Preparation of amorphous silicon-doped Y₂O₃ aerogel enabling nonlinear optical features for ultrafast photonics

2.1 Characterization of amorphous silicon-doped Y₂O₃ aerogel

Based on previous studies [42], [43], [44], it has been established that yttrium ions exhibit a coordination number of eight with water molecules in aqueous solutions, owing to their heightened electrophilicity. In the current investigation, a precursor of yttrium chloride hexahydrate was dissolved in absolute ethanol; the yttrium ions were anticipated to undergo hydration, forming connections with six water molecules, which is the maximum possible coordination due to an insufficient quantity of available water molecules. In this context, it was postulated that the yttrium ions, coordinated with six water molecules, would undergo hydrolysis, releasing hydrogen ions subsequent to the introduction of a propylene oxide (PO) gel initiator. The hydroxyl yttrium species would then proceed to polycondense, leading to the formation of bi-continuously connected Y₂O₃ gel networks through a concurrent phase separation process. The entire detailed process is shown in Figure 1(a). In order to introducing the Si ions, tetraethyl orthosilicate (TEOS)/ethanol (EtOH) solutions with different volume ratios were utilized. In our case, the wet gel which was not treated with TEOS/EtOH solution was labeled as Y₂O₃-1 aerogel, actually consisting of no Si concentration at all. The wet gels treated in 25 vol% and 50 vol% TEOS/EtOH solutions for one day were denoted as Y₂O₃-2 and Y₂O₃-3 aerogels, respectively. Indeed, after the TEOS/EtOH treatment, the aerogel was of more stable structural integrity. The typical microscopic structure and morphology of the prepared amorphous Y₂O₃-3 aerogel were performed with the scanning electron microscopy (SEM) and the transmission electron microscopy (TEM). The corresponding element mapping images of amorphous Y₂O₃ aerogel are shown in Figure 1(b), which clearly show the distribution of elements Y, O, and Si. The typical SEM images with different resolution are displayed in Figure 1(c) and (d). Obviously, SEM images uncovered the irregular pores in the prepared Y₂O₃ aerogel with a size of ∼30 nm. The inset of Figure 1(c) and (d) shows the overall morphology of the sample at a scale of 10 μm. Figure 1(e) shows a typical TEM image captured at a scale of 100 nm, further revealing the porous structure of the Y₂O₃ aerogel material. The SEM and TEM images of Y₂O₃-1 and Y₂O₃-2 can be seen in Figures S1 and S2.

Figure 1:

Fabrication and Characterization. (a) Graphical illustration of the formation of Y₂O₃-3 gel networks. (b) The energy-dispersive X-ray spectroscopy (EDS) elemental mappings of amorphous Y₂O₃ aerogel. (c) And (d) SEM images at different scales. (e) TEM image.

To investigate the optical absorption properties of the prepared amorphous Y₂O₃ aerogel, the sample was ground into powder, then dissolved into anhydrous ethanol solution for ultrasonic and centrifugal treatment, and the supernatant was taken for use. The optical absorption characteristics of the material were measured using a UV–VIS–IR spectrophotometer. Clearly, as shown in Figure 2(a), the amorphous Y₂O₃ material exhibits broad absorption characteristics in the near-infrared range, suggesting the great potential for various photonics applications. Figure 2(b) represents the X-ray diffraction (XRD) analysis of the Y₂O₃ aerogels. It can be seen that with the incorporation of Si, the crystallinity of Y₂O₃ aerogels continues to deteriorate, and the reason is analyzed as follows: The Si⁴⁺ ions have a significantly smaller ionic radius (R _Si4+ = 0.04 nm) compared to the Y³⁺ ions (R _Y3+ = 0.09 nm). As a result Si doping causes stress in the Y₂O₃ lattice, degrading its crystal structure [45]. To confirm the chemical bonding in the amorphous silicon-doped Y₂O₃ aerogel, the X-ray photoelectron spectroscopy (XPS) analysis was carried out. Figure 2(c) shows the distinct evidence of the existence of Y 3d, O 1s, and Si 2p peaks. For the Y 3d binding spectrum, there are two prominent peaks at 158.0 and 160.1 eV in Figure 2(d), which can be assigned to the divisive Y 3d_5/2 and Y 3d_3/2 states [46]. Displayed in Figure 2(e), the XPS spectra of O can be deconvoluted into three peaks at 530.2, 531.5, and 532.8 eV, respectively. The binding energy of 530.2 eV came from the O–Y bond. While the other peaks at 531.5 and 532.8 eV indicated the presence of the Y–O–Si and O–Si bonds, suggesting the incorporation of Si during the material synthesis process. Furthermore, the only peak of Si 2p spectrum in Figure 2(f) is observed at 101.4 eV, which can be attributed to the inhomogeneous broadening in the non-crystalline aerogel.

Figure 2:

Characterization. (a) UV–VIS–IR absorption spectra of Y₂O₃ aerogel. (b) XRD pattern. (c–f) XPS of amorphous Y₂O₃ aerogel.

Additionally, in order to study the effect of silicon concentration on the optical properties of Y₂O₃ aerogels, X-ray diffraction (XRD) patterns, UV–VIS–IR absorption spectra and the Fourier-transformed infrared (FT-IR) spectra were used to analysis the Y₂O₃ aerogels, as shown in Figure S3. Clearly, with the Si concentration increase, the aerogel was getting more and more amorphous, while the absorption peak attributed to the Si–O bond was stronger, indicating that the Si incorporation degraded the crystal structure.

2.2 Nonlinear optical properties in Y₂O₃ aerogels

The nonlinear optical absorption measurement was carried out with an in-line twin-detector system excited with the picosecond 1590-nm fiber laser, as shown in the Supplementary Material. Obviously, the nonlinear transmission curve of the Y₂O₃ aerogels can be fitted by [47]:

here, T, ΔT, I, I _sat, and α ₀ denote the transmission, modulation depth, on-axis pump intensity, the saturation intensity and the unsaturation loss of the prepared samples. Obviously, all samples in Figure 3 possess the typical saturable absorption effect at 1590 nm. The Y₂O₃-3 aerogel exhibits the largest modulation depth of 1.65 % with a relative low saturation intensity of 0.78 MW/cm². The saturable absorption properties enabled the mode-locking performance to generate the ultrafast optical pulses.

Figure 3:

Nonlinear transmission curve of Y₂O₃ aerogel with different silicon concentration. (a) Y₂O₃-1 aerogel. (b) Y₂O₃-2 aerogel. (c) Y₂O₃-3 aerogel.

2.3 Mode-locking EDF laser near 1530 nm

The nonlinear optical (NLO) responses of Y₂O₃-based SAs show that they have admirable saturable absorption characteristics. Therefore, to compare their potential in ultrafast photonics applications, they were coupled into the ring Er-doped fiber laser cavity operating in the super C band. It is worth noting that hereby we only demonstrated the mode-locking operation with the Y₂O₃-3 aerogel SA, owing to the large modulation depth. The detailed experimental results from the passively mode-locked EDFLs with Y₂O₃-1 and Y₂O₃-2 aerogels SAs are attached in the Supplementary Material.

By tuning the polarization state and pump power, the mode-locking pulses of the EDFL near 1530 nm are realized, as shown in Figure 4. Figure 4(a) shows that the mode-locked laser runs in the conventional soliton mode-locking operation when the pump power is 191.76 mW. The spectral diagram in Figure 4(a1) shows clear and symmetrical Kelly sidebands. The central spectral wavelength was 1531.6 nm with a 3-dB bandwidth (Full width at half-maximum, FWHM) of 2.47 nm. In fact, the Kelly sidebands are resulted from the constructive interference of the pulse and dispersive wave in the laser resonator [48]. The pulse train in Figure 4(a2) shows the stable time interval of 133.6 ns between the adjacent pulses, coinciding well with the cavity length of 27.31 m. Figure 4(a3) shows the radio frequency (RF) spectrum with a central frequency of 7.487 MHz and a signal-to-noise ratio (SNR) of about 48.2 dB, indicating a stable mode-locking operation. The stability of the mode-locking state was confirmed by the RF spectrum in the 500-MHz range, as shown in the inset of Figure 4(a3). The autocorrelation trajectory in Figure 4(a4) reveals a FWHM of 1.76 ps, corresponding to a pulse width of 1.14 ps with the squared hyperbolic secant fitting.

Figure 4:

Mode-locking EDF laser near 1530 nm. (a1–a4) Conventional soliton operation; (a1) optical spectrum, (a2) pulse trains, (a3) RF spectrum, inset: the broad-range RF spectrum, (a4) autocorrelation trace. (b1–b4) Harmonic conventional soliton operation; (b1) optical spectrum, (b2) pulse trains, (b3) RF spectrum, inset: the broad-range RF spectrum, (b4) autocorrelation trace. (c1–c4) Harmonic soliton molecules operation; (c1) optical spectrum, (c2) pulse trains, (c3) RF spectrum, inset: the broad-range RF spectrum, (c4) autocorrelation trace.

When the pump power was 245.76 mW, harmonic conventional soliton mode-locking pulses were obtained, as shown in Figure 4(b). Figure 4(b1) shows the spectrum of the harmonic conventional soliton mode-locking operation. The central wavelength was 1532.3 nm, and the 3-dB spectral bandwidth was 2.68 nm with clear and symmetric Kelly sidebands. Figure 4(b2) shows the pulse sequence displayed on the oscilloscope, with adjacent pulses spaced by 2.3 ns. Figure 4(b3) shows the RF spectrum with the SNR of about 44.1 dB, a repetition rate of 434.2 MHz, corresponding to the 58th harmonic mode-locking operation. The stability of the mode-locking state was confirmed by the RF spectrum in the 3-GHz range, as shown in the inset of Figure 4(b3). The autocorrelation trace in Figure 4(b4) possesses a FWHM of 2.14 ps, with a corresponding pulse width of 1.39 ps with the squared hyperbolic secant fitting. The reasons for the formation of harmonic mode-locking operation are explained as follows. For passive mode-locking fiber lasers, with an increase in the pump power, the energy of a single soliton pulse exceeded its maximum saturable energy. Due to the peak power limitation effect and the soliton area theorem, a single soliton pulse was split into multiple pulses to form an unstable multiple-pulse state. In the case of the gain, loss, nonlinearity, and dispersion interaction of the ring cavity, these multiple pulses can equalize via a self-arrangement process and finally form an orderly distributed pulse sequence with uniform spacing, which is referred to as harmonic mode-locking [49], [50].

After obtaining the output of conventional soliton pulses, the pump power was adjusted to 215.76 mW to achieve the output of harmonic soliton molecules pulses, as shown in Figure 4(c). The spectral diagram reveals a central wavelength of 1532.6 nm, and the spectrum exhibits modulation with the modulation period of 1.86 nm, as shown in Figure 4(c1). The pulse sequence diagram shown by the oscilloscope in Figure 4(c2) has an interval of 5.56 ns between the neighbored pulses. In Figure 4(c3), the SNR is about 45 dB, and the repetition frequency is 179.7 MHz, corresponding to the 24th harmonic mode-locking operation. The inset of Figure 4(c3) shows the RF spectrum in the 1-GHz range. The autocorrelation trace with the multi-peak fitting curve is shown in Figure 4(c4), in which the FWHM of the highest peak is 2.17 ps, corresponding to the pulse duration of 1.41 ps and the peak interval is 3.8 ps. The time interval Δt and modulation period Δλ are related as follow [51]:

The time interval was calculated as 4.2 ps, almost coincident with the experimental data. The formation mechanism of the harmonic soliton molecules pulses can be explained as follows. For the fundamental frequency mode-locked single soliton pulse, the peak power of the optical pulse was clamped due to the quantization effect of the soliton energy attributed to the soliton area mechanism and limited gain bandwidth of the active optical fiber. However, with the increase of pump power, owing to the accumulated nonlinear effect and peak power limiting effect of the soliton, a single soliton pulse splits into multiple pulses with low peak energy. The pulses interact with each other through repulsive and attractive forces. Finally, when the repulsive and attractive forces got a dynamical equilibrium, stable bound solitons are generated [52]. Moreover, under the combined action of cavity gain, loss, dispersion, and cumulative nonlinearity, multiple soliton molecules pulses were automatically arranged to exhibit a uniform amplitude and equal time interval within one round trip. Consequently, the repetition rate of the output pulse was amplified to an integral multiple of the fundamental repetition rate, thus leading to the generation of harmonic soliton molecules pulses [49]. In general, there are both short distance and long distance interactions in the formation process of harmonic soliton molecules mode-locking operation: the former leads to the formation of bound state solitons, and the latter rearranges and distributes the bound state solitons uniformly in the resonator, and finally makes the laser work in the harmonic solitons molecules mode-locking operation [53].

2.4 Synchronized dual-wavelength mode-locking near 1530 and 1560 nm

In addition, synchronized dual-wavelength mode-locking near 1530 and 1560 nm are obtained under different pump power and polarization states, as shown in Figure 5. When the pump power was adjusted to 227.76 mW, the harmonic conventional soliton mode-locked EDFL coexisting with the continuous wave can be achieved after fine tune of the polarization controller. The spectrum of the harmonic conventional soliton mode-locking operation is illustrated in Figure 5(a1), showcasing the continuous wave wavelength at 1558.7 nm and the pulsed wave wavelength at 1532.1 nm, with a 3-dB bandwidth of 2.5 nm. The distinct Kelly sidebands, a typical characteristic of soliton operation, are clearly observed in the Figure 5(a1). Figure 5(a2) displays the pulse sequence obtained from the oscilloscope, with a neighboring pulse interval of 4.0 ns. Figure 5(a3) shows the RF spectrum, with the repetition frequency of 247.1 MHz, corresponding to the 33rd harmonic mode-locking operation, and the SNR of approximately 39.8 dB. The inset of Figure 5(a3) shows the RF spectrum in the 1-GHz range, showing the stability of the mode-locking operation.

Figure 5:

Synchronized dual-wavelength mode-locking near 1530 and 1560 nm. (a1–f1) Optical spectra. (a2–f2) Pulse sequence diagrams. (a3–f3) RF spectra, inset: the broad-range RF spectra.

When the pump power was adjusted to 203.76 mW, the output of 22nd harmonic soliton molecules is obtained by tuning the polarization state, as shown in Figure 5(b1). The central wavelength of the soliton molecules was 1532.1 nm and the modulation period was 1.8 nm. There was an obvious continuous wave component in the spectrum with a central wavelength of 1557.1 nm. The adjacent pulse interval was 6 ns, corresponding to repetition rate of 164.7 MHz, and the SNR is as high as 62.6 dB. The inset of Figure 5(b3) shows the RF spectrum in the 500-MHz range. Increasing the pump power to 227.76 mW, the output of the 29th harmonic soliton molecules can be obtained, as shown in Figure 5(c). The central wavelength of the soliton molecules was 1531.5 nm and the modulation period was 1.76 nm. There was also an obvious continuous wave component with a central wavelength of 1555.7 nm. The SNR is about 56.5 dB, the repetition rate is 217.1 MHz, corresponding to an adjacent pulse interval of 4.6 ns. The inset of Figure 5(c3) shows the RF spectrum in the 1-GHz range, which demonstrates the stable mode-locking operations.

Increasing the pump power to 245.76 mW and adjusting the polarization controller, we were able to achieve conventional soliton mode-locking operation and soliton molecules mode-locking operation near 1560 nm. The conventional soliton mode-locked laser operated at 1555.9 nm with a FWHM of 1.85 nm was realized, as depicted in Figure 5(d1). The continuous wave component operated at 1532.6 nm. The pulse sequence displayed by the oscilloscope, as shown in Figure 5(d2), has a pulse interval of 133.6 ns, which matches the cavity length. In Figure 5(d3), the fundamental repetition frequency is 7.487 MHz, which is consistent with the conventional soliton mode-locking operation at the wavelength of 1531.6 nm with an SNR of about 47.6 dB. The inset shows the mode-locking pulse RF spectrum within a span of 500 MHz. Figure 5(e) shows the soliton molecules mode-locking operation near 1560 nm. In Figure 5(e1), we find that the spectrum exhibits modulation with a modulation period of 1.7 nm and the central wavelength of 1557.6 nm and an obvious continuous-wave spectral component at 1531.8 nm. Figure 5(e2) shows the pulse interval is 133.6 ns. The SNR in Figure 5(e3) is about 36.1 dB, the repetition frequency is 7.487 MHz. The RF spectrum in the range from 0–500 MHz was displayed in inset of Figure 5(e3).

Finally, we achieve the synchronized dual-wavelength conventional soliton mode-locking operation near 1530 and 1560 nm. After maintaining the pump power at 245.76 mW, the polarization state was carefully tuned to achieve the dual-wavelength conventional soliton mode-locking operation. The broad gain bandwidth of Er-doped fiber enabled the laser to produce synchronized conventional soliton pulses at two different central wavelengths [54]. The dual-wavelength mode-locking spectrum, shown in Figure 5(f1), displays two separated spectral peaks at 1531.4 nm and 1557.3 nm, with 3-dB bandwidths of 2.2 nm and 2.0 nm, respectively. It can be observed that continuous wave is present in the spectrum. Figure 5(f2) displays the pulse sequence at this point, with a pulse interval of 133.6 ns, agreeing with the conventional soliton mode-locking operation when the pump power was 191.76 mW. In Figure 5(f3), the SNR is about 50.2 dB. The fundamental repetition frequency is 7.487 MHz, and the inset displays the RF spectrum in the range from 0–500 MHz, indicating the stable mode-locking operation. Owing to the wide gain band of EDFL and the high nonlinearity of SA, the laser has an approximate gain level near 1531.4 nm and 1557.3 nm, and can realize the synchronized dual-wavelength pulse output under specific pump power and polarization state. Besides, the continuous light at the top of the spectrum is the result of the interaction between the birefringence effect of the fiber and the nonlinear optical effect of SA [55].

3.1 Synthetic methods

Yttrium chloride hexahydrate (YCl₃·6H₂O, Aladdin, 99.5 %) and PO (Aladdin, 99.5 %) were utilized as the primary raw materials in this process. To initiate the synthesis, YCl₃·6H₂O was dissolved in ethanol (EtOH, Fuyu, 99.5 %) within a polyethylene mold, and the resulting solution was stirred for 30 min until a clear sol solution was obtained. Following the sol preparation, the solution was then cooled in an ice bath until reaching a temperature of 0 °C. Subsequently, a measured quantity of PO was slowly added drop by drop to the sol. In this case, the YCl₃·6H₂O EtOH solution had a molar concentration of 0.5 mol/L. The final sol with the PO addition, the molar ratio of Y³⁺ and PO was set as 1:4. After a 5-min stirring period, the solution was allowed to undergo gelation under standard room temperature conditions (25 °C, 1 bar) and aged for 1 h. During the gelation process, which lasted approximately 10 min, the flowable sol transformed into a gel structure. Two of three wet gels were solidified by immersing in 25 vol% and 50 vol% TEOS/EtOH solution for one day, respectively. To remove any impurities and by-products, all three wet gels were thoroughly washed with EtOH, a process repeated 6 times with 12-h intervals between each washing step. Finally, the wet gel was subjected to supercritical drying at 280 °C under a pressure of 100 bar for 24 h, employing an N₂/EtOH supercritical atmosphere. This step allowed for the effective removal of the liquid phase, resulting in the formation of the desired Y₂O₃ aerogel. The one without treatment in TEOS/EtOH solution, consisting of no silicon, was labeled as Y₂O₃-1 aerogel. The aerogels treated in 25 vol% and 50 vol% TEOS/EtOH solutions were denoted as Y₂O₃-2 and Y₂O₃-3 aerogels, respectively.

3.2 Fabrication of silicon-doped Y₂O₃ aerogel saturable absorber

Amorphous silicon-doped Y₂O₃ aerogel (Y₂O₃-3) was ground into the powder and added to anhydrous ethanol to create a suspension, from which the supernatant was collected. Then, the partial coating of a single-mode fiber (SMF-28e, Corning Inc.) was removed using wire strippers. Then, we fixed the ends of the stripped coated part of the fiber and heat the exposed part of the fiber with an alcohol lamp. In the meantime, the fiber was slowly stretched from both ends to create a tapered shape. The minimum waist diameter was 9.6 µm, and the length of the tapered section was 6.0 mm using a microscope. At this time, the insertion loss was measured to be about 97 %. Finally, the Y₂O₃ supernatant was dropped onto the tapered fiber, while the CW laser beam at 974 nm propagated in the tapered fiber to promote the optical deposition on the tapered fiber. After deposition, the insertion loss of Y₂O₃ – based saturable absorber was measured to be 98 %. The preparation process of the other two Y₂O₃-based SAs is the same as described above. In order to ensure the consistency of experimental conditions as much as possible, the insertion loss of the new tapered fibers was kept consistent with that of the tapered fiber mentioned above when we prepared them, and the insertion loss was still consistent after the deposition of the material. The preparation process is detailed in Supplementary Material.

3.3 Characterization

The X-ray diffraction patterns were obtained by an X-ray diffactometer (Shimadzu XRD-6100). The X-ray photoelectron spectrum was performed on a XPS system (Thermo fisher scientific K-alpha). The morphologic images were carried out with the scanning electron microscope (FEI Quanta 250 FEG) and the transmission electron microscope (FEI Tecnai G2 F20). The UV–VIS–IR absorption and the Fourier transformed infrared absorption were implemented with a UV-VIS-NIR spectrometry (Shimadzu UV-3600) and FTIR spectrometer (Thermo Fisher Scientific Nicolet iS 5).

3.4 In-line twin-detector system

To study the optical properties of the amorphous silicon-doped Y₂O₃ aerogel, we characterized nonlinear absorption properties using a home-made twin-balanced-detector system. A passively NPE mode-locked EDFL running at 1590 nm was utilized as the excitation laser with a pulse duration of 1.1 ps. Then in order to ensure the enough output power for the nonlinear optical absorption measurement, an erbium-doped fiber amplifier (EDFA) was applied to further boost the excitation intensity. A variable optical attenuator (VOA) was used to adjust the laser power input into an optical coupler (OC). The output beam from the OC was divided into two sub-beams by a 50:50 tap coupler: One sub-beam was directly detected with a power meter as the reference and the other beam entered an identical power meter after passing through the saturable absorber. The schematic diagram of in-line balanced twin-detector system is provided in Supplementary Material.

3.5 Laser configuration

The 974-nm laser diode (LD) emitted the pump beam into the laser cavity via a 980/1550 wavelength division multiplexer (WDM). The gain medium of the cavity was a 1.48-meter-long Erbium-doped fiber (EDF, Fibercore I-25) with a group velocity dispersion (GVD) of 40 ps²/km. A polarization-independent isolator (PI–ISO) and polarization controller (PC) were used to maintain unidirectional transmission of the laser beam and adjust the polarization, respectively. The total fiber length of the cavity was 27.31 m, including 25.83 m of single-mode fiber with a GVD of −22.3 ps²/km. A 10:90 tap coupler was used to filter out 10 % of the power and the rest of beam was reflected back to the laser resonator. To analyze the output mode-locked pulses, several measurement instruments were used, including an oscilloscope (MDO4104C, Tektronix Inc.), an InGaAs photodetector (3 GHz bandwidth), a spectrum signal analyzer (FPC1000, Rohde&Schwarz Inc.), a spectral analyzer (6375D, Yokogawa Inc.), and an autocorrelator (FR-103XL, Femtochrome Inc.).