Open Access
Add to BibSonomyAbstract
We report for the first time the conversion of incoherent infrared light around 4.4µm into a near-infrared signal at 810nm in erbium-doped GaGeSbS fibers and bulk glass samples. This energy conversion is made possible by pumping erbium doped chalcogenide samples at 982 nm and simultaneously exciting them with a 4.4µm infrared signal. This result paves the way for the development of an “all-optical“ gas sensor able to detect various gas traces using a remote detection based on commercial silica fibers.
© 2015 Optical Society of America
1. Introduction
Greenhouse gas emissions play a major role in global warming. Within this context, IR gas sensors are key elements to fulfill the crucial need to accurately detect, identify and quantify gas contents. Emissions from rare-earth ions embedded into chalcogenide glasses, which are well-known for their low phonon energies [1–3], can be advantageously used to develop optical IR gas sensors. These rare-earth doped glasses because of their wide IR emission bands can be used as IR emitters enabling the use of differential and multiple detections for the identification of various gases by absorption spectroscopy [4–6].
We have previously developed a CO2 gas sensor prototype based on the transmission through a CO2 gas cell of an IR probe signal from a Dy3+ doped 2S2G fiber [4,6]. A differential detection was implemented using two emission wavelengths within the 4.3µm Dy3+ emission band. One of the wavelengths matches the CO2 absorption while the other wavelength is not absorbed and is thus used as a reference. Based on these promising first results, we are developing a second prototype designed to be an all-optical remote sensor able to accurately measure gas concentrations or changes in gas content [7]. In principle, this remote detection can be achieved using optical fibers to transport the IR signal, after having gone through the gas cell, away from the detection location. However, infrared fibers such as chalcogenides present significant propagation losses (~0.5-2 dB/m) impairing their use over large distances. An interesting solution to this problem is the frequency conversion of the transmitted IR probe signal into visible or near-infrared light in an Er3+ doped Ga-Ge-Sb fiber. This frequency conversion enables thus the transport of the converted signal over few kms using standard silica fibers to finally reach a Si detector. Moreover, a Si detector offers a higher sensitivity and limited electronic noise in comparison with standard IR detectors. The remote aspect of this optical gas sensor evidently separates the gas location from the actual gas detection part increasing the scope of possible applications such as the detection of hazardous gases.
Frequency conversion using non-linear parametric processes necessitates very high electromagnetic field intensities. In a similar way, upconversion processes based on energy transfer require also a high level of excitation density since the interaction takes place between excited ions. Unlike these typical upconversion mechanisms, Excited State Absorption (ESA) can be used to convert a very weak signal from IR to visible following Bloembergen’s original idea [8].
To the best of our knowledge, the only IR to visible conversion using an ESA process in Er3+ doped materials have been limited to the conversion of near-infrared signals typically around 1.5µm into visible light [9–12]. In the present paper, the conversion of a 4.4µm IR signal into an 810 nm radiation is demonstrated for the first time using the ESA process displayed in Fig. 1. The first step of the conversion process consists in the excitation of Er3+ ions into the 4I11/2 level by a continuous-wave (CW) pump laser at 982 nm. During the second step of the process, a modulated probe around 4.4µm from a black body source promotes ions into the 4I9/2 level.
Fig. 1 Er3+ energy level diagram, two-step upconversion mechanism by ESA from 4I11/2 to 4I9/2 resulting in emission around 810nm (4I9/2 to 4I15/2 transition), upconversion (T22) by energy transfer between two Er3+ ions in the 4I11/2 level.
As a result of this two steps excitation, a modulated emission around 810nm (4I9/2 to 4I15/2 transition) arises at the probe modulation frequency. Ideally, following this scheme, each IR photon can be converted into a photon emitted from the 4I9/2 level.
2. Methods
Er3+ doped glass fibers with a single refractive index and bulk samples with the Ga5Ge20Sb10S65 exact composition were used during the experiments with three Er3+ doping levels: 1000, 5000 and 10000 ppm. These glasses were prepared by conventional melting and quenching methods [13]. High purity raw materials and purified sulfur were used for the synthesis of bulk glasses and glass preforms for drawing fibers. Er3+ doped fibers with a 400µm diameter and a 1000 ppm Er3+ content were obtained by drawing a 10 mm diameter and 10 cm long preform.
Room temperature emission spectra were recorded with a MS 260i ORIEL monochromator and detected using a photomultiplier tube (PMT) for wavelengths from 185 nm to 900 nm or an InSb photodiode cooled with liquid nitrogen for the 1µm to 5.5µm spectral region. The 982 nm pump signal from a CW Ti:Sapphire laser was focused onto the sample with a 200mm focal lens. A Thermo Scientific EverGlo IR source along with a long-wave pass filter having a 3.5µm cut-off wavelength was used as IR probe and focused on the different samples with a CaF2 lens having a 25mm focal length. During the conversion experiments, the pump and probe beams were kept collinear so as to maximize their spatial overlap as displayed in Fig. 2. The IR probe was modulated by a mechanical chopper at 70 Hz while the pump was still CW. A lock-in amplifier was used to detect the visible converted signal at the probe modulation frequency in order to reject unwanted signals due to competing upconversion mechanisms which will be discussed in details further in the text. As mentioned earlier, the converted signal is detected using a monochromator and a PMT while the sensor prototype at its final stage will be using a Si detector. While a Si photodiode is less sensitive than a PMT, the sensor will not comprise any monochromator, but simply instead a bandpass filter placed in front of the Si photodiode allowing for more signal to reach the photodiode.
Fig. 2 Experimental setup with the Ti-Sa laser at 982nm and the filtered blackbody source. The small mirror reflecting the pump beam to the sample is of millimeter size, thus not impairing the collection of the fluorescence from the sample toward the monochromator.
3. Results and discussion
3.1 Wavelength conversion
Er3+ is an attractive dopant because of its 4I11/2 to 4I9/2 Mid-IR ESA transition covering a spectral region from 4.2µm to 4.8µm, that matches absorption of various gases such as CO2 [2]. Figure 3 shows the 4I9/2 to 4I11/2 Er3+ doped chalcogenide glass emission spectrum calibrated in emission cross-sections using the well-known füchtbauer-ladenburg equation [14] and the corresponding ESA spectrum calculated using the inverse reciprocity method [14] leading to a maximal ESA cross-section of 2.16 × 10−21 cm2 at 4.4µm. In most host materials, the 4I9/2 level emission is essentially quenched by multiphonon relaxation impairing its use for a realistic application. However, germanium based sulfide glasses exhibit low phonon energies with a maximum frequency around 340cm−1 thus enabling efficient emission from the 4I9/2 level. The 4I9/2 radiative quantum efficiency was estimated to be 64% by comparing the measured lifetime (τexp = 0.7ms) in a 500ppm Er3+ doped Ga-Ge-Sb-S glass with the radiative lifetime (τrad = 1.1ms) calculated using the Judd-Ofelt formalism [15].
Fig. 3 4I9/2 to 4I11/2emission and calculated excited state absorption (ESA) spectra in Er3+ doped Ga-Ge-Sb-S samples after direct excitation of the 4I9/2 level at 806 nm.
An example of IR to visible conversion is given in Fig. 4 for a 104 ppm Er3+ doped sulfide bulk glass sample. When the modulated IR probe is solely sent onto the glass sample (trace 1) no emission signal around 810nm is detected. In a similar way, when the modulated probe is turned off and the CW pump light at 982nm impinges on the chalcogenide glass sample (trace 2), no converted signal can be recorded at the IR probe frequency. However, when both the IR modulated probe and the CW pump simultaneously excite the sample, one can observe a clear emission spectrum around 810 nm corresponding to the 4I9/2 to 4I15/2 emission transition. The same conversion signature is observed in all bulk samples as well as in Er3+ doped sulfide fibers.
Fig. 4 Emission spectra recorded around 810nm with a 104 ppm Er3+ doped Ga-Ge-Sb-S sample under different conditions. 1- modulated IR probe only with a × 10 factor to be able to see the noise, 2- CW pump at 982 nm only and 3-CW pump and modulated probe simultaneous excitation.
The modulation frequency in Fig. 4 was at 70Hz, but, as expected, the converted intensity does not vary much with the modulation frequency except for frequencies higher than 500Hz where the converted intensity decreases significantly since the frequency becomes then too high compared to the 1.36ms lifetime of the 4I11/2 level to allow for an efficient pumping of the level.
3.2 Competing processes
Results presented in Fig. 4 are very promising, but a thorough investigation of the key parameters at play in this conversion process is crucial. One of the most important aspects of this system is the occurrence of competing processes which tend to empty the 4I11/2 level from which the IR ESA transition takes place (Fig. 1). Two prominent competing mechanisms, which are solely due to the pump photon flux at 982nm, can be identified [16]. The first one is the upconversion energy transfer T22 (Fig. 1) in which two excited Er3+ ions in the 4I11/2 level interact with each other exciting one ion into the 4F7/2 level while the other ion returns to the ground-state. The second competing process is the excited state absorption at the pump wavelength exciting Er3+ ions from 4I11/2 to 4F7/2. This ESA process is sensitive to the pump wavelength since the wavelength has to match both 4I11/2 to 4F7/2 and 4I15/2 to 4I11/2 transitions. Following both processes, Er3+ ions are excited in the 4F7/2 level and then relax down to the 4S3/2 level from which a significant part of excited ions ends up in the 4I9/2 level through radiative transitions or cross-relaxation processes [17]. As a result, the population of the 4I9/2 level in our pump-probe experiment can either be due to the IR probe light through the conversion process detailed in Fig. 1 or only to the pumping level following either the T22 upconversion or the pump-related ESA. The use of a modulated IR probe along with a CW pump enables the rejection of these competing processes in the conversion detection.
Nevertheless, these parasitic processes might bring about noise in the converted signal by adding a significant CW background. In order to assess the respective efficiency of these processes, the 810nm emission from the 4I9/2 level was recorded as a function of the 982nm pumping power. Figure 5 shows for a 1000ppm Er3+ doped 8 cm long sulfide fiber on one hand the evolution of the pump-probe converted intensity (Ic) and on the other hand the pump related upconversion intensity (Ip) recorded without the IR probe. The dependence of both processes with the pump power (P) is clearly different. Adjustments of the experimental data with a Pn function are represented in Fig. 5 as solid lines. The pump induced upconversion signal (Ip) exhibits an almost quadratic dependence with an exponent value n of 1.7 while Ic has a slightly sub-linear dependence with n = 0.8. One would expect a linear slope (n = 1) for the Ic intensity with respect to the pump power since the IR probe intensity is kept constant. One could think that this apparent Ic sub-linear behavior is due to the pump related ESA and/or the T22 upconversion mechanisms which reduce the converted intensity by lowering the 4I11/2 level population from which the IR probe absorption originates. To verify that the impact of these two processes is small, one can examine the following 4I11/2 level population rate equation in the steady state regime:
where σp is the Er3+ absorption cross-section at the pump wavelength, Φp the pump photon flux, NT the total Er3+ concentration, τ2 the 4I11/2 level fluorescence lifetime, σESAp the Er3+ excited-state absorption cross-section at the pump wavelength (λ = 982nm) and T22 the upconversion parameter (in cm3.s−1). The maximum pump power density is kept during the experiments at a rather low value of 100W/cm2. Therefore, the excited-state absorption at the pump wavelength can be neglected in Eq. (1) as σESApΦp has a rate of 2.5s−1 (σESAp = 5 × 10−21cm2 [16] measured in similar sulfide glasses) while 1/τ2 is equal to 730s−1 [15]. The 4I11/2 level population can then be derived from Eq. (1) as:The term is equal to 0.06 (σp = 7 × 10−21cm2 [15], NT = 1.15 × 1020cm−3 for 104 ppm, T22 = 2 × 10−17cm3.s−1 [15]) and is therefore very small compared to unity in Eq. (2), allowing this equation to be expanded in Taylor series:which shows that the impact of the T22 upconversion mechanism on the 4I11/2 level population is limited within our experimental pumping conditions. Therefore, the slight deviation of the Ic intensity from a linear slope as a function of the pump power cannot be explained by pump related mechanisms depleting the 4I11/2 level, and remains unclear being possibly due to thermal effects as observed in Er3+ doped chalcogenide glasses [18].Fig. 5 Emission intensities at 810 nm in log scale as a function of the 982nm pump power in a 1000ppm Er3+ 8cm long sulfide fiber. Ip represents the pump related upconversion intensity when the fiber is excited with only the modulated pump and Ic is the converted intensity recorded under simultaneous excitation with the CW pump and a modulated IR probe. Inset: converted intensity to upconversion intensity ratio Ic/Ip versus pump power in log scale. Solid lines are adjustments of experimental data with a Pn fitting function.
The choice of a 1000ppm Er3+ 8cm long sulfide fiber is first motivated by the optical losses at the converted signal wavelength of 810nm which are about 0.4db/cm. These optical losses prevent the use of a much longer fiber, which would in return allow the use of a lower Er3+ concentration.
Intensities in Fig. 5 are given in arbitrary units as it is usually difficult to give an absolute value of radiative intensities in terms of emitted photon irradiance for instance. On the opposite, since both Ic and Ip emissions arise from the same 4I9/2 to 4I15//2 electronic transition, the calculation of the Ic/Ip ratio does not require any calibration and directly gives a meaningful value of the intensity ratio. The inset in Fig. 5 exhibits the Ic/Ip ratio which reaches a maximum value of 0.06 for a pumping power of 50mW indicating that Ip is always larger than Ic. It is therefore important to further describe the Ic/Ip ratio evolution with the pump power. As displayed in the inset of Fig. 5, the Ic/Ip ratio decreases with the pump power. A simple approach based on rate equations can be used to explain this non-linear dependence of the Ic/Ip ratio provided that the pump power density remains at a rather low value such as in this study as emphasized by Eq. (3) showing that these experiments remain within a linear pumping regime. Therefore, any depletion from the 4I15//2 level by the pump can be neglected.
The pump-related upconversion intensity (Ip) recorded without the IR probe is proportional to the 4I9/2 emitting level population n3 which can be simply derived by taking into account the T22 upconversion:
with τ3 the 4I9/2 level fluorescence lifetime and β53 an effective branching ratio representing the percentage of ions from the 4S3/2 level (level 5 in Fig. 1) which ends up in the 4I9/2 level through radiative transitions and cross-relaxation processes [17]. One can notice that for concentrations below 500ppm, the T22 upconversion mechanism is expected to be very limited while for concentrations above 10000ppm the formation of clusters amid an uniform distribution of ions will further increase the impact of this upconversion process [19].Now, in the case of the pump-probe converted intensity (Ic), the main feeding mechanism for the 4I9/2 level is the excited state absorption of the IR probe, which promotes ions from the 4I11/2 level to the 4I9/2 level, and leads to:
with σIR the Er3+ excited-state absorption cross-section at 4.4µm, ΦIR the infrared photon flux and τ3 the 4I9/2 level fluorescence lifetime. Equation (5) does not take into account the T22 upconversion process since the modulation of the IR probe enables the rejection of the CW pump-related upconversion contribution.The Ic/Ip ratio then simply becomes equal to the ratio and by substituting n2 using Eq. (3), one can obtain:
Equation (6) clearly shows that the Ic/Ip ratio decreases with the pump power, more precisely in an inversely proportional fashion which is in agreement with the P−1 fitting of the Ic/Ip experimental ratio displayed in the inset of Fig. 5. However, Eq. (6) can be somewhat misleading by suggesting that the best situation in order to convert the IR signal into a near-infrared one would be to use a low pump power in order to maximize the Ic/Ip ratio. As emphasized by Eq. (5), the converted intensity (Ic) is proportional to the n2 population and therefore to the pumping power. Consequently, the best situation is to use a high pump power to increase the Ic level while filtering in an efficient manner the CW signal induced by Ip. However, by increasing the CW signal to very high values, one will also deteriorates the signal to noise ratio for the Ic signal simply because the initially negligible noise associated to the filtered CW signal will become larger and larger as the CW signal itself increases.The dependence of the Ic/Ip ratio with the dopant concentration is illustrated in Fig. 6 for two different Er3+ doping concentrations in Ga-Ge-Sb-S bulk glass samples (5000 ppm and 104 ppm). As noticed before, whether it is for bulk glass samples (Fig. 6) or fibers (Fig. 5), the Ic/Ip ratio is smaller than one confirming the large value of Ip compared to IC.
Fig. 6 Comparison of the Ic/Ip ratios in log scale for two Ga-Ge-Sb-S bulk samples with different Er3+ concentrations (5000 ppm and 10 000 ppm).
The decrease of the Ic/Ip ratio with the pump power in Fig. 6 for both Er3+ concentrations is, as expected, similar to the result obtained with the 1000ppm Er3+ doped sulfide fiber (Fig. 5). Moreover, for a given pump power the Ic/Ip ratio is larger for the lowest concentration (5000ppm). This result is consistent with Eq. (6) where the Ic/Ip ratio is inversely proportional to the doping concentration and is explained by the fact that Ic has a linear dependence with the dopant concentration (Eq. (5)) while Ip has a quadratic one because of the T22 upconversion mechanism (Eq. (4)). However, the comparison of Fig. 6 and Fig. 5 shows that the Ic/Ip ratio values are higher in bulk samples than in fibers while the Er3+ concentration is lower for the fibers. This apparently contradictory result is explained by the fact that the pump beam is more focused in the case of fibers in order to adequately pump the fiber while for bulk samples a less focused beam allows for a better overlap with the IR probe spot which is on the millimeter scale. As a consequence, the pump photon flux is higher in the case of fibers than for bulk samples, therefore decreasing the IR to near-infrared converted intensity.
Equation (6) despite being the result of a simplified approach is in agreement with experimental results in terms of pump power and dopant concentration dependences. One can notice that a more complete description of the Ic and Ip intensities would require the use of a complete set of rate equations associated with all the relevant Er3+ levels, which would, however, prevent the derivation of tractable analytical solutions such as Eq. (6).
4. Conclusion
We have reported for the first time, to the best of our knowledge, the conversion of an IR light signal around 4.4µm into an 810nm emission signal in both Er3+ doped chalcogenide fibers and bulk glass samples by simultaneously pumping them at 982 nm. The excitation at 982 nm promotes Er3+ ions into the 4I11/2 energy level while the IR to near-IR conversion is made possible by subsequent excited-state absorption from the 4I11/2 to the 4I9/2 level. The resulting conversion emission intensity (Ic) at 810nm from the 4I9/2 level therefore follows the IR light intensity. However, competing mechanisms such as excited absorption at the pump wavelength or more importantly upconversion by energy transfer among Er3+ ions in the 4I9/2 level also lead to an emission (Ip) from the 4I9/2 level. A simple rate equation based approach reveals that our experiment remains within a linear pumping regime and that the Ic/Ip ratio is inversely proportional to the dopant concentration and the pump power. This ratio is in all our experiments smaller than one showing that the unwanted emission Ip solely due to the pump light is larger than the converted emission. However, the use of a low pump power and a low dopant concentration in order to increase the Ic/Ip ratio, reduces the Ic converted intensity in itself. The best solution is therefore, as implemented in this work, to modulate the IR probe signal while keeping a CW excitation. The subsequent rejection of the Ip signal enables an accurate detection of the IR to near-infrared converted intensity. This converted signal at 810 nm precisely reflects the IR signal intensity which is transmitted through a gas cell within a gas sensor. The possibility to implement this photonic conversion in fibers instead of bulk materials is particularly appealing for the development of an all-optical remote sensor since the “conversion” chalocogenide fiber can be directly coupled to commercial silica fibers using standard fiber connectors [4] to transmit the 810nm signal over large distances enabling the development of an “all-optical” fiber based gas sensor.
Acknowledgments
This work has been funded by the ADEME STOCKCO2 program within the COPTIK project.
References and links
1. L. B. Shaw, B. Cole, P. A. Thielen, J. S. Sanghera, and I. D. Aggarwal, “Mid-wave IR and long-wave IR laser potential of rare-earth doped chalcogenide glass fiber,” IEEE J. Quantum Electron. 37(9), 1127–1137 (2001). [CrossRef]
2. T. Schweizer, B. N. Samson, R. C. Moore, D. W. Hewak, and D. N. Payne, “Rare-earth doped chalcogenide glass fibre laser,” Electron. Lett. 33(5), 414–416 (1997). [CrossRef]
3. B. J. Park, H. S. Seo, J. T. Ahn, Y. G. Choi, J. Heo, and W. J. Chung, “Dy(3 +) doped Ge-Ga-Sb-Se glasses and optical fibers for the mid-IR gain media,” J. Ceram. Soc. Jpn. 116(1358), 1087–1091 (2008). [CrossRef]
4. F. Starecki, F. Charpentier, J.-L. Doualan, L. Quetel, K. Michel, R. Chahal, J. Troles, B. Bureau, A. Braud, P. Camy, V. Moizan, and V. Nazabal, “Mid-IR optical sensor for CO2 detection based on fluorescence absorbance of Dy3+:Ga5Ge20Sb10S65 fibers,” Sens. Actuators B Chem. 207, 518–525 (2015). [CrossRef]
5. F. Charpentier, F. Starecki, J. L. Doualan, P. Jóvári, P. Camy, J. Troles, S. Belin, B. Bureau, and V. Nazabal, “Mid-IR luminescence of Dy3+ and Pr3+ doped Ga5Ge20Sb10S(Se)65 bulk glasses and fibers,” Mater. Lett. 101, 21–24 (2013). [CrossRef]
6. F. Charpentier, V. Nazabal, J. Troles, Q. Coulombier, L. Brilland, C. Boussard-Pledel, P. Nemec, H. Lhermite, J. Charrier, F. Smektala, M. Frumar, K. Le Pierres, N. Thybaud, B. Bureau, F. Baldini, and J. Homola, “Infrared optical sensor for CO2 detection,” in Optical Sensors, R.A. Lieberman (Eds.) (SPIE Press, 2009), 735610.
7. P. Camy, J.L. Doualan, and V.Nazabal, ”Chemical species optical sensor operating in infrared”, patent registered WO2011042628 A9.
8. N. Bloembergen, “Solid State Infrared Quantum Counters,” Phys. Rev. Lett. 2(3), 84–85 (1959). [CrossRef]
9. G. F. de Sa, P. A. Santa Cruz, and F. Auzel, “1.5μm to 0.65μm up-conversion by “summation of photons by energy transfer” effect in diode pumped vitroceramics,” J. Lumin. 31-32, 693–695 (1984). [CrossRef]
10. M. R. Brown and W. A. Shand, “Infrared Quantum Counter Action in Er-Doped Fluoride Lattices,” Phys. Rev. Lett. 12(13), 367–369 (1964). [CrossRef]
11. S. E. Stokowski, “Characteristics of a laser‐pumped 1.5‐μm infrared quantum counter,” J. Appl. Phys. 45(7), 2957 (1974). [CrossRef]
12. Yu. O. Barmenkov, A. V. Kir’yanov, A. D. Guzmán-Chávez, J. L. Cruz, and M. V. Andrés, “Excited-state absorption in erbium-doped silica fiber with simultaneous excitation at 977 and 1531 nm,” J. Appl. Phys. 106(8), 083108 (2009). [CrossRef]
13. J. Kobelke, J. Kirchhof, K. Schuster, and A. Schwuchow, “Effects of carbon, hydrocarbon and hydroxide impurities on praseodymium doped arsenic sulfide based glasses,” J. Non-Cryst. Solids 284(1-3), 123–127 (2001). [CrossRef]
14. S. A. Payne, L. L. Chase, L. K. Smith, W. L. Kway, and W. F. Krupke, “Infrared cross section measurements for crystals doped with Er3+, Tm3+and Ho3+,” IEEE J. Quantum Electron. 28(11), 2619–2630 (1992). [CrossRef]
15. V. Moizan, V. Nazabal, J. Troles, P. Houizot, J.-L. Adam, J.-L. Doualan, R. Moncorgé, F. Smektala, G. Gadret, S. Pitois, and G. Canat, “Er3+-doped GeGaSbS glasses for mid-IR fibre laser application: Synthesis and rare earth spectroscopy,” Opt. Mater. 31(1), 39–46 (2008). [CrossRef]
16. K. Kadono, T. Yazawa, S. Jiang, J. Porque, B.-C. Hwang, and N. Peyghambarian, “Rate equation analysis and energy transfer of Er3+-doped Ga2S3-GeS2-La2S3 glasses,” J. Non-Cryst. Solids 331(1–3), 79–90 (2003). [CrossRef]
17. B. Whualkuer Lozano, C. B. de Araújo, Y. Ledemi, and Y. Messaddeq, “Upconversion luminescence in Er3+ doped Ga10Ge25S65 glass and glass-ceramic excited in the near-infrared,” J. Appl. Phys. 113(8), 083520 (2013). [CrossRef]
18. C. J. da Silva and M. T. de Araujo, “Thermal effect on upconversion fluorescence emission in Er3+-doped chalcogenide glasses under anti-Stokes, Stokes and resonant excitation,” Opt. Mater. 22(3), 275–282 (2003). [CrossRef]
19. D. J. Coleman, P. Golding, T. A. King, and S. D. Jackson, “Spectroscopic and energy-transfer parameters for Er3+-doped and Er3+, Pr3+-codoped GeGaS glasses,” JOSA B 19(9), 1982–1989 (2002). [CrossRef]
