chemo-mechanical magneto-rheological finishing (cmmrf) of silicon for microelectronics applications

6
Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications V.K. Jain a , P. Ranjan b , V.K. Suri b , R. Komanduri (1) c, * a Indian Institute of Technology, Kanpur, India b Bhabha Atomic Research Center (BARC), Mumbai, India c School of Mechanical & Aerospace Engineering, Oklahoma State University, 218 Engineering North, Stillwater, OK 74078-5016, USA 1. Introduction Highly finished surfaces are in great demand in mechanical, electronic, and optical industries due to their respective superior properties [1–3]. Chemical mechanical polishing (CMP) process has the ability to finish silicon wafers using a polishing pad with abrasives and alkaline solutions that can react with the workpiece [2,3]. In CMP, chemical reactions between the slurry and the workpiece are used to enhance the finish quality and the degree of planarization [4–6]. The slurry consists of colloidal silica instead of standard abrasives [7]. The abrading forces in the CMP process depend mainly on the response of the polishing pad, which is less controllable, hence, lacks determinism in controlling the finishing action. The magneto-rheological finishing (MRF) process uses magnetically stiffened pad to deterministically finish optically flat, and spherical and aspherical surfaces [8]. The abrading forces in MRF process depend mainly on the behavior of polishing medium (or MR polishing fluid) [9]. Hence, to develop a better process, chemical etching during CMP [2,3,5–7,10] can be employed to improve the quality of the finish and MRF process for controlling the forces to get the best removal rate on the silicon substrate [8– 11]. In this paper, a new process, namely, chemo-mechanical magneto-rheological finishing (CMMRF) is introduced that com- bines the beneficial features of both CMP and MRF without the detrimental effects of either process. During material removal in CMMRF process, two actions simultaneously take place, namely, chemical reaction and mechanical abrasion. Chemical reaction of silicon with cerium oxide, deionized water, and oxygen converts silicon into soft silicon dioxide [12]. The chemically softened layers are then easily swept away by the mechanical action of comparatively soft abrasive and magnetic particles. Water molecules consisting of H + and [OH] ions attack Si–Si bonds to form Si–H and Si–OH, then all the four Si–Si bonds are replaced by Si–OH and a soluble silicate, such as H 2 SiO 3 . If all the four bonds are not broken then the silicon atom will not be detached as a soluble silicate but instead be present as an oxide on the surface. The Si–H bonds are also gradually replaced by Si–OH bonds. It has been observed that formation of soluble silicates is indeed favoured if the protective oxide on the silicon surface is removed: Si þ 3H 2 O ¼ H 2 SiO 3 ðsolublesilicateÞþ 4H þ þ 4e However, if a protective oxide layer is present on the silicon surface, the formation of soluble silicates is unlikely [6,7,12]: SiO 2 þ H 2 O ¼ H 2 SiO 3 ðsolublesilicateÞ Further reaction with soft abrasive (CeO 2 ) [6,7,13] is as follows: 2CeO 2 þ Si ! Ce 2 O 3 þ SiO Ce 2 O 3 þ 1=2O 2 ! 2CeO 2 CeO 2 þ Si þ O 2 ! CeO 2 SiO 2 ðsoftmaterialwitheaseoffinishingÞ Thus, chemical reactions soften the Si workpiece surface and MR polishing pad removes the softened (silicate and silicon oxide) surface layers from the peaks by mechanical action. The MR polishing pad will have abrasive particles (non-magnetic) in CIRP Annals - Manufacturing Technology 59 (2010) 323–328 ARTICLE INFO Keywords: Non-traditional machining Chemical mechanical planarization (CMP) Magneto-rheological finishing (MRF) ABSTRACT A new finishing process, namely, chemo-mechanical magneto-rheological finishing (CMMRF) was developed for polishing silicon blanks that combines the beneficial features of chemical mechanical polishing (CMP) and magneto-rheological finishing (MRF) without the detrimental effects of either process involved. Chemical reactions associated with CMP are used to enhance the finish quality while the magneto-rheological polishing fluid is used to control the magnitude of the forces acting on the workpiece that controls the material removal rates (MRR) and minimizes the surface integrity problems. An apparatus for CMMRF was designed and built for nanometric finishing of silicon substrates. This process is able to finish silicon blanks with nanometric finish, minimal surface defects, and higher removal rates. ß 2010 CIRP. * Corresponding author. Contents lists available at ScienceDirect CIRP Annals - Manufacturing Technology journal homepage: http://ees.elsevier.com/cirp/default.asp 0007-8506/$ – see front matter ß 2010 CIRP. doi:10.1016/j.cirp.2010.03.106

Upload: vk-jain

Post on 26-Jun-2016

216 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications

CIRP Annals - Manufacturing Technology 59 (2010) 323–328

Contents lists available at ScienceDirect

CIRP Annals - Manufacturing Technology

journal homepage: http: / /ees.elsevier.com/cirp/default .asp

Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon formicroelectronics applications

V.K. Jain a, P. Ranjan b, V.K. Suri b, R. Komanduri (1)c,*a Indian Institute of Technology, Kanpur, Indiab Bhabha Atomic Research Center (BARC), Mumbai, Indiac School of Mechanical & Aerospace Engineering, Oklahoma State University, 218 Engineering North, Stillwater, OK 74078-5016, USA

A R T I C L E I N F O

Keywords:

Non-traditional machining

Chemical mechanical planarization (CMP)

Magneto-rheological finishing (MRF)

A B S T R A C T

A new finishing process, namely, chemo-mechanical magneto-rheological finishing (CMMRF) was

developed for polishing silicon blanks that combines the beneficial features of chemical mechanical

polishing (CMP) and magneto-rheological finishing (MRF) without the detrimental effects of either

process involved. Chemical reactions associated with CMP are used to enhance the finish quality while

the magneto-rheological polishing fluid is used to control the magnitude of the forces acting on the

workpiece that controls the material removal rates (MRR) and minimizes the surface integrity problems.

An apparatus for CMMRF was designed and built for nanometric finishing of silicon substrates. This

process is able to finish silicon blanks with nanometric finish, minimal surface defects, and higher

removal rates.

� 2010 CIRP.

1. Introduction

Highly finished surfaces are in great demand in mechanical,electronic, and optical industries due to their respective superiorproperties [1–3]. Chemical mechanical polishing (CMP) processhas the ability to finish silicon wafers using a polishing pad withabrasives and alkaline solutions that can react with the workpiece[2,3]. In CMP, chemical reactions between the slurry and theworkpiece are used to enhance the finish quality and the degree ofplanarization [4–6]. The slurry consists of colloidal silica instead ofstandard abrasives [7]. The abrading forces in the CMP processdepend mainly on the response of the polishing pad, which is lesscontrollable, hence, lacks determinism in controlling the finishingaction. The magneto-rheological finishing (MRF) process usesmagnetically stiffened pad to deterministically finish optically flat,and spherical and aspherical surfaces [8]. The abrading forces inMRF process depend mainly on the behavior of polishing medium(or MR polishing fluid) [9]. Hence, to develop a better process,chemical etching during CMP [2,3,5–7,10] can be employed toimprove the quality of the finish and MRF process for controllingthe forces to get the best removal rate on the silicon substrate [8–11]. In this paper, a new process, namely, chemo-mechanicalmagneto-rheological finishing (CMMRF) is introduced that com-bines the beneficial features of both CMP and MRF without thedetrimental effects of either process.

During material removal in CMMRF process, two actionssimultaneously take place, namely, chemical reaction and

* Corresponding author.

0007-8506/$ – see front matter � 2010 CIRP.

doi:10.1016/j.cirp.2010.03.106

mechanical abrasion. Chemical reaction of silicon with ceriumoxide, deionized water, and oxygen converts silicon into softsilicon dioxide [12]. The chemically softened layers are then easilyswept away by the mechanical action of comparatively softabrasive and magnetic particles.

Water molecules consisting of H+ and [OH]� ions attack Si–Sibonds to form Si–H and Si–OH, then all the four Si–Si bonds arereplaced by Si–OH and a soluble silicate, such as H2SiO3. If all thefour bonds are not broken then the silicon atom will not bedetached as a soluble silicate but instead be present as an oxide onthe surface. The Si–H bonds are also gradually replaced by Si–OHbonds. It has been observed that formation of soluble silicates isindeed favoured if the protective oxide on the silicon surface isremoved:

Si þ 3H2O ¼ H2SiO3ðsolublesilicateÞ þ 4Hþ þ4e�

However, if a protective oxide layer is present on the siliconsurface, the formation of soluble silicates is unlikely [6,7,12]:

SiO2þH2O ¼ H2SiO3ðsolublesilicateÞ

Further reaction with soft abrasive (CeO2) [6,7,13] is as follows:

2CeO2þ Si ! Ce2O3þ SiO

Ce2O3þ1=2O2 ! 2CeO2

CeO2þ Si þ O2 ! CeO2�SiO2ðsoftmaterialwitheaseoffinishingÞ

Thus, chemical reactions soften the Si workpiece surface andMR polishing pad removes the softened (silicate and silicon oxide)surface layers from the peaks by mechanical action. The MRpolishing pad will have abrasive particles (non-magnetic) in

Page 2: Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications

Fig. 2. (a) Magnetically stiffened flexible magnetic brush and (b) squeezed MR

polishing fluid during finishing operation.

Fig. 3. Distribution of magnetic flux density on the surface of the magnet.

V.K. Jain et al. / CIRP Annals - Manufacturing Technology 59 (2010) 323–328324

contact with the workpiece surface and CIP/iron particles incontact with the magnetic pole [14]. A component of the magneticforce acts in a direction normal to the work surface and anothercomponent acts in the tangential direction. When the pad comes incontact with the workpiece surface, it is squeezed and it applies aforce normal to the work surface. The normal force thus adds to thenormal component of the magnetic force to increase abrasivepenetration into the workpiece surface. The tangential force actingon the abrasive particle due to the combined effect of rotation ofthe brush and a tangential component of the magnetic force,removes the material (soft) from the peaks of the workpiecesurface resulting in the formation of plateau on ‘the hills’ of thework surface.

2. Experimental set-up and the mechanism of polishing

A CMMRF apparatus (Fig. 1) was designed and fabricatedkeeping in view the fundamental mechanisms of the process andbasic functional requirements [15,16]. The rotary motion to the MRpolishing brush is provided by the vertical rotational system whichis driven by a three-phase induction motor, and its speed (rpm) iscontrolled by an independent variable frequency drive (VFD).Another motion system controls the feed rate in the x-directionthough VFD. The stroke length is controlled with a relay unit andlimit switches. The magnetic polishing tool is divided into twodifferent zones, according to the intensity of magnetic field(Fig. 2a). Zone 1 is the top surface of the magnet (N-pole). It isidentified as uniform magnetic field zone while Zone 2 is identifiedas non-uniform magnetic field zone.

Fig. 2(a) shows a photograph of the pad (from the bottom side)or formation of flexible magnetic abrasive brush (FMAB) on andaround the magnet. Fig. 2(b) shows the MR fluid segregated aftersqueezing during the finishing operation. The CMMRF processutilizes pad flexibility of smart magneto-rheological (MR) polish-ing fluid brush that acts as a tool. Its rheological behavior iscontrollable by means of magnetic field. In this process, a magneticlevitation force [14] is transferred to the active abrasive particles.

The force on the magnetic particles, under the influence of amagnetic field, is directly dependent on the magnetic flux densityand its gradient. In Zone 1, distribution of magnetic lines of flux isuniform which is responsible for constant magnetic flux densitythroughout the Zone 1. Hence this force is very small. It has alsobeen observed that there is a large gradient of magnetic fluxdensity in Zone 2 (Fig. 3). Hence, the magnetic force is much largerin Zone 2 as compared to Zone 1. It is also observed that themagnetic force becomes lesser as one goes far away from Zone 1.

All the above factors help in segregating the magnetic particlesin Zone 2. Thus, Zone 1 and outer side of Zone 2 do not exhibit alarge magnetic force. That is why a substantial amount of non-

Fig. 1. Schematic diagram of the CMMRF set-up [12].

magnetic particles (abrasives) get piled up in these regions. Thispiling up of the non-magnetic particles happens due to magneticlevitation force. As discussed above, during finishing operationsome reactive abrasives (CeO2) and non-reactive abrasives (Al2O3)diffuse from Zone 1 to Zone 2 during finishing. Some abrasives areconsumed due to chemical reaction while others fracture due tomechanical action.

MR polishing fluid consists of deionized water (DI), abrasives(SiC, Al2O3, or CeO2) and ferromagnetic particles (Fe). To avoidsedimentation of ferromagnetic and abrasive particles, glycerol(soluble in water) was added. Either iron particles or carbonyl ironparticles (CIPs) are used as ferromagnetic particles. MR polishingfluid forms a polishing brush (FMAB) which is magneticallysensitive. DI water as base material is chemically active on the (Si)work surface.

Fine particles provide better surface finish but they haveagglomeration problems due to van der Waals forces, whichgenerate scratches on the work surface. Glycerol is used toprevent solid particles from agglomeration and sedimentation.Two types of MR polishing fluid are prepared, one with single typeof abrasive (CeO2) and another with two types of abrasives CeO2

and SiC (or Al2O3). The first one is softer than the second one. Thefluid compositions are given in Table 1, where the mean diameterof Fe is 55 mm, CIP (HS grade) is 2 mm, CeO2 is 5 mm, and Al2O3 is6 mm.

3. Preliminary experiments

The experiments were conducted using two types of MRpolishing fluid: oil-based and water-based MR polishing fluid.

Page 3: Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications

Table 1Fluid compositions (vol%).

Type of fluid CeO2 SiC/Al2O3 Fe Magnetic particles DI water Glycerine

Fluid 1 5 – 36 – 56.8 2.2

Fluid 2 5 19 – 19 55 2.2

V.K. Jain et al. / CIRP Annals - Manufacturing Technology 59 (2010) 323–328 325

Surface roughness value (Ra) and percent change in Ra value (%DRa = [(initial Ra � final Ra)/initial Ra] � 100) were evaluated.Various process parameters identified were MR polishing fluidcomposition, magnetic field in the working gap, rotational speed ofthe magnet, feed rate to the workpiece and finishing time. Theexperiments were conducted in two phases: (i) finishing with oil-based MR polishing fluid and (ii) finishing with water-based MRpolishing fluid.

With oil-based MR polishing fluid, two sets of experimentswere conducted, one by varying the rotational speed of themagnet and the other by varying the mesh size of the abrasive. Asshown in Fig. 4(a), the maximum value of % DRa was obtained at�140 rpm. Hence, in the second set of experiments, the rotationalspeed was kept constant at 140 rpm and the mesh size was varied.In this case, the best surface finish obtained was at �400 rpm(Fig. 4(b)) and its Ra value was 20 nm but again some finescratches were seen.

Since the scratches were seen on the surface finished by oil-based MR fluid, in the second phase, finishing with DI water-basedMR polishing fluid was carried out (gap: 1–2 mm; rpm: 140; feed:5–6 mm/s). Surface roughness (Ra) after finishing for 7.25 hreduced from 1.21 mm to 40 nm. Fig. 5(a) shows the surface finishwith deionized water giving the surface roughness value as 4.8 A(0.48 nm) (Fig. 5(c)) using different stages of finishing [15], asdiscussed later. Fig. 5(b) shows a mirror finish obtained by water-based MR polishing fluid.

From the preliminary experiments, it is observed that hardabrasive particles produce cutting or scratch marks (or surfacedefects) on the workpiece surface that depends on the shape, sizeand relative hardness of the abrasive. Therefore, if soft abrasivesare used for finishing comparatively softer workpiece, it is likely togive better surface finish with reduced surface defects, if any.Hence, deionized water and CeO2 are used as the constituents of

Fig. 4. (a) Effect of rotational speed on f% change in Ra and (b) critical roughness vs.

abrasive mesh number.

the MR polishing fluid during final finishing by the CMMRFprocess.

Workpiece was finished in four different stages by varying theMR polishing fluid composition as follows:

Stage I (semifinishing). The surface roughness value (Ra)achieved in this stage of finishing is 0.1–0.2 mm. This stage offinishing is mainly done to level the work surface perpendicular tothe axis of rotation of the polishing tool. Constituents of the MRfluid used are abrasives (Al2O3 and CeO2 #2500), magneticparticles (Fe#250-300), and a base material (a mixture of DIwater and glycerol).

Stage II (finishing). This stage reduces the surface roughness upto 80 nm and levels small peaks formed due to chemical reaction ofsilicon and cerium oxide, and mechanical abrasion by iron particleson chemically reacted surface. The composition of MR fluid hadabrasive (CeO2#2500), magnetic particles (Fe#250-300), and basematerial (mixture of DI water and glycerol).

Stage III (fine finishing). Waviness was observed on the finishedsurface after Stage II (Fig. 6(a)). To remove the waviness, hard

Fig. 5. (a) AFM image of the finished surface, (b) optical view of silicon blank mirror

finished using CMMRF. The figure shows mirror reflection of IITK, BARC, DST, and (c)

surface roughness of the finished component (Ra = 4.8 A).

Fig. 6. (a) surface after finishing (Ra = 13.4 nm, Rz = 77.5 nm), (b) after semi ultra

fine finishing (Ra = 0.5 nm, Rz = 10.8 nm), (c) ultra fine finishing (Ra = 0.5 nm,

Rz = 5.2 nm).

Page 4: Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications

Fig. 7. Effect of finishing time on Ra, working gap: 1.0 mm.

Fig. 8. Effect of working gap on Ra, finishing time: 7 h.

V.K. Jain et al. / CIRP Annals - Manufacturing Technology 59 (2010) 323–328326

abrasive (Al2O3#2500) along with CeO2 (#2500) and tinymagnetic particles (CIPs, mean diameter 10 mm, HS grade andbase material (mixture of DI water and glycerol)) were used in theMR polishing fluid. Micrographs of the final finished surface areshown in Fig. 6(b). The total sampling area for each AFMmeasurement was 10 mm � 10 mm. Substantial improvementin this stage over Stage II is observed. However, some pits(Fig. 6(b)) are also seen.

Stage IV (ultrafine finishing). Final polishing was carried outwith very fine soft abrasive particles. Hence, for this stage, thepolishing fluid consists of a mixture of abrasives (CeO2 #2500),magnetic particles (CIP, HS grade), and base material (mixtureof DI water and glycerol). In this stage of finishing, criticallyfinished surface is shown in Fig. 6(c). It shows a very smoothsurface topography as well as roughness profile as compared toFig. 6(b).

Finishing time for all these four stages discussed above wasdetermined according to the time required to achieve criticalroughness value. Finishing time for the first stage was �2 h. Forstages 2 and 3, it was 1 h each, and the last stage of finishing hadvariation of finishing time from 1 to 5 h. Thus, the total time forfinishing was within 5–9 h.

4. Parametric analysis of CMMRF process

After obtaining encouraging results from the preliminaryexperiments, it was decided to conduct a parametric study ofthe process to delineate the effects of various important processparameters. The design of experiments and the analysis of resultswere conducted using the Design Expert1 7.0.0 software. Twentyexperiments were designed according to the central compositerotatable design methodology with three factors (Table 2). Basedon the sequential model sum of squares, lack of fit and p-value, theANOVA confirms the adequacy of the quadratic model (the ModelProb. > F and < 0.05). In the present case of Ra as a response, T, N,TG, TN, GN, T2, G2, and N2 are the significant terms [15]. The finalequation (Eq. (1)) of the response is evolved in terms of actualvalues of the factors:

Ra ¼ 0:29� 34:3� 10�3�T� 27:08� 10�2�Gþ 6:86� 10�4�N

þ 16:2� 103�TG� 70:6� 10�6�TN� 18:83� 10�5�GN

þ 15:38� 10�4�T2 þ 57:44� 10�3�G2 þ 23:0� 10�8�N2

(1)

5. Results and discussion

Based on the model obtained (Eq. (1)) after regression analysis,the effects of finishing time, working gap, and rotational speed ofpolishing brush, on Ra (after finishing) are as follows.

From Eq. (1) and Fig. 7, it can be seen that Ra decreases withincrease in the finishing time because the number of times anabrasive shears the peaks also increases. Hence, Ra value decreasestill it attains critical value at different rpm. Fig. 8 shows anoptimum value of the working gap at which minimum Ra value canbe obtained for different rpm. Beyond the optimum value of theworking gap, an increase in the gap reduces the indentation force

Table 2Coded level and actual values of the process parameters.

Sr. no. Parameters Units Levels

�1.682

1 Finishing time (T) h 4.98

2 Working gap (G) mm 1.40

3 Rotational speed (N) rpm 100

which reduces the depth of indentation. It results in reducedfinishing rate and hence poorer surface finish or higher Ravalue.

At working gap lower than the optimum value, Ra value ishigher with reduced gap due to lack of active abrasives in theworking zone. But beyond the optimum working gap, this problemdeceases because abrasives diffuse away from the inner part ofpolishing pad to the finishing plane/outer zone (Zone 2) due tomagnetic levitation force. Hence, final Ra value decreases withincrease in rpm.

As shown in Fig. 9, the roughness value increases with anincrease in the spindle rotational speed. As rpm increases, Coriolisforce acts in an adverse manner on the magnetically held abrasives

�1 0 1 1.682

5.80 7.00 8.20 9.02

1.52 1.70 1.88 2.00

141 200 259 299

Page 5: Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications

Fig. 9. Effect of rpm on Ra, working gap: 1.7 mm.

Fig. 10. (a) Initial surface topography an

V.K. Jain et al. / CIRP Annals - Manufacturing Technology 59 (2010) 323–328 327

in the working zone and splashes them out of the finishing zone. Itresults in the lack of abrasive particles in the working zone thatmakes it difficult to achieve better surface finish. Therefore,with increase in the spindle speed, Ra value increases. Further, asrpm of the spindle increases, the number of finishing cycles persecond also increases. Hence, it should increase the finishing rateor give a lower Ra value at the same finishing time. Hence, at lowfinishing time, the combined effect of both of these actions slightlyincreases the Ra value. Thus, at higher finishing time, Ra value islower.

6. Optimization study of CMMRF process

Optimization of CMMRF process was carried out using DesignExpert1 to obtain the optimum conditions of various variables,theoretical values of the responses (final Ra value), and theircorresponding desirability values (desirability value of ‘1’ indicatesideal case). The response, namely, Ra value was minimized and thevalues of the process parameters were kept in their respectiveranges (5.8 h < T < 8.2 h), (1.5 < G < 1.88), (141 < N < 259),(0.008 < Ra < 0.030) to make sure that the optimized solutiondoes not fall outside their respective ranges.

d (b) finished surface topography.

Page 6: Chemo-mechanical magneto-rheological finishing (CMMRF) of silicon for microelectronics applications

V.K. Jain et al. / CIRP Annals - Manufacturing Technology 59 (2010) 323–328328

Optimized experimental conditions and three theoreticalsolutions were obtained from the Design Expert Software. Forvalidation of the optimization results, one experiment wasconducted at the optimized parameters (finishing time: 8.2 h,working gap: 1.60 mm, rpm: 259) and the final Ra obtained was10 nm in place of 8.0 nm (theoretical). Hence, there seems to be�20% error. Fig. 10(a) and (b) show the initial and finished surfacetopography of silicon.

7. Conclusions

On the basis of experimental and theoretical work presentedhere, the following conclusions may be drawn:

1. C

MMRF process has the ability to finish silicon workpiece due tothe combined effect of chemical activation as well as mechanicalerosion. The best surface finish obtained is 4.8 A.

2. F

inishing with very fine abrasive results in superior surfacefinish. Implementation of the four different stages of finishing,results in ultrafinished surface without any scratches.

3. I

mportant parameters identified in the process are working gap,finishing time, and magnet rotational speed.

Acknowledgements

The authors acknowledge the financial support provided by theBhabha Atomic Research Center (BARC) for the project entitled‘‘Feasibility Study of Superfinishing Process for Silicon Mirror.’’Authors thank Dr. R. Balasubramaniam of BARC, Mumbai forsuggestions and Mr. Ajay Sidpara of ME Department of IIT – Kanpur(India) for assistance in the preparation of the manuscript. One ofthe authors (RK) thanks the A.H. Nelson, Jr. Endowed Chair inEngineering for additional financial support.

References

[1] McKeown PA (1987) The Role of Precision Engineering in Manufacturing of theFuture. Annals of CIRP 36:495–501.

[2] Venkatesh VC, Inasaki I, Toenshof HK, Nakagawa T, Marinescu ID (1995)Observations on Polishing and Ultraprecision Machining of SemiconductorSubstrate Materials. Annals of CIRP 44/2:611–618.

[3] Komanduri R, Lucca DA, Tani Y (1997) Technological Advances in Fine AbrasiveProcesses. Annals of CIRP 46/2:545–596.

[4] Komanduri R (1996) On Material Removal Mechanisms in Finishing ofAdvanced Ceramics and Glasses. Annals of CIRP 45/1:509–514.

[5] Bhagavatula SR, Komanduri R (1996) On Chemo-mechanical Polishing ofSilicon Nitride with Chromium Oxide Abrasive. Philosophical Magazine A74:1003–1017.

[6] Jiang M, Wood NO, Komanduri R (1998) On the Chemo-mechanical Polishingof Si3N4 Bearing Balls with CeO2. Transactions ASME Journal of EngineeringMaterials and Technology 120:304–312.

[7] Jeng Y-R, Huang P-Y (2004) Impact of Abrasive Particles on the MaterialRemoval Rate in CMP. Electrochemical and Solid-State Letters 7(2):G40.

[8] Magneto-Rheological Finishing. (1998) Center for Optics Manufacturing. U ofRochester. (www.opticam.rochester.edu).

[9] Steckenrider JS, Mueller BL (2003) US Patent No. 6,533,832.[10] Lu H, Obeng Y, Richardson KA (2003) Applicability of Dynamic Mechanical

Analysis for CMP Polyurethane Pad Studies. Materials Characterization 49:177–186.

[11] Golini D, Jacobs SD, Kordonski W, Dumas P (1997) in Ealey MA, Paquin RA,Parsonage TB, (Eds.) Precision Optics Fabrication Using MagnetorheologicalFinishing, SPIE CR67: Advanced Materials for Optics and Precision Structures.SPIE, Bellingham, Wash, pp. 251–274.

[12] Rajagopalan S, Mitra U, Pan S, Gupta K, Lin CM, Sery G, Mitta1 S, Hassejian K, LoWJ, Neubauer G (1993) Reaction of DI Water and Silicon and its Effect on GateOxide Integrity, Intel Corporation, RN3-16, 2200 Mission College Blvd., SantaClara, CA 95052.

[13] Tian YB, Zhou L, Shimizu J, Tashiro Y, Kang RK (2009) Elimination of SurfaceScratch/Texture on the Surface of Single Crystal Si Substrate in Chemo-mechanical Grinding (CMG) Process. Applied Surface Science 255:4205–4211.

[14] Jain VK, Singh P, Kumar P, Sidpara A, Das M, Suri VK, Balasubramaniam R(2009) Some Investigations into Magneto-Rheological Finishing (MRF) of HardMaterials. Proceedings of the ASME International Manufacturing Science andEngineering Conference Held at Purdue University (USA), .

[15] Ranjan P (2009) Development of Nanofinishing Technology for Silicon Sub-strate, M. Tech. Thesis, IIT Kanpur, India.

[16] Singh DK (2006) Investigations into Magnetic Abrasive Finishing of PlaneSurfaces, Ph.D. Thesis, I.I.T. Kanpur, India.