Enhanced etch rate of deep-UV laser induced etching of diamond in low pressure conditions

Diamond is of intense interest for a wide range of applications due to its many extreme material properties. Fields such as quantum engineering,¹ electronics,² MEMS,³ and electrochemistry⁴ are prominent areas in which interest in utilizing diamond-based technologies is growing. Diamond's extreme hardness and chemical resistance, however, also hinder efforts to fabricate complex devices, rendering many standard surface processing techniques developed for other semiconductor platforms inapplicable. Given the promise of diamond devices, it is valuable to develop new techniques for structuring and manipulating diamond surfaces. Ultraviolet (UV) laser induced etching (LIE) of diamond^5–7 is a promising optical technique for the direct-write, high resolution structuring of diamond surfaces.

UV LIE is an unusual process in the context of optical or laser processing of materials in that material removal is via a cold photoejection process, in contrast to the thermal or ablative processes that prevail for other materials. It occurs for laser pulses below the ablation threshold and at a rate quadratic with the pulse laser intensity,⁶ providing a technique with outstanding depth control, even to effective depths corresponding to the partial removal of a single atomic layer.⁸ The etched surfaces are oxygen terminated and largely graphite-free.⁹ Sustained UV LIE to depths of tens to hundreds of nanometers produces nanostructured patterns on the surface, with the pattern type and orientation dependent on the angle between the laser polarization and the crystal lattice orientation.¹⁰ On the {100} surface, for example, regular structures composed of {111} microfacets are formed that vary from extended ridges or grid-like patterns depending upon the polarization direction.

To date, UV LIE has been characterized for a variety of conditions, including power, wavelength, and pulse duration.^5–7 It has been found that the process is suppressed under vacuum,⁵ indicating that the participation of oxygen surface species is integral to the etching mechanism. However, the nature of the oxygen–surface interaction during etching and its role in carbon photoejection has yet to be clarified.

In this paper, we have investigated the effects of pressure and gas composition on the etch rate and nanopattern morphology. An expression for the etch rate as a function of O₂ partial pressure is derived, which is compared with measurements of the etch rate over the pressure range $10−5$ to 760 Torr for atmospheric air, dry air, and pure oxygen. Fair agreement with the model is achieved at low partial pressures of oxygen, but diverges at higher pressure depending on gas composition. In particular, when using pure O₂ at intermediate pressures, we observe a 10-fold increase in the etch rate over etching in ambient conditions accompanied by marked changes in the etched surface pattern. The results are analyzed with reference to the role of water vapor on photoejection probability and surface stabilization.

The model for the etch rate developed here is based on the prevailing assumption that carbon is ejected from the surface in an oxidized form—the etch rate then depends on the amount of oxygen chemisorbed on the surface and the probability that a laser pulse leads to photoejection of a carbon–oxygen species. Since the observed per pulse etch rates are typically much less than an atomic layer,^5,8,11 it can be assumed that the etch rate is proportional to the number density of oxygen surface species N prior to each pulse. To obtain an expression for the steady-state etch rate as a function of pressure, we find the equilibrium condition that the number of ejected atoms per pulse equates to the number of adsorbed O atoms between pulses.

From the kinetic theory of gases, the average O₂ arrival rate per unit area is $Zc=P2πmkT$⁠, where m is the particle mass and P is the O₂ partial pressure. The probability of an incident O₂ molecule resulting in dissociative adsorption is $S(θ)=Sc(1−θ)2$⁠, where S_c is the sticking coefficient and $θ=NNsat$ is the oxygen coverage with $Nsat=1.0×1019$ atoms/m² being the saturated number density of O sites in ambient conditions (1 atm, 20 °C, and ∼50% relative humidity).^12–14 The adsorption rate is $dNdt=2Zc·S(θ)$ so that

and the surface coverage after time t, θ_t, given an initial coverage θ₀, is given by

Each laser pulse ejects carbon atoms with a probability dependent on the particular environment of C–O groups on the surface. Defining the per-atom etch probability R₀, i.e., the probability that a given surface oxygen atom desorbs as a carbon containing molecule, each pulse reduces the oxygen coverage from θ_T, the coverage at the end of the interpulse period T, to $(1−R0)θT$⁠. R₀ is a function of laser parameters, surface temperature, and changes in the proportions of surface moieties, and in the following measurements, it is determined from the etch rate in atmosphere, where $θT=1$ by definition.

In the many pulse regimes, equilibrium is attained in which the number of oxygen species removed per pulse equals the number of species adsorbed during the interpulse period. This results in the condition,

Substitution into Eq. (2) yields the etch rate as a function of the O₂ partial pressure for a given R₀.

Etch rate was measured using a pulsed deep ultraviolet laser at 258 nm with a power of up to 200 mW. The laser system was based on an Innolight FLARE Yb:YAG pump laser operating at 1030 nm producing 4 ns pulses at 8 kHz with the fourth-harmonic generated using consecutive second-harmonic stages. The laser was focused with a 5× objective lens, into a vacuum chamber via a fused silica window, producing a 10 μm spot on a diamond target. The diamond target was a single crystal “optical grade” (<100 ppb N impurity) CVD material from Element Six, and all measurements in this work were performed on the [100] face.

Prior to each experimental run, the chamber was vented to air and the diamond sample was loaded into the sample holder. The chamber was then pumped with a turbomolecular pump and oil-free backing scroll-pump, followed by a bakeout to about 120 °C for at least 24 h to remove residual water and provide a base pressure of ∼10^–8Torr. A flow of gas was introduced into the chamber via a leak valve to control pressure. Optical profilometry (Wyko NT9300) and SEM were used to characterize etch depth and surface texture.

The etch rate behavior was first measured as a function of air pressure (Fig. 1). Below $10−6$ Torr, etch rates were below the detectable limit (⁠$R<5×10−3$ nm/s). Etch rate increased with pressure, saturating at the ambient etch rate above $∼10−1$ Torr. Best fit was achieved for a sticking coefficient of S_c = 0.015 ± 0.003.

As a function of pure O₂ pressure (Fig. 2), the etch rate was found to become significant at pressures above $10−5$ Torr, but the rate increased markedly for higher pressures. A peak in the etch rate is observed at pressures of approximately $3×10−2$ Torr at a value many times the rate observed under ambient conditions. The factor of enhancement over the ambient rate, which was found to vary depending on the chamber preconditioning, was typically in the range of 5–10 times. As the oxygen pressure was increased above 1 Torr, the etch rate decreased with pressure and returns to the ambient value at pressures above 500 Torr. As shown in Fig. 2, the low pressure behavior fits well using an R₀ value 10 times higher than that used for atmosphere.

It was found that increasing the O₂ pressure above 1 Torr led to an irreversible change in the etch rate pressure dependence. Once the chamber was exposed to higher pressures, lowering the pressure, the etch rate did not recover the higher etch rates but, instead, reduced the rate in a similar fashion to that observed for atmospheric air [Fig. 2(vi)]. Recovering the enhancement could be achieved by baking the chamber under high vacuum conditions, as described in the chamber preparation procedure described above.

Using dry air (<5 ppm water vapor) was found to yield a similar result as to pure oxygen (see also Fig. 2), providing a strong indication for the important role of water vapor on the etch rate behavior. We note that up to 5 ppm of water vapor specified by the manufacturer to be present in the O₂ and dry air used in these experiments at 10^–3Torr results in a monolayer time of a few minutes—thus, at such pressures, the chamber walls are likely to be dosed with a comparatively large amounts of water vapor, significant quantities of which may reach the diamond during etching.

Measurements of the temperature dependence of the etch rate have been used previously to determine the activation energy E_a of the photoejection process.⁵ Similar temperature dependence measurements were also performed in the enhancement regime with primary motivation to determine a change in the activation energy that could help explain the enhancement. Using a sample stage heater, the etch rate was measured as a function of temperature and compared with ambient conditions (Fig. 3). Under ambient conditions, the etch rate increased with temperature in good agreement with the literature [$Ea=80±10$ meV (Refs. 5 and 15)]. Under enhancement conditions, however, the etch rate decreased with temperature and an activation energy was unable to be deduced. This reduction of rate was not reversible by returning the sample to a lower temperature, and water vapor pressure was measured to increase with a residual gas analyzer. As a result, the effect of heating is deduced to be dominated by the effect of water vapor desorbed from the chamber walls as the sample and surrounding mount area were heated, which more than outweighs any thermal rate increase.

SEM imaging of the etched surface shows that low pressures of pure O₂ also bring about marked changes in the etched surface morphology. In ambient conditions, and for the laser polarization parallel to the [110] crystallographic direction, ridge-like structures are formed with amplitude and pitch approximately proportional to the etch depth.¹⁰ An example is shown in Fig. 4(a) for an etch depth of 220 nm. For low pressure O₂, however, etching produced surfaces with characteristically finer and longer ridges [Fig. 4(b)]. The ridge period of 70 nm remains approximately constant across the entire etched area and with depth. These features were only observed in the regime of high etch rate enhancement. Several etched pits were observed to contain a central region with larger ridge structures similar to those observed in ambient conditions, whereas the fine, long range ridges were only observed in the outer regions [Fig. 4(c)]. Such a result indicates that the morphology changes may be highly sensitive to local conditions.

The varying etch rate with pressure is interpreted in the context of the model. A good agreement between the model and experiment was observed for atmospheric air and pure O₂ by choosing $Sc=0.015±0.003$ and fitting R₀. The results show that the R₀ is lower when using air or O₂ at high pressures. The results are inconsistent, however, with changes in the sticking coefficient or $Nsat$⁠, which would manifest as a horizontal shift in the pressure curve peak etch rate.

The observed changes in the etch rate and morphology are attributed to the action of water vapor at the diamond surface. The lower R₀ at high pressure, as well as the irreversible nature of the decay of rate enhancement, is consistent with the introduction of water vapor into the system that is then adsorbed on surfaces throughout the chamber to act as a reservoir. Factors such as the bakeout duration, the time between exposures, and chamber wall temperature, therefore, influence the amount of enhancement. The fact that water in the O₂ and dry air are at levels approximately $10−3$ lower than in the atmosphere agrees well with the expected pressure level (approximately 1 Torr) at which the etch rate starts to decrease substantially from its peak value.

There are a number of ways in which a reduction in R₀ can be interpreted physically. As degradation of rate enhancement was observed when the sample was heated to over 200 °C, the suppression of the etch rate cannot be explained by the restriction of oxygen uptake due to the formation of a physisorbed aqueous layer. Water is known to undergo dissociative chemisorption on unterminated diamond to form H and OH groups. This is well studied on the {100} surface,^16–20 but less so for {111} surfaces²¹ that are most representative of the surface facet of etched surfaces.^9,10 The observed variation in R₀ is attributed to the coadsorption of H and OH groups resulting in a surface configuration less conducive to desorption.

Small amounts of adsorbed H and OH groups are well known to significantly alter the structure of diamond surfaces. Adsorption of O on the {111} surface results in a {111}-(2 × 1) surface;^22,23 however, coadsorption of even small fractions of H or OH converts this to {111}-(1 × 1).^{21,22,24–30} Calculations also indicate that water is dissociatively chemisorbed to unterminated {111} surfaces.²¹ It is proposed that it is this suppression of the {111}-(2 × 1) reconstruction that reduces R₀ and suppresses the etch rate. This accounts for large effects on the etch rate induced by such small quantities of water. The accompanying change in morphology observed in our study may also be indicative of a change in the surface construction.

The stabilization of the {111}-(2 × 1) surface potentially influences the etch rate through either the fraction of photoactive surface species or the probability that a photoexcitation results in desorption. Given the E_a = 80 meV, only 6% of excitation events result in desorption at room temperature. A 10-fold increase in R₀ would be achieved if the activation energy of this thermal desorption step were reduced to E_a = 20 meV. Measurement of the activation energy as a function of rate enhancement, obtained by carefully controlling water vapor pressure in the vicinity of the diamond target as it is heated (e.g., using higher purity gases or water getters), is needed to verify any change in E_a.

The observation of high rate etching has a number of practical implications. The processing time of any potential etching application can be reduced by an order of magnitude by utilizing a dry etching environment. Furthermore, the morphological differences observed in the rate-enhanced regime provide a new parameter space for engineering surface features, such as diffraction elements or anti-reflection structures.

In conclusion, UV LIE measurements were carried out in low pressure conditions for atmospheric air, dry air, and pure oxygen. The etch rate was observed to increase by a factor of 10 in the case of low pressures using dry air and oxygen. This was attributed to the suppression of the etch rate in the presence of even trace amounts of water vapor, which acts to reduce the probability for carbon photoejection. These findings provide additional insights into the etching mechanism and offer practical advantages for reducing processing times for micro- and nanostructuring of diamond surfaces.

While this manuscript was being finalized, the authors became aware of a very recent publication³¹ by Gololobov et al. reporting etching at low pressure for the case of 120 fs pulses at 400 nm. As in our case, enhanced etch rates are observed at low pressure that are attributed to the absence of water. Notable differences are observed in the details of the pressure dependence, which are highlighted by the quantitative analysis presented in this work. There is a pressing need for further investigations to understand the details of this interesting phenomenon, including the influence of vacuum conditions and laser parameters.

This material was based on the research sponsored by the Australian Research Council Discovery Grant (No. DP150102054) and the U.S. Air Force Research Laboratory Grant (No. FA2386-18-1-4117).