You are on page 1of 6

ARTICLE IN PRESS

Journal of Luminescence 129 (2009) 14011406

Contents lists available at ScienceDirect

Journal of Luminescence
journal homepage: www.elsevier.com/locate/jlumin

Nonstoichiometric laser materials: Designer wavelengths in neodymium-doped garnets


Brian M. Walsh , Norman P. Barnes
NASA Langley Research Center, Hampton, VA 23681, USA

a r t i c l e in fo
Available online 2 April 2009 Keywords: Nd lasers 0.94 mm lasers Compositional tuning Amplied spontaneous emission (ASE) Voigt lineshape Water vapor

abstract
The tunable nature of lasers provides for a wide range of applications. Most applications rely on nding available laser wavelengths to meet the needs of the research. This article presents the concept of compositional tuning, whereby the laser wavelength is designed by exploiting nonstoichiometry. For research where precise wavelengths are required, such as remote sensing, this is highly advantageous. A theoretical basis for the concept is presented and experimental results in spectroscopic measurements support the theoretical basis. Laser operation nicely demonstrates the validity of the concept of designer lasers. Published by Elsevier B.V.

1. Introduction Stoichiometry is a term derived from the Greek word stoikheion, meaning element and the word metron, meaning measure. In chemistry it is related to conservation of mass, the law of denite proportions and the law of multiple proportions. In a stoichiometric material, the elements composing the material appear as ratios of integers. An example is yttrium aluminum garnet, Y3Al5O12. Nonstoichiometric materials are materials composed of elements that cannot be represented by a ratio of whole numbers. Many crystals, in fact, show some degree of deviation from stoichiometry due to crystallographic point defects that occur in the growth process, resulting in a small excess or deciency of an element. This paper is not concerned with these small accidental deviations in stoichiometry, but large deviations introduced intentionally for the purpose of altering the crystal eld, thus shifting the transition wavelengths of the lanthanide dopant ions. The term compositional tuning is used to refer to this type of method for laser tuning. Compositional tuning creates laser materials with noninteger ratios of elements, thus tuning the laser as chemical composition changes. While the change in wavelength with composition from one structure to the next similar one, for example Y3Al5O12 (YAG) to Y3Ga5O12 (YGG), spans several nanometers, there is a continuous range of intermediate wavelengths that can be realized. For applications that may require very specic wavelengths, such as remote sensing, designer wavelength capability is denitely an advantage. Research for specic applications usually relies on trying to t the application to the lasers available. Compositional
Corresponding author. Tel.: +1757 864 7112; fax: +1757 864 8828.

tuning ts the laser to the specic application. The absorption characteristics of any number of atmospheric constituents exhibit a ne structure in their spectral proles. The ability to design the laser wavelength to match these features is advantageous. A specic example is H2O, water vapor, in the atmosphere. The atmospheric prole of water vapor with altitude is important in atmospheric dynamics on Earth and on extraterrestrial planets. On Mars, the detection of H2O indicates potential areas where life may exist. On Earth, an understanding of global H2O distribution aids in understanding long-term climate trends and short-term prediction of weather. Although fossil fuel emissions are considered a major contributor to the green house effect and global warming, water vapor is also a major green house gas as well. Other atmospheric species of interest that affect life on earth include ozone (O3), sulfur dioxide (SO2), carbon dioxide (CO2), carbon monoxide(CO), and methane (CH4). The usefulness of laser devices designed to specically target absorption features of atmospheric constituents cannot be understated. In the next section, the concept of compositional tuning within the context of nonstoichiometric materials will be discussed. A theoretical basis involving chemistry and the crystallography of materials will form the foundation, followed in the remaining sections by practical applications in spectroscopy and realized in laser demonstrations.

2. Theory Garnets have the general structure {A}3[B]2(C)3O12, where { } is the dodecahedral site, [ ] is an octahedral site, and ( ) is a tetrahedral site. Although YAG is usually written as Y3Al5O12, it is actually Y3Al2Al3O12 because the aluminum atoms occupy sites of

E-mail address: brian.m.walsh@nasa.gov (B.M. Walsh). 0022-2313/$ - see front matter Published by Elsevier B.V. doi:10.1016/j.jlumin.2009.01.021

ARTICLE IN PRESS
1402 B.M. Walsh, N.P. Barnes / Journal of Luminescence 129 (2009) 14011406

Fig. 1. Crystal structure of garnets.

different symmetry. The various sites have different symmetry depending on the number of coordinated oxygen atoms, eight for the dodecahedra, six for the octahedra, and four for the tetrahedra. This is referred to as the coordination number (CN) of the site. The crystal structure showing the relative sizes of the atoms and their oxygen coordination is pictured in Fig. 1. The cation (RE3+, Al3+, Ga3+, etc.) and anion (O2) distance is generally determined by the sum of the ionic radii. The ionic radius of the cations depend on the coordination number of the site. [1] The radius ratio rule of Pauling [2] is a geometric argument that dictates the number of anions that can be placed around a cation. Generally speaking, it predicts that as the cation size decreases the coordination number decreases. In general, the following rules apply: CN 8; 1:0004Rc =Ra 40:732 Dodecahedral sites CN 6; 0:7324Rc =Ra 40:414 Octahedral sites CN 4; 0:4144Rc =Ra 40:225 Tetrahedral sites where Rc is the cation radius, Ra is the anion radius and CN is the coordination number. The radius units are in angstroms. This rule is only applicable in general, as many exceptions exist, but it provides a useful set of guidelines for which ions will t in various sites. This is helpful in deciding the possibilities for atoms that can be used to form compositionally tuned structures. Thus far only the ionic character of the bonds has been considered. In reality, the bond is actually some mixture of ionic and covalent bonding. In an ionic bond, the atoms (cations and anions) are considered as charged particles and electrostatic forces hold them together. The bond strength depends on the ionic charges. In a covalent bond, the atoms satisfy charge balance by sharing their electrons with adjacent orbitals in hybrid or molecular orbitals. The bond strength depends on orbital overlap. The situation is depicted in Fig. 2. It is convenient to discuss ionic and covalent bonds in terms of Paulings theory of electronegativity. [3] Broadly speaking, Pauling dened electronegativity as The power of an atom in a molecule to attract electrons to itself. [4] In more specic terms, Pauling has stated, ythe energy of an actual bond between unlike atoms is greater than (or equal to) the energy of a normal covalent bond between these atoms. Pauling attributed the extra bond energy as due to the additional ionic character of the bond. That is, it is the additional ionic resonance energy that the bond has as compared with the bond between like atoms. In this way, Pauling determined an electronegativity scale and assigned a value to each element of the periodic table based on empirical thermochemical data. In this theory, the fraction of the bond that is ionic can be determined by considering the electronegativity difference between atoms. The relationship is given as [4] f I 1 exp0:25Dw2 (1)

Fig. 2. Ionic and covalent bonding.

bond is mostly covalent. This is also illustrated in Fig. 2, where the nature of the bond is sketched versus electronegativity difference. Some general ranges of Dw values and bond type are highlighted in this gure as well. Compositional tuning of nonstoichiometric materials affects the crystal eld inuence on energy levels. Even though the crystal eld is a small perturbation, it provides enough tunability to be useful over several nanometers. The crystal eld can be characterized by Bnm parameters as Bnm rn Anm (2)

where rn are radial integrals of the 4f rare-earth ions and Anm is a host-dependent parameter given by X Anm q2 Z j C nm fj ; yj =Rn1 (3) e j
j

where qe is the electronic charge, Zj is the ionic charge, Cnm(fj,yj) are spherical harmonics and Rj is the distance separating atoms. If the crystal is composed of atoms that have only ionic bonds then the ionic charges, Zj, would be given by the valence state. However, in the case where the bonds are only partially ionic, the effective charges need to be found for bonds that are ionic and covalent. The electronegativity difference determines the fraction of the bond that is ionic, as given in Eq. (1), however, as originally dened by Pauling, electronegativity is assumed to be a property of an atom before the bond is formed. It is therefore reasonable to consider an effective electronegativity for atoms in crystals. One natural way to do this is to scale them according to a comparison ionic and covalent bond lengths. The effective charges, Z*, can be then considered in the following way. If the bond is entirely covalent then half the charge is involved in bonding. In the other extreme, if the bond were entirely ionic the valence charge would apply. A linear combination of extremes gives the effective charge Z 1 f I Z f IZ 2 (4)

where Dw is the electronegativity difference. From Eq. (1) it is seen that for large Dw the bond is mostly ionic, while for small Dw the

where Z is the valence charge and fI is the fraction of the bond that is ionic. Note that if fI 1 (purely ionic bonding) then Z* Z and if fI 0 (purely covalent bonding) then Z* Z/2, which are the extreme cases. Once the effective electronegativity is calculated, the fraction of the bond that is ionic can be found using Eq. (1). The effective charges can then be calculated using Eq. (4). With the effective charges, the crystal eld parameters can be determined from Eqs. (2) and (3) with knowledge of the crystal structure (spherical

ARTICLE IN PRESS
B.M. Walsh, N.P. Barnes / Journal of Luminescence 129 (2009) 14011406 1403

harmonics and atomic distances) as well as the rare-earth ion (radial integrals). With this knowledge, the compositional tuning of any rare-earth ion in any crystal structure can be quantied [57].

3. Spectroscopy The emission spectroscopy of a wide variety of Nd-doped garnets and mixed garnets were investigated to provide information on line center, line width, and cross section. Of particular interest is the 4F3/2-4I9/2 transition near 0.94 mm as this laser line has applications for H2O lidar measurements. One line of interest is 944 nm due to its strong H2O absorption. Fig. 3 shows the spectra of YAGYSAG mixed garnet materials. The top spectra are Nd:YAG, the middle spectra is Nd:YAG0.82YSAG0.18, and the bottom spectra is Nd:YSAG. This illustrates nicely how the wavelength can be precisely tuned to the desired wavelength by choosing the material composition. In this case, the Nd:YAG line is too long at 946 nm while the Nd:YSAG is too short at 943 nm. Appropriate composition mixtures of YAG and YSAG, in this case 0.82 YAG and 0.18 YSAG put the wavelength at the desired 944 nm. The critical parameters that are measured include the linewidth for assessment of inhomogeneous broadening, the line center wavelength with composition for laser tuning, and the cross-section ratios (0.941.06 mm) for the purpose of mitigating amplied spontaneous emission effects (ASE). [8] Nd lasers at 0.94 mm, operating in Q-switched mode, can be severely limited by ASE effects of the stronger 1.06 mm line operating on the 4 F3/2-4I11/2 transition because this transition originates from the same manifold as the transition at 0.94 mm. It is therefore benecial to choose a material that has a small cross-section ratio. We have performed these emission measurements on several new types of mixed garnet materials. Emission spectra have been

measured in the wavelength range 0.851.45 mm to obtain all transitions from the 4F3/2 lasing manifold, with the exception of the 4F3/2-4I15/2 transition which has a very small branching ratio and can be neglected. Using measured lifetimes and the measured emission spectra, the cross sections of the relevant manifolds are obtained. The mixed structure of these materials produces large inhomogeneous broadening due to the mixed occupancy of the constituent ions making up the host on the dodecahedral, tetrahedral, and octahedral sites of the garnet structure. The ability to shift the wavelength is often accompanied by a lowered cross section due to inhomogeneous broadening. In order to properly make an assessment of the best material in terms of line center, line width, absolute cross section, and cross-section ratio we have t the R1-Z5 line to the Voigt prole. The Voigt prole is a convolution of Lorentzian (homogeneous line) and Gaussian (inhomogeneous line) proles. The tting results are given in Table 1. In Table 1, l is the line center wavelength, DEL is the Lorentzian linewidth (FWHM), DEG is the Gaussian linewidth (FWHM), DEV is the Voigt linewidth (FWHM), sem is the emission cross section P of the R1-Z5 line, and r is the cross-section ratio of $1.06 and $0.94 mm emission cross sections. The term Voigt shape is a parameter that denes the relative weight of Lorentzian and Gaussian contributions to the linewidth. The equation describing the Voigt shape parameter, SV, is given by [9] SV p DEL In 2 DEG (5)

The Voigt shape parameter is an indication of the dominant mechanism for broadening (homogeneous or inhomogeneous). In general, for 0oSVo1 the Voigt prole becomes Gaussian, while for SV41 the prole becomes Lorentzian. At SVE2 the Voigt shape is mainly Lorentzian. The observed spectral linewidth, in this case a Voigt width, is a convolution of Lorentzian and Gaussian linewidths given by [9]

DEV

DEL
2

DE2 L
4

!1=2

DE2 G

(6)

Fig. 3. Compositional tuning in YAGYSAG materials.

It is clear from Table 1 that all the nonstoichiometric mixed garnet materials show some large degree of inhomogeneous broadening accompanied by reduction in the emission cross section, which is to be expected. There are two additional materials that were investigated that do not appear in Table 1. They are YAGxYGG(1x) and Nd:YAGxGGG(1x). These materials have R1-Z5 lines that show a slight splitting into two lines. This is not conducive to Voigt tting so the previous analysis could not be performed. YAG, YGG, and GGG all have SVo1, and are mainly Lorentzian. Their corresponding mixed structures, however, may show this splitting because the inhomogeneous broadening is not sufcient to mask it. In Fig. 4, the cross-section ratio versus R1-Z5 line is plotted for the materials in Table 1. Using electronegativity theory outlined in Section 2, the theoretical prediction for tuning rates can be compared with the measurements. [57] YAGYSAG and YAGYGG (YGAG) [10] materials are chosen for comparison. The results are given in Table 2. The tuning ranges given in Table 2 refer to the change in energy of the R1-Z5 line in going from Nd:YAG to YSAG or YGG, respectively. The theoretical calculations were performed a couple of times with different values for the 4f radial integrals, so both calculated tuning ranges are reported. As can be seen the agreement between the theory and the experiment is quite good, validating the compositional tuning theory based on Paulings electronegativity theory.

ARTICLE IN PRESS
1404 B.M. Walsh, N.P. Barnes / Journal of Luminescence 129 (2009) 14011406

Table 1 Spectroscopic parameters from Voigt tting of the R1-Z5 laser line in the 4F3/2-4I9/2 neodymium transition. Material YSGG YAG1/2YSGG1/2 YAG3/5YSGG2/5 YAG2/3YSGG1/3 YAG3/4YSGG1/4 YAG GSAG YAG0.45GSAG0.55 YAG0.50GSAG0.50 YAG YSAG YAG0.18GSAG0.82 YAG GGG YSAG2/3GGG1/3 YSAG YAG0.03 YAG0.20 YAG0.30 YAG0.40 (YSAG0.98GGG0.02)0.97 (YSAG0.98GGG0.02)0.80 (YSAG0.90GGG0.10)0.70 (YSAG0.90GGG0.10)0.60

l (nm)
937.75 942.59 943.43 944.24 944.84 945.87 944.23 944.51 945.87 943.63 943.93 945.87 937.30 940.96 943.63 943.66 944.16 944.10 944.81

DEL (cm1)
8.33 6.76 7.63 8.35 6.76 8.09 7.37 7.59 8.09 8.15 7.73 8.09 10.19 7.83 8.15 8.68 7.97 8.12 8.36

DEG (cm1)
13.15 28.50 27.61 26.03 25.56 3.85 22.45 22.55 3.85 15.52 17.28 3.85 3.94 28.24 15.52 17.17 19.26 22.23 22.19

DEV (cm1)
21.02 32.89 31.65 28.83 26.66 10.02 26.43 26.66 10.02 20.13 21.58 10.02 11.94 32.42 20.13 22.06 23.65 26.76 26.65

Voigt shape 0.5273 0.1975 0.2303 0.2671 0.3432 1.7400 0.2733 0.2803 1.7400 0.4374 0.3721 1.7400 2.1491 0.2310 0.4374 0.4208 0.3447 0.3043 0.3138

sem R1-Z5 ( 1020 cm2)


1.7073 1.5050 1.4177 1.5780 1.4026 3.7670 1.9673 1.8017 3.7670 2.5825 2.3277 3.7670 1.7500 1.4761 2.5825 2.1152 2.0205 1.8765 1.7903

Cross-section ratio ( 5.57 6.01 6.25 6.30 6.50 7.36 4.44 4.75 7.36 4.64 5.07 7.36 7.66 6.08 4.64 4.75 4.96 5.04 5.27

r)

8.0 7.5 7.0


r (1.06/0.94)

GGG

YAGxYSGG1-x YSAGxGGG1-x YAGxYSAG1-x YAGxGSAG1-x YAGx(YSAGyGGG1-y)1-x

YAG

6.5 6.0 5.5 5.0 YSAG 4.5 4.0 937 YSGG

938

939

940

941 942 943 Wavelength (nm)


P
r

944

945

946

947

Fig. 4. Nd cross-section ratio

versus R1-Z5 wavelength.

Fig. 5. Nd Cross sections in YAG and YAG0.18YSAG0.82.

Table 2 Calculated and measured tuning range. Material Nd:YAGYSAG Nd:YAGYGG Theoretical 25.3/26.1 82.7/95.4 Measured 30 89 Units cm1 cm1

4. Laser performance The ashlamp-pumped laser performance of Nd:GYAG and Nd:YAG0.18YSAG0.82 was compared. The laser consisted of two ashlamp-pumped cavities in a 0.85-m-long folded resonator design, as shown in Fig. 6. The folded resonator design was used to provide discrimination against 1.06 mm oscillation by providing 2 mirror surfaces highly transmitting at this wavelength. An acoustooptic modulator was employed for Q-switching and an external amplier was used to amplify the laser energy from the oscillator. Fig. 7 shows the normal mode and Q-switched laser performance for an Nd:GYAG laser operating at 0.946 mm and an Nd:YAG0.18YSAG0.82, referred to as YAG/YSAG, laser operating at 0.944 mm. GYAG is a compositionally tuned material with properties similar to YAG. Nd:GYAG contains some gadolinium (Gd3+) that replaces some of the yttrium (Y3+) in Nd:YAG, however, it does not have transition wavelengths that are very much different. Its advantage is that it possesses a smaller

Now, as for material selection, it is clear that the Nd-doped nonstoichiometric materials that show the best combination of properties are YAGxGSAG(1x), YAGxYSAG(1x), and YAGx[YSAGy GGG(1y)](1x). This is based on the best choice of materials due to peak cross section, cross-section ratio, and Voigt shape parameter. All these materials can be tuned to 944 nm, the most useful line for H2O DIAL experiments in the 0.91.0 mm region. [11] On balance, the material of choice is YAGxYSAG(1x). The emission cross section of 4F3/2-4I9/2 and 4F3/2-4I11/2 transitions in Nd:YAG and Nd:YAG0.18YSAG0.82 is shown in Fig. 5 for comparison.

ARTICLE IN PRESS
B.M. Walsh, N.P. Barnes / Journal of Luminescence 129 (2009) 14011406 1405

A-O Q-Switch

HR 0.94 HT 1.06

Laser rod

Oscillator PFN Energy meter Amplifier PFN Energy meter

HR 0.94 HT 1.06

Laser rod

Output mirror

Pickoff

Laser rod

Fig. 6. Dual-cavity folded resonator design.

140.0 120.0 100.0 Laser energy (mJ) 80.0 60.0 40.0 20.0 0.0 20 30 40 50 60 70 Electrical energy (J) 80 90 100 Nd:GYAG (NM) Nd:GYAG (QS) Nd:YAG/YSAG (NM) Nd:YAG/YSAG (QS)

Fig. 7. Normal mode and Q-switched operation.

1.06 mm cross section than Nd:YAG, while retaining approximately the same 0.946 mm cross section as in Nd:YAG. This offers an advantage in mitigating amplied spontaneous emission [8]. In normal mode operation, the gain in the cavity is at moderate levels, so ASE is not a problem. In Q-switched operation the gain reaches high levels during holdoff and ASE manifests itself as a sublinear roll-off in laser energy with increasing pump energy. This is seen in the Q-switched data in Fig. 7. Each cavity of the Nd:YAG laser system contained a 5 55 mm2 rod consisting of a 38 mm active length Nd:YAG with bonded ends. The use of bonded ends, undoped sections diffusion bonded to the doped rods, eliminates the unpumped regions. Flashlamppumped systems have unpumped regions near the ends of the rod where they are mounted in the cavity with O-rings to seal them for uid cooling. These regions are not excited by the ashlamp and would mainly serve to increase the effects of ground state absorption if doped with neodymium ions. The Nd:YAG laser system was pumped with 120 ms ashlamp pulse duration and a 0.65 reectivity output coupler was used. Each cavity of the YAG/ YSAG laser system contained a 5 55 mm2 rod without bonded ends, so the active length was 55 mm and contains unpumped regions. As this was the rst attempt to lase Nd:YAG/YSAG materials, bonded ends were not used as a cost-saving measure. The Nd:YAG/YSAG laser system was pumped with 160 ms ashlamp pulse duration and a 0.80 reectivity output coupler was used. The longer pulse width used for the Nd:YAG/YSAG

lasing provided the higher pump energies necessary for respectable laser energies, because the R1-Z5 line in Nd:YAG/YSAG has a smaller cross section than in Nd:YAG. In addition, since Nd:YAG/ YSAG is a new laser material there are still some problems to be worked out regarding their optical quality. Some minor distortion of a heliumneon laser can be plainly seen on passage through our Nd:YAG/YSAG. This will deleteriously affect laser oscillation. So, higher pump energies are necessary because of the lower cross section and lower optical quality than in Nd:YAG. Despite this, the laser performance of Nd:YAG/YSAG is quite good. Q-switched pulses of 100 mJ with a laser pulse width of 220 ns were achieved. The results presented here were performed with a 1 Hz repetition rate. Fig. 7 demonstrates clearly the amplied spontaneous emission on the Q-switched laser energy for 0.94 mm lasing in Nd-doped materials. Nd:YAG begins to roll over at approximately 20 J above threshold, while Nd:YAG/YSAG does not begin to show a roll over until approximately 40 J above threshold. This is a validation that a lower cross-section ratio of 1.060.94 mm helps to substantially reduce the effects of ASE. While the threshold of Nd:YAG/YSAG is approximately 2 times higher than in Nd:YAG and the slope efciency is 2 times lower, improvements in the optical quality of Nd:YAG/YSAG are expected to improve this situation. Despite the higher threshold and lower slope efciency, maximum Q-switched energies of 100 mJ were obtained in both Nd:YAG and Nd:YAG/YSAG before ASE losses became severe.

5. Summary The principles underlying compositional tuning of nonstoichiometric materials have been covered. A model for prediction of tuning rates in these materials has been presented and shown to agree well with experiment. Voigt tting of spectroscopic data has given information on line center wavelength with composition, peak emission cross sections, and degree of inhomogeneous broadening. These considerations aided in the choice of a material with favourable properties for laser application at 944 nm, a designer wavelength for H2O remote-sensing measurements. The ashlamp-pumped laser performance has been evaluated for Nd:GYAG and Nd:YAG/YSAG. Over 100 mJ of Q-switched laser energy at 944 nm was achieved for compositionally tuned Nd:YAG/YSAG material containing 0.18 YAG and 0.82 YSAG. From basic crystallographic principles, theoretical models, and experimental spectroscopy, a broad picture of nonstoichiometric materials and designer wavelengths in Nd garnets has been presented. Laser demonstration shows the successful application of this development.

ARTICLE IN PRESS
1406 B.M. Walsh, N.P. Barnes / Journal of Luminescence 129 (2009) 14011406

References
[1] [2] [3] [4] R.D. Shannon, C.T. Prewitt, Acta Cryst. B 25 (1969) 925. L. Pauling, J. Am. Chem. Soc. 51 (1929) 1010. L. Pauling, J. Am. Chem. Soc. 54 (1932) 3570. L. Pauling, The Nature of the Chemical Bond, Cornell University Press, New York, 1960. [5] N.P. Barnes, B.M. Walsh, E.D. Filer, R.L. Hutcheson, R.W. Equall, OSA TOPS 68 (2002) 280.

N.P. Barnes, Opt. Mater. 26 (2004) 333. N.P. Barnes, Opt. Mater. 27 (2005) 1653. N.P. Barnes, B.M. Walsh, IEEE J. Quantum Electron. 35 (1998) 101. B.M. Walsh, N.P. Barnes, R.L. Hutcheson, R.W. Equall, IEEE J. Quantum Electron. 37 (2001) 1203. [10] B.M. Walsh, N.P. Barnes, R.L. Hutcheson, R.W. Equall, B. Di Bartolo, J. Opt. Soc. Am. B 15 (1998) 2794. [11] W.B. Grant, in: F.J. Duarte (Ed.), Tunable Laser Applications, Marcel Dekker, New York, 1995, p. 241.

[6] [7] [8] [9]

You might also like