The selective etching process is widely used for achieving the desired etch rate in semiconductor fabrication. Parameters such as input power, operating pressure, gas mixture, chamber geometry, and amplitude of the radio-frequency voltage govern the etch rate and etch quality in plasma. In this work, we experimentally investigated the optimum plasma etching conditions required to achieve an anisotropic etch profile and analyzed how the optimum etching can be carried out using an appropriate operating pressure and oxygen concentration. Optical emission spectroscopy was used to measure the concentrations of oxygen and fluorine, and Langmuir probe was used to measure the electron density in the plasma. The oxygen concentration was varied from zero to 100 vol. % for pressures in the range of 20–600 mTorr. The optimum etch conditions are used to study the ion energy distribution given by Kawamura et al., Plasma Sources Sci. Technol. 8, R45 (1999). The results suggest that in addition to O2% and pressure, the DC bias is another crucial parameter for achieving the optimum etch conditions.

The rapid development of technologies that use semiconductors has led to enhanced requirements for the synthesis of optimized semiconductors, such as in terms of the dimensions of the material and surface properties. The effectiveness of the etching process depends critically on the etch profile, e.g., a microelectromechanical system requires an anisotropic etch profile.

Ongoing research on plasma etching1–6 faces several challenges to attainetch uniformity, anisotropic etch profile and plasma operating conditions. For example, change in input power, pressure, and flow rate can influence the ion density and the number of free radicals in the plasma. Consequently, we see changes in the etch rate and etch profile. There are very few studies carried out in last few decades those have linked the discharge parameters to the etching results. Bates et al.7 form such linkages between silicon etch properties and SF6/C4F8/Ar plasma. Zou et al.8 investigated the etch profile for low (30 mTorr) and high pressure (200 mTorr) cases. The results suggest that O2 concentration is the principal factor influencing the etch profile, while the etch rate is strongly affected by the system pressure. Morshed et al.9 studied the effect of oxygen flow rates on the fluorine density in SF6 plasma. The results show that the etch profiles can be controlled by changing the flow rate of oxygen in the plasma. Knizikevičius et al.10 simulated Si and SiO2 etching in SF6–O2 plasma. In our previous work,11 we experimentally investigated the anisotropic etch profile in two geometrically different capacitively coupled plasma sources. The results obtained indicate that the etch profile varies as a function of F and O densities in the plasma chamber. The peak etch rates for both chambers were obtained at 20 vol. % oxygen concentration in the plasma.

In the current work, we extended our research study and investigated the variation in the optimum conditions as a function pressure, O2%, and DC bias. We found that the optimum etch profile can’t be obtained using a single recipe of operating conditions i.e. optimum conditions varies if any single parameter in plasma has changed. The detailed investigation about this feature of the reactive plasma is presented in the work.

The experimental setup used in this study is shown in Fig. 1. A capacitively coupled plasma system, Oxford Instruments Plasma 100 (hereinafter referred to as OIP 100), manufactured by Oxford Instruments, UK, was used.12 The device consisted of two horizontal, parallel stainless-steel electrodes. The bottom electrode was a plate with a diameter of 300 mm, powered at a radio frequency (RF) of 13.56 MHz. The distance between the electrodes was 53 mm. A built-in load lock system of the OIP 100 allowed for the loading of silicon wafers without venting the chamber.

FIG. 1.

Experimental setup of the OIP 100, data reproduced with permission from Alshaltami et al., J. Vac. Sci. Technol. A 35, 031307 (2017). Copyright 2017 American Vacuum Society.

FIG. 1.

Experimental setup of the OIP 100, data reproduced with permission from Alshaltami et al., J. Vac. Sci. Technol. A 35, 031307 (2017). Copyright 2017 American Vacuum Society.

Close modal

A capacitively coupled plasma (CCP), discharge was generated by a 13.56-MHz RF power source with an L-type matching network in the chamber.12 N-type silicon <100> wafers were patterned by a positive photoresist via photolithography.13 The RF power, gas flow rate, and process duration were kept constant at 200 W, 50 sccm, and 5 min, respectively. The oxygen concentration was altered from 0 to 100 vol. % for each value of pressure. The pressure was increased from 20 mTorr to 600 mTorr.

Scanning electron microscope (SEM) was used to analyze the sample surface. The precise composition of the plasma interacting with the surface of the material and the densities of the main reactive species needed to be known to achieve an anisotropic trench. The electron density was measured with a Langmuir probe.14,15 The plasma emission spectra were collected with an Optical emission spectroscopy16 at a resolution of 1.5 mm. The optical fiber was positioned at a quartz window. Emission intensities of 703 nm and 750 nm were recorded for F and Ar, respectively, as a function of O2 concentration for different values of pressure in the SF6–O2 plasma system. Argon was used as an inert gas at a concentration of 4 vol. %. The small amount of Ar had no significant effect on the emission lines of the plasma species.17,18

Ion energy plays an important role in achieving the desired anisotropic profile.19 This is because ion flux and ion energy are mainly controlled by the sheath adjacent to the substrate. The sheath thickness is the main parameter that can alter the path of the ions approaching the substrate and that can result in uniformity in the etching. The sheath thickness depends on the sheath electric field and the ion mean free path, which can be controlled by varying the amplitude of the RF voltage or by changing the DC bias at the substrate electrode. Using these operating parameters, we can optimize the ion energy at the substrate and hence the etching profile.

The sheath voltage waveform is periodic in case of RF plasma and, therefore, the energy E of an ion hitting the target depends on the phase angle Θ at which the ion enters the sheath.20 For a low-frequency regime, i.e. τionrf <<1, where τion is the ion transit time through sheath and τrf is the time period of the RF cycle. Under this condition, the ion energy distribution (IED) is given by,20 

IED=1πeVdc2E212+12δE,
(1)

where E=eVdcsinΘ and Vdc is the DC self-bias voltage. The DC self-bias voltage plays an important role in etching. For instance, a higher DC bias leads to an increase in the ion energy.

In our experiments, we collected the reading data of the DC bias voltage by varying the O2% and pressure. Fig. 2 and Fig. 3 shows DC bias data. The DC bias decreases exponentially with pressure while it increases linearly with O2%. The ion energy decrease due to the fact that DC bias decreases with pressure as higher pressure results in more collisions, and therefore, the ions will lose energy. On the other hand, the bias voltage increases with O2% because the ion energy is proportional to the mass (α m-1/2).21 Since oxygen is lighter than fluorine, we observed the increase in the DC bias in this case.

FIG. 2.

DC bias vs. pressure.

FIG. 2.

DC bias vs. pressure.

Close modal
FIG. 3.

DC bias vs. % O2.

FIG. 3.

DC bias vs. % O2.

Close modal

Optimum etch rate is defined as etch rate corresponding to maximum etch achieved at a given plasma condition as shown in Fig. 4.

FIG. 4.

Schematic of the etch profile where OC means optimum etch condition achieved under a given plasma condition, d is the depth and δ is the lateral depth.

FIG. 4.

Schematic of the etch profile where OC means optimum etch condition achieved under a given plasma condition, d is the depth and δ is the lateral depth.

Close modal

In Fig. 5, we have presented the DC bias data points as a function of pressure corresponding to the optimum etch rate condition. Clearly, the DC bias for optimum etch condition is lower for higher pressure and high O2% and higher for lower pressure and low O2%. This may be due to the decrease in the ion energy at high pressure because of higher collisions in the plasma.

FIG. 5.

DC bias vs. pressure for different % O2 for optimum etch rate.

FIG. 5.

DC bias vs. pressure for different % O2 for optimum etch rate.

Close modal

Using Eq. (1), we calculated the IED for different values of DC bias and the results are plotted in Fig. 6. The IED obtained is bimodal in nature where the two peaks correspond to the maximum and minimum sheath drops where the voltage varies slowly. We found that the peak corresponding to the maximum energy shifts toward the lower energy with increase in pressure; in other words, the broadness of the IED decreases with pressure. However, the ion population corresponding to the maximum energy peak increases with pressure.

FIG. 6.

IED vs. energy for different pressure and % O2.

FIG. 6.

IED vs. energy for different pressure and % O2.

Close modal

The sheath thickness around the substrate electrode is given by Liebermann22 

S=23ε0ens1/22VsTeV3/4,
(2)

Where Vs is voltage across the sheath, e is electron charge, ε0 is vacuum permittivity and Te is electron temperature, and ns is the sheath density = 0.61np. The above equation can be simplified further as follows

S=7544Vs3/4np1/2TeV1/4,
(3)

It is clear from the above equation sheath thickness is the factor of plasma density, electron temperature and sheath voltage.

Optical emission spectroscopy was used to measure the density of the F atom from the intensities of the 703 nm and 750 nm lines.23 The relative intensities of the F atom (IF/IAr) are shown as a function of % O2 in Fig. 7.

FIG. 7.

Relative intensity of the F atom as a function of % O2 in SF6–O2 plasma.

FIG. 7.

Relative intensity of the F atom as a function of % O2 in SF6–O2 plasma.

Close modal

The density of the F atom was related to the intensities of the atomic emission lines of F and Ar as described by Alshaltami et al.11 The estimated density of the F atom can be expressed as:

nF0.47IFIArnAr,
(4)

where nAr is the known density of argon and IF/IAr is the ratio of the corresponding intensities of the F atom (703 nm) to that of the Ar atom (750 nm). The densities of fluorine are plotted in Fig. 8.

FIG. 8.

Values of fluorine densities as a function of % O2 at different pressures.

FIG. 8.

Values of fluorine densities as a function of % O2 at different pressures.

Close modal

The density of F and O atom as a function of power for different set of O2% is shown in Fig. 9 and Fig. 10 respectively. We found that both F and O density increases with the applied power.

FIG. 9.

The relative F atomic density as a function of power at different O2 concentration in 200mT.

FIG. 9.

The relative F atomic density as a function of power at different O2 concentration in 200mT.

Close modal
FIG. 10.

The relative O atomic density as a function of power at different O2 concentration and 200mT.

FIG. 10.

The relative O atomic density as a function of power at different O2 concentration and 200mT.

Close modal

Figure 11 shows the etch rate as a function of oxygen concentration for different pressures. The etch rate was found to vary parabolically for different pressures. For instance, the etch rate is maximum at 20 vol. % O2 level when the pressure is 20 mTorr. Further, the peak etch rate occurs at higher oxygen concentrations on further increase in the pressure to 600 mTorr. This is due to the fact that addition of oxygen increases the conversion of SF6 by reacting with fluorosulfur radicals [Refer Table I: Production of F from O2] and thus prevent their recombination with fluorine to reform SF6.1 This leads to a net increase in the F concentration. However, at higher oxygen level, i.e., above 20 vol. % for the 20 mTorr case, oxygen becomes more dominant over fluorine and, therefore, SF5 recombines with F to form SF6 [Refer Table I: Destruction of F]. The lower pressure in the plasma results in a longer mean free path, thus limiting the collisions, whereas higher pressure results in a shorter mean free path, which promotes the collisions in the plasma. At higher pressures, the level of SF5 density increases, and to compensate, the oxygen level has to be increased to achieve the desired etch profile. Therefore, the peak etch rate occurs at higher O2 level for high pressures.

FIG. 11.

Etch rate as a function of % O2 for different pressures.

FIG. 11.

Etch rate as a function of % O2 for different pressures.

Close modal
TABLE I.

List of probable chemical reactions in CCP dry etch system.10 

Electron-impact ionization Production of F from O2%
S F 6 + e SF 5 + + F + 2 e   SF5 + O2 → SO2F2 + 3F 
S F 5 + e SF 4 + + F + 2 e   SF5 + O2 → SO2F2 + F2 + F 
S F 5 + e SF 5 + + 2 e   SF4 + O2 → SO2F2 + 2F 
S F 4 + e SF 3 + + F + 2 e   SF4 + O2 → SO2F2 + F2 
S F 4 + e SF 4 + + 2 e   SF3 + O2 → SO2F2 + F 
S F 3 + e SF 2 + + F + 2 e   SF5 + O → SOF2 + 3F 
S F 3 + e SF 3 + + 2 e   SF5 + O → SOF2 + F2 + F 
O2 + e → O+ + O + 2e  SF4 + O → SOF2 + 2F 
O 2 + e O 2 + + 2 e   SF4 + O → SOF2 + F2 
Electron-impact dissuasion   SF3 + O → SOF2 + F 
SF6 + e → SF5 + F + e  Destruction of F  
SF5 + e → SF4 + F + e  SF5 + F → SF6 
SF4 + e → SF3 + F + e  SF4 + F → SF5 
SF3 + e → SF2 + F + e  SF3 + F → SF4 
O2 + e → O + O + e  SF2 + F → SF3 
Dissociative Attachment   Surface reaction  
S F 6 + e SF 5 + F   Si + 4F → SiF4 
S F 5 + e SF 4 + F   Surface oxidation  
S F 4 + e SF 3 + F   Si + 2O → SiO2 
S F 3 + e SF 2 + F   Combination  
O2 + e → O + O  SiO2 + F → SiO2
Electron-impact ionization Production of F from O2%
S F 6 + e SF 5 + + F + 2 e   SF5 + O2 → SO2F2 + 3F 
S F 5 + e SF 4 + + F + 2 e   SF5 + O2 → SO2F2 + F2 + F 
S F 5 + e SF 5 + + 2 e   SF4 + O2 → SO2F2 + 2F 
S F 4 + e SF 3 + + F + 2 e   SF4 + O2 → SO2F2 + F2 
S F 4 + e SF 4 + + 2 e   SF3 + O2 → SO2F2 + F 
S F 3 + e SF 2 + + F + 2 e   SF5 + O → SOF2 + 3F 
S F 3 + e SF 3 + + 2 e   SF5 + O → SOF2 + F2 + F 
O2 + e → O+ + O + 2e  SF4 + O → SOF2 + 2F 
O 2 + e O 2 + + 2 e   SF4 + O → SOF2 + F2 
Electron-impact dissuasion   SF3 + O → SOF2 + F 
SF6 + e → SF5 + F + e  Destruction of F  
SF5 + e → SF4 + F + e  SF5 + F → SF6 
SF4 + e → SF3 + F + e  SF4 + F → SF5 
SF3 + e → SF2 + F + e  SF3 + F → SF4 
O2 + e → O + O + e  SF2 + F → SF3 
Dissociative Attachment   Surface reaction  
S F 6 + e SF 5 + F   Si + 4F → SiF4 
S F 5 + e SF 4 + F   Surface oxidation  
S F 4 + e SF 3 + F   Si + 2O → SiO2 
S F 3 + e SF 2 + F   Combination  
O2 + e → O + O  SiO2 + F → SiO2

Morshed et al experimentally investigated that electron density increases with applied power and it is correlated to the concentration of Fluorine.9,24 The increase in power increases the electron density and hence increases the F density due to the electron collision with the SF6 gas. This will result in decrease in the lateral depth. However, we also expect a degradation of photo resist due to increase in power which produces hydrogen gas. This is the only source of hydrogen in the chamber as we are not using any gas producing hydrogen.

On the other side, oxygen concentration increases with applied power. The increase in oxygen concentration execute the oxidation of Si to SiO2 [Refer Table I: Surface oxidation]. Therefore, the balanced mixture of O2 and SF6 is required to achieve the vertical sidewall etching. This balanced mixture may depends on various factors such as power, pressure, DC bias and O2 concentration for optimum vertical sidewall etching.

Figure 12 shows the electron density versus oxygen level for different pressures. The peak value of ne is obtained at 20 mTorr and 20 vol. % O2. As expected, the maximum electron density shifted to a higher O2 concentration with increasing pressure. This result is in good agreement with the profile obtained for F density and etch rates.

FIG. 12.

Electron density as a function of the fraction of O2 corresponding to the pressure values observed in the SF6–O2 plasma system.

FIG. 12.

Electron density as a function of the fraction of O2 corresponding to the pressure values observed in the SF6–O2 plasma system.

Close modal

We collected all the peak etch rate points from Fig. 11 and plotted them in Fig. 13 as a function of pressure. We found that the amount of etch rate is highest for 200 mTorr and 30 vol. % O2 level. This is because at pressures above 200 mTorr, there will be significant electron-neutral collisions in the plasma, which will reduce the density and thermal energy of F and O to perform the required etch process.

FIG. 13.

Optimum of the etch rate as a function of pressures for different oxygen concentrations.

FIG. 13.

Optimum of the etch rate as a function of pressures for different oxygen concentrations.

Close modal

Figure 14 shows the lateral depth versus oxygen concentration at different pressures. For the optimum etch rate, the lateral depth should be minimum to achieve an anisotropic etch profile. For instance, 20 vol. % O2 results in a maximum etch rate at 20 mTorr, and the minimum lateral depth obtained under the same plasma conditions is as shown in Fig. 14. This result is consistent for all different sets of operating pressures.

FIG. 14.

Values of δ/d for each profile as a function of the fraction of oxygen in SF6–O2 discharge.

FIG. 14.

Values of δ/d for each profile as a function of the fraction of oxygen in SF6–O2 discharge.

Close modal

Figure 15 shows the anisotropic etch profile obtained at 100 mTorr and 25 vol. % oxygen concentration. The ratio δ/d for different plasma conditions are plotted in Fig. 14.

FIG. 15.

Etch profile obtained at 100 mTorr and 25 vol. % oxygen concentration, where d is the depth and δ is the lateral depth.

FIG. 15.

Etch profile obtained at 100 mTorr and 25 vol. % oxygen concentration, where d is the depth and δ is the lateral depth.

Close modal

After certain level of O2 [example: O2 – 20 vol. % at 20mT], F concentration decreases with further increase in O2 concentration as shown in Fig. 8. One of the mechanism is that SF5 recombines with F to form SF6 due to the dilution effect causing reduction in F concentration.

Another reason for the reduction of F could be due to the resonance enhanced quenching of the Argon transition 750 nm which is correlated to the electron density. Morshed et al.9 found that the electron density decreases with increased O2 dissociation rate [above O2=30 vol. %] in the plasma. Argon provides stability and enhances anisotropic etching and it decreases with decrease in the electron density in the plasma. Therefore, this will lead to the reduction of Fluorine in the discharge.

The third possible factor for reduction in the F could be the ion energy. The ion energy is inversely related to the mass. Therefore, oxygen energy in the oxygen rich plasma would be more dominating than the F due to its lighter mass. Due to that, the plasma becomes more electronegative and we know that electron density decreases with increase in electronegativity in the plasma.

We collected the data points from the high-energy peaks from Fig. 6 and plotted in Fig. 16 as ion population (IP) versus energy for optimum O2% and pressure. The result shows the nonlinear trend between the population and energy, confirming the inverse correlation between the two parameters. We performed the nonlinear curve fitting method and obtained an equation, as given below.

IP=0.004+120.081+0.007ΔE3.1,
(5)

This equation and the graph provided in Fig. 16 provides information about the selective etching process, i.e., the pressure and O2% level required to achieve an optimum etch rate for a given ion energy.

FIG. 16.

Ion population vs. ion energy.

FIG. 16.

Ion population vs. ion energy.

Close modal

The plasma density is a function of pressure and O2% as shown in Fig. 8. Hence, the sheath thickness will also vary with pressure and O2% in the plasma. Similarly, Sheath voltage is a function of DC bias and DC bias changes with pressure and O2% as shown in Fig. 2 and Fig. 3 In accordance with our results, we found that as we increase the pressure for a given power and O2%, DC bias decreases, plasma density increases up to the optimum condition therefore the sheath thickness should decrease with the pressure as per Eq. (2) provided electron temperature is more or less constant.

In second case, increase in the O2% results in the decrease in the optimum level plasma density and increase in the DC bias, therefore the sheath thickness increases with O2%. Hence, we can conclude that the sheath thickness varies with the O2% and pressure and we need a definite thickness or voltage across sheath in order achieve the optimum etch rate in the system. This optimum condition has been achieved experimentally in our work and correlated with the analytical model of IED.

In this work, we studied the etch rate and etch profile in SF6–O2 plasma as a function of pressure, oxygen concentration, and DC bias voltage. We found that the optimum etch rate conditions strongly dependent on the operating pressure, DC bias and O2%. For instance, if the plasma conditions are 20mTorr, Vdc=370V and 20% O2 that gives the optimum etch rate. However, change in one parameter such as pressure can affect the whole process and therefore the other parameters such as Vdc and O2% has to be adjusted accordingly. On the other hand, the IED in RF plasma and its broadening can provide important information about the optimum etch conditions.

This study can be further benchmarked by measuring the IED using a diagnostic tool such as retarding field analyzer, RFA.25 

The Ministry of Higher Education and Scientific Research, Libya, is gratefully acknowledged for its financial support for this research. The generous support of National Centre for Plasma Science and Technology (NCPST), Dublin City University, Ireland is also gratefully acknowledged.

1.
R.
d’Agostino
and
D. L.
Flamm
,
J. Appl. Phys.
52
,
162
(
1981
).
2.
N.
Layadi
,
J. I.
Colonell
, and
J. T.
Lee
,
Bell Labs Tech. J.
4
,
155
(
1999
).
3.
V. M.
Donnelly
and
A.
Kornblit
,
J. Vac. Sci. Technol. A
31
,
050825
(
2013
).
4.
S.
Rauf
and
A.
Balakrishna
,
J. Vac. Sci. Technol. A
35
,
021308
(
2017
).
5.
A.
Picard
,
G.
Turban
, and
B.
Grolleau
,
J. Phys. D.
19
,
991
(
1986
).
6.
R.
Knizikevičius
and
V.
Kopustinskas
,
Vacuum.
77
,
1
(
2004
).
7.
R. L.
Bates
,
P.
Stephan Thamban
,
M. J.
Goeckner
, and
L. J.
Overzet
,
J. Vac. Sci. Technol. A
32
,
041302
(
2014
).
8.
H.
Zou
,
Microsystem Technologies.
10
,
603
(
2004
).
9.
M.
Morshed
and
S.
Daniels
,
Plasma Science and Technology.
14
,
316
(
2012
).
10.
11.
K. A.
Alshaltami
,
M.
Morshed
,
C.
Gaman
,
J.
Conway
, and
S.
Daniels
,
J. Vac. Sci. Technol. A
35
,
031307
(
2017
).
13.
C. R.
Scott
and
P. M.
Troccolo
, US patent 5, 935,777 (
1999
).
14.
R. L.
Merlino
,
American Journal of Physics.
75
,
1078
(
2007
).
15.
D.
Smith
,
N.
Adams
,
A.
Dean
, and
M.
Church
,
J. Phys. D.
8
,
141
(
1975
).
16.
V.
Donnelly
,
M.
Malyshev
, and
M.
Schabel
,
Plasma Sources Sci. Technol.
11
,
A26
(
2002
).
17.
R.
d’Agostino
,
F.
Cramarossa
,
S.
De Benedictis
,
F.
Fracassi
,
L.
Laska
, and
K.
Mašek
,
Plasma Chem Plasma Process
5
,
239
(
1985
).
18.
J.
Conway
,
S.
Kechkar
,
C.
Gaman
,
M.
Turner
, and
S.
Daniels
,
Plasma Sources Sci. Technol.
22
,
045004
(
2013
).
19.
D.
Gahan
,
B.
Dolinaj
, and
M.
Hopkins
,
Plasma Sources Sci. Technol.
17
,
035026
(
2008
).
20.
E.
Kawamura
,
V.
Vahedi
,
M.
Lieberman
, and
C.
Birdsall
,
Plasma Sources Sci. Technol.
8
,
R45
(
1999
).
21.
M. A.
Lieberman
and
A. J.
Lichtenberg
,
Principles of plasma discharges and materials processing
, 2nd ed. (
Wiley-Interscience
,
Hoboken, NJ
,
2005
), pp.
445
446
.
22.
M. A.
Lieberman
and
A. J.
Lichtenberg
,
Principles of plasma discharges and materials processing
, 2nd ed. (
Wiley-Interscience
,
Hoboken, NJ
,
2005
), pp.
175
.
23.
J.
Coburn
and
M.
Chen
,
J. Appl. Phys.
51
,
3134
(
1980
).
24.
M. M.
Morshed
and
S. M.
Daniels
,
IEEE Trans Plasma Sci
38
,
1512
(
2010
).
25.
D.
Gahan
,
S.
Daniels
, and
C.
Hayden
,
Plasma Sources Sci. Technol.
21
,
024004
(
2012
).