Dynamics of Plasma‐Assisted Epitaxial Silicon Growth Driven by a Hydrogen‐Incorporated Nanostructure for Novel Applications

Plasma‐assisted epitaxially grown silicon (plasma‐epi Si) is a new silicon‐based material with a tailorable nanostructure. Nanovoids can be introduced into plasma‐epi Si during growth, enabling the bottom‐up fabrication of porous Si for applications such as batteries, hydrogen storage, and even explosives. To fully control the nanostructure of plasma‐epi Si, its growth dynamics must be studied. In this study, the correlation between hydrogen incorporation and defect nanostructures in plasma‐epi Si grown under various process conditions is investigated, and the experimental results are supported by molecular dynamics simulations. The nanostructural evolution during growth suggests a model in which plasma‐epi Si shows two growth stages distinguished by different dominant defect nanostructures. In the initial growth stage, the nanostructure can be controlled by the deposition conditions, whereas the nanostructure is dominated by interconnected voids, forming a porous structure. In the subsequent bulk growth stage, the material growth is less sensitive to the deposition conditions, whereas the nanostructure becomes prevalent isolated defects. In the results of this study, different strategies for the plasma‐epi Si growth process for different applications are suggested. In these results, a better understanding of this new material may be provided and the discovery of various applications for bottom‐up‐grown porous Si is facilitated.


Introduction
Microelectronic fabrication involves several thin-film deposition processes.Among the various thin-film deposition methods, epitaxy is utilized to grow thin crystalline films with a precise crystal structure and orientation, [1] which matches the crystal lattice of the growing film with that of the substrate it is deposited on.[8][9][10][11] Despite various advantages, such as a good thermal budget and low cost, plasma-epi Si has not garnered significant attention from the scientific community owing to its poor electrical properties.For example, in the highefficiency silicon heterojunction (SHJ) solar cell process, the formation of plasma-epi Si is considered a process failure that should be suppressed. [2,3,15]8][9][10][11][12][13][14][15][16] Plasma-assisted epitaxially grown silicon (plasma-epi Si) is a new silicon-based material with a tailorable nanostructure.Nanovoids can be introduced into plasma-epi Si during growth, enabling the bottom-up fabrication of porous Si for applications such as batteries, hydrogen storage, and even explosives.To fully control the nanostructure of plasma-epi Si, its growth dynamics must be studied.In this study, the correlation between hydrogen incorporation and defect nanostructures in plasma-epi Si grown under various process conditions is investigated, and the experimental results are supported by molecular dynamics simulations.The nanostructural evolution during growth suggests a model in which plasma-epi Si shows two growth stages distinguished by different dominant defect nanostructures.In the initial growth stage, the nanostructure can be controlled by the deposition conditions, whereas the nanostructure is dominated by interconnected voids, forming a porous structure.In the subsequent bulk growth stage, the material growth is less sensitive to the deposition conditions, whereas the nanostructure becomes prevalent isolated defects.In the results of this study, different strategies for the plasma-epi Si growth process for different applications are suggested.In these results, a better understanding of this new material may be provided and the discovery of various applications for bottomup-grown porous Si is facilitated.
The poor electrical properties of plasma-epi Si can be attributed to the defect nanostructures, which have two primary origins.First, the low growth temperature of plasma-epi Si restricts the surface mobility of reactive species, resulting in incomplete surface reconstruction and crystallographic defects such as dislocations and stacking faults. [16]Second, plasma-epi Si consists of a high concentration of Si─H bonds owing to the hydrogen radicals in the plasma.This hydrogen incorporation can also induce defects such as hydrogen-incorporated nanovoids and platelets. [1,17]espite its poor electrical properties, however, the nanostructure of plasma-epi Si can be easily tailored, making it suitable for various potential applications.In our previous study on the fabrication of epitaxial silicon wafers, we employed plasma-epi Si as a self-releasing sacrificial layer, capitalizing on the existence of hydrogen-incorporated nanovoids. [1]Controlling the concentration and distribution of the nanovoids can produce bottom-upgrown porous silicon (pSi) that can be used for gas sensors, [18] lithium-ion batteries, [19,20] hydrogen storage, [21] and even explosives. [22,23]To manipulate the nanostructure of plasma-epi Si and broaden its applications, its growth dynamics must be studied.However, studies on plasma-epi Si are mostly limited to electronic applications thus far.
Herein, we describe a study on the hydrogen-incorporated nanostructure of plasma-epi Si and its growth dynamics by correlating the hydrogen incorporation and defect nanostructure of the material.First, the key deposition parameters for the growth of plasma-epi Si are discussed, and the manipulation of the nanostructure is presented.The growth dynamics as a function of the relative gas flow ratio are comprehensively studied using spectroscopic ellipsometry (SE), transmission electron microscopy (TEM), hydrogen exodiffusion, and secondary ion mass spectrometry (SIMS).The experimental results are supported by molecular dynamics (MD) simulations.Second, a growth model is proposed, based on the growth of plasma-epi Si with various thicknesses.Finally, based on the findings of the study, growth strategies for various applications are discussed.

Nanostructure Control at the Initial Growth
The nanostructure of the plasma-epi Si varies according to the deposition conditions.In this section, we describe the control of the nanostructure of plasma-epi Si films with similar thicknesses but grown under different deposition conditions.One of the most important deposition conditions is the relative flow ratio of the precursor gases. [15]Atomic hydrogen in the plasma plays three roles, promoting epitaxial growth.First, hydrogen covers the growing film surface and enhances surface diffusion of the impinging species (surface diffusion). [24]Second, hydrogen breaks weak Si─Si bonds, resulting in the removal of disordered and less stable Si─Si bonds (etching). [24]Third, hydrogen penetrates the subsurface of the film and releases considerable energy to increase the effective surface temperature locally (chemical annealing). [24,25]However, opposing effects of hydrogen also exist.The incorporation of a high concentration of hydrogen introduces compressive stress to the film and leads to distortion of the crystal lattice. [26,27]Therefore, hydrogen in the plasma not only promotes epitaxial growth but also distorts the crystalline lattice and induces crystallographic defects when incorporated at high concentrations.The concentration of atomic hydrogen during PECVD process can be controlled via the relative gas flow ratio R defined as where SiH 4 and H 2 represent the gas flow rates (in sccm).In addition to the ratio R, the ion bombardment energy is the second most important parameter in plasma-epi Si growth. [28]he other process parameters are discussed in Supporting Information (Note S1 and Figure S1, Supporting Information).
Figure 1 shows the tailorable nanostructure of plasma-epi Si under various R values.The PECVD-deposited silicon thin films were grown at 200 °C with a target thickness of 50 nm.The thickness deduced by SE modeling for films grown with R values of 0.99, 2.91, 3.84, 5.66, and 9.09 were found to be 55.77, 50.73, 51.6, 43.38, and 46.2 nm, respectively.The SE measurements revealed that these silicon thin films with similar thicknesses exhibited distinct nanostructures.The depositions under R = 0.99-5.66exhibited epitaxial growth, whereas the deposition under R = 9.09, which is the lowest hydrogen dilution, exhibited a-Si:H growth.Pseudo-dielectric functions <ε> obtained via SE measurements and the modeling of the gas flow ratio series are shown in Figure 1a-e.Details of the SE measurements and modeling results, such as film thickness, composition, and cross-section structure deduced by the optical models, can be found in Supporting Information (Note S2 and Figure S2, Supporting Information).Figure 1f-j schematically illustrates the corresponding nanostructures deduced from the SE optical models shown in Figure S2 (Supporting Information).The first interesting aspect is the optically distinguishable interface.The films deposited under R = 0.99-3.84showed interference fringes at photon energies below 3 eV, whereas the film deposited under R = 5.66 showed unrecognizable low-energy interference fringe owing to the optically indistinguishable film from the c-Si substrate.Interference fringes arise due to the refractive index contrast in the cross-section and play a crucial role in determining the position of the interface, enabling the estimation of the thickness of the plasma-epi Si layers.However, in cases where the interference fringes are unrecognizable, such as the film deposited under R = 5.66, alternative methods must be employed to estimate the thickness, which include cross-sectional electron microscopy and SE measurements of co-deposited films on a glass substrate.Modeling of the SE measurements (Figure 1a-c and S2, Supporting Information) shows that the plasma-epi Si films grown under R = 0.99-3.84consisted of porous interface layers of thicknesses up to 33 Å as well as void fractions in their bulk.In contrast, the films grown under R = 5.66 consisted of no porous interface layer and 100% c-Si in their bulk.The second interesting aspect is the bulk properties of the plasma-epi Si films.An imaginary part of the pseudodielectric functions (ε i ) of c-Si shows two distinctive peaks at 3.4 and 4.2 eV, which are attributed to the first two direct transition energies in the dispersion relation of the c-Si. [29,30]n contrast, ε i of a-Si:H shows a broad peak located at approximately 3.6 eV, which is attributed to the peak-to-peak transition of a-Si:H. [30,31]While the epitaxial growth of Si is highly preferred on the (100) surface, [1,15,32] the role of hydrogen is also necessary for plasma-epi Si as discussed previously.For plasma-epi Si, the two c-Si peaks of ε i generally became sharper and higher in amplitude for R = 0.99-5.66.The films deposited under R = 0.99, 2.91, 3.84, and 5.66 showed ε i values of 28, 35, 37, and 38, respectively, at 4.2 eV.A high and sharp ε i peak is interpreted as a compact material without void, whereas a low and broad ε i peak is attributed to porosity and structural defects of the crystalline network. [4,29,30]The low and broad ε i peak is also observed in SE analysis of porous Si (pSi) formed via the anodic etching of c-Si. [33,34][35] Therefore, the low and broad ε i peak for the samples grown with low R (0.99-3.84) indicate not only significant porosity but also single-crystal epitaxial growth, as cross verified via the cross-sectional TEM images shown in Figure 1k-o.As shown in the fast Fourier transform results in the inset of Figure 1k, the lowest R sample (0.99) was also a single crystal but highly porous and inhomogeneous.Meanwhile, the film grown under R = 9.09 was amorphous, as shown in the inset of Figure 1o.More details of Figure 1k-o can be found in Supporting Information (Figure S3, Supporting Information), including bright-field (BF) and high-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) images.The TEM images show good agreement with the SE measurement and modeling results (Figure 1a-e and S2, Supporting Information).The cross-sectional TEM images show that the nanostructure of the films evolved from very porous and inhomogeneous for R = 0.99 to a high-quality epitaxial film grown under R = 5.66.The porosity of samples with lower R (0.99 and 2.91) is clearly observed in HAADF cross-sectional STEM images (Figure S3a,b, Supporting Information).Furthermore, the surface region of the c-Si substrate near the interface was found to be highly defective for films grown under R = 0.99-5.66.Similar microstructural damage can be observed in the case of c-Si exposed to hydrogen plasma. [36]Majority of the crystallographic defects in the subsurface region appeared to be platelets, which were planar in shape and aligned predominantly along the (111) crystallographic planes.The (111) platelets are created by the hydrogen plasma. [17]Furthermore, platelets are known to be the generation centers of dislocation loops. [37]n addition, various defects are found in plasma-epi Si films, such as platelets, dislocations, and nanovoids.This result suggests that the excess hydrogen in the plasma modified not only the nanostructure of the growing film but also that of the substrate.Finally, as shown in Figure 1o, a saw-blade-shaped interface is observed, and the BF STEM image (Figure S3e, Supporting Information) confirms that such interface is a result of epitaxial breakdown.This suggests that the film growth under R = 9.09 started as epitaxial at the initial stage, but epitaxial breakdown occurred after a few nanometers of the initial stage, and it turned to amorphous growth.
Plasma-epi Si consists of Si─H bonds. [1,6,8,15]Given the low growth temperature of 200 °C, hydrogen can be easily incorporated into growing films.Hydrogen exodiffusion provides information on the Si─H bonding energy and configurations. [2]etails on the hydrogen exodiffusion measurement and setup are presented in Experimental Section and Supporting Information (Note S3 and Figure S4, Supporting Information).In Figure 2a, hydrogen exodiffusion shows three significant desorption temperatures, i.e., 350, 480, and 600 °C for plasmaepi Si with R = 3.84.In the case of the hydrogenated c-Si surface, three Si─H binding states can be observed in the hydrogen exodiffusion experiment, as shown in Figure S5 (Supporting Information).β 1 desorption at 480-520 °C is attributed to the desorption of hydrogen related to the monohydride rupture on the Si(100)-(2 Â 1) surface, which consists of isolated Si─H bonds. [38,39][40] Desorption at lower temperature of approximately 250 °C is assigned as β 3 , which indicates higher hydrogen coverage. [39]Moreover, β 3 desorption is attributed to the decomposition of dihydride on the Si(100)-(1 Â 1) surface. [40]The Si(100)-(1 Â 1) surface consists of two dangling bonds per surface atom, and the Si─H bonds exist as dihydrides.Considering that the two neighboring Si─H bonds rupture and form H 2 , the newly formed dangling bonds can be easily reconstructed into dimers to form Si(100)-(2 Â 1).This explains why β 2 and β 3 desorption is found at lower temperatures instead of β 1 desorption.In the case of PECVD of a-Si:H, the low-temperature desorption is attributed to the Si─H rupture from interconnected internal voids or the surface of Si nanocrystals, [2,3,41] whereas the high-temperature desorption is considered to be decomposition of Si─H bonds from isolated voids. [2,42]As mentioned previously, plasma-epi Si was grown at 200 °C and consists of Si─H bonds.Therefore, the hydrogen exodiffusion of plasmaepi Si should be considered from a comprehensive perspective consisting of the hydrogenated surface of c-Si and PECVDdeposited films.In other words, the desorption of hydrogen from plasma-epi Si should occur from not only the surface but also the bulk of the plasma-epi Si films.Therefore, increased β 2 and β 3 desorption at lower temperatures should originate from the internal surface of interconnected voids, which is an indication of porosity.
Additional hydrogen desorption was found at high temperatures from 550 to 650 °C for the plasma-epi Si samples grown with R = 0.99-3.84.In the case of the hydrogenated c-Si surface, a high-temperature α shoulder appeared at 550 °C, which is attributed to the desorption of H 2 trapped in c-Si, i.e., tetrahedral interstitial sites. [40]Analogously, in the case of PECVD-deposited a-Si:H, the desorption shoulder at 600 °C indicates hydrogen desorption during crystallization. [43]Therefore, the desorption at 550-600 °C is related to the release of interstitial H 2 .In addition, the accumulation of hydrogen leads to material strain, [44][45][46] which causes the precipitation of lattice-dilating defects such as platelets and dislocations.Platelets are analogous to dislocations in that they are capable of relaxing strain. [47][50] In our exodiffusion results, a weak but clear desorption shoulder at 650 °C was found for samples grown under R = 0.99-5.66.The shoulder at 650 °C was found to be much clearer for all the exodiffusion results plotted in scale as shown in Supporting Information (Figure S6, Supporting Information).The desorption shoulder at 650 °C should be desorption from the platelets [17,37] and dislocation cores, [51,52] which is denoted as δ in this study.Kveder et al. fabricated deformed c-Si by applying mechanical compression and hydrogenating it using hydrogen plasma.As a result, two additional desorption peaks were found at 650 and 800 °C, which they attributed to hydrogen bonding in dislocations. [52]ased on this discussion, the hydrogen exodiffusion results of the plasma-epi Si films agreed well with the SE and TEM results.In the case of the sample grown with R = 5.66, which resulted in the highest crystalline quality, the desorption at 520 °C was dominant and resembled the hydrogen exodiffusion from the hydrogenated c-Si surface, except for the high-temperature shoulder at 650 °C.The samples grown at lower R exhibited two noticeable desorption characteristics.First, β 1 desorption at 520 °C was significantly reduced.Second, the low-temperature β 3 desorption was shown at 250 °C for the sample grown with R = 0.99.The shift of β 2 desorption to a lower temperature (β 3 ) is ascribed to the higher hydrogen concentration as well as higher porosity of the lower R samples.Furthermore, the weak but clear desorption shoulder at 650 °C is linked to the cross-sectional TEM images shown in Figure 1k-n, revealing hydrogen-induced platelets/dislocations not only in the plasma-epi Si film but also underneath the surface of the c-Si substrate.
The results and discussion of the hydrogen-incorporated nanostructures are also supported by MD simulations.The formation energies, which are normalized by the total number of atoms in each system, provide insights into the thermal stabilities of different hydrogen-incorporated defect configurations.By simulating the formation energies at a temperature of 650 °C, the thermal stabilities of the five possible types of hydrogen-incorporated defect configurations aligned with our experimental observations.The simulated thermal stabilities of the nanostructures were found to be in the following order: dislocation > platelet > interstitial > isolated Si─H > Si─H x in interconnected voids, as shown in Figure 2b. Figure 2c shows ball-and-stick diagrams of the representative hydrogenincorporated defect configurations based on the MD simulation.
As discussed previously, the distinction in the nanostructure is significant for the 50 nm thick plasma-epi Si films grown under various gas flow ratios.As hydrogen incorporation in plasma-epi Si is one of the key factors determining material properties, [15] further analysis of hydrogen in plasma-epi Si was performed to obtain information on the spatial distribution of hydrogen.Figure 3a shows the SIMS profiles of hydrogen in the PECVD-deposited silicon thin films grown at various gas flow ratios.The SIMS profile shows that the plasma-epi Si samples consisted of hydrogen in their bulk, but much more hydrogen was incorporated into the interfacial region.Comparing the two plasma-epi Si films with a-Si:H, a remarkable difference in film composition can be observed in the hydrogen profiles shown in Figure 3a.The film grown under R = 9.09, which was a-Si:H, showed a uniform but slightly increasing hydrogen concentration profile over the thickness from the interface.In contrast, the two plasma-epi Si films grown under R = 0.99 and 5.66 showed slightly decreasing bulk hydrogen profiles of 1.8 Â 10 21 and 3 Â 10 20 cm À3 , respectively.Moreover, much higher amounts of hydrogen were found at the interface, showing concentrations of 3.2 Â 10 21 and 1.28 Â 10 21 cm À3 , respectively.This result is consistent with previously presented results, as well as a series of cross-sectional HAADF STEM images and their vertical intensity profiles (Figure 3b-g).
The HAADF STEM series shows that the three PECVDdeposited silicon films changed from porous and defective epitaxy (R = 0.99) to epitaxial breakdown (R = 9.09).The film grown under R = 5.66 exhibited the highest crystalline quality and a non-distinguishable interface, which is consistent with the results presented in Figure 1d,n.In addition, the use of highly diluted SiH 4 , i.e., that with R = 0.99, resulted in a high hydrogen incorporation and a porous material.As discussed previously, a high concentration of hydrogen induces compressive stress on the film and leads to the distortion of the crystalline lattice, [26,27] resulting in defective materials.][55] This result leads us to the conclusion that deposition under R = 0.99 introduces concentrated stress and porosity due to hydrogen incorporation, resulting in a highly inhomogeneous HAADF STEM cross section, particularly at the interface, as shown in the intensity profile (Figure 3g).This is supported by the high hydrogen concentration obtained from SIMS and the high concentration of voids and defects observed in the cross-sectional TEM images.Comprehensively, the growth of plasma-epi Si starts from a highly hydrogen-incorporated interface, where the initial growth should be particularly defective and porous to relax the strain induced by hydrogen. [15,47]In the case of the plasma-epi Si used as a component of the semiconductor devices (i.e., doped emitter or back-surface field), the defective interface layer should be as thin as possible to avoid recombination loss.

Nanostructure Evolution upon Bulk Growth
In the previous section, we discussed the material properties of the 50 nm thick plasma-epi Si films, grown under various R values.Interestingly, the nanostructure of plasma-epi Si not only depends on the growth conditions but also evolves as the films grow thicker.In this section, we discuss the changes in the nanostructure of plasma-epi Si films upon bulk growth.Figure 4a-f shows a series of SE measurements and modeling results for the plasma-epi Si films grown on the (100) c-Si surfaces of various thicknesses.To demonstrate the evolution of the nanostructure during growth, the R ratio was set to 0.99, representing a defective and porous growth condition.Peak amplitudes of ε i at 3.4 and 4.2 eV evolved during plasma-epi Si growth.The peak amplitudes at 4.2 eV started from a low value of 32 at the initial stage of epitaxial growth and then continuously increased during growth.Moreover, the 490 nm thick plasma-epi Si film showed an ε i amplitude that reached 41 at 4.2 eV, which is close to the value of a clean c-Si surface with native oxide.The modeling results shown in Figure 4 suggest that the growth process of plasmaepi Si moved from an initial stage of highly porous growth toward the growth of a compact bulk phase, thereby evolving the nanostructure.The proposed growth model for plasma-epi Si is schematically shown in Figure 4g.
Although the peak amplitudes of ε i evolved, low-energy interference fringes persisted during growth.As discussed in Section 2.1, the low-energy interference fringes are due to the presence of nanovoids at the interface between the substrate and plasma-epi Si.The presence of the nanovoids at the interface has also been reported for low-temperature epitaxy techniques such as molecular beam epitaxy, [56] hot-wire chemical vapor deposition (CVD), [57] and PECVD. [13,58]This imperfect interface has also been cross-verified via cross-sectional TEM. [7,12,15]35] Therefore, the interface layer can be depicted as the porosity from the initial stage that resides at the interface.
It is also noteworthy that the thickness of the interface layer deduced by the SE model revealed an apparent inconsistency in measurement techniques.According to SE measurement and modeling, this thickness tends to decrease as bulk growth progresses.In contrast, the interface layer thickness in TEM crosssection images remains approximately 5 nm thick in all cases, as shown in Supporting Information (Figure S7, Supporting Information).There are two possible explanations for such a discrepancy in the interface layer thickness.First, it could be attributed to artifacts in the SE modeling of the measurement results.In this work, SE measurements were conducted in reflection mode, which is more sensitive to the surface than it is to the bottom of the sample, particularly for semi-absorbing materials like plasma-epi Si.As the film reaches a thickness of approximately 500 nm, a significant portion of the probe beam should be absorbed by the bulk, thereby reducing the sensitivity of the SE measurements to the interface layer at the bottom, which is only a few nanometers thick.Therefore, as the film grows thicker, the information obtained from the interface layer may become less accurate than that obtained from the surface.The second explanation is related to the modification of the interface layer by hydrogen plasma.Hydrogen can diffuse into c-Si up to approximately 1 μm. [59,60]Prolonged exposure to hydrogen plasma modifies the film's nanostructure, [25] and such structural modifications induced by hydrogen plasma can also occur during the film growth process. [2,5,24]Consequently, as the plasma-epi Si film grows thicker, the interface layer beneath the bulk may be modified by the hydrogen plasma.To minimize the number of fitting parameters in our SE modeling, the composition of the interface layers was fixed at 50% small-grain Si (ud) and 50% voids, and only the thickness of the interface layers was adjusted.Therefore, structural modifications to the interface layer caused by hydrogen plasma during growth may be reflected in the SE modeling.
Figure 5 shows the changes in the hydrogen exodiffusion of the plasma-epi Si films with various thicknesses.To show the low-level signals of thinner samples, the hydrogen partial pressure is shown on a logarithmic scale.The 50 nm thick films exhibited different hydrogen exodiffusion characteristics, as discussed in Section 2.1, which may be attributed to the controllable nanostructure depending on the deposition conditions.Interestingly, their hydrogen exodiffusion results were similar when grown as 500 nm thick films.Commonly, β 1 desorption at 520 °C is the most significant while accompanied with β 2 desorption at 350-400 °C, α shoulder at 550 °C, and desorption from platelet/dislocation (δ) at 650 °C.For desorption from isolated Si─H bonds, the released hydrogen diffuses slowly into the material in the form of atomic hydrogen.The high-temperature desorption of hydrogen is a diffusion-rate-limited process, and the peak temperature shifts to higher values when the film is thicker.Unlike isolated Si─H bonds, desorption from interconnected voids should be much faster and thickness independent because two neighboring Si─H bonds rupture and form H 2 in the open space of the interconnected voids.In case of the film grown with R = 0.99, the β 3 desorption peak was at same temperature of 250 °C for both 50 and 100 nm thick films.Considering that the sample was heated with a constant rate of 5 °C min À1 , desorption at the same temperature from two different thicknesses was associated with interconnected voids.In other words, the sample with R = 0.99 was initially grown to be highly porous, but as it reached to a thickness of 500 nm, its nanostructure became compact bulk with prevalent isolated defects.
The growth model depicted previously was verified by crosssectional TEM images of plasma-epi Si films grown using R = 0.99 having thicknesses of 50, 200, and 500 nm, as shown in Figure 6a-c.Figure 6d,e shows a zoomed-in view of the interface and bulk region of the 500 nm thick plasma-epi Si, as marked by orange squares in Figure 6c.At the initial stage of the growth of the plasma-epi Si, the material appeared to be porous and inhomogeneous, as shown in Figure 6a,b,d.The dominant defects at initial growth appeared to be misfit and threading dislocations, which were directly analogous to the propagation of platelets. [47]As the film grew to a thickness of 500 nm, the defect density significantly decreased in the bulk region of the 500 nm thick plasma-epi Si film, as shown in Figure 6c.Furthermore, in the bulk region, isolated crystallographic defects were mainly observed, such as coffee-bean-shaped (111) platelets, as shown in Figure 6e.
The results of the thickness series suggest a few key points regarding the growth dynamics of plasma-epi Si.First, at the initial growth stage, i.e., at a thickness of a few tens of nanometers, the nanostructure of plasma-epi Si can be controlled via the deposition conditions.Moreover, the nanostructure at the initial stage is porous and interconnected void dominated because of high hydrogen incorporation.The origin of the porosity at the initial growth should be the relaxation of the strain of hydrogen incorporation.Second, after the plasma-epi Si grows thick, the bulk growth becomes less dependent on the deposition conditions because the initial strain owing to hydrogen incorporation is already relaxed.Therefore, the growth of bulk plasma-epi Si is determined by the surface mobility of the reactive species, in which the growth temperature should be the prepotent process parameter.Third, the resulting bulk material shows that dominant defects are isolated ones such as interstitial H 2 , dislocations, and platelets.Based on these results, different deposition strategies can be established for different applications.For example, to improve carrier transport property of plasma-epi Si for electronic applications, the strain at the initial growth stage should be suppressed by adequate substrate cleaning, impurity control, and uniform substrate temperature.Isolated bulk defects should also be carefully controlled by maximizing the surface mobility of the reactive species.For example, growth at high temperatures can be considered.Otherwise, for the application of bottom-up-grown pSi, there is a risk that the desired porosity may be confined primarily to the interfacial region because the hydrogen-induced strain should be already relaxed at the initial growth stage.Therefore, a consistent supply of strain during growth should be considered if thick bottom-up pSi is desired.For example, layer-by-layer hydrogen plasma treatment or profiled gas flow during growth can be considered.

Conclusion
Plasma-epi Si is a new Si-based material that can be fabricated using conventional PECVD tools, while having a tailorable nanostructure and porosity, which has a great potential for applications beyond electronic materials and devices.Herein, we present the growth dynamics of plasma-epi Si driven by a hydrogen-incorporated nanostructure.At the initial growth stage, the nanostructure of the plasma-epi Si was manipulated by controlling the relative gas flow ratio.The initial growth dynamics of plasma-epi Si films grown under various deposition conditions were verified via SE, TEM, and hydrogen exodiffusion measurements.The correlation between hydrogen exodiffusion and the nanostructure of the plasma-epi Si was studied, and the spatial distribution of hydrogen-incorporated defects was monitored using STEM and SIMS.The experimental results were also supported by MD simulations.For the thickness series, we suggest a growth model of plasma-epi Si in which porosity dominates at the initial stage of growth owing to a high hydrogen incorporation, whereas the bulk nanostructure becomes prevalent isolated defects after some hundreds of nanometers.This growth model suggests strategies for various applications.In electronic applications, the strain during initial growth should be suppressed, whereas isolated defects during bulk growth should be carefully controlled by maximizing the surface mobility of the impinging species.In the case of bottom-up pSi growth to the desired thickness, a consistent supply of strain should be considered to maintain porosity throughout the bulk.This study not only provides general insights into various crystallographic defects in c-Si, which is widely recognized as one of the most crucial semiconductor materials, but also presents recent updates on plasma-epi Si.However, the results presented in this study are constrained to the manipulation of material properties by understanding the plasma-epi Si growth dynamics, and further investigations should be conducted to explore the potential of plasma-epi Si for diverse applications of pSi, including but not limited to gas sensors, lithium-ion batteries, hydrogen storage, and even explosives.Overall, this article is expected to contribute to a better understanding of this new material, which will lead to various applications for bottom-up pSi.

Experimental Section
Reactors for Plasma-epi Si Film Growth: Plasma-epi Si films were grown via capacitively coupled-plasma radio-frequency (13.56 MHz) glow discharge PECVD method at a substrate temperature of 200 °C.Polished (100) c-Si substrates were used.To ensure the reliability and feasibility of the growth process, two different plasma reactors were employed to grow the plasma-epi Si films.The nanostructure control in the initial growth stage, as depicted in Figure 1-3, was investigated using a single chamber PECVD reactor (custom-built, Ultech, South Korea), which allows for the high-temperature process of conventional low-pressure CVD.The thickness series of experiments, the results of which are shown in   4-6, was carried out in a cluster tool consisting of five independent PECVD reactors (custom-built, SNTEK, South Korea), which were specifically designed for depositing ordinary thin-film silicon such as a-Si:H.Despite the different designs of the two reactors, consistent results were obtained in terms of film growth behavior under common deposition conditions (e.g., when changing the relative gas flow ratio of SiH 4 and H 2 ).The details of the PECVD reactors can be found in our previous publications. [1,2,4]haracterizations: SE measurements and modeling were performed using an M-2000 ellipsometer (M-2000UI, J. A. Woollam, USA).The recorded SE spectra of the PECVD-deposited silicon films were modeled using the Bruggeman EMA with optically different silicon phases such as single-crystalline silicon (c-Si), large-grain polycrystalline silicon (As), small-grain polycrystalline silicon (ud), and voids. [30]More details on the SE measurements and modeling can be found in Supporting Information (Note S2 and Figure S2, Supporting Information).The hydrogen exodiffusion measurements were performed using a custom-built setup (Realtek, South Korea).Samples deposited on polished (100) c-Si substrates were loaded in a quartz tube under vacuum (10 À8 Torr).The samples were then heated at a constant ramp rate of 5 °C min À1 , from 50 to 800 °C, and the total duration for the hydrogen exodiffusion measurement was approximately 150 min.The sample temperature was measured using a thermocouple with the wires in contact with the sample surface.The change in the partial pressure of the desorbed hydrogen gas was measured using a quadrupole mass spectrometer (QMG 250 F2, Pfeiffer vacuum, Germany), which was part of the custom-built hydrogen exodiffusion setup.Further details on the hydrogen exodiffusion setup can be found in Supporting Information (Note S3 and Figure S4, Supporting Information).The depth profile of the hydrogen concentration was determined using SIMS (IMS-6f Magnetic Sector SIMS, CAMECA, UK).The primary ion source was Cs þ , with an impact energy of 5 keV.Cross-sectional TEM measurements were performed using a field-emission electron microscope (JEM-2100F, JEOL, Japan).
MD Simulations: MD simulations were performed to investigate the thermal stability of hydrogen-incorporated defect configurations.The COMPASS III force field was used in the MD simulations. [61]Previous studies utilized the COMPASS force field in both c-Si [62][63][64][65][66] and crystalline silica [67][68][69] systems.The simulated density value of pristine c-Si systems (i.e., 2.331 g cm À3 at 25 °C) and the lattice constants of an 8 Â 8 Â 8 supercell of a Si unit cell (i.e., 43.433 Å at 25 °C) closely match the experimental values (i.e., 2.329 g cm À3 and 43.446 Å for 8 Si unit cells).The isothermalisobaric ensemble (NPT ensemble), where a constant number of particles (N), pressure (P), and temperature (T ), was simulated for 2 ns with a Berendsen thermostat and barostat [70] using a time step of 1 fs, under pressure and temperature conditions of 1 atm and 650 °C, respectively.The reference temperature of 650 °C was chosen to distinguish the thermal stabilities of various hydrogen-incorporated defects in this work.An atombased summation method with a cutoff distance of 1.25 nm was used for the short-range van der Waals interactions.The particle-particle, particlemesh method with an accuracy of 0.001 kcal mol À1 was used for longrange electrostatic interactions.Formation energies were calculated using the following equations: Formation E ¼ E defect system À E defect system w=o H À n H in defect system Â E H 2 n H in H 2 (2)   where E and n represent the total energy and number of atoms, respectively.The five types of defect systems were modeled based on an

Figure 1 .
Figure 1.Nanostructural evolution of the plasma-enhanced chemical vapor deposition (PECVD)-deposited silicon thin films grown on the (100) c-Si surfaces under various relative gas flow ratios, R. a-e) Pseudo-dielectric functions <ε> obtained via spectroscopic ellipsometry (SE) measurement and modeling.f-j) Schematic representation of the corresponding nanostructures deduced via an optical model of the SE measurement.The blue balls represent Si atoms, the light-blue balls represent dislocated Si atoms, and the red balls represent H atoms. Detailed optical models of SE measurements can be found in Supporting Information (Figure S2, Supporting Information).k-o) Cross-sectional transmission electron microscopy (TEM) images.The interface is marked with yellow arrows.Insets: fast Fourier transform (FFT) results of each corresponding image.

Figure 2 .
Figure 2. a) Hydrogen exodiffusion results of the PECVD-deposited silicon thin films under various R values.In the curve for of R = 9.09, LT and HT refer to low and high temperature, respectively.b) Formation energies of five possible types of hydrogen-incorporated defect configurations.The formation energies are normalized by the total number of atoms in each system.c) Ball-and-stick diagrams of the representative hydrogen-incorporated defect configurations based on the MD simulation.The blue balls represent Si atoms, the light-blue balls represent dislocated Si atoms, and the red balls represent H atoms.

Figure 3 .
Figure 3. a) Secondary ion mass spectrometry hydrogen profiles of the PECVD-deposited silicon thin films grown at various gas flow ratios.b,d,f ) Crosssectional high-angle annular dark-field scanning transmission electron microscopy images and c,e,g) their corresponding vertical intensity profiles along the indicated arrows (black, red, and blue).The film-substrate interface is indicated by yellow lines/arrows.Insets: FFT results of each corresponding image.

Figure 4 .
Figure 4. a-c) Series of SE measurement and d-f ) corresponding optical model of plasma-epi Si films grown on the c-Si(100) surfaces of different thicknesses.g) Schematic diagram of the growth model of plasma-epi Si proposed in this work.

Figure 6 .
Figure 6.Cross-sectional TEM images of a) 50 nm, b) 200 nm, and c) 500 nm plasma-epi Si and magnified views of the marked d) interface and e) bulk regions of the 500 nm plasma-epi Si.

Figure
Figure4-6, was carried out in a cluster tool consisting of five independent PECVD reactors (custom-built, SNTEK, South Korea), which were specifically designed for depositing ordinary thin-film silicon such as a-Si:H.Despite the different designs of the two reactors, consistent results were obtained in terms of film growth behavior under common deposition conditions (e.g., when changing the relative gas flow ratio of SiH 4 and H 2 ).The details of the PECVD reactors can be found in our previous publications.[1,2,4]Characterizations: SE measurements and modeling were performed using an M-2000 ellipsometer (M-2000UI, J. A. Woollam, USA).The recorded SE spectra of the PECVD-deposited silicon films were modeled using the Bruggeman EMA with optically different silicon phases such as single-crystalline silicon (c-Si), large-grain polycrystalline silicon (As), small-grain polycrystalline silicon (ud), and voids.[30]More details on the SE measurements and modeling can be found in Supporting Information (Note S2 and FigureS2, Supporting Information).The hydrogen exodiffusion measurements were performed using a custom-built setup (Realtek, South Korea).Samples deposited on polished (100) c-Si substrates were loaded in a quartz tube under vacuum (10 À8 Torr).The samples were then heated at a constant ramp rate of 5 °C min À1 , from 50 to 800 °C, and the total duration for the hydrogen exodiffusion measurement was approximately 150 min.The sample temperature was measured using a thermocouple with the wires in contact with the sample surface.The change in the partial pressure of the desorbed hydrogen gas was measured using a quadrupole mass spectrometer (QMG 250 F2, Pfeiffer vacuum, Germany), which was part of the custom-built hydrogen exodiffusion setup.Further details on the hydrogen exodiffusion setup can be found in Supporting Information (Note S3 and FigureS4, Supporting Information).The depth profile of the hydrogen concentration was determined using SIMS (IMS-6f Magnetic Sector SIMS, CAMECA, UK).The primary ion source was Cs þ , with an impact energy of 5 keV.Cross-sectional TEM measurements were performed using a field-emission electron microscope (JEM-2100F, JEOL, Japan).MD Simulations: MD simulations were performed to investigate the thermal stability of hydrogen-incorporated defect configurations.The COMPASS III force field was used in the MD simulations.[61]Previous studies utilized the COMPASS force field in both c-Si[62][63][64][65][66] and crystalline silica[67][68][69] systems.The simulated density value of pristine c-Si systems (i.e., 2.331 g cm À3 at 25 °C) and the lattice constants of an 8 Â 8 Â 8 supercell of a Si unit cell (i.e.,43.433Å at 25 °C) closely match the experimental values (i.e., 2.329 g cm À3 and 43.446 Å for 8 Si unit cells).The isothermalisobaric ensemble (NPT ensemble), where a constant number of particles (N), pressure (P), and temperature (T ), was simulated for 2 ns with a Berendsen thermostat and barostat[70] using a time step of 1 fs, under pressure and temperature conditions of 1 atm and 650 °C, respectively.The reference temperature of 650 °C was chosen to distinguish the thermal stabilities of various hydrogen-incorporated defects in this work.An atombased summation method with a cutoff distance of 1.25 nm was used for the short-range van der Waals interactions.The particle-particle, particlemesh method with an accuracy of 0.001 kcal mol À1 was used for longrange electrostatic interactions.Formation energies were calculated using the following equations: