NASA-TN-1U117

Replicate Wolter-I X-ray Mirrors

D. E. EngelhauptThe University of Alabama in Huntsville

R. Rood, S. Fawcett, C. Griffith and R. KhanijowNASA Marshall Space Flight Center

Huntsville, AL

ABSTRACT

Cylindrical (hyperbolic - parabolic Wolter I) mirrorshave been electroformed from nickel over an electrolessnickel-phosphorous (NiP) plated aluminum mandrel insupport of the NASA AXAF-S x-ray spectrometer program.The electroless nickel was diamond turned and polishedto achieve a surface finish of 10 angstroms rms or bet-ter. Gold was then plated on the nickel alloy after anelectrochemical passivation step. Next a heavy layerof pure nickel was plated one millimeter thick withcontrolled stress at zero using a commercial PIDprogram to form the actual mirror. This shell wasremoved from the NiP alloy coated mandrel by cryogeniccooling and contraction of the aluminum to release themirror. It is required that the gold not adhere wellto the NiP but all other plated coatings must exhibitgood adherence. Four mirrors were fabricated from twomandrels prepared by this method. The area of eachpart is 0.7 square meters (7.5 square feet).

Fig. 1(NIPS

Electroformed Wolter I grazing incidence mirror95-05520) REPLICATE WCLTER-I N96-13162

X-RAY MIRRORS (NASA. MarshallSpace Flight Center) 12 p

Unclas

G3/74 0072796

https://ntrs.nasa.gov/search.jsp?R=19960003153 2018-08-29T16:18:02+00:00Z

1.0 INTRODUCTION

Fabrication of first surface optical components bycoating diamond machined aluminum with electrolessnickel and subsequently diamond machining and/orpolishing the nickel has been reported with typicalfinal component surfaces achieved about 20 to 50angstroms rms. The alloy preferred is 11 wt% phos-phorous, Nio (P) where the parenthesis denotes an amor-phous rather than crystalline deposit. A low stressdeposit with sufficient hardness to support an excel-lent surface finish is achieved at low plating rates bysuitable adjustment of the plating parameters.Ref.^1'^'3' Also cylindrical x-ray mirror fabricationby electroforming nickel over an aluminum mandrelcoated with electroless nickel and polished similarlyhas been reported. 1^/5)

In order to overcome numerous problems with qualitycontrol this task extended beyond the previouslyreported efforts. Since the flight mirror surface re-quirement was less than 10 angstroms rms, examinationof the stability of the polished Ni-^fP) surface oxida-tion was examined by SEM, AUGER and ESCA as well aselectrochemical polarization tests to determine the ex-tent of subsequent process degradation on the polishedand inspected surface. Also circularity requirementswere such that even very low residual internal stressin the electroformed nickel would cause excessive dis-tortion if not controlled in real-time. This requireda major effort to achieve the x-ray quality opticalcomponents using the first surface as a mandrel forreplication.

A commercial electroless nickel process was selectedwhich produced the requisite alloy with no indicationsof porosity when operated at low pH and high tempera-ture. Replication of the surface was at the gold in-terface to the nickel - phosphorous requiring a debond-ing oxide film formation without disturbing the 10angstrom rms polished surface prior to gold plating.The gold was plated on site from an alkaline non-cyanide electrolyte for three shells and from an offsite contracted vacuum evaporation system for thefourth shell. This was followed by a 1.0 millimeterthick nickel electroformed shell, with real-time con-trol of the internal stress, which became the actual

optical component. Separation of all four shells wassuccessful by using liquid nitrogen to cool the com-posite Al/Ni3(P)/Ni P Oz/Au/Ni assembly until themandrel shrank sufficiently to release at thegold/NixP Oz layer.

2.0 MANDREL MATERIAL SELECTION

The typical aluminum alloy selected for diamond turnedoptics is 6061 due to good diamond turning response.The material is typically purchased as wrought remeltalloy and contains a large amount of inclusion im-purities. No large item first melt pure aluminum addi-tion 6061 is known to be available in the U.S.

The aluminum mandrel alloy selected for the AXAF-S mir-ror program was 2024 since it has a higher copper con-tent which promotes less porosity in the plated coatingand also since it can be forged. The lower porosity isbelieved to be due to lower inclusions of iron and alsothe dissolution displacement of copper on the includediron particles during preparation for plating which inturn permits better coverage of inclusions by thenickel alloy plating process.

Since the parts are very large optics (0.33 metersdiameter by 0.71 meters long) it was necessary to forgethe mandrel blank. The alloy was received in the "F"condition after forging. Since 2024 is not a commonforging alloy precautions were taken to assure that noresidual stress would be in the material at the time ofdiamond turning noting that it is not necessary or evendesirable to achieve the highest hardness and strengthfor this application. Therefore the 2024 aluminummaterial was taken to T-6 and then overaged to relaxall internal stress from previous steps at the expenseof hardness and tensile strength reduction.

2 .1 Mandrel preparation procedures

1) 21 Hours at 850 deg. F.2) Forge until temperature drops to 700-740 deg. F.3) Reheat at 850 deg. F. for 3 hours.4) Second swage with temperature above 700 deg. F.5) Air cool and ship to MSFC.6) Solution heat treat at 920 deg. F per MIL-H-6088F.7) Boiling water "uphill" quench.

8) Precipitation harden 18 hours at 375 deg. F.9) Stress relief after machining 1 hour maximum at

500 deg. F.10) Reverse quench, LN2 stabilize, steam blast.11) Final machine.12) Post machine stabilize thermal cycle +200 deg. F

to -100 deg. F then to Room Temperature.13) Diamond machine allowing for nickel - phosphorous

plate.

2.2 Electroless nickel plating

The next step was electroless NiP plating. The platingprocesses were selected on the basis that no cyanide orhydrofluoric acid would be used. The Ni^ (P) platingprocess used a citric acid cleaner, HNOo deoxidizer andzincate cycle similar to most other processes but thenan electrolytic nickel strike from a glycolic acidbased solution was used to achieve adherence andcoverage of any inclusion sites. The process produceda very low plating rate at pH 4.2 which was also due inpart to the addition of sub-ppm stabilizer additiveswhich in turn promoted a deposit with outstandingquality performance. This was evidenced by the com-plete absence of micro-porosity over the entire 0.7meter2 (7.5 ft. sq.) surface of each of the twomandrels processed. The Nio (P) was heat treated at 160deg. C (320 deg. F) for 4 hours to promote additionaladhesion without crystallization to Ni^P.

Next the mandrel was sent to the diamond turning andpolishing lab where the surface was produced to lessthan 10 angstroms rms by re-turning the Nio(P) with asingle crystal diamond and polishing witn aluminumoxide and either diamond or silica gel. '

2.3 Electroless nickel passivation

In order to gain better control of the adhesion, ormore appropriately minimized adhesion of the gold, amore fundamental understanding of the nature and be-havior of oxide films on high phosphorous nickel alloyswas undertaken using potentiostatic methods along withSEM, AUGER, ESCA and AFM.

The formation of oxide films on metals proceeds dif-ferently depending upon many conditions. It must beconsidered that the oxide film is a mixture of metal,and oxygen and that unless supplied by another sourcethe metal is nominally diffused from the metal surfaceby a corrosion process. This in turn may well degradethe 10 angstrom polished surface of interest for thereplication process. First, spontaneous oxide forma-tion in humid air or in water occurs for nickel.

The pH, time and temperature determine the extent ofoxidation with a maximum at about 160 - 170 deg. F inpH 7.0 water or high humidity. The film formed willgenerally increase in oxygen toward the surface. Theoxide film is semi-conductive with local sites ofhigher conductance as a result of film formationdefects. The "defect" sites have been studied for manysystems and agreement is that increased ionic andelectron mobility occurs when the atomic arrangementsurrounding atoms of reduced coordination is not of thesame stoichiometry as the bulk film.

This generally causes patches of non-conductive oxidesin a field of ambipolar conductive points. Thus thestoichiometry and thickness for a spontaneous film arenot well controlled. Discolored gold deposits may oc-cur on this type surface and are most likely dendriticspires of gold occurring at the defect sites.Adherence will occur at these sites and the gold willnot plate smoothly across the non-conductive patchesdue to localized high current density at the defectsites which causes patchy microscopic diffusion limiteddeposits which are rough.

The need is therefore to achieve a very thin anduniformly conductive oxide (semi-conductive) film withno defect sites or with such a multitude of defectsites that the conduction appears to be continuous.The initial passivation process studied consisted of abrief immersion of about two seconds in 2% NaOH fol-lowed by 5% I^CrO-^ without current for five seconds.After forming a very thin oxide film and upon immersionin a hexavalent chromium solution a semi-conductivefilm (perhaps II-VI in nature) is formed and subse-quently permits uniform electrodeposition of gold withthe required low adhesion for subsequent release.

Since hexavalent chromium is both toxic and a substan-tial contaminant for other plating operations it wasdesirable to eliminate this step. The short immersiontimes found suitable on the small samples was notachievable on the actual AXAF-S hardware due to thesize and weight requiring a slow hoist.

An anodically formed film with an included cation wasbelieved to be most appropriate. The substitution ofnickel for the chromium would be most desirable. Thusalkaline processes containing nickel complexes were in-vestigated. When the nickel-phosphorous alloy is madeanodic in an alkaline nickel tartrate or other nickelligand forming solution, a conductive nickel rich oxidesurface rather than a non-uniform nickel oxide, phos-phorous surface is formed. For pure nickel in analkaline solution the progression of anodically inducedoxide formation is in the order; Ni(OH)2, NijO^, Ni202and NiC>2. By adding a complexing agent such astartrate, the progression of the oxide formation isreduced to two observable steps (i.e., slow) and a morenegative and stable reduction of the oxide is observedby cyclic voltammetry. See figures 2,3.

Nickel-Phosphorous alloy samples made anodic in analkaline nickel tartrate solution have been found toform a uniformly conductive Ni/P/oxide surface.

This surface formation can be controlled potentiometri-cally to form a film less than 10 angstroms thick con-taining nickel, phosphorous and oxygen which permits areadily deposited and subsequently, removable golddeposit. The presence of carbon was persistent in theoutermost film but appeared even in films formed fromreagent grade sodium hydroxide indicating that the car-bon was adsorbed from the air or from carbonates in thesolution.

2.4 Gold to Nickel-Phosphorous Adhesion

Initial tests plating from an acid gold cyanide solu-tion onto nickel-phosphorous alloy of 6% phosphorouswere inconclusive. By electrolyzing the nickel phos-phorous sample cathodically in an alkaline cleaner andthen passivating in a five gram/liter potassiumdichromate solution it was possible to achieve variouslevels of adhesion.

Using cathodic treatment followed by simple immersionin the chromate solution no release of the gold was ob-served. By making the sample cathodic in the cleanerfor 30 seconds and then anodic in the chromate solu-tion, complete passivation occurred and the highlystressed acid gold cyanide process failed to produce anadherent deposit causing blistering and peeling of thegold.

Reduction of the thin oxide film must not occur due tothe gold deposition process. The acid cyanide gold-cobalt process first employed plates with low ef-ficiency with the evolution of hydrogen concurrentwhich reduces the nickel oxide and possibly complexesnickel with cyanide. This results in adhesion of thegold to the substrate even with an oxide film initiallypresent.

Polished one inch diameter flat samples of 11 wt% phos-phorous NiP deposit were plated in an alkaline goldsulfite process. This process will not readily reducethe oxide film and permits deposition with lowadherence on the passivated NiP surface permittingseparation of the shell later. The Faradaic efficiencyof the gold plating must be near 100% which requirescareful control of the pH to a somewhat high value be-tween 10 and 10.5. The initial passivation consisted ofsimple immersion in 5 grams/liter NaOH for 30 secondsand 5 grams/liter I^CrO-^ for 5 seconds although themirror could not be handled this quickley.

A smooth gold deposit was obtained on the dichromatepassivated surface which readily lifted with cellophanetape. Electrodeposition of nickel onto these flatsamples resulted in excellent repeatable separation atthe gold/oxide layer.

Next tests were conducted to determine proper condi-tions for anodic passivation. Low anodic current from2 to 4 mA/cm from 5 to 10 seconds resulted in filmformation which could be gold plated and subsequentlythe gold could be lifted with cellophane tape. Whensamples were prepared using 5 mA/cm for 10 secondsanodic passivation in 5 gm/liter NaOH, the gold was notuniformly deposited and required replating. Additionaltests were performed using electroless nickel on coni-

cal aluminum samples. Both quantitative tensile test-ing with a load frame and qualitative tape pull releaseof the gold was readily achieved on samples by usingthe alkaline nickel tartrate solution. However whenthe much larger mandrels were prepared it was foundthat this relationship was not linear. A potentialmonitor was set up using a pH meter in the potentialmode connected to a recorder to monitor the formationof the oxide film on the actual mandrel in real-time.This allowed the oxide film growth to be observed todetermine the extent of the film formation. See figure4.

2.5 Gold plating process selected

The passivated Ni-^P) surface was plated using analkaline sulfite gold process at a pH of 10.5. Thepotentiometric passivation in 0.1 M nickel and sulfatebuffered pH 11.0 0.1 M Rochelle salt (sodium potassiumtartrate) was employed. A smooth gold deposit was ob-tained which readily lifted from the electroless Ni inmost cases but a very light haze was observed when ob-served under bright light. Also on occasion during thetesting of coupons the deposit was very dark and sopoorly adherent that it would peel spontaneously. Byreducing the natural oxide cathodically in 0.2 N sodiumhydroxide at 20 rtiA/cm2 for 5 minutes and then formingthe desired film anodically as previously described,better control was achieved.

Samples of the gold electroplated and separated fromthe passivated electroless nickel were subjected toESCA and AUGER analysis for chemical analysis todetermine if the oxide film was separated from the NiPand remained with the part.

The analysis showed some oxide and even some traces ofthe nickel plating process may have been present in theinterface. This was removed by immersion in 25 vol%^2Q2 overnight.

Vacuum deposition of pure gold on a clean Ni-11 wt% Psubstrate after argon sputtering (no electrochemicalpassivation) produced similar results in that overallvery low adherence was achieved.

Locally adherent areas of vacuum evaporated gold weredifficult to release with differential thermal expan-sion from one of the smaller test shells produced withthis process. This resulted in small patches of goldnot releasing from the mandrel on the test piece. For-tunately when the second large mandrel was coated withgold in a vacuum evaporation system the separation ofthe last electroformed nickel shell was readily per-formed with no such anomalous attached spots. This ispossibly due to the relatively long time period thepart remained in air prior to plating which would ex-tend the surface passivation.

2.6 Nickel electroforming

The use of an active sensor to monitor the stress in-duced in the sulfamate process electroformed nickelshell permitted an on-line method of controlling thestress in real-time. The current density first had tobe as uniform as possible. This was mapped using asecond monitor and changes in the configuration of theplating equipment were implemented to provide a uniformfield and subsequently uniform current density andthickness. The thickness of the part was latermeasured by CMM to be within 0.0025 centimeter varianceover the length of the part. Optical measurementsdemonstrated that the circularity and optical figurewere excellent when proper control was implemented butwithout this control the quality was not adequate forx-ray applications. See figure 5.

2.7 Final Cryogenic Separation of Shell:

After electroforming the nickel shell the part wasreturned to the diamond turning center and the shellwas trimmed to exact length on the mandrel. The partwas then cleaned for separation. The mandrel was hol-low and about two inches thick permitting the LN£ to bepumped into the center thus cooling the part from theinside. Care was taken not to locally heat any part ofthe shell to avoid thermal difference stress. Ther-mocouples were used on the outside and when the tem-perature dropped sufficiently an audible click washeard and the shell was free. This occurred slightlyabove freezing water temperature at the outside of theshell. The shell was lifted carefully straight up toavoid scratching the surface. The focus was better at

the longer x-ray wavelengths primarily due to scatter-ing of the x-rays however by the fourth mirror verynear flight quality results were achieved. ^°'

3.0 REFERENCES

1) "Observations on Polishing Electroless NickelPlated Surfaces"

R. Parks and C. Evans, April 1991 ASPE SpringTopical Meeting, Tucson, AZ

2) "Electroless Nickel Plating for Enhanced GoldCoating:A Lesson in Bimetallics"

Steven Scott, April 1991 ASPE Spring Topical Meet-ing, Tucson, AZ

3) "Influence of Phosphorous Content and Heat Treat-ment on the Machinability of Electroless NickelDeposits", Ibid

C. Syn, et al, Lawrence Livermore NationalLaboratories

4) "Results of X-Ray Measurements on ElectroformedMirror Shells for the XMM Project"

0. Citterio, P. Conconi, F. Massoleni, et al, In-terim report for European Space Agency for X-RayAstronomy Jet-X X-Ray Telescope, ESA Contract No.9685/91/NL/MS.

5) "Imaging Characteristics of the Development Modelof the JET-X X-ray Telescope"

0. Citterio, P. Conconi, F. Mazzoleni, et al, SPIEVol. 1546 Multilayer and Grazing IncidenceX-Ray/EUV Optics (1991)

6) "Wolter I X-ray Optics Produced by Diamond Turningand Replication" S. Fawcett, et al of NASA MSFC,Huntsville, AL. SPIE International Symposium onOptics, Imaging and Instrumentation, 24-29 July1994 in San Diego, CA.

L289

teag

S 1.688

-6.488

•1.888

-1.288 I

-fl.588 -7.568 -<.588 -5.598 -4.588

Io9<l/vn)Figure (2)

-1588 -1588 -I.K

0.1N NaOH0.1M Rochelle Salt0 .1M Ni+ +

•4.568 -7.588 -4.5B8 -1588 H.S68 -3.5B9 -2.568 -l.i

Figure 0 . 1 M NaOH R . T .

POTENFIOMETER

RECORDER

J

F3

TIME

POWER SUPPLY

POTENTIAL

AND

CURRENT

REFERENCE

ELECTRODE

PART

CATHODE

Figure 4. Schematic of monitor used to control passivation

Platng

Diaptragm

Press-re

Transducer I

T. OUT

Fluid Filed Crtairber

Housng

Gauge Plating Supply hput

To Anode

GAUGE VOLTAGE

SUPPLY

VOLTMETER

BASIC MOMTCR

Figure 5. Schematic of monitor for control of plating stress