EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH

CERN–ECP/94–20
December 1994

ELASTIC MODULI OF POLYCRYSTALLINE DIAMOND

E. Borchi, M. Bruzzi, U. Biggeri

Dipartimento di Energetica, Firenze, Italy

C. Furetta, F. Lemeilleur
CERN, Geneva, Switzerland

Abstract

The behaviour of crystallite orientation for polycrystalline CVD diamond is examined and its
influence on elastic moduli considered. In particular the case of axial texture is worked out and
some bounds of elastic constants are derived. Finally, a comparison with previously existing
experimental data is carried out.

Paper presented at the Diamond Films ’94 Conference: 5th European Conference on
Diamond, Diamond-like and Related materials, Il Ciocco, Lucca, Italy, 25 –30 September
1994.
Submitted for publication to Diamond and Related Materials.
1 INTRODUCTION
The possibility of growing chemical vapour deposited (CVD) diamond films with the
outstanding properties of natural-bulk diamond has stimulated extensive research efforts. The
research is aimed at harnessing these properties — extreme hardness, wide-band optical
transparency, and high thermal conductivity) — for mechanical, optical and electronic
applications.
Diamond films have been suggested for use in tracking detectors in the highest radiation
areas of the LHC and, towards this, results have been presented on charge collection properties,
radiation hardness and low noise electronics [1, 2]. In addition, minor attention has been
devoted to the study of the mechanical and thermal properties. In particular, Bonnot [3] has
used Raman spectroscopy to analyze diamond-thin films as a function of crystallite orientation.
More recently, the orientation distribution of the diamond crystals in the CVD diamond films
produced by several nitrogen concentrations in CH4/H2 and CH4/H2/N2 gas mixtures was
investigated and the crystalline quality judged by Raman spectroscopy [4]. Using the same
experimental technique for stress and crystallinity as in [100], [110] and [111], oriented
diamond films were also accurately studied [5]. The actual mechanical properties of CVD
diamond films are nevertheless difficult to measure — especially their shear modules.
Moreover, the elastic moduli depend on the incorporated impurities (i.e. nitrogen and boron)
and the crystal orientations in the polycrystalline sample.
In this work an evaluation of the bounds of the elastic moduli of polycrystalline diamond
is presented, using the theory of elasticity both for an isotropic distribution of crystallites and
for a small number of textures, as deduced from the existing experimental data. Specifically
transverse isotropy axial texture is examined by using the Voigt and Reuss averaging
procedure. The results of this study, although restricted to second order elastic constants only,
offer a first partial response to the mechanical characterization of CVD diamond samples. A
more complete study of the behaviour versus texture of the third-order elastic constants of
anisotropic polycrystalline diamond will be presented in a forthcoming paper.

2 ISOTROPIC DIAMOND POLYCRYSTAL

In the case of a macroscopically isotropic polycrystal all the averaging methods share
common assumptions: 1) small grain to specimen size, 2) absence of voids, nonhomogeneities
and so on, 3) cohesion of crystallites occurring through very thin grain-boundary regions
deformed relative to the crystal interiors, 4) randomly oriented grains and 5) grains large
enough so that interfaces remain non-important. On these bases several averaging methods
have been proposed using different starting hypotheses.
For a material such as diamond, composed of cubic crystals, one needs only two
independent constants to completely describe the behaviour: the bulk modulus K and the shear
modulus G. The bulk modulus K is the same as that of the single crystal
C + 2C12 
K =  11 ,
 3  (1)

where C11 and C12 are single-crystal elastic constants.The shear modulus is

(i)
G = C44 , (2)

1
 (i)
where C44 is the C44 elastic constant of the isotropic polycrystalline sample, depending on the
averaging method. From the knowledge of K and G the elastic constants Cij(i) of the isotropic
polycrystalline sample are easily obtained as
= K+ G
(i) 3
C 11
 4

= K− G ,
(i) 2
C 12 (3)
 3
with
(i) (i)
C 11 − C 12 − 2 C (i)
44 = 0 . (4)
In the Voigt model in which one assumes that the polycrystal strain is constant, the shear
module is [6]
G(V) = C (V)
44 = (1 / 5)(C11 − C12 − 3C 44 ) = (1 / 5)(2G1 + 3G 2 ) , (5)
where: G1 = (C11 – C12)/2 and G2 = C44 are the 'zero-order solutions'.
The Reuss method assumes that stress is constant. This lead to the bound [6]:
(R)
5/G = 4/(C11 – C12) + 3/C44 = 2/G1 + 3/G2 . (6)

The upper (u) and lower (l) bound G(HS)

u and G(HS)
l , obtained by Hashin and Shtrikman
using variational principle [7], are expressed in terms of the single-crystal elastic constants by
the equations:
−1
 5 
G(HS) = G1 + 3  
 (G 2 − G1 ) − 4β1 
l

−1
 5 
G(HS) = G2 + 3   (7)
 (G1 − G 2 ) − 6β2 
u

where:
3  (K + 2G1 ) 
β1 = −  
5 [G1 (3K + 4G1 )] 

3  (K + 2G 2 ) 
β2 = −  
5 [G 2 (3K + 4G 2 )] 

Finally, the self-consistent bound by Kroner [8], G(K), lies between G(HS)
l and G(HS)
u . Gairola
and Kroner [9] have established that in the case of macroisotropic polycrystals of cubic
symmetry the shear module G(K) satisfies the cubic equation:
(K)3 (K)2 (K)
G + αG – βG –γ=0 (8)
where:
α = (5C11 + 4C12)/8

β = C44(7C11 – 4C12)/8

γ = C44(C11 + 2C12)(C11 – C12)/8

2
 In Table 1 we give bounds of the effective moduli for macroscopically isotropic
polycrystalline diamond calculated from the previous formulas. Young’s modulus is given by:
−1
E = C44 .

The experimental values for single-crystal elastic constants (in unit of 1011 N/m2) are taken
from Mc Skimin and Andreatch [10]:

C11 = 10.76 , C12 = 1.25 , C44 = 5.78 .

The bounds of Table 1 have been evaluated from Eqs. (5) to (8). The coefficients α, β and γ of
Eq. (8) are all positive. Therefore the sum of the roots G(K) is negative and their product
positive. There is only one real positive root. For obtaining this root a method of successive
approximations proves quite easy and effective.

Table 1
Bounds of various orders for the effective moduli of microscopically isotropic polycrystalline
diamond (units are 1011 N/m2).

Young Shear Poisson Bulk

1011 N/m2 1011 N/m2 1011 N/m2

Voigt 11.48 5.37 0.068 4.42

HS upper 11.43 5.35 0.069 4.42

Kroner 11.42 5.34 0.069 4.42

HS lower 11.42 5.34 0.069 4.42

Reuss 11.39 5.32 0.070 4.42

The results of Table 1 have to be compared with the experimental data by Gray [11],
obtained from the measurement of ultrasonic velocities on six samples of polished CVD
diamond. The sample thickness was 800 µm. The average values presented by Gray are:

Young modulus E = 11.80 ± 6.00 [1011 N/m2]

Shear modulus G = 5.14 [1011 N/m2]
Bulk modulus K = 5.58 ± 12.00 [1011 N/m2]
Poisson ratio σ = 0.148 ± 0.006

3
Other experimental values presented by Sussmann et al. [12] agree substantially with Gray’s
results. The mechanical strength of bulk polycrystalline CVD diamond has been determined
using a bursting pressure technique on samples of 170 µm to 300 µm in thickness and
Young’s moduli values of 9.86 × 1011 to 10.79 × 1011 N/m2 have been obtained. These values
are all close to those published for single crystal natural diamond (10.50 × 1011 N/m2 ) [11].
Values previously reported for CVD diamond are in the range from 5.36 × 1011 to 12.24 ×
1011 N/m2 [13].
The comparison with our calculated values shows that major differences are relative to
the bulk module and Poisson ratio, suggesting a certain degree of anisotropy of the
polycrystalline aggregates (we only remark that in an anisotropic medium the elastic moduli
depend in general on the directions in the medium itself). In this case material characterization
becomes more difficult — either because changes in ultrasonic wave velocity due to the stress
are masked partially by anisotropic effects, or there are more elastic constants to identify.
To complete the theoretical presentation of this section, reported in Table 2 are the elastic
constants C (i) (i)
ij and the elastic compliances S ij of the isotropic polycrystalline diamond.

Table 2
Elastic constants C (i) 11 2
ij in units of 10 N/m , and elastic compliances S (i)
ij in
units of 10–13 m2/N for isotropic polycrystalline diamond.

(i) (i)
Model C 11 C 12 C (i)
44
(i)
S 11 (i)
S 12 S (i)
44
1011 N/m2 1011 N/m2 1011 N/m2 10–13 m2/N 10–13 m2/N 10–13 m2/N

Voigt 11.58 0.84 5.37 8.71 –0.59 18.62

HS upper 11.55 0.85 5.35 8.75 –0.60 18.69

Kroner 11.54 0.86 5.34 8.76 –0.61 18.73

HS lower 11.54 0.86 5.34 8.76 –0.61 18.73

Reuss 11.51 0.87 5.32 8.78 –0.62 18.79

3 THE ANISOTROPY MODEL

From a theoretical point of view the characterization of polycrystalline material in the
presence of texture presents a large degree of uncertainty. The evaluation of the effective moduli
is easy only for Voigt and Reuss models. For diamond in particular, calculations for

4
anisotropic aggregates are generally still lacking, especially as far as third-order elastic constants
are concerned. In this paper we consider only axial anisotropy and limits ourselves to the
evaluation of the second-order constants relative to a small number of textures — namely
complete alignment of the [111], [001] and [110] directions of the single crystal along the
sample symmetry axis (considered as the z-axis). Using the Voigt assumption (constant strains
throughout the polycrystalline material) the elastic constants are [14]:
v v v
C11 = C22 = C11 + (3 / 20)Cγ
v v v
C23 = C13 = C12 − (1 / 5)Cγ
v v
C33 = C11 + (2 / 5)Cγ (9)
v v
C66 = C44 + (1 / 20)Cγ
v
C v44 = C55 v
= C44 − (1 / 5)Cγ
v v
C12 = C12 + (1 / 20)Cγ .

Here C = C11 – C12 – 2C44, and γ is the 'elastic anisotropy index'. For a completely random
distribution of crystallites (isotropic material) one has γ = 0. For complete alignment of the
[111], [001] and [110] directions, respectively, of the single crystals along the specimen
symmetry axis one has γ = –2/3, γ = 1, γ = –1/4. A similar analysis may be carried out by
averaging the moduli of compliance Sij, assuming constant stress in the material (Reuss
model). The results so obtained are identical in form with Eq. (9) after replacing:

S ijR→ C ijV , S ijR→ CijV ; i, j = 1,2,3

S iiR / 4 → C ijV , S iiR / 4 → CiiV ; i, j = 4,5,6

S→ C

where S = S11 – S12 – S44/2, and the experimental values of the crystal compliances are (in
units of 1013 m2 /N):
S11 = 9.52 , S12 = – 0.99 , S44 = 17.30 .

In Tables 3a and 3b the anisotropic elastic constants C (a) (a)

ij and compliances S ij are given
respectively for the considered textures. One notes an overall moderate dependence on the
(a) (a)
model and a strong dependence of the elastic constants C 12 , C 13 and C (a)
33 on the texture.
From the previous results we evaluate the apparent moduli for polycrystalline diamond with
axial texture (Table 4). Whereas Young and shear moduli show weak changes, the Poisson
ratio presents a large variability with texture. The calculated bulk module, on the other hand,
always presents a large discrepancy between it and the experimental value given by Gray [11].
We conclude this section with a brief discussion concerning the dependence of the elastic
moduli on the direction. Although we are considering an anisotropic material, it is desirable to
define elastic moduli as is usually done in isotropic media. Following Wortman and Evans
[15] we define Young modulus as
(a)−1 (a)
E[i' ] = Si' i' , (10)

5
where i' is an assigned direction. When i' is making an angle α with the z-axis of the sample a
simple calculation gives,
(a)
E (a)–1(α) = S11 + (S (a) (a) 4 (a) (a) (a)
33 − S11 ) cos α – 2 (S11 − S13 − 1 / 2S44 ) cos α sin α
2 2 . (11)

Here S(a)ij are the compliance constants suitable for the Reuss (a = R) or the Voigt (a = V)
cases. In Fig. 1 we report the angular variation of Young modulus for diamond suitable for
[001], [110] and [111] textures; Voigt and Reuss results are shown in each case. For α = 0 one
obtains the values reported in Table 4. Analogously the Poisson ratio is defined as
(a) (a) (a)
σ[i' ][ j' ] = Si' j' / Si' i' , (12)

where i' and j' are assigned directions. To give a representation for a sample orientation, values
of Poisson ratio have been reported in Fig. 2 as a function of direction in the [100] plane. In
Fig. 2 both i' and j' are in the [100] plane. In this case the calculation gives

(a)
σ[i' [ (a) (a) (a) (a) (a) 2 2 (a)
][ j' ] = − S13 + (S33 + S11 − 2S13 − S44 ) sin α cos α E[i' ] (α) ] (13)

Table 3a
Elastic constants C (a)
in units of 1011 N/m2, for polycrystalline diamond with axial texture.
ij
For each texture the results obtained in the Voigt (Reuss) model are reported in the first
(second) row.

(a) (a) (a)

Axial Model C 11 C 12 C 13 C (a)
33 C (a)
44 C (a)
66
Texture 1011 N/m2 1011 N/m2 1011 N/m2 1011 N/m2 1011 N/m2 1011 N/m2

Voigt 11.78 0.91 0.57 12.13 5.10 5.44

111

Reuss 11.74 0.95 0.57 12.13 5.06 5.54

Voigt 11.27 0.74 1.25 10.76 5.78 5.27

001

Reuss 11.22 0.79 1.25 10.76 5.78 5.22

Voigt 11.66 0.87 0.74 11.78 5.27 5.40

110

Reuss 11.88 0.94 0.79 11.74 5.22 5.35

6
In particular
(a) (a)
σ[001][010] = −S13 / S(a)
33
(a) (a) (a)
σ[010][001] = −S13 / S11

(a)
The values of σ[001][010] are also reported in Table 4.

Table 3b
Elastic compliances S (a)
ij , in unit of 10
–13 m2/N for polycrystalline diamond with axial texture.

For each texture the results obtained in the Voigt (Reuss) model are reported in the first
(second) row.

(a) (a) (a)

Axial Model S 11 S 12 S 13 S (a)
33 S (a)
44 S (a)
66
Texture 10–13 m2/N 10–13 m2/N 10–13 m2/N 10–13 m2/N 10–13 m2/N 10–13 m2/N

Voigt 8.56 –0.64 –0.37 8.28 19.61 18.38

111

Reuss 8.59 –0.68 –0.37 8.28 19.78 18.05

Voigt 9.01 –0.48 –0.99 9.52 17.30 18.97

001

Reuss 9.06 –0.53 –0.99 9.52 17.30 19.16

Voigt 8.57 –0.53 –0.51 8.55 18.97 18.52

110

Reuss 8.50 –0.64 –0.53 8.59 19.16 18.70

7
 Table 4
Apparent moduli for polycrystalline diamond with axial texture. For each texture the results
obtained in the Voigt (Reuss) model are reported in the first (second) row.
Symbols are related to following definition:
(*) E–1 = S(a) ; (**) G' = C(a) ; (•) σ (a) = – S(a) / S(a) ; (••)K (a) = ( C(a) + 2C(a) )/3.
33 44 13 33 11 12

Axial Model Young(*) Shear(**) Poisson ratio(•) Bulk(••)

Texture 1011 N/m2 1011 N/m2 1011 N/m2

Voigt 11.70 5.27 0.060 4.47

111

Reuss 11.64 5.22 0.062 4.59

Voigt 10.50 5.78 0.104 4.25

001

Reuss 10.50 5.78 0.104 4.27

Voigt 12.07 5.10 0.045 4.53

110

Reuss 12.07 5.06 0.045 4.57

8
Fig. 1 Young’s modulus as a function of direction for diamond.The vertical direction is z (texture axis)
about which the figure has rotational symmetry. Voigt (O) and Reuss (∆) values are reported
together with the isotropic cases respectively (-) and (--) for Voigt and Reuss. Plots (a), (b) and (c)
refer to [001], [111] and [110] axial textures, respectively.

9
Fig. 2 Poisson ratio as a function of direction in the [100] plane. Both i' and j' are in the [100] plane. See
Fig. 1 caption.

10
4 CONCLUSIONS
Diamond has very distinctive elastic features. Compared with silicon, for example,
diamond has a much smaller equilibrium lattice constant (adiam = 3.567 A, asi = 5.429 A),
much larger bulk modules and elastic constants, relatively small C12 and a much smaller
Poisson ratio. A first-principle calculation of elastic properties of silicon and diamond has been
carried out by Fukumoto [16], explaining the main elastic properties of these materials.
As far as the elastic properties of polycrystalline diamond are concerned, the theory of
elasticity, applied to isotropic elastic samples, gives results which only partially explain the
experimental data. On the other hand, the CVD material does not behave isotropically but
presents both complicated microstructure effects and some degree of texture caused by the
crystals aligning themselves in certain preferred orientations during the forming process. In
particular one must expect the crystallinity quality of the films to be poor close to the substrate
[5] and that the state of strain partially produced in these film structures will induce lattice
deformations and distortions from cubic to tetragonal symmetry.
The preliminary results presented in this work show that texture may play an important
role in explaining the mechanical properties of CVD polycrystalline diamond.
In order to obtain a reliable characterization of the material it is necessary, from the
experimental point of view, to perform more quantitative measurements of elastic constant in
different texture limits.

References
[1] L.S. Pan, S. Han, D.R. Kania and K. Okano, Diamond Films and Technology, 2 (1992)
99.
[2] R & D Proposal, CERN/DRDC/P56, May 5, 1994.
[3] A.M. Bonnot, Phys. Rev. B41 (1990) 6040, (see also J.W. Ager III, D. Kirk Veirs and
G.M. Rosenblat, Phys. Rev. B43 (1991) 6491.
[4] R. Locher, C. Wild, N. Herres, D. Behr and P. Koidl, Appl. Phys. Lett. 65 (1994) 34.
[5] S.R. Salis, D.J. Gardiner, M. Bowden, J. Savage and S. Haq, Appl. Phys. Lett. 65
(1994) 43.
[6] H. Ledbetter, in Dynamic Elastic Modulus Measurements in Materials, A. Wolfenden
editor, American Society for Testing and Materials, Philadelphia, 1989.
[7] Z. Hashin and S. Shtrikman, J. Mech. Phys. Solids 10 (1962) 343.
[8] E. Kroner, Zeut. Phys. 151 (1958) 504.
[9] B.K.D. Gairola and E. Kroner, Int. J. Eng. Sci. 19 (1981) 865.
[10] H.J. McSkimin and P. Andreatch, J. Appl. Phys. 43 (1972) 2944.
[11] K.J. Gray, SPIE Vol. 1759, Diamond Optics V (1992) 203.
[12] R.S. Sussmann, J.R. Brandon, G.A. Scarsbrook, C.G. Sweeney, T.J. Valentine,
A.J. Whitehead and C.J.H. Wart, Properties of bulk polycrystalline diamond, seminar
given at CERN by R.S. Sussmann, October 12, 1993.
[13] G.F. Cardinale and R.W. Tustison, J. Vac. Sci. Technol. A9 (1991) 2204.
[14] H. Pursey and H.L. Cox, Phil. Mag. 45 (1954) 295.
[15] J.J. Wortman and R.A. Evans, J. Appl. Phys. 36 (1965) 153.
[16] A. Fukumoto, Phys. Rev. B42 (1990) 7462.

11
Mac/Word/5026F/SQ/dtp