Heat Treatment of Cold Extruded Polycarbonate: Some Implications for Design Engineers

213

Materials Science and Engineering, 18 (1975) 213--220 © Elsevier Sequoia S.A., Lausanne -- Printed in The Netherlands

C.S. LEE, R.M. CADDELL and A.G. ATKINS

Department of Mechanical Engineering, The University of Michigan, Ann Arbor, Mich. (U.S.A.)

(Received in revised form August 6, 1974)

SUMMARY tensile strength as compared with the virgin material; the same cannot be said about yield Polycarbonate was subjected to various stress. A detailed discussion regarding these combinations of mechanical - thermal histories two properties is given in the Appendix since to investigate such effects on subsequent ten- there appears to be confusion in the published sile mechanical properties. This was accom- literature regarding the meaning of these terms. plished by "cold extruding" the material ini- From a design viewpoint, the yield stress tially; nominal "reductions in area" of 18, 40 of a solid is usually of greater concern than is and 64% were used. Cold extruded bars were the tensile strength which merely defines the then heat treated at three temperature levels, maximum load carrying capacity. Thus, studies all being less than T z (150°C) of polycarbonate. which concentrate on tensile strength measure- As compared with the material that was only ments alone may not be of paramount interest cold extruded, it was found that in general, to the design engineer. heat treating tends to raise the yield stress The results reported in this paper suggest while lowering the tensile strength, elastic that the mechanical behavior of cold formed modulus and stress at fracture. The results PC, subjected to subsequent heat treatment, suggest that a desired combination of proper- may be altered to produce a combination of ties may be obtainable by the use of a cold properties most useful to the design engineer. work - heat treating sequence. It appears that the yield stress of the cold extruded material (extrudate) may be increased without a noticeable decrease in tensile strength. INTRODUCTION This suggests a relatively unexplored method for altering the mechanical properties of The influence of cold forming upon the polymers. subsequent mechanical behavior of various polymers has received the attention of several investigators. Cold rolling of sheet material EXPERIMENTAL PROCEDURE was used in some studies [1 - 4] as was cold extrusion [ 5,6]. Cold drawing, in its historical All test specimens were produced from a context in metal forming, has also been re- single bar of PC; it was 19.1 mm (0.750 inch) ported [6]. Increases in tensile strength of in diameter and obtained from a commercial cold rolled polymers have been reported [7] source. and biaxial cold rolling improved the deep A tensile specimen of 6.35 mm (0.250 inch) drawability of polymers [8]. Perhaps because diameter and having a 50.8 mm (2 inch) gage of its attractive combination of properties length was machined from the bar so as to (strength and ductility), polycarbonate (PC) determine the properties of interest of the "as- has been used in many of these studies. received" material. Solid cylinders of 17.1 mm When the nominal amount of "cold working" (0.675 inch) diameter and 76.2 mm (3 inches) exceeds 15 to 20%, PC displays an increase in long were machined from the supply bar pre- 214

histories were loaded at a crosshead speed of /_STEEL ROD 8.33/~m/sec (0.05 cm/min) to produce thir- teen sets of load - extension data. During the early portion of each test, an Instron extenso- meter was employed to sense length changes and to drive the recording device. In those tests where a localized neck formed and even- V / / //% !// / / //I tually stabilized, concurrent measurements EXTRUDED "-'-I Ir"~ of load and neck diameter were monitored to POLY- "~ ~--DIE OUTLET rSEE TABLE i ] CARBONATE [FOR BIAMETERSJ provide continuing information. All diameter measurements were obtained with a pair of Fig. 1. Schematic illustration of extruding operation. point micrometers. Each set of load - extension (or diameter) paratory to being cold extruded. By using data was converted to true stress - true strain three dies of varying outlet diameters, three information using the standard definitions "nominal" reductions were available; these that were about 18, 40 and 64% respectively in terms of percent reduction in area. Figure 1 a - ~ and e = In = 2 In , (1) is a schematic version of the extrusion operation. Four extruded specimens were produced where L and A (or D) correspond to instanta- for each reduction, with one specimen per re- neous values of load and area (diameter) and duction used to provide a tensile specimen as Ao is the original area prior to loading. Every described earlier. The remaining three specimens test was carried to fracture. With specimens per reduction were heated for two hours at that displayed a stable neck, fracture always temperatures of 100 °, 117 ° and 140°C respec- occurred when the neck had fully propagated tively; they were then air cooled. Since the to the shoulders. Those that showed no ten- glass transition temperature (Tg) of PC is dency towards localized necking usually about 150°C, the highest temperature used fractured away from the shoulders. had to be less than Tg to prevent all effects of the prior cold extrusion from being erased. This led to an arbitrary choice of 140°C; the EXPERIMENTAL RESULTS AND DISCUSSION other two temperatures were chosen so as to give a spread of possible heat treatment effects. 1. Dimensional changes Following the heat treatment, each of these Table 1A contains information about the nine specimens was machined to produce a dimensions of specimens at various stages tensile specimen similar to that described prior to being machined for tensile tests. The above. Using an Instron machine, the tensile changes are expressed both in terms of dia- specimens of different mechanical - thermal meter and percent reduction in area from the

*TABLE IB ~rcentrecovery ofdeformation ~rdifferent m~hanical-~ermM ~eatments

Diam. of % Recovery at % Recovery after heat treatment extrudate room temp. (inches) 100°C 117°C 140°C 3min 48 hou~

0.610 23.8 26.8 + 38.9 41.9 57.3 0.524 12.9 15.4 22.7 25,9 41.9 0.407 20.4 21.2 25.0 27.0 40.9

* A starting of 0.675 inch (area of 0.3578 in 2) was used in all calculations. + Sample calculation (subscripts o,e,r refer to original, extruded and recovered sizes) D o = 0.675 inch, A o = 0.3578 in 2 Area reduced = 0.3578 -- 0.2922 = 0.0656 in.2 D e = 0.610 inch, A e = 0.2922 in.2 Area of recovery = 0.3177 -- 0.2922 = 0.0255 in. 2 D r = 0.636 inch, A r = 0.3177 in 2 % Recovery = 0.0255/0.0656 (100) = 38.9. 215

• o • .~1C0

• o °

~D u~ cO

r~ C~1 w,

• • ...... ° o , ° L~. i ,-I o~ o~ O ~J

N -d

• o o

~D Cq 00 CO ~D 00

OOO

0 C~ @

= 000~D C o~ ..= oC "0 ca o,..~ ~.~

0 o O r.D t~. ~J o~ , ~

o~ ~J o~ ~D~D~ o. e~ odd 0 .,-, • ~ o ~ ..4 "0 ~c~c~ 0

~0 o 0 o0 L~- r.D oo o~ c~ N O o o o o o o ~

~ O I> tl- o 0 + od~ O

~e4d ~+ + tl. ,It + 216

pre-extruded diameter. As may be noted, the specimens show significant relaxation imme- .g 5O diately after being extruded. Although a ~-o ;0 greater recovery in the absolute diameter ac- t[+ companied the higher degree of initial cold ~0 o • oea work no consistent trend pertained in regard 0o to the percent recovery of deformation. Ad- 0• 8o # ocw eee • Bo~ @ ditional recovery may be seen for the heat co I0 o3 treated specimens, with the degree of recovery 50 ~ I- ~°ee°° • i : NOTREATMENT HEAT I-- correlating directly with temperature of heat 100"C--2 HR8 treatment. Table 1B contains the information o 117 'C-2HRS in terms of percent recovery of deformation. 140"0 --2 HRS, % o.; o.'2 or3 o.'4 a'5 0/6 o:7 2 2. Tensile true stress - true strain behavior of TRUE STRAIN non-heat treated specimens Fig. 3. Tensile true stress - true strain curves of PC Figure 2 shows the influence of the degree cold worked 18% by extrusion then subjected to va- of cold extrusion on the tensile behavior of rious thermal treatments. PC for the reductions used in this study. It may be noted that the observed behavior is creases the elastic modulus and decreases the quite similar to that for cold rolled PC where fracture strain in a consistent manner whereas similar levels of "cold working" are used. the true stress at fracture first increases but Values of tensile strength, yield stress based then begins to decrease with ever increasing upon a 1% offset, elastic modulus and the amounts of cold work. Tensile strength increases true stress and strain at fracture are listed in with cold work Table 2. Both English and SI units are included where applicable. It may be seen that small 3. Tensile true stress - true strain behavior of amounts of cold working (here 18%) lead to heat treated specimens a substantial decrease in the 1% offset yield Figures 3 - 5 show the influence of heat stress as compared with the material in the treating the extrudates that had experienced "as-received" condition (i.e. no cold work). 18, 40 and 64% cold work respectively. In all With increasing levels of induced cold working, cases, heat treating raises the yield stress however, the yield stress increases but never while lowering the tensile strength. Reference exceeds that of the initial condition even to Table 2 displays the findings more explicity. when the percent cold work was as high as i i the 64% used in this study. Cold working in- 200

I 30 2.00

: x % zo~ ,..' " °##

x 20 ¢5 o9 03 L,I # e e rf I00 I- if) t.O i I • ~aAA wx ~x • IOOuJ x ~e • ~J if) x~ I0)#'J ~t,~` tO #•~ UA I- ~2 :@ ~ NO HEAT x rY I I TREATMEN I F u tO0 C-2HRS NOMI NAL 18~ 96 F II7 C-2HRS A 40~, COLDWORK 1_~ 140C-2HR8 t m 64.~ O0 O.II 0.'2 0.'3 0.~4 O.J5 160 % o'.1 o,'z ot~ 0.'4 o.'s ors oI~ so TRUE STRAI N TRUE STRAI N Fig. 4. Tensile true stress - true strain curves of PC Fig. 2. Tensile true stress - true strain curves of PC for cold worked 40% by extrusion then subjected to va- four levels of nominal per cent cold work by extrusion. rious thermal treatments. 217

CONCLUSIONS

Within the limited range of mechanical - 150 thermal treatments used in this exploratory % 20 i o~ o o study, there is little doubt that the mechanical ~o behavior of PC can be altered to produce de- d ° O9 d finite and controllable changes in yield stress, LD o ° o) Q/ I00 ,,I [1! stress and strain at fracture, the tendency to CO (D avoid local necking, and tensile strength. Al-

LD tJ though the trends may differ, similar treat- D ,0,-p r~ ments have been used for years to control me- NO HEAT 50 I-- !u [-o TREATMENT chanical properties of metallic solids. It is [-u I OO'C--2 HRS. obvious that more extensive studies must be I Lo I 17°C-2HRS. pursued before the observations presented o d.1 d2 ors 0.4 might be considered as showing typical behavior for other polymers; we are currently TRUE STRAIN pursuing such studies. Fig. 5. Tensile true stress - true strain curves of PC cold worked 64% by extrusion then subjected to various thermal treatments. APPENDIX

Heat treating in general also tends to lower Engineers have traditionally differentiated the elastic modulus and the true stress at between the yield stress and ultimate tensile fracture compared with the non-heat treated strength in metals but many workers in poly- extrudate. No general pattern was observed mers confuse these terms and a designer may regarding the influence of heat treament on be caught unawares when looking up polymer the true strain at fracture. strength data from the literature. The common definitions of yield strength 4. Neck formation (Sy) and tensile strength (Su) of ductile me- For any specimen that displayed a sudden tals are illustrated in Fig. 6, with reference tendency to form a sharp localized neck, it to a load/extension {or nominal stress - no- was not possible to transfer from extenso- minal strain) curve for an annealed low carbon meter to micrometer readings in a manner steel. For all practicalpurposes, 'yield strength', that permitted a continuous monitoring of 'elastic limit' and 'limit of proportionality' are strain information. This is reflected on the given by the same stress Sy, where the associated stress - strain curves by a fairly large jump in strain is most often less than 0.2% or so. strain before a smooth continuation of the There is essentially no change in cross-sectional curve proceeds. On Fig. 2 this phenomenon area between O and Y, so that nominal stress may be associated with the curve describing and 'true' stress are the same at yield. the "as-received" material; note that there are The maximum load point U, at which an no points between a strain of about 0.05 and unstable neck initiates, gives the ultimate of 0.50. Similar observations pertain to Figs. tensile strength (or tensile strength) Su. The 3 and 4 with all the specimens that were heat associated strain is quite considerable, say treated. 20%, and there is a marked but uniform re- It would seem reasonable to conclude that duction in cross-sectional area between Y and heat treating the initially cold worked speci- U; the true stress at ultimate is thus greater mens tends to alter the structure in a menner than Su. The unstable neck at U always leads that approaches the "as-received" material to fracture in the region where this neck first and, therefore, one would expect a reversion develops. to an abrupt and localized neck which is char- The flat portion of the curve at Y is really acteristic of the non-cold worked material. a series of ripples in the load trace, with the Thus, in operations where PC is used and lo- associated propagation of Liiders bands. calized necking is detrimental, the use of a Depending on the material and testing arrange- particular mechanical - heat treating history ment, a stress spike Can sometimes occur at Y, will prove desirable. giving an upper and lower yield point. For 218 rE,,3o~ U _ Su

fracture

~Z~ o.o1 Sy ~,e eu 0.001'

e-O

-0 0 E~ 400,000 psi o II ..4

• e....~~ | M~"~" \ o.ol offset " ""' V//I ,/.~ -.~oo4~o.oo2 offset P extension or nominal strain Fig. 6. Nominal stress - strain curves for a polymer that displays stable neck propagation ("cold drawing") as compared with an annealed plain low carbon steel which exhibits a pronounced yield point. Su

II / ,racur.

-6 i I. ~~0.002 orrser ./" ." ~ fracture .=_ £ e-0

"0 _oa I ~ "-..o010ffs.t /~~ ~o.oo2 offset

0 extension or nominal strain Fig. 7. Nominal stress - strain curve for a polymer that does not display localized necking (i.e. does not "cold draw") as compared with a ductile metal which does not exhibit a pronounced yle]d point. 219 design purposes, Sy as shown in Fig. 6 is usual- elongation in tensile tests of metals is likewise ly the important strength parameter. meaningless without reference to the starting Other metals display nominal stress -strain gage length). It is most sensible beyond N to curves as shown in Fig. 7, where there is a measure the strain in terms of 'percentage re- gradual transition from elastic to plastic be- duction of area' (Ao--A)/Ao, or true strain, havior. The flat region at Y of Fig. 6 is not in e = In (Ao/A) = 2 In (Do/D). Then during most evidence and the yield point is defined by an of the stable neck propagation, changes in true offset (proof) method. The associated yield stress and true strain are minor until the neck strain is still of the order of 0.2% or so. has propagated to the shoulders. It is worth The deformation behavior of ductile poly- noting that, depending on the 'suddenness' of mers is characterized by much larger strains the initial neck formation, it may be very dif- than ductile metals, and the Young's moduli ficult experimentally to obtain points between ate also much lower. Many polymers also dis- N and P. This is pointed out in this paper in the play the feature of what is called "cold draw- section entitled "Neck formation". ing", which mechanically is the propagation The region NP and its growth of one stable of a stable neck along the tensile testpiece. neck corresponds with the region YZ on the A typical load - extension (or nominal stress - low carbon steel diagram where there are strain) plot for a polymer, such as polyethylene many local maxima and minima in the load, that 'draws', is shown in the lower half of corresponding to the propagation of many Fig. 6. The precise behavior can be affected Liiders bands. For a detailed explanation of by viscoelasticity and anelasticity, but for mechanical instability the reader may con- practical engineering design, the following sult Vincent [10] or McClintock and Argon description is adequate*. [11]. There is a departure from linearity at M, Polymers which do not 'draw' have continu- (where the strain may be some 1%), and the ously rising load - extension (nominal stress - load curve rises to a local maximum at N strain curves}, similar to Fig. 7 for metals, but (where the strain may be 10%) at which point the strains are larger and the moduli smaller. the stable neck initiates. The load then falls The deformations and strains at N and P in as the neck reduces in cross-sectional area, ductile polymers which draw are much greater until stability is reached and the neck propa- than those at Y and Z. Also the range of strain gates along the testpiece at the essentially between M and N is extremely large compared constant load P. Subsequently, after the neck with the corresponding region for ductile has propagated the length of the test bar, the metals which occur near Y. The point at load increases again and fracture eventually which permanent deformation sets in is some- ensues under rising load, akin to failure in where between M and N. The curvature over brittle materials. The particular polymer a large range of strains between M and N and fracture load may be less than or greater than the local maximum in load at N has led some the load at N, depending on the interaction workers to call N the ultimate tensile strength of the geometry of the propagated neck and for polymers. Reference to Fig. 6 shows that the original shoulders of the testpiece. this is erroneous, if we mean that ultimate After the local load maximum at N, mea- tensile strength is followed by an unstable surements and definitions of strain in terms neck and fracture as is the case with ductile of length, such as the nominal strain given by metals*. Again, it is clear that N and P do not (l--lo)/lo or the 'draw ratio' l/lo, are quite represent upper and lower yield points in the meaningless because the reduction is non-uni- traditional sense. form. Clearly, any value is possible depending Other workers identify N as the local yield on the reference gage length (the percentage point, and the similarity of events between N and P, and Y and Z, makes such an approach justified. However, the strains at N are much * As strain rates are increased, the tendency to "draw" bigger than at Y and from a design point of is decreased. Also, the load - extension (nominal stress - strain) behavior shown in Fig. 6 does not mean that the true stress - true strain behavior will * A point such as N may, however, indicate the maxi- necessarily display a drop after point N (see, e.g., mum load and in that context the use of "tensile references 3 and 9). strength" is consistent with a long standing definition. 220 view, M may be more meaningful as a limiting 3 R.M. Caddell,T. Bates, Jr. and G.S.Y. Yeh, Mater. strain; even then the deformations are greater Sci. Eng., 9 (1972) 223. than traditional metal yield strains. 4 P.H. Rothschild and B. Maxwell, J. Appl. Polymer Sci., V, 16 (1961) 311. The principal objection to using N as a 5 A. Buckley and H.A. Long, Polymer Eng. Sci.,9 yield value, however, relates to those load - (1969) 664. extension curves, as shown in Fig. 7, that 6 C.S. Lee, R.M. Caddell and G.S.Y. Yeh, Mater. show no load maximum. It is impossible to have Sci. Eng., 10 (1972) 241. 7 L.J. Broutman and S. Kalpakjian, Soc. Plastics a consistent definition of yield unless one re- Engrs. J., 25 (1969) 46. sorts to an offset method. Though the offset 8 H.L. Li, P.J. Koch, D.C. Prevorsek and H.J. Os- used is arbitrary, it does have the merit of wald, J. Macromol. Sci., B4 (3) (1970) 687. consistency once a particular value is chosen. 9 R. Raghava, R.M. Caddell, L. Buege and A.G. At- kins, J. Macromol. Sci. B6(4) (1972) 655. 10 P.I. Vincent, Polymer, 1 (1960) 7. 11 F.A. McClintock and A.S. Argon (eds.), Mechanic- REFERENCES al Behavior of Materials, Addison-Wesley, 1966, p. 311. 1 G. Gruenwald, Mod. Plastics, 38 (1960) 137. 2 L.J. Broutman and R.S. Patil, Soc. Plastics Engrs., 28th Ann. Tech. Conf., New York, 1970, p. 20.