Controlling the Nature of Etched Si Nanostructures: High- versus Low-Load Metal-Assisted Catalytic Etching (MACE) of Si Powders

(1)

UEF//eRepository

DSpace https://erepo.uef.fi

Rinnakkaistallenteet Luonnontieteiden ja metsätieteiden tiedekunta

2020

Controlling the Nature of Etched Si Nanostructures: High- versus

Low-Load Metal-Assisted Catalytic Etching (MACE) of Si Powders

Tamarov, Konstantin

American Chemical Society (ACS)

Tieteelliset aikakauslehtiartikkelit

http://dx.doi.org/10.1021/acsami.9b20514

https://erepo.uef.fi/handle/123456789/8390

Downloaded from University of Eastern Finland's eRepository

(2)

1

Supporting Information for

Controlling the nature of etched Si nanostructures:

High versus low load metal-assisted catalytic etching (MACE) of Si powders

Konstantin Tamarov^†, Joseph D. Swanson^‡, Bret A. Unger^‡, Kurt W. Kolasinski^‡*, Alexis T. Ernst^§, Mark Aindow^§, Vesa-Pekka Lehto^† and Joakim Riikonen^†

†Department of Applied Physics, University of Eastern Finland, Kuopio, Finland

‡Department of Chemistry, West Chester University, West Chester, PA, United States

§Department of Materials Science and Engineering, Institute of Materials Science, University of Connecticut, Storrs, CT, United States

*KKolasinski@wcupa.edu

This document is the unedited Author’s version of a Submitted Work that was subsequently accepted for publication in ACS Applied Materials & Interfaces, copyright © American Chemical Society after peer review. To access the

final edited and published work see [https://pubs.acs.org/articlesonrequest/AOR-wzdFn9sVIzrXmjwTqJHd].

(3)

2

Section S1. Experimental.

S1.1. Calculation of Ag layer thickness.

To calculate the thickness of the Ag layer, we assumed that all the injected Ag from the AgNO3

solution was deposed uniformly in the metallic form on the surface of the particles. Therefore, the total volume of metallic Ag is:

𝑉_"# =𝑚_"#

ρ_"# = ν𝑀

ρ_"#,

where ν, M, and ρAg are number of moles, molar mass, and density of Ag, respectively. Then, the thickness of Ag layer is:

ℎ = 𝑉_"#

𝐴_,-𝑚_,-,

where mSi is the mass of Si powder and ASi is the surface area per mass unit of Si powder, which was measured by N2 sorption (Section S1.4).

Therefore, the thickness of Ag layer is:

ℎ =ν𝑀

ρ_"#⋅ 1

𝐴_,-𝑚_,-.

The measured value of surface area was 0.42 m²g^–1 for 44–75 µm Si powder. With these assumptions, the thickness of Ag layer would be 118 nm (~288 atomic layers) or 0.61 nm (1.5 atomic layers) for 4.8 mmol or 0.025 mmol, respectively. Note, that Ag never deposits uniformly on Si. Instead it forms nanoparticles, clusters and/or dendrites.

(5)

4

S1.2. Metal-assisted catalytic etching (MACE) of powders.

Warning: This procedure involves use of HF and a highly exothermic reaction of a strong oxidant with Si powder. Appropriate safety measures and labware must be used to deal with the toxic and corrosive nature of HF. Appropriate thermal management must be employed to dissipate the heat generated by the reaction.

MACE of Si powders has been made in the West Chester University (WCU) and the University of Eastern Finland (UEF) with reagents from different sources, e.g., AgNO3 (Fisher ACS reagent, Alfa Aesar 99.9+ %), HF (Acros Organics 49 %, Merck empula® 38−40 %), H2O2 (Acros Organics 35 % wt.), HNO3 (Fisher ACS reagent, 65 %, Merck emsure®) and glacial CH3COOH (Fisher ACS reagent, Merck, VWR Chemicals). The Si powders were made by milling metallurgical grade (MG) Si particles (Elkem, 99.6 % and 99.999 % purity, polycrystalline), electronics grade (EG) wafer chunks (Union Carbide), p+ and undoped Si wafers of resistivities 10–20 mΩ∙cm and > 100 Ω∙cm (single crystal, Okmetic). Distilled water was used in WCU and deionized water was used in UEF.

Particle sizes in the powders were 2–44 µm, 44–75 µm, 11−25 µm and 2–10 µm. All the powders were washed in ~ 15 g batches by 1-hour sonication in 230 ml of water mixed with 20 ml of 35 wt%

hydrogen peroxide to remove organics. The removal of organics and small embedded particles is a crucial step to obtain uniformly etched Si particles. After the washing, particles were filtered out using a Büchner style funnel and dried at 50 ^oC in a vacuum oven.

For a typical etch, 1.005 g (±0.005 g) or 0.036 mol of Si powder was first weighed in a PTFE cup.

Then, 10 ml of acetic acid was added to the powder, and the suspension was sonicated to break the particle aggregates formed after drying. Next, 20 ml of HF was added to the particles and the suspension was stirred in an ice bath until equilibrium was reached (no bubbles evolving). After that, different volumes of 0.6 M or 0.006 M AgNO3 water solution were injected using a syringe pump.

Digital syringe pumps Legato SPLG100 and World Precision Instruments AL-1000 were used in WCU and UEF, respectively. The 0.6 M AgNO3 was used for etching Si nanowires by typical MACE procedure. The 0.006 M AgNO3 was used for low load MACE to prepare porous Si. The time of AgNO3 injection was varied to keep the injection rate constant. Typical injection rates were 0.17 ml/min and 0.28 ml/min for high and low concentration AgNO3 solutions, respectively. For the two lowest amounts of Ag, the solution was added by dropwise pipetting without the use of a syringe pump. The tube between the syringe and the particle suspension was typically not immersed into the liquid, i.e. the AgNO3 solution was added dropwise. During the injection of AgNO3 solutions, particles were stirred using a magnetic stirrer. After the AgNO3 injection was finished, particles were further stirred for 5 min to complete the Ag deposition.

The final and etch-inducing step is the injection of H2O2 solution. The required volume of 35 wt%

H2O2 was calculated based on the value of H2O2/Si molar ratio, and the solution was subsequently diluted with water to have the total volume of 6 ml. For example, for H2O2/Si molar ratio of 0.9 and 1 g of Si, volume of 35 % H2O2 was 3.2 ml, which was subsequently diluted with 2.8 ml of H2O to obtain total volume of 6 ml. The H2O2 injection time was 30 min for the Ag concentration dependence experiments and was increased to 90 min for the further studies at low Ag load to avoid excess of heat. During the H2O2 injection, particle suspension was stirred and the tube from syringe was immersed into Si particle suspension to provide steady supply of the oxidant. After injection was completed, the particle suspension was further stirred for 10 min to allow the rest of H2O2 to react.

After completion of the etch process, for high load MACE case etched particles sediment easily due to high Ag amount. Therefore, the etching solution was decanted, and particles were rinsed with water

(6)

5

three times. After that, Ag was removed by dissolution in HNO3. The final wash was performed using a Büchner style funnel. The appropriate filter was selected to retain the particles (8 µm filter for particles ≥ 11µm and 1–2 µm filter for particles ≤ 10 µm). Usually, the filtered solution was clear indicating, that no particles passed through the filter and that the etching did not significantly decrease the particle size. While in the filter funnel, the particles were rinsed with water, then pentane. Ag was removed by dissolution with HNO3. The sample was then either dried at 55 ^oC in an oven or at ~40

oC in a vacuum oven.

In the case of low load MACE (Ag amount < 1 mmol), the particles after etching do not sediment due to hydrophobicity and porous structure. Thus, it is not possible to decant HF and rinse particles without losing substantial amount of them. Therefore, the particles are first filtered from HF, rinsed with water and pentane, dried and washed as discussed previously. We denote the low load MACE as LL-MACE.

S1.3. Electron microscopy.

For SEM examination, Si powders were deposited onto carbon tape and mounted onto Al SEM stubs for direct examination of the surface morphologies using an FEI Quanta 400 SEM, a Zeiss Sigma HD VPSEM, and an FEI Verios 460L HRSEM.

The cross-sectional imaging was performed using both an FEI Helios Nanolab 460F1 dual-beam Ga⁺ focused ion beam (FIB) and an FEI Helios Xe⁺ plasma FIB (PFIB). Selected larger particles were sectioned using the Xe⁺ PFIB, and the cross-sectional morphology was examined using the electron column. For STEM analysis, specimens were produced using the Ga⁺ FIB. A protective buffer layer was deposited onto the surfaces of the Si particles in situ to protect the surfaces during ion milling.

This deposition was accomplished using both the electron beam and the ion beam to crack the organometallic precursor. During the milling process, the accelerating voltage used for the ion column was 30 kV and the ion beam currents were reduced iteratively to minimize beam damage. Analysis on the FIB-cut specimens was performed using a FEI Talos F200X operating at 200kV in STEM mode to acquire both bright field and high-angle annular dark field (HAADF) images. In addition, a Super-X silicon drift detector energy-dispersive X-ray spectrometry (EDXS) system on the Talos was used to collect spectrum images from each FIB-cut TEM specimen. In each case, EDXS spectra were acquired by scanning over the region of interest in a grid of 512 x 512 points repeatedly for 5 min, and X-ray maps were then extracted from these data using the intensities in the O, Si & Ga K-peaks, the Ag L-peak, and Pt M-peak.

TEM imaging of the detached nanostructures was performed using JEOL JEM-2100F instrument.

Prior the measurements, approximately 4 mg of particles were placed into 2 ml Eppendorf tubes containing 1 ml of ethanol. The suspensions were then sonicated for 1 hour in an ultrasound bath and centrifuged at 1000 g to separate the big particle cores. The supernatant was then diluted to have slightly colored suspensions, which were then used for TEM imaging. 2.5 µl of suspension was dropped onto the 400 mesh holey carbon film (Cu, Agar Scientific) and dried for 15 min under a lamp. Finally, the films were placed in the instrument for the imaging.

S1.4. Structural characterization.

The specific surface area (SSA), total pore volume and pore size distribution of the Si samples were determined with N2 sorption measurements at –196 °C using a Micromeritics TriStar II instrument.

The SSA was calculated from the adsorption isotherm using BET analysis; pore volume was obtained from the total adsorbed amount at a relative pressure of 0.97; pore size distribution was calculated from the desorption isotherm using BJH theory with a home-written script.

(7)

6

Raman measurements were performed using a Thermo Scientific™ DXR™2xi Raman Imaging Microscope with 50x objective, 795 nm excitation laser and fine grating (400–1800 cm^–1). The power of the laser was 0.2 mW, the scan frequency was 200 Hz and number of scans was 200, i.e. one measurement took 1 s. Such low power and high scan frequency were selected to avoid heating of the sample and the influence of the heating on Si peak position. The Si peak amplitude for each measurement was maximized by adjusting the focus. For each sample, a layer of particles was tightly glued on a quarter of SEM carbon tape, which itself was glued to a glass microscope slide. The glass slide with particles was places into a motorized holder and Raman spectra were acquired. A total of four spectra were taken per sample at four different locations. Before starting the measurements for the samples, the Si peak position was measured for a piece of undoped Si wafer. The peak position of the wafer was then adjusted to 520.5 cm^-1 and the calibration constant was subtracted to obtain the correct peak positions for the samples. The calculation of nanocrystal sizes from the Raman peak position was done using the phonon confinement model.¹ Note, that this model can be applied to calculate the sizes of small nanocrystals and does not allow reliable estimation of crystallite size for e.g. initial Si powder.

To estimate the size of larger crystallites, X-ray powder diffraction measurements were performed using a Bruker D8 Discover instrument with Cu tube. The samples were placed on the zero- background holder and measured for 25–117° two theta range with the step size of 0.0057° and time per step of 0.205 s (total time of measurement ~1 hour). The crystallite sizes were then estimated using Rietveld refinement method by fitting the XRPD spectra with two Si phases and one Ag phase if all Ag was not washed out in several high load MACE samples (Ag crystallite sizes are not shown).

TOPAS® 4.6 software was used to fit the full spectrum, calculate crystallite sizes and errors.

(8)

7

Section S2. Effect of deposited Ag amount.

S2.1. SEM and STEM of deposited Ag before and at the beginning of etching.

This section contains the supplementary SEM and STEM data from deposited Ag nanoparticles before etching (H2O2 injection) and after 2 min of etching. Figure S2.1 supplements Figure 1 with high-angle annular dark-field (HAADF) STEM images and energy-dispersive X-ray spectroscopy (EDXS) maps of cross sections of Si particles after deposition of 4.8 mmol and 0.025 mmol Ag. At high Ag load, both big and small Ag nanoparticles were formed, which started to descend into Si during the deposition. Large dendrites (not present on Figure S2.1a) stayed on the surface of Si and did not descend. In case of low Ag deposition, much finer Ag nanoparticles were formed. They also descended into Si surface but stayed in a narrower near-surface region of ~60 nm in thickness. See the main text for further discussion and Figure 1 for general-view SEM images.

After 2 min etch at high Ag load (4.8 mmol), the smaller Ag nanoparticles descended deeper into Si surface, up to 800 nm, while larger particles and dendrites remained on the surface (Figure S2.2). In some places on the Si surface, a reorganization of larger Ag particles into lines was observed. At these places, tips of Si NW-like structures were preferentially formed (Figure S2.2c,f). At the places, where there was no reorganization, more ridge-like structures were etched (Figure S2.2b,e). Some larger particles started to change their shapes, for example Ag particles > 200 nm across on their short axis but larger > 500 nm on their long axis were found (Figure S2.4).

Eventually, most of Si particles after MACE with 4.8 mmol of Ag end up looking similar to each other with long etch track pores formed by the Ag nanoparticles (Section S2.2). Therefore, there must be further restructuring of Ag into 70–100 nm nanoparticles subsequent to 2 min of etching. These larger Ag nanoparticles then moved cooperatively to produce parallel etch track pores, which can be exfoliated to form Si NWs. That the catalyst motion is cooperative after prolonged etching is evinced by the presence of etch track pores that are not only parallel but also etched to the same depth. This would not happen if the metal nanoparticles etched along uncorrelated paths.

In the low Ag load, much less correlated motion of Ag nanoparticles was observed (Figure S2.3 Figure S2.4). Initial particles size of 10−30 nm did not change after 2 min etch and the particles were found 250−300 nm deep in Si. There was no evidence of high surface roughness and the initial Si surface remained flat. The lack of cooperativity in catalyst motion is evinced by the presence of non- parallel etch track pores of random orientations and lengths after prolonged etching.

(9)

8

Figure S2.1. HAADF STEM and EDXS maps of the cross sections of Si particles after deposition of (a) 4.8 mmol Ag and (b) 0.025 mmol Ag. The scale bars are (a) 100 nm and (b) 50 nm.

(10)

9

Figure S2.2. SEM (a–c, e, f) and STEM (d) images of MG Si particles after 2 min etch in the high Ag load limit (4.8 mmol). (a) general view of a Si particle; (b), (e) ridge-like structures formed during etching; (c) reorganization and alignment of big Ag particles; (d) cross-sectional HAADF images of etched tracks; (f) Si NWs-like structures and Ag nanoparticles on tips of Si NWs.

(11)

10

Figure S2.3. SEM (a–c, e, f) and STEM (d) images of MG Si particles after 2 min etch in the low Ag load limit (0.025 mmol). (a) general view of a Si particle; (b), (c) backscatter SEM images of Ag nanoparticles and pores; (d) cross-sectional HAADF images of etched tracks; (e), (f) secondary electron SEM images of Ag nanoparticles and pores.

(12)

11

Figure S2.4. HAADF STEM images and EDXS maps of the cross sections of Si particles after 2 min MACE with (a) 4.8 mmol Ag and (b) 0.025 mmol of deposited Ag. The scale bars are (a) 300 nm and (b) 200 nm.

(13)

12

S2.2. SEM imaging and pore size distributions of etched Si particles.

Metal-assisted catalytic etching (MACE), presented in this section, was performed using 44–75 µm metallurgical grade (MG) polycrystalline Si powders (99.6 %, Elkem Silicon Materials). The silver nitrate solution injection time was varied to keep the injection rate constant. The injection rates were 0.228 ml min^–1 for high load Ag (0.6 M AgNO3 solution) and 0.417 ml min^–1 for low load Ag (0.006 M AgNO3 solution). The etching time was 30 min.

It was found that, depending on the amount of injected Ag into the suspension of particles, we can obtain nanostructures with completely different morphologies. High Ag load (> 2 mmol) etched particles presented distinct etch track pores created by cooperative motion among the Ag nanoparticles (Figure S2.5). The walls of these etch track pores exfoliated to form silicon nanowires (Si NWs). Numerous particles were fully etched to their cores. The Si NWs exhibited some mesoporosity, which was most pronounced for 4.8 mmol Ag (Figure S2.7, see also Section S3 and Section S4 for pore size distributions).

When the amount of Ag was decreased to 1–2 mmol, the unetched surfaces of particles were observed. Particles etched with 0.1–0.5 mmol Ag, did not show regular thin Si NWs. Instead, they were etched unevenly, had low surface area (Figure 3 of the publication) and no pores (Figure S2.7).

Further decrease of Ag amount to below 0.1 mmol revealed a complete change in the morphology of the etched particles (Figure S2.5). These low Ag load particles presented a random texture of fine pores. In most cases, we observed a bimodal pore size distribution with pore sizes of 2–8 nm and 20–

50 nm. Such a pore size distribution suggests, that pores could be formed both by uncorrelated motion of much smaller Ag nanoparticles and/or by etching that was induced by charge transfer to adsorbing Ag⁺ ions that do not aggregate into Ag nanoparticles. To distinguish between such different morphologies, we define high load MACE (HL-MACE, n(Ag) ≥ 1 mmol) and low load MACE (LL- MACE, n(Ag) ≤ 0.05 mmol).

Pore size distributions for particles etched with different Ag loads are presented on Figure S2.7.

Clearly, there is a transition from regular etch track pores to a mesoporous structure with no mesoporosity in the transition region of 0.5–0.1 mmol. Interestingly, etch track pore walls are not porous at 2–3 mmol Ag, while at 4.8 mmol, the 2–20 nm mesopores are present for all H2O2/Si molar ratios (Section S3) and etching times (Section S4).

Note, that the transition point is determined not only by the Ag amount, but also by the ratio between the total surface area of powder and the Ag amount (see main text for values of this ratio). For example, if the powder consists of smaller particles, its total surface area will be higher, and therefore, it is expected, that the transition amount of Ag to obtain porous particles will also be higher. However, the high amount of Ag will also consume H2O2 faster and thus prevent formation of deep pores and the particles will not be etched through. On the other hand, big Si particles are not etched through as easily as is shown on Figure 2. In summary, we found that we could fully porosify particles below 25 µm by depositing 0.01–0.03 mmol of Ag per gram of Si and 90 min H2O2 injection time.

(14)

13

Figure S2.5. SEM images of MG Si microparticles after MACE with different amount of deposited Ag. H2O2/Si molar ratio was 0.9, etching time was 30 min, Ag nucleation time was varied to keep the same injection rate of Ag solution. Scale bar is 10 µm.

(15)

14

Figure S2.6. SEM images of MG Si microparticles after MACE with different amount of deposited Ag. H2O2/Si molar ratio was 0.9, etching time was 30 min, and Ag nucleation time was varied to keep the same injection rate of Ag solution. Scale bar is 20 µm.

(16)

15

Figure S2.7. Dependence between pore size distribution and the deposited Ag amount. H2O2/Si molar ratio was 0.9, etching time was 30 min, Ag nucleation time was varied to keep the same injection rate of Ag solution. For the Ag amount of 0.025 mmol and higher, HNO3 wash was performed after the etching.

(17)

16

Figure S2.8.Pore size distribution of MG Si powder after MACE of 30 min (red) and 180 min (gray). The amount of deposited Ag was 0.001 mmol and Si/H2O2 molar ratio was 0.9, MG-Si particle size was 44–75 µm.

(18)

17

S2.3. TEM imaging of the detached Si nanostructures.

To get a closer look at the etched nanostructures, we detached them by agitation for 1-hour in an ultrasonic bath. Figures Figure S2.9–Figure S2.12 are TEM images of nanostructures obtained from the HL- and LL- MACE of MG and EG powders (44–75 µm). The TEM images are consistent with the SEM images. At high Ag loads, Si NWs are cleaved by sonication from the walls of the parallel etch track pores that were formed by the cooperative motion of big Ag nanoparticles. At low loads, sonication releases jagged nanoparticles with mesopores that were formed by the randomly moving much smaller Ag nanoparticles or by remote etching.

The thicknesses of the Si NWs formed by HL-MACE were in the range of 90–150 nm for both MG and EG powders, and these values were consistent with the crystallite sizes obtained by X-ray powder diffraction (XRPD) (see Section S2.4).

Figure S2.9. TEM images of nanowires obtained by 1-hour sonication of MG Si powder etched with 4.8 mmol Ag (HL-MACE). Other etching parameters are: 0.9 Si/H2O2 molar ratio, 40 min Ag nucleation time and 30 min etching time.

(19)

18

Figure S2.10. TEM images of nanoparticles obtained by 1-hour sonication of MG Si powder etched with 0.025 mmol Ag (LL-MACE). Other etching parameters are: 0.9 Si/H2O2 molar ratio, 15 min Ag nucleation time and 30 min etching time.

(20)

19

Figure S2.11. TEM images of nanoparticles obtained by 1-hour sonication of EG Si powder etched with 3 mmol Ag (HL-MACE). Other etching parameters are: 0.9 Si/H2O2 molar ratio, 27 min Ag nucleation time and 30 min etching time.

(21)

20

Figure S2.12. TEM images of nanoparticles obtained by 1-hour sonication of EG Si powder etched with 0.025 mmol Ag (LL-MACE). Other etching parameters are: 0.9 Si/H2O2 molar ratio, 15 min Ag nucleation time and 30 min etching time.

(22)

21

S2.4. Analysis of crystallite sizes of etched nanostructures.

The crystallite sizes of the etched powders were analyzed using Raman spectroscopy and XRPD. In Si nanostructures, phonon confinement leads to a shift of the Si Raman peak to smaller wave numbers and this shift can be related to the crystallite size¹. Figure S2.13 shows the Raman spectra of initial MG Si powder and powders after MACE with different amount of Ag. The shift of the peak value becomes more pronounced with the increase of Ag. Figure S2.14 summarizes the values of the peak shifts from the bulk Si value (520.5 cm^-1) and the calculated crystallite sizes using the phonon confinement model¹. The HL-MACE samples typically presented smaller crystallites and their size increased with the decrease of Ag amount. The introduction of the Si nanocrystallites on the surface and mesoporosity of Si NWs can be explained by two processes: remote etching (etching with holes diffused away from the Ag/Si interface) and etching by charge transfer to Ag⁺ ions, that are not agglomerated to form nanoparticles.

In addition to Raman spectroscopy, which mainly provided information on the sizes of the crystallites located on the surfaces, we used XRPD for the analysis of crystallite sizes throughout the particles.

The XRPD diffractograms were fitted with two phases of Si to estimate bigger and smaller crystallite sizes present in the sample and one Ag phase if it was not fully removed with nitric acid (data for Ag sizes not shown). The diffractograms of powders, etched with high Ag load, were typically well fitted with the model of two distinct crystallite sizes (Figure S2.15a). The bigger size was attributed to the diameter of the Si NWs and was in the range of 80–120 nm coinciding with the TEM data. The smaller sizes were similar to the Raman data for high and low Ag, but in the intermediate region, the small crystallite sizes obtained from XRPD were relatively large, on the order of 20 nm or completely absent. This result correlates well with the N2 sorption data (Figure 3, Figure S2.7), which indicated no porosity and very low surface area for these samples. When the Ag amount was less than 0.01 mmol, no smaller crystals were observed. Clearly, with this small Ag amount, not all of the H2O2

was consumed during the 30 min etch, the particles were only partially etched, and the diffractogram mainly consisted of the signal from the unetched cores. Similarly, the calculation of crystallite size from Raman data indicated the presence of only big crystallites, for which the phonon confinement model was not fully applicable any more.

(23)

22

Figure S2.13. Selected Raman spectra for MG Si powders etched with different amount of deposited Ag. Other etching parameters are: 0.9 Si/H2O2 molar ratio and 30 min etching time. Ag nucleation time was varied to keep the same AgNO3 solution injection rate.

(24)

23

Figure S2.14. (a) The dependence between Si Raman peak shift of MACE etched MG Si powder and amount of deposited Ag; (b) calculated crystallite size from the peak shift according to the phonon confinement model¹. The Raman peak was normalized to the undoped Si wafer value.

(25)

24

Figure S2.15. (a) Dependence between crystallite size nanostructures on the MG Si microparticles after MACE and different amount of deposited Ag; (b,c) selected XRPD patterns presented with full range and narrower range, respectively. H2O2/Si molar ratio was 0.9, etching time was 30 min, MG Si particle size was 44–75 µm. Ag injection time was varied to keep constant AgNO3 solution injection rate. The crystallite sizes were determined by Rietveld refinement method by fitting the XRPD spectra with two Si phases (denoted as smaller and bigger size on (a)) and one Ag phase if not all Ag was washed out. Some of the samples had both Si phases, while others did not.

(26)

25

Section S3. HL-MACE: Effect of H

2

O

2

/Si molar ratio.

There are many parameters that may affect the outcome of etching. The results in this section describe the effect of the injected H2O2 amount. We present the data as the H2O2/Si molar ratio and use 30 min injection time and 6 ml of H2O2/water solution for all samples. For example, for H2O2/Si = 0.5, the volume of 35 % H2O2 was 1.54 ml and the volume of water was 4.46 ml. Thus, the amount of H2O2

injected per second during the etching was different.

The dependence between the yield and the H2O2 amount was found to be linear (Figure S3.1a), which was expected, since the H2O2 is consumed during the reaction to remove Si under the Ag nanoparticles. The higher amount of H2O2 thus etched out more Si and produced deeper etch track pores. However, the N2 sorption and XRPD results for molar ratio of 1.0 were found to be rather different. As can be seen from Figure S3.1, this sample has much higher surface area and pore volume, and smaller crystallite size. We attribute these differences to the higher concentration of H2O2 during the etching, which can induce more active dissolution of Ag nanoparticles and promote the remote etching of mesopores.² Also, as the concentration of H2O2 increases, the etching gets closer to the electropolishing regime and etches out more Si resulting in thinner walls of etch track pores.

Figure S3.2 shows the SEM images of MG Si powder after HL-MACE with four different H2O2/Si molar ratios. At the ratio of 0.5, unetched parts are visible, which disappeared with the increase of H2O2 amount. At the highest H2O2 amount, no core was left, and the Si NWs were no longer present as ordered arrays but started to break off.

All the etched Si particles were found to have mesopores (Figure S3.3) with pore sizes of 4–20 nm.

The presence of mesopores was attributed to dissolution and redeposition of Ag, remote etching and participation of impurities in the etching (see also the main text for additional discussion).

(27)

26

Figure S3.1. The dependence between (a) yield, (b) XRPD crystallite size, (c) BET surface area, (d) pore volume, (e) Raman peak shift, and (f) Raman crystallite size and the injected H2O2 amount. The x axis shows the H2O2/Si molar ratio. The amount of deposited Ag 4.8 mmol, Ag nucleation time was 40 min, etching time was 40 min, MG-Si particle size was 44–75 µm.

(28)

27

Figure S3.2. SEM images of 44–75 µm MG Si powders etched with different amount of H2O2. Ag amount was 4.8 mmol, nucleation time was 40 min, etching time was 30 min.

(29)

28

Figure S3.3. Dependence between pore size distribution and the H2O2/Si molar ratio for 4.8 mmol deposited Ag amount. Ag nucleation time was 40 min, etching time (H2O2 injection time) was 40 min, MG-Si particle size was 44–75 µm.

(30)

29

Section S4. HL-MACE: Effect of etching time.

To change the concentration of H2O2 during the etching, we fixed the H2O2/Si molar ratio to 0.9 and varied the H2O2 injection time from 10 min to 75 min (Figure S4.1). The sample, produced with the shortest injection time had a low yield, surface area and pore volume. Substantial unetched parts were also observed (see also Figure S4.3). This could happen due to too high a concentration of H2O2, which resulted in high reaction temperature and approaching the electropolishing region.

When the etching time was slightly increased, a significant increase in surface area and pore volume was observed, which was attributed to the etching of mesopores in the walls of etch track pores as discussed in Section S3. The further increase of etching time up to 40 min resulted in the gradual decrease of yield and pore volume, while the surface area first increased and then decreased (Figure S4.1a,c,d). With the increase of etching time several variables change, which explain the observed trend. First, the slower injection rate translates into a lower steady-state concentration of H2O2

resulting in the decrease of side reactions and temperature. Thus, more H2O2 was used to remove Si under the deposited Ag nanoparticles and the depth of etch track pores and surface area of their walls increased while the yield decreased. Second, the remote etching of mesopores decreased, which eventually led to the decrease of surface area when etching time was more than 30 min.

The further increase of etching time revealed the large variation of the parameters, which is not understood yet. However, it should be noted that N2 adsorption is most sensitive to mesoporosity and is not as useful for pore volume determination for macropores above 50 nm in size. In addition, the non-ideal low value of the intercept for the yield versus molar ratio clearly indicate the presence of side reactions. The effect of these side reactions is likely masking more regular trends in the data.

SEM images of 10 and 15 min etched particles clearly present surfaces without etch track pores (Figure S4.2). The nanostructures were either cleaved during the etching or were completely etched out. Furthermore, the pores were not etched close to the core of the particles, which was observed with the increase of H2O2 injection time. All the samples were found to have mesopores for the similar reasons as discussed before (Figure S4.3).

(31)

30

Figure S4.1. The dependence between (a) yield, (b) XRPD crystallite size, (c) BET surface area, (d) pore volume, (e) Raman peak shift, and (f) Raman crystallite size and the etching time (i.e. H2O2

injection time). The amount of deposited Ag was 4.8 mmol, Ag nucleation time was 40 min, H2O2/Si molar ratio was 0.9, MG-Si particle size was 44–75 µm.

(32)

31

Figure S4.2. SEM images of 44–75 µm MG powders after HL-MACE with different etching time.

The amount of deposited Ag was 4.8 mmol, Ag nucleation time was 40 min, H2O2/Si molar ratio was 0.9.

(33)

32

Figure S4.3. Dependence of pore size distribution on the etching time for 4.8 mmol deposited Ag amount. Ag nucleation time was 40 min, H2O2/Si molar ratio was 0.9, MG-Si particle size was 44–

75 µm.

(34)

33

Section S5. LL-MACE: Effect of etching time.

The effect of Ag amount was studied with 30 min H2O2 injection time. However, the high injection rate can result in the increase of etching solution temperature and decrease of the depth of etched pores. To evaluate the effect of etching time on LL-MACE, the injection time was varied from 30 min to 2 hours. It was found, that yield, surface area and pore volume did not vary significantly for etching times of 60 min and higher (Figure S5.1), while the 30 min sample had lower yield and higher pore volume. The difference for the 30 min etched sample could be due to higher temperature during fast H2O2 injection (see also Section S8), which enhanced Ag particle movement and induced higher dissolution and redeposition of Ag⁺ ions. The latter could result in the highest number of ~4 nm pores among all samples (Figure S5.2). On the other hand, the XRPD data demonstrated the lowest crystallite size for 90 min etch indicating the possibility of more uniform etching that left smaller Si crystals.

We have also performed etching with a very low amount of deposited Ag: 0.001 mmol. In this case, clearly not all H2O2 was consumed during 30 min injection, and even after 180 min injection, bubbles were forming in the solution indicating that etching had yet to cease. The increase from 30 min to 180 min of H2O2 injection resulted in the increase of surface area from 23.3 m² g^–1 to 55.2 m² g^–1 and pore volume from 0.063 cm³ g^–1 to 0.186 cm³ g^–1. Both volume of ~4 nm and 10–20 nm pores increased (Figure S2.8). Furthermore, pores with sizes 20–50 nm appeared. The latter observation might indicate the aggregation of Ag particles and/or formation of dendrite-like clusters that are visible on SEM (for example on Figure S7.3f or Figure S8.2b).

(35)

34

Figure S5.1. The dependence between (a) yield, (b) XRPD crystallite size, (c) BET surface area, (d) pore volume, (e) Raman peak shift, and (f) Raman crystallite size and the etching time (i.e. H2O2

injection time). The amount of deposited Ag was 0.025 mmol, Ag nucleation time was 15 min, H2O2/Si molar ratio was 0.9, MG-Si particle size was 44–75 µm.

(36)

35

Figure S5.2. Dependence between pore size distribution and the etching time for 0.025 mmol deposited Ag amount. Ag nucleation time was 15 min, H2O2/Si molar ratio was 0.9, MG-Si particle size was 44–75 µm.

(37)

36

Section S6. LL-MACE: Effect of H

2

O

2

/Si molar ratio.

The study of low Ag load MACE was done with 2–44 µm MG Si particles. The use of smaller size is necessary to obtain particles, that are etched through completely, because the etched layer in LL- MACE is on the order of a few micrometers thick (Figure 2). The yield was found to decrease linearly with the increase of H2O2/Si molar ratio, and the intercept was 97 ± 4 %, within uncertainty equal to the ideal 100 % value of unetched powder (Figure S6.1a). Surface area and pore volume increased linearly; growth of the pore volume was consistent with the decrease of yield and followed the linear increase trend with close to zero intercept value. The crystallite sizes decreased with the increase of H2O2/Si molar ratio and small crystals of ~ 5 nm were observed when the ratio was 0.6 or higher corresponding to the increase of the number of ~ 4 nm pores (Figure S6.2).

SEM imaging revealed the smoothening of the particles’ edges with the increase of H2O2/Si molar ratio (Figure S6.3), which however did not decrease particle sizes significantly, as no particles passed through the filter during washing.

See also main text for more detailed discussion of the effect of H2O2 amount on LL-MACE.

(38)

37

Figure S6.1. The dependence between (a) yield, (b) XRPD crystallite size, (c) BET surface area, (d) pore volume, (e) Raman peak shift, and (f) Raman crystallite size and the injected H2O2 amount. The x axis shows the H2O2/Si molar ratio. The amount of deposited Ag was 0.025 mmol, Ag nucleation time was 20 min, etching time was 90 min, MG Si particle size was 2–44 µm.

(39)

38

Figure S6.2. Dependence between pore size distribution and the injected H2O2 amount for 0.025 mmol deposited Ag. Ag nucleation time was 20 min, etching time was 90 min, MG Si particle size was 2–44 µm.

(40)

39

Figure S6.3. SEM images of 2–44 µm powders etched with different amount of H2O2. Ag amount was 0.025 mmol, nucleation time was 20 min, etching time was 90 min.

(41)

40

Section S7. MACE of particles of different grades and sizes.

We performed MACE (0.025 mmol Ag) of powders made from different grades of Si chunks and wafers: reclaimed electronics grade (EG) chunks milled to 44–75 µm, 99.999 % pure polycrystalline 2–10 µm (ESMC10) and 11–30 µm (ESPS30) particles, 11–25 µm powder obtained by milling doped p+ (DW) and undoped (UW) wafers.

Figure S7.1 shows the EG Si particles after HL (3 mmol) and LL (0.02 mmol) MACE. The morphologies of the etched nanostructures demonstrate a similar change from correlated etch track pore to random porous structure, as in the case of MG powder, with decreasing Ag load. Surface area (40.1 m² g^–1) and pore volume (0.17 cm³ g^–1) of low Ag load etched samples exhibit similar values to those of MG Si powder. On the other hand, the pore size distribution presents a more pronounced bimodal distribution with two distinct pore sizes (Figure S7.2b) for highly doped EG and less pure MG powders (Figure S7.4).

All different grades of powders were successfully etched by LL-MACE and presented similar exterior surface morphologies (Figure S7.3). Table S7.1 summarizes the results of etching the powders. Note, that after first using a 30 min etch duration, subsequently we switched to 90 min etch duration to reduce heating (see also Section S5 for the effect of etching time and Section S8 for the effect of etching temperature) and to allow for a deeper, more complete etch.

The grade of Si powder had a significant effect on the outcome of LL-MACE. In the pore size distribution, the peak centered about 4 nm is only observed in 99.6 % metallurgical grade Si (MG), low resistivity reclaimed electronics grade Si powder (EG) and powder from a p+ wafer (DW) (Figure S7.4). More pure 99.999 % MG powders (ESPS30) and powder from undoped wafers (UW) exhibited only a broad peak covering 7–30 nm pores. The disappearance of smaller pore sizes for high purity Si powders supports the existence of two distinct etching processes that are involved in the LL- MACE. The first process occurs regardless of doping level and leads to larger mesopores. This is the result of etching in the vicinity of Ag nanoparticles of roughly 7–30 nm that deposit on the Si surface.

Thereafter, these nanoparticles progress along random and uncorrelated paths to form random and uncorrelated 7–30 nm etch track pores. This mechanism is analogous to that of high load MACE;

however, the small particle size and/or lower density of Ag nanoparticles present in LL-MACE results in random uncorrelated etch track pores rather than the highly correlated parallel etch track pores observed after HL-MACE. The mechanism resulting in the etching of smaller pores is less clear. It correlates with the level of impurity and/or doping. It is induced by some combination of the etching of impurities, remote etching from holes that diffuse away from the Ag/Si interface, and etching initiated by adsorption of Ag⁺ ions that do not coalesce into nanoparticles.

Both particle size and grade of Si influenced the yield, surface area and porosity of etched powders (Figure S7.5a−c). The presence of 4 nm pores resulted in the increase of the surface area (highest for 11–25 µm MG and lowest for UW powder). However, since the etch depth was only 5–10 µm from the surface (Figure 2), and due to 7–30 nm pores, the pore volume was the highest in 2‒10 µm ESMC10 sample and, we suppose, that this sample was etched through.

(42)

41

Figure S7.1. SEM images of EG Si microparticles after MACE with (a, b) 3 mmol Ag and (c, d) 0.02 mmol Ag. Particle sizes were 44–75 µm, Si/H2O2 molar ratio was 0.9, etching time was 30 min, Ag nucleation time for high and low Ag concentrations was 40 min and 15 min, respectively.

(43)

42

Figure S7.2. Pore size distributions for EG Si particles etched with (a) 3 mmol Ag and (b) 0.02 mmol Ag.

(44)

43 Sample Particle

size, µm

Etch time, min

Yield, % BET surface area, m²/g

Pore volume, cm³/g

MG 44–75 30 44.4 58.1 0.235

EG 44–75 30 37.9 40.1 0.177

MG 11–25 90 30.0 85.4 0.244

DW 11–25 90 21.0 65.2 0.134

ESMC10 2–10 90 19.6 60.8 0.359

ESPS30 11–30 90 66.4 33.7 0.177

UW 11–25 90 25.4 22.4 0.121

Table S7.1. Summary of the yields, surface areas and pore volumes for powders, milled to different sizes from different Si grades. MG – metallurgical grade Si (99.6 %, polycrystalline); EG – electronics grade reclaimed wafer chunks; ESMC10 – metallurgical grade Si powder (99.999 %, polycrystalline); ESPS30 – metallurgical grade Si powder (99.999 %, polycrystalline); DW –p+

single crystal Si wafer (B-doped, 10–20 mΩ∙cm); UW – undoped wafer (> 100 Ω∙cm). The amount of Ag was 0.025 mmol, Ag nucleation time was 15 min.

(45)

44

Figure S7.3. SEM images of LL-MACE Si powders of different grades. See Table S7.1 for the details of the Si grades and etching parameters.

(46)

45

Figure S7.4. Pore size distribution of different Si powders after LL-MACE. See Table S7.1for the details of the Si grades and etching parameters.

(47)

46

Figure S7.5. (a) yield, (b) XRPD crystallite size, (c) BET surface area, (d) pore volume, (e) Raman peak shift and (f) Raman crystallite size of Si powders after the LL-MACE. See Table S7.1 for the details of the Si grades and etching parameters.

(48)

47

Section S8. Temperature dependence of LL-MACE.

The length of correlated etch track pores has been observed to be linearly dependent on the etching time at different temperatures when Si wafers are etched with MACE using Ag.³ The etching rate increases with the increase of temperature.³ In this section, we present the temperature dependence of LL-MACE (0.025 mmol) of 11–25 µm MG Si powders. The temperature was controlled by placing the PTFE cup either into an ice bath or water bath, the temperature of which was controlled with a heating plate. Both the temperature of the bath and the temperature of the etching solution was monitored with an IR thermometer. Due to poor thermal conductivity of PTFE, the temperature of the etching solution was typically 10–15 degrees higher than the bath temperature. For example, the average temperatures in the etching solutions were 16 °C, 30 °C and 52.5 °C for ice bath, water bath at 18 °C and water bath at 43°C, respectively.

With the increase of temperature, the yield was found to decrease, while the surface area slightly increased (Figure S8.1). Pore volume did not increase with the temperature change from 16 °C to 30 °C but was significantly higher at the highest temperature of 52.5°C. The absence of change in pore volume between 16 °C and 30 °C was accompanied by a noticeable decrease of the volume of large 8–50 nm pores and increase of both volume and average size of small 2–7 nm pores (Figure S8.3, Table S8.1). The increase of temperature up to 52.5 °C led to further increase of volume and average size of 2–7 nm pores. However, the average size and volume of 8–50 nm pores also increased resulting in the decrease in yield.

Thus, the temperature of 30 °C is optimal, if one wants to reduce the volume of big pores and increase the volume and average size of small pores while keeping the yield relatively high. This temperature prevents the aggregation of Ag into bigger particles and improves the etching of small pores.

(49)

48

Figure S8.1. The dependence between (a) yield, (b) XRPD crystallite size, (c) BET surface area, (d) pore volume, (e) Raman peak shift, and (f) Raman crystallite size and the temperature in the etching solution. The values show the average temperatures in the etching solution measured by IR thermometer. The amount of deposited Ag was 0.025 mmol, Ag nucleation time was 20 min, H2O2/Si molar ratio was 0.9, etching time was 90 min, MG-Si particle size was 11–25 µm.

(50)

49

Figure S8.2. SEM images of 11–25 µm MG Si powders, etched at different temperatures. The values show the average temperatures in the etching solution measured by IR thermometer. The amount of deposited Ag was 0.025 mmol, Ag nucleation time was 20 min, H2O2/Si molar ratio was 0.9, etch time was 90 min.

(51)

50

Figure S8.3. Pore size distributions of 11–25 µm MG Si powders after LL-MACE at different temperatures. The amount of deposited Ag was 0.025 mmol, Ag nucleation time was 20 min, H2O2/Si molar ratio was 0.9, etch time was 90 min.

Tempe- rature,

°C

Average size of

small pores, nm

Volume of small pores, cm³/g

Average size of big pores, nm

Volume of big pores,

cm³/g

Average size of all pores, nm

Total pore volume,

cm³/g

16 4.17 0.078 13.82 0.166 6.89 0.244

30 4.34 0.097 16.60 0.139 6.03 0.236

52.5 4.84 0.111 20.72 0.333 6.78 0.444

Table S8.1. Average pore sizes and pore volumes calculated from pore size distributions (Figure S8.3). Pore volume data for small pores were taken for sizes, that are smaller than the local minimum in the distribution (~9 nm for Ag and Au etched samples). Data points for big pores were taken after the local minimum. Average size of all pores was calculated using the full data range.

(52)

51

References

(1) Paillard, V.; Puech, P.; Laguna, M. A.; Carles, R.; Kohn, B.; Huisken, F. Improved One- Phonon Confinement Model for an Accurate Size Determination of Silicon Nanocrystals. J.

Appl. Phys. 1999, 86 (4), 1921.

(2) Kolasinski, K. W. Electron Transfer during Metal-Assisted and Stain Etching of Silicon.

Semicond. Sci. Technol. 2015, 31 (1), 14002.

(3) Cheng, S. L.; Chung, C. H.; Lee, H. C. A Study of the Synthesis, Characterization, and Kinetics of Vertical Silicon Nanowire Arrays on (001)Si Substrates. J. Electrochem. Soc.

2008, 155 (11), D711.

(4) Kolasinski, K. W.; Barclay, W. B.; Sun, Y.; Aindow, M. The Stoichiometry of Metal Assisted Etching (MAE) of Si in V2O5 + HF and HOOH + HF Solutions. Electrochim. Acta 2015, 158, 219–228.

Controlling the Nature of Etched Si Nanostructures: High- versus Low-Load Metal-Assisted Catalytic Etching (MACE) of Si Powders

Controlling the Nature of Etched Si Nanostructures: High- versus

Low-Load Metal-Assisted Catalytic Etching (MACE) of Si Powders

Tamarov, Konstantin

Supporting Information for

Controlling the nature of etched Si nanostructures:

High versus low load metal-assisted catalytic etching (MACE) of Si powders

Contents

Section S1. Experimental.

Section S2. Effect of deposited Ag amount.

Section S3. HL-MACE: Effect of H

O

/Si molar ratio.

Section S4. HL-MACE: Effect of etching time.

Section S5. LL-MACE: Effect of etching time.

Section S6. LL-MACE: Effect of H

O

/Si molar ratio.

Section S7. MACE of particles of different grades and sizes.

Section S8. Temperature dependence of LL-MACE.

References