**4. Ultrasonic cleaning of Si wafers**

Cleaning and conditioning of silicon wafer surfaces and Si/SiO2 interfaces for the manufacturing of photovoltaic and microelectronic devices are in increasing demand with improved performance, reliability of these devices, scaling down to below 10 nm, incorporating extended metallization layers, employing epitaxial layers of compound SiGe and III-V semiconductors [51]. Clean wafer surfaces are crucial in high efficiency solar cell as well as in Ultra-Large Scale Integration (ULSI) fabrication processes, fin-shaped field effect transistors (FinFET), 3D NANDstacked memory devices etc.

mechanisms, including mechanical vibration and appropriate pressure gradients, microcavitation bubbles that oscillate and dance around due to Bjerknes force [39, 61], acoustic streaming flows, etc. One of the most important aspects of using acoustic streaming is the effect of the frequency on the boundary layer [62]. Its thickness decreases and the streaming velocity increases with increasing the sonication frequency. These both remarkably increase the drag force and hence the particle removal efficiency. It is demonstrated that the acoustic streaming induced removal of foreign contaminants with sizes in the dozen nm range is accomplished at frequencies greater than 1 MHz, i.e. in a high-frequency sonication region [63]. The removal of contaminants having sizes down to ≈100 nm is possible at frequen-

*Ultrasonic Processing of Si and SiGe for Photovoltaic Applications*

*DOI: http://dx.doi.org/10.5772/intechopen.96939*

It has furthermore been demonstrated that the caustic etching process can be ultrasonically enhanced producing finer, more homogeneous textured surfaces [65]. This technique inhibits the sunlight reflection from the Si surface thus enhancing

To demonstrate the capabilities of high-power ultrasound in powerfully manipulating surface species, consider the data reported in **Figure 3**. Here, following the method of cleaning from laser-induced cavitation bubbles [66], the wafer surface is covered with a thin layer of grease. This yields the optical transmission spectra indicative of organic contaminants, which are marked by several peaks in spectra 2

and C–H stretching vibrations [–CH3 and –(CH2)n–] at 2850 cm<sup>1</sup> [67]. It is seen in spectra 3 that organic contaminants are effectively removed from the wafer surface upon its exposure to ultrasonic cavitation with the peak acoustic intensity of about

*FTIR spectra of n-Si (a) and p-Si (b) wafers, prior to surface greasing and applying the ultrasound (spectrum 1), covered with a thin layer of vaseline (2) and subsequently cleaned in an ultrasonic bath (3) during 15 min. Reprinted with permission from Podolian A, Nadtochiy A, Kuryliuk V, Korotchenkov O, Schmid J, Drapalik M, Schlosser V. Solar Energy Materials and Solar Cells. 2011;95:765–772. Copyright 2010, Elsevier.*

, –CH2– at 2920 cm<sup>1</sup>

, –CH at 2890 cm<sup>1</sup>

cies smaller than 1 MHz [64].

the performance of solar cells.

**Figure 3.**

**143**

of **Figure 3**, such as C–CH3 at 2960 cm<sup>1</sup>

Deposition of monolayers and self-assembly of nanoparticles in multilevel structures requires the wafer surface to be completely free of any particulate contamination down to a nanometer scale. Moreover, installing a 1 MWp solar module generator implies that more than 2<sup>10</sup><sup>5</sup> Si wafers must be processed.

Furthermore, metallic contaminants on Si wafers cause substantial increase in leakage current in silicon p-n junctions and decrease the oxide breakdown voltage thus deteriorating the minority carrier lifetime [52, 53]. In particular, Cu and Al contaminants worsen the gate oxide integrity [54, 55]. To achieve ultraclean substrate surfaces with high reproducibility, it is also important to note that not only removal of contaminants is effective in improving the cell performance, but also prevention of their redeposition on the wafer surface. In this respect, dilute HF can be effectively used. However, some Cu and Al residues can be found on the wafer surface due to its hydrophobic nature [56].

The contamination of wafer surfaces by particle contaminants is one of the major problems in the industries. One way to increase the yield on fully processed silicon wafers is to use cleaning techniques specifically efficient to remove particle contaminants. Small particles are especially difficult to remove because they are strongly bounded to the substrate by electrostatic forces. It is therefore very important to find an effective way to remove particles from wafers without causing damage to the wafers.

A wide variety of cleaning methods are being used in wafer manufacturing such as brush or water-jet scrubbing of wafer surfaces employed prior to further immersion-type cleaning, scrubbing of rotating wafer surfaces between each processing step, adding cheating agents to the solution aiming to avoid metal adsorption onto Si wafers, cleaning in wet chemical baths, post treatment rinsing and many others [57]. They, however, are known to damage the wafer surface. Moreover, the chemical-type cleaning has inherent danger caused by residues from sulfuric acid, ammonium hydroxide, isopropyl alcohol and other chemical pollutants. For example, an immersion-type cleaning step widely used industrially utilizes RCA Standard Clean 1 (RCA-1) [58].

Wet-chemical processes are still the most widely used method for Si wafer cleaning in the semiconductor industry today [51]. The critical demands of surface purity raised by the International Technology Roadmap for Semiconductors (ITRS) [59] can generally be reduced by utilizing ultrasonic cleaning processes.

In the above context, a substrate independent cleaning process is highly desirable because, opposite to a chemical based cleaning process, it is equally well suited for different substrates and does not modify the surface through the etching, roughening, etc.

While keeping the compatibility with Si wafer standard processing steps, ultrasonic treatment of surfaces can be effective in passing several obstacles to achieving wafer cleaning mentioned above. Ultrasonic cleaning employs an ultrasonically activated liquid with a submerged wafer used to achieve or enhance the removal of surface contaminants [60]. Ultrasonic irradiation involves a variety of complex

### *Ultrasonic Processing of Si and SiGe for Photovoltaic Applications DOI: http://dx.doi.org/10.5772/intechopen.96939*

**4. Ultrasonic cleaning of Si wafers**

*Solar Cells - Theory, Materials and Recent Advances*

surface due to its hydrophobic nature [56].

RCA Standard Clean 1 (RCA-1) [58].

damage to the wafers.

roughening, etc.

**142**

stacked memory devices etc.

Cleaning and conditioning of silicon wafer surfaces and Si/SiO2 interfaces for the

manufacturing of photovoltaic and microelectronic devices are in increasing demand with improved performance, reliability of these devices, scaling down to below 10 nm, incorporating extended metallization layers, employing epitaxial layers of compound SiGe and III-V semiconductors [51]. Clean wafer surfaces are crucial in high efficiency solar cell as well as in Ultra-Large Scale Integration (ULSI) fabrication processes, fin-shaped field effect transistors (FinFET), 3D NAND-

Deposition of monolayers and self-assembly of nanoparticles in multilevel structures requires the wafer surface to be completely free of any particulate contamination down to a nanometer scale. Moreover, installing a 1 MWp solar module

Furthermore, metallic contaminants on Si wafers cause substantial increase in leakage current in silicon p-n junctions and decrease the oxide breakdown voltage thus deteriorating the minority carrier lifetime [52, 53]. In particular, Cu and Al contaminants worsen the gate oxide integrity [54, 55]. To achieve ultraclean substrate surfaces with high reproducibility, it is also important to note that not only removal of contaminants is effective in improving the cell performance, but also prevention of their redeposition on the wafer surface. In this respect, dilute HF can be effectively used. However, some Cu and Al residues can be found on the wafer

The contamination of wafer surfaces by particle contaminants is one of the major problems in the industries. One way to increase the yield on fully processed silicon wafers is to use cleaning techniques specifically efficient to remove particle contaminants. Small particles are especially difficult to remove because they are strongly bounded to the substrate by electrostatic forces. It is therefore very important to find an effective way to remove particles from wafers without causing

A wide variety of cleaning methods are being used in wafer manufacturing such

as brush or water-jet scrubbing of wafer surfaces employed prior to further immersion-type cleaning, scrubbing of rotating wafer surfaces between each processing step, adding cheating agents to the solution aiming to avoid metal adsorption onto Si wafers, cleaning in wet chemical baths, post treatment rinsing and many others [57]. They, however, are known to damage the wafer surface. Moreover, the chemical-type cleaning has inherent danger caused by residues from sulfuric acid, ammonium hydroxide, isopropyl alcohol and other chemical pollutants. For example, an immersion-type cleaning step widely used industrially utilizes

Wet-chemical processes are still the most widely used method for Si wafer cleaning in the semiconductor industry today [51]. The critical demands of surface purity raised by the International Technology Roadmap for Semiconductors (ITRS)

In the above context, a substrate independent cleaning process is highly desirable because, opposite to a chemical based cleaning process, it is equally well suited for different substrates and does not modify the surface through the etching,

While keeping the compatibility with Si wafer standard processing steps, ultrasonic treatment of surfaces can be effective in passing several obstacles to achieving wafer cleaning mentioned above. Ultrasonic cleaning employs an ultrasonically activated liquid with a submerged wafer used to achieve or enhance the removal of surface contaminants [60]. Ultrasonic irradiation involves a variety of complex

[59] can generally be reduced by utilizing ultrasonic cleaning processes.

generator implies that more than 2<sup>10</sup><sup>5</sup> Si wafers must be processed.

mechanisms, including mechanical vibration and appropriate pressure gradients, microcavitation bubbles that oscillate and dance around due to Bjerknes force [39, 61], acoustic streaming flows, etc. One of the most important aspects of using acoustic streaming is the effect of the frequency on the boundary layer [62]. Its thickness decreases and the streaming velocity increases with increasing the sonication frequency. These both remarkably increase the drag force and hence the particle removal efficiency. It is demonstrated that the acoustic streaming induced removal of foreign contaminants with sizes in the dozen nm range is accomplished at frequencies greater than 1 MHz, i.e. in a high-frequency sonication region [63]. The removal of contaminants having sizes down to ≈100 nm is possible at frequencies smaller than 1 MHz [64].

It has furthermore been demonstrated that the caustic etching process can be ultrasonically enhanced producing finer, more homogeneous textured surfaces [65]. This technique inhibits the sunlight reflection from the Si surface thus enhancing the performance of solar cells.

To demonstrate the capabilities of high-power ultrasound in powerfully manipulating surface species, consider the data reported in **Figure 3**. Here, following the method of cleaning from laser-induced cavitation bubbles [66], the wafer surface is covered with a thin layer of grease. This yields the optical transmission spectra indicative of organic contaminants, which are marked by several peaks in spectra 2 of **Figure 3**, such as C–CH3 at 2960 cm<sup>1</sup> , –CH2– at 2920 cm<sup>1</sup> , –CH at 2890 cm<sup>1</sup> and C–H stretching vibrations [–CH3 and –(CH2)n–] at 2850 cm<sup>1</sup> [67]. It is seen in spectra 3 that organic contaminants are effectively removed from the wafer surface upon its exposure to ultrasonic cavitation with the peak acoustic intensity of about

#### **Figure 3.**

*FTIR spectra of n-Si (a) and p-Si (b) wafers, prior to surface greasing and applying the ultrasound (spectrum 1), covered with a thin layer of vaseline (2) and subsequently cleaned in an ultrasonic bath (3) during 15 min. Reprinted with permission from Podolian A, Nadtochiy A, Kuryliuk V, Korotchenkov O, Schmid J, Drapalik M, Schlosser V. Solar Energy Materials and Solar Cells. 2011;95:765–772. Copyright 2010, Elsevier.*

400 W/cm<sup>2</sup> [68], so that the resulting absorption resembles the one taken before the wafer has been exposed to sonication (spectrum 1 in **Figure 3**).

Although cavitation bubble dynamics close to solid surfaces has been given particular attention [68], quite little is known about the streaming along the surface. Therefore, in attempting to explain the removal mechanism behind the cleaning effect observed in **Figure 3** there may be several potential ways. One is that the pressure gradients due to bubble implosion and acoustic streaming would bombard and remove organic contaminants on the silicon surface. The other is that some excited oxygen atoms produced by the sonochemical decomposition of the water adhere to the organic compounds, oxidize them finally decomposing into H2O, O2, H2, CO and CO2, having high vapor pressures allowing the lift-off from the wafer surface [68].

Typical forward and reverse bias *I V* characteristics of the wafers with and without organic contaminants are plotted in **Figure 4**. Here, distilled water and piranha (3:1 volume solution of H2SO4 and 30%-H2O2) are used as a cleaning liquid (curve 2 compared with curves 3 and 4 in **Figure 4**), and both chemical and sonochemical cleaning processes are contrasted (curve 3 compared with curve 4). It is interesting that the cleanings cause an overall decrease in the current through the wafer. This can, in part, be described by the removal of the organic contaminants from the wafer and appropriate quenching of the leaky currents between the basal wafer surfaces. The ultrasonic effect in piranha (curve 4 in **Figure 4**) is obviously greater than that in water (curve 2), as would be expected in reactive chemical agents (cf. curve 2 in **Figure 4**).

ultrasonically cleaned wafer (curve 2 compared with curve 1 in **Figure 5**), which can simply be attributed to an increased density of unsaturated dangling bonds on the wafer surface due to the cavitation-induced local removal of oxide from the

*SPV decays of n-Si, as-purchased (curve 1), ultrasonically cleaned in distilled water for 60 (2) and 120 (3) min. Reprinted with permission from Podolian A, Nadtochiy A, Kuryliuk V, Korotchenkov O, Schmid J, Drapalik M, Schlosser V. Solar Energy Materials and Solar Cells. 2011;95:765–772. Copyright 2010, Elsevier.*

*Ultrasonic Processing of Si and SiGe for Photovoltaic Applications*

*DOI: http://dx.doi.org/10.5772/intechopen.96939*

For sonication times increasing up to 90–150 min, the SPV decay is slightly worsened. For example, curve 3 in **Figure 5** exhibits a long-tail decay constituent at time instants greater than ≈20 μs. The involvement of the above surface states and interface traps is therefore reasonable to assume. In this context, for the initial decays at *t*< 10 μs, when the injected carrier concentration is large compared with the density of the trapping centers *Nt*, the recombination centers mainly determine the excess carrier lifetime. Once the concentration of these carriers becomes exceedingly small, particularly compared with *Nt* (at *t* > 20 μs, curve 3 in **Figure 5**), the SPV decay is determined by *Nt*. The initial decays at *t*≤ 10 μs in curves 2 and 3 are nearly identical, giving some indication of the importance of this cavitation processing step in obtaining clean wafer surfaces. Based upon these results, surface and interface trap generation is likely to be significant at prolonged sonication

A cogent resource for the burgeoning field of the surface passivation coating utilizing hydrocarbon chains [69], which can reduce the density of surface states and increase the recombination lifetime of the majority carriers. Different organic solvents can be used in practical ways for these purposes such as chloroform (CHCl3) and dichloromethane (CH2Cl2). For example, silicone polymers were grown on the Si surfaces with gaseous CHCl3 and Cu catalyst [70], brominated aromatic moieties were successfully prepared from KBr/H2O2 in sonochemically treated chloroform [71]. Recent work in applying ultrasound to chemical reactions demonstrates the promise of the sonochemical approach, yet the bromination of aromatic compounds is not achieved with simple mechanical stirring replacing

This topic is covered in more detail below for two types of samples. Type A sample is a GexSi1-x alloy layer, 100 nm thick, grown on a p-doped Cz-Si wafer. Type B sample is obtained by coating sample A with a 10-nm thick a-Si layer (see

**Figure 6**). The Ge content *x* in the GexSi1-x layers is about 30 at.%.

silicon surface.

**Figure 5.**

sonication.

**145**

times, greater than 60–90 min.

**5. Sonochemical passivation of Si and SiGe**

Perhaps it is best noted here that the cavitation processing affects a sub layer region beneath the wafer surface. Therefore, the air/oxide and oxide/wafer surface state or interface trap densities could be reduced significantly by this processing step.

This would have a similar effect on the photovoltaic response of the wafers since the photo-induced charge carriers are separated in the electric field of the surface space charge region. These carriers would partially screen the fixed surface or interface state charge thus reducing the surface band bending.

As one of the earliest attempt to manipulate the surface photovoltage, **Figure 5** illustrates the sonication effect on the SPV decays. The decay is seen to be faster for

#### **Figure 4.**

*I V curves of the Au Schottky contact to n-Si wafer, as-purchased (1), ultrasonically cleaned in distilled water (2), chemically (3) and ultrasonically (4) processed in a piranha bath. In each case, the cleaning time is 60 min. Reprinted with permission from Podolian A, Nadtochiy A, Kuryliuk V, Korotchenkov O, Schmid J, Drapalik M, Schlosser V. Solar Energy Materials and Solar Cells. 2011;95:765–772. Copyright 2010, Elsevier.*

*Ultrasonic Processing of Si and SiGe for Photovoltaic Applications DOI: http://dx.doi.org/10.5772/intechopen.96939*

**Figure 5.**

400 W/cm<sup>2</sup> [68], so that the resulting absorption resembles the one taken before

Although cavitation bubble dynamics close to solid surfaces has been given particular attention [68], quite little is known about the streaming along the surface. Therefore, in attempting to explain the removal mechanism behind the cleaning effect observed in **Figure 3** there may be several potential ways. One is that the pressure gradients due to bubble implosion and acoustic streaming would bombard and remove organic contaminants on the silicon surface. The other is that some excited oxygen atoms produced by the sonochemical decomposition of the water adhere to the organic compounds, oxidize them finally decomposing into H2O, O2, H2, CO and CO2, having high vapor pressures allowing the lift-off from the wafer

Typical forward and reverse bias *I V* characteristics of the wafers with and without organic contaminants are plotted in **Figure 4**. Here, distilled water and piranha (3:1 volume solution of H2SO4 and 30%-H2O2) are used as a cleaning liquid (curve 2 compared with curves 3 and 4 in **Figure 4**), and both chemical and sonochemical cleaning processes are contrasted (curve 3 compared with curve 4). It is interesting that the cleanings cause an overall decrease in the current through the wafer. This can, in part, be described by the removal of the organic contaminants from the wafer and appropriate quenching of the leaky currents between the basal wafer surfaces. The ultrasonic effect in piranha (curve 4 in **Figure 4**) is obviously greater than that in water (curve 2), as would be expected in reactive chemical

Perhaps it is best noted here that the cavitation processing affects a sub layer region beneath the wafer surface. Therefore, the air/oxide and oxide/wafer surface state or interface trap densities could be reduced significantly by this processing step. This would have a similar effect on the photovoltaic response of the wafers since the photo-induced charge carriers are separated in the electric field of the surface space charge region. These carriers would partially screen the fixed surface or

As one of the earliest attempt to manipulate the surface photovoltage, **Figure 5** illustrates the sonication effect on the SPV decays. The decay is seen to be faster for

*I V curves of the Au Schottky contact to n-Si wafer, as-purchased (1), ultrasonically cleaned in distilled water (2), chemically (3) and ultrasonically (4) processed in a piranha bath. In each case, the cleaning time is 60 min. Reprinted with permission from Podolian A, Nadtochiy A, Kuryliuk V, Korotchenkov O, Schmid J, Drapalik M, Schlosser V. Solar Energy Materials and Solar Cells. 2011;95:765–772. Copyright 2010, Elsevier.*

interface state charge thus reducing the surface band bending.

the wafer has been exposed to sonication (spectrum 1 in **Figure 3**).

*Solar Cells - Theory, Materials and Recent Advances*

surface [68].

**Figure 4.**

**144**

agents (cf. curve 2 in **Figure 4**).

*SPV decays of n-Si, as-purchased (curve 1), ultrasonically cleaned in distilled water for 60 (2) and 120 (3) min. Reprinted with permission from Podolian A, Nadtochiy A, Kuryliuk V, Korotchenkov O, Schmid J, Drapalik M, Schlosser V. Solar Energy Materials and Solar Cells. 2011;95:765–772. Copyright 2010, Elsevier.*

ultrasonically cleaned wafer (curve 2 compared with curve 1 in **Figure 5**), which can simply be attributed to an increased density of unsaturated dangling bonds on the wafer surface due to the cavitation-induced local removal of oxide from the silicon surface.

For sonication times increasing up to 90–150 min, the SPV decay is slightly worsened. For example, curve 3 in **Figure 5** exhibits a long-tail decay constituent at time instants greater than ≈20 μs. The involvement of the above surface states and interface traps is therefore reasonable to assume. In this context, for the initial decays at *t*< 10 μs, when the injected carrier concentration is large compared with the density of the trapping centers *Nt*, the recombination centers mainly determine the excess carrier lifetime. Once the concentration of these carriers becomes exceedingly small, particularly compared with *Nt* (at *t* > 20 μs, curve 3 in **Figure 5**), the SPV decay is determined by *Nt*. The initial decays at *t*≤ 10 μs in curves 2 and 3 are nearly identical, giving some indication of the importance of this cavitation processing step in obtaining clean wafer surfaces. Based upon these results, surface and interface trap generation is likely to be significant at prolonged sonication times, greater than 60–90 min.
