Hot Pressing

Chapter XI HOT PRESSING

There is known a variety of hot-pressing methods for production of structural components. Normally, semi-fabricated materials are processed by using such methods. Therefore we start with a description of fabrication of composite precursors. At the same time, such methods as rolling and explosive welding are discussed in this chapter, although they are not hot-pressing in a strict sense. Some composite materials normally obtained by hot pressing are also described in this chapter.

11.1. Fabrication of composite precursors Hot pressing like many other technological schemes is performed by using premanufactured semi-fabricated products. It is similar to well known prepregs in the fibre reinforced plastics technology. Preliminary bonding of fibres to a matrix appears to be convenient for further fabrication of structural components. A number of methods can be used to prepare composite precursors. Some of them, liquid infiltration being an example, is presented in other chapters.

11.1.1. Plasma sprayed tapes A jet of low temperature plasma had been used for deposition of metals and various coating on a solid surface long before the necessity of obtaining composite semi-fabricated products arose. Hence the suggestion to use such a process to deposit a layer of a matrix onto a set of fibres by Kreider [340] was a natural step. The process is usually conducted in the following way. Fibres are wound onto a cylindrical mandrel with a fixed pitch. Then drops of the molten matrix material are carried by a low temperature plasma jet to the mandrel surface, coating the fibres with a matrix layer. Then a composite layer with a weak matrix is cut along a generatrix of the cylinder and a precursor sheet or tape removed from the mandrel. Normally, the tape has one surface, the bottom one, being smooth, and the other one rough. A stack of monotapes is to be densified and sintered in a process of producing a particular structural component. Besides the factors important in assessing any fabrication method (output, energy consumption and so on), in this particular method we have obviously to be 475

476

Hot pressing

Ch. XI, w

interested in the following factors: properties of the matrix obtained, the influence of the process on the mechanical properties of the fibres, the quality of fibre packing, and the possibility of effective processing of the semi-fabricated material. We shall consider some of these factors, looking mainly at the process of making aluminiummatrix composites. Porosity of the plasma sprayed tape is inevitable; in some cases it can be useful the oxide layer at the fibre/matrix interface is being broken and an interface bond can then be formed as a result of the large displacement at the interface during densification of the matrix [548]. The porosity depends on spraying parameters, such as the electric power, the powder size and so on. The dependence of the porosity on the powder size has a minimum [297]. It should be noted that the porosity of a tape is larger than the initial porosity of a single tape due to roughness of one surface of the monotape mentioned above. During spraying, the chemical composition of a matrix can change due to oxidation if the process is conducted in air or in atmosphere of an impure inert gas. Measurements of oxygen content in aluminium-magnesium alloys after spraying conducted by Rycalin et al. [577] have shown that it changes from the initial value equal to 0.06-0.09% to 0.7-0.8% after spraying in air, to 0.27-0.31% after spraying with a local protection by argon gas, to about 0.25% after spraying in a box with an argon atmosphere, and to about 0.22% after spraying in the same box with a zirconium getter. The same authors studied also the formation of a structure of the aluminiummagnesium alloys after spraying and found temperatures between 560 and 580~ to be optimal for hot pressing. The alloys hot pressed in this temperature interval have the highest tensile strength, although such temperatures are too high for the boron fibres to preserve their strength (see Table 11.3), so the optimal temperature interval would appear to be lower. Decreasing the oxygen content in a matrix permits a reduction in hot pressing temperature (corresponding to a maximum value of the matrix tensile strength) by about 30-40~ Titanium matrix can be sprayed under carefully controlled atmosphere to prevent gettering oxygen or hydrogen from the environments. Kieschke et al. [312] conducted plasma spray deposition of titanium onto silicon carbide fibres using the following spraying conditions: Chamber pressure (Ar)." 150 mbar. Gun-substrate distance." 330 mm. Gun current." 750 A. Plasma gas flow rates." Ar: 11 1/min, He: 25 l/min, H2:81 1/min. Powder size range." 45-63 ~tm.

Processing under these conditions lead to absorption of hydrogen by titanium droplets during their flight life. This causes a specific behaviour of the diffusion barrier mainly composed of yttrium oxide.

Ch. XI, w

477

Fabrication of composite precursors

11.1.2. Powder metallurgy prepregs Although powder metallurgy techniques shall be discussed in the next chapter, we describe here some processes based on using either powder or some other kind of the precursor as a source for the matrix material for making prepregs to be used in hotpressing fabrication routes. These routes are normally used for making ceramic matrix composites.

Slurry impregnation This is perhaps an oldest process for producing a prepreg by the powder metallurgy route. It was described in detail by Phillips [533]. A general scheme of making a tape prepreg from a fibre tow is shown in fig. 11.1. Either a row of single filament or single filament or a tow of the fibres is passed through the tank containing slurry of powdered matrix material suspended in an organic solvent with an organic binder. Slurry impregnates the tow or row of fibres, powder adheres to the fibre and the prepreg is then wound onto a take-up drum and dried. In the case of fibres supplied as a slightly twisted tow, the tow is to be converted in a tape by moving it, for example, through a device which consists of a series of rollers and nozzles producing air jets to fan the fibre out. To stimulate impregnation, the slurry may be agitated by air flow from the bottom of the tank or by some other method. The fibre volume fraction, vf, in the prepreg is controlled by the content of the powder and binder in the slurry, the values of vf between 20 and 60% can be achieved.

Slip casting A powder matrix precursor can be obtained as a layer containing fibres by a method of tape casting. In a particular disclosure [565], to decrease porosity of a silicon nitride matrix obtained by reaction bonding, the matrix precursor contains silicon carbide powder in addition to silicon powder. A tape of the casting slip is

FIBRE

FIBRE

SPOOL

!

TO W

TO W ~

TAPE C O N V E R S I O N

\

i

TANK

T A K E - UP D R A M

I

++':+:+:+:+;+:+:+:+:+;+:':':+;+; :+

Fig. 11.1. A scheme of producing a tape pre-impregnated with slurry.

Hot pressing

478

Ch. XI, w

prepared by mixing two materials mentioned and corresponding sintering aids with a solvent, a surfactant to help to disperse the powder, a polymer binder, and associated plasticizer. The slip is cast over fibres (SCS-6 SiC) wound onto a drum. An adjustable gate is used to control the tape thickness. Before nitriding to convert silicon into silicon nitride, which is done in alumina tube furnace using a flowing mixture of 90% N2/10% H2 at temperatures up to 1370~ the prepregs are taken out from the drum, stacked in a graphite die and hot pressed to remove polymer binder and partially sinter silicon with the matrix. 11.1.3. Other processes

Ion-plating The ion-plating process can be conducted at low temperatures and that is its advantage when it is used to make a precursor from two or more reactive components. On the other hand, it is a slow and energy consuming method to be considered as a real candidate for a large scale production. A schematic drawing of the apparatus suitable to execute ion plating (due to Ohsaki et al. [513]) is presented in fig. 11.2 without showing a mechanism for making the continuous precursor tapes.

Vapour condensation Condensation of a vapour of the matrix material on the relatively cold fibre can have an advantage of a low temperature process. One can expect to exclude chemical reaction during the deposition. Ar~

VImuum out

l fibre

--

High (') AI in

voltage

Fig. 11.2. Schematic draw of the ion-plating apparatus according to Ohsaki et al. [513]

Ch. XI, w

Fabrication of composite precursors

479

The process was carried out for coating thick SiC-fibre with a titanium alloy, a titanium alloy strengthened by disperse yttria particles, intermetallic compounds Ti3A1 and TiA1, and an aluminium alloy (A1-4.3Cr-0.3Fe) [682]. Metal matrices were evaporated by an electron beam by using electron beam accelerated by 10 kW to heat a double evaporation source. The deposition rate occurred to be approximately 5 to 10 ~tm min -1. Coating with titanium and aluminium alloys were performed from a single evaporation source. Dispersion of Y203 in a titanium alloy was produced by evaporating yttrium from the second evaporation source that combines with the oxygen in the solid solution. Coating with titanium aluminides was organized by evaporating Ti-6A1-4V-alloy from one source and pure aluminium from another one. Quench rates for vapour deposition were estimated to be 1013 K/s, compared with 104 to 108 K/s for liquid-quenching methods. Therefore, the potential is expected with vapour quenching for producing matrix alloy with extended solid solubilities, or very fine dispersions of reinforcing phases that can improve matrix properties.

Magnetron sputtering Magnetron sputtering described above (Section 10.2.5) as a method of fibre coating is also used by Dudek et al. [138] as a method of producing composite precursor. The advantages of this method are homogeneity of coating thickness which results in homogeneous fibre distribution in the composite structure, a possibility of a strict control of chemical composition of the coating, and small grain size of the matrix material with corresponding benefits for the composite properties (Section 5.2.5). In particular, this method was used to produce titanium aluminide matrix composites.

Foil precursor Using a foil as matrix precursor has an advantage of retaining during a composite fabrication process of a microstructure of the matrix material normally yielding such characteristics of plasticity and fracture toughness that are wanted to attain sufficiently high values of admissible fibre volume fraction and, correspondingly, high strength and stiffness values of the composite (see Section 5.2.5). To join boron fibres and aluminium alloy matrix, Mileiko and Gryaznov [429] used simple techniques. One of them consists of winding the fibre on a drum similar to that used for the matrix plasma-spraying, then covering the fibre with the foil, and finally pressing periodically the foil into the fibre array (fig. 11.3). The length of the stamp and the pitch of stamping are chosen so as to ensure a minimum fibre damage and not to affect the composite strength. To arrange fibres unidirectionally and uniformly and retain the arrangement during consolidation is possible to punch a ductile wire into a fibre array wound onto a drum or to crossweave a fibre layer with a wire. A problem arises when an alloy which is difficult to obtain in a foil form, is chosen as a matrix. In such a case, a technique called "powder cloth" can be used [59, 345]. The powder of a necessary composition is mixed with an organic binder and a

480

Hot pressing

Ch. XI, w

Drum Fig. 11.3. Joining fibre and matrix foil.

wetting agent and then the dough-like mixture is rolled into a foil-like material. During the processing the wetting agent and binder are evaporated, although it is difficult to ensure that a complete removal of the excess material has taken place. Dobbs et al. [131] reported hot isostatic consolidation of powder to produce titanium aluminide, Ti2AINb, foil which needs just a small degree of cold rolling to achieve a final thickness of about 0.125 mm.

11.2. Processing parameters Choosing hot pressing parameters, one should consider the following processes: (1) removing long wave roughness of either the matrix surfaces or both the matrix and fibre surfaces; (2) removing short wave roughness on surfaces of interest; (3) removing voids at the interfaces; (4) development of matrix material properties; (5) formation of either physical or chemical bonding at the interfaces; (6) growth of the reaction zone at the fibre/matrix interface. The first two processes yield the mechanical consolidation or densification of a fibre/ matrix mixture. The third process is actually sintering that was considered in the previous chapter (Section 11.5). Sintering is also involved in the formation of the microstructure of a plasma-sprayed matrix and the development of matrix properties. The fifth and sixth processes are normally accompanied by the former ones and provide, at first, occurrence of the interface bonding and its increase, then either degradation of the bond or a negative contribution to the composite properties (see Section 11.3.1). Thus, a choice of the process parameters is to be performed taking into account the kinetics of the processes mentioned. At the same time, there can be some limitations to the parameters to be chosen. First, hot pressing can cause an excessive fibre breaking which yields a decrease in

Ch. XI, w

Processing parameters

481

composite strength. Some purely technical or facility limitations can also be imposed.

11.2.1. Densification An instructive experiment was carried out by Shioiri who studied bond formation between two titanium surfaces (see [419]) by measuring the intensity of an ultrasonic pulse reflected from the interface. There were revealed two characteristic rates of decreasing the intensity (fig. 11.4). The first characteristic rate relates to disappearance of long wave roughness of the surfaces and the other one relates to short wave roughness. If the load is removed at t < t t and the specimen kept at high enough temperature, we will never observe the strength of the interface equal to the strength of the bulk material. But if the same is done at t > t I, then the strength of the interface inevitably reaches the strength of the bulk material. Plastic smoothing of the surface is usually modelled [294] by deformation of a series of wedges by rigid smooth surface (as shown schematically in fig. 11.5a). A first study of this problem was undertaken by Ushizki [668] (see also [419]) who obtained asymptotic expressions for the pressure and contact area in the framework of rigid/ideally-plastic formulation of the physical problem. When considering a case of the densification of a plasma-sprayed monotape, Elzey and Wadley [154] performed a study of the problem in a way more suitable for numerical procedures. They modelled an interface between two monotapes as a layer of height h between a smooth plane and a surface composed of a stochastic set of the asperities (fig. 11.5b). The probability density functions describing distributions of the asperity heights and radii are q~h(h) and q~r(r), respectively. Assuming the mutual independence of these functions, the probability density of an asperity of height h and radius r is ~c(h,F) = q)h(h)q~r(F).

(11.1)

Fig. 11.4. Dependence of intensity of a reflected pulse on pressing time for diffusion bonding of two titanium surfaces in Shioiri's experiment (a scheme).

Hot press&g

482 ////////////.

Ch. XI, w

"////////////

~///////////////////~

~////////////////////A

V Fig. 11.5. Schematic of the plastic smoothing during hot pressing. (a) A rigid surface acting on a plastic wedge (~). A real picture of large deformation (fl) is replaced by a simpler scheme (7) to be analyzed. (b) Densification during plastic deformation of a stochastic set of the asperities. (c) A matrix material yielding through a lattice of rigid fibres.

In particular calculations by Elzey and Wadley [154], the normal distribution for the heights and the exponential one for the radii were adopted with justifications found in experimental observations. Hence, q>h(h) - x / ~

exp - ~

~

(11.2)

and q~r(r) -- 2exp ( - 2 r )

(11.3)

where h and h are the mean height and the standard deviation of the heights, respectively, and 2 is a parameter of the radii distribution. The probability density distribution of forces required to cause plastic deformation of asperities will be ~f(h, r) - ~c(h,

r)Fc(h, r,z,~)

where Fc is the contact force required to deform a single asperity.

(11.4)

Processing parameters

Ch. XI, w

483

Integrating eq. (11.4) over all asperity heights and radii encountered in compacting the layer from z0 to z, the total force required to continue the smoothing will be F ( z , ~ ) ---

f Z0jr0~176 ~f(h,r)

(11.5)

drdh

/zZO/o

~c(h,r)Fc(h,r,z,~.)drdh

(11.6)

and the corresponding pressure is (11.7)

q(z,k) = n . F(z,k)

where n is the number of asperities per unit area. As in the Ushizki's solution, the theory of perfect plasticity is applied to find the force acting on a single asperity to cause plastic yielding, that is

(ll.8)

Fc = acac = ]7oyac ~ 2?+r(h - z ) f l o ' y

where ~rc is the contact stress, cry is the yield stress,/3 is a constant derived from a plasticity theory solution,/3 ~ 3, ac is the contact area, ac ~ 2 g r ( h - r). Now eq. (11.7) gives the integral equation for the pressure to cause plastic smoothing of the surface as a function of displacement z. However, solid state bonding is normally conducted at elevated temperature when creep of metals is essential. If one uses a power creep law, eq. (10.36), then it should be noted that at relatively low temperatures the value of O"m is high and the value of the exponent m is large. (Examples of temperature dependencies of O'm and m for some alloys which can be used as matrix materials, are shown in fig. 11.6.) So at low temperatures the rigid-plastic analysis can be considered as a good approximation, if the characteristic stress O'm for a characteristic time t oc qm1 is taken as the yield stress ~rv. But at high temperatures, which is the usual case, we have 1 < m < 3 and a creep problem for large deformations has to be solved. Actually, to analyze a temperature dependence of the consolidation process determined by the matrix creep, well known creep-rate/temperature/stress dependencies can be used. Dorn equation [184] is appropriate: -- B

exp(-Qc/kT)

- A

exp(-Qc/kT)

(11.9)

where E and p are the Young's and shear moduli of the creeping material, respectively, A and B are corresponding constants, and Qc is the activation energy for the creep process. In addition to fig. 11.6, we present values of parameters involved in eq. (11.9) for some titanium based alloys (Table 11.1).

Hot pressing

484

Ch. XI, w

10.0

4

E 3

o ~ 2024-T6 - - o ~ AI-6%Mg

a m

o

7.5 5.0

2

E

2.5

1 450 5

b

1

500 ',,

.

,

.

,

0.0 550 9 1000

4

100

E 3 ....

-.......

2 -

..

rJ-3A~-~. Ti-4AI-3Mo- l V I

600

,

I

,

I

800 T / o c 1000

Fig. 11.6. Temperature dependencies of the creep parameters of two aluminium and two titanium alloys. The Creep law is ~ = rh,(a/t~,,)", r/,, = 10-4s -1. After Rabotnov and Mileiko [559], characteristics of A16Mg alloy at high temperatures were obtained by S.V. Trifonov.

Smoothing the surface carrying a stochastic set of the asperitites (Elzey and Wadley's model, eq. (11.1)-(11.7)) was used to evaluate a kinetics of densification due to removing rough waviness of the surface as a result of creep [154]. The stress, ~, in a creeping asperity under a contact stress, ~ c - Fc/ac where ac - ~zx2, is assumed to be

Ch. XI, w 1.2

Processing

485

parameters

TABLE 11.1

The creep parameters of titanium based alloys. After compilation by Elzey and Wadley [154]. Alloy composition

Young's Modulus, E(T) GPa

Ti-6A1-4V

T < 500~ 9 115-0.056. T T > 500~ 9 172.4-0.16- T T _< 500~ 9 100-0.04. T T > 500~ 9 140-0.12. T 172-0.03 9 T

Ti3AI + Nb (Ti-24Al-11Nb) TiA1

A h -1

m

Qc kJ/mol

8.4.1024

4.0

280

6.0.1017

2.5

285

7.6 9 1022

4.0

300

(11.10)

O " - C10"c

where Cl is a constant. The displacement rate, ~, must scale with strain rate ~ and with the radius of contact, x, which yields ~ - c2~x. Therefore, "Z - - C 2 YlX

( C l tT-------~c

(11.11)

.

\ O'm //

The constants e l and r are determined by satisfying the condition of reducing eq. (11.11) to the perfectly plastic solution at m ~ c~ and to the elastic Hertz solution when m - 1, ~ - e and (r/q0)-ltrm - E, where E is the Young's modulus and r/0 is the time unit. This yields Cl = fl and c2 = 1.36rcfl, where fl was already determined. Hence, ~_ --

O-c

1.36~zfll-mxrl

m (11.12)

.

Taking into account that ac - ~ x 2 on an asperity into eq. (11.7) yields

~

27tr(h -z)

and substituting the force acting

Z' ' n mexp[ m (ll.13)

}-m •

J/O"

r 1-1/m exp(--2r) dr

.

Here eqs. (11.2) and (11.3) have been used and

--

1.36.21/2-mrl(Tzfl)l-1/m.

Equation (11.13) is to be solved numerically. When considering the hot-pressing of a fibrous composite made of a stack of fibres and foil, we see that the matrix, at least at the initial stages of the process, flows through a lattice of the rigid fibres (cf. fig. l l.5c). At the matrix/fibre

Hot pressing

486

Ch. XI, w

b o u n d a r y there exist both normal and tangential components of stresses. Such a process can be modelled as flowing of a rigid-plastic matrix in a convergent channel and a result of one such solution can be found in [658]; namely, the dependence of pressure q on the ratio of df to distance L between fibres is q c~ am ln(1 - d f / L ) -1.

(11.14)

The value of the logarithmic term changes with a factor 2 when the fibre volume fraction changes from 0.2 to 0.5. Goetz et al. [197] derived a closed form expression to estimate the consolidation time of an assemblage depicted schematically in fig. 11.7 when the matrix creeps according to eq. (11.9). Their result can be written as

l)f (df -k- r

{ cx Lm_l

d~

1 exp(Qc/kT)

A---N q

"

(11 15) "

Same authors performed a numerical simulation of the process accounting for the effect of shear friction coefficient, M, at the fibre/matrix interface. They found that time t" to complete 97% of the consolidation is approximately equal to the time to complete the final 3% of the consolidation which is in qualitative agreement with Shiori's observations of the behaviour of plane models (fig. 11.4). It occurred to be convenient to introduce an average effective flow stress # dependent on the effective strain rate ~, which was taken as the quotient of the average effective strain, ~, for a particular geometry, and time t". The simulation results show that for a wide range of the hipping parameters ( Ti3A1 + Nb-matrix, SCS-6 SiC-fibre, q = 10 - 500 MPa, T = 760 - 980~ M -- 0.1 - 1), the ratio, p, of isostatic pressure, q, to the average effective flow stress was a function only of M, so

d

/

Q

Q

,,,

Fig. 11.7. Schematic of the consolidation process of a fibre/foil assemblage.

Processing parameters

Ch. XI, w

487

p = p(M) = q/~. It occurred that p = 1.27 for M = 0.1 and p = 1.61 for M = 1. In order to evaluate the soak time tt (cf. fig. 11.4) necessary to form the complete physical contact, we need to obtain a combination of temperature T and pressure q which provides the result without unwanted effects like fibre degradation (dissolving, breaking) or formation of too thick an interface layer containing products of chemical interaction between the fibre and the matrix. If an optimum combination of fabrication parameters (To,qo, to) is known for a composite with one matrix material, then a first approximation to an optimal set of parameters for a composite with other matrix material can be obtained assuming creep of the matrix being a determining process. Note that for the power creep law, the solution of a creep problem is characterized by an interesting property [558]. Namely, if all external loads increase proportionally to one factor, say 2 and for one value of 2, say 20, a solution of a problem has been obtained, and if we know the stress field r 1j and the dist~lacement field u!1j~ (X), then for an arbitrary value of 2 the stress field will be (2/20)~r,, and the displacement field ()0 m (.), 0 9 ij . 9 be ((2/20) u,, ). Therefore, if creep parameters for two matrix materials O'm ( T ) , a~)(T), m0({), ml(T), and rt0--rtl = rt are known, and it is possible to assume m0 = ml = m in a temperature interval of interest, then temperature T1 is to be chosen such that Cr(m ~ - a(m1). If the possibility of changing the temperature is restricted by the chemical interaction, then the following equation has to be satisfied:

qoa~ ) -

(,0) "m ~

.

(11.16)

If m0 ~: m l then a consequence of the well known Calladine-Drucker's theorem in the mathematical theory of creep [558] should be used. The only generalized force here is q, so choosing the parameters, inequality ql

< 1 \q0j

(11.17)

-

has to be taken into account if ml > m0. This inequality can be especially useful if optimum sets of hot pressing parameters for a number of matrix materials with various values of the exponent m are known.

11.2.2. Sintering stage A final stage of the densification can be influenced by the process of removing small voids at the interfaces (matrix/matrix and fibre/matrix) according to stress induced diffusion as discussed in Section 11.5. An evaluation of the corresponding contribution can be done again in the framework of the geometry depicted in fig. 11.5b and taking the Nabarro-Herring

Hot pressing

488

Ch. XI, w

equation, eq. (10.31), as a base for a description of the diffusion flow [ 154]. Assuming the total flux of matter to be the sum of grain boundary and volume fluxes, that is ~2 ~r O( ( 6 D b nt-

2pDv)~--~c

(11.18)

and the void has a special geometry, one obtains r3

(z - -~ 9(z)~.

(11.19)

Here Db and Dv are the grain boundary (along the interface) and volume diffusion coefficients, respectively, 6 and p are the boundary thickness and the radius of curvature of the neck around a sintering front, r is a contacting sphere radius in the geometry assumed, 9(z) is a function determined by the geometry. Combining eq. (11.18) and (11.19) gives the dependence between the force applied to an asperity and the displacement rate:

r2 kT Fc -- ~ 9(z)ac --~ 6Db -~ 2pDv

(11.20)

which should be also substituted into eq. (11.7) to obtain an integral equation for z. In the case of a plasma-sprayed matrix, densification and sintering at the monotape/monotape interface is accompanied by densification and sintering of the matrix. A study of sintering mechanisms considered in Section 10.5 was also conducted by Elzey and Wadley [154].

11.2.3. Development of matrix properties Formation of the matrix microstructure during hot pressing of a plasma-sprayed matrix precursor is determined by densification and sintering processes discussed above. At the same time, melting and subsequent rapid cooling of a matrix alloy during plasma spraying can essentially change a structure of the future matrix, when compared with the structure of a nominal alloy (which is usually taken from a series of wrought alloys). A possible deterioration of mechanical properties of the alloy can lead to a corresponding deviation of composite properties from those predicted assuming nominal alloy characteristics. That is why the investigation of alloy properties after its spraying and various treatments is of importance. Alipova et al. [11] carried out such a study of the aluminium-zinc-magnesium alloy which could be strengthened by heat-treatment. They directed the study towards fabrication of boron-aluminium composites by hot isostatic pressing. The specimens were obtained by depositing the matrix layer-by-layer in air, the density of the sprayed material being from about 10 to 15% less than that of the nominal alloy. The oxygen content in the original alloy was 0.01%, in sprayed specimens about 0.11% by weight. The strength of these specimens was about 50 MPa. Then the specimens were placed into a vacuum container and isostatically hot pressed in a high

Processing parameters

Ch. XI, w

350

.....

,-

, , , , ....

489

, ~ , , , _

o

5

300

5 b~ 250 200

150

I000

tO0

200

300

~- /

400

m4,n

Fig. 11.8. Dependence of the strength of a plasma sprayed A1-Zn-Mg-matrix alloy on equivalent time of hot pressing. Experimental data by Alipova et al. [11].

t e m p e r a t u r e autoclave. The a u t h o r s gave a table of the pressing p a r a m e t e r s , n a m e l y t e m p e r a t u r e T, pressure q a n d time t, as well as strength a m of final materials. A n analysis of the e x p e r i m e n t a l d a t a was u n d e r t a k e n in [419] a s s u m i n g the existence of the equivalent time expressed as

z-

t

q

exp

-

(11.21)

where n, a . , a n d T. are c o n s t a n t , the latter being d e t e r m i n e d by the activation energy of a process to p r e d e t e r m i n e the result. A set of the c o n s t a n t s can be o b t a i n e d which supplies a best c o r r e l a t i o n between the m a t r i x strength, a m, a n d the equivalent time, z. The c o n s t a n t s are f o u n d to be n = 2, a. = 40 M P a , a n d T. = 1000K. The e x p e r i m e n t a l points on the am-Z plane are s h o w n in fig. 11.81. The e x p e r i m e n t a l d e p e n d e n c e calls for either p o w e r or e x p o n e n t i a l a p p r o x i m a tion. T h e p o w e r a p p r o x i m a t i o n s h o w n in the figure, t h a t is , a m -- a 0

(~00)r

,

fits the e x p e r i m e n t a l d a t a if a0 = 153.9 M P a , z0 = 1 min, a n d r = 0.1254.

1 A quantitative error in calculating the equivalent time in fig. 37 of [419] exists.

(11.22)

490

Ch. XI, w

Hot pressing

The ultimate elongation can be made higher by annealing in vacuum, and a standard heat treatment after annealing leads to a strength increase and an ultimate strain decrease. Similar results are presented in fig. 11.9. One can see that there are optimal sets of the processing parameters for the hot-pressing of a plasma-sprayed matrix depending on particular requirements to the matrix. F r o m a very general consideration supported by the results presented in the previous sections, we could write down eq. (11.21) in the form: -- to (~m'm) q

(11.23)

mexp(-Qc/kT).

However, as we have just seen, a number of the processes are really acting and we do not know a priori which one dominates. Therefore, we could change to a consideration of the corresponding rates determined by different activation energies and pre-exponential terms, then to sum the rates (under some assumptions) and finally to derive the total time. Without doing so and going to point out an empirical nature of eq. (11.21), we write it in terms of purely empirical constants, although for a more general discussion, it is perhaps more convenient to use eq. (11.23). Note that if temperature is changing during the hot pressing procedure and the pressure is constant, we can write a cumulative equivalent time as 7 cx

0~0t e x p ( - Q / k T ( t ) )

(11.24)

dt

where Q is an appropriate activation energy.

l

350, 300 \ *b~ 250

00000

Pre.sed i.n air

~

I

200 i -

150

~

-~

i

-

~

-

O"

\

Yield s~ress Ultimate s~rength

100

527

5

J

1

I

500 T /

I

525

550

~

Fig. 11.9. The yield and ultimate strength of the AI-6Mg alloy after plasma spraying and hot pressing versus temperature of pressing. Experimental data after Gukassyan et al. [211].

Processing parameters

Ch. XI, w

491

TABLE 11.2 Interfacial shear strength in SiC/Ti monofilament composites Matrix

Filament/ coating

Temperature/time/pressure ~

Shear strength MPa

Source

Ti-6A1-4V Ti-6A1-4V Ti-6A1-4V Ti-6A1-4V Ti-6A1-4V Ti-6A1-4V Ti-25Al-10Nb-3V Ti-25Al-10Nb-3V Ti-25Al-10Nb-3V Ti-15V-3A1-3Cr-3Sn Ti-15V-3A1-3(~r-3Sn Ti-15V-3A1-3(3r-3Sn Ti-15V-3A1-3(3r-3Sn Ti-15V-3A1-3('x-3Sn

SiC/C/TiB2 SiC/C SiC/C/TiB2 SiC/C SiC/C SiC/C SiC/C SiC/C SiC/C SiC/C SiC/C SiC/C SiC/C SiC/C

925-0.5-30 925-0.5-30

350 -+-35 192 + 20 90-120 156 165 167 111 93 99 124 167 154 119 131

[469] [469] [693] [718] [718] [718] [718] [718] [718] [718] [718] [718] [718] [718]

As received Annealing 800-50-0 Annealing 800-100-0 As received Annealing 800-50-0 Annealing 800-100-0 As received Annealing 500-50-0 Annealing 500-100-0 Annealing 800-12-0 Annealing 800-50-0

Test methods: fibre fragmentation for the first two lines, push-out for the remaining results.

11.2.4. Bonding The interface strength in a composite is determined by all the processes listed above. The importance of reactions taking place at the interface was already pointed out above, in Section 10.3.1. We refer also to diffusion bonding experiments in SiC/Ti system (Table 11.2).

11.2.5. Fibre damage During the densification of a plasma-sprayed tape, fibre can be bent if it is located between asperities. This can lead to a random breaking of the fibre. Obviously, with the processing temperature increasing and pressure decreasing the risk to break fibre decreases. Actually, such a dependence was observed in hot pressing experiments with SiC/Ti3A1 + Nb [209] and SiCw/A1 composites [254]. In the former case, the direct measurements of the density of the fibre breaks were conducted; in the latter case, the qualitative observation were supported by measuring tensile properties which increased with increasing hot pressing temperature, since aspect ratio of whiskers and the density of composites were improved with increasing fraction of liquid phase. A model to incorporate statistics of the fibre strength and that of a geometry of the structure of a plasma-sprayed matrix [155] as well as an attempt to evaluate a dependence of the composite strength on the fibre damage during the consolidation are also known [144]. Clearly, a severe fibre damage can occur in the case of a nonunidirectional fibre distribution. Also, the problem becomes very important when non-plane elements are densified (see below, Section 11.3.3).

Hot pressing

492

Ch. XI, w

11.2.6. Concluding remarks A knowledge of the parameters determining kinetics of the processes in hotpressing fabrication routes allows to design the fabrication process sufficiently quickly and choose a first approximation to necessary fabrication parameters a priori. Experiments to correct the parameters are necessary, especially if we remember that wanted microstructures of a composite can be different for different loading patterns (see Part II). Still, research efforts are directed at lowing a necessary experimental work. An example is given by Nicolaou et al. [492] suggested to plot time/temperature contours corresponding to full consolidation at various pressures and those corresponding to a given thicknesses of the reaction zone (fig. 11.10). The former family of the contours is given by eq. (11.15), the latter one is a graphical presentation of eq. (10.5). Lines for largest admissible values of the processing time (tmax) and temperature (Tma• are also plotted. There can be plotted regime contours corresponding to a fibre damage that should be avoided. Suppose we are looking for a composite with the reaction thickness zone h2. Then sets of the parameters such as q2, t2, T2 and q3,tl, T1 are possible physically and permissible technically, set q2, t2, T2 is possible physically but prohibited technically. It should be noted that a realization of the process with parameters yielding a processing time longer than it is necessary for full consolidation is obviously

hS

h~

h~

7"

>h2 >h~

Tmax

qt ~--

t2

tl

t, na=

q2

t

Fig. 11.10. A schematic of the processing parameters map suggested by Nicolaou et al. [492]. The lines corresponding to various pressures q present combinations of time, t, and temperature, T, which provide the full consolidation of foil/fibre/foil assemblage. The lines corresponding various values of h give combination of processing parameters which provide given values of the reaction zone thickness, h.

Ch. XI, w

Processing parameters

493

permissible provided it does not yield the reaction zone thickness larger than has been prescribed. With regard to consolidation of a plasma-sprayed semi-fabricate, an analysis by Elzey and Wadley (see above) yields [154] density/pressure maps constructed for various processing temperatures (fig. 11.1 l a) as well as those plotted in density/ temperature coordinates for various pressures (fig. 11.1 l b). These maps are to be calculated accounting for densification mechanisms considered above. It can be also advantageous to choose technological parameters to produce a new material in a composite system based on both an approximate evaluation of the creep process in a composite during the consolidation time (eqs. (11.16) and (11.17)) and a known set of the parameters for a composite of the same system. An example is presented in [419]. The temperature dependencies of creep parameters for aluminium alloys D 16 (analogous to 2024) and A1-6Mg presented in fig. 11.6. F o r a boron-aluminium composite with the D16-matrix the set of the parameters of hot pressing close to an optimal one is

I

~,o~,...------~,~ - _

,~r ~,~

" I

~

I# / /

~'~ .~,

T = const

Ini~at density I

log

"-

I

I

.~9"//

(ql(~y)

/g

/;

q = const

l

~ ~ . ~

Initial density

T/T m

Fig. 11.11. A schematics of the calculated densification maps due to Elzey and Wadley [154] showinghow the density (l-p) grows during the process (p is the porosity). Plastic yielding is supposed to occur instantaneously upon pressure application; contours corresponding to non-linear creep are calculated for various times t; diffusion flow (linear creep) contributes to the densification when the total porosity is sufficiently small.

Hot pressing

494

485~

25MPa-

Ch. XI, w

1.5h

In an experiment [419], a set of optimum parameters for the A1-6Mg-matrix (to maximize the tensile strength) was found as 530 ~ - 30 MPa - 1.5 h These two combinations of the parameters nearly satisfy eq. (11.16). It should be noted that a deviation from these parameters can give composites with lower tensile strength values. An example is the dependence of the tensile strength of boronaluminium composites with the A1-6Mg-matrix (vf - 0.28 + 0.02) on temperature of hot pressing at constant pressure q = 30 MPa and time t = 1.5 h. We have the following: Temperature of hot pressing Average tensile strength

~ MPa

520 396

530 496

550 402

It occurred that the same hot-pressing parameters are nearly optimum for compressive strength. Figure 11.12 illustrates the dependence of the critical compressive stress, cr,, of the same composite on temperature of hot pressing. It can be seen that the difference in critical compressive stresses for the composites fabricated at temperatures 530 and 550~ is most essential at small ratios of thickness h to length l of specimens. It is supposed to be a consequence of decreasing either the yield stress a m of the matrix or the shear strength of the interface (see Section 8.4). Finally, we should remember that using the equivalent time given by eq. (11.21) or eq. (11.23) is a helpful method to adjust fabrication parameters to a particular

!

'

'

!

I

'

'

!

"

0

~ o~ oo

%2

oO~

,,~176 %o

9

T = 520~

9

T = 550~

o

T = 530~

vO i o

0o

olO 0

o

~

'

'~ 60

I/h

8'o

'

100

Fig. 11.12. Critical stresses at compression of boron/aluminium composite with AI-6Mg-matrix (vf =0.28 +0.02 obtained at various temperatures of hot pressing; q - - 3 0 M P a , t - 1.5 h). After Mileiko [419].

Ch. XI, w

495

Techniques

technological conditions. This is especially useful if one has to follow a fabrication regime with changing temperature or pressure as can be observed in a case of making sufficiently large components in a large die (see below, Sections 11.3.2 and 11.3.3).

11.3. Techniques The techniques of hot pressing used for producing metal and ceramic based composites are simple and sufficiently versatile. They provide the possibilities to fabricate plates, as well as structural elements such as tubes, open profiles, shells. 11.3.1. Plates

To obtain plates, it is possible to use a rigid die (see fig. 11.13). The temperature should be about 500~ for aluminium, about 800~ for titanium and titanium based materials, and about 1200~ for nickel based matrices including nickel aluminide Ni3A1. A pressure required is normally between 20 and 100 MPa. So corresponding materials have to be chosen for the die with either a built-in or external heater. A chamber should provide the necessary displacement of a die, and a vacuum or protective atmosphere. Using plasma sprayed precursors, it is possible to conduct hot pressing of aluminium matrix composites in air. Large relative displacements of the surfaces coated with oxide layers lead to fracturing of such layers and provide conditions for FLeXiBLe eLeMeNT

CHAMBER

(VACUUM OR PROTeCTIVe CHAMBER 301NT .,

THERMOCOUPLe

ATMOSPHERE PLATe WITH HeATeR

BLANK PLATe WITH HeATeR

Fig. 11.13. A sketch of a chamber for hot pressing a composite.

496

Ch. XI, w

Hot pressing

bonding. Prewo [548] has shown that the process parameters in this case can be chosen such as to decrease pressing time down to 10 minutes. Using dynamic hot pressing studied in detail by Karpinos et al. [297] also leads to a decrease in the total process time. 11.3.2. Tubes

For consolidation of structural elements like shells and tubes, 'soft' die should be used [419] to follow changes in the curvature of the blank surface during consolidation process. But simple liquid autoclaves are of no use because pressing temperatures are too high for metal- and ceramic-matrix composites. Hence, they should be produced in either more expensive high pressure gas isostat or a kind of quasi-isostat . Weisinger [696] was certainly the first to suggest applying gaseous pressure to densify and sinter boron-aluminium tubes according to the scheme shown in fig. 11.14a. Precursor boron-aluminium tape is rolled on a thin-walled steel mandrel. Then it is inserted into a thick-walled outer steel tube. The steel tubes are welded together at the ends to give a vacuum-tight assembly. Evacuation can be done via a special tube welded to the thick-walled steel tube. After diffusion bonding in a gaseous isostat, the outer tube is machined to almost the same thickness as the inner tube and then both steel tubes are etched in nitric acid.

Z

a

\ \ \ - 1//

\ 1 ..~.~ q

.~,f~ q

b

/

\

Fig. 11.14. A scheme [419] of making a composite shell in a gas isostat. (a) Densification of a blank with the outer rigid wall; (b) The same with the inner rigid wall.

Techniques

Ch. XI, w

Mandrel

Diaphragm .\

die

.

.

/B lank

~ ~ ~ \ ' { ~ \ ~ \ \ \ \ \ \ S x N ~1 Ib,~\\\\\\\'Lxt'<\'L<,~\4X,~?\41 I1 V,,'X6 . . . . . ~,L.r////A II

.

d

Segment

497

!i

I '

II-

-~A-F,7,4

.......

/~

\\

...........

~,/"Jv-/KR\\%"~\~\~\\\;\\~-\\~

;///A \ ~ ' / . 4 / K / / / / A / / / / / / / / / d F////////~/./.4//././,/4~ \ fi" "v)/22 "''rd- ....

| I

/

./..

.....

////A

....

...

'

-"

"///'~Fitting

Fig. 11.15. The assemblage for producing boron/aluminium tubes. After Sarkissyan et al. [582].

This method has many advantages as well as disadvantages. A scheme suggested by Hearn et al. [240] permits decreasing the volume of gas and producing composite tubes without using an isostat. A similar equipment (fig. 11.15) was used by Sarkissyan et al. [582]. The equipment installed in an ordinary furnace allows to produce tubes with the outer diameter of 20 to 75 mm and the length up to 1.5 m. The time/temperature profile is controlled by monitoring the temperature at a reference point of the outer surface of the die. Keeping input energy of the furnace as well as the boundary conditions (that is especially important for the cooling stage) constant, makes the equivalent time given by eq. (11.24) be a function of the maximum temperature To (see fig. 11.16). Therefore, it is not necessary to measure temperatures at various points of the blank. A scheme of the densification with the internal diameter being constant (fig. 11.14b) can be effectively used for making tubes of relatively small diameter, up to 20-30 mm.

~r

_ To

o

\ o

Fig. 11.16. Schematic of the temperature change on the die and composite blank. After Sarkissyan et al. [582].

498

Ch. XI, w

Hot pressing

11.3.3. Shells

Fabrication of metal-matrix-composite shells is also possible by using a 'soft' tool. Again, this can be done either in a gaseous isostat or in a die designed to reduce the volume of the gas. Choosing a scheme of those shown in fig. 11.14, a and b, we should take precautions to prevent both excessive fibre breaks or matrix cracking during densification of the blank when the external diameter is constant or a possible fibre kinking [441] (internal diameter is constant), the latter problem being a more difficult problem to conquer. So the scheme of fig. 11.14a looks preferable. Mileiko and Khvostunkov [438] described fabrication of boron/aluminium shells. The blank, after plasma spraying, is encapsulated into a container shown schematically in fig. 11.17. Winding the fibres and spraying the matrix are performed onto the shell-mandrel made of stainless steel of 18-8 type. The wall thickness of the shell-mandrel is about 1.5 mm. The blank after spraying is inserted, with a minimal gap, into a segmented die two halves of which are kept together by the end rings. The whole set is then inserted into the external shell which encapsulates the blank hermetically by welding. Then the container with a blank is subjected to hot pressing in a gas isostat. The extraction of the boron/aluminium shell from the container is performed by turning both the shell-mandrel and sealing welds and then cutting the external shell in the longitudinal direction. The quality of a shell obtained depends, in particular, on fabrication parameters. They determine a configuration of the fibre-break system, in addition to properties of the matrix and fibre/matrix interface. When a set of parameters is chosen appropriately, then the fibre breaks occurred during consolidation of the blank are distributed homogeneously, so they do not influence the shell performance. Otherwise, in a sufficiently thick shell, the fibre breaks are concentrated along one plane as can be seen in fig. 11.18 where specimen 1006 exhibits white spots illuminating the breaks.

Segmented die N

~

Internal steel shell

~

rl#llll#llllllllWllllllllllllZ#llllil.llllilllll/llll111111lli111~

/

steel shell

Fig. 11.17. The container with a blank. After Mileiko and Khvostunkov [438].

Ch. Xl, w

Techniques

499

Fig. 11.18. Boron/aluminium shells with external diameter 122 mm and wall thickness between 3 and 4 mm. The white spots on the shell surface illuminate the collective fibre breaks. After Mileiko and Khvostunkov [438].

In the case of densification of the blank with the external diameter being constant, the fibres in the internal layer experience strain e ~ (h/r)p where h, r, and p are the wall thickness, internal radius and porosity of the blank, respectively. Hence, for usual values of the porosity, p = 0.2, and ultimate fibre strain, e* = 0.007, we have a m a x i m u m relative thickness of the shell wall h/r ~ 0.035 to be densified without a danger of fibre breaks. Actually, because of the presence of rough defects in fibres and gaps in the setup shown in fig. 11.17, the m a x i m u m wall thickness appears to be even smaller. Consequently, one should be looking for such a densification route as to exclude a possibility of the collective fibre breaks and to provide a structure with r a n d o m fibre breakage. Obviously, to stimulate r a n d o m fibre breakage the main densification stage is to be performed when the fibre/matrix interface strength is relatively low. Experiments with various pressure/time and temperature/time routes (fig. 11.19) were done on shells with h/r = 0.07 to choose a consolidation regime. The results show that in order to exclude both collective fibre breaks along some lines (fig. 11.18, Specimen 1006) and longitudinal cracks resulted from the coalescence of the collective breaks (fig. 11.18, Specimen 995) to achieve r a n d o m fibre breakage as shown in fig. 11.18 (Specimen 1008), it is necessary to accept the following regime: p0 ~ ps ~ 55 MPa, Ts ~ 460-500~ The fabrication route described has been used to produce boron-aluminium shells of various configuration, some examples of the shells are also shown in fig. 2.17. A simplification of the equipment by drastic decreasing the gas volume can be done as shown in fig. 11.20. In this case, a gas-tight bag welded of the sheets of stainless steel is located between the blank and the segments with a conical internal surface pushing outward by a conical puncher loaded by a hydraulic press.

500

Hot pressing

Ch. XI, w

71, p

2ZTC

P~

--

~perature

,.

Fig. 11.19. A scheme of possible technological regimes for the fabrication of composite shells in gaseous isostat. See the text for details.

11.4. Rolling and drawing Rolling and drawing can be used to fabricate composites, mainly processing a composite when the matrix is in a semi-solid state. This case is a part of so called compocasting technology, so it will be considered in the next chapter (Section 13.2.5). Still, occasionally these processes have been used for processing solidstate-matrix blank. Also, rolling and drawing can improve the microstructure and so enhance properties of metal matrix composites. A possibility of producing metal-metal composites by rolling was established in 60s. Perhaps the only essential feature of the process is a necessity to conduct it in vacuum or with a protective atmosphere if the foil is used as a matrix precursor. It can be achieved either using a vacuum-tight container or a vacuum rolling machine. In this way the molybdenum-nickel specimens were obtained by Kopecky et al. [328]. To use rolling for making brittle fibre composites one needs to choose processing parameters more carefully. Gusev et al. [215] were able to get a boron-aluminium composite by rolling plasma sprayed monotapes, the matrix being an A1-Zn-Mg alloy. A composite of good quality was obtained after 5 to 6 passes with total reduction of about 50% , at a temperature of 400-500~ Doble and Toth [132] have reported to obtain a boron-aluminium composite by rolling a package of the matrix foils and the filament mat, kept together by a polystyrene binder. The package is placed in between two stainless steel cover plates

RollhTgand drawh~g

Ch. XI, w

Argon gas

501

Conical s e g m e n t s P r e s s u r i z e d bag / B lank

I

////////,' AA

i i ' l

! "//////// ente d die

Fig. 11.20. An equipment for consolidation of composite shells and tubes. After Mileiko et al. [462]

of about 8 mm thickness. The assembly is preheated either in an argon atmosphere or in air. (In the latter case, it has to be wrapped with a protective foil.) The most important parameters of the process are temperature and rolling pressure. When temperature increases, the pressure necessary to form a bond goes down and at high values of the pressure, fibres tend to fracture. An optimal temperature for the 6061alloy is 565~ which is slightly higher than that for static hot pressing. Rolling speed does not seem to influence the composite strength. The same method has also been used for producing precursor tapes. The largest sizes of the tape obtained by the authors are 1200 x 15 mm. Drawing in an evacuated container, as a method of composite fabrication (fig. 11.21), can be found in some publications. For obtaining graphite-aluminium composites by using precursor wires [189], the temperature of the process is about 500~ the speed about 5 cm/s and the material of the container is inconel. To obtain a composite with a sufficiently high strength, the value of the reduction should be between 5 and 15 % . It has been noted that a minor modification to the fabrication scheme allows tubes with various profiles to be made.

502

Ch. XI, w

Hot pressing

Die o

HeATeR

o

CONTAINER

~_.>..., ;...,;

~,

"' ~

~x~~

/o

o

//

o

COMPOSITe WIRES

ooo ~

Fig. 11.21. Drawing in an evacuated container.

11.5. Explosive welding Explosive welding can be considered as a kind of hot-pressing. The phenomenon, discovered by Lavrentiev at the end of the W W II [122], is an efficient way to bond together two similar or dissimilar materials. In 70s, the explosive welding was intensively used in attempts to develop an economically effective fabrication route for metal matrix composites. Then the activity started to diminish. There is perhaps a hope to reanimate the interest in using the method because it seems to be an easy way to utilize the explosive after the cold war has ended. In the process of explosive welding, the impact of two plates is accompanied, as a rule, by a stationary periodic process leading to a characteristic modulating waved interface (see fig. 11.22). All the important events take place in the vicinity of point

Fig. 11.22. Wave formed at the interface obtained by explosive welding. After Mileiko [419].

Ch. XI, w

Boron/aluminium composites

503

eXpLOsive

~

PLATe

GROUND PLATe

Fig. 11.23. A scheme of explosive welding of two plates.

A, which runs behind the detonation front (see fig. 11.23). Certainly, the conditions for forming the cumulative jet occur periodically here. Processes that occur around point A in the presence of fibres between the plates located either in the direction of the welding front or along normal to the front, are analyzed in [715]. In the first attempts to produce metal matrix composites by explosive welding (see, for example, [108, 281]), the direction of detonation front propagation was chosen to coincide with the fibre direction. In this case some new features were observed and they were associated with the flow of the matrix layer around a more rigid fibre [615]. Many experimental data show that the choice of technological parameters in this case is not a difficult task. The situation is easily reached when the mechanical behaviour of a composite obtained by explosive welding corresponds to that of a composite with an ideal interface bond. Moreover, the very short time of the process excludes the formation of brittle interface layers as well as annealing the reinforcing wires. Another advantage of explosive welding is the possibility of making sheets and plates of large sizes with a high productivity rate. When making composite tubes and shells with non-axial fibre directions it is not clear a priori what will be the result of the interaction of wave-forming processes at the matrix-matrix interface with a set of the reinforcing fibres lying, for example, along the impact front. A corresponding experiment is described by Mileiko et al. [441,443]. An assembly before detonation is shown in fig. 11.24, and an example of a structure can be seen in fig. 11.25. An interesting situation arises when the wave formation is completely excluded, which is a case, for example, when the reinforcing element is a wire mesh [50]. In this case, the bonding mechanism appears to differ from that described above. In going through the mesh, the matrix surfaces are cleaned and then the physical contact arises and the bond forms.

11.6. Boron/aluminium composites It is difficult to distinguish between the material and structural element when dealing with composites. Still, we shall divide the following discussion of the two topics mainly from the point of view of convenience. Although boron/aluminium composites have been constantly mentioned in illustrations of various aspects of the

Hot pressing

504

9

9

9

,

,

,

,

.

.

"

.

.

,

,

Ch. XI, w

.

r

e

c

-

.

.

. ,

,

.

.

.

~

.

9 9

.

:

. .

-

.

.

.

Fig. 11.24. An assembly to fabricate a composite tube with circumferencial reinforcement 9 l - mandrel; 2 and 4 - matrix layers: 3 reinforcement layer; 5 - protective shell; 6 - explosive; 7 - detonator; 8 - cone. After Mileiko and K o n d a k o v [441].

Fig. 11.25. The steel-aluminium composite obtained by explosive welding (Kasperovich and K o n d a k o v ) .

behaviour of MMCs, we are coming back to these composites because, first, it is a material thoroughly studied and its behaviour models the behaviour of various metal matrix composites rather well.

Boron/aluminium composites

Ch. XI, w 1.6

505

11.6.1. Materials Formation of the microstructure of boron/aluminium composites is determined by the processes discussed in detail above, those being plastic yielding and interface bonding as well as diffusion through the interface. The latter process is determined by a low solubility of boron in aluminium and a larger affinity of boron for magnesium which is a common alloying element in aluminium. It was shown in Section 10.2.2 that a heat treatment of a boron/aluminium composite with a matrix containing magnesium can yield formation of magnesium borides in the matrix rather than aluminium borides at the interface. The situation is similar to well studied phenomenon of the internal oxidation of alloys [116]. In that case the diffusion of oxygen from the surface of a solid into the volume leads to oxide precipitation. For the element to react with oxygen it must have an affinity for oxygen higher than that of the main element of the solid solution. When boron penetrates an alloy, such as an aluminium alloy containing magnesium, the sequence of events may be almost the same, only internal boriding occurs instead of internal oxidation [448]. The condition for the process to proceed are clear [116], namely cSDB >> CMgDMg, where c s is the boron concentration on a boron-source/alloy interface, CMg is the magnesium concentration in the solid solution, DB and DMg a r e the diffusion coefficients of boron and magnesium in the alloy, respectively. For the simple case where c s << CMg and the absolute value of the free energy of formation of boride Mg~B~ is large, as in the case of boron/aluminium, the concentration of boron and magnesium in the alloy should be as shown in fig. 11.26.

CMg C

s CB

x:

x

Fig. 11.26. Schematic representation of the dependence of boron and magnesium concentrations in the aluminium matrix on the distance from the interface. After Mileiko et al. [448].

506

Hot pressing

Ch. XI, w

Following the analogy with internal oxidation, the internal boriding front may be assumed to be quite sharp, which means that the thickness of the front layer is small in comparison with xf, the concentration of free magnesium in a solid solution at x < xf may be neglected, and the boride content within the internal boriding volume may be assumed to be independent of x. The larger the value o f ] - G ] the smaller the particle size. The particles are becoming finer as the rate of movement of the boriding front increases. Again, following the same analogy, we can write the equation for the rate of the boriding front as dxf

o~r S 1

~ ~ o d---t-= DB -fl- CMg xf

(11.25)

Obviously, in the axi-symmetric case eq. (11.25) transforms to drf a cs 1 d--t- = DB fl CMg rf l n ( r f / R )

(11.26)

where R is the radius of a cylindrical source of boron (a fibre) which is constant in the case under consideration. Just before linking zones, eq. (11.26) may not describe the situation well enough because of a decrease in the magnesium content at r > rf. The kinetics of the internal boriding just described can yield the formation of dispersion hardened zones around the fibre in a boron reinforced aluminium/ magnesium alloy. The influence of the zones on the failure behaviour and the strength of boron/aluminium composites was analyzed in Section 5.2.2. Here we continue the discussion introducing the dimensionless effective heat-treatment time as - (t/to)exp(-Q/RT)

(11.27)

where to is an arbitrary constant and Q is the activation energy for an appropriate physical process which may be the diffusion of either magnesium or boron in aluminium. With the obvious difference of the diffusion constants for boron and magnesium in mind we take Q = 38.5 kcal/mol as the activation energy for magnesium diffusion in aluminium [314] and obtain the strength versus effective time plot fig. 11.27 for various combinations of temperature and time used in [448,456] to produce boron/aluminium composite specimens. Obviously for changing temperature an integral representation of the effective time similar to that given by eq. (11.24) should be used, or in the case of piececonstant temperature: z = Z(ti/to)exp(-Q/RTi).

(11.28)

i

Figure 11.27 justifies the idea of the effective time. A shift of the critical point to smaller values of the effective time for set 2 of experimental data can be explained by the presence of a larger fibre volume fraction for this set.

Boron/aluminium composites

Ch. XI, w

1400

,

,

,

,,,,i

I

,

,

,

,,,uu

I

,

,

,

,u,,v

I

u

,

507

u

~3 1200 9

1000

0

*

O 800

600

0

0 0

400

~ooooo Set;

! Set 2

-oooo, 200

'

1 0 _1,e

'

I ~ .... I

~

'

' ~ .... I

1 0 -11

|

1 0 -1o~

I ' ~ .... I

l~.

I

I

1 0 -9

Fig. 11.27. Strength of boron/aluminium composites with A1-6Mg matrix as a function of effective heattreatment time according to eqs. (11.27) and (11.28) with to -- 10-l~ min. Set 1 of the experimental data is from Ref. [456], set 2 is from Ref. [448]. The average fibre volume fraction for set 1 is about 0.5, for set 2 is about 0.07 higher. After Mileiko et al. [448].

11.6.2. Structural elements Boron/aluminium composites, as mentioned in Section 2.3.1, have been used in aerospace structures. Therefore, fabrication methods to produce structural elements exist, some of them, namely those for making tubes and shells were described above. The behaviour of shells under hydrostatic pressure was discussed in Section 8.3.2 and that of tubes under axial c o m p r e s s i o n - in Section 8.2. Here we continue the discussion of the mechanical behaviour of b o r o n / a l u m i n i u m tubes with longitudinal reinforcement.

Tube strength versus fabrication regime Introducing the effective time for composite consolidation during hot-pressing according to eqs. (11.23) and (11.24) and following the assumption made above, Section 11.3.2, about dependence of the effective time on a m a x i m u m die temperature, Tmax, reached in the consolidation process (fig. 11.16), we present the tube strength as a function of Tmax. Figure 11.28 shows, first, that such a representation is a reasonable one, and second, similar to the behaviour of the strength of plane specimens (fig. 11.27), the tensile and compressive strength of tubes reaches a m a x i m u m at some value of Tmax, or equivalently, of the effective time.

Hot pressing

508

o 9 9

1800

Ch. XI, w

Tension Compression Compression, fai/ure at the fitting end 9

I

"

I

"

I

"

1600

\14oo b 1200 1000 800 0

600 =

480

I

490

=

l

I

500

510

l

520

T/~ Fig. 11.28. Strength of boron/aluminium tubes with AI-6Mg alloy as a matrix versus the maximum temperature of the outer surface of the die. The tube diameter is 60 mm, the length is ~ 400 mm, the wall thickness is ,-~ 1.5 mm. After Sarkissyan et al. [582].

Fitting ends Unlike the case of tubes as well as other structural elements made of fibre reinforced polymers where the problem of load transfer to the element is usually solved by quite a heavy design of the fitting ends (the problem is similar to that of gripping tube specimen for its testing which has been discussed in [630]), the load can be transferred to a boron/aluminium tube through sufficiently light fitting ends [582]. An example of an effective fitting end design is shown in fig. 11.29. Despite some guidance for choosing characteristic sizes and angles for such a fitting end design can be found as a result of consideration of the stress state and failure behaviour of overlapping joints (see Section 4.7.3), a large a m o u n t of the experimental work is still necessary to optimize the design.

11.7. Silicon-carbide/titanium composites We consider here both composites with titanium alloy and titanium aluminides matrices.

Silicon-carbide/titanium composites

Ch. XI, w

509

I

1.5 I

::!1 Fig. 11.29. An example of the fitting end design for the boron/aluminium tube. The design provides an appropriate load transfer if some precautions are beheld. After Sarkissyan et al. [582].

11.7.1. Titanium alloy matrix

A composite with Sigma BP SiC-fibre coated with a thin TiB2 layer and (~ +/~) titanium matrix (Ti-6A1-4V) reveals a fibre break accumulation failure mechanism [693]. A degradation of the fibre strength during the composite processing lowers the low-strength part of the fibre strength distribution which has no effect on the composite strength. To decrease processing temperature and so decrease the thickness of the reaction zone, it was suggested a special titanium alloy, Ti-4.5A1-3V-2Fe-2Mo, of "high formability" at sufficiently low temperatures [179] to be used as a matrix. It occurred that the consolidation temperature of 750~ for 2 h under a pressure of 150 MPa yields a maximum value of the room temperature composite strength. The composite retained its strength up to 500~ 11.7.2. Titanium-aluminide matrix

Titanium aluminide matrices are less tough than titanium alloy matrices. An example of tensile behaviour of such a composite is given by Brindley et al. [59] who produced and tested Sigma BP SiC-fibre/Ti-24A1-6Al-matrix composite. The fibre is coated by a diffusion barrier composed of TiB2. The matrix consists of ~2(Ti3A1) and/~ phases the latter being introduced to enhance the ductility of the intermetallic alloy. It is known (see Section 10.2.2) that the reaction zone in such a composite is surrounded by a matrix zone depleted of//-phase. Then, the alloy ductility is strongly influenced by oxygen content. The authors revealed just a limited fibre breakage in the composites that had occurred in the vicinity of the failure surface and only immediately before the failure, although cracks within the reaction zone and the/~-depleted zone were revealed at early stages of loading. This

Hot pressing

510

Ch. XI, w

microstructural observations mean that the fibre volume fraction in the composite lies within interval v) - vfB in fig. 5.2 that actually corresponds to a large scatter of the composite strength (fig. 11.30). Because fracture toughness of the matrix is relatively low, some contribution to the microcrack arrest comes from the interface. In fact, the authors measured the interface stress at debonding, r,, and that during sliding restrained by the friction, r**, and found ~, ,,~ 105 MPa and ~** ~ 60 MPa at room temperature. The composite behaviour at elevated temperatures is the same, although experimental data here are even more scarce (fig. 11.31). Still, it is clear that SCS6 SiC/(Ti3A1 + Nb) composite has superior specific strength and creep-rupture strength as compared with the corresponding properties of an IN-100 superalloy (see figs. 11.32 and 11.33). However, the superiority inheres with regard only to the properties along the fibre direction, the transverse properties are too low. To enhance fracture toughness at room temperature of titanium-aluminide-matrix composites, it is necessary to introduce a weak interface (Section 5.8) which can decrease creep and creep rupture resistance of the composites (Sections 6.1.2 and 6.2.2). Hence, it is necessary to optimize the interface properties to balance the two characteristics mentioned. Weber et al. [694] showed that single-crystallinealumina/TiAl-matrix composites with graphite interlayer providing a weak interface can be an example of such a balance. Their creep resistance at about 1000~ as well as effective surface energy at room temperatures look promising.

1500

,

,

~

o 0

r,..b

, ,

G

o

o o o o

1000

0

-It

b

8 ~

500 o o o o o C~= ooooo *****

0 0.0

I

0.1

600 1200 1600

1000 1300 1900

-

I

0.2

ppm ppm ppm

I

0.3

0.4

Fig. 11.30. SiC-fibre/Ti-24Al-11Nb-matrix composite: strength versus fibre volume fraction. The matrix contains various amounts of oxygen. Experimental data after Brindley et al. [59].

Ch. XI, w

Silicon-carbide~titanium composites

1500

I

I

!

511

I

r~ o

1000

8

b

500

425~ 650~ 815~

ooooo ooooo ooooo

I

0

I

0.I

0.o

I

0.2

I

0.3

0.4

0.5

"Of Fig. 11.31. SiC-fibre/Ti-24Al-11Nb-matrix composite: elevated temperature strength versus fibre volume fraction. Experimental data after Brindley et al. [59].

~ . 0

,

I

,

I

'

I

|

..4 "~-. 1.5 b Q. "~'- 1,0 b

_/

TisA I + Nb

0.5

S'~C/(Ti~4 I+Nb) -

0.0

i

0

I

200

i

transverse I

I

400

I

600

T /

I

800

~

Fig. 11.32. Specific strength of Ti3A1 + Nb matrix and SiC/(Ti3A1 + Nb) composite normalized by that for IN-100 superalloy versus testing temperature. Experimental data after Larsen et al. taken from [345].

512

Ch. XI, w

Hot pressing

2.5

I

|

|

i

|

I

|

|

I

l

I

|

i

|

,

I

!

i

i

|

,--~ 2.0 b ~

~

s'c

_

b

It~___~ ~*~'~

i

1.0 T'i,sA l +Nb

0.5 Ss

0.0

l

21

TisA l,+Nb) |

|

i

I

22

i

i

i

i

trans~erse_._..._..__.~ I

i

J

22

i

i

I

2a

i

i

i

i

23

LM

Fig. 11.33. Specific stress of Ti3AI + Nb matrix and SiC/(Ti3AI + Nb) composite normalized by that for IN-100 superalloy versus Larsson-Miller parameter, LM = [20 + log(t/to)]T where t is rupture time in hrs, to = 1 h, and T is testing temperature. Experimental data after Larsen et al. taken from [345].

11.8. Glass- and glass-ceramic matrix composites Two driving forces for the development of glass and glass-ceramic matrix composites are known, these being (i) a versatility of the matrices due to a variety of their physical properties and (ii) processing temperatures lower than that for ceramic matrix composites. Such composites are candidates for elevated temperature structural applications, especially when some physical properties are necessary, missile radomes being an example. Hence, this type of composites has been intensively studied starting certainly with early work by Phillips and co-workers (see Section 2.3.2) and that by Prewo and his co-workers [30, 550]. As a matrix, borosilicate and aluminosilicate glasses are utilized. Glasses can be processed at relatively low temperatures, this minimizes unwanted fibre/matrix interactions, but lowers an upper service temperature by 600-700~ Glass-ceramics are normally found in the following systems: Li20-A1203-MgO-SiO2 - lithium aluminosilicate (LAS), CaO-AI203-SiO2 - c a l c i u m aluminosilicate (CAS), MgO-AI203-SiO2 - magnesium aluminosilicate (MAS), BaO-MgO-AI203-SiO2 - barium/magnesium aluminosilicate (BMAS), BaO-AI203-SiO2 - barium aluminosilicate. The compositions can include some oxide additions as nucleating and fluxing agents. Processing of glass-ceramic matrix composites including crystallization, needs higher temperature than glass-matrix composites. Service temperatures for the LAS system containing /~-spodumene crystalline phase are between 1000 and 1200~

Glass- and glass-ceramic matrix composites

Ch. XI, w

513

those for MAS based on the cordierite composition -1200~ [551] . The highest service temperature is certainly characteristic for barium aluminosilicate matrix based on hexacelsian, although there seems to be no reports on a real usage of such glass-ceramics as a matrix. Moreover, a doubt is expressed in a very possibility of such a use due to a phase transformation of hexacelsian with a large volume effect [721]. Varying heat-treatment regime yields variations in crystal/glass composition of a glass-ceramics; among properties which vary accordingly, is thermal expansion coefficient, e. This makes glass-ceramic matrices be very convenient due to a possibility to adjust values of e for the fibre and matrix and to lower the residual stresses in the composite. Normally, hot pressing of slurry infiltrated monotapes (Section 11.1.2) are used to produce the composite, although some other techniques such as hot matrix transfer into woven preform and hot injection molding of chopped fibres compounds in preforms can be also used to produce elements of complicated shapes [551]. If hot pressing and heat treatment parameters are chosen adequately, the fracture toughness and strength of a composite can be sufficiently high. Some typical values of mechanical properties of such composites are presented in Table 11.3. The failure of a brittle-fibre/brittle-matrix composite (see Section 5.8) is a process involved matrix cracking, fibre pullout, interface debonding, etc. An interaction of those processes determines strength and fracture toughness of the composites and is also revealed in the shape of stress/strain curve. Note that niobium oxide present in some LAS matrices reacts with silicon carbide of the fibre that yields a formation of a thin C / N b C layer at the interface [551] which occurs to satisfy requirements to the interface fracture toughness to deviate the matrix crack [159]. Relatively high fracture toughness of ceramic-fibre/glass-ceramic-matrix composites yields a sufficiently small strength scatter (large value of the Weibull parameter /~). At the same time, a variety of fracture mechanisms involved in failure process of the composites makes them very sensitive to both testing method (loading pattern) and specimen sizes. Jansson and Leckie [279] analyzed the results of testing of a unidirectional SiC/LAS composite in tension, four-point bending and three-point

TABLE 11.3 Mechanical properties of some unidirectional glass- and glass-ceramic composites Fibre

Matrix

vf

Graphite SiC Tyranno SiC Nicalon SiC Nicalon SiC Nicalon

BSG MAS CAS LAS LAS

0.44 0.40 0.37 0.35

Tp~

810

Tc~

1100

1200

Young's Tensile modulus strength GPa MPa 170 80 124

1400

1 Borosilicate glass. Tp- temperature of hot pressing, Tc -crystallization temperature.

Bending Work of Source strength fracture MPa kJ/m 2

785 334

1000 793 475 750

30 50

[4831 [403] [2321 [601] [551]

514

Hot pressing

Ch. XI, w11.8

bending of specimens of rectangular and triangular cross-sections. They observed a few patterns of the crack behaviour as well as the shapes of load/displacement curves. The most important finding is perhaps that the comparison of the strength values obtained from various tests yield an effective value of/3 equal to 21 which corresponds to a very small scatter of the strength values. Composites in a SiC/LAS system retain their room temperature strength up to temperatures of about l l00~ [551]. Work of fracture of SiC/MAS composite changes with the testing temperature reaching a maximum, 45 kJ/m 2, at 500~ Because of high anisotropy of unidirectional composites, cross-plied laminates are to be considered for many possible applications. Pryce and Smith [555] studied microcracking in longitudinal lamina of (0x/90y)s SiC-Nicalon/CAS laminates. They found that the progressive cracking of these lamina is reasonably similar to that observed in unidirectional composites and described approximately by the ACK model (Section 4.4.1). Although, the interaction of the neighbouring plies leads to some differences from what the model predicts. A simple shear-lag model [296] of a lamina subjected to microcracking, incorporated into a stress analysis of a laminate predicts the stress/strain behaviour of SiC-Nicalon/CAS laminates. The strength behaviour of angle-ply carbon-fibre-reinforced borosilicate glass presented in fig. 11.34 can be certainly described by Tsai-Hill failure criterion, eq. (3.132). At least, the authors made such a statement.

lO0O (

800(y tl,

-

b 600-

400

200

%

0 I

,0

45 ~ /

,

,

grad

9O

Fig. 11.34. Nardone and Prewo's experimental data on strength of carbon-fibre/borosilicate-glass-matrix + 0 angle-plies [483].

Ch. XI, w

Graphite-aluminium composites

515

11.9. Graphite-aluminium c o m p o s i t e s

Such composites are normally produced by liquid-phase technologies, so they shall be described in detail in an appropriate place (Section 13.3.1). Hot pressing is just sometimes used to obtain such composites. For example, composites of such a kind can be produced by vacuum hot pressing of prepregs obtained by ion-plating technology disclosed by Ohsaki et al. [513]. The prepregs have been pressed at a temperature of about 540~ and pressure of about 60 MPa for 1 h. The fibres used were of low m o d u l u s - high strength type, the matrix was pure aluminium. The mean composite tensile strength occurs to be low, that is about 800 MPa, the strength scatter at a volume fraction just above 40% seems to be very large which is obviously not unexpected. To optimize fabrication parameters of hot-pressing of graphite/aluminium prepregs, Masson et al. [389] varied time/temperature conditions when performing the consolidation process in vacuum under a pressure of 25 MPa. The results of tensile testing of specimens obtained is presented in fig. 11.35. The failure surface of a composite obtained at 600~ min shows the crack propagation in one plane, that for a composite obtained at 580~ min reveals the failure process based on the fibre breaks accumulation.

700

I

I

I

I

I

i

I

I

'

I

ooooo T = 5 8 0 " C 9 ,,,, T = 6 0 0 " C

\

600

*b 500

400 -

30fl

, "t

0

,

I

,

,

25

t

I

/

40

,

,

I

55

,

70

Fig. 11.35. Tensile strength of graphite/aluminium composites obtained by hot-pressing of prepregs at two temperatures versus consolidation time. The fibre volume fraction changes irregularly from 0.39 + 0.10 to 0.43 + 0.06. Experimental data are after Masson et al. [389].