1. Introduction

Laser-damage testing is a familiar area for the chirped-pulse-amplification, high-power laser community, particularly those working on inertial confinement fusion. As the community strives to build petawatt-class lasers, new innovations must be made, including the fabrication of large-aperture, high-damage-threshold gratings used for pulse compression. Metallic gratings were initially used [1], but the development of gratings etched into multilayer dielectric (MLD) stacks has improved grating damage thresholds [2]. The damage morphology caused by high-power, short-pulse lasers has been well investigated, often at 800 nm and pulse lengths varying from tens of femtoseconds to nanoseconds, for various homogenous materials [3–5] and even multilayer dielectric stacks [6], but little work has been done with gratings [7]. In this letter, we report the observation of a ripple structure arising on test sites irradiated with short pulses (0.7 to 10 ps) at 1053 nm. The ripples appear as a secondary, overlapping grating and produce a strong dispersion effect with a colorful glint visible to the bare eye when viewed in white light.

Ripple formations were first observed in semiconductor materials irradiated with a Q-switched ruby laser [8] and, since then, ripples have been observed in many homogeneous dielectrics [4,9] and semiconductors and metals [10] tested under varying conditions. The new grating-like formation on MLD gratings, however, does not correspond to typical conditions in previous experiments. The theories that describe the ripple formation give a direct relation between the laser wavelength and the ripple period [11], which was not observed in the current morphology. Furthermore, this ripple formation phenomenon is observed to form under a number of conditions that included varying the angle of incidence and changing the beam polarization from s- to p-polarization. Finally similar ripples are found to form, albeit weakly, on gold coated gratings. Here we describe the phenomenology of the formation of periodic microstructures on multilayer dielectric gratings prior to total ablation, with the intention of explaining the underlying mechanism in the future.

2. Experimental setup

2.1 Geometry and laser system

The experimental setup for damage testing (Fig. 1) involves a commercial chirped-pulse–amplification (CPA) laser system (Positive Light, Inc.) producing a compressed pulse with up to 50 mJ of energy. The front-end of the laser system consists of a Time Bandwidth Products^™ GLX-200 diode-pumped, Nd:glass, master oscillator (~200 fs, 6.5 nm of bandwidth at 1053 nm), a four-pass, single-grating stretcher (1740 1/mm groove density, 61° incident angle, ~700-ps output chirped pulse), and a Spitfire Ti: sapphire regenerative amplifier (<0.5 mJ, 3.5 nm of bandwidth). After image relaying through the linear Nd:glass amplifiers (two 9.5-mm and one 12.7-mm-diam rods), the chirped pulse has a “top-hat” spatial profile, ~3 nm of bandwidth, and 65 mJ of energy at the compressor input. A two-grating (1740 1/mm groove density, gold, Lawrence Livermore National Laboratory), double-pass compressor (61° incident angle, 85-cm grating separation) permits the pulse duration to be varied in the range of 500 fs (transform-limited) to 100 ps. The system can be run at a repetition rate of approximately 1 shot per minute, as limited by the glass rods.

To characterize the pulse width throughout its entire range, a commercially available single-shot autocorrelator (Positive Light), a spectral-domain, single-shot autocorrelator [12], and an in-house-designed, scanning autocorrelator have been employed for the pulse-duration measurements. In the standard far-field configuration, the sample is situated in the focal plane of a 2-m focal-length mirror, and the laser beam is focused to a 360-μm spot size on the sample. To increase the illuminated area on the sample, the standard focusing mirror is replaced with a 2.5-m focal-length mirror, thereby effectively putting the sample and the diagnostic CCD in a quasi-far-field position. Damage sites exhibit width of between 100 and 150 μm for the 2-m focal-length far-field configuration, and approximately 1 mm for the 2.5-m focal-length, quasi-far-field configuration.

 

Fig. 1. Damage testing setup. The beam profile in the sample optical plane at the standard f = 2-m-mirror configuration with a 360-μm beam diameter is shown.

Download Full Size | PDF

All-reflective optics are used in the main beam pass to eliminate the influence of self-focusing in the air [14], which for this geometry, is found to occur at 3 J/cm² and 12 J/cm² for pulse durations of 0.6 ps and 10 ps, respectively. A small percentage of the s-polarized beam is picked off and passed through two leaky mirrors so that one portion is directed into the DALSA DS11-01M15-50, 12-bit digital CCD camera, and the remaining energy is directed into the energy meter system (Rm-3700/RjP-445, Laser Probe, Inc.). The CCD is placed in a plane equivalent with the sample optical plane. A commercially available LBA-500PC Laser Beam Profiler system (Spiricon, Inc.) is used to analyze the beam profile. The beam profile and energy are acquired for every shot, and the 90/10 knife-edge method [13] is used for the beam size and fluence calculations.

A typical MLD grating sample has a 1740 l/mm groove density designed to reflect s-polarized, 1053 nm light with 8-10 layers of alternating silica and hafnia. The grating profiles vary with duty cycles between 0.3 and 0.5. Pillars are approximately 350-400 nm tall with either a square or a trapezoidal shape. The sample was tested in a classical mount (beam propagation parallel to the grating vector), and the angle of incidence, theta, was varied (Fig. 2). Additional tests were conducted using conical mounting, where the angle of the incident beam with respect to the grating vector, phi, was varied. The effects of pulse length (700fs to 10 ps) and beam polarization were investigated. The beam polarization was varied between s-polarization and p-polarization in increments of 10 degrees.

 

Fig. 2. Figure showing sample orientation and the incident beam. The x-axis is parallel to the grating grooves. The beam is incident at θ_i, and the first and zero diffraction orders are given by θ_d and θ₀. For a classical mount, ϕ = 0 and the incident beam is parallel to the grating vector, k. When ϕ ≠ 0 the grating is in a conical mount.

Download Full Size | PDF

2.2 Experimental procedure

Two methods of damage testing are usually employed. For 1-on-1 testing, a series of sites is irradiated with a single pulse until the fluence causing damage is determined. A second procedure, referred to as n-on-1 testing, starts with an initial irradiation at a low fluence, 10%–20% below the estimated damage threshold. The fluence is incrementally increased until the site is visibly damaged. A site is determined to be damaged once it can be clearly spotted and differentiated from any nearby dust or debris by visual inspection. The damage site is also imaged onto a CCD camera using a 12× zoom lens to aide the examination by the naked eye.

2.3 Sample inspection

Test sites are analyzed using a Leica DMR polarizing microscope with interchangeable, nominally strain-free bright-field/dark-field objectives. The infinity-corrected optics include a Wollaston prism for differential-interference contrast microscopy, which is required for adequate imaging of the periodic features in the damage morphology. A Leica Wild M3Z stereo microscope is used to capture the macroscopic glint produced by the new grating-like morphology. The ripple formation is also examined using a Digital Instruments, Dimension 3100 atomic force microscope (AFM) with a new Ultrasharp Cantilever (NSC14/Cr-Au/S0, tip height 15-20 μm, tip radius < 50nm).

3. Results

3.1 Ripple characterization

Numerous MLD grating (1740-l/mm) samples were tested in the standard configuration (s-polarization, 61° angle of incidence) in order to determine the damage threshold fluence. A series of test sites produced an unexpected bright, color-shifting glint when they were examined macroscopically under white-light illumination. Altering the experimental geometry slightly, the 2-m focusing mirror was replaced with a 2.5-m focusing mirror to effectively place the sample in the quasi-far field, where the beam uniformity is less than optimal. The same glint was seen on the larger 1-cm test site, indicating that beam quality would not hinder the effect (Fig. 3). [The video shows how the new morphology diffracts light and changes colors as the illumination angle varies.] A unique grating-like damage morphology was found when those sites were examined microscopically.

 

Fig. 3. A test site irradiated with a large, low-quality beam profile still produced a brilliant glint. (Stereo microscope at 400× magnification.) [Media 1]

Download Full Size | PDF

These glints are not seen on 1-on-1 test sites, and they disappear on massively damaged n-on-1 test sites. The microscopic analysis confirmed that the source of the glints, a periodic microstructure, is only present on sites damaged with n-on-1 testing, suggesting that the formation of the overlaying ripples is a cumulative effect that cannot be produced with only a single shot. Sites irradiated three to five times at fluences below the damage threshold display the most uniform periodic structures with the brightest glints (Fig. 4). As the collateral damage on the test site increases, typically resulting from fluence increases as small as a few percent, the grating-like structure is slowly destroyed.

 

Fig. 4. The damage morphology from repeated irradiation (n-on-1 testing) was studied with both (a) a polarizing microscope (500× magnification) and (b) an atomic force microscope. The square in (a) indicated the region depicted in the AFM micrograph. The grating pillars extend from the top left to the bottom right corners. Four full “ripples” are formed as material is removed from the pillars to form ~200 nm troughs.

Download Full Size | PDF

The periodicity of these ripples is typically found to be between 2.0 and 2.4 μm, and occasionally it is possible to see harmonic ripples with a periodicity of ~ 1.1 μm. Atomic force microscopy provides further insight into the ripple structure (Fig. 4). Scanning a pristine area of the grating shows the original pillars to be 361 nm. The AFM confirms that the ripple morphology results from a periodic removal of significant amounts of material along the length of the pillars. Evaluating the pillar height shows that the troughs of the ripples are on the order of ~200 nm deep. In the plateau region of ripples, the pillars retain an average height of ~ 358 nm, indicating the little material has been removed. A measurement along the valleys between the pillars reveals that material is also being removed from the bottom of the valley, creating troughs as deep as 120 nm in some regions. This indicates that the periodic field distribution that is strong enough to remove material is not created just at the top of the pillars, but throughout the structure.

The orientation of the ripples varies across the test site. The “central” ripple is approximately perpendicular to the grating pillars, but successive ripples tilt at an increasing angle. Furthermore the ripples do not fan out with a linearly increasing angle. The angle of the ripples tends to increase for the first 10 to 20 ripples from the origin (defined as the ripple most normal to the grating pillars). Thereafter the ripple angle increases more slowly and seems to asymptote, which makes it feasible to measure relatively uniform ripple spacing at a specific tilt angle at the edges of the damage site (Fig. 5). The slope of the section of the linearly increasing angles has less to do with test angle conditions than where it is located on the damage site. Ripples at the bottom of a site (the area where the beam is first incident) fan out toward larger angles (4-7 degrees) than those toward the top of the site (2-4 degrees). The slope, or rate of change of the angle, also tends to be steeper for fans at the bottom of the site. The number of ripples in the center of the fan that follow the slope is larger toward the top of the site. The ripple orientation on the large site tested in the quasi-far field also fan out from the middle, but the area with nearly perpendicular ripples (to the grating pillars) is larger. These observations suggest that the angle of the incident rays affect the ripple orientation.

 

Fig. 5. Ripples at the middle of the site are designated to have position “0”, and they are typically oriented perpendicular to the grating pillar (angle = 0). The tilt angle of the next 10 to 20 subsequent ripples increases linearly and then starts to asymptote. The tilt of the ripples is larger at the bottom of the test site.

Download Full Size | PDF

3.2 Variation of test parameters

The periodic structures are easily formed and clearly identified on most test sites irradiated with 700-fs pulses. As the pulse length increases toward 10 ps, the structure becomes increasing difficult to identify. This suggests that either the mechanism for creating the ripples is more efficient at shorter pulse lengths or, more likely, thermal diffusion, which can still be somewhat effective at 10ps, masks this effect. This suggestion is supported by the fact that as cleaning techniques have improved and more photoresist, which strongly absorbs at 1053 nm, is removed, ripples are observed on 10-ps sites more frequently.

The angle of incidence, theta (in the plane of incidence) has been varied between 45° and 73° in increments of four degrees. As the incident angle decreases from 61°, the damage behavior “oscillates”, showing high damage thresholds and creating clear glints at 53° and 45° (Fig. 6). At the other, intermediate angles of 49° and 57°, total ablation occurs at relatively low fluences. For angles larger than 61°, the fluence at which ripples are created steadily increases. The other two characteristics studied as a function of angle, the ripple spacing and tilt, show a more regular trend. The extreme values of each data set are found between 65° and 69°, with the data fitting well to a third order polynomial. The MLD stack is designed for a maximum reflectivity at 67.5°, which implies that the stack design influences ripple characteristics.

 

Fig. 6. Both the ripple tilt and spacing vary with incident angle in a similar fashion. The surface fluence for ripple formation initially decreases with angle, but resonance-like behavior is seen at 53° and 45°.

Download Full Size | PDF

We describe a second set of experiment performed using conical mounts, whereby the vertical angle of incidence, theta, remains constant at 61°, while the horizontal angle of incidence, phi, is varied between -15° and +5°, as limited by the experimental set-up. Visually, the color of the glint changes as phi is varied, indicating that the ripple period is also changing. Microscopy shows that not only does the ripple period change, but it differs on both the right and left sides of the test site. When the sample is tilted in the positive phi direction (to the right when looking along the propagation direction), the ripple spacing on the left side of the site (where light impinged upon first) increases, while the ripple period on the farther side decreases. The two sets of ripples intersect abruptly at the center of the test site. The tilt angles for each set of ripples increases on the side closer to the incoming beam and decreases on the further side.

 

Fig. 7. When the incident angle in a conical mount is varied from (-15° to +5°), the ripple period is different on each half of the test site. The side impinged upon first always has a period greater than 2 μm and an increased tilt angle. The ripple period and tilt angle on the farther side always decrease from the value for normal incidence. The site shown has been tested at +5°. The ripple spacing and tilt angle are 2.3μm and 6.8° on the left side and 1.7 μm and 4.7° on the right side.

Download Full Size | PDF

Finally, there is no significant change in ripple characteristics such as period and tilt angle, when the sample is tested in the standard configuration and the beam polarization is changed. As the beam polarization is rotated between s- and p-polarization in 10 degree increments, the fluence for ripple formation (and damage) increases. For polarization directions greater than 50-60 degrees (with respect to s-polarization direction), the damage morphology becomes sensitive to the beam fluence and total ablation starts to destroy the ripples.

Several gratings with different MLD stack designs, which varied in the number and the thickness of layers or had a different sole thickness (remaining material of the top layer into which the grating is etched), have been tested, and the grating-like morphology is present on all of them. The grating profiles have also been closely examined, and no correlation is found between characteristics such as the pillar height, shape, or duty cycle and the ripple period, which varied between 2.0 μm and 2.4 μm. A bare fused-silica sample and a MLD stack witness piece from the same coating run that produced substrates for tested gratings have also been tested. Neither sample displays this new morphology, indicating that the grating structure (pillars) is essential to the production of the ripples. This fact is reinforced by the observation of similar ripples on gold-coated epoxy replica gratings, whose reflectivity is based on completely different principles compared to an MLD grating.

Unfortunately, the dependence of two important grating-design parameters—the line spacing and the laser wavelength—will be difficult to test. MLD samples with a line spacing other than 1740 l/mm are not readily available. The current laser system is limited to operation at 1053 nm. Changing the test wavelength requires not only system modification or a different system, but also redesigned MLD’s that act as high reflectors at the new test wavelength.

The fact that the ripple morphology is seen only on grating samples strongly suggests that grating structure couples two or more waves in the layers or layers directly below the pillars. Clearly two waves are interfering, producing periodic regions where the electric field is strong enough to remove material from the sample. The fact that the ripples do not have a uniform orientation with respect to a specific direction makes it difficult to predict the direction of the two interfering waves. The different ripple spacing that forms when the sample is tilted at some angle phi in the conical mount is similar to the de-phasing of two coupled modes from a resonance condition when the incident beam is oriented lightly off from normal [15]. Investigation of the physical mechanism responsible for the formation of periodic microstructure on MLD gratings is ongoing.

4. Summary

An unusual periodic, grating-like structure is observed when high-power, short-pulse lasers irradiate MLD gratings multiple times with fluences just below the damage threshold. A single irradiation does not produce this morphology or the accompanying bright, diffractive glint, so repeated irradiation is necessary for its formation. The ripple period is found to be between 2.0 and 2.4 μm, and occasionally a finer harmonic spacing with half the period has been observed. The ripple orientation varies across the site, increasing toward the edge. The orientation of the ripples fans out more slowly over the larger area of sites tested with a larger beam size and in the quasi-far field. However the tilt of the ripples at the edge of the large sites is the same as the tilt on smaller sites, suggesting that the angle of the incident rays affect the ripple orientation. No direct correlation has been found between grating characteristics (MLD design and grating profile) and the ripple spacing or tilt. Instead, ripple characteristics are found to be highly dependent upon the angle of incidence of the beam. Varying theta, the angle of incidence in the plane parallel to the grating vector, shows that ripples with a minimum spacing and a maximum tilt angle are formed at incident angles between 65° and 69°. Conical mounting with the angle phi varying between -15° and 5° produces ripples whose spacing and tilt angle increases on the half of the test site that was irradiated first. The ripple spacing and tilt on the further side decreases. Beam polarization does not affect ripple characteristics, only the ease and efficiency with which they are formed. This morphology is not seen on regular MLD stacks or other homogeneous samples without a grating profile.