.

.
.
.
. DATE January 2009
. PAGE 1/10

TIE-33: Design strength of optical glass and ZERODUR®

1. Introduction
The design strength of a glass article is an essential characteristic when it has to survive
mechanical or thermal loads during application. Frequently asked questions are:
what is the strength of optical glass ?
what is the strength of ZERODUR® ?
what thickness is necessary for a ZERODUR® mirror blank with a specified diameter
to endure the forces from attached actuators ?
Such questions concern the strength of glass directly or indirectly. In the following some
general information will be given on the strength of glass. Additionally, a calculation scheme
will be sketched which has been used for a variety of different applications. Data of common
optical glass types are enclosed.

Recently it has been shown that a homogeneous isotropic load of 41 MPa (stress rate
2 MPa/s) on a 1 m2 ZERODUR® surface (without any pre-damage), ground with bound
diamond grain tool D151, leads to a failure probability of less than 0.1%, 95 % confidence
level range. This stress level is 4 times higher than the conservative value of 10 MPa for a
moderate surface treatment.

2. General aspects of glass strength

The strength of glasses and glass-ceramics is not primarily a material property like the
Young's modulus e.g. The more prominent influence factors are:
the crack depth distribution at surfaces and edges
the size of the surface area exposed to tensile stress,
the temporal behavior of the load and
the environmental media.
All fracture of glass has at least two prerequisites: Tensile stress and (micro)flaws at the
surfaces or the edges.
Measurement results for different surface conditions and further information on the bending
strength of ZERODUR® can be found in chapter 6 and literature [4,5,6]. For low stress
applications 10 MPa can be taken as a safe value for the strength of ZERODUR® as long as
there are no heavily damaged surfaces or edges. Coarse grain worked surfaces (D151)
show significantly higher strength values than 10 MPa. Surfaces where all micro cracks have
been removed by etching lie even higher. For higher stress loads it is recommended to
analyze the applicable strength limit in more detail referring to the procedures and data given
in literature.

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 2/10

2.1 Microstructure condition of the glass surface

Milling and grinding of glass causes cracks at the surfaces. The depth of the surface cracks
is roughly equal to the maximum grain size. As a rule, decreasing grains sizes result in less
deep microcracks and in higher strength, and narrow grain size distributions lead to narrow
strength distributions.
Polished surfaces have the highest strengths. However, this is valid only if the microcracks
caused by preceding machining processes were eliminated. This can be achieved by
subsequent grinding with decreasing grain sizes, each grinding process taking off a material
layer at least three to four times as thick as the maximum crack depth from the preceding
step. However, the strength distributions of glasses with polished surfaces scatter
significantly more than those for glasses with ground surfaces.

2.2 Area of the tensile stress loaded surface

The admissible strength for a tensile stress loaded area decreases with increasing area.
Deeper microcracks occur more frequently when the surface area increases.

2.3 Dependence on load history and environment

The fracture toughness of optical glass depends on the load history and the environmental
surrounding. The higher the rate of stress increase is the higher is the bending strength of
the glass part.
Environmental media influence the microcrack growth under long-term tensile stress loads.
Liquid water enhances the crack growth and leads to lower strengths values consequently,
whereas glass within vacuum exhibits the highest strength values.
Humid environment without stress loads are not detrimental to the strength of glass objects.
On the contrary they are even beneficial to a certain extend.

2.4 Recommendations to maintain the glass strength

Cleaning procedures should use liquids in abundance and soft clothes. Avoid rubbing. Check
the glass part for scratches. Scratches reduce the strength significantly!
Frames must not exert forces on glass parts. Avoid point contacts and direct metal contacts.
Adhesive joints should use thick layers of soft glues, in order to compensate shear stresses.

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 3/10

3. Calculation procedure
3.1 Mathematical model
The commonly applied mathematical model, to fit the results of strength tests on glass parts,
is the Weibull distribution with two parameters.

(
F(σ ) = 1 − exp − (σ / σ0 )
λ
) (3-1)

with:
F(σ) - Probability of failure at bending stress σ
σ0 - Characteristic strength (F(σ0) = 63,21 %)
λ - Weibull modulus (slope of the Weibull straight line and a measure for the scatter of the
distribution.)
This distribution function is widely used in fracture statistics and allows deriving predictions
on the failure rates for collectives of identical parts. Basing on laboratory test results obtained
under well-defined conditions one can calculate design strengths for loads and conditions
posed by special application requirements.

3.2 Outline of the calculation procedure

The calculation procedure described in the following is a simplified excerpt of the procedure
that has been published in [1]. The notation used in this document is adopted from this
publication.
From the evaluation of strength test results one obtains σ0(SL,63%,R) and λ for a definite
glass type and surface condition.
σ0(SL,63%,R) is the characteristic strength for the laboratory test surface area SL and stress
increase rate R.
In the model employed the design strength depends on the parameters:
SV - area of the tensile stress loaded surface
RV - stress increase rate or
tV - stress load duration time (when constant)
FV - admissible probability of failure
The design strength is derived from the laboratory strength by dividing it by a factor fFOS,
described in the following chapter.
σk = σ0 / fFOS (3-2)

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 4/10

3.3 Stress factors for safety of design

The factor fFOS is factorized once again:
fFOS = fA ⋅ fP ⋅ fF (3-3)

The formulae for the individual factors are derived on the basis of the Weibull model using
the laws on probability.

3.3.1 Area factor fA

The area factor fA is calculated according to the following formula
fA = (S V / SL )1 / λ (3-4)
This formula assumes constant stress within the loaded area. It is a conservative approach
since in many cases the tensile stresses have a maximum amount and fall off with increasing
distances to that maximum. For a more rigid calculation one has to use Seff,V which is
obtained by weighing the maximum stress with the stress distribution function instead of SV.

3.3.2 Probability factor fp

The formula for the probability factor fp is
1
fp = 1/ λ
(3-5)
⎛ 1 ⎞
⎜⎜ ln ⎟⎟
⎝ 1 − FV ⎠

3.3.3 Fatigue factor fF

The general formula for the calculation of the fatigue factor fF is
1/ n
⎛t ⎞
fF = ⎜⎜ eff , V ⎟
⎟ (3-6)
⎝ t eff ,L ⎠
where
teff,V - effective loading time for the application
teff,L - effective loading time at laboratory
n - environmental stress corrosion constant (for most optical glasses it is in the range from
12 to 20)
In the special case for a stress load constant in time this formula reads explicitly
fF = ( t V ⋅ fA ⋅ fP ⋅ R ⋅ (n + 1) / σ0 )1 / n (3-7)

For application cases with time varying loads teff,V has to be calculated instead of tV by using
a weighing function that describes the load variation with time.

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 5/10

4. Bending strength of ZERODUR®

The following table and diagrams show measurement results for the characteristic strength,
Weibull factor and stress corrosion constant n for ZERODUR®. The test procedure was
according to DIN 52292-1 (ring-on-ring method) with a stress increasing rate of 2 MPa/s,
room climate and a test area of 113 mm2 (R30/6) (and 254 mm2 for R45/9-experiments in
2008, marked by an asterisk in the tab.).
Material Surface Characteristic Weibull Stress corrosion
condition strength σ0 factor λ constant n /
[MPa] Medium
ZERODUR® SiC 600 108.0 16.0 51.7 Air 50% [2]
®
ZERODUR SiC 320 71.3 12.4 59.2 Air 50% [3]
®
ZERODUR SiC 230 57.5 15.7 30.7 Water [3]
ZERODUR® SiC 100 53.6 18.7
ZERODUR® D 15 A 130.6 10.6 (for all
ZERODUR® D 35 78.7 15.7 surfaces)
®
ZERODUR D 64 62.8* 11.0*
®
ZERODUR D 151 54.1* 28.2*
ZERODUR® D 251 48.8 11.1
ZERODUR® Opt. polish 293.8 5.3
ZERODUR® D 64 etched 391.4* 3.0*
ZERODUR® D 151 etched 497* 3.4*
®
ZERODUR SiC 600 at 77K 192.7 10.7
glassy SiC 320 66.6 16.7
ZERODUR®

Table 1: Characteristic strength, Weibull factor and stress corrosion factor of ZERODUR®
*= measurements from 2008 [5]. For more information on the surface condition data please
refer to table 4 in the appendix.

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 6/10

Figure 1: SCHOTT ZERODUR® D151 ground and etched samples together with D64
ground and etched samples [5].

Figure 1 shows the comparison of the fracture probability of samples with D64 and D151
ground surfaces. The D64 distribution has a significantly higher characteristic strength of
62.8 MPa than that of D151 with 54.1 MPa. However, the smaller slope of the D64
distribution leads to a crossover at the failure rate of 10 %. That means for lower target
failure rates than 10 %, as will be requested in most application cases, for strength purposes
it is not worthwhile to grind surfaces with D64 instead of D151.
Additional etching shifts the curves to higher strengths, as can be seen in the right part of
figure 1. Results of D64 samples are shown, where 91 µm are removed by etching.
Additionally, the result of D151 samples with a ground surface etched to 181 µm depth which
is about 50 % higher than the maximum micro crack depths for D151 surfaces of about 120
µm. The D151 etched distribution (characteristic strength: 497 MPa / slope: 3.4) lies higher
than that of D64 ground and etched surfaces. Obviously the grain size of the preceding
grinding process does not play any decisive role anymore, when the etched off layer
thickness is significantly higher than the subsurface crack depths. This result is also of
practical significance. It means that it is not necessary to prepare surfaces with fine grinding
which are to be etched later on as long as a minimum layer thickness is etched off, which is
higher than the maximum crack depth.
It is noticeable, that both distributions for the samples with the ratio of 1.5 between the etch
depth have similar slopes. On the other hand, they have significantly different characteristic
strengths.

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 7/10

The results of the D151 measurement can be extrapolated to application conditions. The
general procedure and methods are described in [4] and [6]. In [4] it was shown that a
homogeneous isotropic load of 41 MPa (stress rate 2 MPa/s) on a 1 m2 ZERODUR® surface
(without any pre-damage), ground with bound diamond grain tool D151, leads to a failure
probability of less than 0.1%, 95 % confidence level range. This stress level is 4 times higher
than the conservative value of 10 MPa for a moderate surface treatment.
It can be generally stated that etching or optical polishing of ground surfaces significantly
improves the characteristic strength of the glass. Etching broadens the strength probability
distribution thus leading to a smaller slope in the cumulative failure probability plot. These
statements are only valid if the surfaces are not damaged by handling etc.

5. Bending strength of optical glass

The following table shows some measurement results for the characteristic strength, Weibull
factor and stress corrosion constant n for selected optical glass types. The test procedure
was according to DIN 52292-1 Double ring method R 30-6 with a stress increasing rate of 2
MPa/s, room climate and a test area of 113 mm2.

Material Surface Characteristic Weibull

condition strength σ0 factor λ
[MPa]
N-BK 7 SiC 600 70.6 30.4
N-BK 7 D 64 50.3 13.3
N-BK 7 D 64 etched 234.7 4.1
N-ZK 7 SiC 600 68.9 14.1
N-BaK 1 SiC 600 58.9 8.2
N-SK 16 SiC 600 62.3 19.3
N-LaK 8 SiC 600 70.0 29.9
F2 SiC 600 57.1 25.0
N-BaSF 64 SiC 600 70.1 23.9
N-LaF 21 SiC 600 75.9 28.6
SF5 SiC 600 55.3 10.6
SF6 SiC 600 49.2 5.4
SF 57 SiC 600 39.1 14.7
KzFS N4 SiC 600 49.4 25.5
UG 11 SiC 600 63.4 4.4
KG 3 SiC 600 61.4 11.3
Table 2: Characteristic strength and Weibull factor of optical glass.

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 8/10

6. Fracture toughness of glass

So far the strength of optical glass was considered from a statistical point of view by
measuring the bending strength of samples and subsequently estimating the failure
probabilities. Looking at a single flaw in a material the maximum bending strength depends
on the size of the flaw and geometry in the material. For example, for of a flaw with a short
depth in a thick plate with tensile forces acting normal to the crack plane ( so called “Mode I”
or “tensile mode” or “crack opening mode”) one can define a stress intensity factor KI by

K I = Yσ0 a ≈ 2σ0 a (6-1)

with σ0 being the nominal stress perpendicular to the crack plane and a the depth of the flaw.
Y is a geometry related factor. A flaw will result in a fracture if KI reaches a critical value.
K I ≥ K IC (6-2)

KIC is the fracture toughness for fracture mode I (tensile forces normal to the crack plane,
crack propagation perpendicular to the forces). KIC is a material constant. For glasses without
additional strengthening the value is typically ≤ 1. Table 3 gives fracture toughness values of
some glasses:
Glass KIC [ MPa m1/2]
N-BK7 1.1
F5 0.9
ZERODUR® 0.9
SF6 0.7
Table 3: Fracture toughness values of some glasses [9,10]
In principle, therefore, if one knows the crack geometries of all of the cracks in a part and the
fracture toughness of the material, then one can predict the strength of that part. As a
practical matter, it is not yet possible to know the crack geometries of all of the cracks in a
part exactly enough in order to make reliable predictions of the strength of the part [11].
For a given nominal stress the plate will break for a critical crack depth ac of
2
⎛K ⎞
ac ≈ ⎜⎜ IC ⎟⎟ (6-3)
⎝ 2σ 0 ⎠
Numerical example: For the characteristic strength of ZERODUR® of samples with D64
surface condition σ0 ≈ 64 MPa (table 1) and ZERODUR® KIC ≈ 0.9 MPa m1/2 the critical flaw
size ac is approx. 49 µm. This flaw size compares to the grain sizes for D64 bonded diamond
grains (table 4).

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 9/10

All oxidic glasses and ceramics subjected to tensile stress at their surfaces in a humid
environment exhibit a slow crack growth as long as the stress intensity factor lies well below
the critical value. The velocity of the crack growth can be approximated by [12]:
n
da ⎛ K (a ) ⎞
= A * ⎜⎜ I ⎟⎟ (6-4)
dt ⎝ K IC ⎠
a denotes the depth of the crack, A is a material specific constant and n is the stress
corrosion constant depending on the material and environmental humidity. As a rule, the
drier the environment is, the higher is n. For glass and glass ceramic materials it is known
that in humid environments without stress load the strength gets even higher to a certain
extent, due to a healing effect (rounding of the microcrack tips due to chemical corrosion).
Typical values for the stress corrosion constant can be found in table 1 and 2. Typically the
sub critical crack grow can start from 0,25*KIC.

7. Appendix: Grain Sizes

Designation Mean size [µm] Max. size [µm] ASTM equiv.
Bonded D251 231 250 60/70
diamond grains
D151 138 150 100/120
acc. DIN 848
D107 98 106 140/170
D64 58 63 230/270
D35 36 40 --
D15A 12.5 15 --
Loose silicon- SiC 100 116 149 --
carbide grains
SiC 230 53 84 --
acc. FEPA Std.
SiC 320 29 49 --
42-D-1984
SiC 600 9 19 --
Table 4: Grain sizes of grinding tools

®
TIE-33: Design strength of optical glass and ZERODUR
.
.
.
.
. DATE January 2009
. PAGE 10/10

9. Literature
[1] G. Exner, SCHOTT Glaswerke Mainz, in Glastechnische Berichte 56 (1983) Nr. 11,
p. 299 - 312
[2] H.Richter, G.Kleer Report V24/83 1983, Fraunhofer Institut für Werkstoffmechanik,
Freiburg
[3] M.J.Viens NASA Technical memorandum 4185 Washington 1990
[4] P. Hartmann, K. Nattermann, Th. Döhring, M. Kuhr, P. Thomas, G. Kling, P. Gath, S.
Lucarelli: “Strength Aspects for the Design of ZERODUR® Glass Ceramics Structures,”
Proc. of SPIE Vol. 6666, (2007)
[5] Hartmann, P., Nattermann, K., Döhring, T., Kuhr, M., Thomas, P., Kling, G., Gath, P.,
Lucarelli, S.: “ZERODUR® Glass Ceramics – Strength Data for the Design of
Structures with High Mechanical Stresses,” Proc. of SPIE Vol. 7018 (2008)
[6] K. Nattermann, P. Hartmann, G. Kling, S. Lucarelli, B. Messerschmidt: “ZERODUR®
Glass Ceramics Design of Structures with High Mechanical Stresses ,” Proc. of SPIE
Vol. 7018 (2008)
[7] F. Kerkhof, H.Richter, D.Stahn Glastech. Berichte 54 (1981) No. 8 p. 265 - 277
[8] G. Exner, O.Lindig Glastech. Berichte 55 (1982) No. 5 p. 107 - 117
[9] The properties of optical glass; H. Bach & N. Neuroth (Editors), Springer Verlag 1998
[10] Viens, Michael J.: “Fracture toughness and crack growth parameters of ZERODUR®”,
NASA technical memo A969903 (1990)
[11] J. Varner: “Strength and Fracture Mechanics of Glass”, ICG Advanced Course 2006,
Strength of Glass, Basics and Test Procedures (2006)
[12] S.M. Wiederhorn, L.H. Bolz: “Stress corrosion and static fatique of glass”, J. Am.
Ceram. Soc. 53, 543-548 (1970)

For more information please contact:

Advanced Optics
SCHOTT AG
Germany
Phone: + 49 (0) 6131/66-1812
Fax: + 49 (0) 3641/2888-9047
E-mail: [email protected]
www.schott.com/advanced_optics

®
TIE-33: Design strength of optical glass and ZERODUR