JOURNAL OF RESEARCH of the National Bureau of Standards-C. Engineer ing and Instrumentation

Vol. 7lC, No.2, April - June 1967

Polymeric Materials for Dielectric Reference Speci mens

Arnold H. Scott* ~nd Joseph R. Kinard, Jr.

Institute for Basic Standards, National Bureau of Standards, Washington, D.C. 20234

(December 1. 1966)

The result s from a study of the effec ts of aging and humidity c hange on certain polymer specimens indi cate it is poss ible to establi s h s table dielec tri c reference specimens. Specimens of polyethylene, polystyre ne, polycarbonat e, poly(t e trafluoroethylene) (PTFE). fluorinat ed e thylene-propylene polymer (FEP), and poly(l ,4-cyclohexy lenedimeth ylene terephthalate) were used in the investigation. Slow changes jn the di electri c prope rties . we re observed on so me s pecimens . over a period as long as three yea rs. Ve ry long timed humidit y runs indi ca te PTFE, and FEP to only a s li ghtl y less degree, are best s uited for di e lec tri c refe re nce specime n prepara tion.

Key Words : Dielectri c constant, di ssipat ion factor, frequ ency, fluorinated ethylene-propylene, humidity, polycarbonate, polyethylene. poly(I ,4-cyclohexyle nedimethyle ne tereph­thalate), polys tyrene, poly(te trafluoroethylene), time, re fe re nce specimens.

1. Introduction

A long term s tudy of ce rtain polymer spec imens under co nsid eration as dielec tri c refe re nce specime ns has been conduc ted. Such reference specim e ns may be used to co mpare the measure me nt accuracy of variou s other laboratories, so th ese s pecime ns must have accurately determined and stable dielec tri c properties_ Specimens from several polymers were found to exhibit s uffic ient s tability to permit their use_ The materials inves ti gated were polye th yle ne (PE), polystyre ne (PS), polycarbonate (PC), poly(tetrafluoro­ethyle ne) (PTFE), fluorinated ethylene-propylene polymer (FEP), and poly(1 ,4-cyclohexyle nedimethylene terephthalate) (T -16)_ All data were taken on flat disk specimens, and the conditions under which the measurements were made were carefully controlled_ The humidity of the surrounding atmosphere as well as the specimen temperature was controlled because the properties of most specimens were functions of humidity_

2. Measuring Equipment

A low-voltage conjugate Schering bridge similar to one described in a paper by Scott and Harris: [1] I was used to measure dielec tri c constant and di ssipation factor from 102 Hz to 105 Hz_ The cell used with the Sc hering bridge was of the guard-ri ng microme ter­elec trode type [2, 3, 4] and was mounted in a chamber

*Present address: 4220 Franklin S tree t, Ke ns ington, Md . 20795. I Figu res in brac kets indicate the litera ture references a t the e nd of this paper.

whose te mperature was controlled by pumping liquid from a te mpe rature bath through coils around the chamber , and whose humidity was maintained by exposing inside the box either P 20 S for < 1.5 percent relative humidity (rh) or a saturated Mg(N03h solution for about 52 percent relative humidity_ The humidity in the chamber was measured with an electri c hygrome ter. . In th e freque ncy range from 1 MHz to 10 MHz a

Q-meter was used for di elec tri c cons tant and dissipa­tion factor measure me nts. The cell used was a two­terminal mi crom e ter- electrod e cell mounted in a box whose temperature and humidity were controlled in a manner similar to that used for the three-terminal cell. A two-terminal Schering bridge was used to coordinate the Q-meter measurements with the more accurate three-terminal data.

3. Specimens

3_1_ Contact Electrodes

Many of the measureme nts were made without con­tact electrodes u sing the air-gap technique [5] with 8 .5 c m diam dis ks from 2 mm to 5 mm thick_ Disks with 3.8 e m diam were pre pared with contact elec­trodes so that specime ns might be measured at higher freque nc y with the Q-meter and used to calibrate two-terminal equipment. Heavy gold electrodes, about 1500 A thick, were evaporated on the faces of these similar specimens to form the contact electrodes_ Care was always taken to see that the resistance across the electrode surface was never more than a few tenths of an ohm.

119

3.2. Surfacing Specimens

At the beginning of this study a modified Schiefer abrasion machine was used to surface the specimens [6]. Even though this method produced specimens with thicknesses uniform to within 6 p." we later used a technique which produced more uniform thicknesses. The second method made use of a vacuum chuck and a fly cutter on a milling machine. The vacuum chuck consisted of a metal vacuum chamber with holes drilled in one side for holding the specimen in place and could be positioned on a milling machine for successive passes made with the cutter. The specimen was turned over or rotated with each pass until the faces became quite parallel. With careful preparation of the cutting tool and proper adjustment of the cutting rate, this technique could produce specimens with thicknesses uniform to within 3 p., over an 8.5 em diam.

3.3. Measuring Thicknesses

The specimen thicknesses were measured with an indicating micrometer having a reference point set by gage blocks. Measurements were made at points around concentric circles as determined by a pattern. These numbers were averaged to determine the meas­ured thickness. Comparisons indicate agreement to 0.05 percent between the micrometer thickness and that determined with the two-fluid electrical technique [7].

3.4_ Storage

Between measurements and for initial conditioning, specimens were stored in glass vessels with humidity controlled in the same manner as in the measuring chamber.

3.5. General Properties

The suppliers indicated the following general proper­ties for the various materials. The polyethylene was the high molecular weight, low pressure type. It had a linear structure with no detectable branching and a crystallinity of from 90 to 95 percent. The T-16 poly­ester was a linear, thermoplastic, crystallizable poly­mer about 36 repeating units long. It had a crystal melting point of around 265°C and a glass transition at about 87°C. The polycarbonate material con­tained no plasticizers and had a molecular weight ranging around 30,000 to 35,000. The polystyrene material was reported to be of high purity. PTFE and FEP specimens were from laboratory grade stock.

The polyethylene specimens and certain of the poly­styrene specimens were compression molded in our laboratory. The polycarbonate specimens were im­pact molded, the T-16 specimens were injection molded, and the FEP and PTFE specimens were com­pression molded, all by the supplier.

3.6. Density

The density of most specimens was measured in order to characterize the materials better and to pro­vide a rough measure of the uniformity of specimens from the same material. Two methods were used for

determining density. In the first method the volume of the specimen was determined by liquid displace­ment. The second method was based on dimensional volume measurements. Since our specimens were very carefully machined disks, they were, in fact, good approximations of right circular cylinders. By meas­uring the diameter with a traveling microscope and using the thickness, we could calculate a dimensionally determined volume. The same mass data was used with each volume to calculate two different densities. Based on repeated measurements and calibration of the traveling microscope, we estimate the dimensional volume measurement to be accurate within about 0.1 percent; however, the liquid displacement densities were consistently larger than the dimensional densities by 0.1 percent to 0.5 percent, even though care was taken to reduce errors caused by insufficient surface wetting of the specimen and by surface tension acting on the suspension wire. Since the dimensional meas­urements also produced much more reasonable rela­tionships between the dielectric constant and density, it was concluded that our liquid displacement measure­ments were less accurate, so all the density data pre­sented were determined from mass and dimensional measurements.

4. Measurement Procedure

The following procedure was used to obtain the dielectric properties of individual specimens as func­tions of time after humidity change. Each specimen was stored in a chamber with the desired relative humidity until it reached equilibrium, i.e., until its dielectric properties no longer changed with time. A timed humidity run was then begun by transferring the specimen to another chamber at the new humidity. Periodic measurements were made with careful note being taken of the elapsed time since the humidity change. All timed humidity run data as well as fre­quency sweep data were taken at 23°C.

lt was observed for some of the electrodeless speci­mens that slightly different values of dielectric constant and dissipation factor were obtained when the speci­men was turned over. The amount of this effect varied from specimen to specimen, and for some disks only the dielectric constant differed significantly. Differences as large as 0.2 percent in the dielectric constant were observed. This behavior was caused by the specimens being slightly curved instead of flat although they were of uniform thickness. In some cases it was possible to partially flatten out the warped specimens, and this reduced the difference. When­ever possible, the same face was kept turned up throughout the timed humidity runs; however, for some of the older data, taken before the effect was dis­covered, this was not done.

s. Results

5.1. Timed Humidity Runs

Figures 1 through 7 show timed humidity runs meas­ured at 1000 Hz for certain specimens of the various

120

materials. For all of these figures, time from humidity change is plotted on the horizontal axis using a logarith· mic scale, and those points marked with arrow heads on the left-hand mal'gin of ome of th e plots indicate zero time starting value , whi c h are measureme nts made at the storage humidity shortl y before the begin­ning of the tim ed humidity run. Even though there are usually curves from only one or two specime ns for a given material, th e trends a re ge nerally re prese nta­tive for several of o ur s pecim e ns. All the dielec tri c properties give n here were meas ured us ing th e air­gap tec hniqu e exce pt when contact electrodes are indicated.

Figure 1 shows a plot for a T -16 specime n going from < 1.5 percent rh to 52 percent rh after being exposed to the low humidity for 455 days. This mate rial does not s how long term stability as indicated by the negative slope of the curves after about 100 days. We sugges t that this aging process may be due to very slow crys tallization of this material. We offer thi s as a possible explanation, even though the s pecimens were ke pt well below the glass transition te mperature, because the materia l was not co mple tely amorphous and the tre nd is indeed in the direc tion of higher c rystallinity. Oth e r specime ns of thi s material in both humidities also showed thi s downward tre nd with time.

Fi gure 2 is a plot for two contact elec trode specI­me ns of polys tyre ne going from 52 percent rh to < 1.5 percent rho Th e s pecime n for c urve A had been in 52 pe rce nt rh for 1600 days, a nd the specime n for c urve B had bee n in 52 percent rh for 74 days.

In fi gure 3 a re c urves for two s pecime ns of poly­carbon ate. The spe cime n producing c urve A was in 52 perce nt rh fo r 992 days before the humidity c ha nge and likewise tbe specime n for c urve B was in < 1.5 per­cent rb for 1096 days before tbe cbange. Tbese curves s how tbe revers ibility of the process, but the long time data see m to indicate an aging process again. Figure 4 shows c urves for a polycarbonate specimen differ­ently prepared by the manufacturer; tbi s specimen had been in 52 percent rh for 174 days. Aside from

>-z 3.1 8 <1 >-VI

3.16 z 0 l.)

l.) 3,14 ir >- 3.12 l.) w .J

3.10 w Ci

3.08'-

l~~~'------'-'--'--~ : : . 7=··· =.. ' :I is 0 .0 1 0 .1 1.0 10 100 10 0 0 10,000

FIGU RE 1.

DAYS AFTER HUMIDITY CHANGE

Timed humidity run Jar a specimen oj 1'- 16 going from < 1.5 percent rh to 52 percent rho

121

:J :S: .. :.: :I Q 14 0

~ 130 o >- 120 -~ z liD.>!. o >= 100 c;' U\ 90 VI Ci

FIGURE 2,

0.01 0,1 1.0 10 100 1000 10,000

DAYS AFTER HUMIDITY CHANGE

Timed humidity runs for two contact electrode specimens oj PS going from 52 percent rh to < 1.5 percent rho

330~;-1 ~' .AB

',', • '0' • = I, ~ ~ ~ 299l G:1 ~ 2.97

Ci u 2.95 "b

2.93~_.L-'--_.1...J'--_L..! __ L..! __ L..l_--'L..l_--'....J

Ii~F~~~ 'j 0.01 0.1 1.0 10 100 1000 10,000

DAYS AF TER HUMIDiTY CHA NGE

FI GURE 3. Timed humidity mns Jar two specimens oj Pc. S pec ime n A v;oi nl! from 52 pe rcent rll 10 < 1.5 pe rc ent rh o and s pe c ime n B going from

< 1.5 perc e nt rh to 52 percent rh o

'"l u 3.00

~! ,~ ~ ~ 2.96

o u 2 .94

2,92

0,01 0.1 1.0 10 100 1000 10,000 DAYS A"FTER HUMIDITY CHANGE

FI GU RE 4, Timed hlUnidit y I'lLIl for a specimen oj differently pre· pared PC go ing Fain 52 percent rh: to < 1,5 percent rho

a lower dielec tri c co nstant and dissipation factor, the be havior indicated is about tbe same.

The data in figure 5 are for a specimen of FEP which was in < 1.5 percent rh for.448 days before the change; these curves sbow a slight upward trend going from < 1.5 percent rh to 52 percent rho However, within the

t;~~ " :' ,: .":":': : : Ii :~ 0: :- : 0-: . '.:--:. :l

.01 0 .1 1.0 10 100 1000 10,000 DAYS AFT ER HUMIDIT Y CHANGE

FIGURE 5, Timed humidity run Jor a specimen oj FEP going from < 1.5 percent rh to 52 percent rh,

01 a I 1.0 10 100 1000 10,000 DAY S AFTER HUMIDIT Y CHANGE

FIGURE 6. Timed humidity runJor a specimen oj PTFE goingJrom < J.5 percent rh to 52 percent rho

"'12 x

200

~ 150

u ii' z ~ g 100-

0: Vi (f)

o 50

10.0

DAYS AFTER HUMIDITY CHANGE

FI GURE 7. Timed humidity runs Jor three specimens oj PE. Specimen A going from < 1.5 pe rce nl rh 1052 percent rh, and spec ime ns Band C going

from S2 pe rcent rh to < 1.S pe rce nt rho

limit of our estimated accuracy, no change is indicated by the curves for the PTFE specimen in figure 6. This specimen was in < 1.5 percent rh for 453 days before the change to 52 percent rho In fact, no change was observed for any specimen of PTFE.

Figure 7 contains curves for specimens of poly­ethylene. Curve A shows the very slow rise in dissi­pation factor going from < 1.5 percent rh to 52 percent rh for a contact electrode specimen which had been in < 1.5 percent rh for 199 days. No change in dielec­tric constant has been observed for any polyethylene specimen. The specimen producing curve B was in 52 percent rh for 700 days before the change while the

810

800

t.D 790 Q x

~ 780 -" u ~

6 7 70

~ ~ 760 o

750

o Q)

1390

1380 Q 1370 ~

o 1360 tJ

« 1350 u.

z 1340 ~

« 1330 &

(f)

1320 0

7400L.c-1 ------�:-----------;�';c0--------;;�0~10

DAYS AFTER HUMIDITY CHANGE

FIGURE 8. An expansion oj timed humidity runs Jar three specimens oj PC going Jrom 52 percent rh to < 1.5 percent I'll..

contact electrode specimen associated with the curve C was in 52 percent rh for 123 days. Taken together these data would seem to indicate that, even though water is sorbed into polyethylene slowly, some of the water is released very quickly.

In figure 2 there are peaks in the PS dissipation factor curves after one day. Figure 8 shows similar peaks for three specimens of PC on an expanded time scale, but for this material the peak is in the region of 10 to 30 days. This strange behavior was observed only on PS and PC specimens. Moreover, it has been observed for specimens measured with contact elec­trodes as well as with the air-gap method. No explanation has been found for this unusual data.

5.2. Frequency Sweeps

Figure 9 contains the dielectric constant and dissipa­tion factor plotted against frequency for a contact electrode specimen of PC; these data were measured at < 1.5 percent rh equilibrium. Figure 10 likewise shows the dissipation factor and dielectric constant for contact electrode specimens of PS at both < 1.5 percent rh and 52 percent rh equilibrium. Figure 11 contains similar plots for two T-16 specimens.

From the other materials - PE, PTFE, and FEP - no specimens showed a significant change in dielectric constant over our frequency range. Dissipation factor data for FEP specimens at both the low and high humidity appear in curves A and B in figure 12. The dissipation factors for PE specimens at < 1.5 percent rh measured less than the limit of our accuracy so were not plotted, but 52 percent rh data appear as curve C in figure 12. .

Results differed slightly from specimen to specimen for most measurements; this was especially true at 52 percent rho We were ultimately concerned more with the careful recording of the behavior of a particu­lar specimen or group of specimens than with the general material, so none of the data presented here should be taken to represent the whole material in a detailed way.

122

2.98

11,000

2 .96

<os> 9,000

f-Z

2.94 ~ 0: '" F' 7,000 z

0 u u ~ 292 u z a: 0 5,000 f-

~ U

":i "- 2.9 0 W iii

'" a a 3,000

2.88

1,000 --- 2.86 10' 10' Ia' 105 10' Id

FREQ UENCY ( Hz)

FI GU RE: 9. Dielectric constant and dissipation factor versus fre· quency fo r a contact electrode specimen of PC at < 1.5 percent rh equil ibrium.

250

'" 200 Q )(

a: 0 f-U

~ 150 z Q f-<t "-V) V)

Ci 100

..-----

50

A

-----

------ S

103 104

FREOUE NCY I H,)

2. 555

2 . 550

2 .545

2.5 40

2 . 535

f­Z <l f­V)

Z o u u a: f­u W .J W

is

FI GU RE: 10. Dielectric constant and dissipation factor versus fre­quency for two contact electrode specimens of PS .

Specimen A is in 52 pe rc (~ nl rh equilibriulll a nd specimen B is in < 1. 5 percent rh (~qu jli briu m .

5,3, Density and Water Absorption

A linear relati onship betwee n the di electri c cons tant and density is a good approximation ove r a res tri c ted de nsity range for polye th yle ne, The co mmon form for thi s re lation is the Clausius-Mossotti equation,

15 ' - 1 E'+2 = Kp ,

wh e re p is the de nsity, 15 ' is the di elec tri c constant , and K is a co ns tant for the mate ri al whi ch de pe nds on the molecular weight and the total polarizab ility per mole-

18 ,000

"'Q x 0: 1'\000 0 f-u ~ z 0 >= <t 10,000 "-v; V)

Ci

6,000

2,000

.--.------.--.-----,--,------,--,------'3.20

A

A

s f­Z

" V) z 8

3.10 u a: f­u W .J W

Ci

"---';:--------'"---':-____ --' __ -'-:-____ -'-__ .1:-____ --' 3.00

10 2 10·

FREOU ENCY I H,)

FI GURE II. Dielectric constant and dissipa tion factor verSliS f re­quency for two sp ecimens of 1'- 16.

Sp eci me n A is in 52 Ile rce nl I'h equilihrium and specim en B is in < 1.5 pc n 'cnl rlt equilibrium.

250

200

'" Q x 0: 0 f-U <t "- 150 z Q f-<t "-v; V)

Ci

100

50

103 104

FREOUENCY (Hz)

FIGURE 12. Dissipation factor versus frequency for specimens of FEP and PE.

Specimen A is FEP at < 1.5 percent rh equilibrium ; specimen B is FEP at 52 perce nt rh equilibrium ; and specimen C is' PE at 52 percent rh equilibrium.

123

I­Z

2.370 -

~ 2 .365 z 8 u a: I-

t:l ...J W

Ci

2 .360 I-

I I

/ /

/ /

•

I

/ /

/ /

I

• /

/ /

•

/ /

I I I /

/

./ / -

/ /

/ /

/ . /

/ • / / -

/

-

2 .355 L-_..J..l __ L-I_...J.' __ L-,_--L-_---L_---1_--.J

.96 2 .963 .964 .965 .966 .967 .968 ,969

DE NSITY ( g / cm3 )

FI GU RE 13. Dielectric constant versus density for specimens of polyethylene.

c ule . The dielectric constant of six PE specime ns is plotted ve rsus density in fi gure 13. A s traight line was fitt ed to the data by a method , useful when the errors in the variables are unknown , as given by Bartle tt [8], and this line is also shown. These dif­fere nces in de nsity were probably caused by the slightly different thermal and mechanical treatme nt received during molding and mac hining.

Some effort was made to determine the amount of water absorbed by certain specimens by ta king the difference between the equilibrium weight at < 1.5 per­ce nt rh and weight a t room humidity whic h was gen­erally 20 percent rh to 40 percent rh and never above 50 percent rho A differe nce in weight of 0.03 percent was obtained from the average change of two PS speci­me ns. Thi s roughl y agrees with the value observed by vo n Hippel a nd Wesson [9] of 0.048 percent for 60 percent rh at 25°C. Likewi se , for P C we observed a 0.13 perce nt c ha nge in weight which compares with the 0.15 perce nt observed by Thompson and Goldblum [10] for 50 percent rho T-hese can not be exact com­parisons s ince the materials were, at bes t , only s imilar in composition, but th ey do s how order of magnitude agreeme nt. We were unable to detect a difference in ~e ight for F~P, PTFE or PE even though the di ssipa­tIOn factor of PE showed a quite definite chanae aoi n o­to 52 pe rce nt. The amount of water absorbed b~ PI has bee n s hown to be quite small (on the order of 0.03 or 0.06 mg/cm 3) [11 , 12], and be yond the sensitivity of our weighing sys te m.

6. Estimated Error Limits

From the use of thi s low-voltage conjugate Scherin o­bridge ove r a lo ng pe riod of time, it is estim ated th a~

repea ted measurem e nts of capaci ta nce over the fre­quency range 100 Hz to 100 kHz could be expected to agree to within ±0.001 pF for measured capacitances above 10 pF and 0.0001 pF for 10 pF a nd below. Based on thi s estimate and comparisons with the two­fluid technique as we ll as repeated de terminati ons , we estimate our dielectric constant measure me nts usin o­this bridge and the air-gap techniqu e to be acc urat~ within ± 0.1 percent.

Previous papers by Scott and Harri s [2 , 13] discuss some of the important problems involved in the meas­urement of small dissipation factors. Signifi cant errors can ari se if both the high and low sides of the circuit have some finit e common impedance to ground. In general, this is the case with a three-terminal cell or capacitor. It is possible, in principle, to comple tely analyze the circ uits involved for a given cell and bridge to allow the mos t accurate di ssipation factor de ter­mina ti on ; howe ver , that study has not ye t been com­pleted for our equipme nt. Heating by th e measuring fi.eld can also be ~ proble m with higher loss specimens, SInce the ~ropertles m ay c ha nge rapidly with te mpera­ture and . fr~qu e~cy , so we have included large per­centage limits WIth the estimated accuracies for the di ssipati on factor. I

From the use of thi s Schering brid cre over a lono­period of time, in the same manner as ~apacitance i~ is es timated that repeated measure me nts of the di s­sipation fac tor of a stable specime n co uld be expec ted to agree within 2 X 10- 6 , and based on thi s and com­pari so~ s with carefully co nstructed low loss capacitors, we es tImate the absol ute and relative diss ipation factor measure me nts using this bridge to be accurate within th e greater of the two limits given in their respec tive columns in table 1.

The di ssipation fac tor measureme nts made usin cr the Q-meter with an extra-high-Q coil a t 1 MHz'" are estimated to be accurate within ± 0.0002. At 10 MHz , this accuracy was reduced somewhat due to the uncertainty of corrections for lead induc tance . The dielectric cons tant measure ments made with thi s Q-me ter are believed _ to be accu~ate within ± 0.2 percent. These estimates were deri ved from measure­ments made on specime ns whose properties at 1 MHz and 10 MHz we re beli eved to be essenti ally the same as at lower frequenc ies, where they could be de ter­mined more accurately.

T ABLE 1. Estimated error limits fo r absolute and relative dissipation f actor measurements made with the low-voltage conjugate Schering bn dge are gwen by greater oj the two values

Freq ue ncy

10' 10' 10' 105

Absol ute e rror limit s

±30 X 10- 6 or ± 2% ±20 x 10- 6 or ± 1% ±20 X 10- 6 or ± 1% ± 40 X ]0- 6 or ± 3%

Relati ve er ro r limits

± 20 X 10- 6 o r ± 2% ± lO X 1O - 6 o r ± 1% ± 10 X 10- 6 or ± 1% ±30 X 10- 6 o r ±3%

7. Discussion

The PC and T -16 timed humidity runs show larae rapid c hanges as well as the long term drift. If the~e

124

materials were to be used for ma king dielec tric refer­ence specimens, it would be necessary to carefully control the humidity, and obviously their usefulness would be limited at bes t. The P C da ta does give a clear indication of the rever i bility of thi s process for the same or a similar s pecim e n, a nd therefore, if a copy of the timed humidity run were se nt along with a specimen, a so me what more acc ura te charac terization could be obtained unde r a large varie ty of conditions_

For the PS s pecime ns, th e c han ge was not nearl y so great ; howe ver, it would s till be necessary to take the humidity into acco unt to obtain an acc uracy be tter than about 1 percent for dielectri c constant and about 1 X 10 - 4 for di ss ipati on fac toL

The PE data show a very rapid initial change in the di ssipation factor whe n the relative humidity is changed from 52 percent rh to < 1.5 percent rh, but quite a slow drift upward on the return to 52 percent rho This behavior coupled with the lack of a c hange in the dielec tri c co nsta nt means that it would be possibl e to predi c t a specime n 's properti es quite well if it we re stored in low humidity before and after a measure me nt eve n if the measure me nt were made at a higher humidit y.

PTFE, and FEP to poss ibly a slightly less degree, are mate rials weJl s uited for dielec tri c refe re nce specime n pre parati on. Alth ough th e FEP dielectri c constant and di ssipa tion fac tor data see m to s how an upward trend go ing from < 1.5 perce nt rh to 52 perce nt rh, lines fitt ed by the leas t squ ares me thod vers us the logarithm of tim e after humidity c ha nge do not have slopes s tati s ti cally diffe re nt from zero. For PTFE, no change in dielec tric co nstant or di ssipation factor is indicated within our e rror limits for the 100 Hz to 10 MHz frequ e ncy ran ge or for a change in humidity .

The authors th a nk th e following manufacturers for their ass is tance by s upplying the mate rials used in thi s stud y. Allied Chemical Corp., Dow Chemical Co., E. 1. du Pont de Nemours and Co., General Elec tri c

Co., Phillips C he mi cal Co ., and Tennessee Eastman Co .

We also thank M. G. Broadhurst and W. P. Harris for th eir ver y valuable ass i tan ce and advice.

8. References

[II A. H. SCOli a nd W. P. Harri s, Res idua l losses in a guard·ring micro me te r-e lec trode holde r for so lid-di s k die lec tri c spec i· me ns, .J. Res. NBS 65C (phys. and C he m.) No.2, ]01 (1961).

[21 A. H. Scott and W. P . Ha rri s, Long. time e ffec ts of humidit y c hanges on th e die lec tri c prope rti es of ce rt ain pol ymers, 1962 Annu a l Report , Confe re nce o n E lec trica l In ul a tion, Na t. Acad. Sc i. -Na t. Rese arch Co unc il , Publ. 1080, p. 4l.

[3] J. L Lauritze n, Jr. , The e ffec ti ve area uf a guarded elec trode, 1963 Annual Report , Confe re nce o n E lec t ri call ns ul a tion, Na t. Acad. Sci. -Nat. Research Coun cil , Pub I. 1141 , p. 67 .

[41 W. P. Harri s, Prec ise determination of the area of gua rde d e lec trodes for accurate die lec tri c meas ure ment s on so lid di s k specimens , 1963 Annual Re port , Conference on Elec· tri cal Ins ulation , Nat. Acad . Sci. - N at. Research Council , Pub!. 1141, p. 71.

[51 A. H. Scott , Techniques for u s ing the air-gap method for the prec ise de te rmina tion of the die lectri c constant and loss a ngle of so lid di s k s pecime ns, 1963 Annual Report , Con· fe rence on Elec t ri ca I I ns ulat ion, Nat. Ac ad. Sc i. -Nat. Research Counc il , Publ. 1141 , p. 74.

[6J H. F. Sc hiefer , L. E. C rea n, a nd J. F. Kras ny, J. Res . NBS 42, 481 (1949), RP1988.

[7] W. P. Harri s a nd A. H. Scott , Prec isc meas ure mc nt of d ie lec tri c c(,lIl s ta nt by th e t wo· Auid techniqu e, 1962 A nnu al Re port , Co nfere nce un Elec t ri ca l ln sul ati otl , Na t. Acad. Sci. - Na t. Research Coun cil , Publ. 1080, p.51.

l8] M. S. Bartle tt , Fitting a s traig ht lin e wh en both vari ab les are s ubjec t to e rror , Bi omet ri cs 5 (Sep t. 1949), No. 3, p. 207.

[9] A. von Hippel a nd L. G. Wesson, Polystyre ne plas tics as high frequency die lectr ics, I ndu s t ri a l a nd Enginee ring Chemistry 38, No. ll , p. 1121 (1946).

llO] R. J. Th ompson and K. R. Goldblum , Polycarbonat e resin, Mode rn Plas ti cs 35 (April 1958), p. 13l.

[Ill c. H. Klut e, Diffu sion of s mall molecules in semicrystalline po lyme rs: wate r in polye th yle ne, J. of Applied Polym e r Scie nce 1, No. 3, p. 340 (1959).

[12] D. W. McCall , J. F. Ambrose, and V. L. Lanza, J. of Polymer Science 26, 165 (1957).

[131 W. P. Ha rri s, Appa rent nega ti ve impe dan ces a nd the ir e ffect on three- te rmin a l die lec tri c loss meas ure ments, 1965 Annu a l Re port , Co nfere nce on Electri ca l In sul a ti on, Na t. Acad. Sci.-Nat. Resea rc h Co unc il , Publ. 1356, p. 83.

(Paper 71C2-251)

125