The mechanical integrity of silicon wafers cut by diamond wire sawing depends on the damage (e.g., micro-cracks) caused by the cutting process. The damage type and extent depends on the material removal mode, i.e., ductile or brittle. This paper investigates the effect of cutting fluid on the mode of material removal in diamond scribing of single crystal silicon, which simulates the material removal process in diamond wire sawing of silicon wafers. We conducted scribing experiments with a diamond tipped indenter in the absence (dry) and in the presence of a water-based cutting fluid. We found that the cutting mode is more ductile when scribing in the presence of cutting fluid compared to dry scribing. We explain the experimental observations by the chemo-mechanical effect of the cutting fluid on silicon, which lowers its hardness and promotes ductile mode material removal.

Diamond wire sawing (DWS) is increasingly the preferred process for high volume production of photovoltaic (PV) crystalline silicon wafers.1 A current challenge faced by the PV industry is the breakage of wafers when processing bare wafers into solar cells. Research suggests that the breakage rates of crystalline silicon photovoltaic wafers are greatly impacted by the surface and subsurface damage produced in the DWS process.2,3 Hence, reducing surface and subsurface damage by enhancing ductile mode material removal of brittle silicon is a goal being pursued by researchers.4,5 The cutting mode of silicon is known to be influenced by the sawing process parameters such as wire speed and workpiece feed rate, by the abrasive properties (shape, size, and concentration),6 and by the material properties (mono- vs. multicrystalline silicon6). In DWS, a water-based cutting fluid is used primarily as a coolant for cutting the silicon ingot with 8–20 μm average diameter diamond abrasives attached to a high carbon steel wire with nickel coating. In this paper, we investigate via diamond scribing experiments an unexplored aspect of the role of cutting fluid, namely, the effect of cutting fluid on the mode of material removal (ductile versus brittle) in diamond wire sawing of mono-crystalline silicon (mono-Si).

Material removal in DWS is due to a combination of scratching and indentation processes, with the former process playing a more dominant role. Hence, DWS can be simulated by diamond scribing experiments [Figs. 1(a) and 1(b)].4,6 In general, the silicon wafer surface is characterized by a mix of ductile and brittle material removal modes.2,5 We report here the results of scribing studies performed with a diamond tipped indenter scribing a micro-electronic grade monocrystalline (100) silicon wafer in air (dry scribing) and in the presence of an industrial grade cutting fluid used in the DWS process. The observations are explained in terms of the chemo-mechanical effect of the fluid on the deformation response of silicon.

FIG. 1.

(a) Cutting action in diamond wire sawing, (b) scribing of silicon by diamond tipped indenter, (c) scribing setup, and (d) cutting fluid on top of sample before scribing.

FIG. 1.

(a) Cutting action in diamond wire sawing, (b) scribing of silicon by diamond tipped indenter, (c) scribing setup, and (d) cutting fluid on top of sample before scribing.

Close modal

Microelectronic grade double-side polished monocrystalline (100) silicon wafers were diced into small 10 mm × 25 mm coupons, which were used as substrates in the diamond scribing tests. The scribing direction was [110]. Scribing was performed using a conical tipped diamond scriber with a 90° included angle. The scribes were of gradually increasing depth of cut ranging from 0 to 5 μm over a length of 5 mm [Fig. 1(b)]. We used increasing depth of cut scribes to identify the critical depth of cut for ductile-to-brittle transition, which is defined as the depth of cut at which the mode of material removal transitions from ductile to brittle and is characterized by the appearance of surface cracks. The diced silicon wafer coupons were fixed to a flat aluminum baseplate bolted to a three-component piezoelectric force dynamometer (Kistler 9256C), which was mounted rigidly on computer-controlled stacked X-Y-Z motion stages (Aerotech ANT-4V) [Fig. 1(c)]. The dynamometer enabled the detection of initial contact of the scriber with the silicon wafer and was also used to measure the dynamic forces generated in the scribing experiments.

All scribing experiments were performed at room temperature and at a constant scribing speed of 100 mm/min. First, we carried out dry scribing tests. Next, we flooded the top of the wafer sample with water-based cutting fluid a few seconds before scribing [Fig. 1(d)]. This procedure was repeated for each case: dry and with water-based cutting fluid. We made 3 to 4 scribes for each test condition to ensure repeatability of the results. For the cutting fluid tests, we used a commercial grade DWS coolant (Ambercut DWC-35) mixed with 97% water by volume (97 ml of water and 3 ml of concentrated cutting fluid). Ambercut DWC-35 is a fixed abrasive wire slicing coolant that is a mixture of synthetic surfactants diluted in water to produce a biodegradable water-based cutting fluid.

The critical depth of cut for ductile-to-brittle transition in an increasing depth of cut scribing test was determined through optical confocal microscopy (Olympus LEXT). The scribe morphology was observed using an optical microscope (Leica DVM6). For selected scribes, we obtained higher resolution images using a scanning electron microscope (Hitachi 8230 SEM). Micro-Raman spectroscopy was performed on selected scribes using a Renishaw InVia Raman spectrometer to determine the phase of silicon in the scribed grooves. The wavelength of the laser in the micro-Raman spectrometer was 488 nm with a surface penetration of ∼0.5–0.6 μm and a spatial resolution of 1 μm, which is smaller than the width of the scribes.

The zeta potential of the cutting fluid in the presence of silicon was characterized using a Malvern Zetasizer (Nano ZS). This instrument uses solid particles of the monocrystalline silicon wafer ranging in size from 0.6 nm to 6 μm dispersed in the water-based cutting fluid to measure the zeta potential. Monocrystalline silicon wafer coupons diced from the same silicon wafer utilized in the scribing experiments were cleaned using acetone and air-blasted to remove any contamination before crushing them into fine powder using a mortar and pestle. The silicon powder was then dispersed in the water-based cutting fluid. The resulting mixture was sonicated in a water bath for 30 min. The sonicated mixture was not a true suspension since a few silicon particles settled at the bottom of the beaker when the solution was extracted from the sonicator. Therefore, for the zeta potential measurements, the sonicated mixture was sampled from the top to minimize the inclusion of large particles of silicon. The sonicated silicon-fluid mixture was subjected to an applied electric field of 5 V in an electrophoresis cell, and the method of electrophoretic light scattering was used to determine the zeta potential.

Figure 2(a) shows a representative optical image of the scribes obtained in dry scribing and in the presence of the cutting fluid. Note that although the X-coordinates of the starting locations of all scribes were the same, the scribe made in the presence of the fluid shows evidence of brittle fracture (dark line) after a longer scribing distance than the dry scribe. Specifically, the average distance to transition from ductile to brittle cutting during dry scribing was 1.27 mm compared to 1.75 mm for scribing with cutting fluid. The average critical depth of cut was greater in the scribes made with cutting fluid compared to the dry scribes [Fig. 2(b)]. For the same travel distance of approximately 1.50 mm from the start of the scribe, the SEM images [Figs. 2(c) and 2(d)] showed more ductile deformation of silicon in the presence of the cutting fluid than in dry scribing. Raman spectra of the scribed grooves after ductile-to-brittle transition are shown in Fig. 3. Even though silicon is a brittle material, it is known to undergo stress-induced phase transformation from Si-I to β-Sn and other phases such as Si-III, Si-XII, and amorphous silicon, depending on the unloading rate.7 Raman spectra of the dry scribe showed a prominent crystalline Si-I peak around 520 cm−1. In comparison, the scribe made with the cutting fluid showed a Si-III peak (470 cm−1) and broad peaks of amorphous silicon (a-Si) in addition to the Si-I peak, thus indicating evidence of phase transformation and ductile mode material removal. It is therefore evident that the cutting fluid enhances ductile mode deformation of silicon.

FIG. 2.

(a) Representative scribes obtained in dry scribing and in scribing with water-based cutting fluid, (b) box plot of the critical depths of cut for the dry and cutting fluid cases; SEM images of the scribes after approximately 1.50 mm of travel (c) for dry, and (d) with fluid.

FIG. 2.

(a) Representative scribes obtained in dry scribing and in scribing with water-based cutting fluid, (b) box plot of the critical depths of cut for the dry and cutting fluid cases; SEM images of the scribes after approximately 1.50 mm of travel (c) for dry, and (d) with fluid.

Close modal
FIG. 3.

Raman spectra of scribes: (a) dry and (b) cutting fluid. Measurements were made inside the grooves after ductile-to-brittle transition.

FIG. 3.

Raman spectra of scribes: (a) dry and (b) cutting fluid. Measurements were made inside the grooves after ductile-to-brittle transition.

Close modal

Figure 4 shows the normal (Z) and the tangential (X) force profiles, plotted against the scribing distance. As expected, the scribing forces are seen to increase with increasing depth of cut. The initial smooth increases in the forces correspond to the ductile portion of the scribes. The transition to brittle material removal is marked by rapidly varying forces. Considering that the rate of increase in the scribing depth is the same for the two cases, scribing with the cutting fluid exhibited brittle fracture after a larger scribing distance, and hence at a higher scribing depth, as seen from the critical depth of cut data shown earlier.

FIG. 4.

Normal (Z in blue) and tangential (X in red) scribing forces in (a) dry scribing and (b) scribing with cutting fluid.

FIG. 4.

Normal (Z in blue) and tangential (X in red) scribing forces in (a) dry scribing and (b) scribing with cutting fluid.

Close modal

Fluids are known to influence the mechanical properties of solids they are in contact with. Prior work investigated the effects of ethanol, acetone, and deionized water in constant-depth scribing of monocrystalline (111) silicon.8–11 Various theories have been proposed to explain this “chemo-mechanical effect.” Research on chemo-mechanical effects dates back to the 1920s when Rebinder's theory of reduction in the hardness of solids due to adsorbed chemical species was proposed.12 Later, Westwood13 found the hardness of a solid and the zeta potential of the fluid in contact with the solid to be correlated. The zeta potential is the electrostatic potential between the Stern layer of the double layer at the liquid-solid interface and the bulk solution.14 Westwood showed that a zero zeta potential corresponds to the maximum hardness of very brittle non-metals.13 Yost and Williams14 found that when intrinsic and doped (n- and p-type) silicon are exposed to NaCl and Na4P2O7, the minimum hardness is correlated with the most negative value of zeta potential. Yost and Williams explained that the change in hardness with the zeta potential is due to the surface charges, which affect the near-surface mobility of dislocations. This explanation is based on prior work where charges produced by electronic doping influenced the dislocation velocities15 and dislocation kink formation.16 In another study, Westbrook and Gilman17 reported softening of silicon by 60% during indentation in the presence of a small potential between the silicon surface and the indenter. Therefore, in the following, we use the theory of chemo-mechanical effects to explain our experimental results.

We made zeta potential measurements on single crystal (100) silicon particles dispersed in water (indicative of atmospheric moisture, as our dry scribing tests were performed in ambient conditions) and in the cutting fluid. Figure 5 shows the zeta potential for the cutting fluid in contact with silicon is more negative compared to water. Vickers micro-hardness measurements showed a reduction in the hardness of silicon in the presence of the cutting fluid (7.32 ± 0.50 GPa) compared to without the cutting fluid (dry) (8.80 ± 0.47 GPa). The difference in the mean hardness for the two cases is statistically significant within a 95% confidence interval. The zeta potential and hardness results support the theory that the cutting fluid lowers the hardness of silicon, which causes more ductile mode material removal and hence yields a larger critical depth of cut than dry scribing.

FIG. 5.

Zeta potentials of silicon particles dispersed in water and cutting fluid.

FIG. 5.

Zeta potentials of silicon particles dispersed in water and cutting fluid.

Close modal

In conclusion, the paper presented the results of an experimental study of the effects of a water-based industrial grade cutting fluid (Ambercut DWC-35) on the mode of material removal in diamond scribing of single crystal (100) silicon. The scribing process simulates the abrasive-workpiece interaction in diamond wire sawing. We found that the water-based cutting fluid, which is used in the diamond wire sawing process for lubrication and cooling, also promotes ductile mode material removal. We explained the results obtained in our dry scribing experiments and in the cutting fluid experiments using the chemo-mechanical effect theory. The fluid-silicon combination yielded a more negative value of the zeta potential and a corresponding reduction in the hardness of silicon, which caused more ductile material removal. These results suggest that in addition to their cooling and lubrication properties, cutting fluids also influence the cutting mode in diamond wire sawing of silicon. Nominally, brittle mode removal would be desirable at the bottom of the kerf to maximize the material removal rate, while ductile mode removal would be desirable on the sides of the kerf to minimize surface and subsurface damage. This suggests a trade-off in the cutting fluid's role in diamond wire sawing.

The authors acknowledge S. Kaminski and Dr. C. Arcona of Saint-Gobain Northboro Research and Development Center for providing the cutting fluid sample used in the study. The authors also acknowledge the support of the National Science Foundation (CMMI Grant No. 1538293) for funding the study. Parts of the work reported in the paper were performed at the Georgia Tech Institute for Electronics and Nanotechnology, a member of the National Nanotechnology Coordinated Infrastructure, which is supported by the National Science Foundation (Grant No. ECCS-1542174).

2.
A.
Kumar
,
S.
Kaminski
,
S. N.
Melkote
, and
C.
Arcona
,
Wear
364–365
,
163
(
2016
).
3.
R.
Buchwald
,
K.
Fröhlich
,
S.
Würzner
,
T.
Lehmann
,
K.
Sunder
, and
H. J.
Möller
,
Energy Procedia
38
(
0
),
901
(
2013
).
4.
A.
Kumar
,
A.
Kovalchenko
,
V.
Pogue
,
E.
Pashchenko
, and
S. N.
Melkote
,
Procedia CIRP
45
,
147
(
2016
).
5.
E.
Cai
,
B.
Tang
,
W. R.
Fahrner
, and
L.
Zhou
, in
26th European Photovoltaic Solar Energy Conference and Exhibition
(
2011
), p.
1884
.
6.
A.
Kumar
,
S. N.
Melkote
,
S.
Kaminski
, and
C.
Arcona
,
J. Am. Ceram. Soc.
100
,
1350
1359
(
2017
).
7.
Y.
Gogotsi
,
G. H.
Zhou
,
S. S.
Ku
, and
S.
Cetinkunt
,
Semicond. Sci. Technol.
16
(
5
),
345
(
2001
).
8.
S.
Danyluk
and
R.
Reaves
,
Wear
77
(
1
),
81
(
1982
).
9.
S.-W.
Lee
,
D.-S.
Lim
, and
S.
Danyluk
,
J. Mater. Sci. Lett.
3
(
7
),
651
(
1984
).
10.
S.
Danyluk
and
J. L.
Clark
,
Wear
103
(
2
),
149
(
1985
).
11.
D.-S.
Lim
and
S.
Danyluk
,
J. Mater. Sci.
23
(
7
),
2607
(
1988
).
12.
P.
Rebinder
, “
Effect of changes in surface energy on the embrittlement, hardness and other properties of crystals
,” in
Proceedings of the 6th Physics Conference
,
Moscow
,
1928
, p.
29
.
13.
A.
Westwood
,
J.
Ahearn
, and
J.
Mills
,
Colloids Surf.
2
(
1
),
1
(
1981
).
14.
G. H.
Yost
and
W. S.
Williams
,
J. Am. Ceram. Soc.
61
(
3–4
),
139
(
1978
).
15.
J.
Patel
,
L.
Testardi
, and
P.
Freeland
,
Phys. Rev. B
13
(
8
),
3548
(
1976
).
16.
P.
Haasen
,
Phys. Status Solidi A
28
(
1
),
145
(
1975
).
17.
J.
Westbrook
and
J.
Gilman
,
J. Appl. Phys.
33
(
7
),
2360
(
1962
).