Introduction

With rapid development of nanoscience and nanotechnology, nanostructures have attracted increasing attention in catalysis, optoelectronics, sensing, and data storage [1,2,3]. Silicon (Si) is the most commonly used substrate and structural material in photovoltaics and electronic industry, and its nanostructures play a significant role in improving the performance of Si-based devices [4,5,6,7]. For instance, various surface nanostructures (e.g., nanocones, nanopyramids and nanowires) fabricated by selective removal of Si from bulk substrates can significantly improve light absorption efficiency of solar cells [8]. Patterned Si substrates were demonstrated to effectively alleviate the effect of wafer defects on epitaxial layers [9]. The use of Si nanogratings can provide better device stability for sensing [10]. Nevertheless, crystal defects induced during fabrication of Si nanostructures can degrade the optical and electrical properties of Si-based devices to some extent, strongly limiting their reliability and lifetime [11]. Specifically, defect-rich surfaces could induce rapid recombination of electron–hole pairs, thereby declining average carrier lifetime, and increase current leakage that may amplify the noise [12]. Consequently, the development of novel fabrication approaches for nondestructive Si nanostructures is highly desirable for advanced applications [13].

Typical methods used for nanostructuring of Si substrates mainly include photolithography, ion beam lithography (IBL), electron beam lithography (EBL), and nanoimprint lithography (NL) [14,15,16]. However, these methods still suffer from various challenges (e.g., low resolution, high cost, and complex operating environment and the limitation of processing materials). By comparison, scanning probe lithography (SPL) is advantageous in terms of low cost, high resolution, and good flexibility. Hence, SPL is becoming an attractive approach for the fabrication of micro/nanostructures [17, 18]. Recently, combining scanning probe microscope (SPM) scratching and chemical etching, friction-induced selective etching (or so-called mechano-chemical SPL) is demonstrated to be useful for the fabrication of micro/nanostructures on Si, gallium arsenide, and quartz surfaces [19,20,21]. This route could address the SPL technical challenges in wear debris, tool wear, and substrate damage to some extent. During friction-induced selective etching process of Si, anisotropic etchants [e.g., potassium hydroxide (KOH) and tetramethylammonium hydroxide (TMAH)] are commonly used for the etching of scanned Si surfaces. The subsurface deformation induced by the scanning process can resist the etching, thereby forming hillock nanostructures [22]. However, the size and shape of nanostructures are restricted by crystallographic properties of Si during the etching of anisotropic etchants. Furthermore, the scanning-induced subsurface defects might impact subsequent applications of fabricated surface nanostructures. Therefore, eliminating the scanning-induced defects while keeping a flexible fabrication process of Si nanostructures is highly desirable.

A mixture of mixed hydrofluoric acid (HF) and nitric acid (HNO3) is the most commonly used isotropic etching system for processing Si surfaces [23]. The mixtures are widely applied for the removal of saw-damage and obtaining low reflectivity surfaces during solar cell manufacturing [24]. Moreover, they provide flexible fabrication of micrometer-scale Si structures and present good compatibility with one-chip circuitry. This makes the fabrication of Si nanostructures without crystal defects possible. However, the etching profile of the mixtures is limited by the isotropic dissolution of Si. More importantly, scaling down the etching dimensionality to the nanoscale is challenging, severely hindering the applications in nano-sized functional devices [25].

In this paper, low-cost and rapid approach consisting of site-controlled amorphous formation and subsequent anisotropic etching in HF/HNO3 mixtures was developed to fabricate nanostructures on Si surfaces. A diamond tip was used to prepare friction-induced hillocks through reciprocating scanning Si surface. The effects of volume ratio of HF/HNO3 mixtures, sliding cycle, normal load, and etching time on the formation of nanostructures were investigated in an effort to determine the optimal fabrication parameters. High-resolution transmission electron microscope (HRTEM) was utilized to view the cross section of fabricated nanostructures. The mechanism of anisotropic etching of scanned regions was clarified based on an electrochemical dissolution model. The results showed that the proposed approach was feasible for controllable fabrication of nanostructures with any layout pattern on Si surfaces.

Materials and methods

B-doped Si(100) wafers covered with a native oxide layer (~ 0.5 nm in thickness) were purchased from MEMC Electronic Materials, Inc., USA. The surface root-mean-square roughness of the wafers measured by an atomic force microscope (AFM, E-Sweep, Hitachi Instruments Inc., Japan) over an area of 2 μm × 2 μm was less than 0.1 nm. Before the fabrication, the Si wafers were ultrasonically cleaned successively in acetone and alcohol for 10 min, and then continuously rinsed with deionized water to remove surface contaminants.

An AFM diamond tip (NC-LC, Adama, Ireland) with a radius R of ~ 400 nm was employed to prepare the friction-induced hillocks by reciprocating sliding under a line-scanning mode. The normal spring constant of the probe cantilever was calibrated as ~ 100 N/m by MFP-3D AFM (Asylum Research, UK) using thermal noise method. Analytical grade HF [40% (w/w)] and HNO3 [68% (w/w)] were purchased from Sinopharm Chemical Reagent Co., Ltd, and were used for performing etching experiments. All AFM images were scanned by contact mode using a Si3N4 tip (MLCT, Veeco Instruments Inc., Plainview, NY, USA) with a spring constant of 0.1 N/m.

The fabrication process mainly consisted of two steps, as shown in Fig. 1. Firstly, the AFM diamond tip was used to produce protrusive hillocks instead of material removal (groove-shaped scratch) on Si surface by reciprocating sliding at an applied normal load (Fn) of 0.5–50 µN. The scanned Si samples were then immersed in HF/HNO3 mixtures at the temperature of 25 ± 2 °C and relative humidity of 50 ± 5%. This process significantly increased the hillock height. The microstructures of the fabricated nanostructures were observed by an HRTEM (Tecnai G2 F20, FEI, USA), and cross-sectional samples were prepared by a focused ion beam (FIB, Nanolab Helios 400S, FEI, Netherlands). To prevent the damage from FIB milling, successive deposition of an epoxy polymer and Pt layer was carried out on sample surfaces before milling process.

Figure 1
figure 1

Surface topographies and corresponding cross-sectional profiles of a friction-induced hillock before and after the etching in HF/HNO3 mixtures for 2 s

Full size image

Results and discussion

Effects of scanning and etching parameters on anisotropic etching

Volume ratio of HF/HNO3 mixtures

The volume ratio of HF/HNO3 mixtures could significantly impact the etching rate of Si wafers [26]. Compared to HF-rich HF/HNO3 mixtures (HF:HNO3 > 1:1, v/v), HNO3-rich HF/HNO3 mixtures (HF:HNO3 < 1:1, v/v) presented better safety and controllability due to their relatively low etching rate. Accordingly, HNO3-rich HF/HNO3 mixtures should be more suitable for the fabrication of Si nanostructures [27, 28]. To investigate the concentration-dependent anisotropic etching process, various volume ratios of HNO3-rich HF/HNO3 mixtures (HF:HNO3 = 1:200, 1:100, 1:80, 1:60, 1:40, 1:20, 1:10, 1:5, and 1:1) were used for the etching of similar friction-induced hillocks produced at Fn = 40 μN and v = 40 μm/s. After the etching of the hillocks for 2 s, the evolution of surface topography in scanned regions showed significant dependence on volume ratio, as shown in Fig. 2. At volume ratio of 1:200, no significant changes in etched topography were observed, demonstrating a ratio with poor anisotropic etching capacity. As the ratio rose from 1:100 to 1:5, the height of fabricated nanostructures dramatically increased, confirming volume ratios with excellent anisotropic etching abilities. At volume ratio of 1:1, discontinuous and irregular nanostructures were noticed, implying unsuitable ratio for Si nanofabrication. Moreover, the fabricated hillock nanostructures presented different cross-sectional profiles as the ratio increased from 1:100 to 1:5. Note that such features could hardly be realized by the etching of anisotropic etchants such as KOH solution due to their intrinsic etching characteristics. The correlation between fabrication height/width and volume ratio of HF/HNO3 mixtures was displayed in Fig. 3a. The fabrication height firstly enhanced from 30 to 173 nm as volume ratio increased from 1:100 to 1:60, and then gradually decreased as volume ratio further rose. Besides, the fabrication width exhibited less sensitivity to the volume ratio of HF/HNO3 mixtures.

Figure 2
figure 2

AFM topography of hillock nanostructures fabricated by Fn = 40 μN at 100 reciprocating cycles and post-etching for 2 s under various volume ratios. The corresponding volume ratios are marked in AFM images

Full size image
Figure 3
figure 3

The effect of volume ratio of HF/HNO3 mixtures on a the height/width of hillock nanostructures and b surface roughness. The dotted lines of Fig. 3a are drawn to guide the eyes for fabrication height/width, and the data are estimated from Fig. 2. The height H and width W of the hillock nanostructure are defined in the inset of Fig. 3a

Full size image

In order to understand the distinctive anisotropic etching more clearly, two Si wafers with similar friction-induced hillocks were etched by pure HF solution and pure HNO3 solution, respectively. Figure S1 (In Supplementary Material) displays AFM images and corresponding cross-sectional profiles of the friction-induced hillocks before and after being etched by pure HF solution (Fig. S1a) or pure HNO3 solution (Fig. S1b). It is indicated that the height of friction-induced hillocks after being etched by pure HF solution for 2 s decreased from 3.6 to 2.7 nm, and that by pure HNO3 solution for 2 s decreased slightly as well. Such results demonstrated that any anisotropic etching can be hardly realized by pure HF solution or pure HNO3 solution. Therefore, the distinctive anisotropic etching in the present study can be further described as synergistic effect of HF and HNO3 in the mixtures.

Besides, note that the performance of Si-based functional surfaces often depends on form accuracy and surface quality [29]. Therefore, it is necessary to investigate the effect of the volume ratio of HF/HNO3 mixtures on the roughness of etched Si surfaces to provide a reference for the fabrication and packaging process of devices. To this end, the surface roughness was characterized by AFM after the etching of Si samples by different volume ratios of HF/HNO3 mixtures for 2 s. The correlation between the surface roughness and volume ratio is presented in Fig. 3b. The surface roughness gradually increased with the volume concentration of HF solution. In particular, the surface roughness was less than 1 nm for volume ratios ranging from 1:200 to 1:40. By comparison, the surface roughness rose to ~ 6.1 nm at the ratio of 1:1. The insets in Fig. 3b displayed the surface topographies etched by HF/HNO3 mixtures at volume ratios of 1:60 and 1:1. Comprehensive consideration of anisotropic etching ability and surface roughness, volume ratios from 1:100 to 1:5 looked optimal for the fabrication of Si nanostructures. If not specified, all etching experiments were performed at ratio 1:60.

Scanning parameters

In addition to volume ratio of the HF/HNO3 mixtures, scanning parameters may also strongly affect the formation of hillock nanostructures. To investigate the effect of sliding cycles N on fabrication height/width, several friction-induced hillocks were respectively prepared under Fn = 50 μN with N = 1, 10, 100, 200, and 400. In Fig. 4, the height of newly formed hillock nanostructures enhanced from 18 to 200 nm as N varied from 1 to 100 after the etching of friction-induced hillocks by HF/HNO3 mixtures for 2 s. Also, the hillock height tended to stabilize at ~ 200 nm for N values beyond 100. The results demonstrated that the etched hillock nanostructures were obviously affected by N, which can be correlated with the thickness of anti-etching layer on friction-induced hillocks. In fact, the generation of friction-induced hillocks on Si surface can be dominated by mechanical interaction, which can be further described as frictional shear-induced structural deformation, such as amorphous Si (a-Si) and dislocations [30]. Correspondingly, the increase of sliding cycles N from 1 to 100 in the present study was beneficial for the formation of a-Si, resulting in the increase of the thickness of surface amorphous layer. As N further rose (beyond 100), it was expected that the thicker a-Si layer can reduce the mechanical interaction between diamond tip and bulk crystalline Si, and hence the thickness of surface amorphous layer almost trended to be stable. Consequently, the fabrication height firstly increased and then gradually trended to be stable as N increased. Certainly, the fabrication width also exhibited a similar variation trend. Therefore, it was suggested that lager N can hardly improve the barrier ability of scanned regions to resist the etching process when N reached 100. In sum, the optimal value of N for anisotropic etching of scanned regions was determined as 100 when considering the fabrication results and efficiency.

Figure 4
figure 4

a AFM images, b cross-sectional profiles, and c fabrication height/width of produced hillock nanostructures under different sliding cycles N. The respective values of N are 1, 10, 100, 200, and 400. The applied normal load is 50 μN and etching time is 2 s

Full size image

The influence of normal load Fn on the fabrication height/width was studied and the data are illustrated in Fig. 5 and S2. At Fn = 0.2 μN (corresponding to the maximum Hertzian contact pressure Pmax of ~ 1.6 GPa) [31], friction-induced hillocks can be hardly formed on Si surface. Correspondingly, no anisotropic etching phenomenon was noticed after the etching of the scanned sample by HF/HNO3 mixtures for 2 s. This might be attributed to the absence of plastic deformation under such loading conditions. As Fn increased to 0.5 μN (corresponding to Pmax of ~ 2.1 GPa), a friction-induced hillock with a height of ~ 0.55 nm was produced on Si surface. After the etching of the friction-induced hillock by HF/HNO3 mixtures for 2 s, the hillock height reached ~ 5 nm. Further increase in Fn from 0.5 to 50 μN, the height of the newly formed hillock nanostructures rose from 5 to 200 nm. Also, the hillock width enhanced from 190 to 420 nm. This may be attributed that higher normal load may induce thicker and wider deformation distributions, thereby yielding higher and wider hillock nanostructures. Consequently, normal loads from 0.5 to 50 μN can easily control the fabrication height/with.

Figure 5
figure 5

The change in fabrication height/width as a function of the applied normal load. The respective values of Fn are 0.2, 0.5, 1, 2, 4, 8, 16, 24, 32, 40, and 50 μN. The etching time is 2 s

Full size image

Etching time

The etching time also strongly affected the formation of fabricated nanostructures in bulk micromachining based on etching technology. In order to investigate the effect of etching time on fabrication height/width, a friction-induced hillock produced under Fn = 50 μN was etched by HF/HNO3 mixtures for 2–10 s. The surface topographies of the as-obtained samples were characterized by AFM every 2 s and the results were gathered in Fig. 6. For the sample etched for 2 s, the height of fabricated hillock nanostructure swiftly rose to ~ 200 nm. As etching time increased from 2 to 10 s, the fabrication height gradually decreased until reaching ~ 15 nm. The width also reduced from 420 to 242 nm. Therefore, the fabrication height/width can easily be controlled by etching time. Note that prolonged etching can result in the collapse and even final disappearance of the fabricated nanostructures. This may be attributed that the subsurface deformation in scanned regions was gradually etched away, resulting in the weakness of the anisotropic etching. In addition, the effect of etching time on surface roughness was also investigated and the data are provided in Fig. S3 of Supplementary Material. As etching time rose from 2 to 10 s, the surface roughness increased from 0.7 to 7.7 nm. Hence, short etching times (within 6 s) should be optimal for the fabrication process.

Figure 6
figure 6

a AFM images, b cross-sectional profiles, and c fabrication height/width of produced hillock nanostructures under different etching time. The respective values of etching time are 2, 4, 6, 8, and 10 s

Full size image

Anisotropic etching mechanism

The above results demonstrated HF/HNO3 mixtures to possess excellent anisotropic etching nanofabrication abilities despite being typical isotropic etching systems of Si. To gain a better understanding of anisotropic etching mechanism of scanned regions, a friction-induced hillock, consisting of a-Si and deformed Si [30], was analyzed based on surface topography of scanned regions before and after being etched by HF solution and HF/HNO3 mixtures, as shown in Fig. 7. After the etching by HF solution for 40 min, the hillock with a height of ~ 5.5 nm evolved to form a groove with a depth of ~ 26 nm and remained to be stable with further etching. This indicated the complete removal of the topmost amorphous layer of the hillock. Further etching in HF/HNO3 mixtures for 2 s led to an increase in the groove depth to ~ 34 nm (Fig. 7b). Thus, one can deduce that a-Si can resist HF/HNO3 mixtures etching, while deformed Si provided little contribution to the formation of hillock nanostructures.

Figure 7
figure 7

a AFM images and b cross-sectional profiles of a friction-induced hillock before and after being etched by HF solution and HF/HNO3 mixtures

Full size image

To clarify the effect of scanning-induced defects on the atomic structures of fabricated hillock nanostructures, cross section of a hillock, which was produced by Fn = 32 μN and post-etching for 2 s, was prepared by FIB milling for subsequent HRTM observations. In Fig. 8, a distinct dividing line could be clearly distinguished between protective layer and the hillock. The enlarged HRTEM images showed almost perfect crystal lattice of the hillock. Moreover, a comparison of selected area electron diffraction (SAED) pattern of the fabricated hillock with that of defect-free region showed no additional diffraction spots or rings, indicating that the interior of the hillock preserved its original single-crystal structure without the presence of a-Si [32]. Note that a-Si layer can be selectively removed by HF solution [28], and hence surface topography evolution of fabricated hillock structure before and after being etched by HF solution can be used to further verify whether a-Si existed near crystal Si on the hillock structure. As shown in Fig. S4a–c, the change in the height and cross-sectional profiles of fabricated hillock structure can be hardly noticed after the etching in HF solution for 5 min, further demonstrating that the amorphous near the crystal Si on the hillock structure was protective layer rather than a-Si (Fig. 8). As a result, scanning-induced subsurface defects including a-Si and deformed Si could be gradually eliminated by the anisotropic etching, yielding nondestructive nanostructures. Besides, the HRTEM image of etched surface from unscanned regions revealed that the etching process did not cause secondary lattice damage to Si substrates.

Figure 8
figure 8

AFM topography and cross-sectional lattice structure of fabricated nanostructures on Si surface. The left panel shows the overall cross-section HRTEM image and AFM topography of a hillock etched by HF/HNO3 mixtures for 2 s. The details of the yellow marked area are magnified in a–d

Full size image

The etching of Si in HF/HNO3 mixtures was reported as an electrochemical process [33]. During the etching, the Si surface was persistently terminated by the hydrogen groups. Oxidizing agents (i.e., HNO3) could provide holes (h+) for the initiation of dissolution reaction, allowing reactive species (e.g., HF or HF2) to attack superficial Si–H bonds. This process allowed the replacement of Si–H bonds by Si–F bonds. Subsequently, the polarized Si–Si back bonds were attacked by the reactive species and thereby resulted in the removal of superficial Si atoms from bulk substrates. The final step involved the complexation of SiF4 by HF to yield H2SiF6.

Based on the above results, a model was proposed to explain the anisotropic etching of friction-induced hillocks. As shown in Fig. 9a, b (Not drawn to scale), the scanning of Si surface by the AFM diamond tip transformed the monocrystalline structure of Si atoms in scanned regions into a-Si or deformed Si. Since the atomic arrangement of a-Si became out of order, the average dangling bond density in scanned regions was lower than in the original Si surface [34]. During the etching in HF/HNO3 mixtures, H atoms quickly attach Si dangling bonds and thereby resulted in the formation of Si–H termination. Correspondingly, the scanned regions presented lower Si–H bond densities (see Fig. 9c, d). Subsequent nucleophilic attacks of reactive species on Si–H bonds and polarization of Si–Si backbonds would be the rate-limiting steps, retarding the break out of Si–H bonds and Si–Si backbonds. This, in turn, slowed down the dissolution rate of Si atoms in the scanned regions, resulting in rapid increment of the height of hillock nanostructures on Si surface.

Figure 9
figure 9

a–b Schematic diagrams showing the scanning-induced deformation process along the sliding direction. c–d Dissolution process of a cross-section of the scanned region in HF/HNO3 mixtures

Full size image

Fabrication of nondestructive nanostructures on Si surface

By controlling tip scanning traces under close-loop AFM mode, various nondestructive nanostructures can be realized on Si surface through subsequent anisotropic etching in HF/HNO3 mixtures. An array of Si nanogratings with a period of 1.3 μm and a height of ~ 80 nm is illustrated in Fig. 10a. This array was fabricated at Fn = 40 μN in a line-scanning mode followed by post-etching for 2 s, and the height of asperities is less than ~ 5 nm. In Fig. 10b, a pentagram structure with a height of ~ 95 nm was observed. This structure was fabricated at Fn = 40 μN followed by post-etching for 2 s. In Fig. 10c, a spiral line structure with a height of ~ 90 nm was produced at Fn = 40 μN followed by post-etching for 2 s. Also, by designing the layout of friction-induced hillocks, a note structure with a height of ~ 90 nm was formed at Fn = 40 μN followed by post-etching for 2 s (Fig. 10d).

Figure 10
figure 10

A series of nondestructive Si nanostructures fabricated by the proposed approach: a nanogratings array, b a pentagram, c spiral line, and d a note

Full size image

In sum, the proposed approach possessed an excellent ability for rapid fabrication of nondestructive nanostructures. Also, it did not require additional fields or templates. In traditional SPM-based direct writing technology, a sharp diamond tip would be applied to produce groove-shaped scratch under high contact pressure at the interface between the tip and sample. Although direct scratching is characterized by high precision, it could be unstable due to probe wear during the scratching process. By contrast, the Pmax of ~ 2.1–9.8 GPa used during the preparation process of friction-induced hillocks was less than critical contact pressure (Py) of initial yield of Si, which can be estimated as ~ 11.3 GPa by Eq. (1):

$$P_{y} = \frac{1}{{0.31\left( {\tau_{c} } \right)_{\max } }} \le 1.61\delta_{y}$$
(1)

Note that Eq. (1) was derived from the relationship between principal shear stress (\(\tau_{c}\)) and Pmax (\(\tau_{c}\) = 0.31 Pmax), and Tresca yield criterion (\(\tau_{c}\)) max ≤ 0.5 \(\delta_{y}\). Here, yield stress (\(\delta_{y}\)) of monocrystalline Si was ~ 7 GPa. Therefore, friction-induced hillocks were produced before yield failure of Si material. Moreover, the scanning method could extend the lifetime of the probe due to the relatively low contact pressure.

On the other hand, anisotropic etchants were widely applied in Si bulk micromachining, while HF/HNO3 mixtures were considerable to be difficult in nanostructuring of Si substrates due to the absence of no preference for particular crystal planes. Fortunately, the friction-induced hillocks exhibited strong anti-etching abilities in HF/HNO3 mixtures, resulting in the formation of hillock nanostructures on Si surface. Moreover, diverse cross-sectional contours of fabricated hillock structures were flexibly realized, which can hardly be produced by anisotropic etchants etching due to their intrinsic etching selectivity. Furthermore, the fabricated hillock structures were defect-free, thereby potentially applying in the patterning of Si substrates for the growth of high-quality epitaxial layers. Also, further design and optimization of scanning tracks may prepare high-quality optical structures, such as diffraction gratings and circular gratings for use in photodetectors [35]. Hence, the proposed approach may open a new SPM-based nanofabrication paradigm when considering the above advantages and potential application value.

Conclusions

Fast and anisotropic etching method was successfully developed for the fabrication of high-quality nanostructures on Si surfaces to overcome the limitation of isotropic etching of Si in HF/HNO3 mixtures. Several affecting parameters were evaluated and the results were analyzed. The proposed method looks very promising for the fabrication of controllable nano-sized functional Si-based devices by HF/HNO3 mixtures. The following conclusions can be drawn:

  1. (1)

    The distinctive anisotropic etching in HF/HNO3 mixtures was significantly dependent on volume ratio, sliding cycle, normal load, and etching time. The mixtures with volume ratios from 1:100 to 1:5 presented excellent anisotropic etching abilities for the fabrication of hillock nanostructures with different cross-sectional profiles. As volume ratio rose, the fabrication height first gradually increased and then decreased, and the maximum height was observed at ratio 1:60 while the fabrication width remained stable. Comprehensive consideration of fabrication results and efficiency determined the most suitable sliding cycle N for anisotropic etching of scanned regions as 100. The fabrication height/width can be effectively controlled by normal load and etching time.

  2. (2)

    AFM topographies of a friction-induced hillock before and after being etched by HF and HF/HNO3 mixtures revealed that a-Si can resist HF/HNO3 mixtures etching, while deformed Si provided little contribution for rapid generation of hillock nanostructures. Moreover, HRTEM images displayed defect-free nanostructures.

  3. (3)

    Further analysis based on an electrochemical dissolution model suggested that lower average dangling bond density in scanned regions can retard the breakup of Si–H bonds and polarized Si–Si backbonds, leading to lower dissolution rates of Si atoms in the scanned regions.