Effect of Different Etching Processes on Surface Defects of Quartz Crystals

Abstract

In high-power laser systems, the loading capacity of fused silica components under 351 nm irradiation is an important factor limiting their ability to increase output power, and in the current study, the damage threshold enhancement of fused silica components after RIE and AMP treatments has been investigated. Sub-surface defects in fused silica components after RIE treatment have also been investigated, but the reason for the high damage threshold could never be explained. Since quartz crystals and fused silica belong to the same silica system, and quartz crystals have more characterisation means than fused silica, we can extrapolate to fused silica by studying quartz crystals. We can extrapolate to fused silica by studying quartz crystals, which are characterised by more means than fused silica, and prove that the extrapolation is correct by characterising fused silica. In this study, the relationship between the etching process and the damage threshold is concluded by studying different RIE etching processes, and the damage threshold of the samples is optimal when the etching depth is 1 um.

1. Introduction

In high-energy, high-density laser devices, damage precursors can cause damage to optical elements, which can degrade the optical performance of the entire device [1,2,3]. Therefore, improving the damage threshold of optical elements in devices such as inertial confinement fusion (ICF) holds significant research value in the field of high-energy-density science [4,5]. Currently, the intrinsic threshold of quartz materials can exceed 100 J/cm² for samples exposed to ultraviolet laser light. However, when the optical element is exposed to 351 or 355 ns laser light, the surface of the sample is susceptible to nonlinear effects, which can cause the sample to break or incur other damage [6,7]. The primary reasons for these damages are related to the damage precursors generated during the process and the environmental contamination of the sample cleaning. In large-scale laser devices, damage to the first optical element typically results in damage to the entire element above the beam. Therefore, reducing damage precursors on the surface of optical elements through processing and other means is a critical issue [8,9,10].

In recent decades, various methods have been developed to eliminate damage precursors [11,12,13]. The primary materials used for processing and polishing optical elements are airbag polishing, chemical-mechanical polishing (CMP), and magnetorheological polishing (MRF) methods [14,15,16,17,18]. However, these methods still leave some defects in the processed optical components. For instance, airbag polishing can result in high surface roughness, while CMP may leave residual Ce elements in the polishing solution. Therefore, subsequent processing is still required to eliminate damage precursors in polished samples. We refer to any surface that has been mechanically or chemically treated so that the surface becomes smooth, flat, free of scratches, oxides and other impurities as a polished surface. The area of a few microns to a few hundred microns below the surface is called the subsurface, and the grains, grain boundaries, voids, etc. of the component are called the interior. Overall, visually, point defects can be caused by varying sizes of abrasive particles during the polishing process or uneven polishing pressures, and usually appear as small pits or bumps on the polished surface. Scratches can be caused by improper grinding, polishing or cleaning during the manufacturing process [19,20]. The commonly used physical processing method is the ion beam etching The commonly used physical processing method is the ion beam etching method, which employs high-energy ions to bombard the surface of the optical element. However, this process can reduce the service life of the optical element [21,22]. The chemical processing method currently used involves etching the sample with HF. However, this method tends to increase the pits on the sample surface during processing, leading to the expansion of defects on the optical element’s surface and a decrease in the damage threshold [23,24]. In our previous experiments, wet etching of single crystal silicon with KOH resulted in the adsorption of reaction products on the optical element’s surface, causing secondary contamination. The AMP method is a method of fused silica subsurface damage precursor treatment using inorganic acid leaching and HF etching. In our previous studies, it was found that the scratches can be passivated by this method, which makes the trend of increasing laser loss threshold become larger and more stable, but it cannot remove the subsurface defect layer [25].

Reactive ion etching (RIE) technology is a combination of wet and dry etching commonly used for etching single crystal silicon materials and silicon dioxide to enhance material transmittance. The etching process can also serve as a means of eliminating damage precursors for optical components. Currently, the technology is also widely used for treating damaged precursors in fused silica optical components. However, during the etching process, fluorocarbon compounds are generated due to the reaction between CHF₃ and SiO₂, which may remain on the optical element’s surface [26,27,28]. Quartz crystals are chemically identical to fused quartz, both being SiO₂. fused quartz is less crystalline and has smaller crystals, while quartz crystals are more crystalline and have larger crystals. Fused quartz usually exhibits a glassy or glass-like appearance, while quartz crystals exhibit a crystalline form. When both types of optical components are indicated to be treated, the surface states obtained using the same process are similar, and a variety of means exist to characterise and test the crystalline material as compared to fused quartz. In addition, we know that pure fused silica consists of a single silica, and that the Si-O bonds in quartz glass are arranged in a short-range ordered and long-range disordered state. The Si-O bonds are often used in extreme conditions due to their strong and stable bonding energy. Typically, fused silica is made by melting high-purity silica sand consisting of quartz crystals. SiO₂, on the other hand, is an atomic crystal whose strong chemical bonds are the determining factor. However, the structure and size of the crystal cell also affects the type and strength of the chemical bonds in the crystal. The amorphous form of fused quartz does not allow us to study it directly from a microscopic point of view. Therefore, we resorted to quartz crystals to investigate whether the microscopic level of fused silica is affected by damage under RIE etching. We comprehensively investigated the performance of polished quartz crystal optics optimized for RIE etching treatment in our research. Firstly, we tested the surface and sub-surface defects of quartz crystals using characterization test techniques to obtain data statistics and identify the types of damage precursors. Secondly, we tested the damage performance to discuss and analyze the relationship between damage precursors and laser damage thresholds.

2. Experimental Section

In our research work, the quartz crystals are purchased from the same batch of 7 pieces of quartz crystals (named vendors A, B, C, D, E, F, and G) processed at an established optical component processing manufacturer (Oriet), which are cut and processed from the same billet by the CMP polishing process, and the sample sizes are all 50 mm × 50 mm × 10 mm. The processing surface is perpendicular to the optical axis. Quartz crystal processing equipment used in the same Oxford equipment productsto change the Ar/CHF3 ratio, pressure, power, measured the etching rate of each etching process on the quartz crystal processing, respectively, as shown in

Table 1. Quartz crystal processing technology table.

Sample	Ar (sccm)	CHF3 (sccm)	Intensity of Pressure (mtorr)	Power (W)	Etching Rate (µm/h)	Etching Time (min)
A	5	72	20	200	2.14 ± 0.01	28.0
B	5	72	20	150	1.50 ± 0.01	40.0
C	5	72	20	250	2.36 ± 0.01	25.5
D	38	39	20	200	1.67 ± 0.01	36.0
E	50	27	20	200	2.00 ± 0.01	30.0
F	5	72	10	200	1.92 ± 0.01	31.5
G	5	72	30	200	1.94 ± 0.01	31.0

. Additionally, it should be noted that prior to etching, the sample had a roughness of 2.889 nm. We used the same Oxford equipment products for quartz crystal processing and varied the Ar/CHF₃ ratio, pressure, and power to measure the etching rate of each etching process on the quartz crystals, as shown in Table 1.

Before etching, the substrate was cleaned with Micro-90 cleaner ultrasonically for one hour at a water temperature of 50 °C. The substrate was then repeatedly rinsed with deionized water (resistance value of 18.2 MΩ/cm), washed, and naturally dried. For reactive ion etching treatment, we used an RF 13.56 MHz etching machine. Before etching, the reaction chamber was plasma cleaned with a mixture of argon and oxygen gas for 15 min. After that, a clean sample was placed into the chamber, which was initially vacuumed to 6.0 × 10⁻³ Pa. Then, reactive gases, including trifluoromethane and argon, were introduced into the chamber, and the inlet gas flow rate was adjusted through the valve. The pressure was adjusted by rotating the plate valve, and then the RF switch was turned on to carry out the reactive ion etching process. The sample stage was kept at room temperature (20 °C) during the etching process. The quartz crystal substrate was only etched on one side, and before each plasma cleaning, we wiped around the sample stage and the walls of the reaction chamber with alcohol or acetone (of superior purity) to remove the historical deposits left at the end of the previous experiments. We placed the samples in the same place on the sample stage each time to provide the same plasma state as far as possible. Figure 1 shows the RIE etching process.

3. Results and Discussion

3.1. Etching Rate Calibration

To ensure the accuracy of our study, we chose an etching depth of 1 μm for all samples. The relationship between etching depth and etching time for different samples is shown in the figure. From Figure 2, we observed that the etching rate increases linearly with increasing etching power. When the etching power is less than 200 W, the etching power is the main factor affecting the etching rate. When the etching power exceeds 200 W, the etching rate will gradually level off. By changing the etching gas ratio, we found that the etching rate is highest when the ratio of Ar to CHF₃ is 5:72, while keeping the total volume of the etching gas constant. We also found that the etching rate is highest when the etching gas pressure is 20 mtorr. Therefore, the etching power ratio is the most significant factor affecting the etching rate when other conditions are constant. Additionally, we characterized and detected the surface morphology and defects of the samples with different etching processes to study the relationship between defect evolution and damage threshold. Based on the characterization of quartz crystals, we extrapolated our findings to fused silica and carried out corresponding characterization and verification of fused silica to establish the relationship between quartz crystals and fused silica damage.

3.2. Surface Roughness Analysis

White light interferometry was used to study the morphology and surface roughness of different polished samples [29]. Figure 3 shows the microscope image of the sample surface with a detection area of 180 μm × 250 μm. The results of the white light interferometric tests on samples with different etching processes are given in Figure 3. Figure 3 (A–G) clearly shows the surface morphology and roughness distribution after etching, and we provide the statistical average of the roughness for different samples. Based on the statistical average of roughness, we found that sample A has the smallest roughness, corresponding to an etching power of 200 W, while B and C correspond to etching powers of 150 W and 250 W, respectively. Additionally, we observed that the surface roughness of sample A is more uniformly distributed. This is mainly due to the fact that increasing the RF power leads to more complete ionization of the reactive gas, resulting in larger initial average kinetic energy obtained by the ions, which strengthens the continuous bombardment of the ions on the surface of the quartz crystal and enhances the anisotropic etching on the surface. Furthermore, by comparing samples A, D, and E, we found that when the total amount of gas is kept constant and the ratio of the etching gas is changed, the roughness of sample E significantly decreased. This is mainly because RIE etching involves both physical and chemical etching processes. Physical etching mainly uses Ar ions to bombard the sample for etching, resulting in a more uniform surface. Chemical etching mainly uses CHF₃ to react with SiO₂ for etching, and the etching process can cause some pits on the sample surface, which increases the roughness.

3.3. Chemical Structure Analysis

X-ray photoelectron spectroscopy (XPS) is used to study the stoichiometry and chemical bonding of Si and C on surfaces that are usually sensitive to the local electronic environment, and we obtained the contents of the three elements of silicon, carbon, and oxygen on the surface of the samples by XPS, as shown in Figure 4. Prior to the XPS measurement, the top carbon contamination layer was cleaned by sputtering with an argon ion beam for 100 s. Figure 4 shows the variation of Si, C and O elements in the samples under different etching processes. It is found that no matter which etching process is chosen for the quartz crystals, the chemical elements of the samples do not change after the etching process is completed, which indicates that the etching process is effective in removing defects on the surface of the samples without introducing new impurity elements. In addition, we can also find that the C content in the samples is different under different etching processes. The C content in samples C, E and F is significantly higher than that in other samples, which has an impact on the effectiveness of the subsequent damage test and may lead to a lower sample threshold. However, in general, the RIE etching process does not cause secondary C contamination as well as other forms of elemental contamination.

3.4. Fluorescence Spectral Analysis

Laser-induced fluorescence (PL) is a measurement method that uses laser excitation to detect fluorescence released by particles. We excited the fluorescence of quartz crystals using an emitter with a center wavelength of approximately 280 nm. The fluorescence results show that the different etching processes affect the sample defects in a similar way, but the peaks do not completely overlap due to subtle differences in the etching process. (See Figure 5).

3.5. Photothermal Weak Absorption Analysis

The process of damage occurrence involves thermal absorption, and the photothermal weak absorption technique involves irradiating a localized area of the sample surface with a low fluence laser. This technique reacts to the defects on the surface or sub-surface of the sample by measuring the weak thermal absorption values of the localized area. A photothermal weak absorption imaging system based on PCI technology was used to study absorption defects on the sample surface [30]. 1 W quasi-continuum laser (3.8 μm) was used as the pump beam and a 5 mW He-Ne laser (632.8 nm) was used as the detection beam. The detection sensitivity of the system is close to 0.4 ppm, the test area is 5 mm × 5 mm, and the detection step size is 100 μm. In Figure 6, we present the photothermal weak absorption test data of different samples and summarize their average absorption. The figure shows that there are numerous discrete absorption defects on the sub-surface of the samples, regardless of the process used to etch the quartz crystals. Sample E has a relatively low average absorption level, while sample G has the highest average absorption level. This is mainly due to the presence of localized pits resulting from chemical reactions during the RIE etching process. Due to the non-uniform density distribution of the pits, thermal pockets are generated when the laser is irradiated, resulting in thermal stresses. Therefore, the results show that the weak absorption detection technique can be used as a non-destructive method for evaluating the etching level of optical processing of quartz crystals. However, due to the random selection of the area during the testing process and the large selection of the test range, it is challenging to determine the morphology and shape of the subsurface defects from a microscopic point of view.

3.6. X-ray Diffraction Analysis

We conducted XRD tests on samples with different machining processes to obtain theta values of the (002) surface and the half-height width of the rocking curve. The theta values are presented in

Table 2. Situation of theta values of the samples before and after etching.

Sample	Pre-Etch (rad)	Post-Etch (rad)
A	0.009	0.007
B	0.008	0.0075
C	0.009	0.0085
D	0.0075	0.008
E	0.007	0.0065
F	0.0085	0.007
G	0.0075	0.0065

.

Quartz crystals belong to the hexagonal crystal system, and since the processing surface is perpendicular to the C-axis, the constant C* at the (002) surface can be calculated using Equations (1) and (2). The values of C for the samples before and after the etching process are presented in

Table 3. C magnitude of samples before and after etching.

Sample	Pre-Etch	Post-Etch
A	0.9807764 × 10−6	1.260998 × 10−6
B	1.103373 × 10−6	1.176932 × 10−6
C	0.980776 × 10−6	1.038469 × 10−6
D	1.176932 × 10−6	1.103373 × 10−6
E	1.260998 × 10−6	1.357998 × 10−6
F	1.038469 × 10−6	1.260998 × 10−6
G	1.176932 × 10−6	1.357998 × 10−6

. The h, k and l in Equation (1) are the crystallographic indices.

$$\frac{1}{d^2}=\frac{4}{{3a}^2}(h^2+hk+k^2)+\frac{l^2}{c^2}$$

(1)

$$2d\sin \theta =n\lambda$$

(2)

It is evident that after etching, the value of the lattice constant C* changes, the crystal cells are deflected, and the distribution of lattice orientation becomes wider, indicating that the lattice orientation of the crystals becomes less organized after plasma etching. Among them, the lattice constant of sample D decreases, while the lattice constant of the other samples increases due to stretching. During the ordinary polishing process, the crystal is compressed, the lattice is squeezed, and the lattice constantly decreases. However, after etching, the lattice constantly increases.

In our previous studies, it is known that in RIE etching, oxygen can increase the etching rate and surface roughness, while fluorine gas can slow down the etching rate and improve the surface quality. Typically, higher gas pressures increase the etch rate, but may also result in increased surface roughness. RF power and frequency have an effect on the plasma density and energy, which affects the etch rate and results. Typically, higher RF power and frequency can increase the etch rate, but may also result in a decrease in surface quality. Combined with the above characterization tests, we found that sample A has medium roughness, low C content, and a high etching rate. Sample B has low roughness, medium C content, and a low etching rate. Sample C has medium roughness and high C content. Sample D has low roughness, low C content, but a low etching rate. Sample E has low roughness but high C content. Sample F has the least amount of roughness but high C content. Sample G has more roughness. In samples A–C, the higher the etching power, the higher the etching rate. However, in the samples with etching power of 250 and 300 W, the increase in the etching rate decelerates significantly with the increase in etching power. Moreover, the carbon content of sample C is significantly higher than that of samples A and B in the XPS test. Therefore, considering the rate and carbon content, we selected an etching power of 200 W. When the etching pressure was changed to 20 mtorr, the etching rate reached the maximum value. In the weak absorption test, the weak absorption value of the sample processed at 20 mtorr was smaller than that of samples processed at 10 mtorr and 30 mtorr. When the total amount of gas was kept constant and the etching gas ratio was changed, the carbon content in the XPS test of the samples processed by etching when the ratio of the etching gas argon to trifluoromethane was 5:72 was lower than that of the samples processed with the other two gas ratios. Therefore, we selected an etching process with an etching power of 200 W, an etching pressure of 20 mtorr, and an etching gas ratio of 5:72 between argon and trifluoromethane. The sample processed by this etching process has a larger etching rate, smaller C content, and smaller weak absorption value.

3.7. Laser Damage Threshold of the Sample

The laser-induced damage threshold (LIDT) of the samples was evaluated using small-beam 1-on-1 testing in a laser damage test facility with a triple-frequency Nd-YAG laser. The wavelength, pulse width, and laser repetition rate were 355 nm, 9.3 ns (FWHM), and 1 Hz, respectively. During the test, we randomly selected ten points on the surface of the samples, and each point was exposed to the same energy of the laser beam. We then calculated the damage probability density with the zero-probability damage threshold. The statistics for the zero-probability damage threshold are shown in Figure 7. The results indicate that different etching processes enhance the laser damage resistance performance differently. The zero-probability damage thresholds of samples A and G are significantly higher than those of the other samples, which is consistent with our characterization of the surface and sub-surface defects of the samples. It has been shown that RIE etching can effectively remove surface/subsurface defects from quartz crystal elements. When a reasonable etching process is used, the introduction of impurities can be effectively avoided without causing secondary defects and contamination to the sample. In addition, it is shown that there is a positive correlation between the damage precursor detection and characterisation results and the damage threshold, and the damage precursor detection technique based on quartz crystals can well evaluate and predict the damage performance of the samples, and it can be analogous to the process of the damage performance study of fused silica, which can provide mutual corroboration of the damage precursor detection of fused silica.

4. Conclusions

Reactive ion etching (RIE) is an effective method for removing and preventing two key types of precursors: impurity element contamination in the redeposited layer, and subsurface damage (SSD) in the polished layer, which are associated with destructive absorption of sub-bandgap light and subsequent damage induced by UV irradiation. As shown by the XPS results above, the optimized RIE process produces a pristine sample surface with a relatively stable stoichiometry and a low concentration of carbon atoms. The damage test results demonstrate that the etched quartz crystal samples are significantly more resistant to damage. Since quartz crystals and fused silica belong to the same silica system, and quartz crystals have more means of characterization than fused silica, we can extrapolate the results of the present study to fused silica to provide guidance for the removal of damage precursors and the improvement of damage threshold in fused silica.