THE CATHOLIC UNIVERSITY OF AMERICA

Technique for Reducing Laser Beam Divergence of Intra-Cavity Nonlinear Conversion

A DISSERTATION

Submitted to the Faculty of the

Department of Electrical Engineering and Computer Sciences

School of Engineering

Of The Catholic University of America

In Partial Fulfillment of the Requirements

For the Degree

Doctor of Engineering

By

Alan D. Hays

Washington, D.C.

2014

Technique for Reducing Laser Beam Divergence of Intra-Cavity Nonlinear Conversion

Alan D. Hays, PhD

Director: Scott Mathews, PhD

During the past few years the Monoblock laser has become the laser-of-choice for laser range-finders. It’s eye-safe 1570 nm emission, high pulse energy, simple construction, and high efficiency, when pumped by a laser-diode stack, provide advantages that are not available with other laser types. Although the relative divergence of the Monoblock output beam is large, it can be reduced to the required <1 mR using a telescope with a large magnification. This solution, however, is not acceptable for applications where the laser and telescope size must be kept to a minimum.

A simple and compact technique for achieving significant reduction in the Monoblock beam divergence using a partial reflector that is placed a short distance from the optical parametric oscillator (OPO) has been developed. Using an ultra-compact 38 mm Monoblock with a 10 mm long KTP OPO, we achieved a beam divergence of <4 mR, corresponding to a

>2.5 X reduction from the unmodified laser. Modeling and experimental results are presented detailing the theory and performance for this novel technique.

This dissertation by Alan D. Hays fulfills the dissertation requirement for the doctoral degree in Electrical Engineering approved by Dr. Scott A. Mathews, as Director, and by Dr. Jessica Ramella-Roman, and Dr. Lew Goldberg as Readers.

______

Dr. Scott A. Mathews, Director

______

Dr. Jessica Ramella-Roman, Reader

______

Dr. Lew Goldberg, Reader

ii Table of Contents

1. Introduction ...... 1

2. Monoblock ...... 4

2.1 Optical Parametric Oscillators ...... 13

2.2 Diode-Pumping of Solid-State Laser ...... 17

3. Etalon Theory ...... 22

3.1 Composite Output Coupler ...... 27

4. Techniques for Transverse Mode Discrimination ...... 35

4.1 Variable Reflectivity Mirrors ...... 35

4.2 Polarization Dependent Output Couplers ...... 38

5. Results and Observations ...... 41

5.2 Experimental Results ...... 47

6. Feedback mirror applied to Nd:YAG laser ...... 63

7. Conclusions and Future Work ...... 70

8. Bibliography ...... 73

iii List of Figures

Figure 1 - Photograph of Monoblock assembly ...... 5

Figure 2 - Schematic of Monoblock detailing optical components and surface coatings ...... 5

Figure 3 CAD model of assembly and test fixture ...... 6

Figure 4 - SMC Output energy distribution for lot shipped ...... 11

Figure 5 - SMC Divergence distribution for lot shipped ...... 11

Figure 6 - Tuning curves for x-cut KTP OPO [19], units shown with labels versus angle ...... 15

Figure 7 - Absorption spectrum of Nd:YAG [24] ...... 18

Figure 8 - Absorption convolution with single and three color pumps ...... 20

Figure 9 - Laser diode array specifications, [27] ...... 21

Figure 10 - Laser diode array spectrum and I-P curve ...... 21

Figure 11 - Schematic of etalon ...... 23

Figure 12 - Transmission through etalon versus phase R=0.7 ...... 24

Figure 13 - Transmission of divergent light through etalon ...... 25

Figure 14 - Schematic of OPO and Feedback Mirror ...... 28

Figure 15 - Effective reflectivity of composite mirror versus angle, in radians. OPO mirror ...... 29

reflectivities of 60% and feedback mirror of 40%, length 10 mm ...... 29

Figure 16 - Minimum reflection node versus composite mirror length ...... 31

Figure 17 - Angular dependence of the effective reflectivity for three composite mirror ...... 32

Figure 18 - Angular dependence of the effective reflectivity for feedback mirror reflectivites ...... 33

Figure 19 - Normal incident maximum reflection of composite mirror ...... 34

Figure 20 - Variable Reflectivity profiles for n = 2, 4, 8 ...... 36

Figure 21 - Transmission vs radius for birefrinegent lens, radial units in mm ...... 39

iv Figure 22 - Schematic of experimental arrangement ...... 41

Figure 23 - Experimental arrangement showing laser diode array, Monoblock assembly, ...... 42 feedback mirror and interference filter to pass only 1.57 µm energy...... 42

Figure 24 - Laser diode pump current pulse (200µsec), 1.064 µm fluorescence build up, and Q ...... 44 switched output ...... 44

Figure 25 - Far field beam divergence using 40% R ...... 48

Figure 26 - Far field beam divergence using 60% R ...... 49

Figure 27 - Monoblock output energies versus feedback mirror position ...... 50

Figure 28 – Laser output energy showing pulse to pulse stability ...... 51

Figure 29 - Beam divergence using convex feedback mirrors ...... 53

Figure 30 - Output energy for convex feedback mirrors...... 53

Figure 31 - Beam divergence as a function of feedback mirror misalignment ...... 54

Figure 32 - Temporal pulselength of the Monoblock output. Peak intensities were normalized for both pulses. (normalized intensity, a.u.) ...... 56

Figure 33 - Monoblock emission spectra without feedback mirror ...... 57

Figure 34 - Monoblock emission spectra with feedback mirror ...... 57

Figure 35 - Far field image with no feedback mirror ...... 60

Figure 36 - Far Field image with feedback mirror at 5 mm ...... 61

Figure 37 - Far field image with feedback mirror at 50 mm ...... 62

Figure 38 - Divergence for 1.064 µm long pulse laser...... 64

Figure 39 - 1.064 µm far field divergence without feedback mirror...... 65

Figure 40 - 1.064 µm far field divergence with feedback mirror at 5 mm...... 66

Figure 41 - 1.064 µm far field divergence with feedback mirror at 50 mm...... 67

Figure 42 - Output energy for 1.064 µm long pulse laser...... 68 v vi

Acknowledgements

I would like to thank the management at the US Army, Night Vision and Electronic Sensors

Directorate, for without their support and assistance, my goal of attaining this degree would not have been possible. Specifically I want to thank Dr. Lew Goldberg for his support and encouragement during my studies.

vii

1. Introduction

Ever since lasers have expanded beyond the confines of a laboratory setting, applications have rapidly grown. Applications have been developed to exploit the unique properties lasers possess, from cutting and welding various materials to remote sensing, lasers have been shown to be a superior solution to conventional methods of material fabrication and measurements. Man portable laser systems have unique requirements due to the operating environments and performance objectives. Unlike installed laser systems, most man portable lasers must operate over wide temperature ranges with little or no thermal control due to power and size restrictions.

Man portable systems further mandate battery power which in turn greatly reduces the available power for laser operation and environmental conditioning.

One of the more important and widely used applications of the laser is for laser rangefinding

(LRF) [1]. Often times, the laser is the only way to accurately determine the distance to an object without physically traversing the distance. For range finding, the laser produces a short optical pulse and the distance is calculated by the time of flight to and from the target. The rangefinder system would consist of a laser transmitter and a receiver with collecting optics and a detector.

The radar equation [2], equation (1), is applicable to laser rangefinders and knowing the sensitivity of the detector, a minimum peak power can be determined for the laser pulse at a given design range. For ranges on the order of several kilometers the laser needs to generate peak powers of tens of kilowatts. This is only possible using Q-switched pulselengths of 5-60 nsec

FWHM. A few millijoules (mJ) of energy over this short of a pulselength can easily produce the

1 2 required high peak powers for detection at long ranges.

= (1) π 2 2 푃푡푟푎푛푠푇 퐴 푅 푃푚푖푛 where R is the range for detection in meters, is the transmitted power in watts, is the

푡푟푎푛푠 target reflectivity, A is the receiver aperture area푃 in meters, is the minimum detectable푇

푚푖푛 power in watts. This equation is a slight modification of the푃 typical radar equation, since at optical wavelength the beam diffraction is small, it assumes all transmitted power strikes the target.

With transmission of the laser pulse through free space, another concern is eye safety. Lasers that generate output wavelengths in the visible or near infrared can pass through the lens of the eye and focus on the retina. This brings a great deal of power to bear on a small focused spot and can exceed the damage threshold for the retina. Typical damage thresholds for this situation are in the tens of microjoules per square centimeter of optical fluence incident on the eye. Since the requirements for detection at long ranges requires significantly higher energies that would exceed eye-damage levels, another solution must be found. The dire consequences of injuring someone by accidently exposing them to this power level would preclude the laser’s use in the field. Fortunately there are alternatives. The size of the transmitting optics can be increased, thus reducing the fluence incident on the pupil of the eye. This can only be done within reason; maintaining a compact system is also of paramount importance and large apertures detract from this goal.

Another means of reducing the possibility of eye injury is to operate in the so called “eye- safe” wavelength region. In this wavelength range around 1.5 microns, which also matches a

3 high atmospheric transmission window [3], the laser light passes through the lens of the eye, but is absorbed by the vitreous humor (the fluid within the eye). Instead of focusing down to a small spot on the retina, the absorbed laser power is distributed over the volume of the fluid and dramatically reduces the amount of energy reaching the retina. The term “eye-safe” is a misnomer, since given enough power even this wavelength range can exceed the damage threshold of the human eye. Operating in this region does increase the damage threshold by a factor of approximately 1000. This added increase in damage threshold allows for free space laser transmission of laser pulses with sufficient energies for LRF applications, while maintaining reasonably sized optical apertures. Several laser source types have been developed to operate in this eye-safe wavelength band. Diode-pumped Erbium glass [4] lasers operating at

1.54 µm emit within the band directly, but have issues with thermal degradation and low gain.

Other sources include those that rely on nonlinear conversion, such as Raman Scattering [5] using a methane gas cell or optical parametric generation [6].

2. Monoblock

The solid-state laser is the most expensive component of the rangefinder system. Typical

military lasers incorporate complicated optical assemblies consisting of many elements. Each

element must be fabricated individually and aligned with a high level of precision. This in turn

requires skilled labor, a significant contributing factor in overall system costs. The Monoblock

laser was developed as a low cost eye-safe optical source, specifically for US Army LRF

applications. The concept is to use a simple all flat face crystal assembly, permanently bonded on

a common crystalline substrate referred to here as the “optical bench”. The plano optical faces facilitate batch process manufacturing. The optical components are fabricated with precise tolerances for face parallelism and optical component to optical component perpendicularity,

thus allowing the optical component edges to act as registration surfaces. Using this technique,

laser alignment is essentially automatic and greatly simplifies the bonding requirements for all components on the optical bench. The optical bench, a simple undoped YAG rail to match CTE of the components, along with the laser elements is shown in Figure 1 and Figure 2.

The Monoblock laser incorporates concepts that were originally developed for monolithic lasers. An example of this type of laser is the monolithic ring laser developed in 1985 by Kane and Byer [7]. The idea was to cut, polish and coat precision surfaces to form a resonator from a single component, in this case the laser gain media. The technique was similarly

4 5

Figure 1 - Photograph of Monoblock assembly

Figure 2 - Schematic of Monoblock detailing optical components and surface coatings

applied to optical parametric oscillators (OPO) crystals to form those resonators and laser designs with monolithic OPOs were reported [8]. As originally developed, the Monoblock [9] was a flash-lamp-pumped Neodymium doped, yttrium aluminum garnet (Nd:YAG) gain media with a saturable absorber Cr4+:YAG, and an intracavity Potassium Titanyl Phosphate (KTP)

OPO for frequency conversion to 1.57 µm. Despite the inherently low electrical efficiency, flashlamp pumping was initially used due to the lack of laser diodes with sufficiently high output

6

Figure 3 CAD model of assembly and test fixture

power. Later, end-pumping with a laser diode stack was used in order to increase the overall electrical efficiency and to allow higher pulse repetition rates due to a decrease in the Monoblock heat-load. Figure 3 shows a CAD pictorial of the fabrication and testing setup used to assemble the Monoblocks. Vacuum chuck arms mounted on pitch/yaw stages allows alignment of the components.

The gain media is a square Nd:YAG laser rod (squrod), a square cross section rod, 3 mm x 3 mm and approximately 18 mm long. The crystal is nominally doped with 1% weight Nd. The

7 back face of the crystal is coated highly reflective for 1.064 µm and highly transmissive at 810 nm for diode-end-pumping, which will be discussed later. A key requirement for the Monoblock is to fabricate the back face of the gain media perpendicular to the side of the crystal to a tolerance of less than ±10 arc-sec. The second surface of the gain media is cut at Brewster's angle, 61° for an optical index of 1.82, which forms a simple polarizer to force the 1.064 nm laser emission to be linearly polarized, which is necessary for effective wavelength conversion by the OPO. The phase matching condition discussed in section 2.1 uses the dispersion of the crystal which is polarization dependent for a biaxial crystal, such as KTP. A single Brewster face will cause the beam to refract, and the deviated laser beam would no longer follow a path consistent with simple alignment procedures. To avoid this beam deviation, a second Brewster face on a small piece of YAG is placed with a small air gap after the gain media to bring the beam back parallel to the optical rail. The air gap is typically less than 0.2 mm in order to prevent vertical shift of the beam. This “end cap” has a Brewster's angle and a plano face that is coated anti-reflective (AR) at 1.064 µm. The air spaced Brewsters faces form a polarizer with an extinction ratio between the S and P polarization of approximately 3 db for the two surfaces [10].

This extinction ratio is sufficient to force only the P polarization to reach threshold. To minimize any wedge effect that could deviate the beam, the end cap Brewster and plano surfaces are also registered to a high precision with its side face similar to the requirements for the squrod gain media. The vertical sides of the squrod have a ground finish to suppress parasitic oscillation before the Q-switch pulse. Without these frosted faces the parasitic oscillation will rob energy stored in the crystal and prevent the laser from exceeding threshold. The top and bottom faces are polished in improve pump confinement within the gain media. The highly divergent rays of

8

the laser diode array are confined in vertical direction by total internal reflection (TIR). This

pump confinement allows for a longer absorption path length and greater energy deposited

within the squrod.

The next component along the optical path is the Cr4+:YAG saturable Q-switch. For

rangefinders that use the time of flight to determine the distance to target, a short optical pulse

length is required. Q-switching produces very short pulses, typically on the order of 5-60 ns

FWHM. The saturable absorber Q-switch works by preventing the laser from reaching threshold,

and allowing storage of energy in the gain media. When enough energy is stored, the gain exceeds the loss of the passive Q-switch and the laser begins to oscillate near threshold. The

relatively low laser flux within the cavity bleaches the absorption of the saturable absorber and

causes it to become transparent. This sudden change from a high loss element to a transparent

one, Q-switches the laser and the optical pulse builds up from the stored energy in the Nd:YAG

crystal. A typical Monoblock will produce 5 mJ with a pulsewidth from 15-25 ns FWHM

depending on the output coupling and unsaturated transmission of the passive Q-switch. A lower

unsaturated transmission of the passive Q-switch will "hold-off" lasing as more energy is stored

in the gain media. The excited state lifetime of Nd:YAG at this dopant level is approximately

230 µsec [11]. Storage efficiency, a measure of the inversion fraction remaining at the end of the

pump pulse, drops significantly if pumping the gain media is longer than the upper state lifetime.

Like the other component discussed previously, the passive Q-switch plano faces must be fabricated to a high degree of parallelism and perpendicular to the sides to prevent significant beam deviation. The optical faces are also AR coated at 1.064 µm with a high damage coatings.

9

The final component of the Monoblock is the monolithic KTP OPO. The input face of the crystal is coated AR at 1.064 µm and HR at 1.57 µm. The other side forms the end of both the

1.064 µm and the 1.57 µm resonators and has a high reflection (HR) coating at 1.064 µm and

60%R at 1.57 µm, which serves as the output coupler for the system. Since the KTP crystal forms its own resonator cavity, extremely precise polishing of the end faces is required to maintain parallelism for alignment of the resonator. With the OPO incorporated into the design, the 1.064 µm radiation oscillates between the first face of the Nd:YAG crystal and the second face of the KTP. This intracavity arrangement takes advantage of the greater fluence inside the

1.064 µm laser cavity. The intracavity OPO concept was first proposed and studied in the early

1970’s [12, 13]. For a simple plano cavity such as the one discussed here, the 1.064 µm fluence within the resonator is a factor of greater than what is coupled out of the cavity, where R 1+푅 is the reflectivity of the resonator1 output−푅 coupler. With a typical output coupler of 70% the

intracavity fluence is 5.6 times greater than what is measured exiting the resonator. The

Monoblock uses highly reflective mirrors to form the 1.064 µm cavity. This greatly increases the

fluence inside the cavity and makes it significantly easier to exceed the OPO threshold for

wavelength conversion. The intracavity design also acts to limit the 1.064 µm fluence within the

cavity. When the fluence levels of the 1.064 µm reach the threshold point for OPO conversion,

the 1.064 µm energy is converted to 1.57 µm and is coupled out of the laser. This effectively

limits the 1.064 µm fluence and prevents optical damage, which would be inevitable for a totally

reflective resonator without any means to couple the radiation out of the cavity.

10

All components are rectangular to facilitate bonding to the optical rail using an adhesive.

Even with the high fabrication precision for parallelism and perpendicularity, the slight residual variations in the components are sufficient to cause significant variation in performance of

Monoblocks assembled from nominally identical parts. Output energies, buildup time and beam divergence can vary by over a factor of two from the mean values found with delivered systems.

Minimum performance requirements placed on the Monoblock manufacturers further selects for assemblies, since those not meeting minimum requirements are not shipped and are not part of the aggregate. The Monoblocks chosen for this study are considered typical and none were selected that were particularly better or worse than the nominal. The average and variation of output parameters were detailed in a report provided by Scientific Materials Corp (SMC) [14].

The representative sampling contains 130 units measured for energy, buildup time and divergence. All measurements were done at the manufacturer’s facilities. Energy and divergence are shown in Figure 4 and Figure 5 for all the units in the delivered lot. Unlike the current study,

SMC used a single bar laser diode for pumping the Monoblock. This accounts for the lower energy, since the current research uses a 9-bar diode stack array, but variance of the energy as a percentage, beam divergence, and buildup time for this work are similar to the measurements made by SMC.

11

Figure 4 - SMC Output energy distribution for lot shipped

Figure 5 - SMC Divergence distribution for lot shipped

In addition to the systems fabricated by vendors, Monoblocks were fabricated in-house using alignment techniques and a bonding method uniquely developed at NVESD. The primary goal of this effort was to improve yield and better understand the most sensitive parameters in

12 constructing these lasers. In the final analysis the laser systems fabricated at NVESD had very similar measured performance result to those fabricated by outside vendors. The most important result was the determination that minimum adhesive bond-lines are critical to maintaining system alignment. Specifying high tolerance and testing for the perpendicularity of the optical surface with respect to the bonding surfaces is critical. Output energy is the dominate parameter that is reduced when components do not maintain optimum alignment and this is most readily seen in

Figure 3. Lasers fabricated in-house showed less variation in energy, beam divergence and build- up time, but primarily due to the hand selection and testing of components done during assembly.

2.1 Optical Parametric Oscillators

Optical Parametric Oscillators (OPO's) use the nonlinear optical properties of the material to

generate different wavelengths from the "pump” wavelength [15, 16, 17]. From the pump wavelength, the signal and idler wavelengths are produced. Conservation of energy requires the sum of the frequencies of the signal and idler photons to equal the frequency of the pump photon:

= + (2)

휈푃푢푚푝 휈푆푖푔푛푎푙 휈퐼푑푙푒푟 where , , are the frequencies of the pump, signal and idler fields,

푃푢푚푝 푆푖푔푛푎푙 퐼푑푙푒푟 respectively.휈 In휈 addition푎푛푑 to 휈energy conservation, for efficient parametric conversion from the

pump to the signal and idler fields, the momentum must be conserved. This is also known as phase matching, where the condition must satisfy:

= + (3)

푘�⃗푃푢푚푝 푘�⃗푆푖푔푛푎푙 푘�⃗퐼푑푙푒푟 here , , are the wave vectors for the frequencies of the pump, signal and idler

푃푢푚푝 푆푖푔푛푎푙 퐼푑푙푒푟 fields,푘�⃗ respectively.푘�⃗ 푘The�⃗ interaction of the intense pump and weak quantum noise of the signal

and idlers is through the second-order nonlinearity of the crystal. This interaction is described by

the coupled wave equations [18]:

+ = (4) 푑퐸푠 푛푘 ∗ 푑푧 훼푠퐸푠 푖휅 휔푘 퐸푝퐸푖 + = (5) 푑퐸푖 푛푘 ∗ 푖 푖 푘 푝 푠 푑푧 훼 퐸13 푖휅 휔 퐸 퐸 14

where and are the electric fields, and are the absorption coefficients at the signal and

푠 푖 푠 푖 idler frequenc퐸 ies퐸 , and κ is the coupling constant훼 훼 defined by:

= (6) 2휔푠휔푖 2 3 휅 푛푠푛푖푛푝휀0푐 푑푒푓푓 The amplification of the signal and idler fields takes place at the expense of the pump field.

The phase matching condition is met using the angular dependence of the material’s refractive

index. Since the only requirement for the relative wavelengths of the signal and idler is for their reciprocal sum to equal the reciprocal of the pump, aligning the crystal axes for noncritical phase matching will generally obtain the most efficient conversion. Using the dispersion of the biaxial

KTP crystal, the phase matching condition is met. If phase matching is possible while the optical

fields propagate along one of the principal axes of the crystal it is called noncritical phase

matching. Noncritical phase matching eliminates walk-off between the pump, signal, and idler

wave vectors, which when present, significantly reduces the spatial overlap of the pump, signal

and idler beams. Also, with noncritical phase matching, the acceptance angle for phase matching

is typically at its greatest and maximizes parametric gain coefficient.

KTP is an orthorhombic crystal and therefore is optically biaxial. There are two orientations

for noncritical phase matching. The most common orientation is x-cut and what is used

exclusively here with the crystal orientation set to θ = 90° and = 90°. The signal and idler

wavelengths are 1.572 µm and 3.29 µm for this orientation. Figure휑 6 shows the calculated signal

15

Figure 6 - Tuning curves for x-cut KTP OPO [19], units shown with labels versus angle

and idler wavelengths as a function of the phase matching angle θ for = 90° in a Type II OPO pumped at 1.064 µm. The calculated acceptance angle, walk-off angle휑 and are also shown

푒푓푓 in the figure. The effective nonlinear coefficient is a maximum of =푑 2.1 / for this

푒푓푓 crystal cut. A plane wave analysis of OPO’s was performed by Siegma푑 n [20] and푝푚 modified푉 to include pump reflection to show the intracavity field strength at the pump wavelength

푝 satisfies: 퐸

= (7) 2 1 1−푅 퐸푝 휅ℒ �퐸푝0 − 2휅ℒ�

16 where is the field strength of the incident pump, R and are the reflectivity of the output

푝0 coupling퐸 at the resonate wavelength and gain length, respectively.ℒ For monolithic OPOs the crystal and OPO cavity length are one and the same.

This is analogous to a conventional laser operating in a long pulse or cw condition. Once the

OPO has reached threshold, the gain is clamped and any additional increase in pump input from the 1.064 µm source goes into conversion to 1.57 µm that is output coupled. The OPO is pumped by a transient Q-switched pulse of the Nd:YAG laser, but there is no energy storage within the

OPO. Once threshold is reached for the intracavity OPO, it acts as the output coupler similar to a conventional laser resonator.

2.2 Diode-Pumping of Solid-State Laser

Significant improvements in overall system electrical efficiency and reduced thermal load can be observed with laser diode pumping of solid-state lasers over flashlamp pumped systems

[21]. The use of laser diodes as a pump source has been discussed and demonstrated since the earliest days of laser development. But it was only during the late 1980s that high power laser diodes arrays became available as a reliable pump source. [22] And even then, the cost of these arrays far exceeded the cost for more traditional lamp pumped sources. Only applications that required extremely long lifetimes, such as space-based lasers or military applications which required high laser efficiency due to battery operation could justify the higher cost. As improvements have been made over time, the cost has reduced significantly and the performance has improved. The primary reason for using laser diodes to pump solid-state laser crystals is the

efficient conversion of electrical energy into optical energy in an absorption band of the gain

media. Nd:YAG has a very strong absorption band around 808 nm and this is the primary

wavelength used by laser diodes to deposit energy into the crystal as shown in Figure 7. The

emission linewidth of 2.5 nm FWHM for the laser diode matches the Nd:YAG absorption width well. Unfortunately the band gap of the AlGaAs material used in the laser diode is temperature dependent. This means the center wavelength of the laser diode emission changes as the junction temperature of the device varies[23]. The variation is typically 0.3 nm/°C, which is significant

for systems that must operate over a wide ambient temperature range.

17 18

Figure 7 - Absorption spectrum of Nd:YAG [24]

If the center wavelength of the laser diode shifts beyond the width of the Nd:YAG absorption peak, most of the energy from the pump will not be deposited within the crystal and the laser will either fail to reach threshold or have dramatically diminished output. If the temperature of the laser diode is stabilized and remains fixed this can be prevented, but handheld military systems typically use batteries with limited energy. Using that power source to keep the laser diodes “on wavelength” by controlling the temperature is very inefficient. Without temperature stabilization, the laser will operate over a limited range of temperature, typically ∆T~20°C; outside of this range the laser performance degrades significantly, with output energies <50% of peak values. A novel technique [25] has been developed to mitigate some of this variation in energy deposited in the gain media at the expense of some efficiency. The laser diode array is made up of several

19 bars, each containing numerous emitters along its length. Each bar has a center wavelength and linewidth. Instead of choosing all the bars mounted in a common array to have the same center wavelength, three different center wavelengths are chosen. By doing this the overall linewidth of the array emission is broadened (see Figure 10). Now the convolution of the Nd:YAG spectrum and the laser diode array are considerably smoother and variations of energy deposition within the gain media are not as severe. Other absorption peaks are utilized in addition to the primary wavelength at 808 nm with the wider array linewidth as the environmental temperature is varied.

A model of the absorption for various laser diode spectral distributions was developed [26] and was used to verify the optimum wavelength distribution of the laser diode array for wide temperature range operation. A comparison is shown in Figure 8 of the absorption variation as a function of operating temperature for an array composed of only a single center wavelength peak and an array composed of three center wavelength peaks. The “three-color” array used in the model is based on the lineshape of the actual laser diode array used in the experimental results.

The broader linewidth allows the system to absorb pump energy and maintain threshold even in regions where the single wavelength emission array could not deposit sufficient energy to maintain threshold.

The laser diode array, model number S6 used for this study was produced by Lasertel, Inc. and contains 9 bars for an emitter area of approximately 3 mm x 3 mm. Each bar is rated at 80W peak power. Figure 9 shows typical performance parameters for this model and an image of the laser diode array. Wavelength spectrum and output versus input current are shown in Figure 10.

The laser diode is typically operated at 80 Amp at 2 pulses per second (PPS) for durations

20

Figure 8 - Absorption convolution with single and three color pumps

ranging from 170-280 µsec. The pump pulse length is adjusted until the threshold for bleaching the passive q-switch of the Monoblock is exceeded. The laser diode when operated at 80 A produces over 700W of peak power. This typically exceeds the Monoblock threshold at approximately 200 µsec of pump duration.

21

Figure 9 - Laser diode array specifications, [27]

160

µ 140 Operated at 200 sec, 2 PPS

120

100

80

Output,mJ 60

40

20

0 10 20 30 40 50 60 70 80 90 Current, A

Figure 10 - Laser diode array spectrum and I-P curve

3. Etalon Theory

All lasers incorporate a Fabry-Perot etalon to form the resonator boundary conditions. Long

before their use as laser resonators, this optical configuration was used for high resolution

spectrometers. A detailed study of the theory and use of Fabry-Perot etalons is discussed in

Hernandez [28]. Any two surfaces either plano or curved that are aligned to be parallel at a fixed

distance apart will form an etalon. Consider a plano parallel optical plate with partially reflecting

surfaces as shown in Figure 11. A monochromatic plane wave is incident on the first surface of

the optic. Part of the wave is reflected, and assuming no scatter, the remaining fraction is

transmitted into the optic. The remaining fraction passes through the material and the strikes the

second surface. Again, the wave is partially reflected back into the optic. This division of the

initial plane wave continues as shown in the figure. The phase lag between any adjacent rays is

given by:

= cos (8) 4휋 훿 휆 푛푙 휃 where n is the index of refraction for the media between the reflective surfaces with thickness .

After a little algebra the dependence of the etalon transmission on the phase lag is given by the푙 well known formula: 훿

( ) = 2 (9) 1−푅 2 푇 1+푅 −2푅 cos 훿 where R is the reflectivity of both surfaces. 22 23

Figure 11 - Schematic of etalon

This derivation assumes both reflecting surfaces have equal transmission, but that is only to

simplify the expression and illustrate the salient points of the Fabry-Perot etalon. As can be seen

in Figure 12 the peak transmission of the etalon is periodic with phase lag and has a period of

2 . The free spectral range (FSR), which is the spacing between adjacent peaks as a function of

wavelength휋 is given by:

= 2 (10) 휆0 Δ휆 2푛푙 cos 휃+휆0 where is the specific wavelength at one peak. The full width half maximum (FWHM) of the

0 peaks 휆 is related by the finesse of the etalon.

훿휆 = = (11) Δ휆 휋 −1 1 훿휆 2 sin ( ) ℱ 4푅 � 1−푅 2

24

As the reflectivity of the surfaces approach unity the finesse becomes large and the linewidth of the etalon becomes very narrow. One of the common uses of a Fabry-Perot resonator is as a scanning interferometer, with its resolution determined by the resonator finesse. When used as an interferometer the distance between the reflective surfaces is varied thus scanning the peak of the transmission. At optical wavelengths, with spacing of a few millimeters and reflective coating of

90%, typical FSR, and values are 35 nm, 1 nm FWHM, and 30 respectively.

훿휆 ℱ

Figure 12 - Transmission through etalon versus phase R=0.7

25

Figure 13 - Transmission of divergent light through etalon

If a convergent or divergent beam passes through an etalon, the transmitted intensity forms a pattern of concentric rings as shown in Figure 13. There is a radial dependence to the reflection or transmission of the beam. The concentric fringes are non-periodic in angle; this is due to the cosine dependence of the phase delay on the ray angle, as indicated by equation (8). The angular variation of the etalon transmission with ray angle, displayed in Figure 12, is fundamental in explaining how the feedback mirror used in this study reduces the OPO beam divergence.

It is a relatively easy modification of the equally reflective surfaces for an etalon to the case of unequal reflectivities. The effective transmission is simply given by:

( )( ) = (12) 1−푅1 1−푅2 푒푓푓 1 2 1 2 푇 1+푅 푅 −2�푅 푅 cos 훿 26

And since = 1 assuming negligible absorption or scattering,

푅푒푓푓 − 푇푒푓푓 = (13) 푅1+푅2−2�푅1푅2 cos 훿 푅푒푓푓 1+푅1푅2−2�푅1푅2 cos 훿 The effective reflectivity is the equation that describes the feedback into the OPO cavity for the etalon formed by the feedback mirror and the original output coupler coated on the monolithic

OPO crystal.

3.1 Composite Output Coupler

The previous discussion of Fabry-Perot etalons detailed the theory and some of its useful

characteristics. The present research employs an etalon as the output coupler for the OPO

resonator as shown below. Now the effective transmission and reflectivity from the OPO cavity

have the properties of an etalon and not simply the coating of the simple output coupler. The new

etalon formed by the original output coupler on the OPO crystal and the additional feedback

mirror act as a composite output coupler. Unique properties are possible with this configuration.

Figure 14 shows the coupled OPO resonator and the composite mirror with the signal and pump wavelengths.

Most of the treatises that discuss using etalons as output couplers only consider the wavelength dependence of the reflectivity modulation [29]. But recalling equation (8), the angle with which the ray strikes the mirrors of the etalon also affects the phase lag. If the effective reflectivity is displayed not as a function of wavelength or the simple lumped phase term, but with respect to the incidence angle, the features of the phenomenon studied here are shown in

Figure 14. The reflectivity is dependent on the angle of incidence of the beam, measured relative to the surface normal of the etalon output coupler. For low divergence rays the etalon reflectivity and feedback into the OPO cavity are nearly constant. As the incidence angle increases, the effective reflectivity begins to drop. This property is similar to the variable reflectivity

27 28

Figure 14 - Schematic of OPO and Feedback Mirror

mirrors (VRM) or birefringent lens used for polarization output coupling [23].

The expression for the effective reflectivity is recast as a function of the incident angle for the coatings on the OPO crystal and the feedback mirror and is given below:

= cos 휃 (14) 푅푂푃푂+푅퐹퐵−2�푅푂푃푂푅퐹퐵 cos 4휋푙 휆 푒푓푓 cos 휃 푅 1+푅푂푃푂푅퐹퐵−2�푅푂푃푂푅퐹퐵 cos 4휋푙 휆 where is the reflectivity of the output coupler coating on the monolithic OPO crystal,

푂푃푂 퐹퐵 is the reflectivity푅 of the feedback mirror, is the separation between the OPO output coupler 푅and the feedback mirror. 푙

The maximum reflectivity, which also forms the highest Q in the cavity, is for normal incidence

29

Figure 15 - Effective reflectivity of composite mirror versus angle, in radians. OPO mirror reflectivities of 60% and feedback mirror of 40%, length 10 mm

on the mirrors, when cos 4 = 1. This requires that 4 = 2 + 1 for = 0 is given by: cos 휃 푙 휋푙 휆 − 휆 푛 휃 = (15) 푅푂푃푂+푅퐹퐵+2�푅푂푃푂푅퐹퐵 푀퐴푋 푅푒푓푓 1+푅푂푃푂푅퐹퐵+2�푅푂푃푂푅퐹퐵 where n is a large integer. Consider an etalon length of L=10 mm, the integer value would be approximately 12738 for 1.57 µm. For a given integer, the OPO signal and idler wavelengths shift a small amount to maintain the highest effective composite mirror reflectivity for normal incident beams and thus the highest Q for the OPO resonator. As the composite mirror length is increased the large integer n for normal incident beams also increases. For any change in length the wavelength of the signal and idler need only change by 0.0001 µm before the next integer hop as the composite mirror length is increased or decreased. The wavelength shift is found for

30

the assumed composite mirror length of 10 mm and the signal wavelength of 1.572 µm and

setting the integer relationship between n and n+1 to determine the required maximum

wavelength shift, since:

4 = 2( + 1) + 1 (16) 푙 휆+Δ휆 푛 With the wavelength of the signal able to maintain the maximum effective reflectivity for

normal incident beams, increasing the separation of the feedback mirror has the result of

reducing the angular value of the first minimum shown in Figure 16. The amount of feedback

into the OPO resonator becomes more angularly dependent as the separation of the composite

mirror increases. As the higher divergent rays see less reflectivity, only those modes that exceed threshold produce an output and the divergence of the OPO output is reduced as the composite mirror separation increases. A measure of how the reflectivity of higher angle rays varies with mirror separation, is to consider the change in the angle of the first minimum with feedback mirror separation.

To find the nodes (minimum reflectivity, see Figure 14) of the composite mirror as a function

of the angle of incident, the derivative of the effective reflectivity expression is taken and set to

zero, exquation (17). Solving for the first node where 0 gives a dependence of the node as

the length of the composite mirror is varied and is shown휃 ≠in equation (18) and Figure 16.

= 0 = sin( [2 + 1] cos ) (17) 푑푅푒푓푓 푑휗 휋 푛 휃0 = cos (18) 푙 −1 4 �휆−1 0 푙 휃 4 �휆

31

As could be expected, the angular position of the first transmission minimum decreases monotonically with the mirror separation, with the decrease being most rapid for separation below approximately 10 mm. It is also insightful to graph the angular dependence of the effective reflectivity on the incidence angle for angles that are only few milliradians from normal incidence. This is done in Figure 17 for several distances of the feedback mirror (R = 40%) away from the OPO output coupler face. It can be seen from the figure that a significant drop in the effective reflectivity for angles of a few milliradians from the normal can be achieved even with

10 mm separation of the feedback mirror. As the distance increases, the effective reflectivity vs. angle curve becomes narrower, leading to better rejection of higher angle rays. This explains

Figure 16 - Minimum reflection node versus composite mirror length

32

Figure 17 - Angular dependence of the effective reflectivity for three composite mirror lengths.

why the beam divergence decreases as the feedback mirror is moved further away from the OPO

output face.

The shape of the central lobe is also dependent on the reflectivities of the feedback mirror and output coupler on the OPO crystal. Typically the OPO output coupler mirror is 60% R.

Changing the reflectivity of the feedback mirror effects the normal incident reflection fraction and also the depth of the reflection minimum. Figure 17 shows the modeled performance of the composite mirrors for feedback mirror reflectivities of 0.20, 0.40, and 0.60. Although the angular reflection minimum and maximum locations are not affected by changing the reflection coefficients of the mirrors, the reflectivity coefficients do have a significant impact on the angular dependence of the effective reflectivity. The experimental testing employed both 40%

33

Figure 18 - Angular dependence of the effective reflectivity for feedback mirror reflectivites of 20%, 40%, and 60%, for composite mirror separation of 10 mm and 60% feedback mirrors. Significantly higher reflective coatings on the feedback mirror would produce too high a normal incident reflectivity approaching 1.0 as shown in Figure 19. This can lead to optical damage of the components. The normal incident reflectivities of the composite mirror for the 40%R and 60%R feedback mirrors are 89% and 94% respectively, for an OPO output coupler reflectivity of 60%.

34

Figure 19 - Normal incident maximum reflection of composite mirror

4. Techniques for Transverse Mode Discrimination

As previously mentioned in the introduction, the system aperture is limited by size and weight concerns, but detection requirements for long range applications dictate a reasonably low

divergence beam in order to concentrate the laser beam on target. Several techniques have been

developed for improving the beam quality of a laser with a given Fresnel number. Two of the

more common methods discussed below are based on reducing the cavity feedback as a function

of radial distance from the center of the resonator. All higher order or aberrated modes require a

larger spot size and divergence product than the fundamental Gaussian beam, but most compact

lasers operate with higher order modes because of significantly higher output power. By

reducing the feedback for these higher order modes below threshold, output energy is usually

reduced slightly, but the remaining modes do not experience the diffraction loss at the aperture

and significantly improved beam quality is possible.

4.1 Variable Reflectivity Mirrors

The use of variable reflectivity mirrors (VRMs) also known as graded reflectivity mirrors

(GRMs) as output couplers has been possible only since the techniques needed to fabricate high

quality, radially dependent reflectivity coatings in the early 1990s became available. These

output couplers have significantly improved the performance of solid-state lasers by reducing

35 36

the feedback of the laser near the limiting aperture where diffraction is present. Even small amounts of beam clipping can have a large impact on the beam's phase front. Initially these output couplers were used with an unstable resonator [30, 31] to replace the "dot" output coupler that had traditionally been used. Now these VRMs have found wide spread applications as resonator output couplers, significantly improving output beam quality. The reflectivity profile is expressed by:

( ) = 푟 푛 (19) −2�휔� 푅 푟 푅0푒 where is the radial dimension, is the peak reflectivity, ω is the spot size on the mirror and n

0 is the super푟 -Gaussian order. Figure푅 20 shows the profile of several VRM’s with varying orders of

n= 2, 4, 8 respectively with a peak reflection of 70% and spot size of unity.

When used for an unstable resonator the VRM is typically applied to a convex surface.

This surface contributes to the round trip magnification. As the beam grows transversely due to

Figure 20 - Variable Reflectivity profiles for n = 2, 4, 8

37 the cavity magnification, more of it is coupled out of the cavity by the radially dependent reflectivity.

4.2 Polarization Dependent Output Couplers

Laser resonators that employ crossed porro prisms [32] as end mirrors to improve tolerance to misalignment must employ another means to couple the laser energy out of the cavity. The total internal reflection of the porro prisms prevent any output coupling from the cavity.

Typically these lasers are polarized so a waveplate and polarizer are used to couple a faction of the circulating power out of the cavity. With the beam making a round trip through a quarter waveplate, 2 phase retardation and rotating the waveplate through 90° output coupling 휋 between 0% �and 100% is possible through the polarizer. This configuration is analogous to a more traditional partial reflector used as an output coupler. To create a radially dependent polarization coupled arrangement, the waveplate is replaced with a birefringent lens [33]. Since the lens has a curved surface and the amount of phase retardation is related to the thickness of material traversed, the retardation becomes dependent on the radial distance away from the center of the lens. By using a birefringent material like crystalline quartz, a lens can be suitably fabricated. Using the methods discussed in Optical Waves in Crystals [34] a Jones matrix calculation was performed to determine the optical properties of a birefringent lens. The radially dependent phase retardation is given by:

= (20) 2휋 2 2 Γ 휆 Δ푛�푡 − 푅 − √푅 − 푟 � where is the wavelength, is the on-axis lens thickness, is the radius of curvature, and is

the radial휆 dimension. The푡 model is for two passes through푅 the birefringent lens. The푟

38 39

calculation assumes linearly polarized light coming from a polarizer passing through the

birefringent lens then reflecting off the cavity mirror before passing through the lens a second

time, then output coupled through the polarizer. This model is summarized in the matrix

expression given below. The angle is referenced to the plane of the polarizer and the fast axis

of the lens. 휃

1 0 cos sin 0 cos sin 1 0 cos sin 0 cos sin 1 = Γ Γ (21) 0 0 sin cos −0푖2 sin cos 0 1 sin cos −0푖2 sin cos 0 휃 휃 푒 Γ 휃 − 휃 휃 − 휃 푒 Γ 휃 휃 � � � � � 푖2� � � � � � � � 푖2� � � � � − 휗 휗 푒 휗 휗 − 휗 휗 푒 − 휗 휗 Figure 21 is the reflection as a function of distance from the center of the lens. The radius of

curvature is 25 mm, lens on-axis thickness of 1.1 mm and the fast axis angle is 45°.The

birefringent lens has several drawbacks which limit is usefulness as an output coupler휃 scheme.

First, the ability to accurately fabricate the lens to have the correct on-axis reflectivity is difficult.

Even small changes in lens thickness can have a dramatic affect on the reflectivity at the on axis peak. Second, the width of the minimum reflectivity is primarily dependent on the lens

Figure 21 - Transmission vs radius for birefrinegent lens, radial units in mm

40

radius of curvature. Unfortunately, to make the on axis reflection peak narrow enough to be

useful the reflectivity from the higher order ‘ringing” becomes a significant feedback at the

aperture. This leads to unwanted diffraction.

An alternative was consider to use a thick plano waveplate instead of the curved birefringent

lens. The divergence of the higher order modes would experience a longer path length going

through the waveplate. Thus the feedback for the cavity would be dependent not on the radial

distance, but to the divergence of the beam, similar to the feedback etalon technique discussed

here. Unfortunately, a simple calculation of the thickness needed for quarter wave retardation

and a 10 mR half angle beam divergence requires a quartz waveplate to be approximately 800 mm thick.

5. Results and Observations

5.1 Experimental Arrangement

Several Monoblock lasers were evaluated during the experimental testing. Each laser system was tested for energy, build-up time, far-field beam divergence, temporal pulse length, and wavelength. A schematic drawing of the experimental arrangement is shown in Figure 22. The output of the laser passes through an interference filter centered at the emission wavelength of

1.57 µm and a beam splitter. A small fraction of the beam from the beam splitter is reflected off a 1m concave mirror. The InGaAs CCD array is placed 50 mm from the mirror to form the far- field beam divergence image. The remaining beam is directed into an energy meter and recorded.

A silicon photodiode measures the 1.064 µm fluorescence build-up and a spectrometer records the spectrum from scatter light off the energy meter. A high speed InGaAs photodiode is used to measure the OPO output pulse.

Figure 22 - Schematic of experimental arrangement 41

42

Figure 23 - Experimental arrangement showing laser diode array, Monoblock assembly, feedback mirror and interference filter to pass only 1.57 µm energy.

The feedback mirror is mounted to a micrometer stage with its distance from the OPO output coupler face varied from 3 mm to 50 mm. Data was taken at regular intervals as the feedback

43 mirror was moved away from the Monoblock. It is aligned normal to the laser beam to form the composite mirror. Typically, as the feedback mirror is moved away from the OPO crystal either little or no realignment is needed to maintain the optimum performance. Alignment sensitively of the feedback mirror is similar to resonator end mirrors and this is borne out in the experimental testing.

To monitor the laser diode drive current, which is nominally set at 80 A, and the onset of lasing (buildup time) a slow photodiode and current monitor shown is used and shown in Figure

24. Without monitoring the buildup time, environmental conditions such as ambient temperature or thermal load within the gain media can vary enough to increase the buildup time beyond the pump pulse length, preventing lasing action. Also the buildup time can shorten to the point that two Q-switched pulses are generated when conditions are optimum. The choice of a passive Q- switch unsaturated transmission, typically 60%, provides for a nominal buildup time of approximately 200 µsec. Spatial alignment of the laser diode array pump and center wavelength at the operating temperature can significantly affect the buildup time as well as output energy. In addition to testing and evaluating Monoblocks fabricated by vendors, a procedure was developed for constructing these lasers from individual components. Alignment techniques and fixtures were developed to accurately align the components to the optical axis. As each component was bonded to the undoped YAG bench, performance measurements were taken of output energy.

Norland 61, a UV curing adhesive, was used exclusively to bond all components to the base.

Minimizing bond line variation is a key requirement to reduce any wedge which would cause an

44

Figure 24 - Laser diode pump current pulse (200µsec), 1.064 µm fluorescence build up, and Q switched output

angular change over temperature.

An important step in the assembly process was to thermally cycle the adhesive bonds as each component was added to the bench. The assembly was cooled to -10 °C, and then heated to 100

°C. This process was repeated three times to relive any stresses in the bond line. Failure to perform this step during the assembly process would result in significant variation in overall system performance of the final configuration as the bonds of the various components settled into their final orientation. These changes in alignment were small enough to be difficult or

45

impossible to measure using a simple back reflection from the alignment laser and are on the

order of 10-20 µrad. Changes greater than this could be measured and showed dramatic changes in system performance such as doubling the buildup time to reach threshold. The smaller changes could easily reduce output energy by 20% from its optimum. The thermal cycling reduced the amount of performance variation, at the cost of added fabrication time and complexity, due to multiple runs in the temperature chamber.

The first component bonded to the bench is the Nd:YAG gain squrod and the undoped YAG

Brewster end cap. Alignment is done actively with a 40% R 1.064 µm output coupler held in a kinematic mount. Once aligned and a minimum long pulse (no Q-switch in cavity) output energy of 30 mJ was attained, the components were bonded to the base. If the required energy was not reached with the nominal 200 µsec, 80A (~150 mJ input energy) drive current of the laser diode array, the components were considered substandard and further assembly was halted. Once the initial subassembly had passed its long pulse tests, the Cr4+:YAG saturable absorber was placed

within the cavity. If the nominal output energy of 16-18 mJ was achieved, verifying Q-switched

operation, the saturable absorber was removed from the cavity. The OPO crystal was then

aligned to the assembly and adjusted for pitch and yaw to minimize the time between the start of

the laser diode array pump pulse and the onset of laser action in the long pulse condition. Since

the coating on the output face of the OPO crystal is HR at 1.064 µm, the cavity is totally reflecting at the fundamental wavelength. The onset of long pulse lasing was typically less than

15 µsec after the start of the pump pulse. At this stage the most critical bond was made attaching

the OPO crystal to the bench. As the UV adhesive is cured, the onset of lasing is monitored, and

46

if this time does not increase beyond 20 µsec, the assembly is considered acceptable. The

assembly is again thermally cycled and the onset of lasing time measured. If the time has not

increased beyond 25 µsec the final step of bonding the passive Q-switch is done. Small

adjustments can be made while aligning the passive Q-switch since optically it is a simple parallel plate. The laser is measured for buildup time, output energy and beam divergence and thermally cycled one more time. Using this procedure the nominal output energy is typically 7-8 mJ, which is 30% higher than the typical output from SMC fabricated Monoblocks, when pumped with the same laser diode array. This improvement is attributed to better component selection during the assembly process and thermal cycling of the adhesive bonds. Other parameters such as beam divergence and Q-switched temporal pulselength show no difference between Monoblocks built in-house and from SMC.

5.2 Experimental Results

The primary observation made was the substantial reduction in beam divergence with the feedback mirror in place. Four different Monoblocks were evaluated, two fabricated by Insight, serial numbers R2-9, R2-11, one fabricated by, Scientific Materials Corp. serial number 5023-01, and one fabricated at NVESD serial number NVL003-A. All four produced nominally the same output energy, temporal pulse shape and beam divergence without the feedback mirror present.

Figure 25 shows the change in far field beam divergence as a function of feedback mirror position for all Monoblock assemblies studied. The same 40% reflective feedback mirror was used for the measurements in this figure. The initial high divergence values are when no feedback mirror was installed. The reduction in beam divergence, as a function of the feedback mirror spacing, follows the theoretical dependence of minimum reflection node angle on mirror separation, shown in Figure 15, reasonably well.

The model and theory agree best when the feedback mirror spacing is large. This corresponds to when the composite mirror has the smallest angular feedback. A similar set of measurements were repeated using the same Monoblock assemblies, but replacing the 40% reflector with a 60%

R feedback mirror. Those results are shown below in Figure 26. The change in beam divergence has a similar dependence to the previous case, but shows a higher divergence when the feedback mirror is close to OPO output coupler. For a given gain in the

47

48

OPO crystal and considering the curves for reflectivity in Figure 17, the 60% feedback mirror

will support higher divergent beams than the 40% feedback mirror. Only those modes that

exceed the laser threshold will oscillate, so a higher reflectivity allows oscillation of higher angle

modes, leading to higher OPO output divergence; this is borne out by comparing results in

Figure 26 with those of Figure 24. For a given composite mirror length, the output divergence for the 60% R is greater than the 40% R, but only by a small amount. In addition to the experimental data shown in Figures 24 and 25, the solid curve for the analytical model from equation (18) is overlaid on the figures.

Figure 25 - Far field beam divergence using 40% R

49

Figure 26 - Far field beam divergence using 60% R

Output energy remains almost constant as the feedback mirror is pulled away from the OPO crystal, see Figure 27. The R2-11 system exhibited the lowest output energy, but also the lowest buildup time for the passively Q-switched 1.064 µm laser. The shorter build up time prevents

higher energy storage and thus less 1.064 µm energy for conversion through the OPO process.

NVL003-A produced the highest energy of all the systems for both 40% and 60% feedback

mirrors. This was probably due to the hand selection of components made at NVESD and greater

control of the assembly process.

50

Figure 27 - Monoblock output energies versus feedback mirror position

For the 5023 system, the 40% R gave higher output energies, but for the R2-9 system just the

opposite was found. The 60% R for the R2-9 system gave the highest output energy. In both

cases the differences are small and variations in conversion efficiency due to transverse mode

content can easily explain this result. This result, of no significant reduction in output energy, coupled with the large reduction in beam divergence means the output brightness of the system has improved by the same factor of approximately 2.5 X as the reduction in beam divergence.

51

Figure 28 – Laser output energy showing pulse to pulse stability

The output energy of all the lasers is very stable and varies less than 5% during operation. In

all the testing of the lasers, the output energy remained steady. Figure 28 shows the output energy for one particular arrangement using a 40% feedback mirror set 30 mm away from the

OPO output face for serial number R2-9 Monoblock. As with all the laser measurements, the laser repetition rate is set to 2 pulses per second.

In addition to the plano feedback mirrors two different radii convex reflectors, both with a reflectivity of 60%, were evaluated. The results are shown in Figure 29. These 1 m and 3 m convex feedback mirrors gave the best reduction in beam divergence, but also had a marked reduction in output energy as the feedback mirror was moved away from the OPO crystal. The

52 energy fall off, Figure 30, is due to the narrow high reflective angular acceptance of the convex mirrors, which now has a radial component due to the convex surface. As the composite mirror length increases, the angular width of this region becomes smaller until the mode area of the

OPO is greater than this width and cannot extract any energy. Once this occurs, the output energy is reduced as a function of feedback mirror separation.

The alignment sensitivity for the convex feedback mirrors is also far more stringent when compared to the plano cases. The plano/convex etalon has greater loss even for perfectly collimated rays, since the convex surface diverges rays that are not on axis when compared to the plano/plano etalons. Unstable laser cavities, which have a convex surface to form the Fabry-

Perot resonator, are very insensitive to misalignment, but have gain to overcome the losses. The sensitivity of the feedback mirror alignment was measured as a function of output beam divergence. A HeNe was bounced off the feedback mirror and aligned to a CCD camera at a predetermined distance to measure the angular deflection. The position of the first moment of the reflected HeNe beam centroid was measured on the CCD camera. As the feedback mirror was misaligned from its optimum orientation for both pitch and yaw, the position of the HeNe probe was recorded along with the divergence of the laser.

53

Figure 29 - Beam divergence using convex feedback mirrors

Figure 30 - Output energy for convex feedback mirrors

54

Figure 31 - Beam divergence as a function of feedback mirror misalignment

The angle calculated from the displacement on the silicon CCD camera and the lever arm distance is equal to twice the angle the feedback mirror actually rotates. Both the vertical and horizontal directions were varied and recorded for the resultant increase in beam divergence. The alignment position for minimum beam divergence was defined as the zero point. Figure 31

55

shows the dependence of beam divergence for feedback mirror misalignment. A 40% R plano feedback mirror was used to take these measurements at a composite mirror length of 10 mm.

The beam divergence increases from its minimum point of approximately 4 mR to 6 mR for +/-

0.5 mR of mirror misalignment. This misalignment sensitivity is not as severe as typical laser

resonator mirror misalignment. For lasers a misalignment of 0.1-0.2 mR will cause appreciable

degradation in performance. When the feedback mirror is misaligned the output energy through

the composite mirror is also reduced dramatically. The output energy transmitted through the

misaligned mirror is essentially the fraction transmitted through the feedback mirror, since it is

no longer acting like an etalon, but simply as a partial reflector.

The temporal pulselength of the Monoblock, with and without a feedback mirror, is shown in

Figure 32 measuring 42 ns and 54 ns FWHM respectively. The small increase in pulselength

with the feedback mirror aligned is due to the increased cavity lifetime of the composite mirror

when compared to the Monoblock without a feedback mirror. From the theoretical analysis, the

composite mirror has a peak reflectivity of 89% R for a 40% R feedback mirror and 60% R OPO

output coupler combination, whereas the Monoblock without the feedback mirror simply has the

60% R coating.

56

Figure 32 - Temporal pulselength of the Monoblock output. Peak intensities were normalized for both pulses.

In addition to the other measurements, it was decided to measure the wavelength of the OPO output and see if there were any changes due to the feedback mirror. An InGaAs array spectrometer was used to capture the emission spectra in a single shot mode. The emission for the configuration without the feedback mirror is shown in Figure 33, with a center wavelength

1572.3 nm and FWHM of 1.3 nm. This is the typical lineshape for a KTP OPO pumped with

Nd:YAG. When the feedback mirror was installed a second peak was seen in the emission

57

Figure 33 - Monoblock emission spectra without feedback mirror

Figure 34 - Monoblock emission spectra with feedback mirror

58

spectra, Figure 34, centered at 1567.1 nm. Initially, it was thought to possibly be a critically

phase-matched mode operating in parallel with the non-critically phase matched mode, since the feedback mirror has angular discrimination. That theory was dismissed since operating at a critically phase matched condition would require the additional emission to operate at wavelengths longer than the non-critically phase matched, see Figure 6. Any phase matching

condition less than 90° would require the signal to emit at a longer wavelength than the 90° non-

critically phase matched condition. So, this could not be the source of the second emission line.

The wavelength of the fundamental 1.064 µm laser was then measured and found to have a

second peak as well when the feedback mirror was aligned. This second peak corresponds to the

1.0615 µm emission line of Nd:YAG. This emission line is almost as strong [35] as the primary

1.064 µm emission. For the non-critically phase matched condition in KTP, the resultant

emission from the second Nd:YAG line matches the wavelength of the second peak found in the

OPO emission. The feedback mirror, although not designed to affect the fundamental resonator,

has enough reflection to form its own composite mirror with the HR mirror at 1.06 µm coated on

the OPO crystal. As the feedback mirror separation was increased, the strength of the second

peak diminished, indicating less unwanted feedback into the fundamental laser resonator. The second peak was never observed when either of the convex feedback mirrors were employed. A

test of the reflectivity of the 40% plano feedback mirror at 1.06 µm showed a value of 5%. In the

future to eliminate the formation of the second wavelength the coating of the feedback filter will

59

require incorporating an antireflective coating at 1.06 µm in addition to the existing partial

reflector at 1.57 µm.

Images of the far field are shown in Figure 35, Figure 36 and Figure 37 for three

arrangements: no feedback mirror, feedback mirror at 5 mm separation, and feedback mirror at

50 mm separation, respectively. A 1 m concave mirror was used to form a focus on the camera at

50 cm. The image formed is the far field of the output beam and its size represents the beam

divergence. Since the focusing mirror has an optical power of 2 diopters, the actual beam

divergence in milliradian (mR) is twice the value given on the camera in millimeters. Figure 35 is the raw divergence of the Monoblock without a feedback mirror in place. The beam divergence is 11.8 mR. Figure 36 and Figure 37 show the far field images for the feedback mirror set at 5 mm and 50 mm away from the OPO crystal. The divergence of the beams at the two distances is 4.8 and 3.8 mR respectively. The divergence plot in Figure 25 shows the reduction, but the image from which the data was obtained shows the dramatic reduction in beam divergence, while maintaining a fairly circularly symmetric beam profile. The vertical lines in the images are due to weak interference fringes of the non-AR coated neutral density filters. The filters are required to attenuate the beam so the detector array is not saturated.

60

Figure 35 - Far field image with no feedback mirror

61

Figure 36 - Far Field image with feedback mirror at 5 mm

62

Figure 37 - Far field image with feedback mirror at 50 mm

6. Feedback mirror applied to Nd:YAG laser

After seeing the success of using the feedback filter with the Monoblock, it was a natural thought to see how this technique would apply to a simple Nd:YAG laser. A Monoblock was

fabricated without the KTP OPO. A 60% R was used as the output coupler. All other

components were the same as the usual Monoblock. Testing began with long pulse operation without the Cr4+:YAG saturable Q-switch installed. The laser diode pump was operated at the

usual 80 A drive current with a pulselength of 200 µsec. The beam divergence change is shown

in Figure 38. The resulting reduction in divergence is present, but not to the extent found when

the OPO was employed. The far field profile was also significantly different from the 1.57 µm

arrangement. With and without the feedback mirror, the far field had a rectangular shape with the

horizontal and vertical divergence converging at different rates to the reasonably equal

divergence at 50 mm composite mirror length.

Images of the divergence for three of the composite mirror lengths are shown below. Unlike

the OPO resonator, the 1.064 µm resonator has a thermal lens within the Nd:YAG gain media

due to the absorption of the 810 nm pump light and the subsequent Stokes shift to 1.064 µm. The

Monoblock is operated at 2 PPS which means approximately 250 mW of thermal load is

deposited into the 3 x 3 mm Nd:YAG squrod. The small thermal load is present in both the OPO

and the 1.064 µm only configurations. This thermal load defines the transverse modes for a

flat/flat cavity. Unlike the OPO resonator, the 1.064 µm cavity, with even the slightest

63

64

Figure 38 - Divergence for 1.064 µm long pulse laser.

positive thermal lens, will support significantly more modes. The OPO resonator, which is also flat/flat but without any significant energy deposited within the KTP crystal, has a much larger fundamental mode. Figure 39, Figure 40, and Figure 41 show the far field divergence profiles for three configurations: without feedback mirror, feedback mirror at 5 mm distance, and feedback mirror at 50 mm distance from output coupler. The horizontal divergence corresponds to the direction the pump diode array fast axis is oriented. The fast axis of the pump array diverges very quickly and fills the entire width of the gain media in a very short distance. The larger pump

65 volume in the horizontal direction leads to higher order modes, reaching threshold with higher divergence. The output energy is shown in Figure 42 and remains relatively constant as the feedback mirror is varied. The initial increase in energy when the feedback mirror was aligned can be attributed to the effective reflectivity of the composite mirror being higher than the resonator output coupler alone. The 60% R output coupler is too low for the quasi-cw long pulse configuration and more appropriate for Q-switched operation. When the feedback mirror was aligned to the over coupled cavity, the higher effective reflectivity more closely matched the optimum reflectivity for maximum output energy.

Figure 39 - 1.064 µm far field divergence without feedback mirror.

66

Figure 40 - 1.064 µm far field divergence with feedback mirror at 5 mm.

67

Figure 41 - 1.064 µm far field divergence with feedback mirror at 50 mm.

68

Figure 42 - Output energy for 1.064 µm long pulse laser.

The passive Cr4+:YAG Q-switch was installed and the divergence of the resonator with the

feedback mirror measured. Without the feedback mirror in place the beam divergence and

pulselength are 3.6 mR and 28 ns FWHM, respectively. The passive Q- switch has already reduced the output beam divergence since it is a brightness dependent filter. Only the lowest order modes with the highest brightness bleach the saturable absorber and exceed threshold. No further reduction in beam divergence is measured with the feedback mirror installed during Q-

69

switched operation. The composite mirror has limited angular discrimination for very low

divergence rays so further reduction beyond what the saturable absorber provides would be

limited. Also, care must be exercised, since the composite mirror increases the effective reflectivity thus increasing the intra-cavity beam intensity. During initial testing the Brewster face of the first 1.064 µm laser fabricated was optically damaged when the feedback mirror reflectivity was 40% R. A 20% R feedback mirror was then used on the subsequent laser assembly for the results presented here.

7. Conclusions and Future Work

The simple addition of a mirror to the output of the Monoblock laser dramatically reduces the beam divergence of the output. A factor of two or more can be obtained, as borne out by the

experimental data, without sacrificing any output energy. The only requirement is the accuracy

needed to align the feedback mirror with the OPO resonator. This alignment tolerance is similar

to that required for typical laser mirror alignment and can easily be achieved.

Unlike the GRM or birefringent lens which function with a radial dependence, the composite

mirror has an angular dependence. Rays that pass through the etalon formed by the composite

mirror normal to the axis are preferably fed back into the OPO resonator and reduce the

threshold for those modes. The angular dependence on the feedback into the cavity is a unique

property of the composite mirror.

This phenomenon may not be required for most laser types, but for low cost systems like the

Monoblock, it can be a significant improvement. Other techniques such as curved optical

surfaces or GRM coatings can also improve the system’s divergence, but at significantly higher

system complexity and cost. Non-plano surfaces can be mass produced, but then require

reasonable skill to align with its inherent labor cost. Since most applications require the

divergence of the beam exiting the system to be less than 1 mR, anything that can reduce the

divergence before the telescope will reduce the need for a large exit aperture. The

70 71

diffraction limit of 1.57 µm wavelength light is approximately 2.0 mm-mR. Laser beams greater

than the diffraction limit require greater magnification to meet the required final beam

divergence for the system. Using the composite mirror technique, the required magnification is

reduced by the same factor as the reduction in beam divergence. To put this in perspective, the

unmodified Monoblock has a beam quality of approximately 30 mm-mR. The beam diameter

and divergence are 2.5 mm and 12 mR respectively. The output is 15X over the diffraction limit.

Without the feedback mirror the typical exit aperture would require a minimum beam diameter

of 30 mm to get the desired divergence of 1 mR, and the aperture must be larger than this to

prevent diffraction of the beam due to clipping. With the feedback mirror installed the same

aperture can be 15 mm without sacrificing beam divergence. Also, as the system development

cycle continues there will always be a demand for longer range operation. The ability to improve

output brightness is directly related to that development path. The feedback mirror will be a

pivotal component in meeting these more stringent requirements while maintaining the size and

weight limitations placed on man portable systems.

Future work will consider other laser systems that include a nonlinear element.

Specifically, the intracavity frequency doubled Nd:YVO4 (neodymium vanadate) laser [36] may

benefit greatly from this technique. This laser is composed of all plano surfaces and mode

control is defined by the size of the pump area. Unfortunately, to generate the output powers

needed for the illumination applications the pump area must be significantly larger than the diffraction limited spot size. The result is an output beam with a very bright core, but considerable power is outside this core center and greatly increases the output beam divergence.

Unlike the Monoblock that uses an intracavity oscillator to generate the 1.57 µm output, the

72 vanadate laser uses a simple KTP frequency double to convert the 1 µm light into the green at

532 nm. Without an oscillator present the feedback mirror may not significantly modify the conversion process. In addition to trying a feedback mirror at the harmonic frequency, a feedback mirror for the fundamental frequency should also be explored. As discussed in this treatise, a feedback mirror at 1 µm did improve the beam divergence in a cw or quasi-cw operating condition. The vanadate laser is cw. If the feedback mirror technique proves useful for this system it will open up the possibility of extremely compact, low cost green lasers that can illuminate at long ranges with minimal size restrictions.

8. Bibliography

1. L. Goldberg, J. Nettleton, B. Schilling, W. Trussell, and A.D. Hays, “Compact laser

sources for laser designating, ranging and active imaging,” Proc. SPIE, vol 6552

2. S.A. Hovanessian, Introduction to Sensor Systems, (Artech House, Inc., Norwood, MA,

1988)

3. R. L. McCally, C. B. Bargeron, J. A. Bonney-Ray, and W. R. Green John Hopkins APL

Tech Dig., vol 26, p 46, 2005

4. A. D. Hays, Vernon King, C. W. Trussel, and S. J. Hamlin, “Eyesafe erbium glass microlaser,” Proc. SPIE, vol 5332

5. Z. Chu, U. N. Singh, and T. D. Wilkerson, “A self-seeded SRS system for the generation of 1.54 µm eye-safe radiation,” Opt. Commun., vol 75, pp. 173-177, 1990

6. R. Burnham, R.A. Stolzenberger, and A Pinto, “ Infared Optical Parametric Oscillator in

Potassium Titanly Phosphate,” IEEE Photonics Tech Let., vol 1, pp. 27-28, 1989

7. Thomas J. Kane and Robert L. Byer, “Monolithic, unidirectional single-mode Nd:YAG ring laser,” Opt. Let., vol 10, pp. 65-67, 1985

8. W. J. Kozlovsky, C. D. Nabors, R. C. Echardt, and R. L. Byer, “Efficient monolithic

MgO:LinbO singly resonate optical parametric oscillator,” Opt. Let., vol 13, pp. 1102-1104,

1988

73

9. J. E. Nettleton, B. W. Schilling, D. N. Barr, and J. S. Lei, “Monoblock laser for a low-

cost, eye-safe, microlaser range finder,” Appl. Opt., vol 39, pp. 2428-2432, 2000

10. F. Jenkins and H. White, Fundementals of Optics, 4th ed. (McGraw-Hill, New York,

1976)

11. W. Koechner, Solid State Laser Engineering, 5th ed. (Springer-Verlag, New York, 1999)

12. J. Falk, J. M. Yarborough, and E. O. Ammann, “Internal Optical Parametric Oscillation,”

IEEE J. Quantum Electron., QE-7, pp. 359-369, 1971

13. T. Chuang, R. Burnham, “Multiband generation of mid infrared by use of periodically

poled lithioum niobate,” Opt. Let., vol 23, pp. 43-44, 1998

14. Sean Groth, “LMR-3000-031 Statistical Performance Analysis,” internal FLIR report,

2011

15. Richard A. Baumgartner and Robert L. Byer, “Optical Parametric Amplification,” IEEE

J. Quantum Electron., QE-15, pp. 432-444, 1979

16. L.R. Marshall, Jeff Kasinski, A.D Hays, and R. Burnham, “Efficient optical parametric

oscillator at 1.6 µm,” Opt. Let., vol 16, pp. 681-683, 1991

17. Kiyoshi Kato and Eiko Takaoka, “ Sellmeier and thermo-optic dispersion formulas for

KTP,” Appl. Opt., vol 41, pp. 5040-5044, 2002

18. A. Yariv, Quantum Electronics, 2nd ed. (Wiley, New York, 1975)

19. Larry R. Marshal and A. Kaz, “Eye-safe output from noncritically phase matched

parametric oscillator,” J. Opt. Soc. Am B, vol 10, pp. 1730-1736, 1993

20. A. E. Siegman, “Nonlinear optical effects: an optical power limiter,” Appl. Opt., vol 1,

pp. 739-744, 1962

74

21. Tso Yee Fan and Robert L. Byer, “Diode Laser-Pumped Solid-State Lasers,” IEEE J.

Quantum Electron., vol 24, pp. 895-912, 1988

22. J. Haden, G. Harnagal, and J. Endriz, “High-energy diode-laser bar pulsing of long

storage time solid-state materials,” Proc. ASSL DPL5, 1990

23. Ring-Ling Chen, “Shift of mode-hopping wavelength in single mode semiconductor

lasers,” Appl Spect, vol. 47, pp. 1716-1719, 1993

24. VLOC, Nd:YAG company catalogue

25. P. Thiagarajan, J. Nettleton, C. Cao, M. Sheldon, J. Helmrich, M. McElhinney, “Multi-

Color QCW Array For Uncooled Pumping of Nd:YAG Laser, Proc. SSDLTR, 2007

26. B. Cole, internal US Army NVESD communiqué, 2012

27. Lasertel, Conductively cooled laser diode catalogue

28. G. Hernandez, Fabry-Perot Interferometers, (Cambridge University Press, Cambridge,

1988)

29. K. Battou, K. Ait Ameur, “Transverse effects in compound resonators,” Opt. Com., vol.

183, pp. 189-194, 2000.

30. S. DSilvestri,Vittorio Magni, Orazio Svelto, and G Valentini, ”Lasers with Super-

Gaussian Mirrors,” IEEE J. Quantum Electronic., vol 26, pp. 1500-1509, 1990

31. M. Morin and M. Poirier, “Stable resonator with a graded reflectivity output coupler for moderate gain lasers,” CLEO CThM17, 2000

32. M.K. Chun and E.A. Teppo, “Laser resonator: an electrooptically Q-switched Porro prism device,” Appl. Opt., vol 15, pp. 1942-1946, 1976

33. G. Giuliani, Y. K. Park, and R. L. Byer, “Radial birefringent element and its application

to laser resonator design,” Opt. Let. vol 5, pp. 491-493, 1980 75

34. Amnon Yariv and Pochi Yeh, Optical Waves in Crystals, (John Wiley & Sons, New

York, 1984)

35. Alexander A. Kaminshii, Laser Crystals, 2nd ed. (Springer-Verlag, New York, 1990)

36. T. Sasaki, T. Kojima, A. Yokotani, O. Oguri, and S. Nakai, “Single-longitudinal-mode operation and second harmonic generation of Nd:YVO4 microchip laser,” Opt. Let. vol 16, pp. 1665-1667, 1991

76