Silicon Nanostructuring Using SF 6 / O 2 Downstram Plasma Etching : Morphological , Optical and Sensing Properties

Silicon (Si) nanostructures were prepared in the downstream of radiofrequency SF6/O2 mixture plasma generated in 13.56 MHz hollow cathode discharge system. Depending on the oxygen percentage in the mixture, the obtained Si nanostructures were characterized for their different properties: etching rate, morphology, optical reflectance, photoluminescence, spectral response and humidity sensing. It is found that the etching rate exhibits a maximum value when the O2 ratio reaches 5%. An interesting defect-induced "violet" luminescence is reported from the Si nanostructures, whose intensity depends on their density. The obtained Si nanostructures have shown to induce a spectral response (SR) enhancement, in comparison with a smooth Si substrate, of about 100 times at 1100 nm wavelength. A very short response time (1 sec) to the humidity was measured for 5% O2 in the SF6/O2 plasma mixture, which was found to be well-correlated with the porosity of the Si nanostructures.


Introduction
5][6][7][8] Plasma is a versatile and environmentally friendly technique which uses gases as reactants to modify materials surface.Gas discharges environments have active species such as ions, free radicals, electrons, molecular fragments and photons that are simultaneously generated and can induce surface modification of different materials.Etching is well-known as a consequence of plasma surface activation that results in morphological changes.SF 6 /O 2 plasma mixture, providing the highly reactive etchant fluorine atoms, is widely used for silicon surface nanostructuring, [9][10][11][12][13][14] where the resulted Si nanostructures exhibited remarkable enhanced antireflective surface and light trapping efficiency, which is of great importance for photovoltaic applications to enhance the light absorption in Si solar cells. 10,11,13he technological motivation behind our present work is to investigate, besides the light anti-reflection property, other potential applications of the Si nanostructures using the maskless SF 6 /O 2 plasma etching, such as light emitting (photoluminescence), photo-sensing (spectral response) and humidity sensing.In addition, this contribution will focus on the possibility of tuning these properties through varying the percentage of oxygen in SF 6 /O 2 plasma mixture, and on the relationship of different obtained properties to the etching rate, the surface morphology and the SiO x F y self-masking layer formation.SF 6 /O 2 plasma mixture will be generated in a low-pressure 13.56 MHz hollow cathode discharge (Plasma Consult GmbH PlasCon HCD-L300 System), which generates an intense (high density) primary plasma and a very homogeneous remote plasma. 7

Experimental Procedure
The experimental set-up of the HCD-L-300 system is described elsewhere. 7A grounded stainless-steel substrate holder (70 x 70 x 1 cm 3 ) was located downstream at a distance of 45 mm below the plasma source.Four silicon wafers n-type Si (<100>, ρ ~1-20 Ω.cm, thickness ~ 500 µm) were etched in SF 6 /O 2 plasma mixture during 30 min under experimental conditions described in Table 1.The applied RF power and the total pressure are fixed at 300 W and 50 Pa, respectively.The SF 6 gas flow rate is fixed at 200 sccm, while the flow rate of O 2 gas was varied as follows: 0 sccm, 10 sccm, 20 sccm and 30 sccm; thus the samples will be given the names SO-0, SO-10, SO-20 and SO-30 accordingly.The etch rate  was determined from the measurement of the etching depth determined from the weight loss of the silicon wafer after etching using a microbalance.The surface morphology of the etched Si was studied using Atomic Force Microscope (AFM -Park Scientific Instruments AP-0100).A Jobin-Yvon, Triax 550, UV/VIS/NIR computerized spectrophotometer is used to measure the optical reflection of the plasma etched silicon wafers, in the condition of near normal incidence (angle of incidence ≅ 1.5º) and within the spectral range 400-1200 nm with a step of 1 nm.The spectral response (SR (A/W) = short current density/incident light power flux) of the bulk Si and the obtained Si nanostructured is measured using a Tungsten-Halogen lamp as a light source, a spectrophotometer (Triax 550), and an automated data acquisition Keithley 237 electrometer in the spectral range of 400-1200 nm.The PL spectra were recorded at room temperature (RT) using a 1 m monochromator (SPEX) equipped with a side-on photomultiplier tube R928, where a He-Cd laser (wavelength of 325 nm) was used as an excitation source.The humidity sensing property of the nanostructured surfaces was performed by exposing the samples to a humid air (RH% = 80%) and recording the dynamical electrical response to obtain the response time of the sensor, while a constant voltage of 10 V is applied to the samples.

Etching rate
After measuring the etching depth, the etching rate of silicon in our SF 6 /O 2 plasma is plotted as a function of O 2 percentage in the discharge in Figure 1.It can be seen that it reaches a maximum value (1450 nm/min) at 5% of O 2 in the mixture.To have an idea about the correlation of the etching rate behavior with the etchant atomic fluorine concentration, we performed measurements of both atomic fluorine [F] and atomic oxygen concentrations [O] in our SF 6 /O 2 plasma discharge using the actinometry optical emission spectroscopy as described in details previously, in 7 for [F], and in 15 for [O].The concentration curves, shown in Figure 2, exhibit a maximum for [F] at about 23% of O 2 in the discharge, and a monotonic increase of [O] with increasing of the oxygen percentage.Knizikevicius 16 performed a simulation of Si etching in SF 6 /O 2 plasma and found that at 27% O 2 , the concentration of F atoms approaches the maximum value.From the two figures 1 and 2, one can conclude that, in our case, the decrease of the etching rate beyond 5% O 2 is not related to the decrease of F radicals concentration in the plasma.It is more probably related to the formation of the SiO x F y passivation layer, whose thickness increases with the oxygen content in the discharge. 9Indeed, it is demonstrated that F and O react with Si to form the selfmasking layer SiO x F y on the surface, which acts as etching inhibitor at certain points on the surface. 4This passivation layer is not very stable and dissociates by heating or under the ion bombardment of the plasma. 9,10Since our system is a remote plasma configuration, where the concentrations of ions and electrons are relatively low and the concentration of free radicals is relatively high in the remote region far from the primary plasma, 17 the behavior of the etching rate as a function of O 2 content in the SF 6 /O 2 plasma, may be related to the fact that the passivation layer resists the degradation by plasma ions bombardment in the downstream etching region.Thus, this layer plays a significant role in the etching rate lowering beyond 5% O 2 content.The competition between the formation of SiO x F y passivation layer and the Si etching by F atoms generally leads to the appearance of a maximum in Si etching rate as a function of O 2 content.The maximum value of the silicon etching rate depends on different operation parameters, such as the design of the plasma system, working pressure, applied power and O 2 content.In a Corial 200 IL operating at a frequency of 13.56 MHz (100 W), Wongwanitwattana et al. 12 obtained a maximum in Si etching rate (450 nm/min) for 5% O 2 in SF 6 / O 2 mixture and a working pressure of 2.6 Pa.D'Agostino et al. 18 found a maximum in Si etching rate (about 400 nm/min) for 40% O 2 content in SF 6 /O 2 capacitively coupled plasma SO-30 (d), together with their texture line profiles taken along the diagonal of the AFM image, which are traced for the same height scale [-500 , 500 nm].These images reveal the formation of voids and roughened surfaces with different sizes of granules ranging from 80 nm to 1 µm.The variation of the total roughness (using Gwyddion software), the evolution of the mean pore and particle sizes (Diameter/Depth, Height), the porosity (the amount of voids within the Si nanostructured layer), and the particle density as a function of O 2 content in the SF 6 /O 2 plasma are listed in Table 2.It can be noticed that, except of particle density, all these parameters has a maximum for the sample SO-10 (5% O 2 ), for which the etching rate has also a maximum.The correlation, in our case of RF remote plasma, between the etching rate and the roughness can be explained as following: For 5% of O 2 in the gas mixture, the SiO x F y layer is relatively thin and large denuded areas of the surface are expected.This results in a considerably higher etching rate and higher accompanied roughness induced by the etching of the large unpassivated areas (Fig. 3b).For higher O 2 content, the thickness of the passivation layer is increased. 9Thicker passivation layers lead to more efficient inhibition of the etching, reducing the etching rate (Figure 1).This should also be accompanied with the reduction in the roughness, since it results from the etching process.

Optical reflectance
Figure 4 displays the measured optical reflectance spectra of the etched Si samples (Table 1) in addition to a smooth Si substrate in the spectral range [400-1200 nm].The inset shows the variation of average reflectance as function of O 2 percentage in the discharge.A significant decrease in the reflectance of the silicon surface after plasma etching can be observed, where the lowest reflectance (about 0.01%) was obtained for the sample SO-10.The evolution of optical reflectance as a function of O 2 percentage seems to be inversely proportional to the roughness of prepared samples, as can be illustrated in Figure 5 (average reflectance as a function of roughness).This means that the more surface is rough, the more light is being trapped in the Si nanostructure.This result was also demonstrated by other research groups. 10,11,13

Photoluminescence
Figure 6 exhibits the room temperature photoluminescence spectra of the etched Si samples listed in Table 1.The PL characteristic is "violet".The peaks positions are centered around 426 nm.It can be seen that there is no significant shift in the wavelength of the PL peak as a function of O 2 percentage in the SF 6 /O 2 plasma mixture in spite of the significant difference in the particle size of the structures shown in Table 2. Therefore, the origin of this violet luminescence can't be attributed to quantum confinement effects. 203 Figure 7 shows the evolution of the integrated PL intensity as a function of the particle density of the Si nanostructures, the PL intensity has a minimum for the sample SO-10 (5 % O 2 ), which corresponds to the lowest particle density (Table 2), the maximum integrated intensity is observed for the sample SO-0 of the highest particle density.This behavior of PL intensity is probably due to the enhancement of the  1) Figure 6.PL spectra of prepared samples (Table 1)

Spectral response
Figure 8 shows the evolution of the spectral response (SR) of a smooth Si substrate, SO-0, SO-10, SO-20 and SO-30 samples.A significant enhancement (about 100 times) of the SR, as compared to the reference sample (Bulk Si) after etching, can be observed around the band gap of Si (1100 nm ~ 1.12 eV).We attribute this significant enhancement of SR to the surface roughening of etched Si, which reduces Silicon Nanostructuring Using SF 6 /O 2 Downstram Plasma Etching: Morphological, Optical and Sensing Properties light reflection and thus improves light trapping. 11,13,14At this wavelength, the SR changes depending on samples preparation conditions.It has a maximum at 5% O 2 (sample SO-10) that has the lowest reflectance (Figure 4).The inset of Figure 8 shows clearly an increase of SR with the decrease of average reflectance.These results are of great interest for photo-sensing applications and efficiency enhancement of silicon solar cells, where one can find some reported results about the enhancement of silicon solar cells performance after surface etching in SF 6 /O 2 plasma.Yoo et al. 11 reported a correlation between c-Si solar cells parameters and the reflectance after etching in SF 6 /O 2 plasma, where the cell efficiency was increased to be about 15.7%, for the ratio (SF 6 /O 2 =2), compared to 14.5% for the ratio (SF 6 /O 2 =3).Lin et al. 14 have successfully fabricated large-area (156 x 156 mm 2 ) anti-reflective (reflectance < 15%) surfaces of polycrystalline solar cells by self-masking RIE process using SF 6 /O 2 /Cl 2 plasma etching, where the solar cell efficiency was improved to be 16.27% comparing to the value of 15.56% when wet etching was used.

Humidity sensing property
By similarity to the Si nanostructures fabricated electrochemically, 24,25 the formation of pores in our nanostructures may result in good humidity sensing properties, which is of considerable importance in the field of silicon humidity sensors technology.
Figure 9 shows the dynamical electrical response to relative humidity (RH%=80%) of different prepared samples.The point (t= 0 sec) corresponds to the moment of sample exposure to the humidity, and the response time (t r ) corresponds to the moment at which the electrical resistance begins to be stable.As seen from the figure, the electrical resistance decreases when the samples are exposed to humidity, except of the sample SO-30.The sensitivity (S) was calculated using the following formula: (1)   where R h is the sensor resistance influenced by the humidity, and R a is the sensor resistance in the ambient air.Table 3 summarizes the obtained response time and sensitivity of samples.
It can be seen that both t r and S are influenced by the plasma preparation conditions of the Si nanostructures.Referring to AFM morphology data of Table 2, one can notice that the shortest response time (1 sec) corresponds to the highest porosity (sample SO-10).Figure 10 shows a clear correlation between the response time and the porosity.The long response time (120 sec) of the sample SO-30 (13% O 2 ) isn't only related to the lowest porosity, but also to a possible  1) and a bulk, smooth silicon wafer.The inset shows the plot of SR values at 1100 nm as a function of average reflectance

Conclusions
Silicon nanostructures have been successfully fabricated in downstream RF SF 6 /O 2 plasma excited in a hollow cathode discharge system.The prepared structures have been characterized for their different properties (etching rate, morphology, optical reflectance, photoluminescence, spectral response and humidity sensing) as a function of O 2 percentage in the discharge.The highest etching rate, highest roughness and porosity, lowest reflectance, highest spectral response and shortest response time to humidity were found for the sample etched in the plasma mixture with 5 % O 2 .The different properties of the Si nanostructures were found to be well correlated, where the evolution of the optical reflectance follows the surface roughness of the samples, the violet PL intensity increases with the particle density of the Si nanostructures, the spectral response is inversely proportional to the optical reflectance and the response time to the humidity depends on the porosity of the Si surface.The passivation layer SiO x F y seems to affect the etching rate, the PL intensity and the response time to humidity.

a
Physics Department, Atomic Energy Commission of Syria (AECS), P.O.Box 6091 Damascus, Syria

Figure 1 .Figure 2 .
Figure 1.Variation of silicon etching rate in SF 6 /O 2 plasma as a function of O 2 percentage Figure 2. Variation of F atoms and O atoms concentration as a function of oxygen percentage in SF 6/ O 2 plasma

Figure 4 .Figure 5 .
Figure 4. Optical reflectance spectra of etched Si samples (Table 1) and smooth silicon crystal.Inset: Variation of the average reflectance as a function of O 2 percentage in SF 6 /O 2 plasma

Figure 7 .
Figure 7. Variation of PL intensity of the fabricated Si nanostructures in the SF 6 /O 2 plasma as a function of the particle density

Figure 8 .
Figure 8. Spectral response (SR) spectra of etched Si samples (Table1) and a bulk, smooth silicon wafer.The inset shows the plot of SR values at 1100 nm as a function of average reflectance

=
passivation effect, where the passivation layer causes slow kinetics of absorption and diffusion for water molecules to pores, and this can explain also the abnormal behavior of the electrical resistance (Figure9) of this sample SO-30.The sensitivity decreases as a function of O 2 percentage, from 17 (SO-0) to 0.4 (SO-30).

Figure 10 .
Figure 10.Plot of the response time to humidity of samples as a function of their porosity

Table 2 .
AFM morphology parameters of the Si nanostructures formed by SF 6 /O 2 plasma etching as a function of O 2 percentage.(CCP)driven by 27 MHz RF generator (45 W) at working pressure of 131 Pa.Zou 19 measured a maximum in Si etching rate (3000 nm/min) for 30% O 2 in SF 6 /O 2 plasma and at a pressure of 26 Pa in an Oxford Instruments PlasmaLab 80+ reactive ion etcher with 13.56 MHz RF coupling (160 W).

Table 3 .
Response time and sensitivity of the Si nanostructures exposed to 80% humidity as a function of O2 percentage in SF6/ O2 plasma.