Two-step excitation and blue fluorescence under

Transcription

1 Vol. 3, No. 11/November 1986/J. Opt. Soc. Am. B 1519 Two-step excitation and blue fluorescence under continuous-wave pumping in Nd:YLF T. Y. Fan and Robert L. Byer Edward L. Ginzton Laboratory, Stanford University, Stanford, California 9435 Received April 18, 1986; accepted July 24, 1986 Near-UV and blue fluorescence from the 4 D3/ 2 and 4 D5/ 2 manifolds in Nd:YLF has been observed at room temperature under cw pumping by a Rhodamine 59 dye laser. Excitation to these manifolds is attributed to two-step excitation involving excited-state absorption from the 4F3/ 2 metastable level. A similar phenomenon has also been observed in Nd:YAG and Nd:glass. The effective excited-state absorption cross section is measured to be (2 1) X 1-2 cm 2 at nm in the ir polarization, and the peak effective stimulated emission cross section is measured to be 5 X 1-2 cm 2 at nm, also in the r polarization. Estimated laser threshold at nm for two-step pumping at nm is 7 mw. INTRODUCTION There are a number of processes besides fluorescence and one-photon absorption that can occur in rare-earth-ion- and transition-metal-ion-doped solids. Two-step excitation and energy transfer are examples of processes that can be important in laser operation. For example, concentration quenching in Nd3+-doped laser materials is an energy transfer phenomenon that reduces the lifetime of the 4F 3 / 2 upper laser level. Energy transfer upconversion (ETU) and twostep excitation have also been used to create visible sources, both incoherent and coherent, from infrared pump photons. 1, 2 In two-step excitation, a single photon is absorbed followed by relaxation to an excited metastable energy level. This is followed by absorption of a second photon from the excited state, excited-state absorption (ESA), to an even higher lying level. This is followed by relaxation, either radiative or nonradiative. In ETU, two nearby ions in excited states interact simultaneously, causing one of the pair to be excited to an even higher-lying level and causing the other to relax. Both of these processes are interesting techniques for obtaining higher-energy photons from lower-energy pump photons because of their large effective nonlinearities.' We have recently noted two-step excitation as well as ETU in Nd:LiYF 4 (YLF).3 This paper will discuss two-step excitation that induces near-uv and blue emission under cw resonant pumping. These processes were discovered while Nd:YLF cw resonantly-pumped miniature lasers were being investigated. 4 Two-step excitation and ETU have been predominantly studied in rare-earth ions such as Er3+, which have long (>1-msec) metastable level lifetimes, but similar processes have also been noted in Nd3+-doped materials. For example, blue and near-uv emission has been observed by two-step excitation in Nd3+-doped compounds such as Nd:YAG (Ref. 5) and Nd:LaF 3 (Ref. 6) under pulsed excitation. However, in pulsed excitation the dynamics of the processes can be quite different from the cw regime because peak pump powers can be higher and pulse times can be much shorter than characteristic times such as metastable level lifetimes. In the case of Nd:YAG, the pump wavelength is different from that in this paper, and consequently the energy levels in two-step excitation are also different. In the case of LaF 3, ETU is the dominant process, but twostep excitation is noted at temperatures approaching that of liquid helium. Upconverted blue fluorescence has also been noted in some fluoride crystals. 7 In Nd:YLF, two-step excitation and subsequent blue fluorescence have been used as a technique to investigate Nd 3 + ion pairs in YLF at low temperatures (<4 K), 8 and the blue fluorescence has been observed under ruby-laser pulsed pumping by a two-photon process. 9 In this paper, blue and UV fluorescence is investigated, and the excitation mechanism is determined. EXPERIMENTAL APPARATUS Nd:YLF is a uniaxial solid-state laser material. Its onephoton absorption spectra, infrared fluorescence spectra, and concentration quenching were previously characterized.'( From previously published data, the dopant concentration was 2 at. % Nd given the measured fluorescence lifetime of 44 Asec for 4 F3/ 2 manifold in our sample.' However, more recent measurements of concentration indicate that the r-polarized absorption line at nm has an absorption coefficient of 1.2 cm-' for 1 at. % Nd doping." By this standard, the concentration of our sample is 1.7 at. % Nd. The experimental apparatus used to record the emission and excitation spectra consisted of an argon-ion-laserpumped cw Rhodamine 59 dye laser for excitation, a lens to focus the pump beam into the sample, a 3134 photomultiplier tube for detection, a Chromatix CT-13 1-m monochromator with a 18-line/mm grating for wavelength selectivity, and a lock-in amplifier for signal extraction. The dyelaser beam was typically chopped at -3 Hz with '3% duty cycle. A calcite polarizer was used to select the polarization for the various measurements. The subsequent signal from the lock-in amplifier was digitized and stored in a computer. The spectral response of the detection system was normalized by using a quartz-halogen lamp.1 2 A Perkin-Elmer /86/ $ Optical Society of America

2 152 J. Opt. Soc. Am. B/Vol. 3, No. 11/November 1986 spectrophotometer was used for measuring absorption spectra. All measurements were performed at room temperature. BLUE AND NEAR-UV FLUORESCENCE The normalized fluorescence spectra for both r(e c) and a(e c) polarizations are shown in Fig. 1. There are four regions in which emission is observed that are near 358 nm, 383 nm, 412 nm, and 45 nm, which correspond to transitions from 4 D3/2 and 4D 5 / 2 manifolds to 4I9/2, 4I11/2, 4I13/2, and 4I15/2, respectively, based on energy-level positions. 9 "1 3 The 4 D3/ 2 and 4 D 5 / 2 manifolds should be thermally coupled at room temperature. In other words, the relative populations are given by a Boltzmann distribution. There is a significant discrepancy of approximately 4 cm-' between the observed energy-level positions in Refs. 9 and 13 for these two manifolds; our data are more consistent with those in Ref. 9. The branching ratio flij from the upper manifold i to a lower manifold j is given by A,. E f XIP(X)dX fi = A _ P E Aik E E >,ikp(x)d X k p k where Aik is the transition rate from the upper manifold i to a lower manifold k, and the sum on k is over all lower manifolds. Note that the sum in the denominator is equal to 1-r, where Tr is the radiative lifetime. The second part of Eq. (1) gives fij in terms of the polarized emission spectra, where IikP(X) is the polarized emission intensity per unit wavelength interval as a function of wavelength X from manifold i to manifold k, and the sum over p is over polarization. The results for fluorescence from the 4 D 3 / D 5 / 2 manifolds are shown in Table 1, where emission to states other than the 4 Ij manifolds has been ignored since no emission was observed experimentally. Electric dipole transitions were assumed so the axial spectrum is identical to that for the r polarization. In this case the sum over polarization becomes 2I(X) + IP(X). One difficulty with this measurement is that the emission to the ground-state manifold, 4I9/2, can be reabsorbed, causing the measured branching ratio to that manifold to be too small. To minimize this problem, the sample was pumped near its surface. The radiative transition rates Aij, and therefore the branching ratios, can be calculated from the Judd-Ofelt theory.' 4 "l 5 The radiative transition rate for electric dipole transitions from a state fn(asl)j) to state fn(is'ls)j/), which denote manifolds 2 S+lLj and 2 S+lLj,', is given by A = [647r4e2] V 3 (2J + 1)-' (1) where v is the transition energy in inverse centimeters, n is the refractive index, Q,\'s are the experimentally determined intensity parameters for the given ion in the given host, and the (fn(asl)jii U()fN(a'S'L')J') are reduced matrix elements. Using the reduced matrix elements,' 6 the Ox parameters for Nd:YLF,"7 and the assumption that the relative populations of 4 D3/ 2 and 4 D5/ 2 are given by a Boltzmann distribution, the calculated radiative transition rates and branching ratios to the 4I term are shown in Table 1; emission to other manifolds was ignored since none was observed. While the Judd-Ofelt theory is correct in predicting strong emission from 4 D3/ 2 and 4 D 5 / 2, the branching ratios among the 4I term are not accurate. The Judd-Ofelt theory also predicts strong emission to 4 F 5 / 2 ; however, this emission was not observed. In the case of emission from 4 F3/ 2, the Judd- Ofelt theory is more accurate in predicting branching ratios. This is most likely because one of the two fundamental I-. z zf > I - -J w a: C', -J w a: WAVELENGTH (a) T. Y. Fan and R. L. Byer I I I I I I Nd:YLF POLARIZED FLUORESCENCE ;a 11I~~~~-- - ~~~~~~~, S. (nm) 7r 39 4 where X, (2a) X=2,4,6 = n(n + 2) (b) Fig. 1. Polarized emission spectrum of Nd:YLF under pumping at (2b) nm. (a) Emission from 35 to 45 nm. (b) Emission from 45 to 46 nm.

3 Table 1. Experimentally Determined and Calculated Branching Ratios Measured Branching Judd-Ofelt Theory Lower Manifold Ratio Branching Ratio 4I9/ / I13/ IJs/ approximations in the theory is worse for emission from 4D 3 / 2 and "D 5 / 2 than for emission from 4F 3 / 2. This can be explained by the following. Electric dipole transitions are not allowed for 4f-4f transitions in a free ion because all 4fN configurations have the same parity. However, by placing the ion in a site with no inversion symmetry, opposite parity terms are admixed into the 4 fn configuration. These are typically 4fN-15d configurations, which usually have higher energies than the 4N configurations. One of the fundamental approximations used in deriving the Judd-Ofelt theory is that the energy difference between the initial manifold, 2 S+lLj, and 4fN-15d configurations and the energy difference between final manifold, 2 S'+lLj,', and the 4fN-15d configurations are equal. That is, E( 2 S+lLj) - E(4fN-15d) and E( 2 S'+lLj,') - E(4fN-15d) are taken to be equal, where E( 2 S+lLj) and E( 2 S'+lLj,') are the energies of the initial and final states, respectively, of the transition, and E(4fN-15d) is the energy of the 4fN-15d configurations. This approximation is much worse for the 4D - 4Ij transitions than for the 4F 3 /2-4Ij transitions since the former are more energetic. The fluorescence lifetime of the 4D 3 / 2 + 4D 5 / 2 manifolds was measured by chopping the Rhodamine 59 pump beam and monitoring the fluorescence around 412 nm. The observed lifetime was 23 J 2 sec. The radiative quantum efficiency 1 is given by where 1/Tr 11T + /Tnr Tf Tr Tnr (3a) + 1 (3b) Tf, Tr, and Tnr are the fluorescence, radiative, and nonradiative lifetimes, respectively. From the measured fluorescence lifetime, the multiphonon relaxation rate, and radiative transition rate calculated from the Judd-Ofelt theory, can be estimated. The multiphoton relaxation rate is given by -= B exp(-kae), (4) Tnr where B and K are characteristic of the host material. E is the energy gap from the initial state to the next lower-lying state. For the 175-cm-' energy gap from 4D 3 / 2 to the next lowest level, 2 P 3 / 2, the multiphonon relaxation rate in YLF (Ref. 18) is 3.2 X 14 sec-1. From Eq. (3b) the radiative lifetime is 9 sec compared with the radiative lifetime calculated from the Judd-Ofelt theory of 5 sec. These correspond to radiative quantum efficiencies of 3 and 5%, respectively. Vol. 3, No. 11/November 1986/J. Opt. Soc. Am. B 1521 SEQUENTIAL TWO-PHOTON EXCITATION The blue fluorescence was first observed when a cw Rhodamine 59 dye laser operating near 59 nm was used to pump a Nd:YLF laser operating at 1.47 Am. This fluorescence decreased substantially when the Nd:YLF laser was aligned, demonstrating that the process is dependent on the population of the 4F 3 / 2 manifold. This indicates that real onephoton transitions must be occurring, as opposed to virtual transitions as in two-photon absorption where intermediate states are not populated. Figure 2 shows the square root of the blue fluorescence intensity versus the pump intensity. The linear relation shows that this process involves two photons. The 4D 3 / 2 + 4D 5 / 2 to 4I13/2 emission at 412 nm is monitored as the pump intensity at nm is varied. At low pump intensities (of the order of 1 W/cm2, corresponding to -.7 on Fig. 2), a square-law dependence of the blue fluorescence is observed; at higher pump intensities, saturation effects can begin to be observed. As previously mentioned, blue fluorescence has been observed in Nd:LaF 3 under pulsed pumping in the same wavelength region. This was attributed to both two-step excitation and ETU, with ETU being the dominant process. 6 ETU in Nd:LaF 3 involves one excited ion in the 4F 3 / 2 manifold interacting with another excited ion in the 4G 5 / G 7 /2 manifold, which leads to relaxation of the first to the 4I9/2 ground manifold and simultaneous excitation of the second to the 211/2 + 4D1/ 2 + 4D 5 / 2 + 4D 3 / L15/ 2 manifolds. It is unlikely that this ETU contributes significantly to excitation in Nd:YLF in the cw regime because the 4G 5 / G 7 /2 population should be low owing to fast multiphonon relaxation from this manifold as opposed to the pulsed pumped regime where these manifolds can have significant population. Two-step excitation and ETU have different excitation wavelength and time dependences. We have used excitation spectra to demonstrate that two-step excitation is the predominant effect. The Rhodamine 59 dye-laser wavelength was tuned and the induced fluorescence monitored. The transition rate for ETU assuming dipole-dipole interaction is given by -a.8 C: 2) a,.6 a). 4_ X.4.2. W' Relative pump intensity Fig. 2. 4D 3 / 2 + 4D,5/2-4I13/2 emission as a function of the pump power at nm. The peak pump intensity and emission have been normalized to 1. The line is a least-squares fit to the lowest five data points.

4 1522 J. Opt. Soc. Am. B/Vol. 3, No. 11/November (a) T. Y. Fan and R. L. Byer as well as the one-photon absorption spectra. To avoid difficulties with normalization, the pump power was kept 4.7 constant for each data point. Clearly, the excitation spectra E do not follow the square of the one-photon absorption spectra and, in fact, some of the peaks in the one-photon absorpz tion spectra do not appear in the excitation spectra. This 3, indicates that two-step excitation is the predominant effect. UL Based on the wavelengths of peak excitation and energy- level positions, the ESA transition is 4 F 3 /2 -D 5 / 2. The two- 2 z step excitation process and subsequent fluorescence are summarized in Fig. 4. I- The ESA cross sections were also measured at the wave- length of peak excitation in each polarization. The Rhoda- 2 mine 59 dye laser was tuned to the peak of the excitation, and the beam was focused into the sample with known power and Gaussian beam waist. This causes a certain level of fluorescence in the sample. Then a He-Cd laser at 325 nm was aligned collinearly with the dye-laser beam. The amount of absorbed power at 325 nm in the sample was measured, and the level of fluorescence was noted. This permits a calibration of the measured level of fluorescence to 4 a pump rate into the 4 Dj states. Then the ESA cross section can be derived from the following. The pump rate as a function of effective ESA cross sec- H tion, esa can be calculated by using a simple rate-equation W approach. The ESA excitation rate, Wesa, is given by 28 a- Wesa(r, z) = cesan(r, z)i(r, z) (6) hvp where hvp is the pump photon energy, N(r, z) is the excitedstate population density of 4F 3 / 2, and I(r, z) is the pump Energy (cm 1 x 1-) (b) Fig. 3. Polarized excitation spectra for 4D 3 / 2 + 4D 5 / 2-4I13/2 emission of Nd:YLF and one-photon absorption spectra. (a) r polarization, (b) a polarization. WETU NN2' (5) where N 1 and N 2 are the population densities of the initial excited energy levels involved in the process. This implies that WETU is proportional to the square of the one-photon absorption coefficient as a function of wavelength. This is because the initial populations are excited by one-photon absorptions from a single monochromatic source and therefore both N 1 and N 2 are proportional to the one-photon absorption coefficient. On the other hand, in two-step excitation, two transitions, the one-photon absorption and ESA transition, must occur at the same wavelength for excitation to occur. Consequently, peaks in the one-photon absorption spectrum may not appear in the excitation spectrum if there is no or little ESA at the same wavelength. The excitation spectra were obtained by using the same experimental apparatus as that used for the fluorescence spectra, except that the Rhodamine 59 dye-laser wavelength was tuned and the induced fluorescence on the 4 D D 5 / 2-4I13/2 was monitored. Figure 3 shows the polarized excitation spectra H tt 3/2 I /2 A ~~~~~~~ 91/2 Fig. 4. Energy levels in two-step excitation and subsequent fluorescence. The pump photons are denoted by hvp, and emission from D 3 / 2 and 4D 5 / 2 by the heavy arrows. The dashed arrow indicates nonradiative decay. 8

5 intensity. The excited-state population N(r, z) in steady state can be found from the equation dn(r, z) = yi(r, z) N(r, z) dt hpp T7 where y is the one-photon absorption coefficient at the pump frequency and -r 1 is the fluorescence lifetime of the F 3 /2 manifold. The assumption of no depletion of the ground state, N(r, z) << Nt, where Nt is the total dopant concentration, has been made as well as the assumption of no depletion of the excited state by ESA, which is the condition aesan(r, z)i(r, z) N(r, z) hvp hv<.(8) Under these assumptions and with the pump being a Gaussian beam with negligible diffraction in the sample, the pump beam intensity is given by I(r, z)= 2 - exp-z), (9) 7rwp2 \ WP2J where P is the incident pump power and wp is the beam radius. Substituting Eq. (9) into Eq. (7), solving for N(r, z), and integrating Wesa(r, z) over the volume, the total pump rate, Wtot, is given by wtot 2(h) 2 Ti (7) 2 [1 - exp(-2-yl)], (1) where is the sample length. This is the expression needed to relate the total ESA pump rate to (-esa. By using Eq. (1), the pump rate calibration from the He-Cd-laser pump, and the signal for a given dye-laser incident pump power, the ESA cross sections were calculated to be esa = (2 1) X 1-2 cm 2 in the r polarization at nm and 'Tesa = (9 4) X 1-21 cm 2 in the a- polarization at nm. These wavelengths are at the peaks of the excitation spectra for their respective polarizations. DISCUSSION It may be possible to use the 4 D 3 / D 5 / 2-4 Ij transitions for laser action. It may even be possible to make a twostep-excitation-pumped laser. This type of laser offers an alternative method to harmonic or sum-frequency generation for obtaining shorter wavelengths of light. The potential advantage is that the effective nonlinearity of two-step excitation is orders of magnitude larger than that for harmonic generation' because real, resonant transitions are involved as opposed to virtual, nonresonant transitions. ETU, which also has high effective nonlinearities, was previously used to provide nearly all the pump power for a BaYi. 2 Yb 75 Er 5 F 8 visible laser at 67 nm under lamp pumping at 77 K 2. However, the threshold of this laser was 17 J. With resonant pumping, it is easier to take advantage of these high nonlinearities because of the high pump densities achievable. Of course, the main disadvantage is that this process depends on resonances, so it is less general. The laser threshold for a two-step-excitation pumped laser in Nd:YLF can be calculated once the stimulated emission cross section is known. The stimulated emission cross section can be derived from Vol. 3, No. 11/November 1986/J. Opt. Soc. Am. B 1523 the fluorescence spectra. The effective stimulated emission cross section, ae, for a uniaxial medium is given by'9 2 3X 2 ep(x) = rn CTrIN IPc(X)c I (Ila) where p denotes polarization, c is the speed of light, and IN is given by (lib) IN = E q(x)dx, q where the sum is over polarization. With n = 1.5, Tr = 9!Lsec, and the assumption that the transitions are electric dipole, the peak effective stimulated emission cross section in each polarization for transitions to each of the 4Ij manifolds is shown in Table 2. It is difficult to assess the error in this calculation, but this cross section is generally difficult to measure accurately. The largest error term is probably Tr, which was derived from the measured fluorescence lifetime and the multiphonon relaxation rate in YLF. This multiphonon relaxation rate could be in error by up to a factor of 2, which would lead to a large change in Tr. One way to determine the accuracy of the cross section is to compare the absorption cross section with the emission cross section. This can be done for the line at nm in the ir polarization that appears in both absorption and emission. This transition is between crystal field splittings Z 2 at 132 cm-' in 4I9/2 and L, at 2813 cm-' in 4D 3 / 2. The absorption coefficient is 2.5 cm-' for 1.7 at. % Nd doping. This gives an effective cross section of 1.1 X 1-2 cm 2. The effective cross section is related to the actual cross section and absorption coefficients by (e = /Ntot = af, (12) where Nt t is the total population of the ground-state manifold, ae is the effective cross section, a. is the actual cross section, and f is the fractional occupation of the crystal field splitting involved in the transition given by a Boltzmann distribution. In Eq. (12), degeneracy has been ignored because all the Nd3+ crystal field splittings in YLF have a degeneracy of 2; consequently all degeneracy factors cancel. So the actual absorption cross section at nm is 4.8 X 1-2 cm 2. The effective stimulated emission cross section is 1.2 X 1-2 cm 2, so the actual cross section is 2.6 X 1-2 cm 2, based on the Boltzmann distribution calculated from the energy levels of 4D3/ 2 and 4D 5 / 2. However, as previously noted, this emission is back to the ground manifold, so reabsorption of the emission causes the measured intensity to be too low. The maximum estimated error is 25%, based on an Table 2. Peak Effective Stimulated Emission Cross Sections Wavelength Lower Cross Section (nm) Manifold Polarization (cm 2 ) I9/2 a.5 X j9/2 7r 1.2 X I11/2 a.8 X I11/2 r 2.1 X I13/2 r 5. X I13/2 a 1.6 X I15/2 a 1.6 X I15/2 r 3. X 1-2

6 1524 J. Opt. Soc. Am. B/Vol. 3, No. 11/November 1986 T. Y. Fan and R. L. Byer Z- z Z I I I I II I I I I Nd :YAG FLUORESCENCE Using Eq. (1), which gives the pump rate assuming no saturation, assuming that the pump at nm is double passed, and taking the values 3 =.5 and wp = wo = 2,m, the calculated threshold at nm is 7 mw. This result also assumes no ESA of the nm radiation. While this incident intensity is in the regime where saturation effects are beginning to occur, this calculation is only approximate anyway because of the large uncertainty on the ESA cross section and stimulated emission cross section. Finally, just as with concentration quenching, two-stepexcitation processes such as that characterized here in Nd:YLF probably occur to varying degrees in all Nd 3 +- doped compounds. For example, we have also observed near-uv and blue emission from Nd-doped LHG-8 phosphate glass under cw Rhodamine 59 dye-laser pumping and from Nd:YAG under cw Rhodamine 59 dye-laser pumping at nm. The latter emission, which was previously noted, 22 is shown in Fig. 5 and can be assigned to transitions from the upper states 2 P 3 / 2 and 4 D 3 / 2 to the 4I term. While we have not studied the mechanisms in detail, we believe that the excitation in both cases can also be attributed to two-step excitation Fig. 5. Near-UV and blue emission in Nd:YAG pumping at nm. This has not been normalized for the spectral response of the detection system. absorption coefficient of 2.5 cm-' and a distance of propagation through the material of.1 cm. Without taking reabsorption into account, the absorption and emission cross sections differ by a factor of 1.8. The calculation of the absorption cross section also has a source of error that is the concentration of Nd; however, it appears that the estimate of radiative quantum efficiency and therefore the stimulated emission cross section may be low based on our comparison. Now a threshold for a two-step excitation pumped laser can be estimated. Assuming a Gaussian pump and negligible diffraction in the laser medium, the pump power, Pth, required to reach threshold for the typical one-photon absorption pumping assuming that one absorbed photon yields one excited ion in the upper laser level is given by2l Pth = hp [72 (WP2 + wo2)], (13) where w is the laser cavity radius and is the round-trip cavity loss. This can be converted to a pump rate by dividing by the pump photon energy. In the case of two-step excitation pumping, the term wp 2 is replaced by wp 2 /2 because the pump rate for two-step excitation goes as the intensity squared, so the radial pump distribution goes as exp(-4r 2 /wp 2 ) as opposed to exp(-2r 2 /wp 2 ), as in one-photon pumping. So the expression for threshold pump rate, Wth, becomes Wth = I [ ( 2 /2 + Wo2)]. (14) SUMMARY We have characterized near-uv and blue fluorescence in Nd:YLF under cw Rhodamine 59 dye-laser pumping. The emission is due to 4 D 3 / D 5 / 2 to 4 Ij transitions, and it appears that excitation to these levels is by a two-stepexcitation process. It would be interesting to compare these cw results with pulse pumping to characterize the differences between the two regimes, because the results obtained in Nd:LaF 3 in a pulsed regime suggest that the dynamics of the process may be quite different. While measurements and calculations indicate that a two-step-excitation pumped laser in Nd 3 + doped systems is a possibility, other rareearth- or transition-metal-ion-doped crystals may be more interesting for this type of laser if an appropriate system can be found because the long metastable lifetimes in some of the other ions will permit more efficient pumping. ACKNOWLEDGMENTS The authors would like to thank the U.S. Office of Naval Research and NASA for funding this research. They would also like to thank H. P. Jenssen for providing the Nd:YLF used in this work. REFERENCES 1. F. Auzel, "Materials and devices using double-pumped phosphors with energy transfer," Proc. IEEE 61, (1973). 2. L. F. Johnson and H. J. Guggenheim, "Infrared-pumped visible laser," Appl. Phys. Lett. 19,44-47 (1971). 3. T. Y. Fan and R. L. Byer, "Nonradiative processes and blue emission in Nd:YLF," in Digest of Topical Meeting on Tunable Solid State Lasers (Optical Society of America, Washington, D.C., 1986), pp T. Y. Fan, G. J. Dixon, and R. L. Byer, "Efficient GaAlAs diode laser pumped operation of Nd:YLF at 1.47,4m with intracavity doubling to nm," Opt. Lett. 11, (1986). 5. G. J. Quarles, G. E. Venikouas, and R. C. Powell, "Sequential two-photon excitation processes of Nd3+ ions in solids," Phys. Rev. B 31, (1985). 6. B. R. Reddy and P. Venkateswarlu, "Energy up-conversion in LaF 3 :Nd 3 +," J. Chem. Phys. 79, (1983). 7. L. Esterowitz, A. Schnitzler, J. Noonan, and J. Bahler, "Rare earth infrared quantum counter," Appl. Opt. 7, (1968). 8. R. B. Barthem, J. C. Vial, and F. Madeore, "Two-photon absorption by ion pairs in solid as a technique for homogeneous linewidth measurement," J. Lumin. 34, (1985). 9. Z. Song, S. Lia, Y. Gui, J. Jiang, D. Hua, S. Wang, and Y. Huo, "The electronic Raman scattering and two-photon fluorescence in LiYF 4 :Nd 3 +," Acta Phys. Sin. 33, (1984). 1. A. L. Harmer, A. Linz, and D. R. Gabbe, "Fluorescence of Nd 3 +

7 in lithium yttrium fluoride," J. Phys. Chem. Solids 3, (1969). 11. H. P. Jenssen, MIT Crystal Physics Laboratory, Cambridge, Mass (personal communication). 12. R. Stair, W. E. Schneider, and J. K. Jackson, "A new standard of spectral irradiance," Appl. Opt. 2, (1963). 13. A. A. S. da Gama, G. F. de Sa, P. Porcher, and P. Caro, "Energy levels of Nd 3 + in LiYF 4," J. Chem. Phys. 75, (1981). 14. B. R. Judd, "Optical absorption intensities of rare-earth ions," Phys. Rev. 127, (1962). 15. G. S. Ofelt, "Intensities of crystal spectra of rare-earth ions," J. Chem. Phys. 37, (1962). 16. H. Jagannath, D. Ramachandra Rao, and P. Venkateswarlu, "The radiative transition rates of Nd 3 + in LaF 3," Indian J. Phys. 52B, (1978). 17. W. F. Krupke, "New rare earth quantum electronic devices: a calculational approach," in IEEE Region Six Conference Re- Vol. 3, No. 11/November 1986/J. Opt. Soc. Am. B 1525 cord (Institute of Electrical and Electronics Engineers, New York, 1974), pp H. P. Jenssen, "Phonon assisted laser transitions and energy transfer in rare earth laser crystals," Tech. Rep. 16 (MIT Crystal Physics Laboratory, Cambridge, Mass., 1971), p S. Singh, R. G. Smith, and L. G. VanUitert, "Stimulated-emission cross section and fluorescent quantum efficiency of Nd 3 + in yttrium aluminum garnet," Phys. Rev. B 1, (1974). 2. B. F. Aull and H. P. Jenssen, "Vibronic interactions in Nd:YAG resulting in nonreciprocity of absorption and stimulated emission cross sections," IEEE J. Quantum Electron. QE-18, (1982). 21. K. Kubodera, K. Otsuka, and S. Miyazawa, "Stable LiNdP 4 O 12 miniature laser," Appl. Opt. 18, 884 (1979). 22. Yu. K. Voron'ko, B. I. Denker, V. V. Osiko, A. M. Prokhorov, and M. I. Timoshechkin, "X-ray fluorescence of rare-earth ions in Y 3 A crystals," Sov. Phys. Dokl. 14, (197).