I/EC C O N D E N S A T E S
Cast cylindrical specimens were usedl1/2 inch in length, with a slenderness ratio of 12. On one occasion, a series of la/, by 41/2 inch cylinders having a slenderness ratio of 13 were tested; Although these also met the ASTM standard for stress-strain measurement, they gave consistently lower values for yield strength (ASTM Method, Note 4). Effect of cure cycle and iron oxide concentration on yield strength and compressive modulus was measured a t 77' and 250' F. Cure cycle is more important than filler loading which has the same effect a t both temperatures. Depending on cure, yield strength for Epon 828 filled with 40% of iron oxide, ranged from about 1000 to 16,000 pounds per square inch and modulus from 0.1 X 10-6 to 1.4 X 10-6 pound. Compressive fatigue was measured
'/z inch diameter and
Effect of Fillers in Epoxy Resins
4
B U L K FILLERS as reinforcing additives for rigid resins have been investigated, but apparently no thorough study has been made. Such information about epoxy resins is increasingly needed; they have a remarkable tolerance for fillers of nearly every variety, and also ' growth of resin tooling demands more knowledge about the effects of random fillers on cast resins. Consequently, the general effect of type and amount of filler on cured Epon 828 was studied. Thermal conductivity is not directly proportional to filler concentration, nor has it an obvious relationship to conductivity of the filler itself. For example, pure cast aluminum has a conductivity of about 100 B.t.u./hour sq. foot/foot/O F. but an equal mixture of aluminum powder and cured resin having a conductivity of 0.1 yields a product having a conductivity of only 1. Similar results were obtained for copper. and to a lesser extent for iron. O n the other hand, alumina imparted roughly half and silica about twice its own conductivity. Thus, using aluminum powder as a filler for heated plastic tools is questionable. I t has only marginal advantage over its oxide and mechanical strength it imparts is markedly inferior. Thermal expansion coefficients are lowered by adding fillers, but for this coefficient, filler-resin-product relation-
ships are complex and conflicting data have been reported. Tests with black iron oxide showed that it does not follow a linear relationship with either weight or volume concentration of filler. Compressive strength is decreased by fillers, but their stiffening effect increases modulus and compressive yield strength. For alumina and black iron oxide, both these properties were increased, most markedly after 30% filler concentration. For aluminum powder, modulus increased similarly, but yield strength decreased strikingly-nearly
50%.
20
-
'0 0
VOLUME%
O
l 20
i
L
U
60
40
70Iron
u
80
Oxide
Effect of filler content on coefficient of expansion System of Epon 828, MD- 10 1 iron oxide, 2 0 phr curing agent Z
1.4~10~
Cured for 2 hours at 80'
C.
+ 2 hours at 150' C.
1.2x106 A. Tabular aluminum oxide T-61; B. Black Iron oxide MD-101; C. Atomized aluminum M D - I O 1
* 1.0x106
3
P v
-aa VI
-0 0
E
0.8XlO~
.-> V
VI v)
2
E, 0
0.6x1Og
0.4X10'
: 20 40 60 Wt.% Filler
AEffect of filler on compressive modulus of Epon 828 Cured for 20 hours at 25' C.
,Effect
-/- 2 hours ot 100'
C. with 8 phr DTA
of flller concentration on compressive strength of Epon 8 2 8 Cured for 20 hours a t 25' C.
+ 2 hours at 100' C. with 8 phr DETA
20 Wt.
40 yo Filler
VOL. 49, NO. 7
60 JULY 1957
1103
.-n
20,000
r-
16,000
-
12,000
-
n w
=
Tested at 77' cure: 24/65' C.; 8. Tested at 77' F., cure: 2 4 1 65' C.,2/15Oo Cf C. Tested a t 2 5 0 F., cure: 2.46 65' C., 21150 C.; D. Tested at 250' F., cure: 2 4 1 65' C.
F.,
c
3
s
I
A.
M n
-=
I
3
Wt .yoFiller AEffect of filler concentration and cure temperature on compressive strengths of Epon 828-curing agent Z
0
20 Wt.%
40 Iron Oxide
on the IS/*-inchcast cylinders mounted in a modified Danley die set in turn mounted in a Cleveland crank press (Kish Industries). Fillers lowered resistance to compressive fatigue, even when the load was considerably below compressive yield strength. Powdered aluminum however, was unique; it imparted a definite degree of malleability to a rigid resin. Deformation us. number of cycles increased steadily for an aluminum-filled specimen, while other fillers showed no deformation prior to failure. Tensile strength loss from adding bulk fillers to a rigid polyester resin has been reported by Linzmeyer. This work on Epon 828 using four fillers-iron oxide, aluminum, asbestos, and mica-show similar results. However, data are limited and conclusions are impossible. I t is reasonable to assume that spherically shaped fillers decrease tensile strength in proportion to their concenbration. Fibrous fillers decrease this strength in low concentrations but increase it at higher concentrations where entwining occurs. This decrease in tensile strength is frequently overlooked in designing many reinforced epoxy articles-e.g., glass cloth laminates, where as much filler as possible is commonly added to c!ieapen the article. If strength, not bulk, is required, a thinner laminate without filler should be as strong and also cheaper because of saving in glass cloth. In encapsulation, fillers are frequerrtly used at maximum concentrations consistent with pourability to decrease the COefficient of thermal expansion, and hence, increase resistance to thermal cycling. Since resistance to tensile stresses from thermal cycling is also decreased, a maximum percentage of filler is not necessarily optimum.
1 104
60
80
4 Effect of filler concentration and cure temperature on compressive modulus of Epon 828-curing agent Z
Impact strength was measured by the Notched Izod method, but a special technique was used, which should give a homogeneous film of resin across the entire surface and prevent projecting filler particles from becoming focal points for stresses. Despite this, however, these data are probably not accurate beyond ~ 0 . foot-pound 1 per inch of notch. Neither iron oxide nor sand had a significant effect on impact strength. Glass fiber, however, when added to an Epon-sand mixture, even in small amounts, caused appreciable increase. T h e heat distortion point, about 81 O C. at 30% of filler, for Epon 815 containing 10 parts per 100 of triethylenetetramine and filled with iron oxide, increased to about 87' C. for 75yo of filler. The curve obtained is similar to that which relates modulus to iron oxide concentration. Abrasion resistance was tested on the Taber abraser for Epon 815 filled with graphite, titanium dioxide, nylon, molybdenum disulfide, aluminum, and iron oxide. Graphite at sufficiently high concentrations (1 5% tested) apparently reduces friction. Molybdenum disulfide at this percentage may have the same effect. Viscosity is the property of uncured liquid resins most markedly affected by fillers. Sometimes-e.g., gunk molding -this is of minor importance because pressure is used in fabrication. HOIVever, in all casting operations such as making plastic tools and encapsulating electronic circuits and components, viscosity of the uncured resin limits filler content. Shape of filler particles affects viscosity-spherical shapes least and fibrous forms most. The smaller the particles and the better the mixing, the greater the viscosity. Absolute viscosity is inde-
INDUSTRIAL AND ENGINEERING CHEMISTRY
pendent of filler density but pourability, or kinematic viscosity, is not. At a given volume concentration, the higher the filler density, the better the pourability. Measurements of kinematic and absolute viscosity were largely not reproducible. Method of mixing had an important bearing. Powdered fillers naturally carry entrapped air into the resin and severe milling is needed to break u p the agglomerates and release air. In excessive, tight milling, filler particles may be broken, especially if coarser particles are present. Therefore, only qualitative observations were madee.g., Epon 828 filled with iron oxide has a maximum pourable viscosity a t about 300 parts of filler per 100 parts of resin; for aluminum or its oxide, this ratio becomes 200 and 150, respectively; for graphite and titanium dioxide about 50; for calcium carbonate and silica, about 30; and for asbestos, about 5. Before effects of fillers on cast resins can be understood, many aspects need further investigation. More accurate impact tests must be devised and thermal expansion and conductivity behavior must be explained. Dependence of mechanical strength on filler concentration and particle shape must be studied, and more data on abrasion resistance and coefficient of friction are needed. HARVEY L. PARRY and RALPH W. HEWITT Union Technical Service Laboratory, Shell Chemical Corp., Union, N. J. The complete manuscript may be obtained from Harvey L. Parry at the above address. Division of Paint, Plastics, and Printing Ink Chemistry, Symposium on Epoxy Resins, 130th Meeting, ACS, Atlantic City, hT.J., September 1956.
