Erik B. Svedberg. NIST & Seagate

Transcription

1 What Can Combinatorial Methods Do for Finding Protective Coatings for EUVL Mirrors? Examples from the Data Storage Industry Erik B. Svedberg NIST & Seagate

2 Data Storage: For 1Tb/in 2 d=6.5nm disc overcoat is less than 2nm

3 Magnetic Thin Film Media CoX alloys are the media of choice in current disk drives. Media parameters depend on its microstructure and can be changed by choosing various compositions of the CoX alloy adding Pt ==> Increases Ku, Hc adding Cr ==> Reduces exchange coupling, Reduces Ms adding B ==> Reduces grain size and by using an underlayer (e.g. Cr, NiAl) Control of film texture Control of grain size Lubricant Overcoat Magnetic layer Intermediate layer Underlayer Seed layer Substrate

4 What Can Combinatorial Methods Do for Finding Protective Coatings for EUVL Mirrors? The disc and the head both needs to be protected from corrosion. Today all drives are exposed to the ambient air Humidity, O -, Cl -, The overcoat adds space between the head and the data and thus signal loss. The thin overcoat needs to cover the whole disc surface with no pinholes. (For the head production the loss of a few heads on a wafer is not too bad.) They need to be dense, ~5-2Å thin, produced at less than 2 C. The driving metric for the functionality of the overcoat has shifted from wear resistance to corrosion protection Density and film integrity are now more important than hardness

5 Head & Disk Corrosion Disk and head corrosion can be divided into two parts: Oxidation (anodic) reaction: M = M +n + ne - M = Co, Fe, Ni, Cu. Reduction (cathodic) reaction: 2H 2 O + O 2 + 4e - = 4OH - For oxidation to take place: Access for H 2 O and O 2 or 4OH - and metal atoms to meet Path for electrons to travel from location of oxidation (anodic) reaction to location of reduction (cathodic) reaction The role of atmospheric pollution (e.g. Cl 2, SO 2, NO ) is to disrupt the protective oxide that forms on some metals (e.g. Ni, Cr, Ta ) Under normal moist air environmental conditions Co and Cu do not form protective oxides Cr and Ta can form protective oxides over Co when part of a CoCr or CoTa or CoCrTa alloy system Pt will not protect Co from corrosion at less than about 75% Pt Co is likely to leached out of the CoPt alloy leaving a hollow structure of Pt behind similar to the CuAu system

6 Lubrication Perfluoropolyether (PFPE) lubricants with molecular weights in the range of 2 6 are most commonly used to lube disk Different end groups denoted by different tradenames e.g. Fomblin Zdol has hydroxl end groups CF 2 CH 2 OH and Fomblin Zdiac has carboxylic acid end groups CF 2 COOH Bonded lubricant is defined as lubricant left on the disk after rinse in a non-polar solvent Mobile lubricant is defined as lubricant that washes off the disk when rinsed in a non-polar solvent

7 Carbon for Corrosion Protection Slider is coated with diamond-like carbon (DLC) (layers approaching 1-2 nm) Media is coated with amorphous carbon (thickness is approaching 2-3 nm) Carbon coating properties are determined by the deposition technique higher deposition energy -> higher sp3 content higher sp3 content -> higher electrical resistivity higher sp3 content -> harder, higher modulus film But, higher hardness, modulus -> higher residual stress (bad - delamination)

8 A plasma beam with macroparticles and neutral atoms is emitted from the cathodic arc spot by a cathodic vacuum arc process Unwanted macroparticles and neutrals are then be filtered out by a cross-magnetic and electric field Only ions within a well-defined energy range are allowed to reach the substrate Nearly 1% carbon ion during deposition Substrate can be biased FCA Deposition Striker Filter field Vacuum arc supply Optimum energy per carbon atom to achieve sp 3 bonding and high density is ~ 1 ev Any magnetic deadlayer thickness created has to be added to head media spacing Scanning magnetic field Carbon plasma Substrate Vacuum chamber

9 Carbon surface modifications N + implantation Objective: achieve higher surface N content without decreasing the bulk density Method: N + implantation from plasma into the surface of high density i-ch Mass Density: ~1.98 g/cm 3 N/C Ratio (ESCA): CoOx comparable to I-CH Transfer function (coded units): N/C ratio = * N2 Flow +.29 * Duration +.5 * N2 Flow * Duration CoOx by ESCA [%] N ions I-CH N-rich CoOx before and after 4-day 8/8 expose i-ch before expose N implanted i-ch before expose i-ch after expose N implanted i-ch after expose Carbon Thickness by ESCA [Å]

10 Carbon surface modifications Ar + etch Objective: increase the number of carbon surface dangling bonds available for lube bonding Method: short duration Ar + ion bombardment to reduce the H content to reactivate the surface Transfer function: Lube: Ztetraol, 1 Å Lube process: vapor lube Ar + Ar + H H H H C H C C C C C C C C H C H C C C C H C Ar_flow: 1, 2 sccm; Sub_bias:, 12V; Duration:.5, 1 sec F C F F CF F F O C F C C C C C C C H C H C C C F Lube Bonded Ratio = Ar_flow Sub_bias Duration Ar_flow*Sub_bias Sub_bias*Duration 1.86 Ar_flow*Sub_bias*Duration

11 Carbon Structure Triangle sp 3 (Diamond) Filtered cathodic arc ta-c Ion-assisted (PECVD, IBD) a-c sp 2 (Graphite) Sputter i-ch a-ch polymer Polymer CH 2 - Tetrahedral amorphous carbon (ta-c) by FCA is a very good media overcoat High sp3 content (up to 83-88%) High density (~2.7 g/cm 3 from XRR vs. IBC~2. g/cm 3 and SC~1.6 g/cm 3 ) High hardness (up to 75 GPa) High resistivity

12 Why Carbon? Low temperature depositions yield amorphous films even at high ad-atom energy No grain boundaries to allow easier diffusion of atmospheric corrosion promoters (moisture, Cl, S ) Chemically inert, but Can enhance bonding of lubricants by adding nitrogen Forms carbides with most of metal films used in heads and disks enhancing adhesion Wear is dominated by Tribochemical processes (burning) No 3-body wear as long as no adhesion failures Excellent topological conformity Cheap and easy to integrate into sputtering process equipment used to sputter magnetic layers (@6-1 disks per hour)

13 Possible carbon design strategy 2 or 3 layer design: N%, Bias, Thk Ratio Bias, H%, Power, Temperature Intermix layer i-cnh for Lube interaction, Flyability: N2 i-ch for Corrosion, Durability (Thickness ratio, Density ) Low energy i-ch for minimizing intermixing while maintaining improved coverage Magnetic layer

14 Atomic layer deposition Alternative coatings Al 2 O 3 Al 2 O 3 by ALD at 1, 15, and 2Å Process not yet optimized for media Ta 2 O 5 1% CoOx detected even before stress exposure CoOx percentage by ESCA [%] CoOx before and after 4-day 8C/8H expose Before expose After expose Al2O3 Thickness [Å]

15 Alternative coatings a-sin x Rf sputtered a-sin x : a-sin x has a low coverage limit of ~1Å (cf. a-cn x ~2Å) a-sin x has 93% of bulk density (3.2g/cm 3 ) cf. DLC ~54% of diamond Film coverage thickness 3Å if relative density 1% However, oxidation seems to be a problem and is accelerated by H 2 B.K. Yen et al. JAP 93(1) p

16 8 / 8 - Testing Measure CoO x content and count surface defects in film after 4 days of 8 C and 8% humidity

17 Corrosion evaluation potentiodynamic scan Potential (V) E vs log( I ) RF 2Å E-1 1.E-9 1.E-8 1.E-7 1.E-6 1.E-5 Current (A) ECR 17Å C21 SINX3PP1 (Ovl) SINX7PP1 (Ovl) C21_PP32 C21: 9Å Si / 13Å FCA Pitting potential for C21 ~ 5-7mV Most of the SiN x films did not show pitting up to 8mV Overall the corrosion resistance of SiN x films is as good as or better than C21 control Pinhole decoration technique: Cu electro-deposition from a sulfate solution at.1v highlights weak areas

18 Summary Ion beam & FCA carbon show the following advantages over sputter carbon: Higher mass density Lower ESCA CoO x take-off thickness Higher surface coverage Some issues for ion beam & FCA carbon: Mag/C interface mixing Additional HMS due to the magnetic dead layer Degradation of corrosion resistance at very thin film <= 2Å Metrology Current ESCA carbon thickness measurement overestimates the higher density IBC film, which is indicated by both XRR and electrical test. FCA carbon is currently the strongest candidate for next media carbon overcoat. More evaluation and development effort is needed, especially on the manufacturability and particle issues. Atomic layer deposition (ALD) is a potential method to engineer the media overcoat one layer at a time

19 Combinatorial Materials Science A two stage process: I. Sample creation II. Sample analysis

20 Combinatorial Materials Science A two stage process: I. Sample creation Sequential deposition Parallel deposition II. Sample analysis Sequential measurements Parallel measurements

21 Combinatorial Materials Science A two stage process: I. Sample creation Sequential deposition Parallel deposition Knowledge II. Sample analysis Sequential measurements Parallel measurements

22 Knowledge Optimal operation point for your film Alloy composition with highest Coercivity Thickness that gives proper protection Transfer Functions Coercivity dependence on composition Ability to create a higher level model

23 Look and see More real Knowledge Knowledge Optimal operation point for your film Alloy composition with highest Coercivity Thickness that gives proper magnetic coupling Transfer Functions Coercivity dependence on composition Ability to create a higher level model

24 Combinatorial Materials Science Sample creation Sequential Or In parallel

25 Multiple parameters Thickness and composition: Combinatorial materials science delivers full control over both thickness and composition of a layer. A Thickness B Composition A Min thickness B A+B Max thickness

26 Combinatorial Materials Science Sequential Or In parallel E.B. Svedberg, in Combinatorial and High-Throughput Discovery and Optimization of Catalysts and Materials, Eds. R.A. Potyrailo and W. Maier, CRC Press, (26).

27 Combinatorial Materials Science: Multilayers High % Cr Co 1-x Cr x Low % Cr Thin Thick Pt Co 1-x Cr x Pt Co 1-x Cr x Substrate E.B. Svedberg, R. Van de Veerdonk, K.J. Howard, and L.D. Madsen, Method for Seed and Underlayer Optimization of Perpendicular Magnetic Recording Media, JVST A, 2(4), 1341 (22).

28 Combinatorial Materials Science E.B. Svedberg, CoCr/Pt Multilayers with Perpendicular Anisotropy and Texture-controlled Coercivity, J. Appl. Phys., 92(2), 124 (22).

29 Combinatorial Materials Science Look and see E.B. Svedberg, CoCr/Pt Multilayers with Perpendicular Anisotropy and Texture-controlled Coercivity, J. Appl. Phys., 92(2), 124 (22).

30 Combinatorial Materials Science 1, Cr, Ta, CrTa, Cr^2, Ta^2, Cr^2Ta Cr Cr Cr Ta CrTa

31 Combinatorial Materials Science 1, Cr, Ta, CrTa, Cr^2, Ta^2, Cr^2Ta Cr Cr Cr Ta CrTa More Knowledge

32 Math: Combinatorial Materials Science Here the coercivity is described as a transfer function from Co, Cr and Ta concentration in a media layer Cr Cr Cr Ta CrTa Remember: The models cannot predict data outside the interval measured. Model checking is needed by residual plots and p-values e R e R %Cr % Pt

33 Math: Combinatorial Materials Science Here the coercivity is described as a transfer function from Co, Cr and Nb concentration in a media layer Cr Cr Cr Ta CrTa Remember: The models cannot predict data outside the interval measured. Model checking is needed by residual plots and p-values e R e R %Cr % Pt

34 :ParameterTable Finding a Model Estimate SE TStat PValue Cr Cr , Cr Cr Ta Ta RSquared , AdjustedRSquared , EstimatedVariance.52841, > DF SumOfSq MeanSq FRatio -.1 PValue Model ANOVATable Error Hc Total Re Re Re Re CrTa Ok Ta NOT Ok!!!!! th Really small errors! Re %Cr %Nb Re %Pt %Ta Hc=k +k 1 Cr +k 2 Cr^2 +k 3 Cr^3 +k 4 Ta +k 5 CrTa Re %Ti y x Cr Cr Cr Ta CrTa

35 :ParameterTable Finding a Model Estimate SE TStat PValue Cr Cr , Cr Cr Ta Ta RSquared , AdjustedRSquared , EstimatedVariance.52841, > DF SumOfSq MeanSq FRatio PValue Model ANOVATable Error Total CrTa Ok Ta NOT Ok!!!!! The P-value can be regarded as the smallest α level at which the data is significant. Here, α is the probability of a type I error. (A type I error is to have the parameter rejected when it is true). The values of P is normally accepted when they are below.5. Thus,.5 is value that indicate the level of risk one is willing to take in order to include that base function in the model Cr Cr Cr Ta CrTa

36 :ParameterTable Finding a Model Estimate SE TStat PValue Cr Cr , Cr Cr Ta Ta RSquared , AdjustedRSquared , EstimatedVariance.52841, > DF SumOfSq MeanSq FRatio -.1 PValue Model ANOVATable Error Hc Total Re Re Re Re CrTa Ok Ta NOT Ok!!!!! th Really small errors! Re %Cr %Nb Re %Pt %Ta Hc=k +k 1 Cr +k 2 Cr^2 +k 3 Cr^3 +k 4 Ta +k 5 CrTa Re %Ti y x Cr Cr Cr Ta CrTa

37 Combinatorial Materials Science Cr Cr Cr Cr Ta Hc 2.5 Ta = Dependence on Cr %Cr Dependence on Cr and CrTa Hc %Cr

38 Hc Cr Cr Cr Cr Ta Dependence on Cr Ta =.65 Combinatorial Materials Science %Cr Dependence on Cr and CrTa Hc %Cr TEM study showed that an alloy with 15% Cr had a concentration of 23% Cr within 1 nm of the grain boundary while the center of the grains were at half the average concentration. Extended x-ray absorption studies, show that the Ta atoms are preferentially distributed to the Cr enriched regions in the CoCr films and that they have a direct effect on the average local environment of the Cr atoms by introducing a large amount of local structural disorder. E.B. Svedberg, R.J.M. Van de Veerdonk, K.J. Howard, and L.D. Madsen, Quantifiable combinatorial materials science approach applied to perpendicular magnetic recording media, J. Appl. Phys., 93(9), 5519 (23).

39 Combinatorial Corrosion studies Fluid-cell AFM in continuous mode is an excellent method for monitoring accelerated corrosion progress in-situ, and generating valuable insight into the corrosion mechanism. Case study Ta y Zr x O protecting Ni

40 Statistical Approach (Cont d) Oxides Thickness Gradient + + TaO - - ZrO Areas most resistant to corrosion

41 Nickel (2nm)

42 Nickel (2nm)

43 Nickel (2nm)

44 Nickel (2nm)

45 Nickel (2nm)

46 1.5 nm Thick Ta/Zr Oxide Dot (8 min.)

47 1.5 nm Thick Ta/Zr Oxide Dot (8 min.)

48 .8 nm Thick Ta/Zr Oxide Dot (4 min.)

49 Model For Ta y Zr x O on Ni Mean Onset FOM md y@m 1 md y@m + 1 md y@m x@mmd x@mmd x@mmd Amount of corrosion+ Onset time FOM

50 md TaO thickness TaO thickn.@ad Developing & Fitting a Model 3 FOM FOM Model FOM =3.9-.2*Ni th+.12*tao th-.3*tao th 2 3 Model md y@m x@mmd Ni thickness Ni thickn.@ad x@mmd md y@m x@mmd md y@m x@mmd Estimate SE TStat PValue ThNi ThTaO ThTaO

51 TaZrO Ni Gradient Vacuum breach md Mean md y@m Onset md y@m FOM TaZrO Ni Gradient No Vacuum breach md y@m x@mmd Mean md y@m x@mmd Onset md y@m x@mmd FOM x@mmd Mean x@mmd Onset x@mmd FOM TaZrO Uniform Ni Vacuum breach md y@m x@mmd md y@m x@mmd y@m md x@mmd

52 So.What Can Combinatorial Methods Really do for Finding Protective Coatings for EUVL Mirrors? Overcoat gradients could be deposited directly onto multilayer structures with all process conditions kept identical. Many samples could be exposed simultaneously to EUVL. Automated scanning of the analysis tools are needed for speed. Transfer functions can quickly reveal complicated interactions and guide the next material choice. Thanks!