1 Introduction

In the laser radiation interaction with matter, various wavelengths are required to match the energy gap needed to excite an individual irradiated object. If the main aim of the interaction is to generate plasma, an ultraviolet laser is usually used. If visible and near-infrared radiation is required, various gas, dye, semiconductor, or solid-state lasers are now available. As for the 2–3 µm radiation generation, solid-state lasers with the active ions of Tm, Ho, or Er (i.e., Tm:YAG, Tm:YAP, Ho:YAG, Cr:ZnSe or Er:YAG, Er:YSGG) can be used [1–3]. Looking further into the infrared region, only gas lasers, e.g., CO (5.5 µm) or CO2 (10.6 µm), are commercially available. There is still a lack of compact, solid-state lasers operating in the 3–10 µm spectral region. These lasers are of another great importance because of their wide variety of fundamental and practical applications such as next-generation imaging devices, near-IR quantum counting devices, remote sensing, molecular and solid-state spectroscopy, clinical and diagnostic analysis, atmospheric sensing, optical metrology, and medicine. Currently, the combination of a laser generator with some nonlinear system (as stimulated Raman generator or optical parametric oscillator) is typically used to cover these wavelengths. The generation of direct mid-infrared (mid-IR) radiation using novel solid-state-laser-active material presents much simpler solution.

In the literature, it is possible to find some unique laser systems based on solid-state-active materials, e.g., Pr:LaCl3 [4], Dy:YLF [5], or Fe:ZnSe [6–13]. A suitable laser-active material fulfilling the above-mentioned requirements, i.e., compact system generating radiation in the mid-IR region, could also be dysprosium-doped lead thiogallate Dy3+:PbGa2S4 (Dy:PGS) [14, 15]. Using this unique active material, laser emission at 4.3 µm was demonstrated at room temperature, pumped by a flash-lamp-pumped Nd:YAG laser generating at a wavelength of 1.32 µm (6H15/2 → 6H9/2 + 6F11/2 transition) [16]. Also, the in-band pumping (6H15/2 → 6H11/2 transition) by Er:YAG (1.66 µm) [17] or Er:YLF (1.73 µm) [18] flash-lamp-pumped lasers was investigated for the 4.3 µm radiation generation (6H11/2 → 6H13/2 transition). The output energies up to 57 µJ at 4.32 µm and 7 mJ at 4.32 µm were obtained for 1.66 µm and 1.73 µm pumping, respectively. Further diode-pumped Nd:YAG generating radiation at 1.32 µm [19] or even direct laser diode at 1.7 µm [20] was also used for pumping the Dy:PGS active medium. For 1.32 µm pumping, a laser operating in the pulsed mode with pulse duration 4 ms, maximum pumping energy 15 mJ, and repetition rate of 20 Hz was used. Energy up to 90 μJ and mean output power 1.8 mW were obtained at 4290 nm and slope efficiency with respect to the absorbed pumping energy higher than 3 % [19]. In the case of laser diode 1.7 µm pumping (5 ms, 20 Hz), 9 % slope efficiency was achieved with the maximum mean output power 9.5 mW at 120 mW of the absorbed power [20]. Stable continuous-wave room-temperature operation was also demonstrated with the output power 67 mW at 970 mW of absorbed 1.7 µm laser pumping radiation [20].

In all these experiments, the only one laser transition from the Dy:PGS energetic scheme was selected. This article discusses the possible gradual generation of radiation of various wavelengths by Dy:PGS laser pumped by a Nd:YAG laser or laser diode operating at 1.32 µm.

2 Dysprosium-doped lead thiogallate active material

Dysprosium-doped lead thiogallate (Dy:PbGa2S4) crystal synthesized by Bridgman technique from the melt in quartz ampoules served as an active medium for the developed mid-IR laser. The nominal Dy3+ ion concentration in the melt was 0.7 wt%, and the investigated crystal sample was 16 mm long with a diameter of 19 mm. The sample was cut straight from the boule with the flat faces perpendicular to the cleavage plane (end-face parallelism ~10″).

The approximate energy diagram of Dy3+ ion constructed from the room-temperature absorption spectrum is shown in Fig. 1. As can be seen from this figure, the pumping with a commercially available 1.32 µm pump source can efficiently excite 6H9/2 + 6F11/2 level of dysprosium with two possible oscillation channels (as shown in Fig. 1) corresponding to H9/2 + 6F11/2 → 6H11/2 (λ = 5.4 µm) and 6H9/2 + 6F11/2 → 6H13/2 (λ = 2.4 µm) transitions. Also, 6H9/2 + 6F11/2 → 6H13/2 transition (λ = 4.3 µm) could be achieved using 1.32 μm pumping.

Fig. 1
figure 1

Energy diagram of Dy3+ ion constructed from room-temperature absorption spectrum

Full size image

Judd–Ofelt analysis of the Dy3+ ions non-polarized absorption spectrum has shown that the calculated radiative lifetimes for H9/2 + 6F11/2, 6H11/2, and 6H13/2 levels were 0.28, 2.3, and 6.2 ms, respectively. The measured room-temperature lifetimes for corresponding levels were 0.16, 2, and 6.2 ms, which results in 57, 90, and 100 % quantum yield of fluorescence from these levels [21]. As could be seen at a glance, despite low non-radiative losses in the lead thiogallate crystal, all possible lasing transitions should be self-terminated due to longer lower laser level lifetime compared to the upper one. Also, branching ratios from H9/2 + 6F11/2 level to 6H11/2 and 6H13/2 levels are not favorable, being only 1 and 12 %. The situation is slightly better for 6H9/2 + 6F11/2→ 6H13/2 transition, for which the branching ratio was calculated to be 20 %.

Non-radiative loss values for long-wavelength mid-IR to H9/2 + 6F11/2 → 6H11/2 and H9/2 + 6F11/2 → 6H13/2 transitions seem to be rather small despite their rather extensive phonon spectrum. The highest phonon frequencies observed in PbGa2S4 crystal spontaneous Raman spectra were 280, 360, and 400 cm−1 [22], the last being not so far from the well-known fluoride materials, where such long-wavelength fluorescence at room temperature was not observed. To find out the frequency of the effective phonon participating in non-radiative quenching, temperature dependence of 6H9/2 and 6H13/2 dysprosium levels lifetime was measured in the 77–335 K temperature interval. The result for 6H9/2 level is shown in Fig. 2. As can be seen from the figure, the 6H9/2 level lifetime at 77 K is 0.275 ms, which is very close to the pure radiative lifetime 0.28 ms calculated using Judd–Ofelt analysis.

Fig. 2
figure 2

Lifetime of 6H9/2 dysprosium level in temperature interval from 77 to 335 K

Full size image

Using the calculated radiative lifetimes, the non-radiative decay rates for 6H9/2 and 6H11/2 levels are calculated and presented in Figs. 3 and 4, respectively. These temperature dependences of non-radiative quenching rates were fitted by the known equation [23]:

$$ W = W_{0}\; \left( {n + 1} \right)^{p} , $$

where

$$ n = \left[ {\exp \left( {\hbar w/kT} \right) - 1} \right]^{ - 1} , $$

where ħw is characteristic phonon energy and p is the number of phonons involved.

Fig. 3
figure 3

Non-radiative decay rates for 6H9/2 level

Full size image
Fig. 4
figure 4

Non-radiative decay rates for 6H11/2 level

Full size image

As can be seen from Figs. 3 and 4, the observed non-radiative decay rates of temperature dependence for both 6H9/2 and 6H11/2 levels could not be fitted using the highest frequency phonons (400 and 360 cm−1) because the slope of these approximation curves is much lower (see 360 cm−1 energy phonon fit in Figs. 3, 4).

Nevertheless, both curves could be well fitted using 280 cm−1 phonon energy with 7 phonons involved in the case of 6H9/2 level (ΔE = 1900 cm−1) and 8 phonons for 6H11/2 level (ΔE = 2250 cm−1). As can be seen from these approximations, rather large number of phonons is required in multiphonon relaxation process for both levels decreasing the multiphonon relaxation probability, thus allowing to have high quantum yield for dysprosium mid-IR transitions in the lead thiogallate crystal. The necessary presence of one additional phonon for the multiphonon relaxation process in the case of 6H11/2 level is understood to lower the multiphonon relaxation rate by nearly two orders of magnitude, resulting thus in 90 % fluorescence quantum yield of this level.

On the other hand, a relatively low multiphonon relaxation rate for 6H9/2 state limits the efficiency of 6H11/2 4.3 µm upper laser level population to below 1.3 µm pumping because only 1 % of the excitation can radiatively reach the 6H11/2 state according to the 6H9/2 state branching ratio. From this point of view, sufficient increase in 4.3 µm Dy:PGS laser efficiency under direct excitation into 6H11/2 level was observed [18, 20] or involving 6H9/2 → 6H11/2, 6H11/2 → 6H13/2 cascade lasing [25] becomes quite obvious.

3 Experimental results and discussion

3.1 Laser generation at 4.3 µm

Though 6H11/2 → 6H13/2 transition is self-terminated and the lifetime of the lower laser level is ~3 times longer than the upper one, the increase in pump pulse duration from 250 µs to 1 ms should result in sufficient increase of 4.3 µm Dy:PGS laser efficiency [25]. However, the development of long-pulse flash-lamp-pumped 1.32 µm laser systems is quite a challenge, so we have used a 1.32 µm diode-pumped Nd:YAG laser as an excitation source to achieve longer pump pulse durations. The Dy:PGS crystal was inserted in a copper holder without any active cooling and placed in a semi-hemispherical stable optical cavity formed by a flat pumping mirror and a concave output coupler. In Fig. 5, the input–output curve for 10 ms pulses is presented, demonstrating slope efficiency as high as 3 %, which is very close to the best obtained efficiency of 4 % for 1.3 µm pumped 4.3 µm lasing [25], where cascade 6H9/2 → 6H11/2, 6H11/2 → 6H13/2 lasing was achieved. Even when the pulse duration was increased up to 200 ms, no sign of transition self-termination was observed (see Fig. 6).

Fig. 5
figure 5

Output power dependence of the Dy:PGS laser at 4.3 µm pumped by 1.318 µm diode-pumped Nd:YAG laser

Full size image
Fig. 6
figure 6

Oscillation of Dy:PGS laser pumped by 1.318 µm diode-pumped Nd:YAG laser. Blue curve (bottom): Dy:PGS oscillations; red curve (up): pumping 1.318 µm pumping pulse

Full size image

Dependence of the average output power on the pump pulse duration (the repetition rate was changed to keep the same duty cycle in each measurement) is shown in Fig. 7. The average output power is seen to increase for pump pulse durations up to approximately 2.5 ms, and then, it is saturated.

Fig. 7
figure 7

Dependence of the 4.3 µm Dy:PGS laser average output power on 1.32 µm pumping pulse duration

Full size image

The 6H15/2→6H9/2 ground-state pump pulse absorption dynamics for long 12 ms pulses was measured during crystal lasing by a weak probe beam. Though the pump and probe beams overlap was not optimized (and thus absolute values may not be correct), the pump absorption behavior is quite well seen from Fig. 8. One can see the bleaching of the ground-state absorption in the first part of the pulse most probably due to excitation accumulation at the long-living lower 6H9/2 state. The tendency is seen to change to ground-state absorption increase after about 2 ms from pump pulse start.

Fig. 8
figure 8

6H15/2 → 6H9/2 ground-state pump pulse absorption dynamics for 12 ms pulse measured during crystal lasing by a weak probe beam. The pump and probe beams overlap was not optimized (the absolute values may not be correct)

Full size image

To find out the nature of this process, fluorescence of Dy:PGS crystal in a broad spectral range was measured using the fiber-coupled Ocean Optics USB4000-VIS–NIR–ES spectrometer. Under 1.3 µm excitation, rather intensive up-conversion fluorescence of the higher lying 6H5/2 level (Fig. 1) with the fluorescence maximum about 820 nm was observed. For comparison, the up-conversion fluorescence spectrum is shown in Fig. 9 together with the corresponding absorption data. Integral intensity of this up-conversion fluorescence was measured depending on 1.3 µm excitation energy, and the result is shown in Fig. 10. As can be seen from this figure, in double logarithmic scale the dependence of fluorescence intensity on excitation energy is close to quadratic. This means that two particles (cross-relaxation) or two-photon (excited-state absorption) process should take place. As Dy3+ concentration in the sample was rather small (0.7 wt%), the cross-relaxation process seems to be less probable. Thus, we suppose 6H13/2 → 6H5/2 excited-state absorption (this transition has good resonance with 6H15/2 → 6H9/2 ground-state absorption transition, as shown in Fig. 1, so pump radiation could be efficiently absorbed) to be the source of this up-conversion process. Long lifetime of 6H13/2 level of 6.6 ms should support this mechanism. This excited-state absorption is playing a negative role due to absorption of a part of pump radiation, and a positive role depopulating the lower lying long lifetime 6H13/2 level with its excitation partially returned to ground state via 6H5/2 → 6H13/2 relaxation and partially populating the upper 6H11/2 laser level through fast multiphonon relaxation. This process should help to remove the self-termination limit of the 6H11/2 → 6H13/2 transition making long pulse or CW 4.3 µm operation possible.

Fig. 9
figure 9

Up-conversion fluorescence spectrum under 1.3 μm pumping

Full size image
Fig. 10
figure 10

Integral intensity of up-conversion fluorescence measured depending on 1.3 μm excitation energy

Full size image

CW operation was investigated under excitation by a commercial 1.31 µm laser diode with a maximum output power of 2 W. The diode was not fiber-coupled and contained optics at the output for fast-axis divergence compensation. The pump was focused by a single spherical lens with a focal length of 100 mm into the Dy:PGS crystal placed in the close to hemispherical cavity formed by back-plane dichroic mirror and output curved (r = 100 mm) output coupler with reflectivity 98 % at 4.3 µm. Due to different divergences for fast and slow laser diode beam axis, the pump beam inside the crystal had elliptical shape. In this case, pure CW operation was achieved without any active cooling of the active crystal with slope efficiency 1 % with respect to the absorbed pump (Fig. 11). The oscillation spectrum in this case was rather broad, the maximum being about 4.32 µm, as shown in Fig. 12.

Fig. 11
figure 11

CW 4.3 µm Dy:PGS laser output power with respect to the 1.31 µm laser diode pumping power

Full size image
Fig. 12
figure 12

CW 4.3 µm Dy:PGS laser output spectrum pumped by the laser diode at 1.31 µm

Full size image

3.2 Laser generation at 5.4 µm

Despite our numerous attempts to obtain 5.4 µm lasing at 6H9/2 → 6H11/2 dysprosium transition under long-pulse excitation by diode-pumped Nd:YAG or diode 1.32 µm lasers, we did not see any sign of such lasing, though in [25] it was observed for flash-lamp Nd:YAG 1.32 µm laser pumping as a cascade 6H9/2 → 6H11/2, 6H11/2 → 6H13/2 lasing. To test 5.4 µm (transition 6H9/2 → 6H11/2) lasing of a Dy:PGS crystal separately, the crystal was pumped by short pulses of a flash-lamp-pumped Nd:YAG 1.32 µm laser (operated at 1 Hz) in a cavity formed by a dichroic plane pumping mirror and a curved (r = 200 mm) output coupler with a reflectivity of 96 % at 5.4 µm. It should be mentioned that both cavity mirrors were reflective only within the 5.1–5.7 µm spectral range, and so further cascade lasing at 4.3 µm was not possible. Figure 13 shows Dy:PGS laser output energy at 5.4 µm versus the 1.32 µm absorbed pump energy. The slope efficiency with respect to absorbed energy is seen to be quite low, about 0.5 %, though, more important should be the pump and oscillation pulses oscillogram shown in the inset. As can be seen from this figure, despite pump duration 200 µs the 5.4 µm oscillations start practically immediately with the pump pulse (~20 µs delay) but finish after approximately 100 µs (half pump pulse duration). This early oscillation pulse termination should be the result of 6H9/2–6H11/2 dysprosium transition self-termination, as lower 6H11/2 level has about an order of magnitude longer lifetime compared to 6H9/2. Even cooling down to liquid nitrogen temperature does not enable a strong increase in efficiency and oscillation pulse duration. Unlike 6H13/2 level, the 6H11/2 is not depopulated by some additional process and the efficiency seems to become much better if cascade 6H9/2 → 6H11/2, 6H11/2 → 6H13/2 lasing quickly depopulating 6H11/2 level takes place. Figure 14 shows the oscillation spectrum of 5.4 µm mid-IR oscillations corresponding to fluorescence curve maximum.

Fig. 13
figure 13

Dy:PGS laser output energy at 5.44 µm as a function of absorbed energy. Inset shows the oscillogram of generated and pumping pulses

Full size image
Fig. 14
figure 14

Dy:PGS laser output spectrum around 5.44 µm together with the air transmission spectrum [24]

Full size image

3.3 Laser generation at 2.4 µm

For the same pumping conditions (pulsed Nd:YAG laser 1.32 µm, 200 µs pulse duration, 1 Hz), laser oscillations at 6H9/2 + 6F11/2 → 6H13/2 dysprosium transition were investigated. The input–output curve is shown in Fig. 15, demonstrating significant increase in the laser efficiency being 8.5 % in this case. Despite using output coupler with higher transmission (90 % at 2.4 µm instead of 96 % for 5.4 µm lasing), now the oscillation pulse is delayed by about ~20 μs from pump pulse start but lasts practically till its end, as can be seen from the inset in Fig. 15, despite ~20 times longer 6H13/2 level lifetime compared to 6H9/2 + 6F11/2. As it was shown above, under 1.3 µm excitation 6H13/2 level could be effectively depopulated due to the excited-state absorption of 1.3 µm pump process removing self-termination. Figure 16 presents the oscillation spectrum around 2.4 µm. Dual-line lasing in this case corresponds to lasing to two different Stark sublevels of 6H13/2 state and was observed earlier for 6H11/2 → 6H13/2 transition (lasing at wavelength 4.3 and 4.6 μm) [18].

Fig. 15
figure 15

Dy:PGS laser output energy at 2.44 µm as a function of absorbed energy. Inset shows the oscillogram of generated and pumping pulses

Full size image
Fig. 16
figure 16

Dy:PGS laser output spectrum at 2.43 µm together with the air transmission spectrum [24]

Full size image

Thus, low multiphonon relaxation rate in the Dy:PGS crystal was shown to result from a rather low effective phonon energy of 280 cm−1 involved in the process, allowing to obtain high quantum yield of both 6H9/2 and 6H11/2 states operating as upper laser levels for 5.4 and 4.3 µm mid-IR transitions, respectively. Excited-state absorption of 1.3 µm pump radiation from 6H13/2 level is suggested to be the mechanism removing self-termination of both 6H11/2 → 6H13/2 (4.3 µm) and 6H9/2 → 6H13/2 (2.4 µm) transitions and the source of efficient depopulation of long lifetime (6.6 ms) 6H13/2 level. The efficiency of excited-state absorption depopulation mechanism is shown to increase with 1.3 µm pump pulse duration. Without cascade lasing, 6H9/2 → 6H11/2 (5.4 µm) transition seems to self-terminate fast (~100 µs) after the start. Lasing at 2.4 µm 6H9/2 → 6H13/2 transition was obtained with the slope efficiency up to 8.5 % at room temperature.

4 Conclusions

Various Dy:PGS laser transitions originating from 6H9/2 + 6F11/2 level suitable for 1.32 µm pumping were investigated. Laser generation at 5.4, 4.3, and also at 2.4 µm was obtained at room temperature. This result proves that Dy:PGS laser is a promising compact source of laser radiation for the interacting experiments in the mid-IR region, not explored so far.