Determination of laser shock treatment conditions for fatigue testing of Ni-based superalloys

HAL Id: jpa-00251764

https://hal.archives-ouvertes.fr/jpa-00251764

Submitted on 1 Jan 1993

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Determination of laser shock treatment conditions for fatigue testing of Ni-based superalloys

Philippe Forget, Michel Jeandin, A. Lyoret

To cite this version:

Philippe Forget, Michel Jeandin, A. Lyoret. Determination of laser shock treatment conditions for fatigue testing of Ni-based superalloys. Journal de Physique IV Proceedings, EDP Sciences, 1993, 03 (C7), pp.C7-921-C7-926. �10.1051/jp4:19937142�. �jpa-00251764�

Determination of laser shock treatment conditions for fatigue testing of Ni-based superalloys

F? FORGET, M. JEANDIN and A. LYORET*

Ecole des Mines de Paris, Centre des Matkriaux PM. Fourt, BP 87,91003 Evry cedex, France

* SNECMA, Centre de Corbeil, Dkpartement "Matkriaux et Prockdks", Bl? 81, 91003 Evry cedex, France

Abstract.

It i s envisaged that laser shock surface treatment may be used t o surface harden and improve the mechanical properties of materials by inducing compressive stresses. This study deals with its application t o the high performance aeronautical Ni-based superalloy Astroloy for turbine discs and its effect on low-cycle fatigue resistance. X-ray diffraction was used t o measure the surface and in-depth stress distributions. The prominent features of laser shock processing have been studied by an analytical approach t o the main physical phenomena occurring successively during the impact. This led t o an adequate treatment of conventionnal cylindrical low-cycle fatigue specimens. Fatigue tests were then conducted on Astroloy a t 550°C. These showed the beneficial effect of laser shock processing.

Nomenclature.

(x,y,z) : Cartesian coordinates z : laser pulse length

e

(r,e,z) ^: cylindrical coordinates z : plasma confinement length

0

t : time

- -

0- : stress tensor

dP : laser power density

R : radius of the irradiated zone

- -

E : strain tensor : Lam6 coefficients

EP : plastic strain tensor a : longitudinal wave velocity L : X-ray diffraction peak width 6 : transversal wave velocity

0- : yield strength c : Rayleigh wave velocity

R 0- = (l+h/Zp) o; ^: Hugoniot elastic limit

HEL

Introduction.

The superalloys used in the aeronautic industry often work a t the limit of their capacities. For example, in the case of Ni-based superalloys, such a s "Astroloy" studied by the French aero engine manufacturer SNECMA as a test material f o r turbine discs, one of these limits is its fatigue resistance. Fatigue cracks generally initiate a t defects such as inclusions or porosities : when the defect is located in the bulk material, i t is not considered as damaging a s a surface defect. So, the enhancement of the capabilities of an engine part can only be achieved if all risks of early fatigue failure ( by surface crack initiation ) a r e eliminated. This can be achieved by an adequate treatment of the surface of the engine part t o protect the material against these damaging defects.

Shot peening is the surface treatment currently used by SNECMA. It induces superficial residual stresses that counteract the applied stress during fatigue loading, thus reducing the harmfulness of the defects. However, although efficient, shot peening exhibits several

JOURNAL DE PHYSIQUE IV

disadvantages such a s possible superficial particle inclusions, problem of accessibility to certain areas and surface roughness modification. These problems can be overcome by using the laser shock treatment, i.e. "laser peening". Laser peening treats a larger surface volume than shot peening [I], and so offers protection to a greater depth against defects of larger sizes.

Principles of laser shock treatment.

The laser generated shock mainly results from a mechanical phenomenon based on a shock wave created by an explosion of laser-irradiated matter. When a material is exposed t o a laser beam with a sufficiently high fluence, a fine layer evaporates thus forming a plasma.

The pressure of the plasma ( up t o a few GPa 1 creates a shock wave which propagates within the material ( Fig.1 1. To induce this phenomenon, a short-pulsed laser Ismust be used in order t o reach a sufficiently high power density : a power density of 10 ~.m-', which is required typically, corresponds t o pulse lengths between 100 ps and 100 ns.

In order t o prevent the material surface from damage during the treatment, the target material is covered by a protective overlay such a s a black adhesive film or a metal foil.

Thus, the plasma is created by the overlay with detrimental effects such a s thermal effects, ablation or matter projection, affecting only the protective overlay. Only the mechanical shock wave is therefore transmitted t o the material.

The material i s also covered by a "confining" transparent overlay such a s water : the plasma then forms in the confined volume between the confining overlay and the surface, and i t s expansion in the direction perpendicular t o the surface is dramatically limited. The pressure of the plasma is thus magnified and its decay time is increased by a factor of 2 t o 3 compared t o the pulse length.

After the shock wave is created at the surface, i t propagates through the bulk material a s a uniaxial plane wave. The state of deformation a t the passage of this stress wave is uniaxial and, if the vertical stress rr is greater than the Hugoniot Elastic Limit rr

zz HEL

( o r yield strength in uniaxial deformation state 1, plastic flow occurs ( Fig.1 1. Since the shock wave is compressive, plastic deformation is negative, i.e. compressive, along the direction of propagation ( or z-axis ) while i t is positive, i.e. tensile, along the direction parallel t o the surface or perpendicular t o the direction of propagation ( that is the r- or 6-axis 1.

After the shock wave, the material returns t o equilibrium f o r which the stresses parallel t o the surface are compressive due t o the reaction of the surrounding undeformed matter.

Protective overlay

r Target material

Figure I : Principle of laser shock treatment and cylindrical coordinate system.

Materials and apparatus.

The aeronautical Ni-based superalloy used is P/M ( Powder Metallurgy 1 polycrystalline Astroloy ( NK17CDAT in the French standards 1 for turbine discs. Its composition is given in Table 1. Its yield strength cY a t room temperature is about 1050 MPa. Astroloy samples consist of mirror-polished 2 x 2 x 1 cm 3 parallelepipeds and 4.37 rnm-diameter cylindrical specimens for fatigue testing.

The laser shock experiments were conducted using the Nd-glass lasers of the L.U.L.I.

laboratory of the "Ecole Polytechnique" and the L.A.L.P. laboratory of the "Etablissement Technique Central de 1'Armement". The pulse length is s = 25 ns approximately. The f i r s t

e

laser delivers highly homogeneous laser beams of up t o 150 J but with a pulse repetition r a t e of 1 every 20 min, while the second one delivers only an energy of 80 J a t every pulse but the pulse repetition r a t e is up t o 1 every minute. The beams are focused using spherical or cylindrical lenses t o form 8 mm-diameter cylindrical spots or 16 x 4 mm2 elliptic spots.

Residual s t r e s s measurements a r e made by X-ray diffraction. The wavelength ho= 1.02 A

of t h e Mn Ka radiation is diffracted by t h e (311) planes of nickel a t a n angle 29

-

Table 1 : Chemical composition ( wt.% ) o f Astroloy.

16.6 1 4 . 8 5.0 3.8 3.5 0 . 0 4 0 . 0 1 5 0 . 0 2 b a l . In-depth r e s i d u a l stress measurements.

The in-depth profiles show t h a t residual stresses remain negative, i.e. compressive, up t o 1 mm in depth ⁽ Fig.2 1. Outside of t h e center of t h e impact, f o r example f o r r = 3 mm, t h e superficial residual s t r e s s reaches -500 t o -700 MPa f o r t h e most efficients shots [I].

This order of magnitude i s in agreement with results on different materials f o r which i t is generally claimed t h a t the maximum superficial residual s t r e s s t h a t can be obtained is a little more than half t h e yield strength.

In-depth profiles a t t h e very center of t h e impact ( r = 0 ) will be discussed later.

Whereas t h e residual s t r e s s becomes zero f o r z = 1 mm, t h e depth of material plastically affected by t h e shock i s g r e a t e r ( about 1.5 mm 1. This value can be determined in t w o ways. I t corresponds t o t h e value of z where t h e measured residual s t r e s s ( without any correction f o r relaxation ) reaches zero : beyond this depth, t h e metal is not plastically deformed, therefore at equilibrium a f t e r electro-polishing of t h e upper layer.

This can also be obtained f r o m t h e curves of the X-ray diffraction peak width determined during t h e measurements. The peak width is actually caracteristic of t h e plastic deformation r a t e within t h e material. Thus, t h e peak width decreases with depth a s does t h e plastic deformation, until it stabilizes a f t e r z = 1.5 mm a t t h e value of undeformed matter.

I I L - 2

Figure 2 : In-depth residual s t r e s s e s Figure 3 : Superficial residual stress p r o f i l e . p r o f i l e .

S u r f a c e r e s i d u a l stress measurements.

Superficial residual s t r e s s profiles along diameters of circular impacts always lead t o t h e same conclusions. According t o t h e cylindrical symmetry of t h e problem, t h e radial and orthoradial stresses, i.e. cr _rr and cQQ, a r e found t o be t h e two non z e r o eigenvalues of t h e s t r e s s tensor.

Except a t t h e very center of the impact ( r = 0 1, t h e compressive s t r e s s distribution i s almost uniform in t h e impact ( Fig.3 1. But in a small a r e a right in t h e center of t h e impact ( f o r r ⁵ 0.5 rnm 1, t h e residual s t r e s s level almost decreases t o zero. This now well-known phenomenon i s independant of t h e laser and occurs f o r a circular impact whatever t h e irradiation conditions. This s t r e s s drop is clearly a superficial phenomenon since in-depth measurements show t h a t the difference between the center ( r = 0 1 and t h e

JOURNAL DE PHYSIQUE IV

surrounding a r e a ( e.g. r = 3 rnrn 1 almost disappears below a depth of 0.2 mm. I t is attributed t o a s t a t e of plastic deformation created by release waves t h a t a c t s against t h e plastic deformation l e f t by the shock wave. However, although t h i s phenomenon has long been described empirically, t h e exact mechanism by which it is produced has only been understood since adequate mathematical models have been developed in t h e past few years [21.

Description o f t h e w a v e propagation d u r i n g a shock experiment.

Modeling of t h e wave propagation due t o a laser shock have been conducted using methods of resolution of Lamb's problem. These methods a r e applied t o a perfectly elastic material loaded by a uniform pressure within t h e circular impacted a r e a during a length r . They

0

consist of t h e resolution of t h e equation of motion ⁽ o r Lam6 - Clapeyron equation 1 using Fourier and Hankel [31 o r Laplace and Hankel [41 transforms. The exact analytical solution can be used t o evaluate t h e kind of plastic deformation created in t h e case of a n elasto -

plastic material, simply by considering t h a t this s o g t i o n is applied everywhere t o t h e material as a path of loading in t h e deformation ( o r r- 1 space [21. The chronology of t h e laser shock experiment i s developed below ( Fig.4 1.

From t = 0, i.e. when t h e plasma pressure i s applied t o t h e material, a plane longitudinal wave is created at t h e surface. This plane wave propagates in t h e vertical o r z-direction. I t induces plastic deformation of t h e f o r m :

rp 0

=

[ ^;

]

0 -2rp

However, a t t h e border of the' impact, release waves a r e created. They consist of two types of waves : a longitudinal o r P-wave which corresponds t o a n edge e f f e c t due t o t h e impossibility f o r a discontinuity t o exist between t h e inner and t h e outer sides of t h e impact, and a transversal o r S-wave which is created by t h e shear rrZ t h a t occurs at t h e side of t h e impact and t h a t induces t h e cliffs t h a t define t h e c r a t e r of t h e process zone.

After t h e plasma dissipates, all waves continue t o propagate. The plane wave keeps propagating towards t h e interior and looses energy by plastic deformation. Both t h e release P- and S-waves leave t h e impact boundary and propagate along torus-like wave surfaces, t h e f i r s t being at the longitudinal velocity a = 1 / ( ~ + 2 ~ ) / p ' = 6000 m.s-' and t h e second a t t h e transvers,al velocity = = 3300 m.s-'. The P-release wave interacts with t h e sample surface z = 0 t o c r e a t e a "head wave" but t h e amplitude of both i s too low t o have any permanent ( i.e. plastic 1 e f f e c t thereafter. On t h e other hand, t h e S-release wave strongly interacts with t h e surface t o form a Rayleigh wave. Since-lthe l a t t e r is a surface wave, i t s energy is radiated horizontally at a velocity c = 3100 m.s and i s thus confined along t h e

R

surface. The amplitude of t h e Rayleigh wave thus decreases very slowly.

When a t time t = R/cR, all t h e Rayleigh waves created from t h e boundary r = R reach t h e center-r = 0, they a l l superpose o r "focus" and t h e amplitude of t h e corresponding s t r a i n field E~~~ increases dramatically. The Rayleigh wave has t h e two components : a P- o r longitudinal type and a S- o r transversal type. Calculations show t h a t t h e P-component of t h e Rayleigh wave creates plastic deformation of t h e type :

=RAY -P rp 0

r = 0 rp

1

( 0 o -2rPJ

I t then tends t o annihilate t h e plastic deformation l e f t by t h e plane wave near r = 0. _The S-component leaves a shearing

O O E

-

r z

where rP is approximately a linear fonction of r f o r r < c z /2. This consequently induces

rz R o

a V-shape plastic deformation of t h e surface near t h e center. After t h e passage of t h e Rayleigh wave, t h e underlying m a t t e r r e a c t s against t h e deformation of t h e Rayleigh wave affected layer and tends t o f l a t t e n t h e V then opening its vertex and creating tensile stresses cr and creB f o r r < c T /2.

rr R o

So, t h e s t r e s s drop a t the center of t h e impact is due t o t h e e f f e c t s of both t h e P- and S-components of t h e Rayleigh waves which a r e emitted a t t h e same distance of t h e center

Emission of : Plane wave P

Release waves P and S

lnteraction of the P - release wave w i t h the surface

-

Interaction of the S - release wave w i t h t h e surface

-

1k I\ ^{RAY '} *+* t Focussing of the Rayleigh wave

w S

^{& -}

- 7 - rri?iV\n,

+ +

Figure 4 : Chronology o f Laser shock treatment.

Advantages d u e t o t h e u s e o f e l l i p t i c impacts.

The interest f r o m elliptic impacts is two-fold. First, a n elliptic shape is more suitable f o r treating elongated specimens like fatigue specimens. Second, i t drastically reduces the amplitude of t h e s t r e s s drop by limiting wave focussing. For a n elliptic impact, release therefore Rayleigh waves a r e emitted from t h e elliptic boundary. Then they can not meet altogether a t t h e same point a t t h e same time. In f a c t , t h e convergence of t h e Rayleigh waves occurs along t h e whole long a x i s x of t h e ellipse, with no focussing effect. The s t r e s s drop is consequently minimal and a f f e c t s t h e stress component s in t h e direction

YY

parallel t o t h e small axis y ( Fig.5 1. The small s t r e s s drop near t h e center of t h e impact, which also corresponds t o a diffraction peak width drop, can not be attributed t o a similar effect but t o a n a r e a where t h e laser intensity is smaller due t o diffraction e f f e c t s of t h e laser light during focussing.

'O:I

Impact

.5 o 5 1 ~ 3 5 x(mml

Figure 5 ^: Superficial residual s t r e s s profiLes f o r an elliptic impact. Hatched the s t r e s s drop Locations.

areas are

F a t i g u e tests.

Treating of t h e fatigue specimens is achieved using 4 elliptic impacts along each of 6 generating lines of t h e specimen. The whole surface of t h e cylindrical portion of t h e specimen i s then completely treated. The mean s t r e s s value on t h e surface a r e wZZ= -400 MPa and w = -700 MPa ⁽ Fig.6 1.

QQ

JOURNAL RE PHYSIQUE IV

Figure 6 : Sequence o f treatment for a fatigue specimen.

Fatigue t e s t s have been conducted by SNECMA a t the service temperature ( T = 550°C ), the applied s t r e s s cr oscillating from 0 t o the maximum stress cr a t a frequency of 0.5 Hz.

max

For cr = 1000 MPa, no failure i s observed ( Fig.7 ). For crmax= 1150 MPa, only one max

early superficial failure ( a t 6000 cycles 1 and one internal failure ( at 40000 cycles ) a r e observed. The other 3 specimens did not fail f o r over 200000 cycles. Every unfailed specimen was further tested a t cr = 1200 MPa where early failures always occurred. The

max

large increase of fatigue life shows a beneficial effect of laser shock treatment t h a t i s actually due t o a protection against superficial defects since those defects led t o failure a t cr = 1200 MPa. However, these results over a small number of tests need t o be confirmed

max

by further tests.

For cr = 1200 MPa ( whether directly o r a f t e r a f i r s t t e s t a t o;nax= 1000 or max

1150 MPa 1, early superficial failure always occurred a s f o r untreated specimens. The compressive residual stresses induced by laser shock treatment probably relax very soon by such a high applied stress.

Shot pccncd

r n4

Figure 7 : Fatigue t e s t s results.

Conclusion.

Adopting a rationnal approach t o understand the mechanisms involved during a laser-generated shock led t o designing and applying an homogeneous treatment t o fatigue specimens. The beneficial effect of laser peening has been shown. Further fatigue t e s t s t o determine residual stress relaxation f o r the various applied stresses a r e in program.

Acknowledgements.

This study has been supported by the "Direction des Recherches Etudes e t Techniques (DRET 1" under contract # 87.017.010.16 and 92.017.00.003, which i s gratefully acknowledged. The authors would also like t o thank Messrs. Lebrun and J i f o r technical assistance and helpful discussion.

Bibliographic.

[I1 FORGET P. e t al., Mat. C Manuf. Proc. 5[1990)501.

[21 FORGET P. and JEANDIN M., Proc. of "Colloque Contraintes RBsiduelles", 22-23 Sept. 1992, Arcueil, DRETETCA pub., Arcueil, France, (199219.

131 AKI K. and RICHARDS P.G., in "Quantitative Seismology", A. Cox ed., W.H. Freeman and Co, New York, NY, U.S.A., (1980)304.

[41 EASON G., J. Inst. Maths Applics 2(1966)299.