Energy-level structure and spectral analysis of Nd 3+ 4f 3 in polycrystalline ceramic garnet Y 3 Al 5 O 12

Transcription

1 JOURNAL OF APPLIED PHYSICS VOLUME 96, NUMBER 6 15 SEPTEMBER 2004 Energy-level structure and spectral analysis of Nd 3+ 4f 3 in polycrystalline ceramic garnet Y 3 Al 5 O 12 John B. Gruber a) Department of Physics, San José State University, San Jose, California Dhiraj K. Sardar and Raylon M. Yow Department of Physics and Astronomy, University of Texas at San Antonio, 6900 North Loop 1604, San Antonio, Texas Toomas H. Allik Science Applications 1nternational Corporation, P.O. Box 632, Fort Belvoir, Virginia Bahram Zandi ARL/Adelphi Laboratory Center, 2800 Powder Mill Road, Adelphi, Maryland (Received 23 April 2004; accepted 1 June 2004) A detailed crystal-field splitting analysis is given for the 26 lowest-energy multiplet manifolds, 2S+1 L J,ofNd 3+ 4f 3 in polycrystalline ceramic garnet Y 3 Al 5 O 12 (YAG). The absorption spectra obtained between 8 K and room temperature, and between 1750 and 350 nm, and the fluorescence spectrum obtained at 8 K and observed between 1450 and 875 nm are analyzed for transitions between individual energy (Stark) levels that characterize the energy-level structure of Nd 3+ ions in D 2 symmetry sites, replacing Y 3+ ions in the garnet host lattice. A model Hamiltonian including atomic and crystal-field terms is diagonalized within the complete 4f 3 SLJM J basis set which includes 364 states. The calculated splitting of the Nd 3+ energy levels by the crystal field is compared with the experimental splitting observed in both the ceramic sample and a single-crystal laser rod. Both samples have approximately the same Nd 3+ concentration, about 1 at. %. By varying the atomic and crystal-field parameters, we obtain a standard deviation of 18 cm 1 between 106 calculated-to-observed Stark levels found between the ground state and the 4 D 1/2 at cm 1. Within this standard deviation the energy-level structure of Nd 3+ is found to be similar in both samples. Low temperature visible and near IR spectra of ceramic Nd 3+ :YAG show it has comparable properties to the crystalline analog American Institute of Physics. [DOI: / ] I. INTRODUCTION The synthesis of polycrystalline ceramic garnets doped with rare earth ions R 3+ as an aggregate of nanocrystalline grains, each randomly oriented with respect to neighboring grains, has drawn attention from numerous investigators, both in terms of fundamental spectroscopic studies now possible, as well as the potential for device applications. 1 6 The R 3+ -doped polycrystalline ceramic Y 3 Al 5 O 12 (YAG) appears to have optical, thermal, and physical properties comparable to single-crystal rods currently in use in numerous laser applications. The Czochralski method of crystal growth has been applied for years in the fabrication of Nd 3+ : YAG, but the rod and slab dimensions have been limited due to the unusable central core of the boule and the strain spokes that radiate to the surface. The laser design engineer has been constrained to Nd 3+ : YAG geometries that can be mined from crystalline boules. However, the ceramic form has significant potential since it can be fabricated to unique and large strainfree dimensions that can be less expensive to fabricate in large quantities of polished and coated geometries. Additionally, ceramic YAG has better doping profiles for the R 3+ ions a) Author to whom correspondence should be addressed; electronic mail: jbgruber@ .sjsu.edu than its single-crystal counterparts. 4 7 The scattering losses are low due to a pore volume concentration of about 1 ppm and grain size of about 10 m diameter as determined from electron microscopy measurements. 4,6,7 Sample fabrication and characterization of some of the physical properties of ceramic Nd 3+ : YAG are described by Lu et al. 3,4 and by Kaminskii et al. 5 Their spectroscopic investigations centered around that part of the Nd 3+ spectrum which is associated with stimulated emission observed at m in the ceramic sample. In the present study, we expand upon those measurements to include a detailed crystal-field splitting analysis of the 26 lowest-energy multiplet manifolds, 2S+1 L J of Nd 3+ 4f 3 in ceramic YAG. These multiplets contain 109 individual energy (Stark) levels observed between the ground-state multiplet manifold, 4 I 9/2, and the 4 D 1/2 multiplet at nm cm 1. The observed energy-level structure of Nd 3+ in the ceramic sample is compared to the experimental Stark levels reported by Kaminskii 8 and by Gruber et al. 9 for Nd 3+ : YAG as large single-crystal laser rods. The calculated energy-level structure of the Nd 3+ 4f 3 ions that replace Y 3+ ions in sites of D 2 symmetry in the lattice, was determined by using a model Hamiltonian that consists of Coulombic, spin orbit, and interconfigurational /2004/96(6)/3050/7/$ American Institute of Physics

2 J. Appl. Phys., Vol. 96, No. 6, 15 September 2004 Gruber et al TABLE I. Absorption spectrum of Nd 3+ 4f 3 in nanocrystalline ceramic garnet. a 2S+1 L J b nm c d Trans. e ceramic E cm 1 f E cm 1 g calc. E cm 1 h crystal 4 I 15/ Z 1 W Z 1 W Z 1 W Z 1 W Z 1 W Z 1 W Z 1 W Z 1 W F 3/ Z 1 R Z 1 R F 5/ Z 1 S Z 1 S Z 1 S H 9/ Z 1 S Z 1 S Z 1 S Z 1 S Z 1 S F 7/ Z 1 A Z 1 A S 3/ Z 1 A Z 1 A Z 1 A Z 1 A F 9/ Z 1 B Z 1 B Z 1 B Z 1 B Z 1 B H 11/ Z 1 C Z 1 C Z 1 C Z 1 C Z 1 C Z 1 C G 5/ Z 1 D Z 1 D Z 1 D G 7/ Z 1 E Z 1 E Z 1 E Z 1 E G 7/ Z 1 F Z 1 F Z 1 F Z 1 F K 13/ Z 1 G Z 1 G Z 1 G Z 1 G Z 1 G

3 3052 J. Appl. Phys., Vol. 96, No. 6, 15 September 2004 Gruber et al. TABLE I. (Continued.) 2S+1 L J b nm c d Trans. e ceramic E cm 1 f E cm 1 g calc. E cm 1 h crystal Z 1 G Z 1 G G 9/ Z 1 H Z 1 H Z 1 H Z 1 H Z 1 H G 9/ Z 1 I Z 1 I Z 1 I Z 1 I Z 1 I D 3/ Z 1 J Z 1 J G 11/ Z 1 K Z 1 K Z 1 K Z 1 K Z 1 K Z 1 K K 15/ Z 1 L Z 1 L Z 1 L Z 1 L Z 1 L Z 1 L Z 1 L Z 1 L P 1/ Z 1 M D 5/ Z 1 N Z 1 N Z 1 N P 3/ Z 1 O sh 0.12 Z 1 O D 3/ Z 1 P Z 1 P D 5/ Z 1 Q Z 1 Q Z 1 Q D 1/ Z 1 Q a Sample temperature is nominally 8 K; garnet is Y 3 Al 5 O 12. b Multiplet manifold of Nd 3+ 4f 3 ; number in parentheses is experimental centroid. c Wavelength in nanometers. d Absorption coefficient in cm 1. e Transitions between Stark levels using labels from Ref. 10. f Energy of Stark level in cm 1. g Calculated splitting using parameters given in Table III. h Reference 9. electronic interaction terms associated with the free ion Nd 3+ and crystal-field terms that split each multiplet, 4f 3 2S+1 L J into J+1/2 energy (Stark) levels. Each Stark level is twofold degenerate for Nd 3+ (a Kramers ion) in a crystal field of D 2 symmetry. The model Hamiltonian was diagonalized within the complete 4f 3 electronic configuration that includes a SLJM J basis set of 364 states. 9 The calculated Stark levels that are obtained are compared to the Nd 3+ Stark levels observed in the ceramic sample and in the single-crystal laser rod. Within a standard deviation of 18 cm 1 based on a fitting of the same 106 experimental-to-calculated levels, the detailed energy-level structure in both samples is equivalent. II. EXPERIMENTAL DETAILS A spectroscopic sample of polycrystalline ceramic Nd 3+ : YAG was obtained from Baikowski International Corpora-

4 J. Appl. Phys., Vol. 96, No. 6, 15 September 2004 Gruber et al FIG. 1. Absorption spectrum of the 4 I 15/2 manifold as transitions Z 1 W 1 8 obtained at 8 K and 300 K for the ceramic sample and at 8 K for the crystal rod sample. FIG. 3. Absorption spectrum of the 2 H 9/2 and 4 F 5/2 manifolds as transitions Z 1 S 1 5 and Z 1 S 1 3, respectively, for the ceramic sample at 8 and 300 K, and at 8 K for the crystal rod sample. tion, Charlotte, NC. The sample disc was 7.21 mm in diameter and 1.71 mm thick. The Nd dopant concentration was quoted as 1.00 at. %. The supplier certified that the sample had no concentration gradient or impurities such as Cr 4+,or other R 3+ ions. Our detailed spectroscopic investigations confirmed these statements. Absorption spectra were obtained between 1750 and 350 nm at nominal temperatures between 8 K and room temperature using an upgraded Cary model 14 R spectrophotometer controlled by a desktop computer. All spectra were taken at 0.2 nm intervals. The spectral bandwidth was about 0.5 nm for measurements above 1200 nm, and 0.05 nm for measurements between 350 and 1200 nm. The Cary model 14 R spectrophotometer is equipped with two light sources and two detectors. A deuterium lamp is used for wavelengths below 350 nm, and a tungsten lamp for wavelengths above 350 nm. A photomultiplier tube is used for wavelengths below 900 nm with a fixed high voltage of 400 V, and a PbS detector is used for wavelengths above 900 nm with a fixed biasing voltage of 6.4 V. The slitwidth, therefore, automatically adjusts depending on the light intensity, thereby giving rise to the variable bandwidths in different spectral ranges. Fluorescence spectra obtained between 800 and 1500 nm at 8 K were obtained by exciting the sample with a nm laser emission from a Spectra Physics Model argon-ion laser. The emission bandwidth was 0.12 nm and the maximum laser power was 2 W. The spectra were analyzed by SPEX Model 340 E monochromator equipped with a liquid nitrogen cooled germanium detector. Spectral resolution was better than 0.8 nm by using a reflection grating with 600 grooves mm blazed at 1 m. The sample was mounted on the cold finger of a CTI Model 22 closed-cycle helium cryogenic refrigerator that provided temperatures between 8 and 300 K. The sample temperature was monitored with a silicon-diode sensor attached to the base of the sample holder and maintained by a Lake Shore control unit. Sample temperatures are likely several degrees higher than reported and thus represent a nominal temperature quoted for the experiment described. FIG. 2. Absorption spectrum of the 4 F 3/2 manifold is shown for the ceramic sample at 8 K; at 300 K temperature-dependent transitions are observed from excited Stark levels in the 4 I 9/2 manifold to 4 F 3/2 (ceramic sample); the 8 K crystal rod spectrum is shown for comparison purposes. FIG. 4. Absorption spectrum of the 4 F 7/2 and 4 S 3/2 manifolds as transitions Z 1 A 1 4 and Z 1 A 1,2, respectively for the ceramic sample at 8 and 300 K, and at 8 K for the crystal rod sample.

5 3054 J. Appl. Phys., Vol. 96, No. 6, 15 September 2004 Gruber et al. FIG. 5. Absorption spectrum of the 2 P 1/2 multiplet manifold at 8 and 300 K in the ceramic sample; temperature-dependent transitions Z 2 M 1,Z 3 M 1,andZ 4 M 1 are observed in 300 K ceramic spectrum; the 8 K absorption spectrum for the crystal rod sample is also shown. III. OBSERVED SPECTRA Table I lists all the absorption that is observed at 8 K between 1750 and 350 nm for Nd 3+ in polycrystalline ceramic YAG. Twenty-three multiplet manifolds, 2S+1 L J, are represented between 4 I 15/2 at the longest wavelengths and 4 D 1/2 at the shortest wavelength listed in the table. Of the 91 FIG. 6. The 8 K fluorescence spectrum from 4 F 3/2 R 1 to 4 I 9/2 Z 1 Z 5 in the ceramic sample. Stark levels associated with these manifolds, all but four are identified by transitions from the ground-state Stark level Z 1 to excited Stark levels. The remaining four Stark levels are associated with transitions presumed to be very weak according to calculations involving the same transitions in singlecrystal Nd 3+ :YAG reported earlier. 9 Many of the strongest absorption peaks within manifolds 4 F 5/2 809 nm 2 H 9/2 795 nm, 4 F 7/2, 4 S 3/2 736 nm, 2 G 7/2 569 nm, and 4 G 7/2 527 nm exhibit satellite structure, similar to that observed TABLE II. Fluorescence spectrum 8 K from 4 F 3/2 of Nd 3+ 4f 3 in nanocrystalline ceramic garnet a. Term. 2S+1 L J b manifold Trans. c nm d I e E cm 1 f E cm 1 g calc. E cm 1 h E cm 1 i crystal 4 I 9/2 R 1 Z R 1 Z R 1 Z R 1 Z R 1 Z I 11/2 R 1 Y R 1 Y R 1 Y R 1 Y R 1 Y R 1 Y I 13/2 R 1 X R 1 X R 1 X R 1 X R 1 X R 1 X sh R 1 X a Emitting Stark level in 4 F 3/2 manifold is cm 1 in absorption at 8 K. b Terminal multiplet manifold; experimental centroid is in parentheses. c Transitions from R 1 (Table I) to Z n,y n, and X n. d Wavelength in nanometers. e Relative intensity within a manifold with 10.0 as maximum. f Energy of transition in cm 1. g Energy difference relative to cm 1 ; taken as the energy of the Stark level in the designated multiplet manifold. h Calculated splitting based on parameters given in Table III. i Reference 9.

6 J. Appl. Phys., Vol. 96, No. 6, 15 September 2004 Gruber et al FIG. 7. The 8 K fluorescence spectrum from 4 F 3/2 R 1 to 4 I 11/2 Y 1 Y 6 in the ceramic sample. FIG. 8. The 8 K fluorescence spectrum from 4 F 3/2 R 1 to 4 I 13/2 X 1 X 7 in the ceramic sample. for Nd 3+ : YAG in single-crystal rods. These satellites have been interpreted in the past as due to Nd 3+ ions in sites associated with defects, that arise during crystal growth. 9,11 Figures 1 5 compare the 8 K absorption spectrum in the ceramic sample with the 8 K spectrum of Nd 3+ in a singlecrystal rod with both samples having approximately the same Nd concentration (about 1 at. %). The multiplet manifolds of Nd 3+ include 4 I 15/2 (Fig. 1), 4 F 3/2 (Fig. 2), 4 F 5/2 and 2 H 9/2 (Fig. 3), 4 F 7/2 and 4 S 3/2 (Fig. 4), and 2 P 1/2 (Fig. 5). At 8 K the absorption peaks observed in each sample nearly superimpose. The spectra were obtained using the same spectrophotometer with samples on a common mount, and the energies cm 1 reported in columns 5 and 7 of Table I represent the splitting of the multiplet manifolds in the ceramic and crystal, respectively. Within experimental resolution many of the experimental Stark levels listed in Table I are in agreement with the levels that are reported by Kaminskii, 8 and Gruber et al. 9 for Nd 3+ in single-crystal rods. The 300 K spectrum is also included in Figs. 1 5 for the ceramic sample. There is a shift of the absorption peaks to lower energy (longer wavelength) observed in comparing the 300 K spectrum to the 8 K spectrum in the ceramic sample. This is perhaps easiest to see in Figs. 2 and 5. A similar shift is also observed in the single-crystal spectrum, but is not shown in order to avoid crowding in the figures. The 300 K spectrum for the ceramic sample also shows temperaturedependent peaks (hot bands) that represent transitions from excited Stark levels in the ground-state manifold, 4 I 9/2,to Stark levels in excited manifolds. Four of the five Stark levels of 4 I 9/2 can be identified in Figs. 2 and 5 from transitions Z n to the Stark levels in 4 F 3/2 and 2 P 1/2 as Z 1 =0,Z 2 =130,Z 3 =200, and Z 4 =309, all in cm 1. These values compare favorably with those identified from the 8 K fluorescence spectrum reported in Table II, and the levels reported by Kaminskii, 8 and Gruber et al. 9 for Nd 3+ in single-crystal YAG. Table II gives the fluorescence spectrum obtained at 8 K from the emitting Stark level 4 F 3/2 R 1 to the Stark levels in manifolds 4 I 9/2, 4 I 11/2, and 4 I 13/2. Figures 6 8 show the details and the relative intensities within a given manifold for 4 I 9/2, (Fig. 6), 4 I 11/2 (Fig. 7), and 4 I 13/2 (Fig. 8). The crystalfield splitting of each manifold is given in Table II, column 6, and can be compared with the splitting obtained from the single-crystal fluorescence at the same temperature and Nd concentration. One fluorescence peak in Fig. 8 4 I 13/2 could not be resolved into two peak representing transitions from R 1 to X 1 and X 2 as reported by Kaminskii. 8 However, the crystal-field splitting of the 4 I 9/2, 4 I 11/2, and 4 I 13/2 of Nd 3+ in the ceramic sample match reasonably well with comparable splittings in the single-crystal sample listed in Table II, and with the experimental Stark levels reported by Kaminskii. 8 IV. ENERGY-LEVEL ANALYSIS A model Hamiltonian expressed in terms of free-ion or atomic parameters and the crystal-field splitting parameters is used to analyze the energy-level structure of the Nd 3+ 4f 3 ions that occupy Y 3+ sites having D 2 symmetry in the garnet structure. Table III identifies the free-ion parameters and the crystal-field splitting parameters taken from Ref. 9 and used TABLE III. Model parameters for Nd 3+ in ceramic YAG a. Free-ion parameter Value cm 1 Crystal-field parameter Value cm 1 E B F B F B F B B B B T B T 3 46 B T 4 39 T N b 106 T cm 1 c 18 T M P a Reference 9. b Number of calculated-to-experimental Strak levels involved. c Standard deviation of calculation (Ref. 9).

7 3056 J. Appl. Phys., Vol. 96, No. 6, 15 September 2004 Gruber et al. in our present calculations. The model Hamiltonian was diagonalized within the complete 4f 3, SLJM J basis set which includes 364 states. In a crystal field of D 2 symmetry each multiplet manifold of Nd 3+ 4f 3 2S+1 L J, splits into J+ 1 2 Kramers doublets that can be assigned experimentally based on an energy-level scheme obtained from analyses of the absorption, fluorescence, and Zeeman data. Because the calculated splitting based on the parameters given in Table III predicts levels within the standard deviation reported earlier 9 for the same levels analyzed in single-crystal laser rods, there is no need to further refine the calculation for the Nd 3+ levels in ceramic YAG by a least-squares fitting analysis. The calculated energy (Stark) levels are reported in column 6 of Table I and column 7 in Table II. Columns 5 7 in Table I and columns 6 8 in Table II allow the reader to compare individual Stark levels observed in the ceramic, and crystal form with the calculated energy levels. Within the standard deviation of 18 cm 1 quoted in Ref. 9, both sets of data are consistent in the details of the interpretation of the crystal-field splitting. In summary, we find that the crystal-field splitting of the energy levels of Nd 3+ are similar in both the ceramic sample and in the laser rod sample containing Nd 3+ in the garnet YAG host. The spectroscopic properties (positions and bandwidths) between 1750 and 350 nm appear to be identical for both ceramic and crystalline Nd 3+ : YAG. No additional inhomogeneous broadening was detected in the ceramic Nd 3+ : YAG sample at 8 K. The particle size in the ceramic host is sufficiently large so that the spectroscopic properties reflect the bulk phase of YAG. Our spectroscopic analysis, together with recent studies reported by Lu et al. 4 that compare the laser performance between Nd 3+ : YAG ceramic and singlecrystal rods, clearly show that this new form of the garnet host is an excellent alternative to Nd 3+ : YAG single crystals. 1 C. Greskovich and J. P. Chernoch, J. Appl. Phys. 44, 4599 (1973). 2 M. Sekita, H. Haneda, Shiransaki, T. Yanagitani, J. Appl. Phys. 69, 3709 (1991). 3 J. Lu, M. Prabhu, J. Xu, K. Ueda, H. Yagi, T. Yanagitani, and A. A. Kaminskii, Appl. Phys. Lett. 77, 3707 (2000). 4 J. Lu, K. Ueda, H. Yagi, T. Yanagitani, Y. Akiyama, and A. A. Kaminskii, J. Alloys Compd. 341, 220 (2002). 5 A. A. Kaminskii et al., Quantum Electronics Conference, IQEC-2002, June 22 27, 2002, Moscow. 6 J. B. Gruber, B. Zandi, and D. K. Sardar (unpublished). 7 Baikowski International Corp. ceramic YAG specifications, for Konoshima Chemical Industry KK, Japan, A. A. Kaminskii, Laser Crystals (Springer, Berlin, 1981). 9 J. B. Gruber, M. E. Hills, T. H. Allik, C. K. Jayasankar, J. R. Quagliano, and F. S. Richardson, Phys. Rev. B (1990). 10 J. B. Gruber, D. K. Sardar, B. Zandi, T. A. Reynolds, T. Alekel, and D. A. Keszler, J. Appl. Phys. 93, 3345 (2003). 11 J. B. Gruber, M. E. Hills, R. M. Macfarlane, C. A. Morrison, G. A. Turner, G. J. Quarles, G. J. Kintz, and L. Esterowitz, Phys. Rev. B 40, 9464 (1989).