1. Introduction

Visible lasers are of considerable interest for laser displays with great color reproduction characteristics [1], ophthalmic medical treatments [2,3], and processing of metals with reduced reflectivity in the visible such as copper and gold [4]. Solid-state lasers directly emitting in the visible based on rare-earth-doped gain media have been recently developed [5]. This progress was driven by the advance of novel blue emitting pump sources, such as indium-gallium-nitride (InGaN) laser diodes (LDs) [6] and frequency-doubled optically pumped semiconductor lasers (2ω-OPSLs) [7]. Crystals doped with praseodymium (Pr) are currently the leading laser materials for direct visible emission because of their strong emission at many lines in the visible region. Many groups reported Pr lasers using the above-mentioned blue pump sources [8–25]. In addition to Pr, fluoride crystals doped with trivalent terbium (Tb) pumped by 2ω-OPSLs [26–28] or laser diodes [29] enabled efficient green and, more interestingly, yellow laser operation.

For NIR lasers passive Q-switching is well established. However, there are limited materials recognized as saturable absorbers for the visible region, since the history of directly emitting visible solid-state lasers is relatively short compared to that of NIR lasers. We summarized the parameters of the materials previously reported to be suitable saturable absorbers in the visible region in Table 1 [30–41].

Yttrium aluminum garnet (Y₃Al₅O₁₂, YAG) doped with tetravalent chromium (Cr⁴⁺) is a representative Q-switching material for 1-µm lasers [42]. Many groups intensively studied its saturation characteristics [43–53] and demonstrated e.g. microchip Nd³⁺:YAG/Cr⁴⁺:YAG lasers generating high peak power pulses [54–56]. We previously reported the saturation parameters of the visible absorption in Cr⁴⁺:YAG, and demonstrated passive Q-switching of a diode-pumped Pr:LiYF₄ (YLF) laser at 640 nm [30]. Spinel (MgAl₂O₄) crystals doped with divalent cobalt (Co²⁺) are known as saturable absorbers for 1.5-µm lasers, e.g. based on Er³⁺ [57,58]. Demesh et al. reported on the saturation of the visible absorption in Co²⁺:MgAl₂O₄ 31]. This saturable absorber Q-switched Pr:YLF lasers at 523, 607, and 640 nm pumped by 2ω-OPSL [31] or blue LDs [32]. GaInP-based SESAMs enabled Q-switching, as well as mode-locking, of Pr lasers [33–35]. However, these SESAMs showed relatively large non-saturable losses in the visible region compared to those for the NIR region. CdTe/CdS-quantum dots (QDs) also exhibit saturable absorption for visible light. Xu et al. applied these QDs for passive Q-switching of Pr:YLF lasers at 607, 640, and 720 nm, and obtained pulses of below 300 ns duration [36]. Graphene, in theory, shows saturable absorption at any wavelength due to its zero-bandgap structure [59], often referred to as ‘Dirac cones’. Since the saturation intensity or fluence significantly increases for shorter wavelengths, the application of graphene as a saturable absorber has been mostly limited to NIR lasers. Yamashita et al. presented an empirical law stating that the saturation intensity of monolayer graphene is proportional to λ^-6, where λ is the wavelength [60]. Despite the resulting very high saturation intensity of graphene in the visible region, Kajikawa et al. reported Q-switching of a Pr³⁺ doped fluoroaluminate fiber laser using few-layer graphene [37]. Transition metal dichalcogenides (TMDs) exhibit a wide direct bandgap resonantly absorbing in the NIR to visible region [61]. There are reports on Q-switching of Pr:ZBLAN fiber lasers at 635 and 607 nm using TMDs in a polyvinyl alcohol (PVA) film [38,39]. Gold nanoparticles and black phosphorus are also suitable saturable absorbers to Q-switch Pr fiber lasers at 635 nm [40,41]. It should be noted, that the above-mentioned reports on passively Q-switched visible lasers are all based on fluoride crystals or glasses doped with Pr. There are no reports on Q-switched Tb³⁺ lasers which can potentially generate high energy pulses thanks to their very long upper state lifetime, typically ≈5 ms in fluorides [29].

In this paper, we present the characterization of transition-metal-doped crystals as saturable absorbers in the visible region. Crystals exhibit higher damage thresholds than films or layers; therefore, they sustain higher pulse energies and peak power during Q-switching. Crystalline saturable absorbers allow very good control of the parameters determining the performance of passively Q-switched lasers: the initial transmission and the modulation depth. These parameters can be tailored just by adapting the absorber’s length and/or the doping concentration of transition metal ions, while the control of these parameters in thin-film or thin-layer saturable absorbers is more challenging. To the best of our knowledge, Cr⁴⁺:YAG and Co²⁺:MgAl₂O₄ are the only crystalline saturable absorbers applied in the visible region reported so far. Transition metal ions in a variety of hosts exhibit different saturation characteristics since the crystal field of the host strongly influences their unshielded 3d-electrons. We investigated saturable absorption of Cr⁴⁺ in YAG and Mg₂SiO₄ (forsterite) as well as Co²⁺ in MgAl₂O₄, ZnGa₂O₄, LiGa₅O₈, YAG, and Gd₃Ga₅O₁₂ (GGG). We focused on the characteristics at the three major emission wavelengths of Pr:YLF lasers at 523, 607, and 640 nm, as well as the green emission wavelength of Tb doped fluorides at 545 nm to investigate the applicability of these crystals for passive Q-switching.

Table 1. Overview of previously reported saturable absorbers for visible lasers.

View Table | View all tables in this article

2. Methods

Dynamics of saturable absorbers are in general described by a four-level model schematically shown in Fig. 1(a). This model is strictly valid only if there is no intermediate level between the ground and excited states, i.e. a single exponential decay of the population change in the excited state occurs. The four-level system has three parameters: recovery time of the ground state τ (referred to as ‘recovery time’ in the following), ground-state absorption (GSA) cross section σ_gs, and excited-state absorption (ESA) cross section σ_es. The transmission of a saturable absorber is then given as:

 

Fig. 1. (a) Four-level model of a saturable absorber. (b) Experimental setups of pump-probe and Z-scan measurements.

Download Full Size | PDF

We measured the recovery time of the saturable absorbers by the pump-probe technique schematically shown in Fig. 1(b). The pump laser was a 10-Hz optical parametric oscillator (OPO) (versaScan, GWU-Lasertechnik) tunable from 410 to 2550 nm. The pulse energy of the OPO was more than 10 mJ and the pulse duration was ≈5 ns in the wavelength range of 400–700 nm. We used a continuous-wave (CW) helium-neon (HeNe) laser at 632.8 nm as the probe because all the investigated crystals exhibit a GSA at this wavelength. We focused the pump and probe beams spatially overlapping into the crystal. We kept the probe beam smaller than the pump beam to ensure a strong modulation of the probe beam, which was detected by a fast Si-photodetector (PD). A narrow bandpass filter transmitting only at 632.8 ± 1.5 nm suppressed unwanted scattered or reflected pump light. The time-dependent transmission of a saturable absorber is expressed as follows:

We performed Z-scan measurements with numerical fits to determine GSA and ESA cross sections of the saturable absorbers under investigation (see Fig. 1(b)). Again, we used the OPO as the pump source. A 1:1 beam splitter (BS) split the pump beam, and one part was directed to an energy meter (PE25-C, Ophir) to monitor fluctuations of the pump pulse energy. The other part was focused into the crystal, and a second energy meter measured the energy of the transmitted pulse. A motor-controlled stage (HPS-170, PI miCos) translated the crystal along the optical axis to vary the input pulse fluence. We recorded the transmission-versus-position curve (hereafter called ‘Z-scan curve’) for every 0.5-mm step using an automated LabVIEW (National Instrument) program. The program averaged the energy for hundred pulses at each position to suppress noise caused by pump energy fluctuations.

The GSA and ESA cross sections were determined by fitting the Z-scan curve with the model described in [50]. The model regards the spatial profile of the pump beam to be elliptical Gaussian because the model overestimates ESA cross sections if a top-hat beam profile is assumed [50]. The variation of the pump beam size in the crystals was taken into account, but the crystals were significantly shorter than the confocal length of the pump beam. Therefore, the fluence could be defined as the pulse energy divided by the beam cross section area averaged among a crystal. We separately measured the small signal absorption coefficient α by a double-beam absorption spectrometer (LAMBDA1050, Perkin Elmer) because the pump pulses in the Z-scan measurements were too energetic to measure the small signal transmission. We defined the figure-of-merit (FOM) of saturable absorbers as the ratio of GSA and ESA cross sections σ_gs/σ_es. An ideal saturable absorber with zero residual absorption has then a FOM of infinity.

3. Results and discussion

3.1 Tetravalent chromium-doped crystals as saturable absorbers

The first material system under investigation was Cr⁴⁺:YAG. Its lattice provides two different Al³⁺ sites: 60% of the Al³⁺ ions occupy tetrahedral sites, the remaining are found on octahedral sites. Cr ions substitute for the Al³⁺ on both sites. Since the Cr ions form Cr³⁺ at the Al³⁺ sites in the absence of oxygen vacancies, divalent charge compensating ions, usually Ca²⁺ or Mg²⁺ substituting respectively for the Y³⁺ or Al³⁺ ions, are codoped to enhance the formation of Cr⁴⁺. We prepared two (111)-cut Cr⁴⁺:YAG crystals: a 0.80 mm long Cr⁴⁺(0.08%):YAG sample and a 0.66 mm long Cr⁴⁺(0.11%):YAG sample. The doping levels were calculated by the absorption coefficient at 1064 nm and the GSA cross section of 6.1×10⁻¹⁹ cm² reported in [50]. Figure 2 shows the absorption spectra of both Cr⁴⁺:YAG samples. The broad absorption band centered at 1 µm corresponds with the well-known transition of Cr⁴⁺ on tetrahedral sites (upward black arrow in the energy diagram of Cr⁴⁺:YAG in Fig. 3). Cr⁴⁺ ions also exhibit strong absorption peaks at 615 and 647 nm (upward orange-red arrow in Fig. 3), and passive Q-switching of Pr:YLF lasers in this range was demonstrated using Cr⁴⁺:YAG [30]. Although the absorption around 1 µm is purely owing to Cr⁴⁺, the visible absorption is due to three different types of Cr ions: Cr³⁺ and Cr⁴⁺ on octahedral sites (Cr³⁺[O_h] and Cr⁴⁺[O_h]) as well as Cr⁴⁺ on tetrahedral sites (Cr⁴⁺[T_d]). Cr³⁺[O_h] exhibits absorption around 600 nm, which overlaps with the Cr⁴⁺[T_d] absorption in the orange-red region [51]. The strong absorption below 500 nm can be attributed to Cr⁴⁺[O_h], and its broad tail at the longer wavelength side is still not zero in the orange-red region. Thus, the presence of Cr³⁺[O_h] and Cr⁴⁺[O_h] gives rise to unwanted absorption not contributing to the saturable absorption process, which apparently increases the non-saturable losses. Since the amount of these unwanted absorption centers depends on the growth conditions as well as the post-growth treatment, e.g. annealing, the absorption spectra of Cr⁴⁺:YAG crystals are in general different. To estimate the amount of absorption due to Cr⁴⁺ [T_d], we resolved the absorption spectra using the data of the peak positions and widths due to Cr⁴⁺[T_d] and Cr⁴⁺[O_h] presented in [51]. Note that the absorption of Cr³⁺[O_h] was subtracted in the data reported here. We determined the amount of Cr⁴⁺[T_d] by fitting the absorption around 1 µm, and found that the measured orange-red absorption was stronger than that expected by the data in [51]. Thus, both Cr⁴⁺:YAG samples contain a non-negligible visible absorption due to Cr³⁺[O_h]. Table 2 lists the estimated absorption coefficients at 607 and 640 nm due to Cr³⁺ and Cr⁴⁺ for the two samples.

 

Fig. 2. Absorption spectra of the two Cr⁴⁺:YAG samples and a Cr⁴⁺:Mg₂SiO₄ (forsterite) sample. The Fresnel reflections on the uncoated facets were taken into account using the Sellmeier equations of undoped YAG [68] and undoped forsterite [69].

Download Full Size | PDF

 

Fig. 3. Energy diagrams of tetrahedrally coordinated Cr⁴⁺ in YAG (D_2d symmetry when the closest ligands are only taken into account) [51] and forsterite (C_s symmetry) [70], as well as tetrahedrally coordinated Co²⁺ in spinel or garnet crystals [76].

Download Full Size | PDF

Table 2. Determined parameters of Cr⁴⁺:YAG and Cr⁴⁺:forsterite crystals.

View Table | View all tables in this article

Table 3. Determined parameters of Co²⁺ doped MgAl₂O₄ spinel crystal.

View Table | View all tables in this article

The ground-state recovery time of Cr⁴⁺(0.08%):YAG and Cr⁴⁺(0.11%):YAG samples was estimated to be 3.8 ± 0.6 µs and 3.9 ± 0.9 µs when excited at 607 nm, respectively (see Fig. 4(a) for Cr⁴⁺(0.08%):YAG). We did not observe a visible change in the recovery time when exciting at 640 nm. The slope −1/τ was determined by fitting to the range except the first 1 µs from the excitation since the decay curves deviated from an exponential function due to a fast quenching observed just after the excitation. The quenching was more pronounced in the sample of higher doping concentration, Cr⁴⁺(0.11%):YAG. These recovery times are in good agreement with the fluorescence lifetime of the NIR emission around 1.4 µm, and we could not detect any fluorescence in the visible region; therefore, the excited ions non-radiatively relax to the ³B₂(³T₂) level and then relax to the ground state by emitting in the NIR. We tuned the excitation wavelength to determine the applicable wavelength range for passive Q-switching. Figure 4(b) shows the transmission change of the Cr⁴⁺(0.08%):YAG sample for different pump wavelength between 580 and 430 nm. We conclude that Cr⁴⁺:YAG is practically useful at wavelengths down to ≈580 nm because of a strong transient color center formation below this wavelength. The long-time-window transmission change shown in Fig. 4(c) reveals the lifetime of this color center to be as long as ≈7 ms. Except the range between 520 nm and 530 nm, bleaching was still observed at shorter wavelengths, where the color formation was clearly visible (see Fig. 4(b)). Even at 450 nm saturable absorption was detected. This observation is consistent with a dip around 525 nm in the excitation spectrum of Cr⁴⁺:YAG presented by K­ück et al. [62].

 

Fig. 4. (a) Time-dependent normalized transmission of Cr⁴⁺(0.08%):YAG and Cr⁴⁺:forsterite, pumped at 607 and 570 nm, respectively. The curve of Cr⁴⁺:forsterite is also shown ten-times magnified in the linear scale. (b) Time-dependent transmission change of Cr⁴⁺(0.08%):YAG when pumped at wavelengths between 580 and 430 nm. (c) Time-dependent transmission change of Cr⁴⁺(0.08%):YAG pumped at 430 nm recorded over 15 ms.

Download Full Size | PDF

Figure 5(a) shows the saturation curves, i.e. fluence vs. transmission, of the Cr⁴⁺:YAG samples at 607 and 640 nm. Both samples exhibit large non-saturable losses, as these saturation curves converge well below 50%. One of the reasons is the existence of unwanted absorption centers, namely, Cr³⁺[O_h] and Cr⁴⁺[O_h]. We determined the GSA and ESA cross sections by numerical fitting to these saturation curves. The resulting parameters minimizing the fitting error are summarized in Table 2. To evaluate the saturation properties of Cr⁴⁺ ions, we excluded the absorption due to Cr³⁺. For example, in the Cr(0.08%):YAG sample, we found that ≈40% of the peak absorption coefficient of 17.9 cm⁻¹ at 607 nm can be attributed to Cr³⁺ absorption. This means, the non-bleachable Cr³⁺ absorption causes significant losses in this sample, reducing the saturated transmission to below 60%. We found the GSA cross sections of the two Cr⁴⁺:YAG samples to be almost identical, but the ESA cross sections of the Cr⁴⁺(0.11%):YAG sample were 30–50% larger than for the lower doped Cr⁴⁺(0.08%):YAG sample. The difference in ESA cross sections might be due to the ratio of Cr⁴⁺[O_h] and Cr⁴⁺[T_d] ions, but also, partially, the inaccurate evaluation of the Cr³⁺ density. Note that the inaccuracy in the Cr³⁺ density only influences the determination of ESA cross sections, not GSA cross sections. As a consequence, to achieve a suitable FOM of Cr⁴⁺:YAG saturable absorbers for the visible, great care has to be taken on growth conditions and post-growth processing to reduce the amount of detrimental Cr³⁺ and Cr⁴⁺[O_h]. Note that a thin slice of ∼100 µm needs to be prepared if we apply the investigated samples for Q-switching experiments.

 

Fig. 5. Saturation curves of (a) Cr⁴⁺:YAG and (b) Cr⁴⁺:forsterite crystals.

Download Full Size | PDF

It is well-known that the 1-µm absorption in Cr⁴⁺:YAG exhibits a polarization dependence due to the local symmetry of the Al³⁺ on tetrahedral sites being reduced from T_d to S₄. The CrO₄^2- tetrahedrons stretched along <100> show different optical response [43]. We could, however, not observe the polarization dependence in the Z-scan measurement, since we used (111)-cut crystals in which the dependence is much less pronounced.

Mg₂SiO₄ (forsterite) doped with Cr⁴⁺ is a laser gain medium emitting at ≈1.2 µm [63], and also enables passive Q-switching of 1-µm lasers [50,64]. Forsterite crystals comprise Mg²⁺ on octahedral sites and Si⁴⁺ on tetrahedral sites. Cr ions substitute for both sites and respectively form Cr³⁺[O_h] and Cr⁴⁺[T_d] [65], despite the large differences in ionic radii [71]. We prepared a 1.08 mm Cr⁴⁺:forsterite sample cut perpendicular to the c-axis. Figure 2 shows its absorption spectra for E||a- and E||b-polarization. Cr⁴⁺[T_d] exhibits absorption peaks at 570 nm (E||a) and at 730 nm (E||b). The weak but broad absorption around 460 nm and the broad tail in the absorption for E||a at 600–800 nm were assigned to Cr³⁺[O_h] with inversion and mirror symmetry, respectively, according to the site-selective excitation spectroscopy in [65].

The recovery time of Cr⁴⁺:forsterite when pumped at 570 nm (E||a) was measured to be 2.6 ± 0.1 µs as shown in Fig. 4(a), which is in reasonable agreement with the fluorescence lifetime of the 1.2 µm emission [66]. The excited ions may undergo a fast non-radiative relaxation to the NIR emitting level, thus the fluorescence lifetime of the emission mainly determines the recovery time of the ground state.

We performed Z-scan measurements for E||a-polarization at 523, 545, and 570 nm (see Fig. 5(b). Bleaching was observed for all wavelengths and the fit-results are summarized in Table 2. The ESA cross sections were determined to be 1.1–1.4 × 10⁻¹⁸ cm², corresponding to a saturated transmission of ≈83% at all three wavelengths. As the estimated ESA cross sections do not strongly depend on the wavelength, we attribute the residual absorption to Cr³⁺[O_h], not contributing to the saturable absorption action. Cr⁴⁺:forsterite exhibited the highest FOM at the absorption peak, but similar to the Cr⁴⁺:YAG crystals, the saturation characteristics of Cr⁴⁺:forsterite should also depend on the growth conditions and post-growth treatment reducing the amount of Cr³⁺. Chen et al. studied the formation mechanisms of Cr⁴⁺ and Cr³⁺ in forsterite and stated that the amount of Cr³⁺ can be suppressed in low Cr doped crystals [67].

We also investigated the saturation characteristics of Cr⁴⁺:forsterite for E||b-polarization at 640 nm. The GSA and ESA cross sections were determined to be (0.95 ± 0.15) × 10⁻¹⁸ cm² and (0.55 ± 0.15) × 10⁻¹⁸ cm², respectively. Even stronger saturation was observed for longer wavelengths, e.g. 720 nm. For passive Q-switching laser in the orange-red region, E||c polarization seems to be best suited due to its strong absorption peaking at 650 nm (upward red arrow in Fig. 3) [70], however, no samples with access to E||c polarization were available for our experiments.

3.2 Divalent cobalt-doped crystals as saturable absorbers

We also prepared samples of Co²⁺ doped MgAl₂O₄, ZnGa₂O₄, LiGa₅O₈, YAG, and GGG. The thickness of the samples was 2.98, 1.30, 0.75, 3.05, and 0.60 mm, respectively. These crystals were polished and uncoated. All crystals possess a cubic lattice. The first three crystals have spinel structure and the others have garnet structure.

A spinel structure crystal provides trivalent octahedral and divalent tetrahedral cation sites. In the case of MgAl₂O₄ and ZnGa₂O₄, the octahedral sites are occupied by Al³⁺ and Ga³⁺ and the tetrahedral sites by Mg²⁺ and Zn²⁺, respectively. Doped Co ions substitute for both sites in the respective ionization state, thus Co³⁺[O_h] and Co²⁺[T_d] ions are formed, respectively. The small mismatch in their ionic radii with the substituted ions [71] makes both configurations likely. In LiGa₅O₈, more precisely written as Ga₂(Li,Ga₃)O₈, the tetrahedral sites are occupied by Ga³⁺, while the octahedral sites are shared by Li⁺ and Ga³⁺ with a ratio of 1:3. To distinguish from normal spinels, the structure of LiGa₅O₈ is called inverse-spinel. Co ions in LiGa₅O₈ substitute for octahedral and tetrahedral Ga³⁺, but not for octahedral Li⁺ due to the large mismatch in their ionic radii. To support the formation of Co²⁺ on the trivalent sites, Si⁴⁺ was codoped for charge compensation.

In the cubic garnet structure, doped Co ions substitute for octahedral and tetrahedral Al³⁺ and Ga³⁺, and Co³⁺[O_h] and Co³⁺[T_d] are formed. Also in the garnets, Si⁴⁺ was codoped for charge compensation.

Figure 6 shows the absorption spectra of the Co²⁺ samples. In all cases, the broad NIR and visible absorptions correspond to the transitions ⁴A₂(⁴F)→⁴T₁(⁴F) and ⁴A₂(⁴F)→⁴T₁(⁴P), respectively (see Fig. 3, right). Only in the spectrum of Co²⁺:YAG, we recognized additional small absorption centered around 700 nm and 1050 nm. This is due to residual Co³⁺ [72]. According to the Tanabe-Sugano diagram [73] the positions of the absorption peaks of transition metals are strongly influenced by the crystal field strength. In a crystalline host with a stronger crystal field, the energy levels blue-shifted. We thus conclude, that LiGa₅O₈ has the weakest crystal field among the investigated crystals. The position of the NIR absorption band may reflect a stronger crystal field in the aluminum oxides (MgAl₂O₄ and YAG) than gallium oxides (ZnGa₂O₄, LiGa₅O₈, and GGG), which is explained by the smaller ionic radius of Al³⁺ in any coordination compared to Ga³⁺ [71]. The reduction from T_d symmetry causes a further splitting of the energy levels and results in broader absorption. It was however not evident in the measured absorption spectra, though the local symmetry of tetrahedral sites in the garnet crystals is reduced to S₄ [43].

 

Fig. 6. Absorption spectra of Co²⁺ doped MgAl₂O₄, ZnGa₂O₄, LiGa₅O₈, YAG, and GGG.

Download Full Size | PDF

Figure 7 and 8 show the time-dependent normalized transmission of all Co²⁺ samples excited at 607 nm from the pump-probe measurements. We clearly observed bleaching in all samples. The recovery times of the garnets in the few nanosecond range were found to be significantly shorter than those of the spinels in the hundreds of nanosecond range.

 

Fig. 7. Time-dependent normalized transmission in Co²⁺ doped (a) MgAl₂O₄, (b) ZnGa₂O₄, and (c) LiGa₅O₈. The excitation wavelength was 607 nm for all the three samples.

Download Full Size | PDF

 

Fig. 8. Time-dependent normalized transmission in Co²⁺ doped (a) YAG, and (b) GGG. The excitation wavelength was 607 nm for both samples.

Download Full Size | PDF

This significant difference is related to the relaxation process from the excited ⁴T₁(⁴P) level (see Fig. 3, right). We excited the Co²⁺ samples at 532 nm and observed three emission bands around 700 nm, 850 nm, and 1200 nm in the spinels, corresponding to the transitions ⁴T₁(⁴P)→⁴A₂(⁴F), ⁴T₁(⁴P)→⁴T₂(⁴F) and ⁴T₁(⁴P)→⁴T₁(⁴F), respectively (dashed arrows in Fig. 3). The 700-nm fluorescence was two orders of magnitude stronger than the NIR fluorescence. We could not detect any fluorescence from the Co²⁺ doped garnets, which indicates that the ⁴T₁(⁴P)-level is rapidly quenched non-radiatively after the excitation. This may be explained by the higher phonon energies of garnet crystals compared to spinels [74,75], causing a stronger coupling of vibrational wave functions with phonons and thus stronger multiphonon relaxation.

We performed Z-scan measurements with Co²⁺:MgAl₂O₄ and Co²⁺:GGG. The three other crystals were not suited for this measurement due to their small aperture below 4 mm². The results are shown in Fig. 9. The determined parameters are summarized in Table 3.

 

Fig. 9. Saturation curves of the (a) Co²⁺:MgAl₂O₄ and (b) Co²⁺:GGG samples.

Download Full Size | PDF

The Co²⁺:MgAl₂O₄ strongly bleached when excited at 523, 545, 607, and 640 nm. We recorded the largest transmission change at 607 nm. Although the initial transmission at 607 nm was lower than at 640 nm, this ratio changed when bleached, indicating a smaller ESA cross section at 607 nm than 640 nm. It is remarkable that the transmission at 607 and 640 nm exceeded 97% when bleached. Despite the unknown GSA and ESA cross sections, we could estimate the FOM of Co²⁺:GGG from the initial and saturated transmission by the equation:

The Co²⁺:GGG bleached at 607 and 640 nm but not at 545 nm or shorter. When we excited at 545 nm, the transmission quickly dropped by >15% even at low fluences of 0.5 J/cm². The decrease in transmission was even more pronounced for shorter pump wavelengths, e.g. 523 nm (not shown in Fig. 9). The saturation characteristics at 607 and 640 nm strongly change around 0.2 and 0.25 J/cm², respectively. This is due to the short recovery time of Co²⁺:GGG of 12 ± 2 ns (see Fig. 8(b)), which does not longer allow to consider the 5-ns pump pulse as instantaneous excitation. Taking this into account for the calculation of GSA and ESA cross sections of the Co²⁺:GGG was beyond the capacity of our model.

The Co²⁺:MgAl₂O₄ excited at 523 and 545 nm and Co²⁺:GGG at 640 nm showed a roll-over at high excitation fluences, which can also not be explained by our model and would require taking into account further energy levels. We performed further Z-scan measurements of these samples with longer pulses from a frequency-doubled Q-switched Nd:YLF laser at 527 nm (Evolution-15, Coherent) with a longer full-width half-maximum pulse duration of 280 ns. We also observed bleaching in both samples, but no roll-over even at the highest fluences of 10 J/cm². Thus, the roll-over may be due to energy upconversion or ESA to the conduction band of the hosts, i.e. charge transfer, due to the high density of excited ions under high intensity 5-ns excitation. The corresponding parity allowed transition is expected to have larger cross sections than the 3d intrashell transitions. This may also explain the decrease Co²⁺:GGG’s transmission under excitation at 545 nm, but further investigations are required for fully understand this phenomenon.

We could also estimate the absorption cross sections of Co²⁺:MgAl₂O₄ and Co²⁺:LiGa₅O₈ in the visible using the absorption spectra (Fig. 6) and their reported GSA cross section values at 1.54 µm [58]. The GSA cross section of Co2+:MgAl₂O₄, for instance, at 607 nm was estimated to be (10.9 ± 1.9) × 10⁻¹⁹ cm² which is in good agreement with our results. In the same way, the GSA cross section of Co²⁺:LiGa₅O₈ at 607 nm was estimated to be (11.4 ± 1.8) × 10⁻¹⁹ cm². It is thus conceivable that Co²⁺ in other spinel structure crystals, including ZnGa₂O₄, exhibits similar values of GSA cross sections.

3.3 Considerations on other crystalline hosts

There is a large number of possible crystalline hosts to incorporate Cr⁴⁺ or Co²⁺ ions, and it is unrealistic to investigate all of these to find the best one. Despite the limited number of crystals investigated here, our results still have useful implications to search for further good saturable absorbers for the visible based on Cr⁴⁺ or Co²⁺.

For Cr⁴⁺, host crystals comprising a single substitution ion on a tetrahedral site are advantageous. Such crystals may exhibit a high FOM, as they suppress unwanted absorption due to Cr³⁺ and/or Cr⁴⁺[O_h]. The Cr⁴⁺:Mg₂SiO₄ (forsterite) crystal showed the best FOM at its peak absorption wavelength; however, we still observed residual absorption by Cr³⁺[O_h] substituting Mg²⁺ ions. In this respect, for instance, LiAlO₂ is an attractive host crystal for chromium because its lattice only comprises tetrahedral substitution sites.

For Co²⁺, we need to choose crystalline hosts with low phonon energies to feature a long recovery time; otherwise, the saturation intensity becomes too high. It is however not trivial to evaluate or predict the interaction between phonons and Co²⁺ ions. We can instead qualitatively check the recovery time of the red fluorescence of a given Co²⁺ doped crystal when exciting the visible absorption band. The presence of red fluorescence indicates that radiative relaxation dominates over phonon-induced non-radiative relaxation, and such crystals may have recovery times up to a microsecond, as seen for the spinel crystals. On the other hand, the absence of red fluorescence indicates very short recovery times due to multiphonon processes as observed in the garnets. The validity of this rule for other lattice structures is supported by our investigations of a Co²⁺:Ca₂MgSi₂O₇ (åkermanite) crystal. The absorption spectrum for the polarization parallel to the crystal’s a-axis was similar to the spectra shown in Fig. 6, but we could not observe any fluorescence from the crystal when the visible band was excited and the recovery time was, as expected, as short as 5 ns.

3.4 Passive Q-switching of Pr and Tb visible lasers using the saturable absorbers

For the application of our saturable absorber materials in passively Q-switched lasers, we need to consider appropriate resonator designs for each saturable absorber. The saturation intensity of the saturable absorber plays a critical role to obtain efficient Q-switching. Passive Q-switching is obtained only if the saturable absorber bleaches prior to the gain medium; otherwise, the laser operates in CW mode. The properties of the saturable absorbers investigated here are summarized in Table 4. In this section, we present resonator design considerations for two representative visible laser gain media: Pr:YLF and Tb:LiLuF₄ (LLF). In the case of Pr:YLF laser [5], the saturation intensities of the visible emission at 523, 607, and 640 nm are calculated to be ≈360, ≈60, and ≈36 kW/cm², respectively. In the case of Tb:LLF laser [5], the saturation intensity at 545 nm is calculated to be ≈46 kW/cm².

Table 4. Summarized properties of the investigated saturable absorbers.^a

View Table | View all tables in this article

It was shown, that Cr⁴⁺:YAG can be utilized for Q-switching Pr:YLF lasers at 607 and 640 nm [30]. Since the saturation intensity of Cr⁴⁺:YAG is roughly half as high as in Pr:YLF, Cr⁴⁺:YAG bleaches prior to this gain medium, if the beam sizes in both elements are comparable. Cr⁴⁺:forsterite is useful to Q-switch green lasers, i.e. Pr:YLF at 523 nm and Tb:LLF at 545 nm. The saturation intensity at 523 nm is lower for Cr⁴⁺:forsterite than for Pr:YLF. This condition loosens resonator design restrictions, because a strong focus in the Cr⁴⁺:forsterite is typically not required. Therefore, it is conceivable to miniaturize the laser resonator to shorten the pulse duration to few-ns in a so-called microchip laser [77]. Cr⁴⁺:forsterite may also be suitable to Q-switch Tb:LLF laser at 545 nm as well. Thanks to the long upper state lifetime of Tb, high energy, high peak power pulses are expected from such a Q-switched Tb:LLF laser. Note that high energy pulses tend to damage components in resonators, particularly coatings of crystals and mirrors. Thus, a resonator configuration which does not strongly focus the beam on these components is required.

Crystals doped with Co²⁺ exhibit a broad visible absorption band covering the green to red region, which makes them suitable to Q-switch both Pr:YLF and Tb:LLF lasers at all transitions in this range. All the investigated Co²⁺ crystals have, however, more than an order of magnitude shorter recovery times than the Cr⁴⁺ crystals, which results in high saturation intensities. For Co²⁺:MgAl₂O, the saturation intensity was calculated to be ≈360 kW/cm² at 607 nm and thus nearly an order of magnitude higher than in Pr:YLF and Tb:LLF. Accordingly, Co²⁺:MgAl₂O₄ saturable absorbers should be placed in a strong focus to bleach them prior to the gain medium. This is applicable to the other Co²⁺ spinels since they are expected to have saturation intensities of comparable scale. The Co²⁺ garnets must have higher saturation intensities owing to their short recovery times, which put restrictions on the resonator design. The observed decrease in transmission at high fluence level (Fig. 9) further indicates that the performance of lasers Q-switched with Co²⁺ doped garnet saturable absorbers can degrade when the intensity in the crystal is too high. This must be further studied in forthcoming laser experiments.

4. Conclusion

We investigated the saturation characteristics of the visible absorption in oxide crystals doped with Cr⁴⁺ or Co²⁺. These transition-metal-doped crystals were found to exhibit useful properties as saturable absorbers to Q-switch visible lasers. The absorption in crystals doped with Cr⁴⁺ shows a strong dependence on the choice of the host crystal, as applicable wavelengths are significantly differing between YAG and forsterite. On the other hand, the absorption in crystals doped with Co²⁺ is less sensitive to the host crystal. Instead, the recovery times of Co²⁺ ions in different host crystals differ from a few nanosecond to a microsecond. This results in saturation intensities varying over two or three orders of magnitude. Co²⁺:MgAl₂O₄ is a very suitable saturable absorber to Q-switch orange-red lasers compared to Cr⁴⁺:YAG due to its higher FOM. Co²⁺:MgAl₂O₄ can also enable a better Q-switching performance at 523 nm than Cr⁴⁺:forsterite because the FOM of Cr⁴⁺:forsterite is limited by its low GSA at this wavelength. Cr⁴⁺:forsterite, in turn, should be a better choice for green lasers of longer wavelength, e.g. at 545 nm, and yellow lasers. Other transition metal ions of divalent, trivalent, or tetravalent charge states exhibiting an absorption in the visible region may also be suitable for the application as a saturable absorber in the visible and should be investigated in future. We expect that this report guides the further research on passively Q-switched visible solid-state lasers. First experiments in this respect are in progress in our lab and will be presented in a separate report.