We report on a spectroscopic analysis of ErCl3 and 1% Er3+:YCl3 to determine their potential as possible laser sources at 3.5 and 4.5 μm. Concentration quenching of the low lying excited states is reported to be surprisingly very weak in this system. Although some shortening of the lifetimes is measured in the fully concentrated system, they retain lifetimes that are of order several milliseconds or more. A Judd-Ofelt analysis is performed and the projected gains for the 3.5 and 4.5 μm transitions are calculated. Successful growth techniques of erbium doped chlorides are also described.
©1997 Optical Society of America
Developments in remote sensing and counter measures applications require the generation of efficient mid-infrared (mid-IR) sources in the 3.5 μm – 5 μm spectral range, over which an atmospheric window exists. The low lying manifolds of rare earth ions are excellent sources of mid-IR emission provided they are not multiphonon quenched. It has been shown that certain chloride materials in which the maximum allowed phonon frequencies are small compared to these transitions, make good hosts for rare earth ions. In these low phonon materials, transitions that are normally quenched by multiphonon decay tend to radiate more efficiently in the infrared. For example, laser action has recently been demonstrated at room temperature in Pr:LaCl3 crystals at 7.2 μm, and just below room temperature at 5.2 μm. [1,2] Although praseodymium has several mid-infrared transitions within the above spectral range, an efficient pumping scheme has not yet been demonstrated.
In this work, we examine trivalent erbium doped chlorides, specifically ErCl3 and 1% Er3+:YCl3. These low phonon materials should provide the necessary quantum efficiency and energy storage lifetimes for efficient mid-IR emission, while simultaneously eliminating some of the problems of multiphonon quenching. The pertinent transitions are the 4F9/2 → 4I9/2 (3.5 μm) and 4I9/2 → 4I11/2 (4.5 μm) as shown in Figure 1. Both transitions are readily pumped at 660 nm and 800 nm using conventional solid state lasers. The 4I9/2 → 4I11/2 transition has an energy gap of approximately 2200 cm-1, while the YCl3 crystal has a maximum phonon frequency of about 200 cm-1. Therefore, a multiphonon decay process would be about 11th order, resulting in multiphonon decay rates around a Hz or less. [3,4] In comparison, a LiYF crystal has a maximum phonon frequency of ~ 400 cm-1 making multiphonon decay approximately a 5th order process and a much more dominant loss mechanism. In addition to being a low phonon host material, the YCl3 crystal field weakly splits the Stark components of the Er3+ ion, effectively increasing the cross sections to these states.
While the 3.5 μm and 4.5 μm transitions have shown laser action in the past, no efficient, room temperature laser has been demonstrated. [5–7] Problems with multiphonon quenching and concentration quenching have inhibited the efficiency of these transitions in hosts such as ZBLAN glass fibers and LiYF. Most systems exhibit some amount of energy transfer between dopant ions as concentration is increased, manifesting itself as an additional loss or gain mechanism in the excited state populations. Nonlinear energy transfer can significantly shorten the excited state lifetimes, and can critically inhibit the efficiency of laser transitions. It has been found, however, that the Er3+:YCl3 system exhibits very weak concentration quenching at 100% dopant levels, indicating the possibility of concentration scaling of the absorption coefficient, and hence the gain. We report here on the growth and spectroscopy of ErCl3 and Er3+:YCl3 crystals to assess their potential as mid-IR laser materials. A Judd-Ofelt analysis leading to gain calculations is provided along with a discussion of the effects of concentration on the radiative decay rates. A short discussion of alternate hosts is also provided.
2. Crystal growth
It has been found that any growth using ErCl3 necessitated rigorous processing procedures due to the strong hygroscopic nature and reactivity of ErCl3, as well as variations in the quality of starting material. All handling had to be done in a dry chamber purged with N2 boil off from a liquid nitrogen dewar. Most importantly, it should be noted that raising the temperature of the material to above ~300 °C under vacuum or in air, results in the removal of chlorine gas and the subsequent reduction of Er3+ to a lower oxidation state. The remaining material would then contain large quantities of black, insoluble particulates having a much higher melting point than ErCl3. A flame test confirmed that the particulates contained erbium, and chlorine was detected in the vacuum system after pumping on the dry powder at temperatures above 300 °C. Raising the temperature and melting the material under a light Cl2 atmosphere (~20% atm at room temperature) was found to prevent this reaction.
Crystals were grown by a vertical Bridgman-Stockbarger method. All processing and growths took place in a low -OH quartz ampoule, which had been cleaned and baked out prior to the processing of starting material. The growth region of the ampoule had a tapered point to initiate seeding of the crystal. Growths were carried out under an atmosphere of Cl2 at the melting point of the material. Pulling rates were nominally 1 mm/hr with a zone gradient of ~10 °C/cm at the transition point.
Typical growth procedures are as follows. Powdered ErCl3 is baked out under vacuum at 75° C for 24 hours to remove any excess water vapor. The dry material is then melted, under pure Cl2 gas and allowed to filter through a silica wool filter and a 200-micron tapered quartz aperture into another tapered region of the ampoule. Filtering is necessary to remove any particulates and unconverted oxychlorides that may have been present in the starting material. In addition, the growth ampoule has a pyrolytic carbon coating to prevent reaction of the material with the wall of the tube. Such reactions were found to result in spurious nucleations with the possibility of additional oxychloride formation, which can hinder the growth of single crystals and increase scattering centers. Once the melt has filtered, the growth region is raised into the top of the furnace for subsequent lowering through the gradient. The resulting single crystals are scatter free, with sections approximately 3–5 mm thick and 1.5 cm wide. Photographs of cleaved pieces of ErCl3 and 1% Er3+:YCl3 are shown in Figure 2. Note that the cleavage planes are approximately 30 degrees to the growth axis.
Initially, erbium doped LaCl3 was considered for this investigation. However, it was found that the LaCl3 crystals did not incorporate Er3+ well into the matrix, possibly due to the differing crystal structures. LaCl3 is hexagonal with a C3h space group, while ErCl3 is monoclinic with space group C2/m. Wet chemical analysis of a 5% doped Er3+:LaCl3 indicated that only 0.2% of the Er3+ was incorporated into the host. Alternatively, Er3+ does incorporate well into a YCl3 matrix with no observable stratification of species. The YCl3 crystal structure is monoclinic with space group C2/m. Consequently, our study focuses on ErCl3 and Er3+:YCl3 for spectroscopic analysis and calculation.
3. Spectroscopic experiments
Absorption spectra, fluorescence spectra, and fluorescence lifetime measurements were taken on both the fully concentrated ErCl3 and the 1% Er3+:YCl3 samples. Single crystal samples of ErCl3 and 1% Er3+:YCl3 were cleaved and mounted in a CaF2 cell using CCl4 as an index matching fluid. CCl4 was chosen due to its flat absorption spectra in the mid-IR and near-IR spectral ranges, as well as its chemical inertness to ErCl3.
Unpolarized room temperature absorption spectra for the 4I15/2 → 4I13/2, 4I11/2 , 4I9/2 , and 4F9/2 transitions in ErCl3 were taken on a Nicolet Model 750 Magna IR spectrometer, and are shown in Figure 3. Strong absorption lines at 660 nm and 800 nm suitable for pumping of the 4F9/2 and 4I9/2 states are visible.
The fluorescence spectra were obtained by pumping the 4F9/2 and the 4I9/2 at 660 nm and 800 nm respectively. The pump sources were a short pulsed Ti:Sapphire laser (800 nm) and a Q-switched Nd:YAG pumped dye laser (660 nm at 5 ns pulse width). The fluorescence spectra from 3 μm – 5 μm of were collected using an InSb detector and notch filters at 0.1 μm increments. Unpolarized room temperature fluorescence spectra for the 4F9/2 → 4I9/2 (3 μm – 4 μm) and 4I9/2 → 4I11/2 (4 μm – 5 μm) transitions are shown in Figure 4. The 4F9/2 → 4I9/2 transition peaks at 3.7 μm. The 4I9/2 → 4I11/2 transition peaks at 4.3 μm although strong fluorescence is present at 4.5 μm.
Fluorescence lifetime experiments were conducted using a tunable, pulsed IR parametric oscillator driven by a Q-switched Nd:YAG laser. Several millijoule pulses of duration 4 ns at 800 nm were used to excite the 4I15/2 → 4I9/2 transition. The 4.5 μm fluorescence (4I9/2 → 4I11/2) was collected using a 4 μm long pass filter and an InSb detector. The 940 nm fluorescence (4I11/2 → 4I15/2) was obtained using a silicon detector and a long pass 930 nm filter. The 1.5 μm fluorescence (4I13/2 → 4I15/2) was collected using a 1.4 μm long pass filter, a red filter, a notch filter to block 800 nm, and an InAs detector. Both the 940 nm and the 1.5 μm fluorescence transients exhibited the expected rise time associated with the decay of the 4I9/2 pump level. The dye laser source at 660 nm was used to pump the 4I15/2 → 4F9/2 transition in the ErCl3 crystal, and the 3.5 μm fluorescence (4F9/2 → 4I9/2) collected using an InSb detector and notch filters. To within the noise of the data, all 4I multiplets decayed exponentially in this pump regime. Only the 4F9/2 level exhibited a nonexponential lifetime with a fast initial decay and a slower exponential tail. The lifetimes are shown in Table 1.
A Judd-Ofelt analysis was performed on both samples in order to calculate the cross sections of the pertinent transitions and predict the lifetimes for use in characterization of the system. [8,9] The electric dipole linestrengths, , for the four transitions represented in Figure 1 were calculated from absorption spectra by
where k(λ) is the absorption coefficient at wavelength λ,λ̅ is the mean wavelength associated with the transition, n is the index of refraction at the mean wavelength, 2J+1 is the number of levels in the upper J manifold and N is the rare-earth ion density. is the magnetic dipole linestrength which was calculated from intermediate coupled wavefunctions. The electric dipole linestrengths were fit by a least squares fit to
From the Judd-Ofelt parameters, linestrengths were calculated for all transitions using Eq. 2, and the intermanifold spontaneous emission rates AJJ’ determined from
The branching ratios, , were then determined from these values. The effective cross section, Σ, which is the stimulated emission cross section integrated over the total band, was calculated from
Table 3 lists the results of these calculations for the ErCl3 sample.
Table 4 lists the values for the 1% Er3+:YCl3 obtained in a similar analysis as that outlined above. However, note that due to the weak signal of the absorption spectra for the 1% Er3+:YCl3 sample, the Judd-Ofelt parameters were found from the least squares fit of the measured linestrengths to Eq. 2 as well as the measured lifetimes to Eq. 3. As these transients were observed to be exponential in nature, this procedure is justified.
The results of the radiative calculations, and the experimentally observed lifetimes are tabulated for ErCl3 and 1% Er3+:YCl3 in Table 5. Recall that the multiphonon rates are expected to be negligible in this system, and are therefore not calculated.
Comparison of the experimentally measured lifetimes and the calculated radiative lifetimes of the ErCl3 crystal in Table 4 shows that, in general, the lifetimes are in agreement (~ a factor of 2). Sources of error include noise in the fluorescence transients, and the use of only four transitions in the absorption spectra to calculate the Judd-Ofelt parameters. The measured lifetimes of the 1% Er3+:YCl3 are also in reasonable agreement with the calculated lifetimes. In addition to the errors describe for the fully concentrated system, other errors in the 1% sample stem from the low concentration and hence weak absorption spectra needed for the calculation. Although no stratification of the dopant material was observed, it is usual for some degree of concentration gradient to manifest itself in the melt.  Hence, the exact concentration of the crystal is uncertain to approximately 15 % based on previous systems where a wet chemical analysis was performed.
One of the most interesting results of this investigation is that the concentration can be varied from 1% to 100% without significantly affecting the lifetimes, as evidenced in Table 5. As mentioned earlier, only the 4F9/2 level in the fully concentrated system exhibited a nonexponential transient after 660 nm excitation. This has previously been attributed to upconversion to the 4S3/2 level in other host materials, and could be a possible explanation here. [5,7,12] It should be noted that upconversion of the 4I11/2 and the 4I13/2 has been investigated in other high phonon host materials such as LiYF4 , YAG, and some glass structures.  Although some green fluorescence was visible under both 800 nm and 660 nm excitation, none of the 4I fluorescence transients show any evidence for strong nonlinear effects suggesting the process is very weak in this system. Future investigations of the origin of the green fluorescence are warranted. Although the 1% sample had very weak absorption and would be insufficient for laser emission, the absence of any strong nonlinear energy transfer in both samples implies that the dopant concentration in YCl3 can be varied continuously up to 100% without concentration quenching, enabling concentration scaling of the absorption coefficient.
Coincident with concentration scaling of the absorption coefficient comes scaling of the small signal gain. Using the above data, one can estimate the gain coefficient for the 4.5 μm transition in ErCl3. The gain coefficient is given by
where I is the pump intensity at frequency ν . A moderate pump intensity at 800 nm would be 0.5 W in a 100 μm spot size, with typical pulse widths of order τ = 1 ms. The absorption coefficient, evaluated from figure 1, is 15 cm-1 at 800 nm. From the fluorescence spectra of Figure 4, we can determine an effective linewidth, ∆νeff , for the 4I9/2 → 4I11/2 transition to be ≈; 200 cm-1, and Table 3 gives the integrated emission cross section Σ to be 0.23 × 10-18 cm. Using these parameters, we calculate a gain coefficient of 0.13 cm-1 for the 4.5 μm transition. Note that due to the low resolution fluorescence data of Figure 4, the effective linewidth, ∆νeff , is larger than the actual linewidth, and hence the gain coefficient is underestimated. The absorption coefficient of 15 cm-1 at 800 nm suggests that a crystal 2 mm thick would absorb approximately 90% of the pump intensity, assuming no other losses. This suggests a 2% gain per pass at 4.5 μm. The long lifetime and respectable gain of the 4I9/2 level coupled with the exponential decay in this system makes this an attractive transition for a 4.5 μm laser. However, without any nonlinear mechanisms in the lower states, the system would suffer from self termination, making CW operation unlikely.  The advantage is that the short crystal lengths required for high absorption with moderate gain would allow a compact laser system to be constructed, readily pumped with commercial laser diodes.
Now consider the 3.5 μm transition in ErCl3. The absorption coefficient at 660 nm is 65 cm-1, while the integrated emission cross section Σ is also 0.23 × 10-18 cm. From Figure 4, we find ∆νeff ≈ 120 cm-1 for the 4F9/2 → 4I9/2 transition. Using a pulse width of τ = 0.15 ms and the same pump intensity as before, we calculate a gain of 0.12 cm-1 for the 3.5 μm transition (4F9/2 → 4I9/2). For this transition, even thinner crystals (0.5 mm) can effectively absorb the pump beam, allowing very compact systems with commercial laser diodes.
Laser emission of the Er3+ 4F9/2 → 4I9/2 transition has been previously reported in LiYF, but in that system multiphonon quenching inhibited the laser efficiency.  The lifetimes reported for the 4F9/2 and the 4I9/2 levels were 60 μs and 6 μs at room temperature, respectively. In addition to the multiphonon quenching in LiYF, excited state absorption (ESA) at the laser wavelength was also speculated to occur from the 4F9/2 to the 4S3/2 level due to an overlap with the 4F9/2 → 4I9/2 emission. The ErCl3 system has the advantage of a blue shifted 4F9/2 → 4S3/2 absorption and a red shifted 4F9/2 → 4I9/2 emission, making this ESA process nonresonant. Although some nonlinear mechanism was observed in the 4F9/2 → 4I9/2 fluorescence, the lifetime is long enough in ErCl3 (~ 200 μs) to allow efficient lasing. Determining the nature of the 4F9/2 nonlinearity in ErCl3 is a goal for future work.
Apart from single laser transitions, several modes of operation are evident from Figure 1 and the lifetimes in Table 5. Due to the absence of multiphonon decay between manifolds, as well as any apparent energy transfer, cascade lasing may be possible. Pumping at 660 nm and lasing at 3.5 μm could result in the sequential lasing of the lower two transitions (at 4.5 μm and 2.9 μm) provided the nonlinearity in the 4F9/2 is weak enough and not a loss mechanism. Similarly, pumping at 800 nm could result in 4.5 μm and 2.9 μm lasing. As can be seen from Table 3, the 4I11/2 → 4I15/2 transition has a higher integrated cross section than either the 3.5 μm or 4.5 μm transitions favoring stimulated emission.
As a final note, we mention that lightly doped (≪1 %) erbium chalcogenide glasses have also been investigated in related work.  We have measured the lifetimes of the 4F9/2 level to be 0.1 ms in lightly doped BaInGeGaS and GeAsGaS glasses, while the 4I9/2 lifetime was measured at 0.6 ms in both glasses. These two systems could provide alternative hosts for Er3+. Although these materials are easier to grow and are not hygroscopic, they have several disadvantages. The most detrimental problem of these glasses is that they do not incorporate Er3+ well into their structure. This prevents concentrations higher than a fraction of a percent from being doped into the material, necessitating the growth of longer samples. However, in a related problem, typical samples exhibit large numbers of scattering centers preventing long, high quality samples from being grown. In addition, these glasses readily form S-H bonds, which have extrinsic absorption at 4 μm, overlapping the fluorescence of the trivalent erbium. It is apparent that although these glasses show some promise as low phonon hosts, there are several problems to overcome.
We have demonstrated growth of laser quality single crystals of ErCl3 and 1% Er3+:YCl3. In performing the spectroscopic analysis, it was found that nonlinear energy transfer was very weak or absent in both the 1% and 100% doped systems, resulting in only a slight shortening of the lifetimes. The low lying excited states of ErCl3 were found to exhibit lifetimes on the order of several hundred microseconds to several milliseconds suitable for pulsed laser oscillation at 3.5 μm and 4.5 μm, with the possibility of cascade lasing. Due to the high dopant concentrations possible in this system, respectable gains and cross sections, laser devices operating at these wavelengths could be made compact, requiring short interaction lengths for high gain.
The authors would like to acknowledge the support of the Office of Naval Research. This work was performed while S. Kirkpatrick held a National Research Council - U. S. Naval Research Laboratory research associateship. The work of J. Ganem was partially supported by the Cottrell College Science Award of research Corporation and acknowledgment is made to the donors of the Petroleum Research Fund, administered by the ACS, for additional support of this research.
References and links
1. S. R. Bowman, J. Ganem, B. J. Feldman, and A. W. Kueny, “Infrared laser characteristics of praseodymium-doped lanthanum trichloride,” IEEE J. Quantum Electron. 30, 2925–2928, (1994). [CrossRef]
2. S. R. Bowman, L. B. Shaw, B. J. Feldman, and J. Ganem, “A 7-μm Praseodymium-Based Solid State Laser,” IEEE J. Quantum Electron. , 32, 646–649 (1996). [CrossRef]
3. A. Riseberg and M. J. Weber, “Relaxation phenomena in rare earth luminescence,” in Progress in Optics. New York: North Holland, 14 (1976).
4. W. Moos, “Spectroscopic relaxation processes of rare earth ions in crystals,” J. Lumin. 1, 106–112, (1968). [CrossRef]
5. J. F. Pinto, G. H. Rosenblatt, and L. Esterowitz, ”Continuous-wave laser action in Er 3+:YLF at 3.41 μm,” Electron. Lett. , 30, 1596–1598 (1994). [CrossRef]
6. A. A. Kaminskii Izv. Akad. Nauk SSSR, Ser. Fiz., 45, 348 (1981).
7. H. Tobben, “Room Temperature CW Fiber Laser at 3.5 μm in Er3+ Doped ZBLAN Glass,” Electron. Lett. , 28, 1361–1362 (1992). [CrossRef]
8. B. R. Judd, “Optical absorption intensities in rare earth ions,” Phys Rev. 127, 750–751 (1963). [CrossRef]
9. G. S. Ofelt, “Intensities of crystal spectra of rare earth ions,” J. Chem. Phys. 37, 511–520 (1963). [CrossRef]
10. M. J. Weber, “Probabilities for Radiative and Nonradiative Decay of Er3+ in LaF3,” Phys. Rev. 157 (1967). [CrossRef]
11. B. P. Sobolev, in Growth of Crystals, edited by E. I. Givargizov and S. A. Grinberg, (Plenum, New York, 1992), vol. 18, pp.197-211. [CrossRef]
12. Pollnau, W. Lüthy, and H. P. Weber, ”Influence of Normal and Inverse Upconversion Processes on the Continuous Wave Operation of the Er3+ 3 mm Crystal Laser,” in Advanced Solid State Lasers, Vol. 20 of OSA Proceedings Series (Optical Society of America, Washington, D.C., 1994), p 163–167.
13. Bogdanov, W. E. K. Gibbs, D. J. Booth, J. S. Javornickzky, P. J. Newman, and D. R MacFarlane, “Fluorescence from highly-doped erbium fluorozirconate glasses pumped at 800 nm,” Optics Commun. 132, 73–76, (1996). [CrossRef]
14. L. B. Shaw, D. Schaasma, J. Moon, B. Harbison, J. Sanghera, and I. D. Aggarwal, “Evaluation of the IR Transitions in Rare Earth Doped Chalcogenide Glasses,” in Conference on Lasers and Electro-Optics, Vol 11 of 1997 OSA Technical Digest Series, (Optical Society of America, Washington, D.C., 1997), p.255.