Modeling of the impurity-induced silicon nanocone growth by low energy helium plasma irradiation

Quan SHI(石權),Shin KAJITA,Shuyu DAI(戴舒宇),Shuangyuan FENG(馮雙園) and Noriyasu OHNO

1 Graduate School of Engineering,Nagoya University,Nagoya 464-8603,Japan

2 Institute of Materials and Systems for Sustainability,Nagoya University,Nagoya 464-8603,Japan

3 Key Laboratory of Materials Modification by Laser,Ion and Electron Beams Ministry of Education,School of Physics,Dalian University of Technology,Dalian 116024,People’s Republic of China

Abstract The formation mechanism of nanocone structure on silicon(Si)surface irradiated by helium plasma has been investigated by experiments and simulations.Impurity(molybdenum)aggregated as shields on Si was found to be a key factor to form a high density of nanocone in our previous study.Here to concrete this theory,a simulation work has been developed with SURO code based on the impurity concentration measurement of the nanocones by using electron dispersive x-ray spectroscopy.The formation process of the nanocone from a flat surface was presented.The modeling structure under an inclining ion incident direction was in good agreement with the experimental result.Moreover,the redeposition effect was proposed as another important process of nanocone formation based on results from the comparison of the cone diameter and sputtering yield between cases with and without the redeposition effect.

Keywords:black silicon,helium plasma,nanocone,simulation

1.Introduction

Silicon with a high density of nanocone morphology,known as black silicon(Si),is one of the promising materials for the solar cell and field emitter[1,2].Various techniques have been conducted to fabricate nanocones on the Si surface such as reactive ion etching,wet chemical etching,and laser pulses[3-5].Helium(He)plasma irradiation,which is relatively simple and economical,is another method to form nanostructure on the surface of material.Kajita et al have observed cone structure,of which He bubbles on the tip,on titanium and stainless steel[6,7].As for the Si material,dense nanocone was formed with crystal nature remained at low energy(＜100 eV)of He plasma[8-10].Recently,we found that the formation of nanocone structure was strongly affected by the deposition of impurity on the Si surface[11].On Si material,impurity-induced nanostructure was broadly reported with argon(Ar)plasma irradiation.Nanorods have formed at various temperatures with the supply of molybdenum(Mo)seed atoms.Ozaydin et al and Tanemura et al have reported that nanodot structure cannot be identified without Mo impurity introduced[12,13].

However,the mechanism of impurity-induced Si nanocone is still in discussion.Iron and chromium(Cr)impurities are seemed as melted catalytic-sphere to lead the nanocone growth in the 1500 eV of Ar plasma environment[14].At a lower Ar ion energy(300 eV),Ozaydin et al concluded that the tensile stress plays a dominant role in driving the nanodot formation[12].The shielding effect of impurity was also considered to contribute to the formation of cone structure on both Si[15]and Cr[16]substrates,because impurities were observed on the tip of the cone.

Simulation is a persuasive method to verify the mechanism of nanocone formation.Time development of conical shape has been produced with pure Ar ion bombardment[17].In that model,the cone structure was initially developed from a small protuberance,which represents the asperity on the surface,and finally eroded.However,the nanocone structures in our previous study[11]were steady for long-time irradiation and the original surface was smooth in nanoscale.Moreover,a black dot,which was likely to be the impurity shield,was observed on the tip of cone.Thus,it is meaningful to simulate the cone formation based on the shielding effect.

Several issues related with plasma surface interaction have been investigated with the three-dimensional Monte Carlo code SURO[18,19].In this work,we upgraded SURO to develop the formation of Si nanocone.Redeposition of Si was found to be another important process during the irradiation.

Figure 1.(a)TEM of a single nanocone.The EDX mapping of(b)Si and(c)Mo on the nanocone in(a).(d)The distribution of Mo atoms ratio calculated by Mo atoms/(Mo atoms+Si atoms).Three specific regions of the Mo ratio were marked.

2.Preparation of simulation

2.1.Impurity distribution

In our previous study,Si nanocone was fabricated with He plasma at 57 eV as a result of the Mo seeding from the sample mask[11].To support the simulation model,electron dispersive x-ray spectroscopy(EDX)mapping was used to examine the distribution of impurities of the cone,as shown in figure 1.The observed cone was shown by transmission electron microscopy in figure 1(a).Figures 1(b)and(c)show the intensity of Si and Mo respectively for K line.They suggest that the intensity of Si is higher at the bottom than that of the top region.Moreover,size of Si presenting region,is smaller than Mo.With the data from figures 1(b)and(c),the atomic concentration of Mo,CMo,was deduced from

where CSiis the atomic concentration of Si,kMo,Siis a constant that accounts for the relative detection efficiency which is approximately 4 according to[20],IMoand ISiare the x-ray intensities for Mo and Si,respectively.The distribution of the Mo concentration is shown in figure 1(d).One can see that the average concentration of Mo is high(about 0.3)at the profile of the nanocone.This phenomenon suggests the present of a Mo deposition layer,mixed with Si,corresponding to the profile region of the cone with higher transparency shown in figure 1(a).On the tip,the ratio of Mo in the deposition layer is even higher(about 0.6)which indicates the surface diffusion or adatoms.This result is in agreement with the hypothesis of the shielding effect.The atomic concentration of Mo on the tip of cone was set as an input factor in the following simulation model.

2.2.Simulation model

In SURO code,the size of the simulation region is 0.5×0.5 μm2separated into 50×50 meshes.To perform the impurity effect,Mo impurities are set initially at the center of the entire region occupying 4 meshes as shown in figure 2(b).The fractions of Mo in those meshes were simply set as 0.6,while the rest of the meshes were pure Si.Information on height,species of particles,number of particles,the slope of the local region,etc is stored in each mesh.The information will be updated after a single time loop of 0.01 s.

Figure 2.(a)A schematic of SURO code,(b)simulation region set up from top view,and(c)the polar angle and azimuthal angle used in the local coordinate.

The code can be mainly separated into two segments,i.e.sputtering and redeposition,as shown in figure 2(a).First,He ions randomly incident from the top in a vertical direction towards the Si surface.Sputtering happens when ions hit the Si surface which belongs to a certain mesh.Then the changes in surface height at that mesh will be calculated based on the sputtering information.The results in section 3.1 were deduced from only the first part.

In the second part,the sputtered Si atoms will leave the surface at a certain angle and energy.In the local coordinate,as shown in figure 2(c),the polar angle and the azimuthal angle are assumed as the sine[21]and uniform distribution,respectively.The energy of sputtered Si follows the Thompson distribution[22].The system will trace the sputtered Si atoms until they re-deposit on the surface or exceed the simulation region.

Because sputtering is the most important process in the simulation,we test the energy and angular dependence of sputtering yield for both Si and Mo materials and compared with the results from A-CAT[23]code and some experiments[24]as shown in figure 3.The sputtering yield is deduced from the following equations:

Here,E0is the projectile energy,α is the angle of incident ion to the surface normal,f and αoptare used as fitting parameters.

Figures 3(a)and(c)are the energy dependence of sputtering yield at the normal incident direction on Mo and Si,respectively.The sputtering yield calculated in SURO code is slightly higher than the A-CAT code.However,the results are consistent with the experimental data well.It is shown that the sputtering yield of Mo is much smaller than that of Si.Because the He ion energy in our previous experiment was～60 eV,all the results presented in section 3 were obtained at the ion energy of 60 eV.Moreover,the angular dependence of sputtering yields has been considered as an important factor that leads to the formation of conical structure in many research[17];thus,we benchmark the angular dependence of sputtering yields for Mo and Si as shown in figures 3(b)and(c).In order to approach the experiment result in[11],the angular dependence of sputtering yields for Mo and Si are calculated under 60 eV of He ion energy.The sputtering yield for Mo is zero in the A-CAT calculation at 60 eV.On Si material,the result calculated from SURO has a similar profile with that from A-CAT.

3.Results and discussion

3.1.Nanocone formation

Figure 4 shows the formation process of Mo-induced Si nanocone.Cone structures at 20,100,and 200 s are shown in figures 4(a)-(c),respectively.The original height of the surface was 0 μm.Because the sputtering yield of Mo is smaller than that of Si,a small protuberance formed from the flat surface where Mo impurities located.As the erosion goes deeper,the cone shape formed and the height gradually becomes larger.The angular dependence of sputtering yield was thought to be an important factor for the formation of nanocone[17].Besides,the incident angle for a maximum sputtering yield was considered to significantly influence the cone shape[25,26].However,a similar result was not shown in our research.The height of the cone will keep increasing as the irradiation time passes,while the diameter,i.e.the width of the bottom of the cone,is not clearly extended and only consistent with the Mo covered region.It is probably because that protuberance without the coverage of Mo will be sputtered immediately,due to the higher sputtering yield at a tilted incident angle.Hence,we conclude that the shielding effect of Mo is a primary reason for the nanocone formation.The diameter of the cone will be discussed again in section 3.2 with redeposition effect.

Figure 3.Energy(a),(c)dependence of sputtering yield for the normal incident,and angle(b),(d)dependence of sputtering yield at He ion energy of 60 eV on Mo(a),(b)and Si(c),(d),respectively,from SURO,A-CAT code,and experimental results.

Figure 4.Cone structure evolution at(a)t=20 s,(b)t=100 s,(c)t=200 s,from the side view.

The direction of Si nanocone pointed was found to be followed the incident ion in our previous study[11].As shown in figure 5(a),the cone structures are inclined to the surface.Generally,the direction of the incident ion is perpendicular to the target surface because of the sheath layer.However,the configuration of the sheath was changed close to the sample mask which was made of Mo.With the influence of the mask,the direction of the electric field was inclined away from the mask.In figure 5(a)we rotated the image to make the direction of incident ion perpendicular to the horizontal for a convenient comparison with the simulation result in figure 5(b).

In this study,we changed the angle between the incident He ions and the surface.Because only unique value of height is allowed for one(x,y)coordinate in our model,we tilt the surface and keep the incident ions parallel to the z axis instead of inclining the incident direction.As shown in figure 5(b),the growth direction is not normal to the surface but parallel to the direction of the incident ions.This result suggested that,the direction of the cone is strongly affected by the angle of the incident ion.

3.2.Redeposition effect

Figure 5.(a)Rotated image of nanocone formed on the surface with oblique incident He ion in the experiment.(b)Nanocone formed on the tilted surface.The directions of incident ions in(a)and(b)are set the same for convenient comparison.

Figure 6.Cross-section of the nanocone with(red dashed)and without(blue solid)redeposition effect after 1000 s.

Figure 6 shows the cross-section of the nanocone with and without redeposition effect after He ions irradiation for 1000 s.At the base of the surface and the tip of the cone,the height of the redeposition(the second part introduced in section 2.2)case is almost the same as the case without the redeposition effect.At the bottom of the cone,however,the height of the former is much higher than that of the latter.Thus,the redeposition is more likely to occur at the bottom of the cone other than uniformly on the whole surface.This is because that sputtered Si atom left the surface at a certain angle which avoids it to redeposit at the same position where it had been sputtered.However,it possibly hit the cone which protruded from the surface.We assumed that if the hit energy is smaller than the surface binding energy,deposition occurs at the hit point.Consequently,the height of the surface at the hit point is increased according to the number of deposited atoms.Considering the structure of the cone,the upper region was too small for the sputtered atoms to reach,while the bottom can receive the atoms from the ground region of the surface.This phenomenon can be considered as one of the processes to increase the cone diameter.Moreover,the heavier redeposition of Si at the bottom region diluted the concentration of Mo which explains the lower Mo ratio at the bottom of the deposition layer observed in the EDX mapping in figure 1.

In the former study,the effective sputtering yield for black Si was found to be significantly decreased compared with that on the flat surface.Here,we developed a nanocone array with 49 cones to approach the black Si case.Morphologies of Si surface developed with and without the redeposition effect are shown in figures 7(a)and(b),respectively.By comparing the erosion depth of(a)and(b)at the center of the sample as shown in figure 7(c),one can recognize that the redeposition effect becomes heavier than the single cone case in figure 6.This can be explained by the enhanced redeposition from the surrounding cone.Moreover,the redeposition effect is one of the important process leading to the reduction of the effective sputtering yield which was observed in[11].Without the redeposition process,the sputtering yield is roughly2.3 ×10?2which is almost the same as the value(2.31 ×10?2)calculated on the flat Si surface in figure 3(c).

However,this reduction is still far from the real condition in which the effective sputtering yield of black Si was 80%less than that on a pristine surface.This is probably because the surface in our simulation model almost consists of pure Si other than the tip of cones.While in the actual irradiation,even if not as much as tips,the Si substrate is also mixed with Mo.The effective sputtering yield of Cr cone surface exposed to a He plasma at an incident ion energy of 80 eV has been measured during the irradiation[16].The decrease of effective sputtering yield can be separated into two phases:(1)a rapid decrease by 50% up to an ion fluence of 5 ×1024m?2at the beginning;(2)a linear mild decrease of 1.3% per the ion fluence of 1 ×1024m?2.Considering the measurement of the atomic concentration of the Si nanocone surface performed by Ying et al[14],the deposition of impurities on the sample was saturated in a very low ion fluence.Thus,we believe that the first phase of the sputtering yield reduction is due to the impurity accumulation on the surface.The second phase might be explained by the redeposition effect.Figure 7(d)shows the reduction of the effective sputtering yield with increasing the ion fluence in the redeposition model.The reduction rate was 3.6% per 1 ×1024m?2of the fluence which is in the same order of magnitude as the result in[16].This result suggests that the formation of dense nanocone will reduce the effective sputtering yield due to the enhanced line-of-sight redeposition.

Figure 7.Surface morphology with the structure of 49 nanocones with(a)and without(b)redeposition effect after 1000 s.(c)Comparison in the cross-section view of the nanocone at the center of(a)and(b).(d)Effective sputtering yield with redeposition effect as a function of ion fluence.

4.Summary

In this study,we measured the distribution of Mo on the Si nanocone formed by He plasma irradiation.A high density Mo zone,which is considered as a shield for the seed of nanocone formation,with a Mo fraction of 0.6 was observed on the tip of cone.Based on the shield effect model,we reproduced the development of cone structure with time initiated from a flat Si surface by SURO code.The direction of the cone structure follows the incident ion in a tilted substrate case,which is consistent with the experimental result.Redeposition of sputtered Si atoms was suggested to be another important process in the formation of nanocone.The redeposition effect will enlarge the diameter of the cone.Moreover,it could be the reason for the gentle decrease of the sputtering yield during the long period of irradiation.

Acknowledgments

This work was supported in part by a Grant-in Aid for Scientific Research(Nos.17KK0132,19H01874)from the Japan Society for the Promotion of Science(JSPS).The contribution by Dr Shuyu Dai was also supported by National MCF Energy R&D Program of China(Nos.2018YFE0311100 and 2018YFE0303105)and National Natural Science Foundation of China(No.12075047).