Crystalline Flat Surface Recovered by High-Temperature Annealing after Laser Ablation

Abstract

Ultra-short laser pulses (1030 nm/230 fs) were used to laser ablate the surface of crystalline sapphire (Al₂O₃) at high intensity per pulse 20–200 TW/cm${}^2$/pulse. Laser-ablated patterns were annealed at a high temperature of 1500 ${}^{°}$C. Surface reconstruction took place, removing the ablation debris field at the edges of ablated pits in oxygen flow (O₂ flow). Partial reconstruction of ripples was also observed when multi-pulse ablated surfaces were annealed at high temperature in O₂ flow. Back-side ablation of a 0.5-mm-thick Al₂O₃ produced high surface roughness ∼$1\mathsf{\mu }$m which was reduced to ∼$0.2\mathsf{\mu }$m by high-temperature annealing at 1500 ${}^{°}$C for 2 h in O₂. Improvement of surface quality was due to restructuring of the crystalline surface and sublimation, while the defined 3D shape of a micro-lens was not altered after HTA (no thermal morphing).

1. Introduction: High Intensity > TW/cm${}^{\mathbf{2}}$ Crystal Processing

Matter at extreme conditions of pressures above 1 megabar (${10}^{11}$ Pa) and high temperatures as in the interior of planets—the warm dense matter (WDM)—is an active field of research [1]. High-pressure and high-temperature phases of materials can be created by ultra-fast thermal quenching of a small modified volume inside transparent material, e.g., sapphire [2,3]. Internally confined micro-explosions create WDM conditions similar to the centre of the Earth. For point-like energy deposition, it is informative to present Sedov’s solution, where the radius, R, of the shock wave is increasing in time t as [4,5]:

$$R\left(t\right)={\left(\frac{E_p}{K\left(\gamma \right){\rho }_0}\right)}^{1/(2+i)}{t}^{2/(2+i)},$$

(1)

where $E_p$ is the instantaneous (in space and time) release of energy, $K\left(\gamma \right)=(0.6$–$0.8)$ is the constant dependent on the adiabatic coefficient $\gamma =C_p/C_v$ ($\gamma =1.4$ for air), ${\rho }_0$ is the mass density in an unperturbed medium, and i is the dimensionality coefficient $i=3$ for spherical and $i=2$ for cylindrical explosion; here, $C_{p,v}$ is the heat capacities at constant pressure and volume, respectively. Such a shock wave launched from the focal volume creates highly dynamic conditions of high pressures and temperatures, which lead to formation of new materials [2,3]. In addition, metastable amorphous phases can be produced as shown for sapphire [6] due to ultra-fast thermal quenching of small volumes modified by fs-laser pulse-triggered explosions. Spatially confined, energy deposition of ∼40 fs onto a water surface triggers shock waves inside the water and the surrounding air which causes emissions of X-rays and THz [7]. This can be a single optical cycle (∼1–3 ps) due to the point-like energy deposition in time and space.

The micro-explosion hydrodynamics was shown to follow the established and tested macroscopic versions of strong explosions [8]. Evolution of the micro-explosion is characterized by the independent parameters: energy E (or pressure P), density $\rho$, and radius r, while the two other parameters required for the full description of the hydrodynamic processes, the velocity and time, can be expressed as $v=\sqrt{P/\rho }\equiv \sqrt{E/\left(r^3\rho \right)}$ and $t=r^{5/2}\sqrt{\rho /E}$, respectively. It is shown that micro-explosions triggered by tightly focused fs-laser pulses are self-similar with macro-explosions made by a ${10}^{21}$ large energy deposition when the time t and space r are ${10}^7$ times larger [8]. All hydrodynamic phenomena with the same initial pressure and density but differing in space and time scales are described by the same hydrodynamic equations. In astrophysics, these same equations are used to describe the propagation of shock waves in self-gravitating perfect gases for isothermal flows [9], studying the dynamics of the envelope in supernova remnants [10,11], and the expansion of supernova shells [12].

Strong structural modifications induced by high-intensity ultra-short laser pulses could be modified by a post-processing steps of high temperature annealing (HTA) as recently demonstrated for micro-optical elements laser ablated out of crystalline sapphire Al₂O₃ [13]. HTA is a commonly used process for changing the surface morphology of different materials [14,15,16,17]. It has been used to make the particle size of high-temperature alloy thin films smaller [14], to create ceramic devices from polymers [17], and to smooth out germanium layers on silicon substrates [15] and terraces of chemically etched SrTiO${}_3$ [16]. The roughness evolution of rapidly annealed surfaces is well understood; as the annealing time increases, the roughness decreases linearly [18]. Germanium, silicon, and some alloys have lower melting temperatures, making them easier to control. Sapphire, on the other hand, has a high melting temperature at 2053 ${}^{°}$C requiring more specialised equipment to reach these temperatures. In this study, it is shown that using temperatures between 1400 ${}^{°}$C to 1600 ${}^{°}$C and different gaseous environments, it is possible to alter the surface morphology of crystalline sapphire. In glasses, the softening point, at which a material does not become liquid as it would at the melting point but becomes malleable [19] is used; however, in crystals, only thermally activated surface diffusion on an atomic scale as well as sublimation is occurring. Sublimation is the process by which a solid material turns to gas without passing through the liquid phase. In crystals, sublimation causes them to create step-like structures, [20] as has been shown with GaN(0001) [21], GaAs(001) [22,23], and Si(111) [24,25]. For each of these materials, both Monte Carlo simulation and experimental results showed agreement on the appearance of step-like structures and in the case of Si(111), it was possible to nano-texture the surface with the sublimation method. The theory here is to balance the removal of material through the sublimation process with the growth of the crystalline structure to create the step surface. Sapphire sublimation has been used before on sapphire spheres 1 mm in diameter with much longer annealing times of 7.5 h and 1.5 h and at temperatures of 1500 ${}^{°}$C and 1850 ${}^{°}$C, respectively, in H₂ gas [26]. The vaporisation temperature for Al₂O₃ at a pressure of 1 atmosphere ranges from 2114.35 ${}^{°}$C to 2249.35 ${}^{°}$C [27]. The advantage of HTA of crystals is the preservation of large micro-scale structures and patterns, while reconstruction and smoothing of laser-treated regions takes place at nanoscale.

Here, we show the results of laser ablation of sapphire (Al₂O₃) and its high-temperature annealing for surface roughness reduction and reconstruction of surfaces with ablated ripples. The conditions to recover a crystalline smooth surface up to the rim of the ablated pits were found, which can be useful for applications to 2D materials. It is shown that back-side ablation of Al₂O₃ in air which produced roughness of ∼$1\mathsf{\mu }$m was reduced to <200 nm after HTA at $1500{}^{°}$C for 2 h.

2. Samples and Methods

2.1. Sapphire

The PHAROS (Light Conversion Ltd., Vilnius, Lithuania) laser system was used to irradiate the surface of the Al₂O₃ on the front and the back-side. The Al₂O₃ samples have ∼0.5 mm thicknesses. Al₂O₃ is a hard and inert material that requires ∼10 TW/cm${}^2$ intensities for ablation at near-IR 800–1000 nm wavelengths where multi-photon absorption is required to seed an efficient avalanche generation of free electrons for a wide bandgap ∼8–9 eV dielectric [28]. Among its virtues is high transparency in UV-to-IR spectral range (200–4000 nm) and high heat conductivity. This makes sapphire optics a promising target for industrial and space applications.

2.2. Laser Ablation Conditions

The PHAROS (Light Conversion Ltd.) laser system with $\lambda =1030$ nm wavelength, $t_p=230$ fs pulse duration, focusing with objectives lenses with numerical apertures of $NA=0.26$ and $NA=0.14$, resulted in focal spots calculated by Airy disk formula $2r=1.22\lambda /NA$ of ≈$4.8\mathsf{\mu }$m and ≈$8.9\mathsf{\mu }$m, respectively. The Rayleigh length is $z_R=\pi r^2/\lambda =35.62\mathsf{\mu }$m and $122.6\mathsf{\mu }$m, which is half of the depth of focus (in air). For example, for the pulse energy of $E_p=1\mathsf{\mu }$J (on the sample), fluence was $F_p=5.45$ J/cm${}^2$ and average intensity $I_p=24$ TW/cm${}^2$. Laser repetition rate was $f_r=200$ kHz and linear density of pulses was varied for larger or smaller overlap (typically ${10}^2$–${10}^3$ pulses/mm), which affected the material removal rate. The laser writing velocity was up to 15 mm/s for a position sensitive output, which was important for the high-precision ablation of micro-lenses in a multi-pass scanning mode. In general, this is a 2D/3D sculpturing method that can reach sub-micron precision when it is performed in the light-induced back-side wet etching (LIBWE) mode [29]. Specific details of laser patterning, if different from those listed above, are shown where applicable.

2.3. High-Temperature Annealing

High-temperature annealing (HTA) was carried out using a tube furnace (Model RHTC, Nabeterhm, Germany). The recipe starts by ramping the temperature up to 1500 ${}^{°}$C at 150 ${}^{°}$C an hour. The next step is holding the temperature at 1500 ${}^{°}$C for 2 hours and afterwards ramping down to 1400 ${}^{°}$C at 200 ${}^{°}$C per hour. The final step is cooling down to 30 ${}^{°}$C at 100 ${}^{°}$C for an hour to finish the procedure. The extra step down to 1400 ${}^{°}$C before cooling to 30 ${}^{°}$C is used as a test for the limits of the tube furnace. This may be important if rapid quenching is required for further experimentation on the lenses. Currently, rates of up to 500 ${}^{°}$C per hour have been tested.

3. Results and Discussion

3.1. Roughness Reduction of Ablated Sapphire Surfaces

Front-side ablation was used with laser pulses traveling from air-into-Al₂O₃; back-side ablation occurs when the focus is placed onto the exit plane (Al₂O₃-to-air). This is usually possible with samples thinner than 0.5 mm and at less tight focusing with $NA<0.4$; see Section 3.3 for the back-side ablation and HTA. It took ∼30 min to ablate the profile of a 1-mm-diameter lens (Figure 1a) with a height of $h=15\mathsf{\mu }$m. The surface roughness of the aspherical ablated surface was ∼$200\pm 50$ nm and it was reduced to $70\pm 30$ nm after 2 h annealing (Figure 1c). However, light scattering was still strong and the surface remained visibly white. Initial roughness was also dependent on the focusing conditions and pulse overlap during ablation. Other smoothing strategies need to be tested, e.g., ablation with more focused pulses. When ablation took place at the very top of the lens profile (surface of sapphire), there was a strong change in ablation conditions, which resulted in a non-ablated plateau, which is ≈$150\mathsf{\mu }$m in diameter (Figure 1). This is a known issue and can be circumvented by placing a micro-layer of a matching refractive index material, e.g., a silica film, that allows to ablate sapphire [30]; the added layer is later washed out in a solvent/acid.

The effect of HTA was examined through the ablation of crystalline flat c-plane Al₂O₃, which led to better understanding of surface changes. Since the maximum temperature used was $1500{}^{°}$C, which is much lower than the melting point of Al₂O₃∼$2050{}^{°}$C, surface restructuring by surface tension-driven flow of the molten phase was absent and other contributions from surface diffusion at atomic level, coarsening of debris nanoparticles, and sublimation took place. Sapphire has a melting point of 2053 ${}^{°}$C and boiling temperature of 2980 ${}^{°}$C. It has recently been shown that lower temperatures with larger times can also yield surface re-flow, which can be understood from length of diffusion scaling $\propto \sqrt{D_s\times t}$, where $D_s$ [cm${}^2$/s] is the surface diffusion coefficient and t is time. Temperatures of 1100 ${}^{°}$C and 1200 ${}^{°}$C at 3, 6, 9, 18, and 30 h (in air ambience) have been tested and showed smoothing of c-plane sapphire terraces in step configurations [31]. Laser-modified regions under high-intensity irradiation in air might also have surface formation of ceramic ALON phase (AlN)${}_x·$(Al₂O₃)${}_{1-x}$ with x = 0.30–0.37 or AlN [32]. When sapphire was placed in N₂ flow, the crystalline flat c-plane surface of Al₂O₃ changed to a nanotextured surface as shown by the SEM images in Figure A1a, and the entire piece of Al₂O₃ looks frosted and translucent as shown by the optical image in Figure A1b.

Generation of defects, vacancies, and interstitials under high intensity fs-pulses is taking place, e.g., F${}^{+}$ color center of the oxygen vacancy with a trapped electron created by fs-laser machining of sapphire is fully annealed at 1100${}^{°}$ [33]. Frenkel pair defects interstitial—vacancy observed earlier by MeV-electron irradiated high-pressure cubic-BN was induced by fs-laser pulses [34] at similar conditions as used in this study. Chemical bond-breaking by high-intensity laser pulses can create metal-enriched regions on the surface of oxides and nitrides due to formation of volatile O₂ and N₂. Such laser-ablated surfaces of oxides rich with defects and debris and depleted of oxygen are expected to be restructured by HTA. We chose O₂ flow in order to facilitate surface migration of Al atoms to form terraces of Al₂O₃ under oxygen supply. This is one probable contribution to the formation of smooth crystalline surface.

3.2. Recovery of Crystalline Smooth Surface on Ablated Sapphire

When focusing using a less tight objective lens $NA<0.14$ and carrying out single-pulse ablation with HTA in N₂ flow, there were more debris left after annealing as compared with samples annealed in O₂ (Figure 2). Nitrogen-rich atmosphere was not favorable to recovery of crystalline surface as compared with oxygen flow (see discussion above). Also, a pattern of single pulse ablation showed non-periodic structure with close-to-wavelength feature length. In addition, there was no typical ablation crater, rather a nano-textured surface resembling thermally quenched molten phase. The pulse fluence used was $F_p\approx 10$ J/cm${}^2$ or average intensity of $I_p\approx 43$ TW/cm${}^2$, which is smaller by more than an order of magnitude than the self-clamping intensity of fs-pulse ∼${10}^{14}$ W/cm${}^2$ when multi-filaments are formed in air [35], and could cause a random pattern of dot-like modifications similar to those in the SEM image (Figure 2). Formation and quenching of a molten phase is more probable. Under such high intensity 43 TW/cm${}^2$/pulse exceeding the ablation threshold of sapphire by ∼5 times, formation of a highly excited plasma skin layer of tens of nanometers was expected during the laser pulse [36]. Light-matter interaction takes place as if there is a nano-film of different optical permittivity $\epsilon \equiv \tilde{n}$ placed on dielectric sapphire; $\tilde{n}=n+i\kappa$ is the complex refractive index. Such transient material state described by changing permittivity $\epsilon \left(t\right)$ is called Die-Met (from dielectric and metal-like plasma) [36]. It has strong absorption due to reduced n and increased $\kappa$ due to electron generation and is also referred to as epsilon-near-zero (ENZ) state $\epsilon ={(n-\kappa )}^2+i2n\kappa$ [37].

The ablation of thin films, which proceeds after the end of the laser pulse, follows an imprint of an fs-laser pulse intensity. Such modulation has $(\lambda /n)/2$ period and, under multi-pulse exposure, reveals itself as ripples [38]. After HTA, a more orderly ripple pattern emerged with period smaller than $\mathsf{\Lambda }<\lambda$. The period of ripple-like pattern which evolved after HTA (Figure 2) resembles ripples ablated under multi-pulse exposure. We put forward here a conjecture (for future validation) that light intensity modulation ∼$\lambda /2$ occurred due to the presence of highly excited plasma nano-layer. Such modification of sub-surface region at the sapphire-plasma interface is hidden from SEM observation in Figure 2, which shows surface of a quenched molten phase. HTA was responsible for restructuring the surface and the imprinted modification was revealed. Figure 3 shows closeup SEM views of the ripples with evident re-crystallisation following c-plane of Al₂O₃ (hexagonal pits). Even though HTA temperature protocol was the same, the surface quality of HTA-treated ablated surface had more debris as compared with that made under O₂ flow (Figure 4).

Figure 4 shows scanning electron microscopy (SEM) images of single and five-pulse ablated sites after HTA in O₂ flow. A single pulse of $E_p=11.25\mathsf{\mu }$J which corresponds to a fluence of $F_p=61.32$ J/cm${}^2$ and an average intensity of $I_p=266$ TW/cm${}^2$ ablated a hole of 300 nm depth and a re-healed area of 4.4 $\mathsf{\mu }$m diameter, which almost perfectly matches the focal diameter of 4.8 $\mathsf{\mu }$m. After annealing, a perfectly smooth surface was observed near the rim of the ablated pit. When the number of pulses was two or more, formation of ripples [38] was observed (see structure of period $\mathsf{\Lambda }\approx 1\mathsf{\mu }$m at the bottom (Figure 4b)). Recovering a smooth surface during HTA in O₂ flow is a promising outcome for opto-electronic applications of 2D materials.

3.3. HTA Treatment of Back-Side Ablated Al₂O₃

Since front-surface ablation caused extensive deposition of debris, a back-side ablation was used to mitigate this issue. No liquid immersion was used as in a previous study [13]. Ablation by circular scan started from the largest diameter and was closing towards the center in concentric circles following trajectory ablating an aspherical micro-lens of $125\mathsf{\mu }$m radius. After all-layer ablation, the depth of focus was changed and the next layer (into the sample from the back-side) was ablated. This protocol was robust against formation of debris and free from defects such as random non-ablated regions. The results are shown in Figure 5. Bruker ContourGT Optical Profiler was used to map surface geometry and surface roughness. The ablation steps according to the protocol of focal position are clearly discernable together with the roughness. The height step of $\Delta z=0.8\mathsf{\mu }$m was used for most reliable ablation of asphetric dome; the step inside Al₂O₃ becomes $n\times \Delta z$ due to refractive index $n\approx 1.7$ at the wavelength of laser printing. The fluence per pulse $F_p=4.7$ J/cm${}^2$ was approximately twice the threshold of sapphire ablation for the focal spot of $r=0.61\lambda /NA=1.4\mathsf{\mu }$m radius. Up to $1\mathsf{\mu }$m roughness (min-max) was observed on as ablated surfaces after ultrasonic washing. Samples were Cr-coated for SEM observation which corroborated reduction of roughness after HTA. Approximately five-times reduction of surface roughness down to ∼200 nm was observed for the laser ablation at $NA=0.45$ focusing. Fabrication of the micro-lens shown in Figure 5 took ∼15 min at laser repetition rate of 200 kHz and linear scan speed of ∼15 mm/s (pulse-to-pulse overlap was $\Delta x=1.3\mathsf{\mu }$m as well as the line-to-line separation and was comparable with the radius r of the focal spot). The designed aspherical lens has a height profile:

$$Z\left(s\right)=\frac{C{s}^2}{1+\sqrt{1-1(1+k)C^2{s}^2}}+A_4{s}^4+A_6{s}^6+A_8{s}^8+A_{10}{s}^{10},$$

(2)

where the coefficients $C=1/0.15$, $k=0$, $A_4=0.0093$, $A_6=-0.005$, $A_8=-0.0056$, $A_{10}=0.0005$, and s is the radial distance from the optical axis.

HTA made the surface smoother and the steps in z-direction less steep. HTA is responsible for the evolution of the overall aspheric shape becoming closer to the designed shape. There was almost no change in the base diameter and height of the lens (Figure 5a,b). Ablation at a fluence only twice that of the ablation threshold of the crystalline host material contributed to a good definition of the shape of the micro-lens. In glasses, melting and surface tension-induced reshaping is taking place even without HTA [39].

The main cause of the reduction in roughness was surface restructuring during the HTA process, as illustrated in Figure 6. Sublimation could be contributing to the discernible difference in the height change of the optical profiles (Figure 5). The steps along the ablation scans become less steep. This study shows that smoother surfaces of crystalline lenses can be achieved using HTA in O₂. Smaller height steps will be tested to achieve tens-of-nm surface roughness required for optical applications.

4. Conclusions and Outlook

Sapphire ablation with different numerical apertures ($NA=0.26$ and 0.14) and similar pulse energies ($10\mathsf{\mu }$J and $6.25\mathsf{\mu }$J) are compared. SEM images showed the formation of ripples with period close to the wavelength of the laser irradiation for the $NA=0.26$ and interestingly, it was half of the wavelength of the laser PHAROS (Light Conversion Ltd.) for $NA=0.14$. After the samples had undergone high-temperature annealing at 1500 ${}^{°}$C for 2 to 4 h in either an O₂ or N₂ environment, these ripples underwent restructuring and crystalline step-like structures become apparent. It was also shown that annealing in an O₂ environment reduced the amount of debris and the surface became smooth. In the case of HTA in N₂, however, pieces of debris were still present and their coarsening took place. Detailed X-ray Photoelectron Spectroscopy (XPS) experiments are planned for detailed understanding of material modification on the surface during HTA in O₂ and N₂ flow. Apparently, the O₂ flow was an important factor for restoration of a smooth c-plane surface by HTA.

We show that the restructuring of laser-ablated crystalline surfaces can be applied to the fabrication of micro-optical elements in reducing the surface roughness and allowing for tighter control over the surface profile while maintaining the precision of a shape definition. This is different from glasses [40], where HTA causes a surface tension-driven reflow and formation of spherical surfaces which might be different from the required shape for a defined optical function. Back-side ablation in air was used to define an aspherical micro-lens profile, which was not altered by HTA while surface roughness was markedly reduced. The true extent of the control over surface roughness is yet to be fully explored using this technique. The optimal time of annealing and the optimal temperature to achieve the most restructuring without sacrificing the surface profile will be determined in future studies. Sublimation via solid-to-gas occurring along with solid-fusion-vaporisation stages during HTA needs better understanding and can be different in other crystals.