Stephen D. FantoneOptical Activities in Industry is re-ported by Stephen D. Fantone, OptikosCorporation, 286 Cardinal MedeirosAvenue, Cambridge, MA 02141. Ste-phen welcomes letters, news and com-ments for this column which should besent to him at the above address.
Examination of the polished surface character offused silica
A. A. Tesar, B. A. Fuchs, and P. Paul Hed
Investigation of the surface character of fused silica polished with various compounds dispersed in wateridentified pH 4 as the optimum condition for high quality. Analyses support the conclusion that at thispH redeposition of hydrated material onto the surface during polishing is limited. Comparativepolishing results for Zerodur are included. Improvement of the laser-damage threshold of a coating onthe pH 4 polished fused silica is suggested.
IntroductionThe polished surface character of optical componentshas a tremendous impact on performance. Opticalcoating processes have undergone intense develop-ment over the past two decades. However, for only afew specialized applications such as laser gyros havepolishing processes shown recent progress. The per-formance-limiting factor of many optical components
The authors are with Lawrence Livermore National Laboratory,Livermore, California 94551.
is now the polished surface. In many cases thepolished surface becomes a substrate-coating inter-face. The nature and effects of the interface in thefinished optic are often disguised once they have beencleaned and coated, and thus they are often over-looked.
The importance of surface quality can be consid-ered in three fundamental areas. First, the opticalperformance of the substrate surface with respect toabsorption, scatter, and stability in changing environ-ments is critical for uncoated and transmissive optics.Second, the ability to clean a substrate depends on thesurface character. Avoiding contamination andstaining during cleaning as well as maintainingsmoothness are concerns. Third, studies continue
to indicate that substrate surface quality has a pro-nounced effect on the resistance of coatings to laserdamage and laser aging. Some related concerns aresubsurface polishing damage,' coating defects nucle-ated at the interface, coating adhesion, and coatingdelamination with time. For our purposes surfacequality implies a level of surface smoothness and theabsence of observable surface or subsurface defects.
While studying the effect of polishing compoundsand slurry conditions on the surface character offused silica, we found it helpful to separate theobserved phenomena in two categories. We foundthat (1) roughness could be minimized by controllingredeposition and chemical reactions (i.e., smoothnessis chemistry dependent) and (2) the polishing processefficiency (removal rate) could be considered some-what separately as dependent on mechanical parame-ters. Fused silica3 and Zerodur 4 substrate materialswere studied because of their widespread use forhigh-power laser optics. It is fortunate, however,that fused silica is also a simple glass. Multicompo-nent glasses add significant complexity to polishingbehaviors because of compositional changes whenthey are exposed to water for long periods of time.5This is also the case for Zerodur, a multicomponentglass-ceramic material. Surface analyses of Zero-dur, polished in the same manner as fused silica, areprovided in this study for comparison purposes.Full control of the Zerodur-polishing process, how-ever, is still being pursued. We analyzed the sur-faces resulting from careful and reproducible polishingexperiments by using complementary tools that in-cluded Nomarski microscopy, optical heterodyne pro-filometry (OHP), atomic force microscopy (AFM), andelectron spectroscopy for chemical analysis (ESCA).Hydrogen forward scattering and ellipsometry analy-ses of the polished surfaces are discussed elsewhere. 6 7
Experimental ProcedureA standard method was designed for all polishingexperiments to permit direct comparison of the results.Samples were prepared of optical-grade fused silicaand Zerodur. The 4.8-cm-diameter, 1.3-cm-thick,beveled disks were polished on all surfaces. One4.8-cm disk of each material was drilled to hold four1.2-cm-diameter analysis samples. The 0.4-cm-thick samples were held in countersunk impressionsflush to the surface by wax on the back surface. Wedrilled a small channel from the impression to theother side of the holder to allow for removal of thesmall parts from the holder without contacting thepolished surface. Samples were prepolished to re-move all subsurface grinding damage on a 120-cmcommercial planetary polishing machine. After be-ing prepolished, the surface figure was better thanX/4 with a roughness of 4-6 A rms. The small,experimental, continuous planetary machine 8 con-sists of a rotating 35-cm-diameter lap and three diskseptums with synchronized rotation. The lap rota-tion speed of 3 rpm resulted in an effective relativevelocity to the samples of 5 cm/s. The weights of the
other 4.8-cm-diameter samples were measured to±0.0001 g. A 416-g lead weight was then taped toeach disk, resulting in a total weight of 470 g and apressure of 26 g/cm2.
A new pitch lap was prepared for each experiment.Commercial Gugolz 73 pitch was used as the lapmaterial. The pitch was heated slowly until it be-came fluid, and then it was poured onto a prewarmedaluminum plate. The plate was then inverted onto asilicon rubber mesh lying on a polyethylene film-covered granite table. A square mesh was used as atemplate for grooves in the pitch. The mesh con-sisted of 0.2-cm ridges spaced 1.0 cm apart. Thepitch was pressed flat in this manner before the filmand mesh were removed and it was placed on themachine. The initial lap temperature of2l.6C (710F)typically rose 0.2C (1F) over the entire experiment.
The slurry was stirred constantly in a Pyrex beakerbefore its application to the lap. The flow rate of theslurry onto the lap was limited to 1.2 mL/min. Theslurry was not recirculated, and the excess flowedfrom the lap at the outer edge or into a center well.For a 6-h period prior to adding the samples, weconditioned the lap using a 15-cm-diameter Pyrexdisk. The conditioner remained on the lap for theduration of the experiment. The actual experimentconsisted of four consecutive 90-min periods of polish-ing. After each 90 min the three samples wereremoved. The two solid samples were rinsed thor-oughly with water, wiped with acetone, and thenweighed. The analysis samples in their holder weresubmerged in distilled water until they were returnedto the lap. The temperature and pH of the slurrywere also monitored at that time. The sample drag(or grip) was monitored during each run by manuallypushing the conditioner and a sample a distance of
2 cm and noting the resistance. Although thismonitoring was not quantitative, a comparison of thedrag between runs was categorized as skating (nodrag), low, medium, or high. All parts were rinsed indeionized water and then wiped with tissues wet witha 50% acetone, 50% methanol mixture before beingdried.
Once the 6-h polishing experiment was completed,surface roughness measurements were made with anOHP. The rms roughness reported is an average ofthree measurements for each sample. Observationsof the surface features when Nomarski microscopywas used were also recorded. One fused silica sam-ple from each experiment was etched and examinedfor subsurface polishing damage by an establishedmethod.' Preston coefficients were calculated fromthe weight-loss data. The Preston coefficient is ameasure of the efficiency of the polishing process.Because the mechanical parameters affecting theremoval rate are normalized, this coefficient poten-tially permits comparison with results from experi-ments with slightly different mechanical parameters.8In this situation we found it more useful than simplyrecording removal rates. The Preston coefficient C
CeO2 (Hastilite PO) 89 0.78 0.58 La, Nd, Y, Al, and Na presentbCeO2 (Opaline) 99.5 0.93 0.82 No Nd or Al detected, white in colorbZrO2 (TAM) 99.9 0.85 0.74 Electrochemical grade 3009A1203 (Ceralox) 99.99 - 0.60 Spherical particlesY203 (Research Chemicals) 99.9 2.0 1.2 Sharp edges, high aspect ratioYF 3 (Research Chemicals) 99.9 - 1.1 Rounded particles, agglomerated
aParticle size measurements using Horiba centrifugal sedimentation.bAnalysis by emission spectroscopy and inductively coupled plasma mass spectroscopy.
in cm2/dyn, or cm3/dyn cm, is given by polishing machines. For this reason several pH 7experiments included both Zerodur and fused silica
AM samples simultaneously being polished on the lap.C = pLAS'
where AM is the mass of material removed in grams, Resultsp is the substrate density in g/cm3 , L is the load in A summary of some of the pertinent parameters ofdynes, and AS is the total contact path length trav- each polishing experiment is given in Table 2. Drageled in centimeters. results from a combination of friction, suction, and
The slurries studied included several oxide com- negative lift as the sample maintains some level ofpounds and one fluoride compound at 0.14-vol. % contact with the lap.8 As previously shown9 there issolids in deionized water. Information on the pow- a good correlation between drag and the Prestonders is in Table 1. Both ceria and zirconia are coefficient (and thus the removal rate). No signifi-commonly used for glass polishing in the optics cant differences were observed with the 6 h of lapindustry. Little information on the polishing behav- conditioning time before we introduced the samplesior of the other three compounds was available. We and experiments where this did not occur.9 Theadjusted several slurries to pH 4.0 ± 0.2 by adding most efficient fused-silica-polishing process occurredcitric acid. Each slurry was stirred for at least 24 h with a fine cerium-oxide-polishing compound at pH 4before being used. Minor adjustments of additional as reported previously.9 The lower drag (and thuscitric acid were rarely required during the experi- low polishing efficiency) of fused silica with Opalinements to maintain pH in the stirring beaker. Expe- may be related to the larger particle size. Whenrience suggests that polishing Zerodur and fused Zerodur was polished concurrently with fused silicasilica simultaneously at pH 4 can have a deleterious on the lap at pH 7, the efficiency and drag wereeffect on the polishing efficiency and quality.9 similar to previous reports for separate polishing atHowever, a material resembling Zerodur is commonly pH 7, although the roughness is slightly higher.9
used as a conditioner with commercial continuous The polishing efficiencies of both fused silica and
Table 2. Parameters of Polishing Experiments
Preston Coefficient OHP Roughness
Compound pH Material Drag (x 10-14 cm 2/dyne) rms (A) P-V(A)
CeO2 (Hastilite PO) 7 SiO2a Medium 11 7 45CeO2 (Hastilite PO) 7 Zerodurb Medium 7 6 37CeO2 (Hastilite PO) 4 SiO2 High 26 2 7CeO2 (Hastilite PO) 4 Zerodur Medium 9 - -
Zerodur were again significantly higher at pH 4 thanat pH 7 with the zirconia-polishing compound. Thealumina compound resulted in low drag and corre-spondingly low polishing efficiencies for both samplematerials. The corresponding roughness measure-ments were also high. Yttria and yttrium fluorideprevented drag between the sample surfaces and thelap. These compounds did not promote polishingbehaviors. No analysis was pursued on the resultingsurfaces. We have also produced enhanced contact,efficiency, and superior surfaces on spherical parts offused silica and Zerodur with ceria at pH 4 by using aconventional spindle machine. This indicates thatthe pH effect is not limited to flat surfaces.
A good correlation exists between the roughnessmeasurements made by using OHP and AFM lineprofiles. The rms and peak-to-valley (P-V) valuesare listed in Table 3. The AFM line profile examineda distance of 4 [im as chosen by the operator. Thisaccounts for the extremely smooth surface measure-ment made on fused silica polished by the aluminaslurry when measured between severe pits, whichleads to a very high average OHP measurement of
300 A P-V. The other sample surfaces showvarying levels of texture and haze with Nomarskimicroscopy. The surface morphologies are clarifiedfurther when AFM is used. The corresponding im-ages are shown in Figs. 1-11. It is reassuring thatthe roughness values listed in Table 3 comparefavorably with the images. The more severe thetexture and haze, the higher the roughness value.
Figure 2. Nomarski micrograph at 200x and AFM surface profileof Zerodur polished at pH7 with Hastelite PO.
Figure 1. Nomarski micrograph at 200x and AFM surface profile Figure 3. Nomarski micrograph at 200x and AFM surface profileof fused silica polished at pH7 with Hastelite PO. of fused silica polished at pH4 with Hastelite PO.
Figure 4. Nomarski micrograph at 200x and AFM surface profile Figure 6. Nomarski micrograph at 200x and AFM surface profileof fused silica polished at pH4 with Opaline. of fused silica polished at pH7 with zirconia.
Figure5. Nomarski micrograph at 200x andAFM surface profile Figure 7. Nomarski micrograph at 200x and AFM surface profileof Zerodur polished at pH4 with Opaline. of Zerodur polislhed at p1H7 with zirconia.
Figure8. Nomarskimicrographat200xandAFMsurfaceprofile Figure 10. Nomarski micrograph at 200x and AFM surfaceof fused silica polished at pH4 with zirconia. profile of fused silica polished at pH4 with alumina.
Figure9. Nomarskimicrographat200xandAFMsurfaceprofile Figure 11. Nomarski micrograph at 200x and AFM surfaceof Zerodur polished at pH4 with zirconia. profile of Zerodur polished at pH4 with alumina.
aPolished concurrent with Zerodur samples on lap.bPolished concurrent with SiO 2 samples on lap.CAFM line profile of selected smooth surface area.
During polishing, material is hydrated and re- talline reaction products were detected on zirconia-moved from the sample surfaces. Depending on the polished silica. The presence of Al and Zr on thechemical environment, we can redeposit some of the fused-silica surfaces polished with compounds contain-hydrated material back onto the surface. The tex- ing these metals was detected by electron spectros-ture and haze observed is hydrated silica either in a copy for chemical analysis (ESCA). Table 4 presentscontinuous layer or as discrete particles. The soft, a list of elements identified on each surface by ESCA.hydrated material supports plastic deformation such Zerodur contains many trace elements, thus makingas grooves and indentations that form the texture identification of contaminants difficult. However, itthat is observed. The very thick, hydrated layer is clear that fused-silica and Zerodur samples withdisplayed in Fig. 1 shows such grooves clearly. high texture also contain detectable levels of slurryAdjusting the pH of the slurry to 4 decreases the contamination in the redeposited layer. For exam-ability of hydrated silica to reabsorb to the fused- ple, use of Hastilite PO ceria at pH 7 resulted in asilica surface. The resulting surfaces are smoother silica surface with an rms of 7 A and detectable Ceas demonstrated graphically by the corresponding and La contamination. The texture of fused silicaNomarski and AFM images. polished with the same slurry adjusted to pH 4 was
Other chemical effects include the chemical reac- not observable, and Ce and La contamination was nottion between the polishing compound and the sample detected. With the experimental conditions de-material. X-ray diffraction of the fused-silica sur- scribed, ceria polishing of fused silica at pH 7 simulta-face polished with alumina identified a reaction prod- neously with Zerodur produces surfaces that areuct AL2Si4 o10. It is suspected that the alumina actu- similar to fused silica polished separately. However,ally react with the fused-silica surface (not the when fused silica is polished simultaneously withhydrated silica in the slurry) leading to severe defects Zerodur at pH 4 the polishing efficiency and qualityin an otherwise extremely smooth surface. No crys- both deteriorate. It is suggested that hydrated Al
Table 4. ESCA Composition Dataa
Composition (at. %)
Compound pH Material Si C 0 N Ce La Zr Al P Ti Mg Ca K
aThe depth analyzed is approximately the top 100 A of the surface.bPolished concurrent with Zerodur samples on lap.CPolished concurrent with SiO2 samples on lap.
ions from the Zerodur may be affecting the slurry.10We did not observe subsurface polishing damage afteretching on the fused silica polished with ceria orzirconia oxide. The pitch was examined by usingscanning electron microscopy/energy dispersive spec-troscopy (SEM/EDS) after fused silica was polishedwith pH 7 and pH 4 ceria slurries. In the case of pH7 the lap became highly loaded with embedded ceriathat was covered with silica. At pH 4, however,there was much less ceria embedded, and no silica wasdetected.
Several 4.8-cm disks with varying polishing condi-tions were commercially coated with an electron-beam evaporated, high-reflection, quarter-wave stackcoating. All samples were coated in the same run.The outer layer was a half-wave silica overcoat. Thecoating is greater than 99.9% reflective at 578 nmwith <2-ppm absorption.' A difference in surfacetextures could still be viewed through the coating.Laser-damage testing of these coatings provided thelower and upper damage thresholds listed in Table 5following the method given previously.' The result-ing catastrophic damage thresholds are an indicationof the possible effects of substrate surface quality onlaser-damage thresholds of the coating. Long-termlaser-exposure tests have not been completed. Thedamage thresholds of the coated silica and Zerodursubstrates covered approximately the same ranges sothat neither substrate material appeared to be asuperior coating support. Surface quality does af-fect the laser-damage threshold of the coating on thesilica substrates. Smoother silica surfaces (withlower texture) polished with ceria do exhibit higherdamage thresholds than those with texture or pol-ished with other compounds leading to surface con-tamination. Zerodur surfaces polished with zirconiaappear to promote higher laser-damage thresholds ofthe coating regardless of the smoothness of thepolished surfaces.
Discussion
The laser-damage results indicate that the polishingcompound and substrate surface quality can affectthe coating damage threshold of fused silica. Fused
silica and Zerodur must be treated as two differentmaterials, and therefore optimized polishing proce-dures may differ. The polishing of Zerodur is beingstudied separately. The redeposition of hydratedsilica is noted to form a plastic layer that supportsmicrodeformations that appear as texture in Nomar-ski images. The hydrated silica is sometimes alsoredeposited as particles. This is not a new concept initself. Beilby" in 1921 discussed the formation of ahydrated layer during polishing. In this study aqualitative relationship was observed between thepresence of redeposited material and roughness.The minimization of a Beilby layer can contributepositively to performance. The coating-substrateinterface will be mechanically more stable. Coatingadhesion may also be enhanced. The amount ofwater at the interface that could diffuse with thermalor vacuum cycling is reduced. Mobile water couldlead to microscopic areas of high pressure at theinterface. Furthermore the redeposited material wasshown to hold contamination at the surface. Theresults indicate that adjusting the slurry to pH 4 andisolating the surface from contamination (especiallyother optical materials on the polishing lap) canminimize redeposition and maximize polishing effi-ciency. Polishing fused silica with a ceria or zirconiaslurry at pH 4 varies from water polishing in twomain respects. First, the ceria particles are sus-pended in the slurry rather than embedded in the lap.Second, polishing was continued for many hourswithout degradation of the resulting surfaces. Wa-ter polishing, however, can result in degradation ofsurfaces after a relatively short time, sometimeswithin an hour. It is suggested that several factorscontribute to the minimized redeposition and maxi-mized polishing efficiency when a pH 4 slurry is usedas follows.
Zeta potential numbers are used to describe thecharged nature of a surface. For most oxides inwater the isoelectric point (iep) coincides with thepoint of zero charge, i.e., the pH at which the zetapotential becomes zero. Values of pH for the iep ofspecified compounds often disagree for several reasons.Reliable zeta potential measurements are difficult toproduce and rely on empirical models. Furthermoresmall amounts of contaminants or aging of surfacescan change the surface character greatly. We willuse these numbers to suggest a model for the fused-silica polishing phenomena observed. Several pro-cesses appear to be working simultaneously to pro-duce the effects observed at pH 4 with fused silica.The iep of ceria is at - pH 7.5. At pH 4, ceriasurfaces support a positive charge. The high surfacecharge breaks up agglomerates and maintains theceria at a smaller particle size. This suggests thatthe number of particles interacting with the surfaceincreases, and the load supported by any one particlemay decrease. A corresponding increase in slurryviscosity around pH 4 as the suspended silica gelswould also enhance the drag during polishing andthus increase the removal rate. Exhibiting no charge,
the hydrated silica is not absorbed onto the ceriaparticles or on the pitch.' 2 Fresh'3 fused silica alsoexhibits no charge at pH 4.14 Therefore there is noattraction and no subsequent absorption or redeposi-tion of material onto the fused-silica surface. Thepolishing activity is carried out on a fresh silicasurface rather than a plastic, hydrated, redepositedlayer. This may enhance the removal rate further.Hypothetically the presence of hydrated Al ions in theslurry could affect the silica surface charge.
ConclusionsWhether or not the suggested model is essentiallyvalid, the results of this study can have a great impacton improving fused-silica polished surface quality andefficiency. Only a single polishing process is re-quired to produce high-quality fused-silica surfaceswhen a ceria slurry at pH 4 is used. Unlike waterpolishing the slurry and lap can be used for a longperiod without rapid deterioration of the polishedsurfaces. The improved surfaces are not limited to aflat geometry. It is indicated that a reduction in theredeposition of hydrated silica at the interface canimprove a coated optic's laser-damage threshold.
The authors acknowledge the contributions of JimYoshiyama, Lawrence Livermore National Labora-tory, for SEM/EDS analysis; Eric Langland, SurfaceScience Laboratories, Inc., for the ESCA analysis;and Chris Stolz and Tim Sarginson, Lawrence Liver-more National Laboratory, for assistance in damagetesting.
This research was performed under the auspices ofthe U. S. Department of Energy by Lawrence Liver-more National Laboratory under Contract W-7405-Eng-48.
References and Notes1. A. A. Tesar, N. Brown, J. R. Taylor, and C. J. Stolz, "Subsur-
face polishing damage of fused silica: nature and effect ondamage threshold of coated surfaces," in Laser-Induced Dam-age in Optical Materials '90, H. E. Bennett, L. L. Chase, A. H.Guenther, B. Newman, and M. J. Soileau, eds., Proc. Soc.Photo-Opt. Instrum. Eng. 1441, 154-172. (1990).
2. K. H. Guenther, "Nodular defects in dielectric multilayers andthick single layers," Appl. Opt. 20, 1034-1038 (1981); "Theinfluence of substrate surface on the performance of opticalcoatings," Thin Solid Films 77, 239-251 (1981).
5. A. Tesar and T. Oja, "Cursory examination of the zetapotential behaviors of two optical materials," UCRL-ID-109373 (National Technical Information Service, Springfield,Va., 1992).
6. R. Musket, R. Daley, and A. Tesar, "Forward recoil spectros-copy of polished fused silica surfaces." Nucl. Instrum. Methods(to be published).
7. K. Vedam, P. Chindaudom, and A. Tesar, "Characterization ofsurface layers on polished fused silica by ellipsometry," Mater.Res. Bull. (to be published).
8. N. Brown, "Optical fabrication," UCRL-misc-4476 (LawrenceLivermore National Laboratory, Livermore, Calif., 1990).
9. A. Tesar and B. Fuchs, "Removal rates of fused silica withcerium oxide/pitch polishing," in Advanced Optical Manufac-turing and Testing, II, V. J. Doherty, ed., Proc. Soc. Photo-Opt. Instrum. Eng. 1531, 80-90 (1991).
10. The presence of hydrated aluminum ions appears to influencepolishing behavior. The dissolution of 2 g of AJCl 3.6H20 in1000 mL of ceria slurry at pH 4 quickly resulted in decreaseddrag. The redeposition of material onto fused silica samplesurfaces became apparent within 30 min.
11. G. T. Beilby, Aggression and Flow of Solids (Macmillan,London, V, 1921), Sec. p. 81.
12. SEM/EDS analysis of used pitch and ceria surfaces.13. The word fresh is used to describe a silica surface that is newly
formed. The authors are aware that such a surface willcontain hydroxyl groups to some degree when it is formed in anenvironment containing water. A fresh surface, however, hasnot been heat treated, cleaned, or otherwise chemically treated.
14. F. Grieser, R. N. Lamb, G. R. Wiese, D. E. Yates, R. Cooper,and T. W. Healy, "Thermal and radiation control of theelectrical double layer properties of silica and glass," RadiatPhys. Chem. 23, 43-48 (1984). Note that the specific iep ofcolloidal silica in water is an area of controversy. The doublelayer formed between colloidal silica and water is sensitive toprocessing and minute levels of contamination (especiallychlorine, alkali salts, and acids). It is unfortunate that mostpublished experiments have examined systems where saltshave been added to render the suspension conductive.
