VOL UME 18 NUMBER 1

APPLIED PHYSICS LETTERS

1 JANUARY 1971

BIREFRINGENCE COMPENSATION AND TEMoo MODE ENHANCEMENT IN A Nd: YAG LASER W. C. Scott and M. de Wit
Texas Instruments Inc. , Dallas, Texas 75222 (Received 18 September 1970)
A Nd: YAG laser has been operated under strong thermal focusing and birefringent conditions and has yielded both linearly polarized output with negligible power loss and TEMoo output of 50-70% of the multimode power. The laser consisted of two Nd: YAG rods separated by a 90 crystal quartz polarization rotator together with a convex mirror and used no special cooling techniques. The performance of this laser, which agreed with theoretical predictions, indicates that this design approach may be generally applicable to the problem of thermally induced aberrations in such systems.

Thermally induced optical distortions in laser rods have been the subject of numerous investigations. 1-1 Basically, a long cylindrical laser rod with uniform internal heat generation and constant surface temperature assumes a quadratic radial temperature dependence. This leads to similar behavior in both the index of refraction and the thermal strain distribution. The former leads to simple thermal focusing, while the latter leads to bifocusing and birefringence via the photoelastic effect. These effects, in turn, severely degrade laser performance both in the fraction of multimode power obtainable in the TEMoo mode and in the fraction of unpolarized multimode power which can be obtained linearly polarized. Osterink and Foster have recently published extensive theoretical and experimental results on the extent of these phenomena. 1 Various methods of compensation for these effects have been suggested and/or tried to date. Compensation for simple thermal focusing via a convex mirrorz or a concave surface ground directly on the laser rod 5 have been employed with some degree of success to enhance TEMoo output. In addition, birefringence compensation techniques using in-cavity polarization rotation have been suggested. 8 This letter is to report the simultaneous use of a convex mirror and an in-cavity polarization rotator to achieve large increases in the fraction of multimode power obtainable in the TEMoo mode and as linearly polarized output power. The resonator configuration employed, consisting of a 50-cm convex mirror, two flat-ended 3 x 50- mm krypton lamp pumped Nd : YAG rods separated by a 90 crystal quartz rotator, a Brewster plate, and a 10-m output mirror, is shown in Fig. 1. The component spacings and lamp pumping powers were determined via a specific procedure of experimental measurements and theoretical calculations which will now be summarized. Direct measurements of the average back focal distance of each rod versus lamp input power were first taken using an expanded parallel beam from a 6328-A helium-neon laser. These measurements
0

yielded the average radial index variation coefficient bz as a function of pumping power, where n(r) =no(l- bzr z/2) via the relation
d=cot[L(b z)1/Z]!no(b z)1/Z. (1)

This in turn determined the optical transfer matrix

Equation (2), with the values for b 2 adjusted for birefringence, 8 and incorporated into a paraxial ray analysis of the resonator as reviewed by Kogelnik and Li, 9 was then used to calculate a resonator configuration with the following specific set of properties: (1) The theoretical beam width of the Gaussian mode was approximately two-thirds of the rod diameter inside both laser rods; (2) a hypothetical "mean" ray, i. e., a ray with focusing properties midway between those of radial and those of tangential polarization, formed a waist halfway between the two rods; (3) the complex beam parameters associated with the two polarization "modes" of the resonator, i. e., radial in one rod, tangential in the other, and vice-versa, were matched as well as possible throughout the cavity. The first property tends to fill both laser rods with the TEMoo mode, thus enhancing its power at the expense of higher orders, while the second and third properties tend to nullify the eff ects of birefringence. The second condition causes each part of the beam to pass through nearly identical
BREWSTER

R'-50cm

PLATE

FIG. 1. Schematic diagram of birefringence-compensated laser. 3

Downloaded 10 Jan 2011 to 200.17.113.203. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

VOLUME 18 NUMBER 1

APPLIED PHYSICS LETTERS

1 JANUARY 1971

22
1

" ' - " - - - - - - - ........ ....

'---- ................. ...

2.0

;:

,. ...
I-

1.8

rt'

1.6

14 I. 2

L----=~___=_"__-'--=

........_
1.8

__'_''--...u.--'

1.0

1.4

2.2

P (kWATT) B

FIG. 2. Laser output versus input power to pump lamps. Dotted lines are plane-polarized power out; solid lines are the ratios PJPII

regions of the two rods with the result that the birefringence induced by one rod is cancelled, to first order, by the other. The third condition is to make the mode profile inside the resonator as independent of polarization as possible, whereupon the maximum uncoupling of Cartesian polarizations will result. This, in turn, allows the Brewster plate to discriminate against one linear polarization without simultaneously attenuating the other via the birefringent coupling mechanism. While expressions for the proper resonator dimensions and pumping powers can be obtained analytically, the actual calculation was performed more readily on a digital computer. Computed optimum component spacings are shown in Fig. 1, and computed optimum lamp input powers were PA =1650 Wand PB =1450 W. These pumping levels corresponded to rod focal lengths of the order of 50 cm and multi mode output power, using closely spaced long-radius mirrors, of 9 W. Experimental results of the birefringence compensation achieved with this system are shown in Fig. 2. The resonator was set up with the computed optimum dimensions and optically aligned. The two lamp input powers were varied independently over a range of more than a kilowatt while the laser output power polarized parallel and perpendicular to the sense of the Brewster plate was measured. Under these conditions the laser operates in a few low-order modes, TEMpI, with P and 1 generally less than 2. The solid contour lines of perpendicular output power divided by parallel output power show a well-defined minimum at input power settings close to the computed optimum. At the optimum settings this ratio was approximately 1. f11o. Furthermore, at this setting the polarized output power was ;;. 95% of the unpolarized output power obtained with the Brewster plate removed. It will be noted from Fig. 2 that, if the input powers were varied such that the rods remain thermally "matched", the degradation in perform4

ance is much slower than if they were varied separately. Furthermore, the decrease in total power at high pumping levels is consistent with thermally induced resonator instability. Higher powers require a more convex mirror for a stable resonator. Finally, with the Brewster plate removed, an aperture was introduced into the cavity to force TEMoo mode operation. Small variations in pumping powers, together with small adjustments in 1A and lB' were made while TEMoo power was measured. The best results obtained were 6 W TEMoo at pump levels of approximately 1600 W to each lamp. This pumping level yields 9 W multimode power in a resonator with closely spaced long-radius mirrors. With the Brewster plate back in, the best results were 5t W TEMoo at pump levels yielding 11 W multimode. In conclusion, a two rod Nd : YAG laser has been operated under strong thermal focusing and birefringent conditions and has yielded both linearly polarized output with negligible power loss and a large fraction (50-70%) of its output in the TEMoo mode. These results were obtained with standard flat-ended rods and without any special cooling techniques. While the 1600-W pump powers, as dictated by the 50-cm mirror, used in this experiment were low, the energy density in the 3-mm rods was high. Quarter-in. -diam rods would require "" 7 kW of pump power to produce equivalent internal heating (in this case ""125 W/cm3). Similarly, the low efficiency arose primarily from a measured 6-7% one-way loss in each laser rod. The high internal losses of these particular rods were also demonstrated by the fact that the optimum output mirror for either single rod was 15% transmitting. These facts indicate that compensation to a similar degree should be obtainable both in higher-power Nd: YAG lasers and in other laser systems as well. The technical assistance of W. J. Koenning is gratefully acknowledged.
IF. W. QueUe, Jr., Appl. opt. ,2,633 (1966). 2C.M. Stickley, IEEE J. Quantum Electron. QE-2, 511 (1966). 3E . P . Riedel and G.D. Baldwin, J. Appl. Phys. 38, 2720 (1967). 4G. D. Baldwin and E. P. Riedel, J. Appl. Phys. 38, 2726 (1967). 5L.M. OsterinkandJ.D. Foster, Appl. Phys. Letters 12,128 (1968). -sL. M. Osterink and J. D. Foster, Tech. Report No. RADC-TR-68-182, 1968 (unpublished). 7J .D. Foster and L.M. Osterink, J. Appl. Phys. 41, 3656 (1970). 8L.M. Osterink and J.D. Foster, IEEE International Electron Devices Meeting, Washington, D. C. , 1969 (unpublished) . 9H. Kogelnik and T. Li, Appl. Opt. ,2,1550 (1966).

Downloaded 10 Jan 2011 to 200.17.113.203. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions