Open Access
Add to BibSonomy
Abstract
Using high power quasi-cw pulse pumping, we show that energy transfer upconversion (ETU) processes in highly doped Dy3+ double clad ZBLAN fibers creates a pathway for significant excitation loss that clamps the gain. For a 4 mol.% Dy3+-doped fiber, we establish that the pump absorption is non-saturable up to a maximum launched (peak) pump power of 100 W. We propose that this arises from a co-operative three-ion ETU process. Additionally, the high power pulsed pumping of Tm3+, Dy3+-co-doped fiber produces laser relaxation spikes that appear after the pump pulse, suggesting that ETU dominates all other process during pumping.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Direct diode pumping of fiber lasers almost always requires the fiber to be double clad. The effective absorption coefficient of the fiber is then approximately proportional to the core-to-cladding area ratio ($\sim$1%) which means the core dopant concentration needs to be relatively high, i.e., typically >1 mol.% so that sufficient pump absorption takes place. For lasers that use fluoride glass (usually ZBLAN) double clad fiber, the most widely studied examples are those that use Er$^{3+}$ or Ho$^{3+}$ doping [1–3]. These fibers have rare earth concentrations >3 mol.% so that beneficial energy transfer processes can also be exploited. For these systems, energy transfer processes negate population bottlenecking and increases the gain.
Recently, there has been strong interest in developing Dy$^{3+}$-doped ZBLAN fibers for the creation of $>3\;\mathrm{\mu} \textrm {m}$ [4] and more recently 576 nm [5] light. To date, all demonstrations of lasing involving singly Dy$^{3+}$-doped fluoride result from fibers that have a single cladding and are core pumped. It is natural therefore, to extend this work to direct diode pumping using double clad fiber that has the core doped to much higher concentrations compared to core pumping. This has led to the idea of co-doping Dy$^{3+}$ with sensitizer ions such as Er$^{3+}$, Yb$^{3+}$ or Tm$^{3+}$ [6] and the recent demonstrations of directly diode pumped Tm$^{3+}$, Dy$^{3+}$ [7] and Er$^{3+}$, Dy$^{3+}$-co-doped [8] fiber lasers. The resulting slope efficiencies were surprisingly low however, being 0.75% and 5.7%, respectively. In this investigation, we show conclusively that the Tm$^{3+}$, Dy$^{3+}$-co-doped ZBLAN fiber and a high concentration 4 mol.% Dy3+-doped double clad ZBLAN fiber suffer from ETU that restricts the available gain. We measure strong non-saturable absorption associated with the 4 mol.% Dy$^{3+}$-doped double clad fluoride fiber which we attribute to co-operative energy transfer between three excited Dy$^{3+}$ ions. Compared to other ground-state terminating laser systems, Dy$^{3+}$ requires relatively large population inversions which provides a higher probability for energy transfer amongst excited states. These energy transfer processes dominate essentially all other processes.
The energy level structure of the Dy$^{3+}$ ion is characterized by two wide groupings of relatively closely spaced energy levels; a lower grouping spanning 0 – $13,000~\textrm {cm}^{-1}$ and a higher grouping spanning the region $>21,000~\textrm {cm}^{-1}$. As a result, only two laser transitions have been successfully demonstrated to lase using a fluoride glass host. Low phonon energy crystals do allow a third transition (i.e., $^{6}H_{11/2}\rightarrow ^{6}H_{13/2}$ at $4\;\mathrm{\mu} \textrm {m}$ ) to lase [9], and there is some prospect for this transition to lase using fluoride glass fiber [10]. It has also been shown spectroscopically using bulk ZBLAN glass samples [11], that the $^{6}H_{13/2}$ energy level does not suffer from appreciable concentration quenching up to a concentration of 4 mol.%. We have verified this experimentally in 4 mol.% Dy$^{3+}$ ZBLAN fiber whereby a lifetime of nominally ${540}\;\mathrm{\mu} \textrm {s}$ was measured using the frequency domain method. There are no potential ETU processes between pairs of Dy$^{3+}$ ions excited to the $^{6}H_{13/2}$ level that are resonant in fluoride glass. Doping fluoride fibers to at least this concentration level was therefore thought possible and moderate power double clad fiber lasers using 800 nm pumping demonstrable [12].
In a first experiment, a 4 mol.% Dy$^{3+}$ double clad fiber, with core diameter of $15\;\mathrm{\mu} \textrm {m}$ and $130\;\mathrm{\mu} \textrm {m}$ flat to flat double D-shaped outer cladding (Le Verre Fluoré, France) was pumped with a quasi-continuous-wave (QCW) driven (Spectra Diode Labs SDL 922) 805 nm diode laser (DILAS), up to a maximum launched peak pump power of 100 W. This pump wavelength has been shown to create sufficient gain and lasing efficiency when a Dy$^{3+}$(0.2 mol.%)-doped ZBLAN fiber was core pumped using a Ti:sapphire laser [12]. In the present case, however, no lasing was observed despite the use of a pair of high reflectivity mirrors (each broadband 98% reflecting around the $3\;\mathrm{\mu} \textrm {m}$ band) to form the resonator.
Figure 1 shows the measured and full dynamic rate equation calculation of transmitted pump power as a function of launched pump peak power ($200\;\mathrm{\mu} \textrm {s}$ pulses at 100 Hz). There is a clear departure between the simulation and experiment with the experimental absorption appearing to be non-saturable. We note that for all simulations of Dy$^{3+}$-doped fluoride fiber lasers, some bleaching of the pump absorption is needed to create sufficient gain for lasing, meaning pump absorption saturation is always expected. The nominal background loss at $3\;\mathrm{\mu} \textrm {m}$ is 0.2 dB/m and the OH concentration <1ppm molar. The results from a detailed spectroscopic study of Dy$^{3+}$-doped ZBLAN [11] established some minor quenching of the lifetime is present with increased Dy$^{3+}$ concentration. The presence of OH in the glass was shown to cause a non-radiative pathway for de-excitation of the upper laser level. This pathway is most likely limited in the present fiber due to the much lower OH concentration.
Fig. 1. Measured and calculated transmitted pump power as a function of the launched power for a 87 cm long Dy$^{3+}$(4 mol.%)-doped ZBLAN double clad fiber. The pump pulses were ${200}\;\mathrm{\mu} \textrm {s}$ long and the diode laser operated at 100 Hz. The launch efficiency was estimated to be >98%. The numerical model was written in Python and has been verified using core pumped Dy$^{3+}$-doped ZBLAN laser systems
Figure 2 shows the measured and calculated temporal characteristic of the transmitted pump for a nominal pump peak power of 100 W. The measurements clearly show that the transmitted pump does not increase across the duration of the pump pulse whereas the calculations show a steady increase in the transmitted pump power relative to a constant pump. The calculated saturation power for this fiber is approximately 3.5 W; it is therefore expected that at such large values of pump power the transmitted pump should increase relative to the input due to significant bleaching of the ground state, as is seen clearly in simulation.
Fig. 2. (a) Measured and (b) calculated transmitted pump power as a function of time during a QCW pump pulse for a 87 cm long Dy$^{3+}$(4 mol.%)-doped ZBLAN double clad fiber. The pump pulses were ${200}\;\mathrm{\mu} \textrm {m}$ long and the peak pump power 97 W.
One of the defining characteristics of the $^{6}H_{13/2}\rightarrow ^{6}H_{15/2}$ transition, is the strong overlap between emission and absorption, see Fig. 3. The difference between the emission and absorption cross sections, what we call the differential cross section, is relatively small being approximately ${6.8}\;{\times }\;10^{-26}\;\textrm {m}^{-2}$ at a lasing wavelength of ${3}\;\mathrm{\mu} \textrm {m}$. Consequently, to reach sufficient gain, relatively large fractional inversions are needed. Typical inversion fractions for other ground state terminating transitions such as those related to Yb$^{3+}$ and Tm$^{3+}$ doped silicate glass are a few percent because the absorption cross sections at the associated lasing wavelengths are relatively small. At threshold, therefore, a large proportion of the Dy$^{3+}$ ions must be excited to the $^{6}H_{13/2}$ level and the probability for higher order energy transfer processes is consequently higher. We propose that it is the relatively large excited ion fraction that allows multiple-ion energy transfer processes to occur in Dy$^{3+}$-doped ZBLAN glass.
Fig. 3. Measured emission and absorption cross sections for Dy$^{3+}$-ZBLAN glass. Inset: differential cross section.
Figure 4 shows the simplified energy diagram for three Dy$^{3+}$ ions. It has been shown [11] that the probability for ETU between two Dy$^{3+}$ ions excited to the $^{6}H_{13/2}$ level is very low because the final energy state of the process is not resonant (to within a maximum phonon energy of the glass) with a metastable energy level of Dy$^{3+}$. Shown in Fig. 4 is a co-operative three-ion ETU process which is resonant. Here, two Dy$^{3+}$ ions excited to the $^{6}H_{13/2}$ level transfer energy to a third excited Dy$^{3+}$ ion, leaving two relaxed to the ground state and the recipient ion excited to the $^{6}H_{5/2}$ level. As result of the somewhat clustered lower grouping of energy levels (spanning from the $^{6}H_{15/2}$ to the $^{6}F_{3/2}$ levels), it is not possible to observe fluorescence from this final state in fluoride glass. The existence of multiple-ion energy transfer processes has been suggested previously for several other rare earth ion systems [13,14] in which the concentration of rare earths in the host is relatively high. Note that energy migration of the $^{6}H_{13/2}$ excitation would enhance the overall rate of three-ion ETU.
Fig. 4. Simplified energy level diagram of three Dy$^{3+}$ ions showing a resonant three-ion ETU process. The red arrows represent de-excitation and the blue arrow represents upconversion excitation. This process would decrease the overall inversion by 67%.
To illustrate further the role of energy transfer in concentrated Dy$^{3+}$-doped double clad ZBLAN fibers, a previously reported Tm$^{3+}$, Dy$^{3+}$-doped fluoride fiber [7] was pumped in the same manner as above and the resonator closed with a lower reflectivity mirror set, i.e., 98% reflecting input dichroic and 50% output coupler. The lasing characteristic from this arrangement is shown in Fig. 5. We observe clear saturation of the laser slope efficiency as a function of the pump pulse energy from the QCW-driven diode laser to a value identical with the previously reported slope efficiency measured for CW pumping [7].
Fig. 5. Measured output as a function of launched pump energy. Slope efficiency ($\eta$) for two regimes indicated. The fiber was Tm$^{3+}$(5 mol.%), Dy$^{3+}$(0.5 mol.%)-doped double clad ZBLAN and was 1.46 m long. The pump pulses were ${200}\;\mathrm{\mu} \textrm {s}$ long and the diode laser operated at 100 Hz.
The measurement of the output laser pulse characteristic relative to the QCW pump characteristic for two values of the peak pump power is shown in Fig. 6. During the initial higher slope efficiency phase of lasing, the laser output pulse appears after the pump pulse. Shorter pump pulses down to ${50}\;\mathrm{\mu} \textrm {s}$ resulted in identical behavior. Clearly the gain is not high enough during pumping to overcome the loss and a higher population inversion is achieved after the pump has ceased. At higher peak pump power, the laser pulse initiates within the final portion of pump pulse and the laser output (within the pump pulse) was characterized by a series of relaxation spikes of almost equidistant temporal separation.
Fig. 6. Measured Tm$^{3+}$/Dy$^{3+}$ co-doped fiber temporal laser output characteristic (blue) relative to the launched pump (red) for (a) 2.9 mJ and (b) 7.8 mJ launched pump energy.
Figure 7 shows an ETU process [7] that is resonant, and which would cause gain clamping and pump excitation loss during the duration of the pump. The final state of the process decays entirely non-radiatively in a fluoride glass host, in much the same way the three-ion ETU process does. Thus, ETU results in heating of the fiber and no observable fluorescence. Cross relaxation amongst the Tm$^{3+}$ ions would have no measurable benefit as it promotes more ETU.
Fig. 7. Simplified energy level diagram for the Tm$^{3+}$, Dy$^{3+}$-co-doped ZBLAN system showing ground state absorption (GSA), cross relaxation (CR), energy transfer (ET) and energy transfer upconversion (ETU).
It is clear from these measurements that power scaling the $^{6}H_{13/2}\rightarrow ^{6}H_{15/2}$ transition of the Dy$^{3+}$ ion cannot involve high doping concentrations. The relatively small differential cross section for this $3\;\mathrm{\mu} \textrm {m}$ ground state terminating transition forces large excitation densities at threshold, even for relatively high-Q cavity arrangements. Combined with the possibility for energy migration of the $^{6}H_{13/2}$ level excitation, the probability of multiple-ion energy transfer and very high rates of energy transfer is not negligible. In addition, if some OH is present, it too would act as an energy sink whose energy transfer rate is also enhanced by the large fraction of excited Dy$^{3+}$ ions and energy migration. Future power scaling experiments will need lower (i.e., <1 mol.%) Dy$^{3+}$ concentrations; it has already been shown [7] using core pumping of the upper laser directly that 0.5 mol.% Dy$^{3+}$ is within scope for the production of high (>90%) efficiency. Increasing the brightness of the diode pump is also a necessary requirement because large inversions at threshold are needed. Future double clad ZBLAN fibers will therefore need high core to cladding area ratios, moderately low Dy$^{3+}$ concentrations and high brightness diode lasers for pumping. Together, this combination offers the potential for significant power scaling at wavelengths beyond ${3}\;\mathrm{\mu} \textrm {m}$. For co-doped systems, sensitizers such as Yb$^{3+}$ [6] offer an alternative approach because there is no ETU process that is resonant to within 3 maximum phonon energies of ZBLAN glass.
In summary, our measurements show that energy transfer in concentrated Dy$^{3+}$-doped ZBLAN double clad fibers is detrimental to the operation of the mid-infrared laser transition. Energy transfer upconversion is enhanced in Dy$^{3+}$-doped glass because the small differential cross section forces relatively large population inversions and strong interactions between excited states. Future development of highly doped and co-doped systems will likely need to take account of the limitations presented by ETU with judicious choice of the fiber and pump parameters.
Funding
Asian Office of Aerospace RD (FA2386-19-1-0043).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. K. Goya, H. Uehara, D. Konishi, R. Sahara, M. Murakami, and S. Tokita, “Stable 35-W Er: ZBLAN fiber laser with CaF2 end caps,” Appl. Phys. Express 12(10), 102007 (2019). [CrossRef]
2. Y. O. Aydin, V. Fortin, R. Vallée, and M. Bernier, “Towards power scaling of 2. 8 μ m fiber lasers,” Opt. Lett. 43(18), 4542–4545 (2018). [CrossRef]
3. S. Crawford, D. D. Hudson, and S. D. Jackson, “High-Power Broadly Tunable 3μm Fiber laser for the Measurement of Optical Fiber Loss,” IEEE Photonics J. 7(3), 1–9 (2015). [CrossRef]
4. M. R. Majewski and S. D. Jackson, “Highly efficient mid-infrared dysprosium fiber laser,” Opt. Lett. 41(10), 2173 (2016). [CrossRef]
5. H. Wang, J. Zou, C. Dong, T. Du, B. Xu, H. Xu, Z. Cai, and Z. Luo, “High-efficiency, yellow-light Dy3+ -doped fiber laser with wavelength tuning from 568.7 to 581.9 nm,” Opt. Lett. 44(17), 4423–4426 (2019). [CrossRef]
6. M. R. Majewski, R. I. Woodward, and S. D. Jackson, “Dysprosium Mid-Infrared Lasers: Current Status and Future Prospects,” Laser Photonics Rev. 14(3), 1900195 (2020). [CrossRef]
7. M. R. Majewski, M. Z. Amin, T. Berthelot, and S. D. Jackson, “Directly diode-pumped mid-infrared dysprosium fiber laser,” Opt. Lett. 44(22), 5549–5552 (2019). [CrossRef]
8. J. Wang, X. Zhu, R. A. Norwood, and N. Peyghambarian, “Beyond 3 μm Dy3+/Er3+ co-doped ZBLAN fiber lasers pumped by 976 nm laser diode,” Appl. Phys. Lett. 118(15), 151101 (2021). [CrossRef]
9. H. Jelínková, M. E. Doroshenko, M. Jelínek, J. Šulc, V. V. Osiko, V. V. Badikov, and D. V. Badikov, “Dysprosium-doped PbGa2S4 laser generating at 4.3 μm directly pumped by 1.7 μm laser diode,” Opt. Lett. 38(16), 3040–3043 (2013). [CrossRef]
10. M. Majewski and S. D. Jackson, “Numerical design of 4m-class dysprosium fluoride fiber lasers,” J. Lightwave Technol. 14, 1 (2021). [CrossRef]
11. L. Gomes, A. F. H. Librantz, and S. D. Jackson, “Energy level decay and excited state absorption processes in dysprosium-doped fluoride glass,” J. Appl. Phys. 107(5), 053103 (2010). [CrossRef]
12. M. Z. Amin, M. R. Majewski, R. I. Woodward, A. Fuerbach, and S. D. Jackson, “Novel Near-infrared Pump Wavelengths for Dysprosium Fiber Lasers,” J. Lightwave Technol. 38(20), 5801–5808 (2020). [CrossRef]
13. A. Lupei, V. Lupei, S. Georgescu, I. Ursu, V. I. Zhekov, T. M. Murina, and A. M. Prokhorov, “Many-body energy-transfer processes between Er3+ ions in yttrium aluminum garnet,” Phys. Rev. B 41(16), 10923–10932 (1990). [CrossRef]
14. V. K. Bogdanov, D. J. Booth, and W. E. Gibbs, “The role of a three-ion energy transfer process in the violet fluorescence in highly doped Er3+:ZB(L)AN glasses,” J. Non-Cryst. Solids 333(1), 56–60 (2004). [CrossRef]
