Impact of energy-transfer-upconversion on the performance of hybrid Er:YAG lasers

Transcription

1 Imact of energy-transfer-uconversion on the erformance of hybrid Er:YAG lasers Ji Won Kim, J. K. Sahu and W. A. Clarkson Otoelectronics Research Centre, University of Southamton, Southamton, SO17 1BJ, UK ABSTRACT Using a hybrid fiber-bulk laser scheme based on Er:YAG, we have achieved ~60 W and ~30 W of continuous-wave outut at 1645nm and 1617nm resectively, and Q-switched ulse energies u to ~30 mj (limited by coating damage). Investigation of various factors influencing laser erformance has revealed that energy-transfer-uconversion can have a very detrimental imact on efficiency, even in continuous-wave mode of oeration. In this aer we reort on the results of this study, discuss various measures for reducing energy-transfer-uconversion and its effect on laser erformance, and consider the rosects for further increase in outut ower and ulse energy. 1. Introduction Erbium-doed solid-state laser sources oerating in the eyesafe wavelength regime around μm have alications in a number of areas including free-sace communications, remote sensing and ranging. For many of these alications, the requirement for high outut ower and/or high ulse energy is often accomanied by the need for high efficiency and good beam quality. The standard aroach for roducing laser outut in the required μm wavelength region is via direct diode uming of erbium-ytterbium co-doed bulk glass or crystal lasers using diode lasers at nm. However, this aroach suffers from the roblem that a large fraction of the um ower is converted to heat in the bulk laser material resulting in strong thermal lensing, which can severely degrade laser beam quality and efficiency, and thermally-induced stress, which can cause catastrohic failure of the bulk material 1. This roblem can be avoided by using a cladding-umed fiber laser configuration. Fibers benefit from a geometry which allows simle thermal management and hence offer a high degree of immunity from the thermal effects which are so detrimental to conventional bulk solid-state lasers. Recent rogress in scaling outut ower from cladding-umed Er,Yb fiber lasers has been dramatic with maximum reorted ower levels exceeding 100 W 2. Unfortunately, due to their long device lengths and small core size, fiber lasers suffer from detrimental nonlinear effects, esecially when oerating in the high eak ower ulsed regime, which can limit efficiency. Furthermore, ulse energies are limited by amlified sontaneous emission and by damage to the fiber facets. For this reason, attention has focussed on resonant (in-band) uming of Erdoed bulk crystal lasers using direct diode uming at 1470 nm or 1532 nm 3,4 or, alternatively, a hybrid fiber-bulk laser configuration 5. The former aroach has the attraction of simlicity, but suffers from the drawback that the currently available InGaAsP/InP diode um lasers have relatively broad linewidths (~10 nm) and oor beam quality, which imose severe constraints on the choice crystal, doing level and resonator configuration, limiting flexibility and making it difficult to achieve high efficiency at room temerature. The hybrid laser aroach is requires an intermediate (fiber) laser stage as a brightness and wavelength converted, but offers very high overall efficiency and flexibility in choice of bulk gain medium, resonator design and mode of oeration. The rationale behind the hybrid laser aroach is to combine advantages of cladding-umed fiber lasers for efficient cw high-ower generation with the energy storage and high ulse energy caabilities of bulk solid-state lasers. This has the imortant advantage that most of the heat generated via quantum defect heating (tyically ~40 %) is deosited in the fiber, with only ~6-7 % of the fiber laser outut ower converted to heat in the bulk crystal. The net result is that thermal effects in the bulk Er laser are dramatically reduced leading to the rosect of much imroved efficiency, beam quality and higher outut ower. The outut beam quality from the fibre laser is determined by the waveguiding roerties of the active-ion-doed core, which can easily be tailored to roduce a single-satial-mode outut beam with little, if any, imact on beam quality due to thermal lensing. Also, by emloying wavelengthdeendent feedback rovided by an in-fiber Bragg grating or by an external cavity containing a simle diffraction grating, the Er,Yb fiber laser s wavelength can be tuned to recisely coincide with the strongest absortion line in the Er-doed bulk crystal (i.e nm in Er:YAG). Using this hybrid laser scheme, we have achieved ~60 W and ~30 W of continuous-wave outut at 1645 nm and 1617 nm resectively, and Q-switched ulse energies u to ~30 mj (limited by coating damage) 5-7.

2 However, investigation of various factors influencing laser erformance has revealed that energy-transferuconversion (ETU) can have a very detrimental imact on efficiency, even in continuous-wave mode of oeration. ETU is the rocess by which energy is transferred between two neighbouring excited ions (donor ion to accetor ion). In these case of Er:YAG (see Fig. 1), energy is transferred between neighbouring ions in the uer laser level, 4 I 13/2, and the accetor is excited to a higher energy level, 4 I 9/2, whilst the donor decays to a lower laser level, 4 I 15/2. The accetor ion then decays non-radiatively (via multi-honon relaxation) back to the uer laser level. Hence ETU converts two excited ions into one excited ion lus extra heat, resulting in a dramatic reduction in the effective uer laser level lifetime and a large increase in the fractional heat loading. The net result can be a significant degradation in laser erformance, articularly in ulsed (Q-switched) lasers or in more comlicated (higher loss) cw laser configurations requiring a high uer-laser-level oulation Thus, an accurate knowledge of the uconversion arameter and its imact on laser erformance is very imortant for ower scaling of Er-doed solid-state lasers. In this aer we resent the results of a reliminary study of the imact of energy-transfer-uconversion on the erformance of hybrid lasers based on Er:YAG in continuous-wave mode of oeration. We show that very low Er 3+ concentrations (<0.25 at.%) are required to render the imact of ETU negligible in certain oerating regimes. We also consider the rosects for further increase in outut ower and ulse energy. 4 I 9/ nm 4 I 13/2 Ion 1 Ion 2 Heat Laser 1617 or 1645 nm 4 I 15/2 Fig. 1. Energy-transfer-uconversion in Er:YAG. 2. Exeriment The exerimental set-u used to investigate the influence of ETU on laser erformance in Er:YAG is shown in Fig. 2. The main urose of our study was to see how the choice Er 3+ doing level affects the erformance. The loss of excitation due to ETU deends on the ETU rate arameter and excitation density in the uer laser level, both of which are deendent on the doing level. To erform this study, we emloyed a simle two-mirror resonator design for the Er:YAG laser comrising a lane inut couler mirror (IC) with high reflectivity (>99.8 %) at the lasing wavelength of 1645 nm and high transmission (>98.0 %) at the um wavelength of 1532 nm and a concave mirror (OC) with reflectivity of 95 % at 1550 ~ 1650 nm. The latter served as the outut couler and also ensured that unabsorbed um light was incident on the Er:YAG crystal for a second time. Pum light was rovided by a cladding-umed Er,Yb fiber laser (EYDFL) constructed in-house (see Fig. 2). The emloyed an Er,Yb double-clad fiber (EYDF) with an Er,Yb codoed hosho-silicate core of 30 μm diameter and 0.22 NA, surrounded by a ure silica D-shaed inner-cladding of 400 μm diameter with a calculated NA of The effective absortion coefficient for um light at 975 nm (launched into the inner-cladding) was measured to be ~7.1 db/m, and hence a relatively short fiber length of ~2.5 m was selected for our exeriment. Both end sections of the fiber were carefully mounted in water-cooled V-groove heat sinks maintained at 17 o C to revent thermal damage to the fiber coating due to unlaunched um ower and heat generated in the core due to quantum defect heating. Pum ower was rovided by two 975 nm nine-diode-bar um modules, which were olarisation-combined and then the resulting beam was slit into two beams of roughly equal ower to allow uming of the EYDF from both ends to sread the heat load. Pum light was launched into the fiber ends using dichroic mirrors with high reflectivity (>99.5% at 45 ) at the um wavelength, and high transmission (>98%) at nm to allow extraction of the signal beam. Feedback for lasing was rovided by a erendicularly-cleaved facet at the outut end of

3 Pum source at 975 nm Grating M1 Er, Yb fiber M nm Er:YAG L nm OC IC Fig. 2. Schematic diagram of the Er,Yb fiber laser and the Er:YAG laser resonator. the fiber and, at the oosite end, by a simle external cavity comrising an 120 mm focal length collimating lens and the relica diffraction grating (600 lines/mm), which was aligned in the Littrow configuration to rovide wavelength selective feedback for adjusting the lasing wavelength. The fiber end facet nearest the grating was angle-olished at ~14º to suress arasitic lasing between the two fiber end facets. Using this configuration, the EYDFL roduced u to 120 W of outut at 1532 nm in a beam with M 2 < 5 for ~440 W of launched um ower (Fig. 3). 125 Outut ower (W) Launched um ower (W) Fig. 3. Er,Yb fiber laser outut ower at 1532 nm versus launched um ower. The um beam from the EYDFL was focussed to a beam waist radius of 180 µm inside the Er:YAG crystal and the radius of curvature of concave outut couler (100 mm) and cavity length (15-50 mm) were selected to give a matched

4 TEM oo radius deending on the strength of thermal lensing in the laser rod. Er:YAG rods with five different doing levels were tested in the laser resonator. The doing levels (4.0 at%, 2.0 at.%, 1.0 at.%, 0.5 at.% and 0.25 at.%) and corresonding rod lengths (3.5 mm, 7.0 mm, 15 mm, 29 mm and 58 mm) were selected so that the fraction of um light absorbed for a single-ass of the um light was ~ 98% and hence the same for each rod. The corresonding absortion coefficient at 1532 nm for a 1.0 at.% Er doing level is 1.4 cm -1. The end faces of each rod were antireflection coated over the range nm (i.e. at the um and lasing wavelengths) and the rods were mounted in water-cooled aluminium heat-sinks maintained a temerature of 17 C ositioned in close roximity to the lane mirror (IC). In each case the laser erformance (i.e. threshold um ower and laser outut ower as a function of um ower) was measured. The threshold um ower was determined as the ower required for the onset of relaxation oscillation, detected using an InGaAs hotodetector and an oscilloscoe. 3. Results and discussion Table 1 shows the measured threshold um ower and sloe efficiency for the Er:YAG lasers with different Er concentrations. For low Er concentration levels (i.e at.% and 0.5 at.%) the threshold um is aroximately the same at just over 1 W. However, for higher Er concentration levels there is a dramatic increase in threshold um ower to ~ 7.8 W for the 4.0 at.% doing level. There is also a marked decrease in sloe efficiency at higher doing levels. In the case of the 4.0 at.%, lasing was terminated at slightly higher ower than threshold, suggesting a much higher fractional heat loading than for crystals with lower doing levels. It should be stressed the Er:YAG resonator is not designed for high efficiency as the outut couling transmission is too low and the rod length too long, and hence with a modified design much higher ower levels and efficiencies can be achieved at low doing levels 6. The urose of this study is a comarison between Er:YAG rods with different doing levels under otherwise identical conditions. The results summarised in Table 1 show very clearly that there is a strong deendence of laser erformance at 1645 nm on doing level under cw oerating conditions % Er:YAG 0.5 % Er:YAG 1.0 % Er:YAG 2.0 % Er:YAG 4.0 % Er:YAG Threshold um ower (W) Sloe efficiency (%) Table 1. Threshold um ower and sloe efficiency of Er:YAG crystals with different Er concentrations. The standard exression for threshold um ower, P th for a quasi-three-level laser is 11 π hν Pth= 0 L 2 fbη τ σ 2 2 δ ( ) w + w0 + fan 2 (1) where ν is the laser frequency, η is the um quantum efficiency, τ is the fluorescent lifetime, w and w 0 are the um and laser mode radii resectively, δ is the round-tri cavity loss including outut couler transmission, σ is the emission cross section, N 0 is the active ion oulation, L is the length of the laser rod and, f a and f b are the fractional oulations in the lower and uer laser levels resectively. The values for N 0 L, w, w 0, and δ are the same for each laser studied, so, neglecting the temerature deendence of the thermal occuancy factors, f a and f b, should yield the same values for threshold um ower. Using f b = 0.211, f b = , τ = 6 ms, w 0 = w = 180 μm, σ = cm 2, N 0 L = cm -2 and δ = 0.1, then from equation (1) we calculate the threshold um ower as ~0.8 W. This is in reasonable agreement with the measured threshold ower for the 0.25 at.% and 0.5 at.% Er concentrations, but not at higher concentration levels. Recently, we reorted on an analytical model for threshold um ower in the quasi-three-level laser taking into account the effect of ETU 12. The resulting modified exression for threshold um ower is

5 2 2 hν π w δ a Wu α τ δ α L( K Q) Pth= + Q + + 2σ τ η 4σ 2 hν π w L ( K Q) 2 a τ W u η α L( K Q ) Q + 4 (2) where α is the effective absortion coefficient for the um, W u is the uconversion arameter, a=1/(f a +f b ), b=f a /(f a +f b ), K = a + 2abτ W N, and b 2 2 u t Q = K 4a τ Wu Nt + b WuNt. Using equation (2) and by measuring the increase in τ threshold um ower as a function of cavity loss in a modified resonator design, we have determined the uconversion arameter for a 1.0 at.% Er:YAG crystal to be cm 3 /s. This value is in good agreement with the reviously reorted value determined in the standard way (i.e. via measurement of fluorescence decay times). If we assume that the ETU arameter is scales linearly with doing concentration 13, then from equation (2) we can calculate the threshold um ower as a function of Er concentration. The results (lotted in Fig. 4) indicate that the trend is in reasonably good agreement with that observed exerimentally. For comarison Fig. 4 also show a lot of the redicted threshold um ower as a function of doing level in the absence of ETU, but including the temerature deendence of the thermal occuancy factors (i.e. f a and f b ) assuming quantum defect heating is the sole source of heat generation. It can be seen that under our exerimental oerating conditions there is a negligible increase in three-level character due to the increase in thermal loading density with higher doing levels. This serves once again to confirm the imortance of the role of ETU on laser erformance. 16 Threshold um ower (W) Exerimental value Theoretical value including ETU Calculated value including the temerature deendence in the absence of ETU Doing concentration (%) Fig. 4. Exerimental (oen circle) and calculated threshold um ower (solid-line) as a function of a doing concentration. The dashed line shows the effect of quantum defect heating on threshold um ower in the absence of ETU. Fig.5 shows Er:YAG laser outut ower at 1645 nm versus um ower for 2.0 at.%, 1.0 at.%, 0.5 at.% and 0.25 at.% doing levels. It can be seen that in addition to the increase in threshold um ower with doing level, there is a very marked decrease in outut ower and sloe efficiency. This is artly due to the higher threshold um ower and hence oeration at a maximum um ower that is fewer times the threshold um ower. However, there is also a contribution due to thermal effects (i.e. aberrated thermal lensing) due to the extra heat generated by ETU. These results emhasize the need to use low Er 3+ doing levels for high efficiency even when oerating in cw mode.

6 Outut ower (W) % Er:YAG 0.5% Er:YAG 1.0% Er:YAG 2.0% Er:YAG sloe eff : 36 % sloe eff : 33 % Incident um ower (W) Fig. 5. Er:YAG laser outut ower as a function of incident um ower for different Er 3+ doing levels. We have also investigated oeration on the 1617 nm line. This transition terminates in a higher lying Stark level than the 1645 nm transition and hence has a much more ronounced three-level character requiring ~ 14.6 % of the Er ions to be excited to the uer laser level manifold to reach transarency. Thus, oeration at 1617 nm requires a resonator design with additional wavelength discrimination to suress the 1645 nm line or a much higher outut couled transmission. The resonator used in our exeriments (shown in Fig. 6) emloyed a simle four mirror-folded cavity comrising a lane inut couler (IC) with high reflectivity (>99.8 %) at the lasing wavelength of 1645 nm and nm Er, Yb fiber laser R1 IC Er:YAG Outut R nm OC Fig. 6. Schematic diagram of the Er:YAG resonator design for oeration at 1617 nm. IC : inut couler mirror with AR at 1532 nm and HR at nm. OC : outut couler mirror with a transmission (T) of 50 % at nm. high transmission (>98.0 %) at the um wavelength of 1532 nm, two concave mirrors (R1) with radius of curvature 100 mm and high reflectivity at the lasing wavelength and a lane outut couler with a relatively high transmission of 50 % at the lasing wavelength. Laser erformance was investigated for Er:YAG rods with Er doing levels of 1.0 at.%, 0.5 at.% and 0.25 at.% and corresonding rod lengths 15 mm, 29 mm and 58 mm. We were unable to achieve lasing using rods with higher doing levels. Figure 7 shows the laser outut ower versus incident um ower for the three doing levels. It can be seen that laser erformance is more strongly deendent on the Er doing level than for 1645 nm

7 oeration with very ronounced roll-over in outut ower for 1.0 at.% and 0.5 at.% doing levels. Only the 0.25 at.% doing level shows no sign of a ower roll over, yielding a maximum outut ower of ~26 W for 72 W of um ower. We attribute the difference in erformance for the different Er doing levels to ETU, which, due to the need for a higher excitation density to reach threshold than was the case for 1645 nm oeration, has a much more severe imact on laser erformance. The roll-over in ower is due to the increased thermal loading which results from ETU. Outut ower (W) % Er:YAG 0.5 % Er:YAG 1.0 % Er:YAG Incident um ower (W) Fig. 7. Er:YAG laser outut ower at 1617 nm versus incident um ower. 4. Conclusion We have studied the effect of energy-transfer-uconversion on the erformance of hybrid (fibre-laser-umed) Er:YAG lasers with different Er doing levels. Our results shown that uconversion can have a very detrimental imact on erformance (esecially for oeration on the 1617 nm line) unless a very low Er doing level (i.e. <0.25 at.%) is used even in continuous-wave mode of oeration. This is a surrising result and serves to further confirm the benefits of the hybrid laser architecture over direct diode uming as the latter requires high doing levels due oor um beam quality. This work was funded by the Electro-Magnetic Remote Sensing (EMRS) Defence Technology Centre, established by the UK Ministry of Defence. Reference 1. V. I. Zhekov, T. M. Murina, A. M. Prokhorov, M. I. Studenikin, S. Georgescu, V. Luei, and I. Ursu, Cooerative rocess in Y 3 Al 5 O 12 :Er 3+ crystals, Sov. J. Quantum Electron. 16, 274 (1986). 2. Y.Jeong, S.Yoo, C.A.Codemard, J.Nilsson, J.K.Sahu, D.N.Payne, R.Horley, P.W.Turner, L.M.B.Hickey, A.Harker, M.Lovelady, A.Pier, Erbium:ytterbium co-doed large-core fiber laser with 297W continuous-wave outut ower, IEEE J. Sel. To. Quantum Electron. 13, 573 (2007). 3. Dmitri Garbuzov, Igor Kudryashov and Mark Dubinskii, Resonantly diode laser umed 1.6-μm-erbium-doed yttrium aluminium garnet solid-state laser, Al. Phys. Lett. 86, (2005). 4. Dmitri Garbuzov, Igor Kudryashov and Mark Dubinskii, 110 W (0.9 J) ulsed ower from resonantly diode-laserumed 1.6 μm Er:YAG laser, Al. Phys. Lett. 87, (2005). 5. D. Y. Shen, J. K. Sahu and W. A. Clarkson, Highly efficient in-band umed Er:YAG lasers with >60 W of outut at 1645 nm, Ot. Lett 31, 754 (2006). 6. D. Y. Shen, J. K. Sahu and W. A. Clarkson, Electro-otically Q-switched Er:YAG laser in-band umed by an Er Yb fiber laser, Advanced Solid-State Photonics 2006 Nevada WD4 (2006).

8 7. J. W. Kim, J. K. Sahu, W. A. Clarkson, High ower in-band umed Er:YAG laser at 1617 nm, CLEO/Euroe- IQEC 2007 Munich CA6-4-TUE, (2007). 8. P. J. Hardman, W. A. Clarkson, G. J. Friel, M. Pollnau, D. C. Hanna, Energy-transfer uconversion and thermal lensing in high-ower end-umed Nd:YLF laser crystals, IEEE Journal of Quantum Electronics 35, 647, (1999). 9. C. Jacinto, S. L. Oliveira, T. Catunda, A. A. Andrade, J. D. Myers, Uconversion effect on fluorescence quantum efficiency and heat generation in Nd 3+ -doed materials, Ot. Exress, 13, 2040 (2005). 10. S. Bjurshagen, R. Koch, Modeling of energy-transfer uconversion and thermal effects in end-umed quasi-threelevel lasers, Al. Ot. 43, 4753 (2004). 11. T. Y. Fan, R. L. Byer, Modeling and cw oeration of a quasi-three-level 946 nm Nd:YAG laser, IEEE J. Quantum Electron. 23, 605 (1987). 12. J. W. Kim, I. O. Musgrave, M. J. Yarrow, W. A. Clarkson, Simle technique for measuring the energy-transferuconversion arameter in solid-state laser materials, CLEO/Euroe 2007 Munich CA-40-MON (2007). 13. M. O. Iskandarov and A. A. Nikitichev, A. I. Steanov, Quasi-two-level Er:Y 3 Al 5 O 12 laser for the 1.6 μm range, J. Ot. Technol. 68, 885 (2001).