Surface defects incorporated diamond machining of silicon

This paper reports the performance enhancement benefits in diamond turning of the silicon wafer by incorporation of the surface defect machining (SDM) method. The hybrid micromachining methods usually require additional hardware to leverage the added advantage of hybrid technologies such as laser heating, cryogenic cooling, electric pulse or ultrasonic elliptical vibration. The SDM method tested in this paper does not require any such additional baggage and is easy to implement in a sequential micro-machining mode. This paper made use of Raman spectroscopy data, average surface roughness data and imaging data of the cutting chips of silicon for drawing a comparison between conventional single-point diamond turning (SPDT) and SDM while incorporating surface defects in the (i) circumferential and (ii) radial directions. Complementary 3D finite element analysis (FEA) was performed to analyse the cutting forces and the evolution of residual stress on the machined wafer. It was found that the surface defects generated in the circumferential direction with an interspacing of 1 mm revealed the lowest average surface roughness (Ra) of 3.2 nm as opposed to 8 nm Ra obtained through conventional SPDT using the same cutting parameters. The observation of the Raman spectroscopy performed on the cutting chips showed remnants of phase transformation during the micromachining process in all cases. FEA was used to extract quantifiable information about the residual stress as well as the sub-surface integrity and it was discovered that the grooves made in the circumferential direction gave the best machining performance. The information being reported here is expected to provide an avalanche of opportunities in the SPDT area for low-cost machining solution for a range of other nominal hard, brittle materials such as SiC, ZnSe and GaAs as well as hard steels.

This paper reports the performance enhancement benefits in diamond turning of the silicon wafer by incorporation of the surface defect machining (SDM) method. The hybrid micromachining methods usually require additional hardware to leverage the added advantage of hybrid technologies such as laser heating, cryogenic cooling, electric pulse or ultrasonic elliptical vibration. The SDM method tested in this paper does not require any such additional baggage and is easy to implement in a sequential micro-machining mode. This paper made use of Raman spectroscopy data, average surface roughness data and imaging data of the cutting chips of silicon for drawing a comparison between conventional single-point diamond turning (SPDT) and SDM while incorporating surface defects in the (i) circumferential and (ii) radial directions. Complementary 3D finite element analysis (FEA) was performed to analyse the cutting forces and the evolution of residual stress on the machined wafer. It was found that the surface defects generated in the circumferential direction with an interspacing of 1 mm revealed the lowest average surface roughness (Ra) of 3.2 nm as opposed to 8 nm Ra obtained through conventional SPDT using the same cutting parameters. The observation of the Raman spectroscopy performed on the cutting chips showed remnants of phase transformation during the micromachining process in all cases. FEA was used to extract quantifiable information about the residual stress as well as the sub-surface integrity and it was discovered that the grooves made in the circumferential direction gave the best machining performance.
Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
Silicon has been the prime material of modern optics/photonics due to its unique engineering characteristics. The utilisation of silicon is ubiquitous in the field of optoelectronics, MEMS, space and defence industries. The drive for miniaturisation in the production of gadgets making use of silicon is now requiring reduced energy consumption and faster productivity to produce silicon optics. Considerable research work on silicon has been done that has advanced our understanding of this topic since the 1990s. For example, Blackley et al [1] gave a fundamental understanding of ductile mode machining of germanium using the single-point diamond turning (SPDT) process.
Shibata et al [2] examined the critical influence of the crystallographic orientation on the finished machined surface and concluded that the (111) surface of silicon offers the best machinability. It was proposed that the ductility observed during micromachining of silicon is broadly an outcome of the high-pressure phase transformation (HPPT) [3]. Using Raman spectroscopy, it was demonstrated that the stress state in silicon during cutting causes the diamond cubic (α-silicon) structure to undergo a metastable body centred tetragonal (β-silicon) structural transformation and then depending on the release of strain rate (or cutting speed), the ductile phase back-transforms to the amorphous room temperature phase of silicon. More recently, a direct amorphization without any prior Si-II transition observed from the MD simulation was also proposed to be another ductility mechanism [4]. However, a key problem during SPDT is the in-process degradation and wear of the diamond tool causing worsening of the quality of the machined surface. It is widely known that the wear volume of the diamond tool scales with the cutting distance and that the tool's flank face undergoes relatively more wear than its rake face [5,6].
In an attempt to improve the machinability of difficultto-cut, hard and brittle materials like silicon, a range of performance enhancement methods have been proposed and augmented in recent times including the incorporation of laser, cryogenic and electroplasticity concepts [7], all these hybrid and sequential micromachining approaches possess their distinct advantages.
Surface defect machining (SDM) is a recent advancement that was explicitly reported for improving the quality of hard turning during cutting of hard steels of up to 69 HRC [8][9][10]. Lately, the insights obtained from molecular dynamics simulation study on processing silicon carbide (SiC) hinted at exploiting this route even for the machining of hard, brittle materials [11]. The MD simulation results showed that the surface defects bring a reduction in shear plane angle, shear plane area and side flow with less metallurgical transformations [11]. Hard, brittle materials usually possess low fracture toughness and silicon is by far the most classic example. Built on previous learnings about SDM, this feasibility study was performed to evaluate the quality of machining in SPDT of the silicon wafer. The aim was to quantify the performance improvement in the extent of residual stresses, improved tool life and ultimately to achieve an improved machined surface finish through the SPDT method with the view of eliminating the need for post-machining polishing.
In particular, this paper expands the idea of the SDM method in the SPDT context, which in essence is an expansion of the pulse laser pre-treated machining method [12]. Recently, a smoothed particle hydrodynamics study on SDM on silicon reported reduced cutting pressure [13]. However, experimental evidence is lacking as this is the first paper experimentally demonstrating the surface defect micromachining of silicon.
Another motive behind this work was to select the shape of surface patterns in such a way that the time to produce the surface defects can be reduced and to make it convenient and cost-efficient. It may be recalled from the aforementioned papers that the surface defects can be generated either by conventional mechanical methods or by using a laser. This work produced these defects using the same SPDT machine with the diamond tool itself just prior to machining. Thus, it saved using any additional machine tool or instrumentation. The experimental results thus obtained were complimented by numerical finite element analysis (FEA) to demonstrate the salient aspects concerning the effectiveness of the proposed surface defect micromachining of silicon.

Machine setup
SPDT experiments were performed on a 3-axis Nanoform-200 (Precitech) machine tool that is aesthetically engineered to fabricate planar, spherical, aspherical and freeform surfaces.  The SPDT experimental setup of the machine is shown in figure 1. The way it works is that a silicon wafer is held on an air bearing spindle through a vacuum chuck by an air pressure of 8-10 bar such that it self-balances and achieves thermal stability over a period of time. A diamond tool is levelled vertically and laterally using an optical tool setter and this process is referred to as the centering process. After the centering process, the tool is fed into the workpiece at a certain depth of cut at a prescribed feed rate while coolant is sprayed into the cutting zone. In this case, the coolant used was Clairsol 330. Experiments were carried out using a spindle speed of 1000 rpm, feed rate of 3 mm min −1 and 10 micron depth of cut.

Workpiece and tool materials
A (111) oriented single crystal silicon workpiece having Ø 45 mm and thickness 6.5 mm was used for the micromachining experiments. The diamond cutting tool with a negative rake of 10 o and 1.5 mm of nose radius procured from Contour Fine Tooling Ltd was used for cutting. The details of the experimental parameters used in this study are shown in tables 1-3 corresponding to groove making, plan of experiments for making groove and SPDT turning respectively.

Surface defect patterns
In the past, a variety of surface defect patterns were explored through FEA simulations during conventional turret lathe machining [9]. Two particular grooves came to light from those preliminary investigations which are being tested here. Grooves or defects made in the circumferential direction and the radial direction were generated (see figure 2) using the same SPDT experimental setup just before carrying the SPDT. These two defects, as well as the machined surface generated after the experiments, were characterised using a contact type surface profilometer (Form Talysurf PGI-120 Model from Taylor Hobson) and non-contact coherence correlation interferometer (CCI) together with a scanning electron microscope. Further details and drawings of these patterns are discussed in the next sections. Figure 3 shows the dimensions of grooves made in silicon with (a) 0.5 mm interspacing and (b) 1 mm interspacing. The depth of grooves was maintained as 5 microns with an opening width on the surface of about 0.1 mm. They were obtained by precise manoeuvring in the Z traverse direction. No coolant was used for better visualisation of the grooves while fabricating them. The tool movement was constrained to allow movement only in the X direction and the feed rate of 2 mm min −1 played a vital role in generating the fine groove profiles without any side flow as measured and characterized using the metrology tools as shown in figures 4 and 5.

Radial direction patterns.
The patterns in the radial direction were developed in Solid Works as per the details and dimensions shown in figure 6. To fabricate these patterns, the diamond tool was mounted in such a way that the rake face approaches along the Z plane of the machine. It may be noted that the pattern generation, in this case, used the c-axis while the spindle was kept stationary. The tool was moved against the workpiece in the Z direction and scratches were made with a depth of cut of 0.3 µm and 0.5 µm at two feed rates ((a) 12 000 mm min −1 and (b) 1000 mm min −1 ). The grooves generated in a straight line at two different feed rates with two different profile morphologies are shown in figure 7. By using the C axis in SPDT, the spindle was indexed by 10 o increment for every groove and eighteen straight line grooves were formed to mimic a star pattern. The feed rate of 12 000 mm min −1 yielded a better groove profile than the one obtained at the feed rate of 1000 mm min −1 because the dynamics of circular grooves were matched with the radial grooves at higher feed rate. For creating the circumferential patterns, the workpiece was rotated and the diamond tool was kept static yet engaged with the workpiece at a fixed depth of cut without any feed motion. Contrarily, the workpiece for creating radial patterns was kept static (headstock in brake mode) while the tool was provided an axial motion at the required depth of cut. Normally, an increasing rotation of the workpiece during turning gives a smoother finish and analogously the fast cutting operation to create radial pattern was performed by increasing the axial movement of the tool to get better finish and profile accuracies. Thus, the axial tool movement feed of 12 000 mm min −1 offered a better profile than 1000 mm min −1 . A detailed microscopic inspection of the     grooves was performed using an SEM (see figure 8) to ensure that the programmable numbers were in close compliance with measured dimensions of the grooves.

Machining of the controlled sample using conventional
SPDT method. Measurement of the surface roughness was one of the primary drivers of the work to improve the quality of machined surface obtained by SPDT. To benchmark the results of SDM incorporated SPDT, a conventional cutting trial was performed using the parameters shown in table 3. The average roughness (Ra) measured on several locations from the conventional SPDT was in the range of about 8 to 10 nm.

Quality of machined surface having pre-machined circumferential grooves.
SDM was developed with the motivation to enhance and improve the shearing of the workpiece during cutting. A comparison of roughness parameters obtained from the machining results while providing grooves in the circumferential direction in two different cases with two interspacings of 0.5 mm and 1 mm respectively are shown in figure 9. The results for machined roughness for an interspacing of 1 mm were observed to be better than for the interspacing of 0.5 mm and there seems to be a strong correlation between this groove interspacing and the width of cut. The results hint at the fact that the interspacing should be equal to or larger than the width of the cut. It may be recalled that the width of cut is governed by the nose radius and undeformed chip thickness (cut depth in 2D) and therefore for the nose radius of 1.5 mm and depth of cut of 5 µm used in this work, 1 mm spacing seems to be better than the interspacing of 0.5 mm. An illustrative explanation of this phenomena is shown in figure 9 to show the importance of keeping the interspacing larger than the width of the cut to avoid jerky contact between the tool and the workpiece.    provided and conventional SPDT was performed. It was found that regardless of the feed rate at which radial grooves were made, the roughness results were consistently poorer (both in terms of surface roughness and sub-surface damage) than the circumferential grooving patterns. This is the reason why in the subsequent results discussed in the paper, only a representative example of the radial groove is considered as one of the results was dropped after seeing no advantage of providing radial grooves. The average mean value of roughness (Ra) and peak to valley height (Rt) were about 8.7 nm and 664 nm during conventional SPDT and they improved when using a circumferential surface defect pattern with 1 mm spacing such that their values were about 3.2 nm and 55 nm respectively.

Cutting chips analysis
Investigation and analysis of cutting chip morphology forms an important aspect of machinability studies as they can reveal the underlying physical phenomena that govern the material removal process and the subsequent surface finish. The chip morphology is largely influenced by the machining conditions as well as the tool-work piece interaction pattern such as the cutting method (continuous, discontinuous or intermittent cutting), tool geometry and the material being cut. The SEM image of the chips produced during the SDM of Si in case of circumferential grooving patterns (interspacing of 0.5 mm and 1 mm) and radial grooving patterns is shown in figure 11. As shown in figure 11(a), the nano-dust sort of chips (continuous but less width) were observed while cutting Si having circumferential defects with 0.5 mm spacing (tool width of cut < pitch of the grooves). One can imagine that the pitch of grooves is less than the width of cut led to the partition of the chip into two pieces and the material does not attain enough plastic conditions in the process and the cutting tool and workpiece contact is somewhat jerky. Such morphological patterns in the generated chips are referred to as partially ductile chips and they can be attributed to uniform material removal rate during cutting, leading to a machined surface with moderate surface integrity. On the other hand, during SDM of Si wafer having circumferential defects with 1.0 mm spacing, the chip characteristics were found to be mid-sized and broken into few micron lengths pieces as shown in figure 11(b). This indicated healthy ductile-regime machining conditions as well as the periodic breaking of the chips (to avoid forming long continuous ribbons), which was one of the original motives behind the development of SDM that it eases cutting load by the periodic breaking of chips.
The case of SDM of Si while providing radial grooves was very different than the other two cases as the chips showed cracking, non-uniform broken size pieces as shown in figure 11(c). Such chips were striated and exemplified debris generation and conditions of cutting completely unfavourable for achieving ductile and plastic conditions.

Study of cutting chips with Raman spectroscopy
The cutting chips of the silicon wafer were examined using laser Raman spectroscopy (model-nVia Raman; make-Renishaw). The wavelength of laser used was 532 nm and the output laser power was set to 10 mW. A 20X objective lens with a numerical aperture of 0.40 was used for the measurement.
It is generally believed that loading with a hydrostatic pressure of about 10 to 12 GPa causes the crystalline silicon (Si-I) to transform to a metastable Beta tin phase of silicon (Si-II). The Beta tin phase or the Si-II phase is no more semiconductive and conducts electricity which is why this highpressure phase transformation is said to cause metallisation of silicon and this metastable form is why metallic silicon can be machined and deformed plastically [14][15][16]. However, the rate of release of load (in turn cutting speed in the context of cutting) governs how this metallic form stabilises to atmospheric and room temperature phases. For example, extremely rapid unloading can cause a direct amorphisation whereas the rate of slow unloading forms other crystalline phases such as Si-XII, Si-III, Si-IV, Si-III before eventually leading to amorphisation.
In the present analysis, the pristine silicon wafer prior to machining showed a representative peak at 520 cm −1 . Various texts on Raman spectroscopy of silicon report the Raman peak around the value of 521 cm −1 . At times depending on the wafer condition or the sensitivity of the Raman instrument, a peak may show a slight shift due to the presence of residual stress (strained crystal can shift the peak to the left or right depending on the state of stress i.e. tensile or compressive). Figure 12 shows the analysed area and the corresponding Raman spectra obtained from the selected machining zone in different cases. The peaks identified from the Raman spectra alluded to the presence of a-Si (peaks corresponding to 160 cm −1 and 300 cm −1 ), crystalline Si-IV (peaks corresponding to ∼507 cm −1 ), crystalline Si-III (peak corresponding to 170 cm −1 ) and a-SiO2 (peaks corresponding to 263 cm −1 ). Si-III is a BCC (bc8) form of silicon which exist between 0 to 2 GPa pressure range while Si-IV is a hexagonal diamond cubic form of silicon, which forms as a result of the martensitic transformation of Si-I (crystalline silicon) such that its relative volume becomes 0.98 with respect to Si-I [4,17].

FEA of SDM of silicon
To complement the reported experimental observations, an FEA study was carried out. A three-dimensional (3D) stress dynamic explicit arbitrary Lagrangian-Eulerian (ALE) numerical model was developed using the FEA software Abaqus. In particular, the localised deformation at the contact of the tool and workpiece was modelled by using dynamic explicit ALE formulation. ALE helps to maintain the highquality mesh during large deformation or loss of materials that generally occurs during machining. The explicit dynamic analysis procedure was based on using very small timesteps. An extensive set of prior trials were carried out to adjust the FEM solver parameters. The developed model was aimed at prediction of the machining forces, changes in shear plane  angle, surface roughness and the evolution of residual stress. The model was built on the previous work by accounting for realistic material behaviour, friction consideration, damage model and by employing realistic geometry of the cutting tool [18]. Further details of the model development are discussed below.

Description and assumptions made for the simulation model development
Initially, 3D solid geometries of tool and workpiece were created. The following assumptions were made in the present model for the simplification of the process. • The cutting length simulated was relatively short (0.5 mm), tool wear was therefore neglected in this study. • As the tool wear was neglected, it was modelled as a rigid body. • For simplicity, silicon workpiece material was assumed to be isotropic and homogeneous, albeit, silicon is known to be highly anisotropic but as the simulation was an orthogonal scratch model, the effect of anisotropy was neglected. • The workpiece material was an elastic-plastic type.
• The workpiece material was assumed to be free from initial internal residual stresses. Figure 13 represents the schematic CAD FEM models of the SDM process developed to be simulated in a scratching mode considering the nose radius effects. The workpiece was modelled as a rectangular block with dimensions 1.5 × 0.5 × 0.1 mm. The single point diamond tool was modelled to have a −10 o rake angle, 10 o clearance angle and to 1.5 mm nose radius.
The SDM process involves non-linear, discrete and complex interactions between the tool and workpiece, as well as intense plastic deformation. Thus, it is essential to select the right type of element type in the simulation. Here, an eightnode linear bricks element (C3D8R) was chosen to mesh the tool and the workpiece. C3D8R is a general-purpose linear brick element, with reduced integration (1 integration point) and provides hourglass control and element deletion (for workpiece). The mesh density in the cutting region was kept fine whereas the region far from the cutting zone was discretized with coarser mesh. As for the boundary condition, the workpiece was constrained at the bottom to imitate the clamping action during machining. The depth of cut was set to be 10 µm and cutting velocity of 2.36 m s −1 was prescribed to the tool. The material properties of silicon and diamond considered in the FEA analysis are listed in table 4. As for the material constitutive model, a Johnson and Cook (J-C) model was used. The J-C constants used to simulate silicon in this work are tabulated in table 5. FEM can simulate the chip separation naturally and therefore chip separation was achieved without introducing any physical, geometrical separation criteria or damage model. The contact between the cutting tool and the workpiece was defined by a modified Coulomb friction model. The coefficient of friction between the tool and workpiece was specified as 0.2. After the development of geometric models of workpiece and tool, material properties, the material damage law and friction law were applied. Next, the theory governing the numerical solution of FEA is presented.

Theory of the FEA numerical solver
The differential equation of motion governing the mechanical response of a system of finite elements can be written as: where m is the mass, c is the damping coefficient, k is the stiffness coefficient. Equation (1) in matrix form can also be written as: In the present work, explicit formulation was employed which uses central difference scheme to discretize the equations. The acceleration equation can be written as: Also, the velocity change is calculated by integrating the acceleration term explicitly through time using the central difference method. The change in velocity obtained is then added to velocity from the middle of the previous step and is used to calculate the velocities at the middle of the current step using: Likewise, the displacement is calculated by integrating velocity through time, which is then used to update the displacements at the end of time step using: Further, an explicit time integration scheme was used to solve the transient problem. It was originally developed to solve high-speed dynamic problems which were difficult to simulate using the implicit method. After the development of the numerical model, the results obtained from the numerical simulations were compared with the experimental results under similar process conditions. These are explained in the following sections.

Analysis of the machining forces, chip morphology, machining stresses and shear plane angle
At the end of the simulation (see supplementary material, available online at (https://stacks.iop.org/IJEM/02/045102/ mmedia)), the cutting outputs such as forces and stresses were extracted from the simulation. Figures 14(a) and (b) show the variations in the axial cutting force or friction force (Fx) and tangential cutting force or the thrust force (Fy) during conventional cutting, cutting defects with a circumferential (circular) pattern and radial pattern. From figure 14, the cutting load can be seen to be minimal when using circumferential pattern defects.    Besides cutting forces, the simulation was also used to estimate the shear plane angle in each case. The shear plane angle was extracted by measuring the angle between the horizontal line and the line along the stress directions. From this analysis, the values of shear plane angle were estimated to be 52.53 o , 48.33 o , 50.94 o and 56.69 o for the cases of conventional cutting and SDM incorporated cutting with circumferential and radial patterns respectively. All these results are provided in table 6. Clearly, the shear plane, much like the cutting forces, was least while using circumferential patterns.
The reduced shear plane angle shows the dominance of friction force over the normal force thus explaining the enhanced cutting action of the tool and leading to improved surface quality as observed during experiments while using circumferential defect patterns. Next in figure 15, the distribution of von Mises stress comparing conventional cutting with the SDM assisted cutting is shown. Typically, silicon was observed to yield at about 9 to 12 GPa which is consistent with the conditions of flow of silicon during micromachining reported by various papers in the past. It may be seen that the stress gradient (distribution) in the case of circumferential pattern seems more concentrated and is low compared to the other two cases. It indicates that only the cutting zone is affected by the stress and this leads to lesser sub-surface damage which would suggest differences between surface qualities of machining. The peak von Mises stress in the cutting zone was found to vary between 10 to 15 GPa in excellent correlation with the reported experimental and simulation values during ductile-mode  cutting of silicon. From the same FEA simulations, an important entity, namely, the residual stress on the machined surface was obtained by calculating the X-direction stress on the machined surface. As such, residual stress represents the extent of distortion caused on the surface due to the manufacturing process. It results due to the partial recovery of the highly strained region that undergoes high compression in the wake of the tool and recovers only partially when the tool passes by this strained zone. The development of high residual stress on the machined surface sometimes requires a postmachining process such as polishing and etching. Having a reliable simulation tool for estimation of the residual stress allows production engineers to select appropriate cutting conditions in advance to minimise these. In this study, the residual stresses were extracted from the workpiece by selecting two nodes (which were chosen to be above and below the machined surface). Table 7 corresponding to figure 16 shows the measurement of the stress vector measured in the 'X' direction in all configurations. These stress values should not be confused with the von Mises stresses as von Mises stress is a scalar entity incorporating all stress vectors. The figure 16 data consist of two parts: (i) incipient flow stress whereby the relative region of the cutting zone experiences a high degree of compression upon being approached by the tool and (ii) the change of stress sign from negative (compression) to positive (tensile) occurs after the material separates into chips and machined surface. The stress measurement on the machine surface after chip separation was averaged, herein referred to as steady-state residual stress and being reported here in all the experiments. The residual stresses while cutting silicon with circumferential defects was lesser in magnitude in comparison to while cutting silicon in a conventional way or while providing radial defects.
Jasinevicius et al [21] have proposed to use ϖ = ϖ 0 + 0.52 L as a formula to estimate the extent of residual stress in diamond-turned silicon wafers based on their micro-Raman spectroscopy analysis of the (100) oriented silicon wafer. Here, ϖ is the experimental peak obtained using Raman spectroscopy, ϖ 0 is the theoretical characteristic peak of silicon (521.6 cm −1 ) and L is the residual stress measured in an area in Kilobar. While cutting silicon at a depth of cut of about 0.1 µm at a feed rate of 1 µm/rev they estimated the residual stress on the machined surface to be about 221.59 MPa. The value of stress seems well compared to what is obtained from the FEA. It is noteworthy that Jasinevicius et al [21] observed that the residual stresses keep on decreasing with an increasing depth of cut and gradually becomes 0 while using a feed rate of 1 µm/rev and depth of cut of 10 µm. SDM with incorporating circumferential defects turns out to be useful in achieved quality of machined surface with low residual stresses.
Furthermore, it was observed that the fluctuations in the residual stresses in the case of the circumferential pattern was more periodic compared to the other configurations. This indicated the non-uniform distribution of stresses on the machined surface in the case of conventional cut and radial defects and that the stress conditions were inhomogeneous. Figure 17 shows the simulated machined surface topography comparing conventional cutting with the SDM incorporated SPDT. It may be seen that the surface profile obtained from the simulation in the case of circumferential pattern seems to have more uniformity compared to conventional cutting and cutting with SDM containing radial grooves. These results can also be reasonably compared (qualitatively) to the results obtained in the experimental study. The machined surface and sub-surface during machining with SDM having radial grooves can be seen to have jerkiness or less-uniformity, which is identical to the experimental observations, where cracking in the cutting chips was observed. On the other hand, in the SDM with circumferential defects there is more homogenous cutting chip generation and the stress distribution in the cutting zone while doing SDM containing circumferential defects at a spacing of 1 mm seems most focused compared to all other cutting configurations. Overall, it seems that the circumferential patterned defects produced better surface finish with less damage and defects in the sub-surface.

Conclusions
An experimental and numerical simulation investigation was pursued to highlight the underlying benefits of using SDM of silicon, which is classed as a nominally hard, brittle material. In the past, the SDM method was found to work successfully while machining hard steels by virtue of a reduced shear plane angle but the current investigation now shows that even ductile-mode machining of silicon can be improved by embracing the concept of SDM into diamond turning. Taking an example of the (111) surface of a silicon wafer, two types of surface defects were made (circumferential and radial directional grooves) and the quality of machining in both cases was benchmarked to conventional SPDT.
(a) The incorporation of SDM into the diamond turning of silicon provided favourable advantages over conventional diamond turning of silicon. The direction and orientation of surface defects to leverage this additional advantage are critical, for instance, prior machining defects made in the circumferential direction of the wafer, was observed to contribute to improving the cutting performance compared to SDM with radial patterns. The radial patterns deteriorated the quality of machining while machining silicon. (b) Cutting chip morphology showed distinct patterns for example, the SDM with radially patterned grooves resulted in striated featured chips and debris arising from brittle fracture. In comparison to this, the circumferential pattern showed improved ductility, however, the importance of width of cut was shown to be larger than the interspacing of the grooves was recognised to be an essential condition to get fully desirable benefits of the SDM incorporate SPDT of brittle materials.
(c) Laser Raman spectroscopy revealed amorphisation of silicon as well as the presence of Si-IV and a-SiO 2 phases confirming the influential role of tribo-chemistry during precision machining of silicon with a diamond tool.
Overall, this novel work on surface defect incorporated machining of silicon has opened new possibilities to apply this technique to a wider range of other hard, brittle materials such as SiC, ZnSe and GaAs as well as hard steels. Care must be taken to optimise the geometry of surface defects together with the machining parameters to obtain full advantages of SDM.

Data statement
All data in the manuscript will be available through Cranfield University data repository (10.17862/cranfield.rd.12442886).