Design and Construction of a High Energy, High Average Power Nd:Glass Slab Amplifier. Dale Martz Department of Electrical & Computer Engineering

Design and Construction of a High Energy, High Average Power Nd:Glass Slab Amplifier Dale Martz Department of Electrical & Computer Engineering 7/19/2006

Outline Introduction Nd:Glass Slab Nd:Glass Material Properties Slab vs. Rod Geometry Discussion of the Rod Geometry Advantages of the Slab Design of the Amplifier Head and Support Systems Mounting Cooling Design of the Pulse Forming Network (PFN) Design Specifications & Considerations Simmering of Lamps Results Single Pass Gain Conclusions Future Work 2

Introduction EUV lasers require high energy pumping by optical lasers Ex: 13.2 nm Ni-Like Cd Laser* 1 µw @ 5 Hz 300-350 mj 120 ps pre-pulse 1 J 8 ps heating pump pulse Current Ti-Sapphire (800 nm) pump laser has 3 stages of amplification 3 rd stage is pumped by a Nd:YAG laser Infrared radiation (1064 nm) up-converted to 2 nd harmonic 5 J of green light (532 nm) Repetition rate of 5 Hz *Weith et al. - Opt. Lett. 31, 1994-1996 (2006) 3

Replacement Pump Laser Schematic 4

Slab Amplifier Specifications Must amplify infrared radiation at 1053 nm to be doubled to 527 nm Green pump should have 14-18 J per arm (both sides of the Ti- Sapph should be pumped) Infrared output for one arm should have 20-25 J of energy A gain of 3.3x per pass has been shown to produce the desired energy after eight passes* Input beam to the slab will be 12 cm x 8 mm 60 mj with a 20 ns pulse width FWHM at 1053 nm and a 1 nm bandwidth Must be able to operate at > 1 Hz *Dane et al. Journal of Quantum Electronics 31, 148-163 (1995) 5

Nd:Glass Similar to Nd:YAG Four level system Has an amorphous structure Not crystalline Glass has high energy storage potential Large Volume Small stimulated emission cross-section 3.5x10-20 cm 2 6

Why a Slab Geometry? Generation of good beam quality at high energies and high average power Due to increased repetition rate of pumping, temperature gradients across the cross-sectional area of gain medium develop Thermal focusing Stressed induced biaxial focusing Stressed induced birefringence resulting in depolarization 7

Thermal Loading in a Rod Temperature Distribution T(K) Temperature Gradient G(10 4 K/m) A rod with 14 cm 2 cross-sectional area, with 1250 W heat loading and cooling to 35 o C on the outer surface Temperature differential of 485 o C 8

Change in the Index of Refraction Due to Temperature Change Given the temperature distribution modeled for a 1250 W heat load the change in the index of refraction can be calculated dn dt n( x) = no + *( T ( x) To ) dn dt Where is the temperature coefficient of refractive index For the glass that is used, dn dt 6 2*10 o / C 9

Thermal Focusing (Lensing) Given the temperature distribution modeled for a 1250 W heat load the change in the index of refraction can be calculated dn n( x) = no + *( T ( x) To ) dt Using this index profile, Hecht gives an expression to find the focal length of the thermal lens 2 2 n ( r) = n ( r + f f ) / d max * Approximation assuming focal length much larger than the rod length The resulting focal length that fits this profile is: f = 52 cm from the end of the rod (92 cm from the beginning of the rod) 10

Thermal Focusing (Lensing) - continued Because the resulting focal length of 52 cm is on the order of the length of the rod (40 cm), ray tracing was employed to verify the example The focal length was found to be: f = 45 cm from the end of the rod (85 cm total) - lensing is severe 11

Thermal Loading in a Slab A slab with 14 cm 2 crosssectional area 1250 W heat loading equal to the heat loading on the rod Cooling to 35 o C on the pump faces Temperature differential of 38 o C Pump Faces 12

Discussion of Phase-Front Aberrations As the rod caused spherical focusing, the slab will cause cylindrical focusing To avoid aberrations and focusing effects, the beam s phase-front should pass through an averaged (non-uniform) temperature/stress environment The slab geometry allows this by letting the beam take a zig-zag optical path through the gain medium 13 Pump Faces

APG-1 from Schott North America APG-1 is an advanced phosphate based laser glass 40 cm x 14 cm x 1 cm Upper-level lifetime = 361 µs Lower-level lifetime = 192-380 ps Emission peak at 1053.9 nm Emission bandwidth of 27.8 nm Index of refraction n = 1.526 The doping of the glass is 2.7% by weight of Nd 3-3.5 x10 20 Nd atoms/cm 3 Thermal Cond. = 0.78 W/m*K 14

Use of Flashlamps to Pump Emission spectrum of the flashlamp (Fenix Tech.) Absorption spectrum of the Nd:Glass slab (Schott N.A.) 15

Frame and Mounts for Slab Frame, window seals and mounts are made from titanium Windows form thin 2.5 mm cooling channels Brass seals do not touch slab, but provide a method to seal with an o-ring Bottom & Top seals are made from Delrin plastic Water enters and exits through these pieces *Designed by Dave Alessi 18

Reflector Cavity Cavity Made from Delrin plastic Defuse Reflector Made from Spectralon TM, resistant to deionized water Flow tubes Provides cooling channel for flashlamps Doped with cerium to block UV light 19

Parasitic Oscillations Absorbing Glass Fits inside titanium mount Absorbing glass n = 1.524 3 mm thick HA-30 (Hoya Corp.) Attached via index of refraction matched elastomer Also used to attach the titanium braces to the slab Allows for a mechanical cushion between the metal and the glass Entrance faces are parallel but tilted 1.5 degree wedge 20

Nd:Glass Slab Amplifier Head 21

Cooling Specifications Use of deionized water as a coolant c p = 4.184 J/g o C - specific heat µ = 0.00764 g/cm*s - dynamic viscosity Chosen flow rate of (m p ) 3 L/s (kg/s) or 180 L/min Corresponds to the removal of 1250 W of heat P = m * c * t Should keep the temperature differential across the 14 cm dimension of the slab close to 0.1 o C Reynolds number ratio of the inertia and viscous forces Turbulent if > 2300 p p Our channel is 10,870 sufficient Re = 4*m p /(µ*w p ) - w p = 2*(0.25 + 36) = 72.5 cm 22

Fluid Flow p total 2 m of 1 tube 90 smooth bend p = 3. 50 psi 18psi( w /1" tubing) Entrance, expansion loss p 0. 28 psi 90 degree sharp turn at entrance p 0. 22 psi smooth bend p 0. 28 psi Small entrance expansion p 0. 58 psi Slab cooling channels p 0. 18 psi Lamp water jackets p 1. 75 psi reflector cavity left side 90 smooth bend and splitter p 1. 11psi reflector cavity right side smooth bend p 0. 28 psi 90 smooth bend Pump Heat Exchanger p = 4 psi 3 m of 1 tube p = 5. 25 psi 90 smooth bend 23

Cooling Test Cooling of a un-doped glass slab installed in the amplifier head (2.5 kj/shot) 24

Pulse Forming Network (PFN) A simple RLC circuit Pulse requirements 270 µs current pulse 2.5 kj/pulse electrical energy per pulse to all four lamps Critically damped to ensure minimization of ringing The amplifier has four lamps to pump both sides of the slab Two PFN units for one amplifier head (one PFN unit for two lamps in series) Over designing the PFN unit: 2 kj/pulse (4 kj/pulse total) For a 145.16 µf Capacitor, 5.3 kv charge voltage corresponds to 2 kj 25

Variable Resistance Variable resistance Two lamps in series R = k / o I or V = ko * I Each lamp is filled with 170 Torr of Xe w/ 18 mm bore gives an arc length of 15.35 in for one lamp or 30.7 in. for two with a k o = 22.82, or 45.64 when running two in series* *Information given by Fenix Technology (vendor of the flashlamps) 26

Simulation of PFN 27

Lighting and Simmering Problem: 1 kv per inch of arc length required to light lamps Solution: Creation of an electric field between anode and the wall of the flashtube Use of high turn-ration pulse transformers (1:36) to convert -520 V pulses to -18.7 kv 20-30 Hz The lamps are simmered with two 60 W simmer units providing 1A of current (max. voltage 1500 V) 1.6 A boost after current pulse for 8 ms 28

Preliminary Circuit Design The switch for this RLC circuit consists of two thyristors in series from Dynex Semi- Conductor (DCR1050F) Rated for 4 kv 1 ka of current RMS 12 15 ka nonrepetitive surge currents 29

Measured PFN Pulse 30

Problem with Blow Out 31

Schematic of PFN w/ Added High- Voltage Transformer 32

Resulting Pulse with HV Transformer 33

Resulting Pulse with HV Transformer 34

Power Delivered to the Flashlamps Not the energy stored in capacitor Tolerance of components Loss in components increases as charge voltage increases Lamp resistance gets small Inductor Wire 35

Single Pass Gain Results Experimental Setup: 6 mm diameter, 20 ns, 1 mj pulses from a single mode Nd:YLF oscillator are used 38

Measured Single Pass Gain Compared to similar work done by: Dane et al. Journal of Quantum Electronics 31, 148-163 (1995) 39

Spatial Gain Results Along 14 cm Dimension 40

Future Work Obtain four passes of amplification Preliminary results are advantageous Possibly phase-conjugate after four passes Achieve 20-25 J output after eight passes Successfully double to 527 nm Operate amplifier at high average power (>1Hz) Pump both sides of the Ti-Sapphire amplifier by operating two units simultaneously 41

Acknowledgements Thanks to my committee members: Dr. Jorge Rocca, Dr. Siu Au Lee and Dr. Carmen Menoni I would also acknowledge everyone who has helped me during my time at the lab David Alessi, Mike Grisham, Scott Heinbuch, Brad Luther, Paul Platte, Mike Purvis, Brendan Reagan and David Springer. 42

Questions? 43