Thesis. Submitted to. The School of Engineering of the UNIVERSITY OF DAYTON. In Partial Fulfillment of the Requirements for.

Transcription

1 SEEDED, GAIN-SWITCHED CHROMIUM DOPED ZINC SELENIDE AMPLIFIER Thesis Submitted to The School of Engineering of the UNIVERSITY OF DAYTON In Partial Fulfillment of the Requirements for The Degree Master of Science in Electro-Optics By Sean A. McDaniel UNIVERSITY OF DAYTON Dayton, Ohio August, 2012

2 SEEDED, GAIN-SWITCHED CHROMIUM DOPED ZINC SELENIDE AMPLIFIER Name: McDaniel, Sean A. APPROVED BY: Peter E. Powers, PhD. Advisory Committee Chairman Professor, Department of Physics Andrew Sarangan, PhD. Committee Member Electro-Optics Program Patrick A. Berry, PhD. Committee Member Research Physicist Air Force Research Lab Kenneth L. Schepler, Ph.D. Committee Member Research Physicist Air Force Research Lab John G.Weber, Ph.D. Associate Dean School of Engineering Tony E. Saliba, Ph.D. Dean, School of Engineering & Wilke Distinguished Professor ii

3 ABSTRACT SEEDED, GAIN-SWITCHED CHROMIUM DOPED ZINC SELENIDE AMPLIFIER Name: McDaniel, Sean A. University of Dayton Advisor: Dr. Peter E. Powers Many scientific and military applications require pulsed laser sources with high peak power output which are tunable throughout the midwave infrared (mid-ir) spectral region. In this report we discuss the design, construction, and characterization of a gain-switched chromium-doped zinc selenide (Cr:ZnSe) amplifier pumped by a Q- switched holmium-doped yttrium aluminum garnate (Ho:YAG) laser and seeded by a free running continuous wave (CW) Cr:ZnSe laser. The amplifier pump laser was constructed using a 0.5% Ho-doped, Brewster-cut YAG rod. In CW operation, powers of up to 3.68 W and a slope efficiency of 45% were obtained. In pulsed operation at 1 khz pulse repetition frequency (PRF), pulse energies of 2.6 mj per pulse were obtained with temporal pulse widths less than 100 ns. The output wavelength of the pump was 2.1 μm with a spectral width less than 1 nm. iii

4 The pulsed Ho:YAG laser was then used to gain-switch a Cr:ZnSe single-pass pulsed amplifier seeded with a CW Cr:ZnSe laser with a free-running wavelength of 2.4 μm and a spectral linewidth of 50 nm. The output of the gain-switched amplifier yields a pulsed beam at the seed wavelength demonstrating high gain. By design, this source will follow the lasing wavelength of the seed beam allowing for tunability over the entire emission wavelength of Cr:ZnSe with proper seed design. The end result of this work was a versatile, pulsed mid-ir source. iv

5 TABLE OF CONTENTS ABSTRACT... iii TABLE OF CONTENTS... v LIST OF FIGURES... vii LIST OF TABLES... xi CHAPTER 1 INTRODUCTION BACKGROUND PROBLEM STATEMENT RESEARCH OBJECTIVES... 5 CHAPTER 2 DESIGN AND CONSTRUCTION OF THE AMPLIFIER PUMP LASER LASER MATERIAL SELECTION CHARACTERIZATION Ho:YAG CAVITY DESIGN CONSTRUCTION Ho:YAG CHARACTERIZATION CW Ho:YAG CHARACTERIZATION (Q-SWITCHED) CHAPTER 3 CONSTRUCTION OF A GAIN-SWITCHED POWER AMPLIFIER v

6 3.1 SEED SELECTION BEAM COMBINATION CHAPTER 4 CHARACTERIZATION OF THE GAIN-SWITCHED AMPLIFIER GAIN SWITCHED TESTING PULSE ENERGIES OTHER RESULTS CHAPTER 5 CONCLUSION WORKS CITED APPENDIX: M 2 MATLAB CODE vi

7 LIST OF FIGURES Figure 1: Atmospheric transmission from 1-12 μm (1)... 1 Figure 2: Gain-switched setup... 4 Figure 3: MOPA configuration with relative amplification behavior. The red arrow represents the behavior of the seed beam Figure 4: Emission and Absorption cross section of Cr:ZnSe... 8 Figure 5: Measured Ho:YAG absorption spectrum of.2% and 2% samples. Green blocks are the wavelength of peak absorption. Note the spectrophotometer switches filters at 2000 nm Figure 6: Laser transition of Ho 3+ ions Figure 7: Output of Tm fiber laser measured by the Yokogawa OSA Figure 8: Literature emission cross section of Ho:YAG (16) Figure 9: Optical transition of Ho 3+ that leads to absorption occurring at emission wavelengths 13 Figure 10: End-pumped, linear cavity for optical pumping of Ho:YAG Figure 11: End pumped, folded cavity for optical pumping of Ho:YAG Figure 12: LASCAD cavity for the Ho:YAG laser in X-Z and Y-Z planes. The green lines show the mode shape while the yellow gradient shows the location of the Ho:YAG crystal Figure 13: Distance vs. Focal length of a Galilean telescope. Red and blue lines correspond to F1 and F vii

8 Figure 14: Ho:YAG crystal, mount and incident HeNe (note, the direction of the arrow is the direction of propagation) Figure 15: Brewster cut Ho:YAG geometry Figure 16: Alignment spots from Ho:YAG cavity. The beams from M1 and M3 are aligned to the center spot Figure 17: Polarization of the thulium pump laser. Each plot plot s y-axis is scaled for viewing.. 22 Figure 18: Transmission of HO50 and HO09 optics Figure 19: Ho:YAG cavity with curved outcoupler Figure 20: Slope efficiency of Ho:YAG with available outcouplers. The data points for each experimental measurement are shown in the legend along with the calculated slope efficiencies Figure 21: Output spectrum of variable outcouplers Figure 22: M 2 plot of Ho:YAG output (CW) Figure 23: Q-switched slope efficiency Figure 24: Setup of Ho:YAG laser Figure 25: Q-switch pulse width Figure 26: Emission spectrum of free-running Cr:ZnSe seed laser. Note that this spectrum changes over time due to thermal fluctuations, power fluctuations and lack of an intracavity tuning element Figure 27: Cr:ZnSe Seed laser pumped by a thulium fiber laser (blue) outputting at peak emission, 2450 nm (red) Figure 28: Gain switching setup. The pump and seed beams are combined using the combining mirror then focused into the amplifier using the 15 cm lens. The beam is then collimated using the 7.5 cm lens viii

9 Figure 29: Beam overlap of pump and seed beams Figure 30: Overlap of the pump and seed beams with seed and pump shown separately Figure 31: Emission spectra after chromium amplifier showing lines from the Ho:YAG, Cr:ZnSe seed laser and an unknown line at 2500 nm Figure 32: Emission through chromium amplifier after realignment Figure 33: Scan with 300 g/mm grating Figure 34: Gain switching scans with Cr 2+ : ZnSe (blue) and without (red) Figure 35: Chromium signal with lock-in triggered at 10kHz. The horizontal dotted lines represent the approximate noise level. The scan was taken with a step size of.1 nm and integration time of 100 ms Figure 36: Crystal length required for 90% absorption Figure 37: Chromium output with crystal (blue) and without crystal (red) Figure 38: Realigned and unclipped chromium signal triggered off an optical chopper at 100 Hz Figure 39: Ho:YAG emission with chromium amplifier (red) and without (blue) chopped by an optical chopper Figure 40: Cr 2+ signal triggered at 10kHz, red trace is un-maximized and blue is maximized Figure 42: Monochromator scans with increasing pump power Figure 43: Incident pump power versus max counts from the scans in Figure Figure 44: Transmission of CO62 optic used for combining and splitting of beams Figure 45: Transmission of LP2000 filter Figure 46: Gain switch measurement setup Figure 47: Scope traces of gain-switched pulse (blue) and pump pulse (red) Figure 48: Gain-switched pulses from the pump beam (red) and seed (blue) ix

10 Figure 49: Scope trace with the seed signal blocked Figure 50: Scope trace with the pump signal blocked Figure 51: Measured absorption of the pump beam through Cr:ZnSe amplifier and CO62 splitting mirror of the Ho:YAG pump beam Figure 52: Extracted pulse energies from Cr:ZnSe amplifier at 10kHz PRF corrected for mirror reflectivities Figure 53: Gain for the gain-switched amplifier Figure 54: Peak pulse power of the gain-switched pulses plotted against the absorbed pump power Figure 55: Peak power gain of the Cr:ZnSe seed laser x

11 LIST OF TABLES Table 1: Material properties of ZnSe host material... 2 Table 2: Power absorbed for Cr182 sample Table 3: Pulse energy absorbed for Cr182 sample at 10 khz PRF Table 4: Specifications for Cr:ZnSe crystals Table 5: Comparison of results xi

12 CHAPTER 1 INTRODUCTION 1.1 BACKGROUND Midwave-infrared (Mid-IR) laser sources are in great demand for military and commercial applications dependent on the atmospheric transmission windows, shown in Figure 1, as well as organic and inorganic chemical absorption features in the 2-5 and 8-12 μm regions. These laser sources have other uses that include manufacturing, medical and scientific applications. Figure 1: Atmospheric transmission from 1-12 μm (1) Many of these research-driving applications require high peak powers to achieve a desired effect, necessitating that the source operate in a pulsed mode. Additionally, 1

13 these applications can require the source to be wavelength-agile to tune in and out of spectral features. These combined requirements mean that a tunable, pulsed source is of great importance to this area of work. Laser radar and infrared counter measures (IRCM) are a few specific areas that require sources with temporal and spectral robustness. A solid state laser system with short pulses allows for an improvement in spatial and temporal resolution ideal for accuracy in laser radar systems. High pulse energy and broad tunability also makes for an ideal system for IRCM. Townes and Schawlow laid the groundwork for the first solid state laser in 1958 with their description of infrared and optical masers (2), which lead to Maiman s demonstration of the first ruby (chromium-doped sapphire) laser in 1960 (3). Transitionmetal (TM)-doped, II-VI host materials have started to be used more commonly, with output in the 2 to 5 µm spectral region. II-VI compounds, consisting of a group II metal and a group VI chalcogen, have excellent IR transmission characteristics. Zinc selenide was used as the crystal host due its favorable properties seen in Table 1. Table 1: Material properties of ZnSe host material Property ZnSe κ (Wm -1 K -1 ) 18 dn/dt(10-6 /K) 70 n (~λ lasing ) 2.45 Hardness (Knoop) 160 2

14 In Cr:ZnSe, the Cr 2+ ion displaces Zn in the crystal structure. This displacement has optical transitions with low non-radiative transition rates at room temperature resulting from low phonon vibration frequencies (4). For this reason, Cr:ZnSe has become a widely available, tunable laser medium that can be tuned across a large wavelength band from 2 to 3 µm (5). Such lasers have been demonstrated in several different configurations [ (6), (7), (8)]. The large tuning range in Cr:ZnSe occurs due to the crystal field splitting of the electronic energy levels. The allowed transitions of the crystal field splitting give rise to the large emission range of Cr:ZnSe. Development of high pulse energy Cr:ZnSe lasers is inhibited by the excited state lifetime of chromium (9). The short excited state lifetime, 5-7 μs, hampers the energy storage ability of Cr:ZnSe. This limits the energy output of a chromium laser when used in a standard actively Q-switched cavity due to the fast decay time of the excited state population. The fast decay time does not allow fluence levels to build up sufficiently. To counteract the lack of energy storage in Cr:ZnSe, pulsed operation usually occurs in a mode-locked or gain-switched configuration. For sub-mhz PRFs and temporal widths greater than 100 ps, gain-switched operation is required. Gain-switching is an initial buildup phenomenon in lasers where fast pumping causes a population inversion and thus gain, to go considerably higher than threshold before the laser oscillation has time to build up (10). In this process, the pumping rate is much faster than the rate of spontaneous emission allowing the high population inversion to take place. Following this excitation, the extraction from the gain-switched 3

15 process must also be fast compared to the spontaneous emission rate. This extraction is accomplished by either making the cavity lifetime short compared to the emission lifetime or making the gain of the amplifier large. In general, high pump pulse energy leads to shorter build up time for the gain-switched pulse. Thus, a short gain-switched pulse can be obtained. In Cr:ZnSe this approach is especially effective because, as a fourlevel laser, the transitions from pump level to lasing level are very fast. Additionally, the pulse energies achievable are no longer limited by the short upper-state lifetime, but by the available gain. Gain-switched pumping is a technique for obtaining a pulsed signal by modulating the gain of a laser cavity. One way this technique works is by using a pulsed source as a pump while seeding an amplifier with seed signal. A visual interpretation of this concept can be seen in Figure 2. Pump Signal Amplifier Gainswitched seed Leftover pump Seed Signal Figure 2: Gain-switched setup 4

16 It is seen that the pulsed pump signal and the seed signal are coincident on the amplifier. The gain in the amplifier is then modulated as the pulses from the pump travels through the amplifier. The seed signal picks the wavelength that is amplified by the pump signal. 1.2 PROBLEM STATEMENT Operation of a cavity with high-pulse energy has associated risk of damage to optical components and coatings. A system is required which can have robust operation under these requirements. In most configurations, a high pulse energy is needed to gain-switch Cr:ZnSe. High pulse energies associated with these configurations can damage certain optical components. A seeded gain-switched amplifier removes most of the sensitive elements from high incident pulse energy theoretically allowing for high pulse energies to be extracted from the amplifier. This configuration also allows for expansion of the beam diameter to keep peak intensities below damage levels. 1.3 RESEARCH OBJECTIVES The objective of this research was to design and build a gain-switched Cr:ZnSe amplifier using a MOPA (Master Oscillator Power Amplifier) configuration, see Figure 3. MOPAs have previously been used in CW operation to improve beam quality and output tunability (11). This work will focus on the amplifier using an existing CW Cr:ZnSe oscillator. The project is then divided into two separate sections. The sections are as follows; design, selection and construction of a pump source (Chapter 2), construction 5

17 of the seeded Cr:ZnSe amplifier and construction and characterization of the Cr:ZnSe gain-switched amplifier (Chapter 3 & 4). The end result of this project is an operational seeded gain-switched amplifier capable of producing a tunable, high pulse energy output. Figure 3: MOPA configuration with relative amplification behavior. The red arrow represents the behavior of the seed beam. 6

18 CHAPTER 2 DESIGN AND CONSTRUCTION OF THE AMPLIFIER PUMP LASER 2.1 LASER MATERIAL SELECTION In order for gain-switching to be feasible, a high pulse energy pump is needed with a frequency that is within the absorption spectrum of Cr:ZnSe. Several factors led to the selection of a Ho:YAG laser as the pump source. First, in-band pumping of Ho:YAG using a 1900 nm source has previously been demonstrated which reduces risk and development time [ (12), (13)]. Second, the lifetime for Ho:YAG is on the order of milliseconds, making it ideal for efficient energy storage. In order to most effectively pump the Cr:ZnSe amplifier, high pulse energy is needed. This means that for continuous pumping, efficient energy storage is needed from the laser crystal and thus a long upper-state lifetime is desired. When comparing Ho:YAG to other pump sources such as an OPO (optical parametric oscillator)pumped by a Q-Switched Nd:YAG laser (14), the simplicity, efficiency and high pulse energy make Ho:YAG an ideal choice for this experiment. 7

19 Figure 4: Emission and Absorption cross section of Cr:ZnSe 2.2 CHARACTERIZATION In order to better understand Ho:YAG, a spectroscopic scan was made of two test samples using a Cary 5000 spectrophotometer with 1 nm resolution. The samples had a doping concentration of 0.2 and 2%. The absorption spectra of these two samples can be seen in Figure 5. The 2% sample was 4.7 mm thick and the 0.2% sample was 5.05 mm thick. The spectra in Figure 5 were corrected for thickness variations. 8

20 1 0.9 Absorption of Ho Samples HO18(2%) HO17(0.2%) OD Wavelength (nm) Figure 5: Measured Ho:YAG absorption spectrum of.2% and 2% samples. Green blocks are the wavelength of peak absorption. Note the spectrophotometer switches filters at 2000 nm. From the absorption spectrum shown in Figure 5, it s seen that the most efficient pumping wavelength for Ho:YAG is 1907 nm. The absorption line at 1907 nm is due to the transitions within the 5 I 5 8 I 7 manifolds of the Holmium dopant (15). The energy level diagram for Ho 3+ ions can be seen in Figure 6. The wavelength produced by the thulium 3 H 3 4 H 6 transition makes a thulium laser ideal for optically pumping Ho:YAG. 9

21 Figure 6: Laser transition of Ho 3+ ions An IPG thulium fiber laser operating at 1908nm was purchased specifically for this application. The laser output was measured with a Yokogawa AQ6375 OSA (optical spectrum analyzer) and can be seen in Figure 7. This figure shows that the laser was emitting at nm with a spectral width of 1.01 nm or less. The stated spectral width is limited due to the measurement settings on the OSA. Spectral widths of the order of 1 nm generally do not matter because the peak absorption region of Cr:ZnSe is very broad. This nm wavelength is ideal for pumping Ho:YAG due to the overlap of the absorption peak at 1907 nm with the emission of the Tm laser. 10

22 Figure 7: Output of Tm fiber laser measured by the Yokogawa OSA After determining the pump wavelength, the emission spectrum was examined in order to confirm that the lasing wavelength will be correct for pumping Cr:ZnSe. The emission spectrum for Cr:ZnSe is shown in Figure 4. Comparing the emission spectrum from the literature (16), shown in Figure 8, to the absorption cross section, Figure 4, shows that the emission of Ho:YAG overlaps the absorption spectrum of Cr:ZnSe. The peaks of emission lead to the laser lines of Ho:YAG occurring around 2090nm. This ensures that Ho:YAG is suitable for pumping of Cr:ZnSe. 11

23 x Emission of Ho:YAG em Wavelength (nm) Figure 8: Literature emission cross section of Ho:YAG (16) The emission peaks match the peaks in the absorption spectra at 2100 nm. There is a small amount of reabsorption of emitted light at the lasing wavelength but the amount of emitted light is approximately a factor of 4 higher than the amount reabsorbed. At room temperature, the lower manifold has a significant population distribution due to the thermal excitation. Under optical excitation, the absorbers are excited to the upper manifold which has an excited state lifetime of approximately 7 ms. In terms of energy, it can be thought that optical transitions can happen in either direction; the same transition that yields lasing at 2100 nm can also lead to absorption at 2100 nm. Figure 9 shows a likely transition that could cause absorption at lasing wavelengths. 12

24 Figure 9: Optical transition of Ho 3+ that leads to absorption occurring at emission wavelengths 2.3 Ho:YAG CAVITY DESIGN There are two main designs that could be used for fiber pumping of Ho:YAG. The first design is an end-pumped hemispherical resonator (17). Figure 10 shows the layout of this cavity where M1 is a planar mirror AR(anti-reflective) coated at 1900 nm and HR(high-reflective) at 2100 nm and M2 is a plano/concave mirror coated to be an outcoupler for the 2100 nm wavelength. Figure 10: End-pumped, linear cavity for optical pumping of Ho:YAG 13

25 The Ho:YAG crystal is Brewster-cut, which enforces polarization in the cavity output. In theory, the pump radiation could be double-passed by having the outcoupler coated to reflect the pump, but the AO Q-switch could cause problems with double passing of the pump. The second design is a folded cavity, end-pumped, double-pass hemispherical resonator [ (18), (19)]. This design, shown in Figure 11, offers several advantages over the end pumped cavity. The folded design allows for a more compact footprint so it can easily be fit onto an optical breadboard. It also allows for easier incorporation of the AO Q-switch and more efficient double pass of the pump beam without interfering with the Q-switch. In Figure 11, M3 is HR at 1900 nm and AR at 2100 nm. M2 is HR at 2100 nm and AR at 1900 nm at a 45 angle of incidence. M1 is a variable percentage outcoupler coated for 2100 nm. 14

26 Figure 11: End pumped, folded cavity for optical pumping of Ho:YAG The end pumped, folded cavity was chosen for its small space requirements and versatile configuration. Using the configuration shown in Figure 11, a cavity was modeled in LASCAD (Gaussian ABCD matrix software). This model required knowing specific lengths of the cavity. In order to approximate these parameters, components were laid out on a breadboard to approximate the dimensions needed. A total cavity length of 13 cm was obtained. The Ho:YAG crystal was 5 cm long and the AO Q-switch was 4 cm long. These values were inputted into LASCAD which uses the ABCD matrix method to determine the cavity mode size, Figure

27 Figure 12: LASCAD cavity for the Ho:YAG laser in X-Z and Y-Z planes. The green lines show the mode shape while the yellow gradient shows the location of the Ho:YAG crystal The supported mode size inside the cavity with a planar outcoupler and a 50 cm focal length mirror is around an 800 µm 1/e 2 diameter. Knowing the supported mode size, the pump beam needed to be resized to allow for a well matched mode overlap. The thulium fiber laser produced an original spot size of 4.2 mm. In order to obtain optimal mode overlap, the mode sizes of the input beam and supported cavity mode size need to be the same. A Galilean telescope design was used to resize the pump beam. The focal lengths of each lens in a Galilean telescope were found using Equations (1) and (2). 16

28 (1) (2) In the above equations, d is the spacing between the lenses and MP is the magnification of the lenses. The magnification power is found by dividing the focal lengths of the available optics; in this case, the magnification power is 0.2. Plotting equations (1) and (2) with respect to d, the spacing between the two lenses can be found. This plot is shown in Figure 13. Figure 13: Distance vs. Focal length of a Galilean telescope. Red and blue lines correspond to F1 and F2. The green horizontal lines are the available -5 cm and 25 cm focal length lenses which intersect the red and blue lines at 20 cm. This implies that in order for a collimated beam to exit the setup, the distance between the two lenses needs to be 20 cm. 17

29 2.4 CONSTRUCTION Using Figure 11 as a guide, a suitable Ho:YAG cavity was constructed. To keep the cavity compact, all components were placed as close as possible to each other. A shorter cavity means a lower threshold when lasing occurs (20). To start, a HeNe laser was aligned to two irises along with the fiber laser. This is done to ensure a straight, level, coaligned beam. The HeNe laser was used to aid in alignment and for safety due to the non-visible, high power lasers being used. Next, the Galilean telescope was constructed using the available -5 cm and 25 cm focal length lenses that are AR coated for 1.9 µm. These lenses were installed in a lens tube with an adjustable end to allow for fine adjustment of the length. The distance between the two lenses needed to be 20 cm. This is of course different for longer wavelengths so the lenses were aligned then adjusted to give a collimated beam for 1.9 µm pump laser. This distance turned out to be 18.6 cm. Now with a well collimated beam, the input coupler, M2, was inserted. M2 is an in-coupler coated AR at 1.9 µm, 45º, S polarization and HR at 2.1 µm, 45º, P polarization to maximize the mirror performance while matching the cavity s preferred polarization induced by the Brewster cut crystal. The reflection from the HeNe was aligned so that the incident beam and the reflected beam were exactly perpendicular. This confirms that the cavity is exactly an L shape and the two legs of the cavity will be perpendicular. The Brewster cut Ho:YAG crystal was mounted in a water-cooled aluminum heatsink. The mounted Ho:YAG crystal was then attached to a 1 post and a translation stage. The mount was then positioned so the HeNe was incident on the front 18

30 crystal face at Brewster s angle as shown in Figure 15 and Figure 14. Brewster s angle is found using Equation 3. ( ) (3) For Ho:YAG with a refractive index of 1.8, at an interface with air, Brewster s angle is found to be 61. Figure 14: Ho:YAG crystal, mount and incident HeNe (note, the direction of the arrow is the direction of propagation) 19

31 Figure 15: Brewster cut Ho:YAG geometry Mirror M1, HR at 2.1 µm, was placed where the incident beam exited the Brewster cut crystal. The HeNe was then aligned back through the crystal onto the original spot on the input coupler. The spot was reflected perpendicular to the input beam onto mirror M3. In turn, the beam is reflected back onto the spot on the input coupler by M3. This allowed for the cavity to be roughly aligned. To ensure fine alignment, the beams passing though M2 were coaligned. These beams can be seen in Figure

32 Figure 16: Alignment spots from Ho:YAG cavity. The beams from M1 and M3 are aligned to the center spot. The three spots in Figure 16 are always stationary. The multiple reflections are from the input coupler, front face and back face. When aligning M1 and M3, two small spots can be seen. These spots were aligned on top of each other, then aligned to the center spot in the above figure. At this point, the thulium pump laser was turned on and the cavity was aligned in attempt to cause lasing. After fine alignment of the system, two main issues arose. The first problem encountered was a severe amount of scattering from the front surface of the Brewster crystal. This was due to instability in the pump polarization. The instability can be seen in Figure 17. The fluctuations are quite severe. The P -polarization changes by several hundred milliwatts over the course of twenty minutes. The specifications that were shipped with the IPG thulium laser quote an approximately stabilized output after 2 hours. 21

33 Figure 17: Polarization of the thulium pump laser. Each plot plot s y-axis is scaled for viewing The graph above was obtained by placing a polarizing beam cube in front of the output of the thulium laser. The polarizing beam cube works by splitting the polarization of the beam into S and P polarizations. The P polarization transmits straight through the cube while the S polarization is reflected by the dielectric coating on the angled face of the beam cube. From Figure 17, it s seen that the S and P polarizations vary greatly over the course of the 20 minute measurement. To guarantee that only P polarization is incident on the crystal, a polarizing beam cube splitter and a half wave plate were installed on the thulium pump laser. The second problem encountered was feedback into the fiber laser through the incoupler. This was due to the pump reflection at M3. The amount of the pump beam reflected back through the Ho:YAG sample and out the incoupler was enough to cause feedback instability in the fiber laser. To remedy this 22

34 situation, the cavity end mirrors were reversed. The 50 cm curved mirror at M3 was replaced with the mirror at M1, a flat mirror that was HR at 2.1 µm and AR at 1.9 µm. The transmission through both sides of these mirrors (M1 is labeled HO09 and M3 is HO50) was measured using a Cary 5000 spectrophotometer. The scans comparing these two mirrors can be seen in Figure 18. Figure 18: Transmission of HO50 and HO09 optics In Figure 18, HO09 is the 50cm curved mirror and HO50 is the flat mirror that replaced the curved mirror. By switching these two mirrors, it is seen that the transmission at 1.9 µm is increased by approximately 50%. The new setup is seen in Figure 19. The only disadvantage to this setup is that the output is now coupling out of the curved mirror. This tends to create a larger spot size with faster beam divergence than if the beam was coupled out of the flat mirror. 23

35 HO09 HO50 Figure 19: Ho:YAG cavity with curved outcoupler The fiber pump laser was again turned on. The cavity was aligned using the same procedure as before. The only difference in the alignment was the alignment spots from each mirror switched positions. The alignment procedure was iterated though. Each time, the signal seen by the photodiode was maximized by adjusting the alignment of the cavity mirrors. Once the system was in alignment, the signal exhibited spiking. This phenomenon is a precursor to lasing. After adjusting the cavity mirrors more, lasing occurred. 2.5 Ho:YAG CHARACTERIZATION CW Referring back to Figure 5, it is seen that the emission peaks in Ho:YAG are relatively close together. Lasing can occur on either of these peaks while optically pumping with a 1907 nm source. The lasing wavelength is chosen by losses in the cavity. The variable percentage outcouplers introduce enough loss to the cavity to allow the 24

36 mirrors to choose the lasing wavelength. The available outcouplers are 90%, 75%, 70% and 50% reflective. With each of the respective outcouplers, the mirror was installed and aligned. A slope efficiency measurement and a spectrum analysis were taken of each outcoupler. The slope efficiency graph for the available outcouplers is seen in Figure 20. Output Power(W) Slope Eff of all OCs 70% Exp % Exp % Exp % Exp Input Power(W) Figure 20: Slope efficiency of Ho:YAG with available outcouplers. The data points for each experimental measurement are shown in the legend along with the calculated slope efficiencies. The difference in outcoupler affects the threshold and max attainable power slightly. This can also be attributed to thermal effects in the crystal. The slope efficiencies are shown in the graph above as the solid lines. It should be stated that the measured losses varied substantially over time. This is due to the instability of the Tm fiber laser. Power ranges being dumped by the beam cube and passed through the crystal varied by tens, if not hundreds, of milliwatts over the course of the power reading. The power losses 25

37 attributed to the beamcube, unabsorbed pump and the laser output were all recorded. This ensures that the slope efficiency is due to the absorbed pump power, not the total power. Looking at the lasing profile with respect to each outcoupler, the varying slope efficiencies can be explained. Figure 21: Output spectrum of variable outcouplers For the output couplers under 90%, the center wavelength is around 2090 nm. For the 90% output coupler, the lasing wavelength is around 2122 nm. The difference in lasing wavelengths is due to reabsorption losses within the laser crystal. The spectral width of each of the above graphs is approximately 1 nm. 26

38 The absorption through the crystal can be estimated by measuring the losses at the polarizing beam cube splitter and the amount of pump that makes it through the crystal. By measuring these values and subtracting them from the input power, the absorption efficiency can be fitted to the model for Ho:YAG (13). ( ) (4) In equation 4, G is the ratio between output and input power, crystal, P is the number of passes of the pump beam and is the length of the is the doping concentration in percentage. The values of, and G are 0.5%, 5cm and.046 respectively. Using these values, is equal to 86.2%. The absorption could be double passed to improve the efficiency. The 50% outcoupler was chosen for use in the final setup due to the lasing wavelength and slope efficiency. The lasing wavelength of the 50% outcoupler was 2.09 µm and the slope efficiency was the highest with a value of 45.6%. This slope efficiency was compared to other published slope efficiencies; 74% (13), 65% (21) and 14% (19). Setting the cavity with a 50% outcoupler allowed for characterization of the output beam. Ho:YAG readily emits in the range of µm. From Figure 20 and Figure 21, it is seen that the slope efficiency changes with output wavelength. With the 50% outcoupler set, the beam quality of the beam exiting the Ho:YAG cavity was characterized with a Thor Labs scanning slit beam profiler (Model BP109-IR2). The data obtained from the profiler was then input into the M 2 Matlab code. 27

39 D x (z) ( m) Beam Profile (X-Dim) M 2 = 1.52 z x (0) = cm w 0 (x) = m D y (z) ( m) Beam Profile (Y-Dim) M 2 = 1.45 z y (0) = cm w 0 (y) = m z (cm) z (cm) Gaussian Beam Fit 400 x(z),y(z) ( m) z (cm) Figure 22: M 2 plot of Ho:YAG output (CW) The Ho:YAG output has an M 2 of 1.52 in the x-direction and 1.45 in the y-direction. A larger M 2 is expected in the x-direction due to the astigmatism introduced by the Brewster cut Ho:YAG crystal. M 2 values as high as 1.8 (18) and as low as 1.15 (22) have been reported. The values in Figure 22 fall within the range of acceptable M 2 values. 2.6 Ho:YAG CHARACTERIZATION (Q-SWITCHED) With the Ho:YAG characterized running CW, the Q-switch was now inserted and aligned. This of course increased the losses of the cavity thus raising the threshold power. The threshold power for either case is approximately 8W. Allowing for proper warm up time for the Tm fiber pump laser, the Q-switch was then turned on. Operating at 1 khz rep rate, the slope efficiency in Figure 23 was collected. 28

40 5 Slope Eff using a 50% OC Outpower(mJ) Input Power(W) Figure 23: Q-switched slope efficiency The slope efficiency obtained for the Ho:YAG laser is 39.8%. This is close to the value for the CW laser but slightly lower due to losses introduced by the Q-switch. The pulse energy at 13.8W of pump is approximately 2.6mJ. The final setup of the holmium laser is seen in Figure 24. The pump beam path is highlighted in red while the holmium florescence/laser path is highlighted in blue. The shared path of the pump and florescence/laser is highlighted in purple. 29

41 Figure 24: Setup of Ho:YAG laser At a repetition rate of 1 khz, a pulse width of 100 ns was obtained using an extended range InGaAs detector. The pulse trace from the oscilloscope is seen below. Pumping the cavity with the maximum available power shortened the pulse width to nearly 50 ns pulses. Figure 25: Q-switch pulse width 30

42 CHAPTER 3 CONSTRUCTION OF A GAIN-SWITCHED POWER AMPLIFIER 3.1 SEED SELECTION Gain switching is a method of producing laser pulses by manipulating the gain of a material. In the case of this project, the gain in a Cr:ZnSe sample is modulated by the Ho:YAG pump source which is Q-switched. The Ho:YAG pulse at 2100 nm is in the absorption region of the Cr:ZnSe thus allowing the 2.1 µm pulse to optically pump the Cr:ZnSe. In order to gain switch pump the Cr:ZnSe, a seed source was used to select the output wavelength. A seed source needs to emit within the emission band of the amplifier. Therefore, the seed signal was chosen to be a free-running CW Cr:ZnSe laser (11). This laser was already built and available for use. It was pumped by an IPG thulium fiber laser and the laser cavity is shown in Figure 27. The Cr:ZnSe laser free runs at 2450 nm and has a spectral width of almost 50 nm at 100 mw of power, Figure 26. x Emission of free-running Cr:ZnSe Counts Wavelength (nm) Figure 26: Emission spectrum of free-running Cr:ZnSe seed laser. Note that this spectrum changes over time due to thermal fluctuations, power fluctuations and lack of an intracavity tuning element. 31

43 Figure 27: Cr:ZnSe Seed laser pumped by a thulium fiber laser (blue) outputting at peak emission, 2450 nm (red) 3.2 BEAM COMBINATION After establishing an amplifier pump source and a seed source, the pump and seed beams needed to be combined prior to entering the power amplifier. This was accomplished using a dichroic mirror that was AR at 2.1 µm and HR at 2.45 µm. The two beams were combined through this mirror and aligned to 2 irises and 2 lenses which ensured that the two beams were coaligned over the sample region. The 2 lenses used were a 15 cm focal length lens AR coated for 2.1 µm and a 7.5 cm lens also AR coated for 2.1 µm. The setup can be seen in Figure

44 Combining Mirror 15 cm lens 7.5 cm lens Figure 28: Gain switching setup. The pump and seed beams are combined using the combining mirror then focused into the amplifier using the 15 cm lens. The beam is then collimated using the 7.5 cm lens. The key to an efficient gain switched process is overlap between the pump and seed beams. The beams have to be overlapped in order to maximize the interaction of the two beams within the crystal. For this gain-switched process, where the pump needs to be fully utilized, the pump beam must be fully overlapped by the seed beam in the crystal as shown in Figure 29. If the seed beam is too small, the pump energy is not efficiently deposited into the seed beam. Thus, there are less chromium ions available with higher intensities, which will lead to saturation at lower powers than expected. 33

45 Figure 29: Beam overlap of pump and seed beams The beam overlap efficiency at a single point in the amplifier can be expressed by equation 5. (5) From this overlap integral, and are intensities of the pump and seed beams and is the differential area of the beam size. The ideal beam overlap efficiency would be 1 when the two beams are perfectly overlapped. For the case shown in Figure 29, the theoretical beam overlap efficiency will be less than 1 due to the pump beam being encompassed by the seed beam. Ignoring diffraction effects and knowing that the seed beam is larger than the pump beam, a generic non-focusing case can be used to estimate the beam overlap. Assuming an intensity of the form shown in equation 6, the beam overlap efficiency becomes equation 7. (6) 34

46 (7) The values for (pump) and (seed) were measured and are approximately 200 µm and 250 µm respectively. These spot sizes approximately represent the beam overlap at two points at either side of the amplifier. These values are not spot sizes at the focal plane of the focusing lens. Large spot sizes away from the focus were used as a precaution to ensure the operation above the camera s damage threshold. Using these values and assuming that I 0 is the same for both beams, the integration was performed. From the integration, it was found that. Eta shows that the mode overlap is not perfect but the seed is slightly under filled. This result is slightly ambiguous. Usually a result of 1 is desired but due to the nature of this amplifier an overlap efficiency of less than 1 is acceptable as long as the pump beam is inside the seed beam. To check this result, an Electro-Physics PV320 camera was used to observe the beams. The camera was set up so that both beams were incident on the array. The beams were again aligned to the irises and observed with the camera. The beams are well aligned when the irises close symmetrically around the beam. Figure 30, shows the pump and seed beams along with the overlap of the two beams. Comparing these three pictures confirms that the seed beam is slightly larger than the pump beam ensuring that the pump beam is fully utilized. 35

47 Figure 30: Overlap of the pump and seed beams with seed and pump shown separately 36

48 CHAPTER 4 CHARACTERIZATION OF THE GAIN-SWITCHED AMPLIFIER 4.1 GAIN SWITCHED TESTING After coaligning the beams and ensuring a good optic axis, the Cr:ZnSe sample was placed in the space between the two irises. The first sample used, labeled Cr91, was an uncoated sample placed at Brewster s angle. Brewster s angle for Cr:ZnSe is. The sample dimensions were 2.5x6.8x7.7mm 3 and id had a doping concentration of 6x10 18 cm -3. Due to the crystal being at Brewster s angle, the beams had to be realigned into the monochromator. The beams were then steered and aligned to an Acton SP2500 monochromator allowing the spectral content of the beams to be analyzed. The scan is seen in Figure 31. The monochromator scans were initially used to investigate the behavior of the Cr:ZnSe amplifier. Using the monochromator, scans over the operating region, 2.1 μm to 2.5 μm, could be taken and examined without major realignment of the system. This was advantageous compared to using a dichroic mirror or other setup to examine amplifier behavior. The counts from the monochromator detector were used for initial investigations. 37

49 3.5 x 105 Emission of Ho:YAG and Cr through Cr Amp(brewster) Counts Wavelength (nm) Figure 31: Emission spectra after chromium amplifier showing lines from the Ho:YAG, Cr:ZnSe seed laser and an unknown line at 2500 nm. There are three distinct lines in this scan. The line at 2100 nm is left over from the Ho:YAG pump laser. The small peak around 2450 nm is from the free running Cr:ZnSe laser. The line at 2500 nm does not correspond to anything that was put into the system, which called for further investigation of what was actually happening. Time was taken to realign the monochromator and ensure that the beams were incident on the proper locations. Another scan was taken after this realignment process and is shown in Figure 32. This is more characteristic of gain switching though the output is shifted to 2500nm leading us to believe that gain switching was not actually occurring. 38

50 3.5 x 105 Emission of Ho:YAG and Cr through Cr Amp Counts Wavelength (nm) Figure 32: Emission through chromium amplifier after realignment showing lines at 2100 nm and 2500 nm The persisting peaks at 2500 nm lead to further investigation of the source causing the lines. Scans were taken one at a time with either the pump or seed beam blocked. Blocking the seed beam, a scan of the Ho:YAG pump beam was taken. This scan showed the same mystery line at 2500 nm. The scan with the seed beam blocked shows the line at 2500 nm can be attributed to the pump source. The uncoated Cr:ZnSe sample was replaced with a coated sample placed at normal incidence. The sample, CR182, is 7.77mm x 6.54mm x 1.9mm and has a doping concentration of 6.88e18 cm -3. The sample was replaced due to concern that the Cr:ZnSe amplifier was lasing between the two faces of the crystal. The Cr182 sample had AR coatings from 2-3 μm to ensure that no lasing was occurring due to reflections from the crystal faces. 39

51 The monochromator was checked for light leaks to ensure there was no extraneous signal. The Acton monochromator has 3 gratings, a 600 g/mm blazed at 1.6 microns, 300 g/mm blazed at 2 microns and 150g/mm blazed at 4 microns. Both the 600 g/mm grating and the 300 g/mm are suitable for use with the wavelengths in this experiment. To start, the 600 g/mm grating was used. To check the reproducibility of the phenomenon, the grating was switched to the 300 g/mm. A scan was taken from 2000 nm to 2510 nm with the 300 g/mm grating in place. The scan is displayed in Figure 33. From this figure, it is seen that the line at 2500nm has disappeared. This means that the narrow line at 2500 nm was actually an artifact of the grating. The artifact is known as a Lyman ghost. Ghost lines that appear at large spectral distances from the source line are called Lyman ghosts. These lines are caused by compounded periodic errors in the spacing of the grating grooves (23). A grating of n lines can be broken up into m groups of n lines. The intensity at any point of the spectrum of a grating of n lines is proportional to equation 9. ( ( ) ( ) ) (8) 40

52 5 x 105 Emission of Ho:YAG and Cr through Cr Amp 4 3 Counts Wavelength (nm) Figure 33: Scan with 300 g/mm grating In equation 9, a is equal to equation 10 where e is the spacing of the grating and and are the incident and reflected angles respectively. ( ( ) ( )) (9) Consider a grating of m groups. The intensity is then shown in equation 11. ( ( ) ( ) ) (10) In equation 11, for orders of the factor (n-1) will vanish for any A. This is due to vanishing for ( ). Assuming that there is an irregularity in the spacing of n, will not vanish at intervals of a. Allowing n to have a periodic error allows for fractional order lines to occur. Thus, will have appreciable values at intervals of the 41

53 irregularity. For example, consider a grating where every other groove is misplaced. This causes ghosts in half order positions. After changing the grating to 300g/mm blazed at 2 μm, the line at 2500 nm disappeared. This allowed more accurate spectra to be taken. Ensuring that the system was well aligned, new spectra was taken with the Acton monochromator. Scans were taken with the Cr 2+ : ZnSe amplifier crystal in place and with the crystal removed. Both scans can be seen in Figure x 105 Emission of Ho:YAG and Cr through Cr Amp 4 3 Counts Wavelength (nm) Figure 34: Gain switching scans with Cr 2+ : ZnSe (blue) and without (red) In Figure 34, there are slight differences in the two spectra. It appears that some of the Ho:YAG (2.09 μm) pump is being absorbed when comparing the shapes of the lines at 2.09 μm. The pump beam was operating at low power with significant attenuation in Figure 34 Looking at the seed lines, it appears that with the crystal in place the output 42

54 signal increased. This, however, does not confirm that the system is gain switching. To test whether the amplifier was gain switching, the SR320 lock-in amplifier was triggered off the 10 khz signal driving the Q-switch. This allows the monochromator to pick out anything that has close to a 10 khz signal. The signal obtained from the monochromator is shown in Figure x 104 Emission of Ho:YAG and Cr through Cr Amp Counts Wavelength (nm) Figure 35: Chromium signal with lock-in triggered at 10kHz. The horizontal dotted lines represent the approximate noise level. The scan was taken with a step size of.1 nm and integration time of 100 ms. The two horizontal lines spanning the graph are the estimated noise level based on the signal fluctuations with no signal present. The signal to noise ratio is about 3:1 so the signal is discernible over the noise. The low signal level can be attributed to several factors including low pulse energy and low signal. The average power from the Ho:YAG pump was 400 mw at a PRF of 10 khz for this experiment, giving an average pulse energy of 40 µj per pulse. The average pulse energy was calculated using equation

55 ( ) ( ) ( ) (11) This was a relatively small amount of pump energy which may lead to the signal not being amplified by a noticeable amount. Another factor that could contribute to low signal level in the monochromator is the amount of the pump absorbed by the Cr: ZnSe amplifier. The amount of pump absorbed was determined by measuring the amount of power present with the crystal in place and with the crystal taken out while pumping CW at around 400 mw. These measurements were measured using a GenTec MAESTRO power meter and a Newport model 818P power head averaged over 2 minutes and are displayed in Table 2. Table 2: Power absorbed for Cr182 sample Power without crystal(mw) Power with crystal(mw) Power absorbed(mw) Power absorbed(%) The same type of measurement was done with the pump laser operating Q-switched at 10 khz PRF. These measurements were taken as average power and then converted into pulse energies using equation 12 from above. Table 3: Pulse energy absorbed for Cr182 sample at 10 khz PRF Energy without crystal(μj) Energy with crystal(µj) Energy absorbed(µj) Energy absorbed (%)

56 From Table 2 and Table 3, the power absorbed by the Cr182 sample is around 50%. A way to raise the signal is to deposit more power into the crystal. To do this, a crystal with either higher doping or longer interaction length is needed. To calculate the length needed to get at least 90% absorption, equations 12 is solved for length (L), where T (transmission) = 10% and alpha is a sample-dependent absorption coefficient. (12) Using this equation, the graph shown in Figure 36 was produced. To determine which sample should be used for higher absorption, a crystal with the appropriate absorption coefficient and length must be used. There are several candidates that could be used for higher absorption; Cr108, Cr109 and Cr74. Specifications for these samples are listed in Table 4. 5 Crystal length required for 90% absorption Length (cm) (cm -1 ) Figure 36: Crystal length required for 90% absorption 45

57 Table 4: Specifications for Cr:ZnSe crystals Dimenstions (mm) Doping (cm^-3) α (cm^-1) Expected absorption Cr x6.62x E % Cr x6.15x E % Cr74 5x5x E % Cr x6.54x E % Closer examination of the specifications in Table 4 led to Cr108 being used. Cr108 was chosen over Cr109 due to the longer length to ensure most of the pump was absorbed. Referring to Figure 36, for α=2.1 cm -1 the ideal length to obtain 90% absorption is 1.1 cm. With Cr108 being slightly longer, this means that slightly more power will be absorbed. For Cr109, the length needed for 90% absorption is also around 1.1 cm which with Cr109 being 8.49 mm long will lead to the power absorbed being less than 90%. Finally, Cr74 was not chosen due to it being cut at Brewster s angle. Using Cr74 would cause the beam to walk off due to the angled faces much like the walk off seen in Figure 15. Thus, the beam and monochromator would need to be realigned. The Cr182 sample was taken out and replaced with the Cr108 sample. The system was realigned to ensure the overlapped beams were properly aligned to the Acton monochromator. Two scans were taken with the Cr:ZnSe amplifier in place and then again with it removed. The signal shown in Figure 37 was triggered off a chopper wheel that was placed before the monochromator entrance. 46

58 12000 Emission of Cr through Cr Amp Counts Wavelength (nm) Figure 37: Chromium output with crystal (blue) and without crystal (red) Comparing the traces from Figure 37 shows a small amount of gain in the edges of the seed signal. After some investigation, it was thought that the beam was being partially clipped due to the increased length of Cr108. The sample was realigned and care was taken to make sure that the beam was not being clipped. Another scan was taken after realigning the sample and is shown in Figure

59 14000 Emission of Cr through Cr Amp Counts Wavelength (nm) Figure 38: Realigned and unclipped chromium signal triggered off an optical chopper at 100 Hz The above figure shows increased gain across the entire seed signal. This ensures that the sample is aligned and the beams are coaligned. Likewise, if gain is seen at the seed wavelength then loss should be seen at the pump wavelength. Figure 39 shows a scan of the pump beam with and without the crystal in place. Just as expected, the pump beam is being absorbed by the Cr:ZnSe amplifier and some of that absorbed power is being emitted at the seed wavelength. 48

60 3.5 x 105 Emission of Ho through Cr Amp Counts Wavelength (nm) Figure 39: Ho:YAG emission with chromium amplifier (red) and without (blue) chopped by an optical chopper The scans seen in Figure 37 and Figure 38 does not necessarily prove that the amplifier is gain switching. In order to definitively demonstrate gain switching, the trigger signal was changed to trigger off the 10 khz signal that drives the Ho:YAG pump. Chopping the beam with an optical chopper will show average power differences. Triggering the lock-in amplifier at 10 khz will determine if there is a signal that corresponds to the 10 khz repetition rate at the seed wavelength. The corresponding scan, seen in Figure 40, shows that a signal is present at the seed wavelength with the appropriate 10 khz repetition rate. 49

61 Emission of Ho:YAG and Cr through Cr Amp un-maximized maximized Counts Wavelength (nm) Figure 40: Cr 2+ signal triggered at 10kHz, red trace is un-maximized and blue is maximized From this point, the overlap of the beams was adjusted to maximize the signal seen by photo-detector in the monochromator. The blue trace in Figure 40 is the maximized signal. While keeping the lock-in amplifier triggered at 10 khz, each beam was blocked to establish that the signal was not seen while either beam was blocked. No signal was seen that exceeded the noise level which confirms that the signal behaved as expected. When the pump is blocked, no signal was seen with a 10 khz repetition rate. Likewise, when the seed was blocked, no signal was discernable at the seed wavelength. The combination of signal present at 10 khz and no signal present when either the pump or seed is blocked lends itself to the amplifier gain switching. Now that the amplifier is confirmed to be gain switching, a preliminary method of measuring pulse energies was formed. This method will measure gain to ensure that the Cr:ZnSe amplifier has an increasing trend as the pump power is increased. The 50

62 method involved using the Acton monochromator to look at the general amplitude of the signal seen at 10 khz. The scans, seen in Figure 41, increase in power from left to right. Fluctuations in pump power can lead to slight differences in these traces. Figure 41: Monochromator scans with increasing pump power The pump pulse energies from left to right are 42.4 µj, 33.9 µj, 52 µj, 105 µj, 125 µj, 133 µj, 161 µj, 180 µj, 186 µj, 209 µj and 223 µj. From this point, the max counts of each scan in Figure 41 were taken. The monochromator counts were then averaged and the average counts were plotted against the incident pump power to obtain the graph seen in Figure 42. This shows that the amplifier output increases as pump power is increased. In other words, because this graph is approximately linear means that the amplifier is not being saturated by the incident pump beam. Thus, power scaling in this 51

63 regime will be linear with increasing pump power. Some points in the graph deviate from the fitted line. The deviation of these points is due to fluctuations in the seed beam power caused by the fluctuations in the seed s pump laser Pump Power vs Monochromator Counts Average Counts Pump Power (W) Figure 42: Incident pump power versus max counts from the scans in Figure 41 With sufficient evidence of gain switching and correct behavior in the amplifier, the temporal nature of the pulses from the amplifier were examined with two photodetectors. This involved using two photodetectors to look at the temporal properties of the pulses. With the two beams collinear, distinguishing the pump and amplified seed pulses become very difficult. This is due to the Ho:YAG pulse being much stronger than the gain switched pulse. To see the gain switched pulse over the Ho:YAG pulse, most, if not all, of the pump needs to be extinguished. This was done in a way similar to the combination of the beams discussed before. The same type of mirror that 52

64 was used to combine the beams was used to separate them. The spectral response of this splitting mirror is shown in Figure Transmission of CO Transmission Wavelength (nm) Figure 43: Transmission of CO62 optic used for combining and splitting of beams The splitting mirror, CO62, passes 99% of the pump wavelength (2.1 µm). Assuming that the laser is run at the highest pulse energy tested, 2.6mJ, the 1% of the leftover pump is approximately 25 µj of energy. Depending on the absorption efficiency of the amplifier, the induced gain switched pulse could be on the order of micro Joules. To mitigate any effects of the leftover pump, 3 splitting mirrors were used to extinguish the pump. Using three mirrors, only about 2.59 nj (13 μw) of pump energy will be present at the detector. The detector being used is a Newport 818P pyroelectric detector with a noise equivalent power of 0.5 µw. Converting the noise equivalent power into in an energy reveals that the noise equivalent energy is 0.5 nj. Thus, the power detector could possibly see the pump pulses still. To ensure this was not the case, a set of three long pass filters were inserted before the power meter. The spectral response of the LP

65 filters is shown in Figure 44. The filter transmits 65% at 2.09 µm. This further reduced the pump signal to a theoretical max of 0.11 nj. This value is far enough below the noise equivalent power of the detector to confirm only pulse energies from the gain switching process can be seen. For measuring temporal characteristics of the pulses, an extended range InGaAs detector was used. This detector can see energies well below.11 nj. This small amount of energy was neglected because the energy from the gain switched pulse will be much higher than.11 nj. 1 Transmission of LP Transmission Wavelength (nm) Figure 44: Transmission of LP2000 filter The setup of the gain switched amplifier with detectors is displayed in Figure 45. Two photodetectors were used in this setup. One was used to examine the pulses from the Cr:ZnSe seed beam while the other was used to observe the behavior of the Ho:YAG pump pulses. 54

66 Figure 45: Gain switch measurement setup The setup in Figure 45 was then used to look at the temporal characteristics of the pulses. Two scope traces were taken. In Figure 46, the blue trace is the detector facing the Cr:ZnSe amplifier. The blue trace detects the input pulse, fluorescence and gained signal. The red trace is the photodetector facing the Ho:YAG pump laser to observe the characteristics of the input pulse.. There are a couple of key features present in these traces that need to be addressed. The pulse from the detector looking at the Cr:ZnSe amplifier (blue) can see the fluorescence lifetime and a pulse similar in temporal response to the pump. The lifetime for Cr:ZnSe is well known to be approximately 5-8 µs [ (24), (25)]. Examining the pulse, it is seen that the fluorescence signal will have a lifetime in the range of 5-8 µs. 55