Ultrafast mode-locking in highly stacked Ti₃C₂T_x MXenes for 1.9-μm infrared femtosecond pulsed lasers

1 Introduction

Since the discovery of graphene [1], 2D materials have attracted great research attention due to their unusual outstanding physical properties [2], [3], [4], [5], [6], [7]. Particularly in recent years, photonic technologies using 2D materials have been among the most important nanophotonic issues, which include exciton-trion-based photoluminescence [8], [9], single photon emission [10], [11], exotic nonlinear optics [12], [13], [14], and plasmonics [15], [16], [17]. In these applications, obtaining and maintaining well-defined 2D structures are essential for their advanced photonic operation, which have required complicated and/or sophisticated processes compared to that of their bulk counterparts. Mechanical exfoliation has been an efficient way to provide high-quality 2D materials from van der Waals layered materials through the physical expansion of interlayer spacing using sticky tapes [18]. However, this method is not suitable for mass production of 2D materials. On the contrary, chemical exfoliation can rather easily produce a large amount of 2D materials from their parent layered materials using intercalation and/or chemical surface modification of the constituent layers, but this method likely deteriorates the structures of obtained 2D materials and needs complicated chemical processes [19], [20]. Generally, 2D materials are prone to surface oxidation under air and/or aqueous environments and to avoid this, they are often coated by protective films or mixed with polymers.

2D transition metal carbides called MXenes are new emerging 2D materials [21], [22] which possess superb physical properties in various fields such as electrochemical capacitance [23], [24], chemical catalysis [25], composite reinforcement [26], electromagnetic interference shielding [27], light-to-heat conversion for solar harvesting [28], quantum dots [29], and photothermal therapy [30] and detection [31], as well as including energy-efficient water purification [32] which was first addressed by graphene technologies [33], [34].

Despite numerous studies on 2D materials for saturable absorption performance in laser optics [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], metallic 2D materials had not been investigated until our recent study because metallic conductors are generally known as poor or incompetent saturable absorbers (SAs). The astounding saturable absorption and mode-locking performance of metallic 2D transition-metal carbonitrides were first observed in our study in which near-infrared mode-locked femtosecond lasers were successfully generated using 2D transition-metal carbonitride saturable absorbers (SAs) [52], opening the door to MXene-based nonlinear optics [52], [53], [54], [55], [56], [57]. Especially, we witnessed its well-operating performance even with a large degree of layer stacking, suggesting the need for in-depth researches on this intriguingly phenomenon [52].

According to the variation of chemical elements, atomic structures, and termination species, MXenes can possess hundreds of different 2D material members. Among such a wide range of choices, titanium carbide (Ti₃C₂T_x) has been the most popular and predominantly employed in practical applications, overwhelming the use of other MXene materials. Based on our work described above, Jiang et al. showed the promising potential of mono- and/or few-layer Ti₃C₂T_x SAs in pulsed laser technologies [54], but which still required sophisticated processes for delamination and dispersion of Ti₃C₂T_x as usual in 2D material application.

On the other hand, concerning the laser spectral range, long-wavelength laser pulses beyond the 1500-nm telecommunication near-infrared regime have recently become of great interest due to their critical role in medical, military, and security technologies. However, many 2D materials are not applicable to this important field due to their inappropriate bandgap sizes or inferior material properties. Particularly, fiber laser systems in the spectral range of 1800–2000 nm have attracted great attention due to its prominent merits as described next. As water has a high absorption rate at 1950 nm, when a 1900-nm laser is irradiated onto biological tissues, ultrafine heating can be attained allowing exquisite tissue cutting in medical applications, compared to 1000 and/or 1500-nm lasers. Furthermore, lasers of this spectral range can suppress bleeding through coagulation in surgical applications, indicating its superb potential in medical applications. In the aspect of micromachining, most of plastics have remarkable absorption around 2000 nm and thus precise and efficient mechanical processing such as cutting, welding, and marking can be achieved using 1900-nm lasers.

This study is aimed to investigate whether the superb saturable absorption performance of Ti₃C₂T_x MXenes could be maintained against significant layer stacking and if it is attained, unveil the origin of this phenomenon. The second goal of this research is to explore whether their spectral range for mode-locking could be extended to the 1900 nm-infrared region. These two factors are both crucial for the development of advanced practical photonic devices based on Ti₃C₂T_x MXenes.

First, we theoretically showed that the electronic and optical properties of Ti₃C₂T_x MXenes are almost unaffected by layer stacking, particularly around the Fermi energy level, due to the intrinsic presence of their unique surface termination, in stark contrast to what observed in other 2D materials such as graphene and MoS₂. This fact was experimentally validated further by showing that hundreds-of-layer stacked Ti₃C₂T_x SAs, which are definitely beyond the regime of mono- and/or few-layer 2D materials, can be excellent optical modulators with a large modulation depth and can produce ultrafast pulsed lasers with a duration as short as 897 fs in a fiber cavity, serving as superb mode-lockers in 1914-nm infrared laser applications. These results suggest that highly stacked Ti₃C₂T_x could widely serve as material platforms for developing cost-effective, high-performance nanophotonic devices operating at long-wavelength conditions, which include 3000-nm deep infrared or terahertz laser systems.

2 Theoretical electronic and optical characterization

Density functional theory (DFT) calculations were performed for in-depth study of the layer stacking effect on the electronic and optical properties of Ti₃C₂T_x MXene, graphene, and MoS₂ 2D materials as implemented in the Atomistic Toolkit package [58], [59]. Specifically, the electronic structures of these three 2D materials were comparatively investigated for their monolayer and bulk-thick forms (Figure 1). The density mesh cutoff energy was set as 150 hartree and the Monkhorst-Pack grid of k-sampling for the electronic structure calculations of bulk-thick (or monolayer) forms were 6 × 6 × 4 (6 × 6 × 1), 7 × 7 × 5 (7 × 7 × 1), and 12 × 12 × 3 (12 × 12 × 1) for Ti₃C₂T_x MXene, graphene, and MoS₂, respectively. The optimized dimensions of their hexagonal systems were 6.24 × 6.24 × 9.94 Å, 3.25 × 3.25 × 12.89 Å, and 4.96 × 4.96 × 6.87 Å, respectively. To obtain optical absorption spectra, the susceptibility tensor was first calculated using the Kubo-Greenwood formula shown below [60]:

where πnmi is the i-component of the dipole matrix element between states m and n, f is the Fermi function, Γ is the broadening parameter, and V is the volume element. Then, the frequency-dependent dielectric constants are calculated using the relation of ε(ω) = 1 + χ(ω). From the imaginary part of dielectric constants, we finally obtained the optical absorption spectra of the systems.

Figure 1:

The atomic geometries and electronic band structures of (top) monolayer and (bottom) bulk-stacked forms of (a, d) Ti₃C₂T_x MXene, (b, e) graphene, and (c, f) MoS₂. The magenta dotted circles marked in (e) indicate the electronic dispersion relation around the Fermi energy level, which changes from (top) a linear to (bottom) parabolic regime upon layer stacking. The magenta arrows marked in (f) indicate electronic transition which changes from (top) a direct and (bottom) indirect mode upon layer stacking, along with distinct quantitative variation.

The Dirac linear dispersion observed in graphene, which can induce massless fermions, has significantly changed to a normal parabolic dispersion upon layer stacking along with the appearance of additional electronic bands. Meanwhile, the electronic band gap of monolayer MoS₂ distinctly decreased upon layer stacking and changed its mode from a direct to indirect transition. as generally seen in 2D transition-metal dichalcogenide (TMDC) semiconductors. In contrast to such considerable electronic variations of graphene and MoS₂, we found that the electronic band structure of bulk-thick Ti₃C₂T_x remained almost the same as that of monolayer Ti₃C₂T_x, particularly at the important region around the Fermi energy level.

Ti₃C₂T_x is composed of a main structural body, which have five atomic layers of Ti–C–Ti–C–Ti, and surface terminations that are located on the top and bottom Ti layers (Figure 2a). We suppose that the above intriguing result stemmed from its intrinsically existing surface termination, serving as a buffer-like separator between Ti₃C₂T_x bodies. The electronic density map measured along the top Ti layer of Ti₃C₂T_x indicated that electron distribution around the Ti atoms was almost unaffected by layer stacking (Figure 2b).

Figure 2:

The electron density distribution, optical properties, and height profiles of Ti₃C₂T_x MXenes.

(a) The layer-wise description of the Ti₃C₂T_x MXene structure. (b) The calculated electron density maps (measured along the top Ti layer) and (c) the optical absorption spectra of monolayer and highly stacked Ti₃C₂T_x MXenes. (d) The SEM image of highly stacked Ti₃C₂T_x flakes, clearly exhibiting the appearance of bulk-thick layer stacking when they are located vertically. (e) The height profiles of Ti₃C₂T_x flakes measured using AFM.

Besides the fact that the presence of surface termination can effectively protect the characteristics of MXene monolayer from the influence of other adjacent ones, it is worth noting that surface termination engineering can also be a powerful means to regulate the electronic and/or optical properties of the MXene monolayer itself [21], [22], [61].

The stacking energy of graphene was calculated as 6.898 meV Å⁻², while that of Ti₃C₂T_x MXene was 56.691 meV Å⁻², which was 8.2 times magnitude larger than that of graphene. Such a large stacking energy of Ti₃C₂T_x is attributed to hydrogen bonds formed between the surface terminations of adjacent MXene layers, whereas the stacking of graphene is governed by van der Waals interaction. It should be noted that we assumed a rather high density of hydrogen bonds between MXene layers to examine the interlayer bonding effect at a severe condition. Based on the number of hydrogen bonds involved in the layer stacking of the Ti₃C₂T_x MXenes, the average hydrogen-bond energy was calculated as 46.148 kJ/mol, which is reasonable considering the well-known hydrogen-bond energies of F–H⋯F (161.5 kJ mol⁻¹) and O–H⋯O (21 kJ mol⁻¹), validating our computation method. Meanwhile, the van der Waals bond energy of the stacked graphene system was evaluated as small as 1.773 kJ/mol, showing an excellent agreement with the reported values of 1.112 and 1.717 kJ mol⁻¹ for AA and AB stacking of graphene, respectively. It is noteworthy that surface termination can offer a good interlayer mechanical stability of Ti₃C₂T_x MXenes.

In order to further support the robust layer-independent photonic operation of Ti₃C₂T_x MXene, we also performed DFT calculations on the optical absorption behaviors of monolayer and bulk-thick Ti₃C₂T_x MXenes, and observed that the optical absorption properties of Ti₃C₂T_x MXene was well conserved against layer stacking (Figure 2c). Based on these theoretical results, we suppose that as-obtained highly stacked Ti₃C₂T_x MXene could serve as a superb nanophotonic platform in a wide range of engineering applications, providing a character-conserved, tightly packed 2D material assembly with a good mechanical stability. For the experimental investigation of optical absorption, we also performed an IR spectrum measurement although it may not be exactly the same with the theoretical calculation because an IR spectrum includes atomic vibration, randomly oriented illumination onto MXene planes and so on. However, estimation for the range of significant optical absorption (or qualitative analysis) would be possible. Indeed, using an IR measurement, we were able to confirm significant optical absorption of Ti₃C₂T_x MXene in the range of 400–2200 nm (Figure S1 in Supplementary Material).

3 Material and device preparation

For our experiments, we used commercially available Ti₃C₂T_x MXene materials (Invisible Inc.) which was made by selective etching of Al atomic layers from Ti₃AlC₂ MAX materials using a HF solution followed by repeated centrifuging and decanting. Details of the MXene synthesis can be found elsewhere [21], [22]. The XPS measurement of these samples showed the typical peaks of Ti₃C₂T_x (Figure S2 in Supplementary Material). For the preparation of MXene flakes suitable for our research, we mildly sonicated the MXene samples in deionized water (3000 rpm for 30 min) to obtain highly stacked Ti₃C₂T_x flakes, which were thick enough to be beyond the 2D nanomaterial regime, but also sufficiently small to make a homogeneous film coating (Figure 2d). The aqueous solution of these Ti₃C₂T_x flakes (∼1 mg mL⁻¹) was dropped onto a slide glass and dried in air for 24 h for characterization using atomic force microscopy (AFM). The AFM analysis showed that the thickness and lateral dimension of the Ti₃C₂T_x flakes spanned over 80–800 nm and 0.5–5 µm, respectively (Figure 2e).

This measured thickness corresponds to tens or hundreds of Ti₃C₂T_x layer stacking because the thickness of Ti₃C₂T_x monolayer is known as 0.98 nm, ensuring that the obtained Ti₃C₂T_x flakes are definitely beyond the mono- or few-layered 2D material regime. The Ti₃C₂T_x solution was then mixed with polyvinyl alcohol (PVA, 10 mg for the solution of 1.5 mL) and drop-cast onto a side-polished fiber device which had been prepared with a careful control of the side-to-core depth to be 6 μm for efficient evanescent coupling between Ti₃C₂T_x SAs and the laser inside the fiber cavity.

The minimum insertion loss and polarization-dependent loss of this Ti₃C₂T_x-deposited fiber-optic device were measured as 5.2 and 2.0 dB, respectively, at a wavelength of 1900 nm, ensuring tolerable optical losses. The insertion loss indicates the power loss resulting from the addition of the fiber device to the system. This loss can vary depending on the laser polarization, from which the concept of polarization-dependent loss comes out and it means the difference between the minimum and maximum insertion losses. The symmetry of the fabricated device is significantly broken due to side-polishing of the fiber and thus, the loss of this device is greatly polarization-dependent. To measure the polarization-dependent loss of the Ti₃C₂T_x-deposited side-polished fiber optical device, we placed a polarization controller (PC) between a reference input beam source and the fabricated device and measured the variation of output loss as gradually changing the input beam polarization using the PC.

4 Nonlinear optical absorption performance

In order to experimentally validate our theoretical prediction for the superb photonic performance of highly stacked Ti₃C₂T_x MXene, we investigated the intensity-dependent optical transmission behavior of the fiber-optic device whose polished side was deposited with the PVA-mixed Ti₃C₂T_x film (Figure 3a and b) as described in the previous section. The following formula was used for the fitting of the nonlinear optical transmission curve of the Ti₃C₂T_x SA device [62]:

where T(I) is the transmission, ΔT is the modulation depth, I is the input pulse energy, I_sat is the saturation energy, and T_ns is the nonsaturable loss. Noticeably, significant nonlinear optical absorption has occurred in the Ti₃C₂T_x SA device for the input of a 0.887-ps mode-locked laser pulses with a wavelength of 1900 nm and a repetition rate of 36.94 MHz. The saturation intensity of the Ti₃C₂T_x SA device was 18.6 MW/cm² and its modulation depth was measured as high as 15% (Figure 3c), indicating the superb modulation performance of highly stacked Ti₃C₂T_x SAs.

Figure 3:

The nonlinear optical absorption behavior of highly stacked Ti₃C₂T_x MXenes.

(a) The photo image and (b) schematic cross-sectional image of an optical fiber device whose side-polished surface was deposited with highly stacked Ti₃C₂T_x SAs. (c) The experimentally measured nonlinear optical transmission curve of the Ti₃C₂T_x-deposited, side-polished fiber device.

5 Laser mode-locking performance

Besides nonlinear optical transmission properties, another important characteristic of SAs is passive mode-locking for ultrafast pulsed laser production, which has particularly been of great interest for the advance of long-wavelength laser technologies including 1.9-μm infrared region in recent years. In order to explore the mode-locking potential of highly stacked Ti₃C₂T_x SAs,

We fabricated a ring-cavity Tm-Ho-codoped fiber laser system incorporated with the prepared fiber-optic device whose polished side was deposited with hundreds-of-layer stacked Ti₃C₂T_x SAs. The schematic of our laser system is depicted in Figure 4a. Noticeably, mode-locked femtosecond laser pulses were readily generated by controlling the pump power with careful tuning of the polarization. Most of components in our laser system are not (or weakly) polarization-dependent, however, the Ti₃C₂T_x SA device component is symmetry-broken due to side-polishing (Figure 3b) and thus, highly polarization-dependent. Consequently, a PC was required in our laser system to find an optimal condition of evanescent interaction between the SA and the laser beam for proper mode-locking operation, which can be achieved by gradually changing the polarization state of the laser beam using the PC (Figure 4a).

Figure 4:

Laser mode-locking performance of highly stacked Ti₃C₂T_x SAs.

(a) The schematic of the ring-cavity Tm-Ho-codoped fiber laser system. (b) The optical spectrum, (c) autocorrelation trace, (d) oscilloscope trace, and (e) electrical spectrum of output laser pulses. WDM, LD, SNR, and RBW denote wavelength division multiplexing, laser diode, signal to noise ratio, and resolution bandwidth, respectively.

The output power of mode-locked lasers was 12.5 mW at a pump power of 274 mW. The period and repetition rate of the output pulses were measured as 59.6 ns and 16.77 MHz, respectively, which were compatible with the round-trip time and fundamental resonance frequency of the 12.33-m fiber cavity of our laser system (Figure 4d). The optical spectrum of the output pulses and the sech² fitting curve for solitons showed that the center wavelength and 3-dB bandwidth were 1913.7 and 4.33 nm, respectively (Figure 4b). From the analysis using a two-photon absorption-based autocorrelator, the temporal width of the output pulses was measured as short as 897 fs (Figure 4c), indicating the generation of femtosecond pulsed laser. Noticeably, the time-bandwidth product was 0.318, achieving almost transform-limited pulses. The electrical spectrum showed a sharp peak at a fundamental repetition rate of 16.77 MHz with a signal-to-noise ratio of 66 dB (Figure 4e), indicating that quite stable mode-locked lasers were successfully produced using highly stacked Ti₃C₂T_x SAs.

The previous study for mode-locked lasers using a nanoscale few-layered Ti₃C₂T_x SA at the wavelength of 2000 nm [63] showed a larger pulse width (2.18 ps) than that (897 fs) of this study using a bulk-thick Ti₃C₂T_x SA. Hence, we conclude that, at least highly stacked Ti₃C₂T_x can be as a good SA as nanoscale few-layered Ti₃C₂T_x and comparison of current results even indicates its better performance. We also compared the nonlinear optical absorption and mode-locking characteristics of a highly stacked Ti₃C₂T_x MXene SA with those of various other SA materials at the wavelength of 1.8–2.0 μm and convinced its excellent performance as summarized in the Table 1.

6 Conclusions

For the first time, using a combined theoretical and experimental approach, we showed that Ti₃C₂T_x MXene can well preserve its electronic and optical characteristics against severe layer stacking and thus, highly stacked Ti₃C₂T_x MXene can serve as an excellent SA and/or mode-locker for 1.9-μm infrared optical switching and/or ultrafast pulsed laser production. The highly stacked Ti₃C₂T_x MXene SA has the modulation depth as high as 15% and readily produced mode-locked femtosecond pulsed lasers with a duration as short as 897 fs at the wavelength of 1914 nm in a ring fiber cavity. Theoretical calculations indicated that intrinsically existing surface terminations play a decisive role in this stacking-independent photonic performance of Ti₃C₂T_x MXenes.

By employing highly stacked Ti₃C₂T_x MXenes in photonic applications, we can not only avoid complicated processes for delamination and dispersion of 2D materials, but also provide characteristic-conserved, tightly assembled 2D material platforms for advanced nanophotonic performance. This work suggests that the application of this work would not be limited to SAs, but also could be extended to numerous other photonic areas, providing many useful insights into the design of nanophotonic systems based on MXene 2D materials.