A Quantitative Comparison of Three Grolising Techniques for the ...

A Quantitative Comparison of Three Grolising Techniques for the Precessions TM Process D.D. Walker a,b, A. Baldwin d, R. Evans a, R. Freeman b, S. Hamidi f, P. Shore d X. Tonnellier c, S. Wei b, C. Williams e, G. Yu a a

University College London, at the National Facility for Ultra-Precision Surfaces, OpTIC Technium, Fford William Morgan, St Asaph Business Park, St Asaph, N. Wales, LL17 0JD b

c

Zeeko Ltd, 4 Vulcan Court, Vulcan Way, Hermitage Industrial Estate, Coalville, Leicester, UK, LE67 3FW

Cranfield University, at the National Facility for Ultra-Precision Surfaces, OpTIC Technium, Fford

William Morgan, St Asaph Business Park, St Asaph, N. Wales, LL17 0JD d

Precision Engineering Centre, Building 70, Cranfield University, Bedfordshire, MK43 0AL e

f

QioptiQ, Glascoed Road, St Asaph, Denbighshire, LL17 0LL

OpTIC Technium, Fford William Morgan, St Asaph Business Park, St Asaph, N. Wales, LL17 0JD ABSTRACT

The ‘Zeeko Classic’ polishing process is implemented in a series of CNC machine-tools. The standard tooling utilizes inflated membranes (‘bonnet’) covered with standard polishing cloths, and flooded by a supply of re-circulating polishing slurry. The usual input quality is a part off a precision CNC grinding machine, and the process both polishes and corrects form. In this paper we demonstrate how dynamic range can be substantially extended using three distinct Zeeko Grolishing processes that are hybrids between loose-abrasive polishing and bound-abrasive grinding. The output quality and volumetric removal rates of these processes are compared and contrasted. Finally, we note how these hybrid processes can extend the capabilities of the machine from polishing and form control, to smoothing parts with inferior input-quality, removing larger volumes of material during form control, and addressing harder materials. Keywords: Polish, grolish, Precessions, texture, sub-surface damage

1 INTRODUCTION Previous papers have detailed the basic operation of the PrecessionsTM CNC polishing process 1-11, as first implemented on a 200mm capacity automated machine-tool, and subsequently in a family of machines up to 1.2m capacity. The overall concept of all the Zeeko machines comprises 7-axis CNC control of the position and orientation of a tool as it traverses the surface of the part being worked (rotating or static), where the tool creates a local spot of action. There are then three main functions that the machine can perform: 1.

Grolishing. Fast removal using a variety of techniques intermediate between grinding and polishing.

2.

‘Pre-polishing’. This removes surface and sub-surface features from a prior grinding or grolishing process, whilst preserving surface form.

3.

‘Form correction’. This uses numerical optimization of dwell-times and, if required, other variables, based on measurements of both the surface and the removal profile (tool ‘influence function’).

In this paper we investigate the grolishing techniques, from the perspectives of volumetric removal rate, surface texture and sub-surface damage. The principal objectives have been to ‘sift’ a wide range of processes to assess their viability, and to identify preliminary process-parameters and performance. Experimental results on Zerodur have been obtained using a Zeeko IRP600 machine (Figure 10). We also include process results on silicon carbide, obtained using a dedicated CNC-controlled test-rig furnished with a standard Zeeko tool spindle (Figure 15). We report on three generic grolishing techniques, subdivided into different detailed techniques, as follows: 1.

Hard tooling

2.

Bound abrasives on inflated bonnet tooling (40 mm radius of curvature)

3.

Loose abrasives on inflated bonnet tooling (40 mm radius of curvature)

2. GROLISHING TOOLS USED IN THE EXPERIMENTS The tools are shown in Figure 1-9 below.

Figure 1 metal-bonded diamond pellets on hard tool

Figure 4 Single metal-bonded diamond pellet on 40mm bonnet

Figure 7 Polyurethane on 40mm bonnet

Figure 2 resin-bonded diamond hard ring tool

Figure 5 stainless steel washers on 40mm bonnet

Figure 8 3M Trizact TM on 40mm bonnet

Figure 3 nickel-bonded diamond pellets on 40mm bonnet

Figure 6 brass washers on 40mm bonnet

Figure 9 New bonnet with no active surface attached

Figures 1 and 2 above show hard tools based on tapered aluminium fixtures that interface to the tool-spindle Schunk chuck of the Zeeko machine and Test Rig. The taper defines a fixed precession angle, and the machine CNC must then be set accordingly to achieve that angle. Figures 3-9 show tools based on standard 40 mm radius of curvature bonnets. In Figure 1, the diamond pellets are inset into machined recesses in the metal fixture; in Figure 2 a resin-bonded diamond ring is bonded in place. Figure 3 shows a ring of metal-bonded diamond pellets inset into raised rubber lands moulded into a special Zeeko bonnet. This moulding geometry provides support whilst maintaining the flexibility of the bonnet. The tool is used at a fixed precession angle. In Figure 4, a single diamond pellet provides pole-down grolishing. Figures 5 and 6 show stainless steel and brass washers cemented to a standard bonnet for use with loose abrasives. Figure 7 is a standard bonnet with polyurethane cloth, also used with loose abrasives. Figure 8 is an example of a standard bonnet covered with 3M Trizact TM diamond pads. For comparison, Figure 9 shows a new bonnet.

3. ZERODUR EXPERIMENTS 3.1 Experimental procedure The experiments on Zerodur were conducted using the IRP600 machine (Figure 1) located in the National Facility at the OpTIC Technium in North Wales. Table 1 summarizes the principal experiment parameters for all the experiments on Zerodur, and Table 2 the detailed parameters and results for each experiment. For each experiment, touch-on of the tool with the part was established using the sensitive load cell in the tool spindle of the machine. The tool was then advanced towards the part by a further distance (‘Z-offset’) in Table 2, to pressurize the tool against the part. Volumetric removal rate was determined by scanning the processed part with a Taylor Hobson Form Talysurf profilometer, computing the volume removed over a well-defined area. This was divided by the time for that area. Surface textures were measured using an ADE Phase-Shift Micro-AXM surface-mapping microscope over a 0.8mm square area. Sub-surface damage was measured by programming the Zeeko machine to polish a wedged trench into the part, shallow at one end and deep at the other. The part was then etched with hydrofluoric acid to expose cracks. Photomicrographs along the trench were acquired to determine the distance along the trench (and hence the depth of material) at which the exposed cracks petered out. (See Figure 11 for an example) . Table 1 Parameters for all experiments on Zerodur Coolant Delivery

Water Mains to Drain

Initial texture

0.6μm

Initial Form

2μm over 100mm

Sample Size (mm)

100 x 100

Grolished area

80mm circle

Raster spacing Material Origin

0.5mm Zerodur Schott Figure 10 IRP600 at OpTIC

10mm, 0.31μm deep

15mm, 3.14μm deep

20mm, 5.61μm deep

25mm, 8.72μm deep

Figure 11 Example of four micrographs along etched trench for sub surface damage measurement

3.2 Results on Zerodur

millimeters

Brass bonded pellet on hard tool pole down

Nickel bonded pellet on bonnet pole down

Brass bonded pellet on bonnet pole down

Nickel bonded pellets on precessed bonnet

C5 Al Oxide on st.steel washers

C20 Al Oxide on st.steel oashers

9 um Trizact on bonnet

Ref. tool Figure # Precess Angle Surface direction Z offset mm Pressure bar H Speed rpm Feed mm/min Time min Final texture Sa (nm) Vol Rem rate mm3 /min ~ Form variation % Sub surface damage

Nickel bonded pellet on hard tool pole down Test No

D15 15µm Resin bonded ring on hard tool

Table 2 Detailed Experimental Conditions and Results for Zerodur

1

2

3

4

5

6

7

8

9

1

1

4

4

2

3

5

5

8

P Down

P Down

P Down

P Down

20 deg

22 deg

20 deg

20 deg

20 deg

1

1

1

1

All 4

All 4

All 4

All 4

All 4

.05

.05

0.5

0.5

0.3

0.5

0.5

0.5

0.5

0.5

0.5

1.5

0.5

0.5

1.5

1.5

1.5

1.5

500

500

1000

1000

500

1000

200

200

500

500

500

500

1000

1000

500

500

500

500

20

20

20

10

40

80

80

80

80

260

220

170

200

400

500

250

600

550

1.9

4.2

0.9

0.6

7.2

37

3

10

5.3

33

66

33

25

9

0.3

3

7.5

9

10 μm

In progress

18 μm

9 μm

17 μm

9 μm

14 μm

13 μm

In progress

Figure 12 Example of highly non-uniform removal with a hard tool (brass pellet poledown) . Note that the uniformity was systematically better with 1000mm/min feed-rate.

Baseline experimental conditions were established from previous experiments. As general guidance, loose abrasives have been found to require slower tool-speeds (circa 200rpm) The two results on hard tools (Tests 1,2) gave highly non-uniform removal, as per Figure 12. This is believed to be in part due to the mode of operation of hard tooling on the Zeeko machine, in contrast to the bonnet tooling for which the machine was designed. Machine linear and rotary axes utilize rolling element bearings. The bonnet then acts as a spring element, de-sensitizing the pressure exerted on the part, to errors of motion. The spring element also dampens vibrations, analogous to an inflated automobile tire. In the case of a hard tool, the spring element is essentially absent, and the system-stiffness is defined by that of the machine and the load cell in the spindle. The contact pressure with the part (and hence volumetric removal rate) is then critically dependent upon the precision of the tracking of the tool over the surface. Furthermore, dynamic effects in the machine can directly propagate into the part. With a Z offset of 50μm as per Table 2, a 10μm positional error would typically lead to a 20% variation in volumetric removal rate.

millimeters

microns

The pole-down configuration also gave inferior uniformity of removal compared with pellets on a precessed tool. In both cases, this is believed to be due to repetitive glazing of the pellets followed by re-exposure of the diamond. Poledown maintains continuous tool-contact with the surface, which amplified the problem, whereas the precessed tool gives an intermittent contact for a single pellet, allowing recovery. Nevertheless, it is interesting that the resin-bonded ring on a hard tool gave comparatively good results (which was not expected), and this requires further study.

Figure 13 Uniformity of removal using nickel bonded pellets on precessed bonnet (Test 6)

Figure 14 Uniformity of removal using C5 loose-abrasive on stainless steel washers on precessed bonnet (Test 7)

The nickel-bonded pellets on a precessed bonnet gave by far the highest removal rate and uniformity of removal, as can bee seen from Table 2 and Figure 13. However, it gave a comparatively rough result (Sa = 500nm), as one would expect from such an aggressive process. At the other extreme, the C5 loose abrasive on stainless steel washers on a precessed bonnet gave a low removal rate, but a good compromise between texture and uniformity of removal (Table 2 and Figure 14). Clearly, a two-stage process would combine the best features of both.

4 RESULTS ON SILICON CARBIDE 4.1 Overall Experimental Details The investigation of grolishing as applied to silicon carbide was conducted using the Precessions Test Rig shown in Figure 15. This is a simple mechanical system comprising commercial stepper-motor driven X, Y and Z slides and turntable, a manual angle stage, and a standard Zeeko polishing head. Simple raster and spiral tool-path software moves the polishing bonnet over a flat part, and the manual angle-stage permits a precession angle to be pre-set. The Test Rig is ideal for process development for investigating materials and abrasives without tying up a Zeeko machine, or needing to take special precautions to avoid its contamination by aggressive abrasives.

Table 3 Parameters for all experiments on SiC Feed Rate (mm/min)

5

Bonnet Pressure (bar)

1.5

H Axis Speed (rpm)

1000

Precess Angle

20 º

Precess Orientations

0 and 90 º

Material

Sintered SiC

Size (mm)

100 x 100

Origin

Boostec Figure 15 Precessions Test Rig

Polyurethane Cloth

St. Steel Washers

Brass Washers

9µm 3M Trizact

4

5

6

7

1mm

2mm

2mm

2mm

40mm Bonnet

40mm Bonnet

40mm Bonnet

40mm Bonnet

2

7

5

6

10mm line contact

10mm line contact

ø13mm

Water

Water

Water

8µm Paste

OD 11.5 mm ID 5.0 mm 8µm Paste

Re-circ.

Re-circ.

Re-circ.

Spread

3 Ground Sa = 233 nm, P-V = 5.75µm



152

136

2.38 NA

D12 (12µm) Resin Bonded Diamond Ring Tool

3 2, 1 & 0.5mm Rigid Ring Tool



Table 4 Experimental Conditions and Results for SiC



Ref. Fig. No.

4

2

Spot Size

10mm line contact

Parameter

SiC Test No Raster spacing Tool Type

Slurry Delivery Sample No Initial Condition Final Sa (nm) Final P-V (µm) Vol.removal rate

OD 9.5 mm ID 6.0 mm

ø10mm

8µm Paste

Water

Spread

Spread

Spread





172

28.7

152

159

142

1.224

2.255

0.729

6.77

5.32

5.21

NA

NA

Not measured

Not measured

0.18 mm3 /min

3.3 mm3 /min

Two of the three tests with the hard tool (2 and 1mm raster spacing) introduced cusps into the surface with a period equal to the raster spacing, and typical depths of 1 and 0.5 microns respectively. This was due to angular errors in the precise orientation of the hard grinding surface to that of the part (‘angle of attack’). This sensitivity to angular errors introduces significant process variability (hence no volumetric removal rates are quoted), and is a disadvantage of the hard tool. In contrast, the inflated bonnet-based tools are self adjusting and very tolerant of angular errors. 4.2 Surface-evolution during grolishing SiC with polyurethane cloth In this experiment, the raster tool-path (path of centre of polishing spot) covered an area of the surface 50mm by 50mm, which represented the fully-polished area. Volumetric removal rate was determined by integrating the volume within an influence function, over a measured grolishing time. Texture Sa was measured over a 0.8mm square patch of surface, using an ADE phase shift Micro-XAM white light interferometer surface-mapping microscope. Table 5 Progress of texture (grolishing with polyurethane) Sa P-V Polishing Runs (nm) (µm) Starting condition 152 3.3 After run 2 63 2.2 After run 4 39 2.0 After run 6 36 2.1 After run 8 32 1.3 After run 10 28 1.0 After last run 13 29 0.73

Figure 17 SiC surface after run 13 (3D map)

Figure 16 SiC surface after run 13 (surface map)

Average Roughness Measurements

Peak to Valley Measurements 70 60

2 50

1.5 P-V (µm) 1

Sa (nm)

Measured Texture (nm)

2.5

40 Sa (nm) 30 20

0.5 10

0

0

2

4

6

8

10

13

No. Polishing Runs

1

2

3

4

5

6

No. Polishing Runs

Figure 18 Polyurethane grolishing: Evolution of P-to-V (left and Sa (right) Detailed inspection of the micrographs clearly shows the existence of deep pinholes in the sintered Boustec material, consistent with its known porosity. It is these defects that are limiting the convergence of the P-to-V and Sa texture values as the grolishing process proceeds (Figure 18). In real use, the SiC would be over-coated with a homogeneous CVD layer.

5. CONCLUSION We have undertaken a preliminary investigation of variations on three generic types of grolishing processes, as applied to Schott Zerodur and Boustec silicon carbide. The hard tools with bound abrasives have been found to exhibit

significant process variability that renders them unviable. This is believed to be because the Zeeko CNC machine-tools are optimised for polishing with compliant inflated-membrane tools (‘bonnets’). These de-sensitise the process to mechanical errors, which retain their full effect with hard tools. In contrast, several highly serviceable grolishing techniques have been identified, implemented by applying different surfaces to standard polishing bonnets. Nickel bonded pellets on a precessed R40mm bonnet have provided by far the highest volumetric removal rate of the methods investigated for Zerodur (37mm3/min), together with the best surface uniformity and lowest sub-surface damage. Remarkably, the simple expedient of attaching stainless steel washers to a bonnet and using C5 or C20 aluminium oxide also provides an effective process, in terms of removal rate, texture and sub-surface damage. On silicon carbide, brass washers with diamond paste, and 3M Trizact, both on R40 bonnets, have been shown to provide serviceable processes. The results have demonstrated that the dynamic range of the Zeeko CNC polishing machines can be substantially extended using simple grolishing techniques. By extending the methods to larger bonnets, it is anticipated that a squarelaw on volumetric removal rate will be achieved (e.g. a factor of 16 on a standard R160mm bonnet), and further work will be undertaken to demonstrate this. The main roles of these processes will be i) to provide an intermediate step between classical form-generation and polishing, ii) to enable ‘cusping’ resulting from certain ultra-precision grinding machine to be removed rapidly, iii) to accelerate processing of hard materials such as silicon carbide, and iv) to provide a rapid aspherisation technique from an initial spherical form, and v) to provide an on-machine capability for applying bevels.

6. ACKNOWLEDGEMENTS We acknowledge support from the UK Particle Physics and Astronomy Research Council, and from the UK Research Councils under the Basic Technology initiative. The National Facility for Ultra Precision Surfaces is run jointly by University College London, Cranfield University, and the OpTIC Technium. D. Walker acknowledges a Royal Society Industry Fellowship.

7. REFERENCES 1. “A Novel Automated Process for Aspheric Surfaces” R,G. Bingham, D.D. Walker, D-H. Kim, D. Brooks, R. Freeman, D. Riley , Proc. SPIE 45th Annual Meeting, 2000, Vol. 4093 'Current Developments in Lens Optical Design and Engineering'; pp445-448 2. “The Zeeko/UCL Process for Polishing Large Lenses and Prisms” D.D. Walker, R. Freeman, G. McCavana, R. Morton, D. Riley, J. Simms, D. Brooks, A. King, proc. Large Lenses and Mirrors conference, UCL, March 2001, pub. SPIE, pp 106-111 3. “The first aspheric form and texture results from a production machine embodying the Precessions process”, D.D. Walker, D. Brooks , R. Freeman, A. King , G.McCavana, R. Morton, D. Riley, J. Simms, proc. Proc. SPIE 46th Annual Meeting, San Diego, 2001, vol. 4451, 2001, pp267-276 4. “Novel CNC polishing process for control of form and texture on aspheric surfaces”, D.D. Walker, A.T.H. Beaucamp, D. Brooks, R. Freeman, A. King, G. McCavana, R. Morton, D. Riley, J. Simms, proc. SPIE 47th Annual Mtg, Seattle, 2002, vol. 4451, pp267-276 5. “The Precessions process for efficient production of aspheric optics for large telescopes and their instrumentation” D.D. Walker, A.T.H. Beaucamp , R.G. Bingham, D. Brooks, R. Freeman , S.W. Kim, A. King, G. McCavana , R. Morton , D. Riley, J. Simms , Proc. SPIE Astronomical Telescopes and Instrumentation Meeting, Hawaii, 2002, Vol. 4842, pp73-84 6. ‘The ‘Precessions’ Tooling for Polishing and Figuring Flat, Spherical and Aspheric Surfaces’, D.D. Walker, D. Brooks, A. King, R. Freeman, R. Morton, G. McCavana, S-W Kim , Optics Express, Published by Optical Society of America on http://www.opticsexpress.org/, Vol. 11, issue 8, 2003, pp958-964 7. ‘Precessions Aspheric Polishing:- New Results from the Development Programme’, D.D. Walker, A.T.H. Beaucamp, R.G. Bingham, D. Brooks, R. Freeman, S.W. Kim, A. King, G. McCavana, R. Morton, D. Riley, J. Simms , Proc.

SPIE’s 48th Annual Meeting, the International Symposium on Optical Science and Technology, ‘Optical Manufacturing and Testing V’, San Diego, Vol. 5180, 2003, pp15-28 8. ‘First Results on Freeform polishing using the Precessions Process’, D. Walker, A. Beaucamp, C. Dunn, R. Freeman, A. Marek, G. McCavana, R. Morton, D. Riley, ASPE Winter Topical Meeting on ‘Free-Form Optics: Design, fabrication, Metrology, Assembly’, CD-Rom, ISBN 1-887706-33-X 2004 9. “New Results from the Precessions Polishing Process Scaled to Larger Sizes”, D.D. Walker, A.T.H. Beaucamp, D. Brooks, V. Doubrovski, M. Cassie, C. Dunn, R. Freeman, A. King, M. Libert, G. McCavana, R. Morton, D. Riley, J. Simms, Proc. SPIE Astronomical Telescopes and Instrumentation Meeting, Glasgow, 2004 10 ‘New results extending the Precessions process to smoothing ground aspheres and producing freeform parts’, D. D. Walker, A. T. H. Beaucamp, V. Doubrovski, C. Dunn, R. Freeman, G. McCavana, R. Morton, D. Riley, J. Simms, X. Wei, Proc. SPIE Vol. 5869, San Diego, Aug. 2005, pp 79-87 11. ‘Zeeko 1 metre polishing system’, D. Walker, R Freeman, G Hobbs, A King, G. McCavana, R. Morton, D. Riley, J. Simms, Proc. 7th Int. Conf Lamdamap, Cranfield, UK June 2005, p 240. 12. ‘Automated optical fabrication – First results from the new “Precessions” 1.2m CNC polishing machine’, D.D. Walker , A.T.H. Beaucamp, V. Doubrovski, C. Dunn, R. Evans, R. Freeman, J. Kelchner, G. McCavana, R., D. Riley, J. Simms, G. Yu, X. Wei, SPIE, Orlando, 2006, SPIE Vol. 627 paper 309 pp1-9