D. W. MCCALL,D. C. DOUGLASS, AND D. R. FALCONE
998
Nuclear Magnetic Relaxation in Polytetrafluoroethylene
by D. W. McCall, D. C. Douglass, and D. R. Falcone Bell Telephone Laboratories, Inc.,Murray
Hill,New Jersey (Received August 86, 1966)
Pulse methods have been employed in a nuclear magnetic resonance study of polytetrafluoroethylene. TI,Tz, and TlP(“rotating frame”) data have been obtained as a function of temperature. The results are interpreted in terms of molecular motions and correlated with dielectric and mechanical relaxation results. In a general way the correlation is good. The low-temperature y transition is identified as an amorphous phenomenon and is detected in TI, Tz, and Tip. The crystalline transitions near room temperature are observed and evidence is presented for an unusual influence of the crystalline structure on the motions of molecules in amorphous regions. The higher-temperature 0 transition is identified as a crystalline phenomenon. Molecular motional mechanisms are discussed.
I. Introduction In this paper we report some fluorine nuclear magnetic relaxation measurements for polytetrafluoroethylene (PTFE) and attempt to interpret these results in terms of molecular motional mechanisms. A large amount of structural and relaxation data have been published and it is well known that PTFE exhibits a very complex pattern of behavior. The substance is partially crystalline and both amorphous and crystalline relaxations occur. There are a t least two crystalline transitions (aside from fusion) and a t least two amorphous transitions. It is a matter of considerable interest to observe the manner in which the various nmr relaxation times reflect these transitions. As we will show, the nmr relaxation times give important information concerning the molecular nature of the transitions. Previous nmr results can be summarized as follows. Wilson and Pake’ suggested the broad-narrow resonance decomposition as a method for measuring the degree of crystallinity. Wilson and Pake2 also made a study of the relaxation times TI and Tz by steadystate (;.e,, resonance-width and saturation) methods. They identified a sharp narrowing of the broad or crystalline piart of the resonance at 293°K and the appearance and continued narrowing of the narrow or amorphous part above about 200°K. A Tla3minimum was observed a t about 270°K. TI,decreased sharply a t 293 “K. A resonance width-temperature study by Smith4a The Journal of PhyeieCl Chembtry
indicates resonance narrowing near 210°K and some resonance-width effects near room temperature. The full resonance width a t low temperatures is about 9 gauss. S l i ~ h t e r , in ~’~ a similar study, did not observe extensive narrowing until the temperature had been raised to about 280°K. Slichter’s samples appear to have been unusually highly crystalline. Powles and Kail6 carried out resonance-width measurements that confirmed a narrowing near 220°K for a specimen that had been melted and cooled slowly. Powles and Kail also observed that PTFE as polymerized (;.e., not having been melted) did not show narrowing until the temperature had been raised to about 280°K. The latter material was shown to be much more highly crystalline than the former. Thus the superficial discrepancy between Smith’s data and Slichter’s data was resolved, as Slichter suggested in his paper, (1) C. W. Wilson, 111, and G. E. Pake, J. Polymer Sci., 10, 503 (1953). (2) C. W. Wilson, 111, and G. E. Pake, J. Chem. Phys., 27, 116 (1957). (3) Subscripts a and c are used herein to denote amorphous and crystalline parameters. The distinction between a and c in this paper is based entirely on Tz, T k 2 T%. This reflects the fact that an increase in !l’2 implies molecular mobility and the generally accepted notion that molecules in the amorphous regions are freer than molecules in crystalline regions. Although the terminology employed carries morphological implications, we emphasize that we know little about the morphology of PTFE and the nmr results do not reveal such features directly. (4) (a) J. A. S. Smith, Discussions Faraday SOC.,19, 207 (1966); (b) W.P. Slichter, J. Polymer Sci., 24, 173 (1957). (5) J. G.Powles and J. A. E. Kail, ibdd., 31, 183 (1968).
NUCLEAR MAGNETIC RELAXATION IN POLYTETRAFLUOROETHYLENE
999
on the basis of differing sample crystallinity. Eby and Sinnott6 have further confirmed this conclusion by measuring resonance widths of PTFE specimens of 48,71, and 83% crystallinity. A very neat experiment on PTFE fibers was carried out by Hyndman and Origlio.' They measured resonance width as a function of temperature, for the broad part of the resonance, for various fiber orientations. When the fiber axis was parallel to the magnetic field no abrupt change in width was observed in the roomtemperature range. At other orientations the width decreased sharply as the temperature was raised through about 288°K. Recently, Iwayanagi and Miura* have discussed broadline nmr for PTFE in terms of three resonance widths. An intermediate resonance is discerned above 240°K. This component shows the sharp transition at 293"K, falling from about 3 to 1gauss in width. Trappeniers, Gerritsma, and Oostingg employed pulse methods to measure Tl for PTFE. They observed a minimum in TI, at 295"K, a sharp change in 21' , at 298"K, and a shallow minimum in TI, at 441°K. TIa is decreasing as the temperature is raised through the melting point indicating a minimum in TI,at some temperature above 600°K. Reddish, Powles, and Hunt lo have confirmed these results. Trappeniers, Gerritsma, and Oostingg have also pointed out the influence of oxygen gas on T1 at low temperatures. Our measurements are in general agreement with the published work cited. The main emphasis in this paper is on extending the range of nmr experiments and on fitting the nmr results into a general framework with the results of dielectric and mechanical relaxation.
70% crystallinity based on the Wilson-Pake assumption.' X-Ray measurements confirm the latter range. Three types of nmr data are reported here-T2, Tl, and TI,. Tz is determined from oscilloscope pictures of the free-induction decay following a 90" pulse. It will be recalled that Tz 2 / y 6 H , where y is the gyromagnetic ratio (for F19) and 6H is the steady-state resonance width. The free-induction decay is a superposition of a short T2 signal and a long T2signal. The former we identify with the crystalline part of the sample and the latter with the amorphous part.' Extracting individual T2 values from the decay curves is not always easy because of both fundamental and practical problems. The free-induction decay tails can be fitted by a sum of a gaussian term, exp(-t2/ T2c2),and a lorentzian term, exp(--t/T2,). The coefficients of the two decay functions provide a measure of the number of nuclei and hence the amount of material represented by the individual decay terms. This enables one to use pulse data for determining the degree of crystallinity according to the method of Wilson and Pake.' Fitting the free-induction signals is tedious at best. It must be kept in mind that the nuclear signal is not seen for a period t, after the end of the 90" p ~ 1 s e . l ~During this recovery period the long and short T2components have decayed by different amounts. TI, the spin-lattice relaxation time, is measured by observing the amplitude of the free-induction decay following a 180-90" pulse pair. If the pulse separation is 7s,a simple resonance will follow
II. Experimental Section
v(t) = vco(t)(i - 2e-7JT1~)
The spectrometer employed is similar to one described previously." It has been modified for operation with a single-coil probe and the 90" phase-shift method for aligning the magnetization12with the radiofrequency field (for "rotating frame'' measurements) was adopted part way through the present study. The spectrometer is operated at 30 mHz, correspondingto an applied field of about 7500 gauss. Specimen temperature is controlled by means of a gas flow thermostat. The data are probably accurate to within 1°K. The material chosen for study was a commercial PTFE (Allied Chemical Co., Plastics Division, designated Halon, G-80). It is believed to be substantially linear and very pure as indicated by its low dielectric loss. The density is about 2.17 g/cm3, which indicates about 55% crystallinity.6 Our nmr results yield 60-
V(t> = Vo(t>(l - 2e-rs''1)
(1)
In PTFE the resonance is more complex and can be described by the sum of two such terms
+
vuo(t)(i - 2e-78//T*a)(2) (6) R. K. Eby and K. M. Sinnott, J . A p p l . Phys., 32, 1765 (1961). (7) D.Hyndman and G . F. Origlio, ibid., 31, 1849 (1960). (8).Cited in N. Saito, K. Okano, 5. Iwayanagi, and T. Hideshima. Solzd State Phys., 14,414 (1963). We are indebted to Drs. Iwayanagi and Miura for providing us with a preprint of an article dealing with their recent nmr results for PTFE. They plan to publish their article in Section C of J . Polymer Sci. (9) N. J. Trappeniers, C. J. Gerritsma, and P. H. Oosting, Physica, 30, 997 (1964). (10) W. Reddish, J. G. Powles, and B. I. Hunt, Polymer Letters, 3 , 671 (1965). (11) G. P. Jones, D. C. Douglass, and D. W. McCall, Rev. Sei. Instr., 36, 1460 (1965). (12) 9. R. Hartmann and E. L. Hahn, Phys. Rev., 128, 2042 (1962). (13) This recovery time is a consequence of receiver saturation and "ringing" in the resonant circuit. In our apparatus, which employs a damping pulse, tr is about 6 psec.
Volume 71, Number 4 March 1967
D. W. MCCALL,D. C. DOUGLASS, AND D. R. FALCONE
1000
Vco(t) and V8"(t) correspond to the short and long T2 components mentioned above. In PTFE, TI, > Tla. Detailed fitting of these functions is particularly tedious and we have resorted to making a visual identification14 of 7 8 8 , defined by (1 - 2e-'"'T'a) = 0, and rea. Obviously, TI, = rs8/ln 2 and TI, = rSc/ In 2. Experiments have shown that the TI values obtained by this method agree satisfactorily with TI values obtained from the appropriate plot, eq 2. There is some ambiguity in identifying 7 8 8 but the uncertainty in TI8 should be no larger than about 20%. Relative T I Svalues are not subject to this difficulty if a consistent procedure is adopted. The temperatures at which T I minima occur are not affected. T,,, the longitudinal relaxation time in the rotating frame, is measured by observing the amplitude of the free-induction decay that follows the termination of a resonant pulse of radiation, the nuclear magnetization having been aligned with the rotating field, HI,a t the beginning of the pulse. 1 5 3 When the free-induction signal amplitude falls exponentially with pulse length, 7rf
V(t) = Vo(t) exp(-rrf/Tlp)
(3)
defines TI,,. PTFE exhibits more complex behavior that has been analyzed as a sum of three exponential terms. The two long T,, terms correspond to Tzo. The magnitudes of these two times are probably affected by the presence of oxygen. The short TI, corresponds to Tz8. Rotating frame measurements were made a t two power levels, HI = 3 gauss and HI = 12 gauss. The procedure for getting into the rotating frame was different for the two levels but this procedural detail does not influence the interpretation 2f Tl,. Steady-state magnetic resonance spectra have been obtained with a Varian DA-60 spectrometer system.178 The resonances were nearly symmetrical supporting a highly linear structure.17b Both modulation and slow-sweep techniques were employed. With the former method derivative spectra are obtained, and with the latter method the absorption is observed directly. We can neither confirm nor deny the existence of three distinct resonance widths on the basis of these observations. At 270°K a superposition of two resonances suffices.
III. Results Figure 1 shows the temperature dependence of T2. The behavior is generally consistent with resonancewidth studies, considering T2 E 2/y3H. I n Figure 1 a sharp increase in Tza occurs near 293°K in complete agreement with earlier workers.1fjJ The magnitude of the increase is about a factor of 2 a t the transition The Journal of Physical Chemivtry
PTFE
10 -100
-50
o
50 loa TEMPERATURE ('C)
ISO
2w
25a
Figure 1. Temperature dependence of the spin-spin relaxation time, T,.
followed by a gradual increase to a factor of 3 as the temperature is raised to about 400°K. The scatter of the data simply indicates that it is difficult to separate T2, and TZa quantitatively, particularly when Tz8and Tzcas well as TI, and TICare not very different. TZ8is seen to increase gradually over the temperature range studied. T2a has increased by a factor of 2 when the temperature is about 220°K. This agrees with previous reports.'l6J Figure 2 shows the temperature dependence of TI. TIa shows an abrupt transition a t about 293°K. This change was seen by Wilson and Pake2and Trappeniers, Gerritsma, and O o ~ t i n g . ~The latter authors referred to this feature as a TI minimurn but it seems clear that it is not a T1 minimum in the usual sense. A fairly sharp increase in TI, occurs near 310°K. The minimum in T1, near 340°K was also observed by Trappeniers, Gerritsma, and Oostingg although they obtained a ~
~~~
(14) This visual estimation of the two T I values is convenient and
useful. It relies upon the fact that crystalline and amorphous signals have different T I and TZvalues. The nuclear signal is examined a t a time -2Tz8 after a 180-90° pulse sequence. The crystalline signal is negligible at this point as Tz0 < Tza. The pulse separation is adjusted until the amorphous signal is zero and ssa is taken as this pulse separation. In determining Tl0,we assumed that the amorphous signal exhibits a smooth, approximately exponential decay. The nuclear signal is examined just after the 90° pulse and the pulse separation is adjusted until the decay is almost flat. This pulse separation is taken to be rso,i.e., the interval that nulls the crystalline signal. (15) (a) D. Ailion and C. P. Slichter. Phys. Rev. Letters. 12, 168 (1964); (b) C. P. Slichter and D. Ailion, Phys. Rev., 135, A1099
(1964).
(16) (a) D. C. Look and I. J. Lowe, J. Chem. Phys., 44, 2995 (1966); (b) D. C. Look, I. J. Lowe, and J. A. Northby, ibid.,44, 3441 (1966). (17) (a) We are indebted to Mr. E. W. Anderson for recording these spectra; (b) C. W. Wilson, 111, J . Polymer Sci., 56, 516 (1962).
NUCLEAR MAGNETIC RELAXATION IN POLYTETRAFLUOROETWLENE
1001
2 0
1
H, HI
= 12 GAUSS = 3 GAUSS
s 2
-
10-1
Y
5
2
2 10-2 5
a 10-3 5
Figure 2. Temperature dependence of the spin-lattice relaxation time, TI.
2 4
10-4 -150
-100
-50
0
50
100
150
200
TEMPERATURE f°C)
lower minimum T1,--O.6 sec compared with -0.8 sec in the present study. I n either case this is a very shallow minimum. TI, exhibits a pronounced minimum a t about 280°K. Figure 2 shows two curves near this minimum representing two different criteria for separating TI. and TI,. The minimum occurs a t 280 5°K. Wilson and Pake2 observed this minimum a t about 270°K; Trappeniers, Gerritsma, and Oostingg found it a t 294°K (although their points do not define the minimum precisely), and Reddish, Powles, and Huntlo found it a t 295°K. The latter authors did not publish their T I us. temperature curve. These temperature differences are too large to be ascribed to ordinary experimental uncertainty even though the minimum does not appear to be sharp. We find the minimum value for TI. to be -0.12 f 0.02 sec compared with -0.08 sec published previou~ly.~T I , seems to be falling off a t high temperatures suggesting another TI, minimum above the melting point.g T I , is plotted as a function of temperature in Figure 3. Below about 290"K, three time constants are observed, the longer two corresponding to TeCand the shorter T I , to T2,. Between 290 and 310"K, two TIP values suffice. The longer T I , in this range corresponds to Tzc. Finally, above 310°K a single TIP describes the rotating frame decay curves. These general features were apparent for both H I = 3 gauss andH1 = 12gauss. T,,,, the shortest time constant at temperatures below 310"K, exhibits a well-defined minimum a t about
Figure 3. Temperature dependence of the rotating frame relaxation time, Tip.
218°K for HI = 3 gauss and 223°K for HI = 12 gauss. Corresponding minimum values of T,,, are 0.3 and 1.0 msec. Abrupt changes in TI,, occur near 290 and 310°K. The two TI,, values fall continuously as the temperature is raised and come together a t about 290°K. The single TI,, merges with TI,, a t about 310°K. The single T I , passes through a shallow minimum near 330°K for H I = 12 gauss and 320°K for H I = 3 gauss. The corresponding T,, minima are 4.8 and 1.4 msec. Our steady-state spectra are considerably more symmetrical than those of Iwayanagi and Miura.* We note that they report three Tz values well above the temperature range for which we detect three or two T I , values. Our data do not exhibit three resonance widths in any obvious way.
IV. Discussion The over-all relaxation behavior of PTFE is indicated on the transition "map" shown in Figure 4. Table I summarizes the observations. The data plotted have been assembled from a number of literature sources, those listed by Eby and SinnottJ6among others. We have attempted to select data obtained for materials that have been sintered or molded and fall into the medium crystallinity range. Frequencies of maximum loss taken from isotherms or, more often, temperatures of maximum loss a t conVolume 71, Number $.
March 1967
D. W. MCCALL, D. C. DOUGLASS, AND D. R. FALCONE
1002
Table 1 (.* gl...
Transition phase Mechanical activity Dielectric activity Nmr activity
0
1
Amorphous Yes Very weak
6H T, TI TIP
No 1
No
30-
190
(I,
glaw 11
Probably crystalline Yes NO“
Crystalline Yes No
Crystalline Yes
No
Amorphous Yes Yes
Yes Yes 1
No Yes Yes
Yes Yes Yes
Yes Yes Yes
Activity seen only (Krum and lluller~8):see text.
-2
I
2
I
I
J
I
I 4
9o)fr
I
, I
I
6
Figure 4. Correlation diagram for various relaxation phenomena in PTFE. Nmr results m e indicated by filled circles and were calculated as indicated in the text. The nmr parameters t h a t correspond to the various points are indicated on the graph. Dynamic mechanical relaxation results are was taken as the frequency of indicated by open circles. measurement a t the temperature of maximum loss. Dielectric results are indicated by triangles. yo was tnken to he the frequency of maximum loss when sufficient data were available (a or g l w I1 process). Higher-temperature dielectric data were analyzed in the same menner used for taken as the frequency of mechanieabloss data, i.e., measurement a t the temperature of maximum loss. A part of the scatter observed in this figure probably a r k s from these different analyses, i.e., the use of isothermal us. constantfrequency plots.
TIpe(3),where T1,(3) is the value for HI = 3 gauss (see eq 6 below.). The following aspects of PTPE relaxation are reasonably well established. (1) A low-temperature (y or glass 11) relaxation is evident in mechanical, dielectric, and nmr data. It is clearly a phenomenon associated with the amorphous regions of the polymer. The activation energy is often quoted as 1s kcal/mole. (2) A firsborder crystalline transition occurs just below room temperature (“the 19” transition”). The lowtemperature form is triclinic and the room-temperature form is hexagonal..l.”.zl The molecules in the crystalline regions begin to rotate about their long axes a t this transition. This is clearly indicated by the results of Hyndman and Origlio’ based on nmr resonance widths for PTFE fibers. (3) A firsborder crystalline transition occurs just above room temperature (“the 30” transition”). The high-temperature form is pseudo-hexagonal.lg (4) Relaxation phenomena associated with the two firsborder transitions are observed above and below room temperaturez2 (0transition region). (5) Rather high-temperature (> 4 W K ) mechanical relaxations (CY or glass I) are observed. Both amorphous and crystalline phenomena are involved. (6) A sharp crystalline melting point is observed at about 600°K. ’We will now take up the various features individually and interpret the accompanying nmr behavior in detail. A. The y M Glass II Transition. Tz increases gradually above 200°K. It has increased by approximately an order of magnitude a t 400°K. This indicates fairly extensive molecular motion but by no means liquidlike mobility. The middle of the transi-
stant frequency were used for the dielectric and m e chanical points. For nmr resonance-width results, the correlation frequency is taken to be (76H)LT/2S at the temperature a t which the resonance width has fallen to half its low-temperature value, (~H)LT. (IS) N. Bloernbergen. E. M. Purcell. and R. V . Pound. Phya. Re.. 73. 679 (1948). Tz is used similarly. va ~/(*Tz)LT,where (TZ)LT (19) D.C. Douglaas and G . P. Jones, J . C h .Phya.. 45,966 (1966). is Tz below the transition range.’* Temperatures for (20) C. W. Bun” and E. R. Howells, Nolure, 174, 649 (1954). where vo is the TI minima correspond to yo s &v0, (21) E. 8. Clark and L. T.Muus, Z.Krisl.. 117, 119 (1962). resonance frequency.I8 A t T,, minima,lS yo ZZ 1 / 2 r (22) K. H.nlem and E. Jenckel, K 0 W - Z . . IM). 97 (1966). ~
NUCLEAR MAGNETIC RELAXATION IN POLYTETRAFLUOROETHYLENE
tion can be relied upon to give an indication of that temperature for which V, = (T6H)LT/2X1 but attempts t o fit the T2 variation in a detailed way, as Eby and Sinnott did for the resonance widthls are unprofitable. The analysis underlying such a procedure is invalid for systems with distributions of correlation times. The Tl minimum as determined by this study and by Wilson and Pake2 correlates very well with the assembled data. The other T I minimaeJOdeviate appreciably from the locus of other relaxation data. Reddish, Powles, and Huntlo have previously pointed out this discrepancy. A dielectric loss maximum is observed23in our PTFE near 300 mHz a t 298"K, in agreement with the result of Reddish.l0 This could be interpreted as an indication that the specimens used in this study are similar to those used by Reddish, Powles, and Hunt;. The absolute value of T I at the minimum can be estimated as18
(T1)min E ( ~ T v ~ T ~ ) L=T 0.03 ~ / ~sec /Z
(4) compared with about 0.12 sec observed. This is rather close agreement, as such comparisons go, and can be explained in terms of correlation time distribuor spin diffusion.26*27 The T,,, minima a t HI = 3 and 12 gauss fit in very well with the bulk of published relaxation data. The magnitude of T,,, a t the minimum can be estimated from'g which yields 0.6 X sec a t 3 gauss and 2.4 X sec a t 12 gauss. These are to be compared with the observed values 3 X low4and 10 X sec. Thus the factor of 4 discrepancy between the predicted and measured ('Z'Jmin carries over to the (T1pa)min. A distribution of correlation times is the most likely explanation of these results. At temperatures and radiofrequency fields sufficiently low that r c >> T2 and H1 E 6H, Slichter and Ailion16 have shown that
(6) The two lowest nmr points in Figure 4 were calculated from Tlpsa t H1 =: 3 gauss using this relation. Reasonably good correlation is observed but the slope deviates appreciably. The deviation is probably the result of another relaxation mechanism. Rotating frame decay curves indicate the presence of a fourth, very short time constant at the lowest temperatures studied. The dielectric results of Krum and Muller28 provide additional evidence for an additional low-temperature relaxation. It is possible that this relaxation is to VC
'/z~T1,
1003
be identified with the low-temperature process observed by Eby and Wilson.29 They suggest that -CF2H groups, in low concentration, give rise to relaxation. B. The 19" Transition and the SO" Transition. T2, increases sharply by a factor of about 2 while T2,is unaffected by the 19" transition. This confirms the result of Hyndman and Origlio.? An abrupt decrease in TICis observed but T I ,varies smoothly through the transition region. Tlpaundergoes a sharp change a t the 19" transition. This is surprising for two reasons; the transition is a crystalline one, and T1, shows no effect. I n seeking an understanding of these experimental results the following considerations are pertinent. At 19" TI, is dominated by the y process; the minimum in TI, occurs only 10" below. T,,, is not near its y minimum and is consequently more susceptible to other relaxation phenomena. Tlpais shorter than T,,, so the molecular motions controlling T,,, must be occurring in the amorphous regions. Of course, molecular motions are occurring in the crystalline regions as well. Between the 19" transition and the 30" transisec. tion T,,, decreases to quite a low value, 0.6 X Estimating that the appropriate ( T ~ & )isL T about 40 X sec, eq 5 yields (Tlpa)mina t about 0.5 X low3 sec, H1 = 3 gauss. Thus we suggest that the molecular motions controlling T,,, near room temperature must be distinct from the y mechanism and have correlation frequencies near 2 X lo4Hz a t 35". Krum and Muller2* observed a corresponding dielectric loss peak but none was found in our material.23 The specimen studied by Krum and Muller may have had polar impurity groups that made this relaxation and others dielectrically active. Much of what has been said about the 19" transition also applies to the 30" transition. The latter appears to be somewhat less sharp and, in our specimen, occurs nearer to 40 than 30". The T,,, and T I , data give the impression that the motions turned on at the 19" transition are turned off a t the 30" transition. Clark and Muus2I propose that the helical molecular conformation changes to a more extended helix at the 19" transition. This results in a hexagonal arrange(23) We are indebted to Mr. G . R. Johnson, Bell Telephone Laboratories, for this result. (24) D. W. McCall, D . C. Douglass, and E. W. Anderson, J . Chem. Phys., 30, 1272 (1959). (25) T.M. Connor, Trans. Faraday Soc., 60, 1574 (1964). (26) N. Bloembergen, Physica, 15, 386 (1949). (27) D.W. McCall and D. C. Douglass, Polymer, 4, 433 (1963). (28) F. Krum and F. H. Muller, Kolloid-Z., 164, 81 (1959). (29) R. K. Eby and F. C. Wilson, J. Appl. Phys., 33, 2951 (1962).
Volume 71, Number 4 March 1967
1004
D. W. MCCALL,D. C. DOUGLASS, AND D. R. FALCONE
ment and hindered rotation occurs between the preferred orientations. At the 30" transition, further untwisting occurs and the conformation becomes irregular. The rotational hindrances cease to be important. The nmr results indicate a very unusual influence of the room-temperature crystal structure on molecular motions in the amorphous regions. The nature of the molecular motion in the roomtemperature range is not entirely clear. We have found that motions occur in both the crystalline and amorphous regions. The latter motions set in sharply a t the 19' transition and must be extensive enough that the local field is modulated by a large fraction of its average v:rlue. Inasmuch as the y transition has already signaled rather general motions in the amorphous regions, considerable activity must be present. Impurity or boundary region motions are not sufficient to explain the data. However, the fact that rotation about the long molecular axes in the crystalline regions triggers the amorphous process might suggest that the amorphous regions have rather small average dimensions, say a few angstroms. C. The p Transition. The /3 transition is observed as a minimum in TI,. This minimum is shallow, suggesting that the motion involved does not change the local magnetic fields drastically or, alternatively, that only a small fraction of the nuclei have the motional freedom. The minor change in TZc in the appropriate temperature range, -375"K, would be consistent with the first possibility. The nmr data clearly indicate that the p process occurs in crystalline regions. T,, exhibits a minimum in the vicinity of 325"K, considerably too low a temperature to be identified with the p mechanism. Krum and Muller% observed a corresponding dielectric effect although other workers have not.n,2g,30We cannot distinguish T,,, from T,,, in this region and T1. and TIc are not too different in the p region, -440°K.
The molecular interpretation of the p process is speculative at best. We propose longitudinal (i.e., parallel to the helical axes) motions of the molecules, which would affect only the intermolecular dipolar fields, as the operative mechanism. The closeness of the crystalline and amorphous TIand T,, values could be interpreted as an indication that phase-boundary motions are involved. D. The CY or Glass I Transition. The CY or glass I transition is not detected by nmr in any obvious way. T1,is decreasing a t the highest temperature studied and it may be presumed that a minimum will be reached, apparently above the melting point.g Tz and T1, do not give an indication of such a process. Referring again to the work of Krum and R/Iuller,28 a dielectric loss is observed in the CY region which increases in intensity with increasing crystallinity. Its temperature is independent of frequency, suggesting that it may be indicative of another first-order transition. The intensity of this loss increases markedly a t low frequencies and it is best interpreted in terms of an ionic conduction mechanism ( L e . , a "Maxwell-Wagner" process). This dielectric observation is of interest in the context of the present paper because the minimum in TI,, occurs a t the same temperatures, -440°K. Thus the T1, minimum that we have associated with the p transition would also be consistent with a diffuse first-order crystalline transition. In any case, there is some evidence% that crystalline motions also occur in the a! region.
The Journal of Physical Chemistry
Acknowledgment. We are indebted to W. P. Slichter, W. Reddish, and E. S. Clark for valuable comments and discussions. ~
(30) G. P. Mikhailov, 9. P. Kabin, and A. L. Smolianski, J . Tech. Phys. USSR,2 5 , 2179 (1955).