Multi-gate FET system has undergone tremendous advancements over the last decade and the biggest challenge for this system has been doping thin-body silicon. Ion implantation is a crucial enabler for multi-gate devices in the production of power semiconductor devices. When ion implantation was modeled into Si FinFETs using atomistic simulation, dopant retention was dramatically reduced by backscattering for low energy & tilt angles and by the transmission for high angles which are both relative to the top surface.

Scientifically is not yet clear whether switching to non-Si materials would alter the issues related to energetic ions striking thin body semiconductors, whether via ion or plasma-assisted implant. Target sputtering, dose retention, and damage development will be impacted by variations in density, mass, average atomic mass, atomic spacing, surface binding energies, and lattice. The use of Binary Collision Approximation models and a physics-based simulation approach is a crucial part. This study is thoroughly focused on examining how to target physical characteristics affect dose retention and target damage.

Modelling Set-ups using SRIM software:

The Stopping and Range Matter (SRIM) is modeling software that offers a technique to investigate the physics and ion implantation process. To properly understand and assess the harm produced by energetic ions, the Lattice Binding Energy (LBE), the Displacement Energy (DE), and the Surface Binding Energy (SBE) of the components were constantly monitored throughout the experiment.

Fig 1: a) Backscattering, Transmission, and Sputtering.

b) Angles shown relative to the sidewall of the fin structure.

(right side) the angles in the forthcoming graphs are defined relative to the sidewall of the fin structure.  

Table 1:-  Parameters used in the SRIM software with energy values set to default

Backscattering, Transmission, and Sputtering are the possible sources of dopant loss during the ion implantation into the fin structures.

Results and trends shown in the experiment:

The same predicted range implants, Sb implanted at 6 keV and B implanted at 2 keV, are contrasted in several plots. The intention was to eliminate that variable from our study because extracted variables like retained dosage, transmitted and backscattered ions will be highly reliant on implant energy and anticipated range.

The BF2 (at. mass 49) is more harmful than the B ion (at. Mass 11). Therefore, BF2 inflicts greater damage and has a narrower Gaussian profile shape. All of the discussion comparing B to Sb also applies to the comparison of B to BF2.

A) Backscattering:

In Backscattering, ions are expelled from the target after one or more collisions and are one of the outcomes of ion implantation. The simulation trends for the 2 keV and 6 keV Sb implants are displayed in Fig 2.

Fig 2: Simulations showing backscattered ions from 10000, (a) Sb implant at 6 keV,

 (b) B implant at 2 keV

Therefore, lighter targets are better for doping retention when hit by energetic ions as they suppress backscattering.

B) Transmission:

When ion implantation happens, some ions penetrate through the substrate and come out of the other side, known as Transmission. A significant decrease in transmission for higher incident angles can be observed in Fig 3. It is also observed that the transmission rate is primarily a factor of density and secondly affected by the atomic density of the target material.

Fig 3: Simulations showing transmitted ions from 10000 Sb ions implanted at 6 keV

Hence, the lighter and less dense targets are worse for doping retention when hit by energetic ions as they enhance transmission.

C) Retained Dose:

The retained dose can be calculated using the following formula: -

Retain dose = implant dose - (backscattered + transmitted)

The data for retained ions in the substrate after a 10000 ion Sb 6 keV implant and after a 10000 ion B 2 keV implant is shown in Fig 4. Dose retention depends on the angle of incidence. Heavier ions (Sb)are better for dose retention because, at a particular angle, lighter ions are more likely to depart the target due to transmission and backscattering.

Fig. 4: Simulations showing retained ions after (a) Sb implant at 6 keV, (b) B implant at 2keV

D) Damage generation:

Damage generation is the total energy loss of the incident ions. The incident ions can either transfer their energy to the target or can cause damage due to ionization (loss in energy caused by target electrons) and phonons (vibrations of atoms in the crystal lattice).

When the target lattice is bombarded with incident ions, some target atoms get displaced from their lattice position creating vacancies in the target. The number of vacancies depends on the displacement energy. If the energy of the atom that reverts has energy greater than the displacement energy, atoms will collide with the atoms creating more vacancies.

Fig. 5: (a) Sb implant at 6 keV and (b) B implant at 2 keV

On increasing the angle, the normal implant energy is reduced by a factor of the cosine of the incident angle which generates fewer vacancies.

To achieve a greater number of vacancies.

(i) Higher mass of the incoming ion

(ii) Higher energy of the incoming ion

(iii) Target material should have low displacement energy

E) Sputtering:

The sputtering process is the forceful ejection of material from a solid surface due to energy exchange by the bombardment of ions. The ejected target material gets deposited on the substrate when the energy of the ion is greater than the surface binding energy of the substrate.

The width of the target material tells the number of atoms sputtered.

Sputtered Width = 

Fig 6 shows the simulation results for sputtering from B and Sb implants. At high incident angles sputtering decreases.

Fig. 6: Sputtered width after (a) Sb implant at 6 keV (b) B implant at 2 keV

As SBE is inversely proportional to sputtering, the yields in GaAs and GaN are higher than  Ge, Si, and C with low SBE and atomic density. We can reduce the ejection of the target atoms by masking the target which means the amount of energy to get the desired profile of impurities should increase to be able to pass through the mask. Heavier ions, higher energy implants, low SBE, and low atomic density result in greater sputtering yield.

Fig. 7: Sputtering yield of Ga and As in GaAs due to an Sb 6 keV implant.

Limitations of SRIM software while performing the experiment:

The experimental data generated by SRIM software is not accurate for newer materials as compared to Si. To get accurate modeling, atomic density, and average atomic mass should be known.

BCA (Binary collision approximation) helps us to know the collisions between incoming ions and the target atoms. SRIM and BCA records are not precise about the effect of damage as they consider the target to be amorphous. This model is limited to simulating only 2D structures. Simulation of 3D structures gives a complete geometrical description of devices. To get an accurate model one should take care of the thermal budget stability, dopant activation limits, clustering, and surface trapping.

Conclusion:

Throughout the investigation, it was learned how the attack of energetic ions on a target leads to sputtering, dose retention, and damage formation generated in thin-body semiconductors and how they change with change in the target physical material properties.

Considering the target material, we found that less dense materials have an optimum implant angle of 30° for dose retention.  It was seen that denser materials have a wider range of implant angles than less dense materials and higher lattice binding energy indicated more damage generation. Low SBE and low atomic density result in target removal even for moderate implant doses, energies, and species. Apart from this, changes were also observed in the composition of alloys when binary and ternary alloys are subjected to preferential sputtering.

References:

[1] L. Pelaz, R. Duffy, M. Aboy, L. Marques, P. Lopez, I. Santos, et al., "Atomistic modeling of impurity ion implantation in ultra-thin-body Si devices," Electron Devices Meeting, pp. 1-4, 12/15 2008.

[2] C. Borschel and C. Ronning, "Ion beam irradiation of nanostructures – A 3D Monte Carlo simulation code," Nuclear Instruments and Methods in Physics Research Section B: Beam

Interactions with Materials and Atoms, vol. 269, pp. 2133-2138, 10/1/ 2011.

[3] M. J. H. van Dal, G. Vellianitis, B. Duriez, G. Doornbos, H. Chih-Hua, L. Bi-Hui, et al., "Germanium p-Channel FinFET Fabricated by Aspect Ratio Trapping," Electron Devices, IEEE Transactions on, vol. 61, pp. 430-436, 2014.

[4] M.J.H. van Dal, B. Duriez, G. Vellianitis, G. Doornbos, R. Oxland, M. Holland, et al., "Ge n-channel FinFET with Optimized Gate Stack and Contacts," IEDM, 2014.

[5] N. Waldron, C. Merckling, L. Teugels, P. Ong, S. A. U. Ibrahim, F. Sebaai, et al., "InGaAs Gate-All-Around Nanowire Devices on 300mm Si Substrates," Electron Device Letters, IEEE, vol. 35, pp. 1097-1099, 2014.