SPIE Proceedings [SPIE Lasers and Applications in Science and Engineering - San Jose, CA (Saturday 22 January 2005)] Solid State Lasers XIV: Technology and Devices - Overview of the

Overview of the Lucia laser program: towards 100 Joules, nanosecond pulses, kW averaged power, based on Ytterbium Diode Pumped Solid

State Laser

J.-C. Chanteloup*, H. Yu*,°, G. Bourdet*, C. Dambrine*, S. Ferré*, A. Fülöp+, S. Le Moal*, A. Pichot*, G. Le Touzé* and Z. Zhao

*Laboratoire pour l’Utilisation des Lasers Intense (LULI), Unité Mixte de Recherche 7605, École Polytechnique-CNRS–CEA-Université Paris 6.

École Polytechnique, Route de Saclay, 91128 Palaiseau CEDEX - France

+Laselec S.A. Laselec IdF, École Polytechnique, 91128 Palaiseau CEDEX - France

°Research Center of Laser Fusion, CAEP P.O.Box: 919-988, 621900 - Mianyang city, Sichuan province - P.R. of China

Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, P.O. Box 800-211, Shanghai 201800, P.R. of China

ABSTRACT

We present the current status of the Lucia laser being built at the LULI laboratory, the national civil facility for intense laser matter interaction in France. This diode pumped laser will deliver a 100 Joules, 10 ns, 10 Hz pulse train from Yb:YAG using 4400 power diode laser bars. We first focus on the amplifier stage by describing the reasons for selecting our extraction architecture. Thermal issues and solutions for both laser and pumping heads are then described. Finally, we emphasize more specifically the need for long-lifetime high-laser-damage-threshold coatings and optics.

Key words: QCW diodes, Diode Pumped Solid State Laser (DPSSL), Diodes array, Yb:YAG, Laser damage threshold.

INTRODUCTION

Lucia is a Diode Pumped Solid State Laser (DPSSL) based on the Yb3+ lasing ion hosted in YAG. When frequency converted, the output 1030 nm, 100 Joules, 10 ns, 10 Hz pulse train will be an efficient pump source for high repetition rate PetaWatt class solid state laser like OPCPA [1] or Titanium-sapphire laser chains. Such sources could have applications in protontherapy for instance. In terms of industrial applications, direct use of the Lucia radiation would allow a 20-times [m²/hour] increase in metallic surface peening. In the field of laser-matter interaction physics, Lucia will motivate the development of high repetition rate targets and diagnostic capabilities. In section 1 we present the comparative study leading us to select our 3-pass extraction architecture. Both pump and extraction laser radiations experience an internal reflection inside the gain medium, allowing thus a doubling of the effective pump and amplification paths. This extraction scheme favors the overall efficiency. Two oscillators have been build, the Lucia Front-End (LFE) designed to deliver 1 Joule per pulse and an industrial version (~100mJ) we refer here as the Industrial Oscillator (IO). Both are described in reference 2. In section 2 we discuss the fact that thermal management issues arise while operating at 10 Hz the Pump Delivery Optical System (PDOS); experimental and modeling studies performed on PDOS of both oscillators are presented.

Solid State Lasers XIV: Technology and Devices, edited by Hanna J. Hoffman,Ramesh K. Shori, Proceedings of SPIE Vol. 5707 (SPIE, Bellingham, WA, 2005)0277-786X/05/$15 · doi: 10.1117/12.588341

105

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 02/20/2014 Terms of Use: http://spiedl.org/terms

In the same section 2, we also give a brief description of the 220 Joules diodes panel used for pumping the two amplifierheads.In section 3, we cover two key aspects of the laser head. We briefly present two different architectures for cooling thecrystal and we summarize our current work on damage threshold testing at high repetition rate (~40J/cm²–10Hz –1 hour).

1. Extraction architecture

1.1. Multiplexing

YAG crystal exhibits isotropic behavior regarding spectroscopic properties, therefore the stored energy in the mediumcan be extracted in several passes by either angular or vectorial (polarization) multiplexing (figure 1). As an example, theMercury laser [3] relies on the first technique since its Yb:SFAP gain medium exhibits a strong anisotropy for extractioncross section (7.3 vs 1.5.10-20 cm²).

Figure 1: Four passes energy extraction with vectorial (left) and angular (right) multiplexing architectures. Amplifiers and deformablemirror planes are conjugated with telescopes not shown in these schematic views. Gain media are AR coated for both pump and

extraction wavelengths.

Vectorial multiplexing relies on the use of a set of polarization optics. The input low-energy linearly-polarized pulseperforms two round trips into the ring (undergoing a 90° polarization rotation after first round). A second 90° rotationsends it towards the 0° deformable mirror (through the beam splitter). The amplified pulse goes back then for anotherpair of round trips before being extracted along the same path it came from.Although vectorial multiplexing appears as an elegant compact option for the Lucia laser, such extraction architecturepresents weaknesses. The main drawbacks for this extraction architecture result from efficiency and damage thresholdissues for polarization optics. Indeed, splitting and rotation efficiencies of beam-splitter and rotation-plate have to beextremely good in order to avoid any counter-propagative amplification. Also, these optics would have to maintain suchhigh efficiency while sustaining relatively high laser fluence (~10 to 20 J/cm²) for a long time (~108 shots). Also, unlessrelying on the use of a kW average power Pockels cell or a Faraday rotator, separation between input and output beamwould still require a certain amount of angular multiplexing.

1.2. Gain medium surfaces optical functions

Because of a high pump saturation intensity (28 kW/cm² at room temperature) and the two-energy-levels properties ofYb:YAG, a high pumping intensity (~20 kW/cm²) and a relatively high doping concentration of ytterbium ions (10 at %)are required to obtain a high efficiency from Lucia (~20 %) from a reasonably short crystal.The level of output energy we look for (100 J) implies a gain medium having a transverse dimension in the fewcentimeters range. Also, heat removing from the laser gain medium is critical in terms of thermal management, a keyfactor for high repetition rate operation of DPSSL. High average power resulting from running Lucia at 10 Hz repetition

106 Proc. of SPIE Vol. 5707


rate and Yb:YAG thermal properties (K= 7 W/mK for our doping) impose consequently a relatively thin thickness (in themillimeter range).It is moreover important to achieve a uniform pump energy deposition across the extraction beam section in order toobtain a homogeneous and efficient amplification. It follows consequently that any side pumping geometry should be ruled out.Finally, the gain medium should absorb the pumping light ( pump = 940 nm) as much as possible in order to increase thesystem efficiency. In order to satisfy this constraint while pumping our thin gain medium, a one-side double-passpumping scheme is required.These considerations lead to the following characteristics for the Lucia gain media: ~4 cm by ~2 mm Yb:YAG diskswith a High Reflectivity (HR) coating ( pump) on one side and Anti Reflection (AR) coating ( pump) on the other side.Based on these characteristics, important differences arise while studying different combinations of HR/AR coatings (at

extraction = 1030 nm) on surfaces of the gain media.The pumping parameters used for the following calculations are as follows: Two-heads/One-side-pumping architecturewith a 1 ms – 20 kW/cm2 pump intensity for each disk. The injected laser pulse duration is 10 ns.

1.2.1. AR/AR coating for extraction wavelength (1030nm)

Let us first study cases when both faces of the disk are AR coated at 1030 nm, i.e. extraction geometries described onfigure 1 where the extraction beam is going through the gain media.Two cases can then be explored: - Case (1a): AR/AR coating for extraction and AR/HR coating for pump.

- Case (1b): AR/AR coating for both pump and extraction.

As pointed out above, geometry of case 1b does not lead to an efficient extraction. The calculations show that onlyabout 72 % of the pump light is absorbed in a single pass (1b), while more than 93% is absorbed after two passes ofabsorption (with a back reflection for the pump, 1a). As expected, the system efficiency will be much higher for case 1abecause of a resulting high saturation pump intensity of Yb:YAG. For a laser injection intensity of 10 MW/cm², themaximum system efficiency s, defined as the ratio between the laser fluence output and the pump fluence, is 28.6 %after the 4th pass for 1a while it is 18.5 % after the 6th pass for 1b (figure 2). Reaching the same fraction of absorbedpower (~90 %) would be possible for case 1b, if we accept working with a twice longer crystal (~0.3 cm).

(1a) (1b)

Figure 2: Output laser fluence versus amplification pass number. The difference between (1a) and (1b) is that there is no reflection of pump light at the medium back surface for (1b): Rp= 0.0 vs Rp= 0.99. The pump light not absorbed is 6.6 % for (1a) and 28 % for (1b).

Total losses per pass is (1-Tpass) = 10 %.

Yet, solution 1b suffers from the fact it implies the use of a dichroic coating with performances difficult to achieve:AR for 1030 nm, HR for 940 nm, high damage threshold (~30 J/cm²) and long lifetime (108 shots). In order to overcome

Proc. of SPIE Vol. 5707 107


this issue, an alternative solution consists in coating the medium back surface with HR for 941 nm and 1030 nm whereasthe front surface is AR coated for both wavelengths.

1.2.2. AR/HR coating for all wavelengths

We explore here a configuration where the medium back surface is HR coated for 941nm and 1030nm and the frontsurface is AR coated also for both wavelengths. The pump light and injected laser parameters are the same as usedabove. The amplification stage architecture is consequently somehow modified (figure 3).

Figure 3: Four pass energy extraction architecture based on angular multiplexing. Gain media are AR/HR coated for both pump andextraction wavelengths. Cooling is performed through the HR coated surface.

The calculation results show that, for a laser injection intensity of 10 MW/cm², the maximum system efficiency s,is 32.8 % in the 3rd pass. Such configuration is really unique. During each pass, the stored energy within each slab is extracted twice. So, the nominal N-pass amplification is in fact equivalent to 2 N passes amplification, and moreinteresting, the single-pass-losses are only counted for N times. As a result, with the same pumping and laserinjection, the laser intensity will be amplified more quickly and reach the same output fluence with fewer passes compared with the case of section 1.2.1. And thus it is no surprise that this configuration exhibits higher systemefficiency (figure 4).

Figure 4: Output laser fluence versus amplification pass when the gain medium is AR/HR coated for both pump and extractionwavelengths.

Moreover, this (1 N losses / 2 N amplification) property relaxes the requirement in term of energy injection level(figure 5).



Figure 5: Spatial small signal gain coefficients versus distance through gain media for each passes of energy extraction. The label“Pass-1: +z” stands for the extracted laser beam first travel through the crystal (AR to HR face). The label “Pass-1: -z” stands for the travel after reflection onto the HR surface. The injected laser intensities are 10 MW/cm2 (left) and 10 kW/cm2 (right). It appears that a very low laser injection can also extract the stored energy efficiently because of equivalent 8-times extraction. And more interesting,the efficiency for a lower laser injection is a little higher than that of a higher injection, because of single-pass-losses of laser light.

1.3. Relationship between the system performance and temperature of Yb:YAG

According to our current thermal modeling, the gain medium temperature would reach approximately 80 C (~353 K).The relationships between the output laser fluence and the temperature for each pass of amplification are shown in figure 6.

Figure 6: Output laser fluence versus pass number for several gain medium temperatures. Pump intensity delivered at the entrancesurface of the gain medium is 20 kW/cm² and the medium back surface reflectivity is 99 %. The total single-pass-loss of laser beam is10%. Each temperature is obtained with the heat exchange coefficient in the parentheses. The Operating Point (OP) for Lucia is close

to the diamond curve.



These curves show that our laser system output will significantly decrease above 80°C. Additionally, from Fig. 6 and 7, we can find that a fourth pass would be useless on Lucia since the gain will be less than the loss during this fourth pass,consequently lowering the overall system efficiency (Our OP indeed induces a temperature below 400°K).

Figure 7: Relationship between the system efficiency (calculated after 3rd and 4th pass) and the temperature. Because of FP (pump fluence) not including the loss (1-T) of Pump Delivery Optical System, the overall optical-to-optical efficiency should be Tot= ,

where T is about 80 %.

1.4. 3 pass - 100 Joules level amplification through angular multiplexing

Figure 8 describes the Lucia amplifier stage architecture. The combined used of angular multiplexing and adaptive opticsallows high efficiency extraction in 3 passes from the two water-cooled Yb:YAG amplifiers. Injection and pick-upmirrors are located in the vicinity of foci to facilitate angular multiplexing.

Figure 8: Lucia amplifying stage layout. The seed pulse coming from the LFE is inserted into this amplifying stage through theInjection mirror. It then encounters a pair of amplifier. The amplified pulse is redirected for a second pass by Pick-up and Reverser

mirrors. Just before pass 3, the pulse experiences wave front shaping while being reflected by a Deformable mirror. A few degrees tilton this mirror allows the output pulse to travel above the Pick-up mirror (see insert).



2. Pumping headWe have developed two oscillators in the framework of the Lucia program. Detailed descriptions of these two unstableresonators are given in reference [2].

2.1. Pump Delivery Optical System (PDOS)Delivering diode light at pump onto the gain medium with a maximum transmission (~80 %) and a good homogeneity isthe main task assigned to the Pump Delivery Optical System (PDOS). Experimental values for these quantities are givenin reference [2].

2.1.1. Thermal issues with silica light ductOne key element of the PDOS is the pair of mirrors (called light duct) used to concentrate the diode light along the slowaxis of the 25 bars stacks. Our first set of light ducts was based on silica mirror having a metallic (SiO2 protected)coating. This solution appeared satisfying for the Industrial Oscillator (IO), although a small, but easily detectable(~10°C), temperature rise (figure 9) was observed. This leads us to study more carefully the thermal behavior of thisoptics.

Figure 9: Light ducts thermal study: (A) visible light side view; (B) Finite element modeling of the IO silica light duct; (C) Thermalimaging (HgCdTe sensor) before lightening the diodes (absolute temperatures are in °C); (D) Relative temperature (°C) elevation

image obtained after reaching the thermal steady state (~20 minutes). The nose temperature increase by ~10°C.

We achieved a good understanding of the origin of the observed thermal rise by cross-evaluation of our finite elementthermal code with a thermal imaging experiment based on a HgCdTe sensor. The heat source is the diode light non-absorbed by the silver layer (figures 10 and 11). The low thermal conductivity (< 1 W/mK) of silica produces anaccumulation of heat localized at the nose of the light duct.



0,0

5,0

10,0

15,0

20,0

25,0

30,0

35,0

40,0

40°-45°

45°-50°

50°-55°

60-65°

60-65°

65°-70°

70°-75°

75°-80°

80°-85°

85°-90°

Incident angle

Frac

tion

of ra

ys(%

)

97,50

98,00

98,50

99,00

99,50

100,00

0 20 40 60 80

Angle (°)

Refle

ctiv

ity (%

)

Figure 10: Angular distribution of pump rays onto the light duct reflective layer (left) and its reflectivity (right).

Total optical power absorbed by each IO light duct is about 200 mW with a maximum density around 60 mW/cm² (dutycycle is 1 % - 1 ms/10Hz). This leads to a maximum 12.5°C temperature elevation according to the modeling whereasthe experiment shows a 11+/- 1°C elevation.

Figure 11: Deposited pump light peak intensity distribution onto the light duct silver layer: Two dimensions map (left) and axiallineout (right).

Confident in our modeling, we used it to for the Lucia Front-End silica light duct. We observed that the thermal loadwould be about three times higher than for the IO light duct, consequently inducing a maximum temperature of ~60°Cfor the nose. Such value would affect the silver coating optical performance on the long run and induce turbulence in theair in the vicinity of the nose, i.e. the laser head. Such an effect would perturb seriously the oscillator beam path bygenerating optical distortions. We choose then to engineer a water-cooled metallic light duct for the LFE. A singlechannel of water localized at the nose appeared to be enough (figure 12). The modeling predicts a maximum temperatureof 30°C without water cooling and almost no temperature rise when water-cooled. The same technology will be used forthe amplifiers light ducts where a maximum pump light peak intensity of 2.3 kW/cm² is expected.



Figure 12: Metallic light ducts used for the Lucia Front-End Pump Optical Delivery System. The maximum pump light peak intensitydeposited near the nose of this mirror is around 1.6 kW/cm².

2.2. Diode panels

Diodes panels are manufactured by DILAS, GmbH; one panel is required per amplifier head. Each of them is made of8 columns of 11 stacks. Each stack is made of 25 diode bars, leading to a total number of 2200 bars able to deliver amaximum of 264 kW peak power (or 264 J). The 88 stacks are fixed onto a water-cooled copper plate.The stacks will be driven with a 10 Hz, 1 ms, 140 A square electrical signals with a DC offset varyingbetween 1 and 5 A. This bias current (smaller than the threshold current) allows a fine adjustment of the wavelength (~1 nm/A).

Figure 13: Computer generated view of one of the 88 stacks (right) arranged in 8 columns in the diode panel (left).

3. Laser Head

Pumping and laser heads of both Lucia front-end and amplifier stage are somehow homothetic in geometry. We then useour LFE to explore scalable architectures for diode light delivery (see section 2) as well as cooling mechanisms.The LFE by itself is described in reference [2]. Two key aspects require a specific attention at the laser head level:Gain medium cooling and coating.



3.1. Gain medium cooling

The two approaches currently under investigation to address the heat removal issue are described on figure 14. Both relyon the use of room temperature water. They differ in the way the heat is removed from the Yb:YAG disk. Architecture 1(left of fig. 14) allows water to be in direct contact with the HR surface of the disk whereas the alternative architecture(right) relies on the use of a glue acting as a thermal bridge and holding the crystal onto a CuW plate. In both cases thekey issue is to maintain the stringent performances of the HR (99 % reflectivity – 30 J/cm² damage threshold) throughoutlifetime of 5.108 shots. The influence of water or glue on this coating needs a complete experimental analysis.

Figure 14: Top and side views of Lucia Front-End laser heads. Disk diameter is 12 mm with a thickness of 1.6 mm.

In both configurations optical phase distortions due to thermally induced surface curvature and index of refractionvariation have been modeled using finite element codes. We are using thermal imaging to evaluate temperature and werely on interferometery to have access to optical phase distortion in order to check code predictions.We explore also different structures for the gain medium; for instance we consider working with a composite structure ofdoped and undoped Yb:YAG both in crystalline and ceramic form. Although we already have at our disposal 4 cm² 10%doped ceramic Yb:YAG, so far only the crystal based laser head have been activated and we obtain 165 mJ in freerunning regime from our Industrial Oscillator whereas 500 mJ have been reached with the LFE.

3.2. Laser damage fluence issues

Operational conditions of Lucia imply severe constraints in terms of laser induced damage threshold of its optics. Alloptics (bulk materials and coatings) should indeed be able to sustain a laser fluence of at least 15 J/cm² in the nanosecondregime, at 10 Hz and for a large number of shots, typically 108. One must indeed keep in mind that extraction fluenceshould be kept above saturation fluence in order to quickly extract stored energy form the amplifying medium (saturationfluence for our Yb:YAG gain medium is around 9 J/cm²)The curves of figure 15 show evolution through amplification of both energy and maximum fluence levels. Thehorizontal axis lists the hundred and so interfaces encountered by the amplified beam. The beam path is describedon figure 8. The energy curve (right scale) exhibits a four-step amplification pattern for each 3 passes. This feature is dueto Lucia double-pass/two-amplifiers extraction geometry described in section 1. Injection level is around a few mJwhereas an output power in the range of 100 J is expected. The fluence curve (left scale) exhibits 3 peaks where the injection and pick-up mirrors are located. The maximumfluence is reached on the AR surface of the second amplifier after completion of the third pass.



Figure 15: This diagram shows logarithmic evolution of energy (right scale) and maximum fluence (left scale) levels along the opticalpath of the pulse. The horizontal axis lists the hundred and so interfaces encountered by the amplified beam.

Crystal or Ceramic Yb:YAG bulk laser damage fluence threshold should not be an issue since it is known to be above100 J/cm² [4]. Polishing is a key factor for achieving high damage threshold values. Yb:YAG crystal and ceramics we are using are respectively 5 and 3 Å rms polished.In order to match our requirements in terms of damage fluence threshold, we are testing several AR and HR coatingstechnologies, namely SolGel for AR, HfO2/SiO2 for AR and HR.Yb:YAG SolGel spin coated samples are currently under test for R/1 damage. To match our specifications, thesecoatings follow an intermediate technical path between the colloidal and the polymeric routes.We have already shown (figure 16) that vaporized Hafnia (HfO2) under Oxygen atmosphere was giving satisfying resultsfor HR function on silica. We will soon test this technology on YAG.

Figure 16: This graph shows the distribution of approximately 100.000 shots at a fluence varying between 30 and 45 J/cm²for 10 ns – 1064 nm laser pulses. Three sites were tested for an hour at 10 Hz on a silica sample coated with a HR [HfO2/SiO2] coating

(CEA/SAGEM technology). Reflectivity measured before and after testing on the different sites did not reveal any degradation.



SUMMARY

We described the architectural choice of the DPSSL Lucia and pointed out the key issues which are thermal management and laser fluence damage threshold of optics. Lucia, together with Mercury, Halna [5,6], and Polaris[7] pioneer the field of high repetition rate DPSSL and will offer new opportunities ranging from pump source for high repetition rate PW laser system, laser peening, laser driven inertial fusion,…

ACKNOWLEDGEMENTS

Work funded by the Ministère de la Recherche, through the Délégation Régionale à la Recherche et à la Technologie. Calculations presented on figure 15 have been obtained by using the software CEA/MIRÓ V.5.24.

REFERENCES

[1] I.N. Ross, P. Matousek, G.H.C. New and K. Osvay, “Analysis and optimization of optical parametric chirped pulse amplification”, JOSA B, Vol.19, No.12, pp. 2945-2956 (2002). [2] A. Fülöp, G. Bourdet, J.-C. Chanteloup, C. Dambrine, S. Ferré, S. Le Moal, A. Pichot, G. Le Touzé, H. Yu and Z. Zhao, “Diode pumped Yb:YAG V-shape unstable super gaussian laser resonators, for 10 Hz – 100 joules class laser”, this conference. [3] C. Bibeau et al., "Mercury and beyond: Diode-pumped Solid-State Lasers for Inertial Fusion Energy", Comptes Rendus de l'Académie des Sciences Série IV-Physique-Astrophysique, Vol. 1, No. 6, pp. 745-749 (2000). [4] J.-F. Bisson, Y. Feng, A. Shirakawa, H. Yoneda, J. Lu, H. Yagi, T. Yanagitani and K.-I. Ueda, “Laser Damage Threshold of Ceramic YAG”, Jpn. J. Appl. Phys. Vol. 42, Part 2, No. 8B, pp. L 1025–L 1027 (2003). [5] T. Kawashima, T. Kanabe, O. Matsumoto, R. Yasuhara, T. Sekine, T. Kurita, M. Yamanaka, H. Furukawa, M. Miyamoto, T. Kanzaki, H. Kan, N. Miyanaga, M. Nakatsuka, Y. Izawa, S. Nakai, and C. Yamanaka, "HALNA-100 DPSSL driver for inertial fusion energy", Conference on Inertial Fusion Science and Applications 2003, B.A. Hammel, D.D. Meyerhofer, J. Meyer-ter-Vehn and H. Azechi ed., Am. Nucl. Soc., pp. 568-571 (2004). [6] T. Kawashima, O. Matsumoto, M. Miyamoto, T. Sekine, T. Kurita, S. Matsuoka, T. Kanzaki, H. Kan, T. Kanabe, R. Yasuhara, Y. Fukumoto, T. Ashizuka, M. Yamanaka, T. Norimatsu, N. Miyanaga, M. Nakatsuka, Y. Izawa, H. Furukawa, S. Motokoshi, C. Yamanaka, H. Nakano and S. Nakai, "Diode-pumped zig-zag slab laser for inertial fusion energy", OSA TOPS Vol. 94, Advanced Solid-State Photonics, Gregory J. Quarles, ed. pp. 282-287 (2004). [7] J. Hein, S. Podleska, M. Siebold, M. Hellwing, R. Bödefeld, R. Sauerbrey, D. Ehrt and W. Wintzer, “Diode-pumped chirped pulse amplification to the joule level”, Appl. Phys. B, Vol. 79, No. 4, pp. 419-422 (2004).



Documents

SPIE Proceedings [SPIE Lasers and Applications in Science and Engineering - San Jose, CA (Saturday 22 January 2005)] Solid State Lasers XIV: Technology and Devices - Overview of the