1 Introduction

Mid-infrared (mid-IR) laser sources with emission wavelengths from 3 μm to 5 μm, and within the atmospheric window, are promising light sources in the fields of optical communication, remote sensing technology, and semiconductors [1,2,3]. As this band is in the strong absorption zone of water (~104 cm−1), the penetration depth when mid-IR light is used to cut through skin or biological tissue is only a few micrometers [4, 5]. The cubic rare-earth sesquioxide crystal, lutetium oxide (Lu2O3), is a promising mid-IR laser matrix owing to its excellent properties. The effective phonon energy of Lu2O3 is 430 cm−1, which is quite low compared to the host material yttrium aluminum garnet (YAG) at 700 cm−1, implying low nonradiative transition rates and, therefore, higher quantum efficiency [6]. The thermal conductivity of undoped Lu2O3 (12.5 W/(m·K)) is higher than that of undoped YAG (11 W/(m·K)). If the crystals are doped with 3% ytterbium ions, the thermal conductivity drops considerably to 6.6 W/(m·K) and 6.8 W·(m·K)−1, for Sc2O3 and YAG respectively. However, in Lu2O3, the thermal conductivity changes only slightly to 11.0 W/(m·K). These properties make Lu2O3 attractive for high-power solid-state lasers [7, 8].

To use Lu2O3 for high-power laser devices, an atomically smooth and damage-free surface is required. However, Lu2O3 is difficult to machine because of its physical properties and chemical inertness. Lu2O3 is a hard and brittle material, with a hardness of approximately 10.8 GPa and a melting point up to 2 400 °C [9,10,11]. It has a strong stability in acid solution. For example, it remains almost unchanged when placed in aqua regia for 24 h. The study of sesquioxide crystal materials mainly focuses on crystal growth, spectral characteristics, and laser properties [12]. At present, no studies on ultrasmooth finishing of Lu2O3 have been reported.

Traditional cutting [13, 14], grinding [15], and polishing processes [16] are the main methods of processing laser crystals. These processing methods mainly employ mechanical contact to remove workpiece materials. However, the stability of the machining equipment is an important factor in determining the accuracy of the crystals. It is also affected by external disturbances in the relative displacement between the workpiece and tool owing to vibration and thermal deformation because the surfaces of these crystals are created by a contact removal mechanism [17]. Inevitably, the surface quality and lattice integrity of crystals are damaged because of mechanical machining, resulting in surface/subsurface damage and residual stress, which directly reduces the damage threshold. Therefore, several unconventional noncontact techniques for hard and brittle material with nanometer-level form accuracy have been reported. Namba and Tsuwa [18] proposed float polishing in which a surface roughness 1 nm Ra could be obtained for sapphire single crystals. Mori et al. [19] demonstrated that minute atomic-size removal was achieved with no damage by elastic emission machining. Gormley et al. [20] suggested hydroplane polishing, which produces damage-free surfaces of gallium arsenide and Indium phosphide. Li et al. [21] proposed chemical-mechanical polishing, which is now widely used as the finishing process of 4H-SiC substrates. Kordonski et al. [22] used magnetorheological finishing to reduce the surface roughness of optical glasses to less than 1 nm Ra. Yang et al. [23] used slurryless electrochemical-mechanical polishing to obtain an atomically smooth SiC surface. However, the removal rates of these finishing processes are extremely low. Plasma surface treatment technology can create surfaces that are difficult to achieve with traditional processing methods. Arnold et al. [24] presented plasma jet machining (PJM), a method suitable for optical surfaces of ultralow expansion glass; the volume removal rate was up to 50 mm3/min and the surface roughness was less than 3 nm Ra. On the other hand, Sun et al. [25] and Yamamura et al. [26] proposed plasma chemical-vaporization machining (PCVM) and plasma-assisted polishing. These plasma technologies remove material by chemical reactions, forming volatile species, or reducing the hardness of modified surfaces, and are mainly used in silicon-based materials. Therefore, it is necessary to develop an efficient finishing process for the ultrasmooth surfaces of sesquioxide crystals.

In this study, a novel finishing approach, i.e., plasma-assisted etching (PaE), combining the irradiation of hydrogen plasma for surface modification and inorganic acid etching for surface removal was used to realize ultrasmooth surfaces of crystal samples. The surface chemical structure of Lu2O3 modified by irradiation of hydrogen plasma was investigated by X-ray photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM), and the etching results of Lu2O3 using inorganic acid were analyzed.

2 Plasma-assisted etching

In plasma etching processes, such as PJM and PCVM, a damage-free surface can be obtained because of the chemical removal characteristics of this processes [24, 25]. However, these processing methods require the plasma to act on the substrate to generate volatile gases; otherwise, the chemical reaction cannot continue. Wet etching is widely used in the semiconductor industry, which requires etching of the substrate material.

In the process under development, irradiation of reactive plasma was employed to modify the surfaces of a chemical inert material to form a more active layer. Subsequently, acid etching was used to preferentially remove the active layer. Figure 1 shows a schematic of PaE process. After hydrogen with a flow rate of 30 standard-state cubic centimeters per minute (cm3/min) was supplied to the Diener plasma generator, some hydrogen was excited, becoming reactive radicals at radio frequency (RF) (f = 13.56 MHz). The reactive radicals caused chemical reactions with the surface atoms. As a result, hydrogen ions were introduced into the sesquioxide, transforming it into hydroxide. Additionally, hydroxide is more likely to react with acids than sesquioxide. After plasma modification, inorganic acid was used to remove the modified layer to obtain an ultrasmooth surface without surface/subsurface damage. The experimental parameters are listed in Table 1.

Fig. 1
figure 1

Schematic of plasma-assisted etching

Table 1 Experimental parameters in plasma-assisted etching

3 Experimental investigations

3.1 Plasma modification process

XPS and TEM were employed in this experiment to investigate the modified surfaces and confirm the generation of a reacted layer on Lu2O3 with hydrogen plasma. Figures 2a, b show the X-ray photoelectron spectra of the processed surfaces, which correspond to Lu 4f and O 1s, respectively. Subfigures (i), (ii), and (iii) of Fig. 2a, b show the surface before hydrogen plasma irradiation, the surface after hydrogen plasma irradiation, and the surface after PaE, respectively. The peaks corresponding to the Lu–O bond (7.5, 8.7, 529.7 and 532.5 eV) can be observed in Figs. 2a (i) and 2b (i) [27, 28]. The peaks observed at 10.4 eV and 531.6 eV were identified as the Lu–OH bond in Figs. 2a (ii) and 2b (ii), corresponding to Lu(OH)x, a mixture of LuOOH and Lu(OH)3 [28,29,30]. These results indicate that the irradiation of hydrogen plasma converted the sesquioxide into hydroxide on the surface of Lu2O3, and it is thought that the hydrogenation species in this reaction system was hydrogen radical because only hydrogen gas was used to excite the plasma. After PaE, the peak intensity of the Lu–OH bond disappeared, as shown in Figs. 2a (iii) and 2b (iii). Additionally, it is thought that the inorganic acid removed the modified layer easily. The main chemical reaction in this process is described as follows.

Fig. 2
figure 2

XPS spectra of the processed Lu2O3 surfaces, which correspond to a Lu 4f7/2, b O 1s ((i) before plasma treatment, (ii) after plasma treatment, (iii) after acid etching)

$${\text{Lu}}_{ 2} {\text{O}}_{ 3} {\text{ + H}}^{ *} \mathop{\longrightarrow}\limits{{}}{\text{Lu(OH)}}_{x}$$
(1)
$${\text{Lu(OH)}}_{x} {\text{ + H}}^{ + } \mathop{\longrightarrow}\limits{{}}{\text{Lu}}^{{ 3 { + }}} {\text{ + H}}_{ 2} {\text{O}}$$
(2)

TEM is a very direct and clear way to observe subsurface changes. Figure 3 shows the TEM analysis of the Lu2O3 surface before and after irradiation by hydrogen plasma for 2 h, and the illustration is the fast Fourier transform (FFT) pattern for the white selected area [31,32,33]. The modified layer was covered by a Pt layer to protect it during the focused ion beam fabrication process. As shown in the low-resolution TEM images in Figs. 3a and b, a modified layer with a thickness of approximately 65 nm was generated after the plasma treatment. In the high-resolution TEM images and FFT pattern in Fig. 3, the Lu2O3 crystals changed from a homogeneous lattice to a disordered lattice after the hydrogen plasma treatment.

Fig. 3
figure 3

TEM images of Lu2O3 surface a before plasma treatment, b irradiated by hydrogen plasma for 2 h

By combining the characterization results of TEM and XPS, it can be proved that the modified layer was formed by converting oxides into hydroxides, and the thickness of the modified layer was 65 nm.

3.2 Acid etching process

Lutetium oxide crystals have strong chemical inertia and extremely low reaction rates with acid. Figure 4 shows the two-dimensional (2-D) white light interferometry (WLI) image of morphological changes after the reaction of lutetium oxide with nitric acids or aqua regia for 1 h. The three 2-D contour curves are almost identical, which indicates that Lu2O3 hardly reacts with these two acids.

Fig. 4
figure 4

2-D WLI image of Lu2O3 after etching

Copper tape was used to cover half of the sample before plasma treatment to observe the removal effect after PaE. Then, the copper tape was removed and the modified Lu2O3 sample was etched with HNO3 to remove the modified layer. The etching results were obtained by comparing the surfaces before and after HNO3 etching, as shown in Figs. 5 and 6. A boundary between the modified area and the nonmodified area can be observed in Fig. 5. The gradual disappearance of scratches from the nonmodified area to the modified area indicates the effectiveness of the etching removal. The high-magnification SEM image in Fig. 5 shows that numerous scratches are exposed at the boundary. This could be the subsurface damage from previous processing which was exposed after PaE processing, and the scratches in the PaE processing area may have disappeared. Figure 6 shows a WLI image of the Lu2O3 sample, where the height of the modified area is lower than that of the nonmodified area by 25.7 nm. Moreover, increasing the acid etching time and changing the type of acid did not result in the removal of samples, proving the feasibility of PaE.

Fig. 5
figure 5

SEM images of the Lu2O3 surface processed by PaE

Fig. 6
figure 6

Etching boundary of Lu2O3 wafer by HNO3 for 1 h

3.3 Surface and subsurface damage

To maximize the crystal laser performance, the application of Lu2O3 substrates requires not only a flat surface with no scratches, but also no subsurface damage. The surface quality of Lu2O3 substrates is guaranteed after chemical-mechanical polishing. An ultraprecision lathe was used to carve crosslines on the surface of the Lu2O3 sample, thus ensure repeatability of the measurement position. As shown in Fig. 7, an area of 50 μm × 50 μm at 200 μm from the center of the crossline was selected as the marked area. Figure 8 shows the WLI and SEM images of the processed surface in the marked area; the surface roughness decreased slightly from 0.40 nm to 0.38 nm. Therefore, a flat surface can be guaranteed after PaE treatment. As for the subsurface damage, the scratches on the boundary between the modified and unmodified areas in Fig. 5 imply subsurface damage exposure after inorganic acid etching. The subsurface damage of Lu2O3 can be evaluated by TEM observation. However, the measurement range of TEM is quite small.

Fig. 7
figure 7

SEM images of the marking point on Lu2O3 surface a cross line area, b marked measuring area

Fig. 8
figure 8

SEM and WLI images of Lu2O3 surface at the marked point a before PaE processing the arithmetical mean height of the surface (Sa) 0.40 nm, b after PaE processing Sa 0.38 nm

Raman spectroscopy is a nondestructive and efficient analysis method for characterizing subsurface damage [34]. This technique was used to characterize subsurface damage of marked areas before plasma treatment, after plasma treatment, and after acid etching; the results are shown in Fig. 9. The surface of Lu2O3 was modified with a layer of approximately 65 nm, as shown in Fig. 3c. In contrast, the Raman (532 nm laser) resolution depth was much higher than the layer thickness; the modified layer on the surface significantly affected the monocrystalline properties and the peaks decreased. The Raman peaks were higher after inorganic acid etching than before plasma treatment, indicating that the modified layer and the subsurface damages were both removed. Meanwhile, the disappearance of the XPS peak intensities of the Lu–OH bond after acid etching indicates that the modified layer can be removed easily, which means that an atomically flat Lu2O3 surface is obtained.

Fig. 9
figure 9

Raman spectra of Lu2O3 under different treatment conditions collected with 532 nm laser

4 Conclusions

A nanometric finishing approach combining irradiation of hydrogen plasma with inorganic acid etching was proposed to achieve high-integrity Lu2O3 surfaces without introducing subsurface damage. The results of the XPS and TEM measurements show that the irradiation of hydrogen plasma thoroughly converted the sesquioxide into hydroxide. The SEM, WLI, and Raman spectroscopy results indicate that the irradiation of hydrogen plasma increases the etching rate of Lu2O3 without introducing scratches or crystallographic subsurface damage. The experimental investigation reveal that PaE can achieve an atomically smooth surface in finishing a Lu2O3 substrate without introducing scratches or subsurface damage.