cr:ZnS laser with graphene-SA generated ultrashort pulses with a duration of 41 fs [
10], and continuous tuning across a broad wavelength spectrum (~300 nm) in the mode-locked operation was also demonstrated [
11].
As shown in Table III in Ref. [
2], a few mid-IR Cr
2+-doped solid-state lasers with saturable absorbers were successful in achieving ultrashort pulses of <100 fs duration. However, to achieve < 100 fs pulses, the amount of intracavity group delay dispersion (GDD) should be precisely controlled [
4,
10]. Furthermore, it is essential to optimize intracavity dispersion to attain a uniform spectral profile across a broad range, a crucial factor in achieving ultrashort pulse durations. However, the fabrication of thick multilayer mirrors with uniform dispersion properties for mid-IR optical components poses a significant challenge [
3].
Table 1 summarize the results of passively mode-locked sub-100 femtosecond Cr:ZnS solid-state lasers, which were updated based on Table III in Ref. [
2]. The most successful results for achieving sub-100 fs pulses from Cr:ZnS lasers utilize Kerr-lens mode-locking techniques. However, only a few groups successfully generated ultrashort pulses using different passive mode-locking techniques, such as SESAMs, SWCNT, and graphene saturable absorbers. The results of this article are marked with bold style at the end of
Table 1.
In this study, we report the simple generation of ultrashort pulses (<100 fs) in the mid-IR spectral range using the enhanced self-phase modulation effect within the Cr:ZnS oscillator. For this purpose, we employed ZnSe as the base plate for developing transmission-type graphene saturable absorbers. ZnSe exhibits a higher nonlinear refractive index (n
2 = 113 ± 12 × 10
−20 m
2/W at 2.3 μm) compared to the ZnS (n
2 = 47.8 ± 5 × 10
−20 m
2/W), which is the host material for the Cr:ZnS laser crystal [
22]. Furthermore, to increase intracavity power further, we decreased the output coupler’s transmission from 15% to 5% in the cavity, as compared to a previously reported similar laser oscillator in Ref. [
11]. Through the enhanced self-phase modulation effect in the oscillator, we successfully demonstrated a sub-100-fs Cr:ZnS laser with graphene-ZnSe saturable absorber, even in the presence of non-uniform round-trip dispersion spectra.
The femtosecond Cr:ZnS laser delivered an average output power of up to 330 mW at a repetition rate of 233 MHz. It produced slightly chirped 99 fs pulses with an 81.8 nm bandwidth centered at 2373 nm. With extracavity dispersion compensation using a combination of 3 mm and 2 mm thick ZnSe plates, the pulse width was compressed from 99 fs to 73 fs, improving the time–bandwidth product from 0.431 to 0.318. Assuming a sech2 pulse shape (0.315), the pulses were almost transform-limited.
2. Passive Optical Switches Based on Graphene-ZnSe and Configuration of Cr:ZnS Laser
A similar fabrication technique detailed in reference [
11] was applied for developing a high-quality graphene saturable absorber. Initially, we synthesized monolayer graphene through chemical vapor deposition on a copper (Cu) foil, followed by prompt cooling. Subsequently, we applied a supportive layer of 5 wt% polymethyl-methacrylate (PMMA) onto the graphene using a spin-coating technique. Following that, the copper layer underwent wet etching with ferric chloride (FeCl
3), which facilitated the transfer of both the graphene and PMMA layer onto 1 inch diameter zinc selenide (ZnSe) substrates. Finally, the PMMA layer was removed using acetone, resulting in the formation of a monolayer graphene saturable absorber on the ZnSe substrate, referred to as graphene-ZnSe SA.
Figure 1a depicts the linear optical characteristics of the graphene-ZnSe SA, and the insets show a photo of the developed graphene-ZnSe SA, with four dots on the ZnSe substrate, indicating the boundaries of the transferred monolayer graphene. The linear transmission curve reveals that the monolayer graphene deposited on a ZnSe substrate exhibits approximately 2% absorption near 2.3 μm [
Appendix A], which is comparable to the theoretical value of πα = 2.3%. However, we have not yet characterized the nonlinear optical properties, such as the resonant response of the saturable absorption, the saturation fluence, and the modulation depth of the graphene-ZnSe SA near 2.3 μm. Nevertheless, we expect a nonlinear response of the passive devices similar to ~2 ps reported in the reference [
11], and comparable modulation depth of ~0.4% and saturation fluence of ~14 μJ/cm
2 measured at near-IR wavelengths [
23].
For passive mode-locking of a polycrystalline Cr:ZnS laser employing graphene-ZnSe saturable absorbers, an astigmatically compensated Z-fold cavity configuration was used, as shown in
Figure 1b. Linearly polarized 20 W Er-doped fiber laser (IPG Photonics) operating at 1550 nm was used as the pumping source. A rectangular, polycrystalline Cr:ZnS crystal, supplied by IPG Photonics, had a dimension of 6 mm (W) × 2 mm (H) × 5 mm (L). The 5 mm long Cr:ZnS crystal possesses a Cr
2+ concentration of around 8.56 × 10
18 cm
−3, which yields a small signal absorption rate of about 94.2% at pump wavelength. After wrapping the crystal with indium foil to ensure efficient heat dissipation from the gain medium, the laser crystal was mounted in a water-cooled copper block whose temperature was stabilized at 15 °C, as shown in the inset of
Figure 1b. The crystal was positioned at the Brewster angle to reduce reflection losses. The cavity waist for the gain medium was formed by two concave mirrors with a radius of curvature (ROC) of −100 mm, M1 and M2. As a focusing optic for the pump beam, a 100 mm plano-convex lens (L) was utilized. The waist size of the focused pump beam was calculated to be approximately 57 μm, while the waist size of the resonator beam varied between 65 μm and 75 μm depending on the distance between M1 and M2. These two beam sizes were similar, which enabled efficient energy transfer from the pump to the resonator beam. Two additional concave mirrors with ROC of −75 mm, CM1 and CM2, formed the second waist (ω
0 ≈ 80–90 μm) in the longer cavity arm for the transmission-type graphene-ZnSe SA to obtain adequate energy fluence for bleaching the absorption of the passive devices. To minimize the insertion losses, the graphene-ZnSe SA was mounted at the Brewster angle near the focal point. At the other end of the resonator, a flat-wedged output coupler (OC) with 5% transmission at the lasing wavelength was installed. The distances between the optical components are as follows: M1 to the OC is approximately 188 mm, M1 to M2 is around 113 mm, M2 to M3 is 228 mm, and M3 to M4 is 114 mm. Consequently, the total round-trip distance covers roughly 643 mm. The design of the laser cavity is compact and occupies a footprint of 30 cm × 20 cm, except for the pump beam line. For extracavity dispersion compensation, 3 mm and 2 mm thick ZnSe plates were installed at the Brewster angle between the OC and the autocorrelator as shown in
Figure 1b. To protect the oscillator from environmental variables like airflow and humidity, we encased Cr:ZnS laser only using acrylic blocks, but did not purge dry nitrogen gas to reduce humidity.
Figure 2a illustrates the wavelength-dependent group delay dispersion (GDD) of different optical components in the laser cavity: the high reflective mirror (M), chirped mirror (CM), output coupler (OC), a unit length (1 mm) of Cr:ZnS crystal, and ZnSe substrate. The theoretical GDD spectra of Cr:ZnS and ZnSe in
Figure 2a were calculated using the Sellmeier equation for the individual materials, and the dispersion spectra for OC, M, and CM were provided by the producer (Layertec GmbH). Although concave mirrors of M1 and M2 in
Figure 1b are characterized as “GDD reflection optimized” by Layertec, they were not specifically optimized for higher-order dispersion.
While the GDD of the 5% OC is almost close to 0 fs
2 in the lasing spectral ranges, the focusing concave mirrors (M1 and M2) exhibit negative dispersion between 2100 nm and 2600 nm. However, these mirrors were not designed to have ideally flat dispersion profiles within the spectral ranges; the “GDD-optimized” mirror (M) introduces an inhomogeneous and relatively strong negative GDD near 2430 nm as shown in
Figure 2a. To mitigate the strong negative dispersion induced by M1 and M2 around 2430 nm, we employed a specially designed chirped mirror (CM) with a dispersion characteristic of −300 (± 300) fs
2 between 2000 and 2700 nm, as depicted in
Figure 2a. With two chirped mirrors (CM1 and CM2), the second focusing region was formed in the longer arm as shown in
Figure 1b. This approach enabled to achieve flattened net-GDD profiles around 2400 nm, depicted in
Figure 2b. In
Figure 2b, the comprehensive round-trip GDD spectra are presented, incorporating parameters like the laser crystal’s thickness, the ZnSe substrate’s thickness for the SA, and the count of GDD-optimized mirrors and chirped mirrors. By employing four concave mirrors (M1, M2, CM1, and CM2), we effectively compensated for the positive dispersion introduced by the Cr:ZnS laser crystal and ZnSe substrate of the SA. As shown in
Figure 2b, the calculated overall intracavity dispersion was approximately −1200 fs
2 at ~ 2370 nm, denoted by the dashed vertical green lines, indicating the center wavelength of the passively mode-locked Cr:ZnS laser using the graphene-ZnSe SA.
However, it is significant to note that the GDD values exhibited for the mirrors in
Figure 2a are based on simulation data under normal incidence (0 degrees). Consequently, due to the folding angle of 15 degrees introduced by the three mirrors (M1, M2, and CM1), the practical cavity round-trip dispersion might change from the approximated values shown in
Figure 2b. In this configuration, soliton mode-locking within the negative dispersion regime was achieved through slight vibrations applied to the output coupler [
24].
3. Femtosecond Cr:ZnS Laser with a Graphene-ZnSe Saturable Absorber
Utilizing a graphene-ZnSe saturable absorber, we successfully demonstrated a passively mode-locked Cr:ZnS laser and studied its output characteristics, as depicted in
Figure 3.
Figure 3a presents the measured laser output powers, with the solid vertical black line indicating the mode-locking threshold achieved at an output power of 104 mW at incident pump power of 2.19 W. By employing 5% output couplers in the oscillator, we achieved output powers of up to 330 mW at 5.9 W pump power. The Cr:ZnS laser exhibited stable mode-locking across the entire power range above the mode-locking threshold and could be maintained for several hours. Additionally, no visible damage to the graphene-ZnSe saturable absorber was observed. While higher output power could be achieved by increasing the pump power, this led to undesirable continuous wave (CW) components and multiple pulsing that disrupted the stable mode-locked operation. The spectrum broadening with increasing output power is shown in
Figure 3b. Each output spectrum was measured using the WaveScan USB (A.P.E GmbH) with a wavelength resolution of ±0.2 nm within the range of 1000 to 2600 nm.
Figure 4 displays the laser spectrum and the intensity autocorrelation trace, both recorded at an output power of approximately 300 mW. The autocorrelation trace was obtained using the PulseCheck USB MIR from A.P.E GmbH. The spectral bandwidth (FWHM) of the pulses measured 81.8 nm at a central wavelength of 2373 nm. The slightly noisy spectrum observed above 2400 nm in
Figure 4a is attributed to intracavity atmospheric absorption [
3]. In
Figure 4b, indicated by a black circle, the concurrently measured pulse width was 99 fs, leading to a time–bandwidth product of 0.431. This value indicates that the generated pulses were slightly chirped.
The intracavity positive dispersion of the Cr:ZnS laser was only compensated with mirrors possessing negative dispersion properties, leading to difficulties in controlling the intracavity dispersion accurately. To achieve Fourier transform-limited pulses, we introduced an extracavity dispersion compensator using ZnSe plates, which provided a positive GDD of 210 fs2 per 1 mm thick ZnSe at 2370 nm. To minimize reflection losses, these plates were positioned at the Brewster angle between the output coupler and the autocorrelator. Additionally, high reflectance (HR) mirrors (~0 fs2 @ 45 degrees) were employed to guide the beam toward the autocorrelator and minimize additional dispersion from external optical components.
As shown in
Figure 4b, with the installation of a 3 mm thick ZnSe plate, the pulse width decreased from 99 fs to 81 fs, yielding a time–bandwidth product of 0.353, which was close to the transform-limited value for a sech
2 pulse (0.315). Adding an extra 2 mm thick ZnSe plate further reduced the pulse duration to 73 fs, resulting in a time–bandwidth product of 0.318. This value was almost identical to the theoretical value for a sech
2 pulse (0.315), as shown in
Figure 5. Compared to other laser cavities with similar round-trip GDD values of around −1000 fs
2 and relatively flat dispersion profiles [
1,
7], the Cr:ZnS laser with graphene-ZnSe successfully generates < 100 fs pulses through the enhanced self-phase modulation effect induced by the ZnSe substrate of the saturable absorber, even in the presence of inhomogeneous dispersion spectra. Therefore, an even shorter pulse width can be achieved by further optimizing the negative dispersion profile to be flat.
The radio-frequency (RF) spectrum of the femtosecond Cr:ZnS laser was also measured in different spans to verify the stable mode-locked operation, as shown in
Figure 6. The first beat note at 233.13 MHz displayed a pedestal peak separation of around 62 dB above carrier (dBc), recorded with a resolution bandwidth of 100 Hz within a 200 kHz span. This high signal-to-noise ratio of RF spectra and wider 1 GHz span measurement (
Figure 6b) clearly indicates a stable and clean CW mode-locked operation without any undesired laser operation, such as Q-switching instability or multiple-pulsing.
4. Discussion
Transmission-type graphene SAs offer three distinct advantages over reflection-type SAs, despite the need for a second focusing regime in the oscillator to ensure adequate energy fluence for the saturable absorber (SA):
Transmitting graphene SAs typically consist of a graphene layer with widespread linear optical absorption properties and a transparent supporting substrate. This design allows the graphene SA to operate across an exceptionally wide spectral range.
Precise control of energy fluence on the graphene SA is achieved by simply adjusting its position within the focusing regime for laser performance optimization.
Changing the substrate material of the transmitting SA enables control over the amount of self-phase modulation effect within the resonator.
Especially noteworthy is the third advantage of the transmitting graphene SA, which assists in the relatively easy achievement of sub-100 fs pulses in the mid-IR spectrum regime. As we explained in the Introduction section, achieving sub-100 fs pulses from polycrystalline Cr:ZnS lasers demands precise control of intracavity dispersion and the establishment of a uniform dispersion profile across a broad spectral range [
4]. However, fabricating mid-IR mirrors with uniform dispersion is not a trivial task due to thick multilayer dielectric coatings [
3]. Therefore, by simply exchanging the previous widely used SA’s baseplate with low nonlinearity for a baseplate with high nonlinearity, it is possible to achieve shorter pulse durations through the enhanced nonlinear self-phase modulation effect.
Until now, CaF
2 substrates with low nonlinearity have been widely employed as the baseplate for fabricating transmitting graphene SAs in the mid-IR spectral regime [
11,
12]. CaF
2 possesses a small nonlinear refractive index (n
2 = 1.09 ± 0.11 × 10
−20 m
2/W), resulting in negligible nonlinear effects within the cavity. The nonlinearity of CaF
2 is approximately two orders of magnitude smaller than that of ZnSe [
22]. However, ZnSe has higher refractive indices (n = 2.43) compared to CaF
2 (n = 1.42) at 2340 nm; ZnSe has a Brewster angle of 67.7°, while CaF
2 has a Brewster angle of 54.9°. Consequently, by increasing the effective focused beam spot size on the graphene-ZnSe SA with a high Brewster’s angle, both the laser mode-locking threshold and the sensitivity of optical alignment would also be increased. However, the enhanced self-phase modulation effects induced by the ZnSe substrate of the graphene SA meaningfully facilitated the relatively easier attainment of ultrashort pulses in the mid-IR range.