## Abstract

It was experimentally found that electronic structures of Russell-Saunders manifolds in Nd:YAG depended on the Nd^{3+}-doping concentration (*C*
_{Nd}) and its fabrication process. Both of the bandwidth and the branching ratio in fluorescent transitions in Nd:YAG varied almost linearly depending on *C*
_{Nd}, and a fabrication process has its own diluted limit of the bandwidth and the branching ratio. Also dependences of Stark splitting in Nd:YAG were also observed. Nd^{3+}-doping causes 1.9% and 4.5% reduction in the stimulated emission cross section of Nd:YAG per 1at.% of *C*
_{Nd} at 1.064 μm and 1.319 μm, respectively.

©2011 Optical Society of America

## 1. Introduction

The ceramic laser technology has become an indispensable technological element in solid-state lasers due to its advantages such as higher fracture toughness [1] and larger crystal size [2]. Recent advances of this technology realized a demonstration of more than 100-kW cw laser oscillation with TEM_{00} transverse mode by using of YAG ceramics [3]. From the viewpoint of developing high power solid-state lasers, the ceramic laser technology can supply laser media with high concentration of luminous rare-earth ions that have larger ionic radii than replaced ions: such as Nd^{3+} in YAG. While this advantage enables the higher pump absorption efficiency [4], it brings the quenching of radiative quantum efficiency [5]. Therefore, it is important for optimizing laser oscillators to determine the best doping concentration of luminous ions in each laser host material.

Since Nd:YAG is the most useful solid-state laser medium, there have been many researches for the definition of the figure of merit for Nd^{3+}-doping concentration (*C*
_{Nd}) in Nd:YAG. In these previous works it was assumed that the cross section of optical transitions was invariant. Only quenching of fluorescence lifetimes has been considered as the origin of *C*
_{Nd}-dependence in spectroscopic properties of Nd:YAG by using of Dexter-Förster theory [6,7]. On the contrary, a few works that paid attentions to the change of cross section induced by Nd^{3+}-doping exist [5], where the accordance of the radiative quantum efficiency between a theory and experiments suggest that total line strength of decay process is invariant. Also authors reported that there was a certain dependence of fluorescent profiles in concentrated Nd:YAG [8]. However, all these researches were still processed on the assumption that Einstein's A and B coefficients of optical transitions in Nd:YAG were invariant.

In addition, even though the influences of the fabrication process to spectroscopic properties are often ignored, there are some reports that referred to the manifold splitting variation due to the crystal field variation caused by the elaboration procedure [9]. This fact strongly suggests that there should be differences in spectroscopic properties between single-crystalline laser media and ceramic laser media.

In this work, detailed fluorescent characteristics of Nd:YAG fabricated by different processes were studied by spectroscopic evaluations with high resolution, and we discussed the variation of the stimulated emission cross section in Nd:YAG due to the structural changes of Russell-Saunders manifolds introduced by not only *C*
_{Nd} but also different fabrication processes.

## 2. Theory and method

The stimulated emission cross section *σ*(*ν*) at frequency *ν* is expressed with Einstein's B coefficient and the line-shape function *g*(*ν*) by

*h*and

*c*are Plank's constant and the speed of light, respectively. Einstein's B coefficient for the optical transition accompanied from an electronic initial state

*i*to a final state

*f*is defined by Fermi's golden rule as

*μ*are the wave-vector of the electronic state

*j*and the electric dipole moment operator, respectively.

*λ*indicates polarization component. Though

*C*

_{Nd}is not included in Eqs. (1) and (2) definitely, perturbation to $|{\psi}_{j}\u3009$ can be induced by the modified crystal field

*V*

_{C}influenced by

*C*

_{Nd}or other factors as crystal point defects. If the expansion

*V*

_{C}in terms of the strain

*ε*is defined byStark splitting

*V*between $|{\psi}_{i}\u3009$ and $|{\psi}_{f}\u3009$ is given by

_{if}^{2}

^{S}^{+1}

*L*. Higher order terms of

_{J}*V*

_{C}give the spectral line broadening caused by nonlinear interactions between electronic states and lattice vibrations. If Raman effects are dominant for determining the bandwidth Δ

*ν*of the transition from $|{\psi}_{i}\u3009$ to $|{\psi}_{f}\u3009$, Δ

_{if}*ν*is given as FWHM in the fluorescence by

_{if}*k*,

*ρ*,

*v*, and

*Θ*are Boltzmann's constant, the density, the velocity of the phonon, and Debye temperature of the crystal, respectively [10]. Since

*Θ*of Nd:YAG is considered to be independent on

*C*

_{Nd}[11], ${\left|\u3008{\psi}_{f}|{V}_{1}|{\psi}_{i}\u3009\right|}^{2}/\Delta {\nu}_{if}$ becomes a constant from Eq. (5).

By using of the assumption that *σ*(*ν*) has a lorentzian spectral profile, the line-shape function *g _{if}* (

*ν*) of the transition from the state

*i*to

*f*is expressed by

*ν*is the center frequency of the transition from

_{if}*i*to

*f*. By considering Eq. (1) and the relation between fluorescent intensities and

*σ*(

*ν*) [12], the fluorescent intensity

*I*(

_{if}*ν*) of the transition from level

*i*to

*f*is

*η*is the efficiency of a measurement defined by the experimental setup. Similarly the fluorescent intensity

*I*(

_{jk}*ν*) of the transition from manifold

*j*to

*k*is the summation of

*I*(

_{if}*ν*) as

*f*is the fractional population of level

_{i}*i*in the manifold

*j*. Equation (8) directly gives the branching ratio

*b*of transition from level

_{if}*i*in manifold

*j*to level

*f*in manifold

*k*, and

*b*is

_{if}The maximum value of the cross section *σ _{if}^{m}* of the transition from level

*i*to level

*f*is the most important value for cavity designs of high power lasers. From Eqs. (1) and (6)

*σ*is

_{if}^{m}Δ*ν _{if}*,

*ν*, and

_{if}*B*can be obtained by the least square fitting of experimentally measured fluorescent spectra to Eq. (7), and they give the relative values of the matrix elements of

_{if}*μ*,

*V*

_{0}, and

*V*

_{1}between optical transitions in each sample.

## 3. Experimental setup

There are two types of transparent YAG ceramics used in this work. Six samples in the first group were fabricated by sintering of carbonated precursor from wet synthesized powder (WS) with *C*
_{Nd} from 0.4 to 4.8at.%, and other four samples were fabricated by sintering with solid-state reaction (SS) between raw sesquioxide powders with *C*
_{Nd} from 1.1 to 5.4at.%. Also twelve YAG single crystals grown by Czochralski method (CZ) were evaluated in this work with *C*
_{Nd} from 0.06 to 1.4at.%. Manufactures of the samples in this work were Konoshima Chemicals Co., Ltd (WS), World-Lab Co., Ltd. (SS), and Scientific materials Co (CZ). These samples had parallel mirror polished surfaces with the thickness of 1 mm. Typical grain sizes of WS and SS ceramics were about several-*μ*m and several tens-μm, respectively.

*C*
_{Nd} of these samples were calibrated by the linearity of the absorption coefficient at 808 nm with the resolution of 0.2 nm measured by a spectrophotometer (U-3500, Hitachi Co.). These samples were considered to have no clustering of Nd^{3+} from the dependence of the fluorescent lifetime on *C*
_{Nd} [5].

In order to obtain fluorescence spectra of Nd:YAG, samples were pumped by 10-mW radiation with 100Hz repetition rate and 10% duty ratio from a fiber coupled 808-nm laser diode (LIMO40-F400-DL808, LIMO GmbH). Fluorescences were analyzed by a monochromater (TRIAX-550, JOBIN YVON) with less than 0.05-nm resolution, and were detected by linear-InGaAs-array detector (IGA512-1x1, JOBIN YVON). The calibration of the wavelength was done within an error of 0.2 cm^{−1} by Ne lamp (Pencil style calibration lamp 6032, Oriel), Xe lamp (Pencil style calibration lamp 6033, Oriel) and Hg lamp (Pen-Ray Mercury lamp, UVP). Samples were sandwiched by copper plates of which temperature was maintained at 23°C by a thermo-electric controller (LDT-5948, ILX Lightwave Co.).

In order to evaluate the center wavelengths, the bandwidths, and line intensities of fluorescence, the least square fitting was carried out with typical errors of 0.02 cm^{−1}, 1%, and 2%, respectively. The accuracy of evaluations in this work is summarized in Table 1
, where *δA* means a reproducible error of *A*.

## 4. Results

When we compare the spectral shapes of various samples, it can be assumed the invariant line strength as stated in the introduction. Therefore, comparing of spectral profiles normalized by their integral with wavelength should be appropriate. Figure 1(a)
shows normalized fluorescent intensities of Nd^{3+}-diluted ceramics (WS) and Nd^{3+}-concentrated ceramics (SS). 0.8at.% WS sample has the spectrum width of 0.7 nm at 1.064 μm, and it is 63% of 1.1 nm that is a well-known value for Nd:YAG [13]. In the case of 1.319 μm shown in Fig. 1(b), observed bandwidth 0.7 nm of 0.4at.% WS samples is 58% of 1.2 nm reported in [13]. Since measured spectral bandwidths depends on *C*
_{Nd}, peak intensities of 5.4at.% SS sample at 1.064-μm and 1.319-μm were decreased 17.2% and 30.9% from above sample, respectively.Even though in the same *C*
_{Nd} conditions, there is a difference in the bandwidth between single crystal and ceramics as shown in Fig. 2(a)
. As a result of this difference in line broadenings, 1.8% and 3.2% smaller peak intensities of 1.3at.% Nd:YAG single crystal than ceramics (WS) at 1.064-μm and 1.319-μm, respectively, were detected experimentally. The label of each Stark level that is used in this work is summerized in Fig. 2(b).

## 5. Discussions

#### 5.1 Matrix element of 0th order term of crystal field

Information of Stark-splitting in Russell-Saunders manifolds of ^{4}
*F* and ^{4}
*I* in Nd^{3+}-doped laser materials is very important, because laser excitation and oscillation wavelengths coincide to the intervals between these levels. Stark splitting defined by the matrix elements of *V*
_{0} with Eq. (4) can be evaluated by the difference of *ν _{if}*.

Stark splitting of ^{4}
*F*
_{3/2} in Nd:YAG is shown in Fig. 3(a)
. This splitting was defined by the difference between the frequencies of the transition from *R*
_{i} to *Y*
_{1}, where generally *R*
_{i}, *Y*
_{i}, and *X*
_{i} are used as labels of the *i*-th lower Stark level of ^{4}
*F*
_{3/2}, ^{4}
*I*
_{11/2}, and ^{4}
*I*
_{13/2} manifolds, respectively. Though changes of Stark splitting by introducing Nd^{3+} can be observed with the order of a percent, the existence of the differences in splitting between single crystals and ceramics is quite important because *R*
_{1} and *R*
_{2} are the laser upper levels of Nd:YAG.

Also Fig. 3(b) shows Stark splitting of ^{4}
*I*
_{13/2} in Nd:YAG calculated from the transition from *R*
_{i} to *X*
_{1} and *X*
_{2}, which are the lower laser levels of 1.319-μm oscillation. While the splitting of *X*
_{1} and *X*
_{2} in single crystals is similar to that of WS ceramics, the difference between WS and SS ceramics was observed. The obtained variation of this splitting among our samples reached to 14%.

The variation of matrix elements of *V*
_{0} should imply the change of branching ratio of otical transitions according to Judd-Ofelt optical transition scheme [14,15]. However, changes of *ν _{if}* in fluorescence from Nd:YAG depending on

*C*

_{Nd}or fabrication process are negligible because crystal field effects discussed in this section are much smaller than the splitting between

^{2}

^{S}^{+1}

*L*manifolds.

_{J}#### 5.2 Matrix element of 1st order term of crystal field

Matrix element of *V*
_{1} brings the line broadening of the fluorescence due to Raman interactions as described in Eq. (5), where the ratio of ${\left|\u3008{\psi}_{f}|{V}_{1}|{\psi}_{i}\u3009\right|}^{2}$ among different samples coincides to the ratio of Δ*ν _{if}*. Figures 4(a)
and 4(b) show the ratio of ${\left|\u3008{\psi}_{f}|{V}_{1}|{\psi}_{i}\u3009\right|}^{2}$ in the transition from

^{4}

*F*

_{3/2}to

^{4}

*I*

_{11/2}, where the differences in ${\left|\u3008{\psi}_{f}|{V}_{1}|{\psi}_{i}\u3009\right|}^{2}$ between fabrication process of CZ, WS, and SS are found. The linear dependence of ${\left|\u3008{\psi}_{f}|{V}_{1}|{\psi}_{i}\u3009\right|}^{2}$ on

*C*

_{Nd}is also discovered experimentally.

These variations of ${\left|\u3008{\psi}_{f}|{V}_{1}|{\psi}_{i}\u3009\right|}^{2}$ directly relate *σ _{if}^{m}* as in Eq. (10). Therefore, what are discussed in this section give the crucial influence to the laser oscillation characteristics of Nd:YAG. The variation of

*σ*is summarized in Sect. 5.5.

_{if}^{m}#### 5.3 Matrix element of dipole moment

As discussed in sect. 5.1 matrix elements of a dipole transition moment in each sample should also be influenced by Nd^{3+}-doping and fabrication processes, and Figs. 5(a)
and 5(b) show the ratio of ${\left|\u3008{\psi}_{f}|\mu |{\psi}_{i}\u3009\right|}^{2}$ of each sample in the transition from ^{4}
*F*
_{3/2} to ^{4}
*I*
_{11/2}. The dependence on fabrication processes of the branching ratio in Nd:YAG was indicated clearly in Fig. 5(a), where branching ratios of single crystals and ceramics are invariant and variant, respectively. On the contrary, dependence of ${\left|\u3008{\psi}_{f}|\mu |{\psi}_{i}\u3009\right|}^{2}$ on *C*
_{Nd} in ^{4}
*F*
_{3/2} to *Y*
_{3} without differences between fabrication process is found in Fig. 5(b).

These differences of ${\left|\u3008{\psi}_{f}|\mu |{\psi}_{i}\u3009\right|}^{2}$ among samples directly indicate that Einstein's A and B coefficients of one sample can be different from those of another sample. The reason why the variation of the ratios is so different between Figs. 5(a) and 5(b) is not revealed definitely, while it can be due to the different representation of the crystal field in each Stark level.

#### 5.4 Branching ratio

Since Einstein's A and B coefficients are not invariant as discussed above, branching ratios should also be influenced by the amount of doped Nd^{3+}. Figure 6(a)
shows the *C*
_{Nd}-dependence of branching ratio in transitions from ^{4}
*F*
_{3/2} to ^{4}
*I*
_{11/2} of Nd:YAG. Although in other peaks *C*
_{Nd}-dependence was hardly observed, branching ratios of the transition from *R*
_{2} to *Y*
_{3} (1064.1 nm) and from *R*
_{1} to *Y*
_{2} (1064.5 nm) transitions have strong *C*
_{Nd}-dependence. This variation in the branching ratio should enhance the reduction of *σ*
_{em} at 1.064-μm due to the line-broadening introduced by Nd^{3+}-doping. On the contrary, the change of branching ratio suppresses the reduction of *σ*
_{em} at 1.319-μm transition as shown in Fig. 6(b).Past spectroscopic evaluations of rare-earth doped laser media by use of the concentration quenching analysis [16] have assumed the invariance of branching ratios. Hence, there is a possibility of errors under designing of advanced laser oscillators by use of the physical properties value that has been believed until nowadays. This possibility is expressed numerically by newly introduced parameter *A*(*λ*) in the next section.

#### 5.5 Stimulated emission cross section

The maximum value of the cross section *σ _{if}^{m}* contains Δ

*ν*,

_{if}*ν*, and

_{if}*B*that depend on

_{if}*C*

_{Nd}as expressed in Eq. (10). We can evaluate the relative value of

*σ*by using of the Δ

_{if}^{m}*ν*,

_{if}*ν*, and the relative value of

_{if}*B*that were experimentally obtained in this work. Figures 7(a) and 7(b) show the estimated variation of

_{if}*σ*of Nd:YAG at 1.064-μm and 1.319-μm, respectively. The stimulated emission cross section

_{if}^{m}*σ*

_{em}(

*λ*) can be approximated by

*λ*is wavelength,

*σ*

_{em}

^{0}(

*λ*) is a diluted limit of

*σ*

_{em}(

*λ*), and

*A*(

*λ*) is a fitting parameter. While

*σ*

_{em}

^{0}(

*λ*) depends on synthesizing processes of samples,

*A*(

*λ*) is hardly influenced by their fabrication processes. At 1.064-μm and 1.319-μm,

*A*(

*λ*) is 0.019/at.% and 0.045/at.%, respectively.

When we estimate laser performances with highly concentrated Nd:YAG, it is important to consider not *σ*
_{em} but the product of *σ*
_{em} and fluorescent lifetime *τ*. In the case of 5at.% Nd:YAG, the radiative quantum efficiency is 21% [5]. Therefore, *σ*
_{em}
*τ* of 5at.% Nd:YAG is decreased to 19% and 16% at 1.064-μm and 1.319-μm, respectively.

## 6. Conclusion

We experimentally confirmed that wave-vectors of electronic states in Russell-Saunders manifolds ^{4}
*F* and ^{4}
*I* in Nd:YAG depended on both of *C*
_{Nd} and its fabrication process. An error of *C*
_{Nd} in this work was below 0.4%. This dependence should be due to the perturbation introduced by the crystal field *V*
_{C} that is influenced by *C*
_{Nd} or other factors such as crystal defects. These variations of electronic states cause following changes in spectroscopic characteristics of Nd:YAG:

- 1. Bandwidth of each fluorescence peak in Nd:YAG varies linearly depending on
*C*_{Nd}. - 2. Bandwidth of ceramics (SS) is narrower than other samples with same
*C*_{Nd}. - 3. Branching ratio of some transitions in Nd:YAG depends on
*C*_{Nd}severely.

In this work at the utmost 16% and 29% reduction of *σ*
_{em} at 1064 μm and 1319 μm were experimentally found respectively, while the deviation in evaluations of *σ*
_{em} was below 3%. Therefore, it is necessary for constructing laser oscillators with highly concentrated Nd:YAG to consider not only degradation of radiative quantum efficiency for the sake of concentration quenching but also the reduction of laser gain due to decreased *σ*
_{em} by Nd^{3+}-doping.

Since the temperature was fixed at 23°C in this work, the evaluation of temperature dependence should be a future work in order to validate the presented results. Also the detailed studies on the branching ratio between Russell-Saunders manifolds will give interesting characteristics for physics of the solid-state laser materials. Above studies are also very useful for future detailed analyses of absorption bands in order to improve the diode-pumped solid-state lasers.

It is important for improving the material performance to clarify the relationship between results of this work and Nd^{3+}-local environments such as defect, disorder, impurities, homogeneity, and etc. In order to solve this problem, the development of the detection method for every distinct environment of Nd^{3+} in laser media must be required. Deep theoretical discussions based on the evaluated Nd^{3+}-local environments will bring the complete design rule of Nd^{3+}-doped laser media.

## Acknowledgments

This work was partially supported by Genesis Research Institute, and by the Special Coordination Funds for Promoting Science and Technology of the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

## References and links

**1. **A. A. Kaminskii, M. S. Akchurin, R. V. Gainutdinov, K. Takaichi, A. Shirakava, H. Yagi, T. Yanagitani, and K. Ueda, “Microhardness and fracture toughness of Y_{2}O_{3}- and Y_{3}Al_{5}O_{12}-based nanocrystalline laser ceramics,” Crystallogr. Rep. **50**(5), 869–873 (2005). [CrossRef]

**2. **R. M. Yamamoto, B. S. Bhachu, K. P. Cutter, S. N. Fochs, S. A. Letts, C. W. Parks, M. D. Rotter, and T. F. Soules, “The use of large transparent ceramics in a high powered, diode pumped solid-state laser,” Proceedings of OSA Topical meeting on Advanced Solid-State Photonics 2008, WC5, Nara, Japan (Jan. 2008).

**3. **S. J. McNaught, H. Komine, S. B. Weiss, R. Simpson, A. M. F. Johnson, J. Machan, C. P. Asman, M. Weber, G. C. Jones, M. M. Valley, A. Jankevics, D. Burchman, M. McClellan, J. Sollee, J. Marmo, and H. Injeyan, “100 kW coherently combined slab MOPAs,” in *Proceedings of Conference on Quantum Electronics and Laser Science Conference on Lasers and Electro-Optics*, CLEO/QELS, CThA1, Baltimore, MA, USA (2009).

**4. **I. Shoji, S. Kurimura, Y. Sato, T. Taira, A. Ikesue, and K. Yoshida, “Optical properties and laser characteristics of highly Nd^{3+}-doped Y_{3}Al_{5}O_{12} ceramics,” Appl. Phys. Lett. **77**(7), 939–941 (2000). [CrossRef]

**5. **V. Lupei, A. Lupei, S. Georgescu, T. Taira, Y. Sato, and A. Ikesue, “The effect of Nd concentration on the spectroscopic and emission decay properties of highly doped Nd:YAG ceramics,” Phys. Rev. B **64**(9), 092102 (2001). [CrossRef]

**6. **D. L. Dexter, “A theory of sensitized luminescence in solids,” J. Chem. Phys. **21**(5), 836–850 (1953). [CrossRef]

**7. **T. Förster, “Intermolecular energy migration and fluorescence,” Ann. Phys. **2**, 55–75 (1948).

**8. **Y. Sato, T. Taira, and A. Ikesue, “A study on influences of Nd^{3+}-doping concentration upon spectroscopic properties of Nd:Y_{3}Al_{5}O_{12} ceramics,” in *Proceedings of OSA Topical meeting on Advanced Solid-State Photonics 2009*, WB18, Denver, CO, USA (Feb. 2009).

**9. **F. S. Ermeneux, C. Goutaudier, R. Moncorge, M. T. Cohen-Adad, M. Bettinelli, and E. Cavalli, “Comparative optical characterization of various Nd^{3+}:YVO_{4} single crystals,” Opt. Mater. **13**(2), 193–204 (1999). [CrossRef]

**10. **T. Kushida, “Linewidths and thermal shifts of spectral lines in neodymium-doped yttrium aluminum garnet and calcium fluorophosphate,” Phys. Rev. **185**(2), 500–508 (1969). [CrossRef]

**11. **Y. Sato, J. Akiyama, and T. Taira, “Effects of rare-earth doping on thermal conductivity in Y_{3}Al_{5}O_{12} crystals,” Opt. Mater. **31**(5), 720–724 (2009). [CrossRef]

**12. **T. Kushida, H. M. Marcos, and J. E. Geusic, “Laser transition cross section and fluorescence branching ratio for Nd^{3+} in yttrium aluminum garnet,” Phys. Rev. **167**(2), 289–291 (1968). [CrossRef]

**13. **T. Taira, “RE^{3+}-ion-doped YAG ceramic lasers,” IEEE J. Sel. Top. Quantum Electron. **13**(3), 798–809 (2007). [CrossRef]

**14. **B. R. Judd, “Optical absorption intensities of rare-earth ions,” Phys. Rev. **127**(3), 750–761 (1962). [CrossRef]

**15. **G. S. Ofelt, “Intensities of crystal spectra of rare-earth ions,” J. Chem. Phys. **37**(3), 511–520 (1962). [CrossRef]

**16. **A. Blumen and J. Manz, “On the concentration and time dependence of the energy transfer to randomly distributed acceptors,” J. Chem. Phys. **71**(11), 4694–4702 (1979). [CrossRef]