1. Introduction

Diamond comprises the material of choice for Raman lasers due to its large Raman frequency shift (∼1330 cm⁻¹), large Raman gain (∼10 cm/GW at ∼1 μm), and giant transparency window, extending from the DUV to the THz [1]. Its excellent thermal properties along with negligible birefringence make it an ideal material for high-power Raman lasing with greatly reduced thermal effects [1,2]. Continuous progress in the growth of diamond crystals has improved the Raman laser output performances reported in past years [3,4], with high slope efficiencies up to 84% demonstrated [5]. Its large transparency range enabled the development of bulk Raman lasers across the UV [6], visible [7,8], near-IR [9], and even mid-infrared [10] regions of the spectrum. In addition, Raman conversion in diamond has been already demonstrated at power levels approaching the kilowatt-level, with high output beam quality [11].

One drawback of diamond is its significant Fresnel reflection losses of 29%, caused by its high refractive index (n = 2.4 in the near-IR). For efficient laser operation, high quality anti-reflection (AR) coatings are mandatory [5]. Manufacturing high power AR coatings in diamond surfaces with ultra-wide bandwidth (to cover multiple Stokes orders) is challenging, being damage threshold and adherence to the surface the main bottlenecks. One possible solution is to build diamond meta-surfaces, increasing dramatically the damage threshold without the need of additional materials. This approach has become of common use in CO₂ laser windows in the mid-IR [12], where can theoretically achieve almost 99.9% average transmission over large bandwidths with 10 × higher laser induced damage threshold. However, methods for forming such sub-micrometer structures in diamond are very limited. In fact, traditional ablation methods cannot achieve the necessary resolution for operation in the visible or near-IR.

With the discovery of laser induced periodic surface structures (LIPSS), feature sizes well below the diffraction limit can be also obtained in a simple manner [13]. LIPSS can be formed on surface of nearly all kinds of materials, including metals, semiconductors, and dielectrics. Many articles have published the results of LIPSS formed by different continuous and pulsed lasers in diamond films [14–16]. Surface processing has been typically demonstrated with near-infrared lasers, producing periodic structures below the excitation wavelength with high aspect ratios by operating close to the ablation threshold and at high repetition rates [17]. There is, however, a scarcity of detailed studies of the nanoscale morphology (including roughness) of the structures and their photonic properties.

In this work, we explore and optimize the conditions for producing high fidelity, low roughness diamond nano-structures using UV ultrafast pulses instead, and characterize their optical properties. We show that the generated nano-structures hold potential as moth-eye anti-reflection coatings for Raman lasers pumped by commercial Nd- or Yb- doped lasers operating at ~1 μm and with Stokes orders located in the range ~1.2-1.5 μm. Grating-like structures were fabricated using the LIPSS mechanism with aspect ratios close to 1:1, high reproducibility and accuracy, and very small surface roughness (<1.4 nm). Combining numerical modeling and AFM/SEM analysis, it was found that this method could be a simple alternative for the fabrication of moth-eye anti-reflection coatings in diamond.

2. Experiments

The creation of surface ripples in dielectric materials under laser irradiation is a well-known method to nanotexture a surface, where the ripple orientation can be finely controlled with the polarization direction. Typically, the anisotropy of the produced structures is presumably due to a heat conduction alteration during fabrication [18], which affects the laser writing itself and, in the end, the performance of the produced optical element. Therefore, fine control over the laser parameters as well as the use high quality diamond substrates are key in order to fabricate smooth ‘optical grade’ nanostructures.

The experiments were carried out in open air atmosphere with a Ti:Sapphire laser system (Coherent Inc. Libra) consisting of a mode-locked oscillator and a regenerative amplifier producing 130 fs pulses at a 1 kHz repetition rate with a central wavelength of 800 nm. The output was subsequently frequency doubled to 400 nm as depicted in Fig. 1. The pulse energy was adjusted with a two-step setup: a variable attenuator formed by a half-wave plate and a low dispersion polarizer and neutral density reflective filters. The ~8 mm diameter laser beam was focused on the samples using a 10 × microscope objective with a NA of 0.16 to a 1.5 μm spot at FWHM. It is also important to remark that the diameter of a micro-machined structures width depends on the material’s ablation threshold and the non-linear response, which typically yields an interaction area smaller than the beam waist. A three-dimensional translational stage was used to move the sample under the laser beam with variable velocity. The laser parameters (spot size, fluence, scan velocity) were adjusted to produce highly homogeneous low-spatial-frequency LIPSS (LSFL) nanopatterns. Finally a CCD camera was used for the online monitoring of the structuring process.

 

Fig. 1 Schematic layout of the femtosecond laser machining setup for processing diamond at 400 nm. HWP: Half-wave plate, PBS: Polarizing beam splitter, ND: Neutral density filter.

Download Full Size | PDF

Single crystal Type IIa (Electronic grade, Element6) with [100] orientation diamond samples were used in the study. The surfaces were cleaned using subsequent baths of boiling aqua regia and chromic acid to ensure clean surface before laser treatment. Atomic force microscope measurements yielded an average surface roughness of around 1 nm prior to experiments.

2.1 Generation of ultra-low roughness LSFL nanopatterns

Generally, the spatial period Λ of LIPSS generated structures is known to depend on the laser wavelength λ, the polarization of the laser electric field, and the number of impinging laser pulses. The surface ripple period can be slightly less than λ (low spatial frequency LIPSS, LSFL), succeeding the in-plane weak interference of the incident transverse laser pulse and almost transverse surface polaritons. Alternatively, is also possible to generate surface ripples with much smaller periods by interference of quasi-monochromatic surface plasmons with the incident wave or among themselves, producing ripples with periods much lower than λ (high spatial frequency LIPSS, HSFL) [15]. In this regime, a typical periodicity of approximately ~λ/4 is found for diamond surfaces [16]. In our experiments, we selected a laser wavelength of λ = 400 nm in order to produce structures in the 100-400 nm range LSFL-type ripples, useful for optical applications from the near-IR and the visible spectral ranges.

The laser fluence and scanning speed was adjusted to produce a range of diamond structures with varied characteristics. The ablation threshold was measured to be around ~2 J/cm², consistent with previous measurements at 400 nm employing ~100 fs pulses [19]. Figure 2(a) shows in red color the parameter space where high optical quality LSFL type LIPSS occur, with a measured period of ~400 nm. Fluences in the range of 2-6 J/cm², and scan speeds between 100 – 300 μm/s produced the desired LSFL structures with high fidelity as shown in the FEG-SEM microscope images in Fig. 2(b), whereas for lower scanning speeds of 100 μm/s, HSFL structures appear as well as morphological irregularities that decrease the optical quality of the structures. For higher fluences beyond 6 J/cm², strong ablation occurs and HSFL structures dominate the irradiated areas. The results show that in order to produce high quality LSFL structures, the total number of photons impinging a certain location in the diamond surface remains fairly constant in the range of 2 - 6 J/cm². For example, machining at 4 J/cm² and 200 μm/s, the same surface location is exposed to a sequence of approximately ~10 pulses. For the case of 6 J/cm² and 300 μm/s, the number of pulses per location is reduced to 6.6, while the fluence was increased by a factor of 1/0.66. Analogously, for the case of 3 J/cm² and 150 μm/s, we find that the photon dose per location remains constant at around ~8 × 10¹⁹ photons/cm².

 

Fig. 2 (a) LSFL structures as a function of scanning speed and laser fluence. (b) Morphology of the nano-ripples for different scanning speeds at a fixed fluence. Scale bar = 1 μm.

Download Full Size | PDF

A scanning atomic force (AFM) microscope “JPK NanoWizard” produced by JPK Instruments was used for the surface roughness investigation. The AFM nano-tip used was a TAP300-G rotated monolithic silicon probe with a tip radius <10 nm, with half cone angle of 25°. The results of the measurements are shown in Fig. 3. Overall, the average surface roughness (R_a) was around 1 nm for the whole processed area. Here we considered R_a as the average of the individual heights (asperities) and depths from the arithmetic mean elevation of the profile. The worst values were found to be around the ‘crests’, where the typical R_a was around 1.4 nm. Figure 3(a) shows the average of ten periodic profiles in red, overlapped with the local standard deviation in gray. Figure 3(b) depicts the results of the AFM measurements over 5 periods (2 μm in x-axis) and Fig. 3(c) shows a lineout of the structures along the x-axis. The maximum slope measured in the diamond surface is about 35°.

 

Fig. 3 Structures generated at 4 J/cm² and 200 μm/s. (a) Average and standard deviation of structure profile. (b) AFM scan of the irradiated area. (c) Line-out of the surface structures over 2 μm.

Download Full Size | PDF

2.2 Large area nanostructuring using LSFL LIPSS

In order to produce useful anti-reflection coatings, spatially coherent nano-structures are necessary. This is a potential limitation of using LIPSS since the irradiated areas are in the order of tens to hundreds of μm². In principle, this can be overcome by overlapping multiple adjacent spots, as it was recently demonstrated in silicon [20], where ripples showed spatial coherence from well-separated spots arising from elastic or surface diffusion properties, as well as a long and far-reaching perturbation of the lattice structure. That finding opened the doorway to structuring large surface areas without the need of strict phase control of the laser. Developing this technique in other media is of major interest as a simple alternative to the fabrication of photonic structures, although to the best of our knowledge has not been demonstrated yet.

Here, we demonstrate the spatial coherence characteristics of the diamond grooves produced in different adjacent scans in a large area by slightly overlapping each laser pass as shown in Fig. 4(a). We rotated the laser beam polarization by ~10 degrees to study in detail the continuity of the grooves between independent passes. As shown in Fig. 4(b), the continuity of the LIPSS patterns extends indefinitely along multiple independent passes.

 

Fig. 4 (a) Schematic of the multi-pass direct-laser nanostructuring over large areas. (b) FEG-SEM image of the produced highly ordered nano-ripples.

Download Full Size | PDF

3. Simulations

In this section we study the photonic properties of the generated structures and their potential as anti-reflection coatings for Raman lasers. Since the produced LIPSS in the diamond surface present features much smaller than the wavelength of infrared light, they exhibit similar behavior as meta-surfaces, with a tailored effective index that will affect the transmission and reflection of light by means of the well known “moth eye” principle [21]. Typically, for cylindrical structures, in order to get a minimum in the reflectivity of the surface, the relation between the structure period Λ and the light wavelength λ should be around Λ/ λ ≈0.4. In this way, for example, an optimized antireflection structure for Stokes orders in diamond from 1.26 to 1.49 μm (pumped by commercial Nd- and Yb- doped lasers at 1.06 μm), would require structures with periods of approximately 400 - 500 nm. We explore the potential of the large area LIPSS generated structures as meta-material for anti-reflection coatings at near-IR wavelengths, although the concept can be extended to any wavelength of interest.

The response of light to the LIPSS morphology and structure was studied by performing simulations using Lumerical© FDTD Solutions v8.17, implementing the diamond material using Sellmeier equations. The LIPSS structure was transferred to the simulator directly from the AFM measurements.

Figure 5(a) depicts the simulated diamond structure, with 400 nm periodicity and 120 nm in depth, resembling the fabricated structures in section 2.2. Figure 5(b) shows the results of the simulations carried out for incoming light polarized in the same direction of the LIPSS pattern and perpendicular to it, and the Fresnel losses of an untreated surface for comparison. For both polarizations, the treated surface enhances the transmission of light into the substrate for wavelengths above 1.2 μm. We found that the maximum transmission of close to 100% occurs at around 1.25 μm. The reason for the drop in transmission below 1.2 μm comes from the fact that the plot shows only the zero-order transmission efficiency. For shorter wavelengths below 1.2 μm, there are considerable contributions of higher diffraction orders.

 

Fig. 5 (a) Schematic of the simulated structure fitted from AFM measurements. (b) Computed dependence of the reflectance on wavelength and polarization of LIPSS surface and untreated diamond surface.

Download Full Size | PDF

4. Conclusions

We studied and optimized the conditions for producing high fidelity, low roughness (<1.4 nm) diamond nano-structures over large areas. We also found that this method could be a simple alternative for the fabrication of moth-eye type antireflection coatings in diamond, useful for applications in high power diamond Raman laser development in the near-IR.