HNBR and long term serviceability in automotive lubricants: structure property relationships.

The analytical investigation done previously[ref. 16] puts

forth the hypothesis that hydrogenated nitrile (HNBR) rubber

in the presence of nucleophilic and basic species can lead to a

chemical reduction of the acrylonitrile group. Lubricant

additive packages found in today's automotive fluids contain

both nucleophilic and basic constituents. Automatic

transmission fluid (ATF) was used in this study.

When subjected to these nucleophilic and basic

constituents, the cyano group of HNBR may be converted to an

isocyanide or isonitrile ion as shown in figure 1[refs. 1 and 16].

Isocyanide and isonitrile ions carry a partial dipole

moment opposite that of their acrylonitrile precursor [(.sup.-]C =

[N.sup.+])[ref. 4]. This charge configuration causes both species to

be metastable at best. The presence of water or oxygen,

coupled with heat, can further advance these structures to an

imide, imine or potentially an aldehyde[ref. 1] (figure 1). For

this article, the scope will be limited to the proposed

mechanism of acrylonitrile sites possibly being converted to

isocyanide or isonitrile ions initially, then to an imide crosslink.

We will attempt to further investigate what was found in the

previous work through dynamic mechanical characterization.

[FIGURE 1, ILLUSTRATION OMITTED]

Experimental

HNBR samples were formulated and molded as described in

the experimental discussion of part 1[ref. 16]. Agings were

done in Scott blocks using ATF at 150 [degrees] C, in accordance with

ASTM D471. Samples for this portion of the investigation

were 3.3 mm thick, compared to 2 mm used previously. This

was done to facilitate dynamic mechanical characterization.

Data were obtained using a Rheometrics RDA-11 in the

torsion rectangular mode. A sample of dimensions 29.4 mm

x 12.8 mm x 3.3 mm was used for each analysis. Data were

collected over the frequency range of 0.1 to 100 rad/sec. at a

data collection rate of 10 points per decade of frequency at

each temperature step. The temperature sweep of the data

ranged from -150 [degrees] C to 150 [degrees] C for the unaged and 3,500 hours

samples, and -75 [degrees] C to 150 [degrees] C for the remaining samples,

using a four degrees per step test method.

Results

Elastic shear modulus scans

Figure 2 compares the storage modulus, G', of unaged

HNBR with four samples aged in ATF for the listed time

periods. All of the data are extracted from a temperature

sweep scan taken at 0.1 rad./sec.

[FIGURE 2, GRAPH OMITTED]

Initiating the review of data in the low temperature

regime, the unaged HNBR exhibits a drop in shear modulus at

-35 [degrees] C. This drop is attributed to the main glass transition

temperature, Tg ([Alpha] relaxation), of the unaged HNBR, and

will be referenced by [Tg.sub.1], hereafter. All of the aged samples

exhibit the same drop in shear modulus at roughly the same

temperature.

Progressing through [Tg.sub.1], it can be seen in figure 2 that the

transition region for the aged samples becomes broader as the

aging time increases. Samples aged 1,500 hours and longer

show the establishment of a plateau modulus, between -20 [degrees] C

and 30 [degrees] C. This first plateau region will be referenced by [G.sub.N],

hereafter. A second transition, starting at approximately

40 [degrees] C, appears in samples aged for 1,500 hours and longer.

This second transition is not seen in the unaged or 750

hours aged samples. This transition will be referenced as [Tg.sub.2].

hereafter. Continuing above this transition, the long term

aged samples enter their equilibrium plateau, commonly

referred to as the equilibrium plateau modulus ([G.sub.E]). The

equilibrium plateau modulus had been reached for the

unaged and 750 hours samples by -10 [degrees] C and 20 [degrees] C,

respectively.

One additional note from the elastic shear modulus scans,

the equilibrium plateau modulus for the unaged and 750 hour

samples appears to be increasing with temperature. We feel

this is an artifact of the test procedure and not a marching

modulus, as described in the elastomer industry, which is the

result of incomplete cure. This gradual increase is probably

due to the onset of aging caused by the long time between

measurements at each temperature, approximately 14

minutes.

Returning, our focus to the first plateau modulus ([G.sub.N]) in

figure 2, initially it could not be determined whether this

plateau was due to additional crosslinking of residual

unsaturation or other potential crosslinking sites (i.e. tertiary carbon

on the acrylonitrile group, unsaturation in 1,2 sites, etc.) after

curing. Other potential causes could be increased

crystallinity. or some type of ionic behavior.

In the first part, an investigation was done to determine

whether this increase in shear modulus was due to unreacted

crosslinkable sites. Twice the typical amount of curative

(peroxide) was added to the base polymer matrix and

vulcanized[ref. 16]. The dynamic scans in figure 2 indicated the

increase in shear modulus was from 1.0E7 dynes/[cm.sup.2] to 1.2

to 1.6E8 dynes/[cm.sup.2], a change that could not be achieved

solely due to unreacted crosslinkable sites.

This was based on determining the change required in the

number average molecular weight between crosslinks ([M.sub.c]) to

achieve such a shift in shear modulus. The calculation was

done assuming affine deformation. Having assumed this, the

total work done on the network during deformation was then

calculated using the equations shown in table 1[refs. 11-15].

Again, we felt this was not a highly probable explanation for

the storage shear modulus ([G.sub.N]) change seen in figure 2.

[TABULAR DATA NOT REPRODUCIBLE IN ASCII]

When the long term aged samples (1,500 hours and

greater) reach their equilibrium plateau ([G.sub.E]), the value is

higher than the unaged or 750 hours samples. This difference

indicates that some additional crosslinking has occurred (a

reduction in [M.sub.c]). Although this appears to be a small amount

on the graph, keep in mind that the y-axis is logarithmic 

(figure 2).

Next, we chose to examine the potential of increased

crystallinity. Documented research has shown that highly

crystalline materials may give scans similar to the aged scans

seen in figure 2[refs. 2 and 3]. However, we could not think

of a reasonable hypothesis why one would experience such

an increase in crystallinity. The HNBR tested was a

crosslinked amorphous elastomer. Aging was done in a free,

undeformed state at elevated temperature, conditions not suitable

for significantly increasing crystallinity in an elastomer.

There are proposed theories and discussions that the

highly saturated grades of HNBR ([is less than] 1% unsaturation remaining

after hydrogenation of the NBR precursor) give high ultimate

strengths due to strain crystallization of the mostly olefinic

backbone. Again, this does not seem to be a plausible

mechanism for the shear modulus changes seen in figure 2, as the

strain applied to the samples was extremely small (0.1%) and

elastomers do not crystallize in shear.

The final hypothesis to investigate was ionic effects.

Review of prior work on the viscoelastic characterization of

ionomers, mainly the research of Eisenberg and Navratil[refs. 6-9],

Yeo and Eisenberg[ref. 10], and Ward and

Tobolsky[ref. 5], showed results that closely paralleled our

results. This, coupled with the results proposed from part 1 of

this work, led us to investigate this hypothesis further.

The research of Eisenberg et al., was on styrene-based and

perfluorinated ionomers. Tobolsky et al., researched

copolymers of a-olefins and carboxylic acids. Their viscoelastic and

dynamic mechanical data closely resemble what we have

found in aged HNBR. Two key hypotheses from their work

which we used as a fundamental basis for ours are: 1) neither

the ion size nor valence play a major role in influencing the

viscoelasticity of the polymer matrix. The concentration of

the ionic species present is the predominate factor. 2) The

formation of ionic behavior may cause the material to behave

thermo-rheologically complex. Thereby, time-temperature

superposition (TTS) could break down in most cases, but not

in every case. The concentration of ionic species present

dictates whether TTS is valid or not.

Having established what we could propose from the

elastic shear modulus scans. and how that led us to investigate

the characterization of ionic species and ionomers, we next

reviewed our scans in light of the findings of Eisenberg et al.

and Tobolsky et al.

Eisenberg and co-workers proposed that below a critical

ion concentration, these ions probably exist as multiplets and

act as pseudo-crosslinks. These pseudo-crosslinks slow down

the primary diffusional relaxation mechanism, but still obey

TTS. Above the critical ion concentration, the ion multiplets

act as crosslinks of finite lifetime and TTS becomes invalid.

Tobolsky and co-workers also found the same results in

their a-olefin and carboxylic acid copolymers at least five

years prior to Eisenberg's published findings.

To examine this concept of TTS failure with the presence

of ionic character, we went back and attempted to do TTS on

the elastic shear modulus scans of the unaged HNBR, and

samples aged 1,500 hours and 3,500 hours. The scans were

not compensated for the change in Tg because there was very

little change after aging. The results are shown in figure 3.

[FIGURE 3, GRAPH OMITTED]

Using [Tg.sub.1] as the reference temperature, all of the scans

were shifted using the WLF equation. None of the samples

showed "strict" adherence to the WLF equation. Again, two

features are apparent.

This follows the same type trend that Eisenber, and

Tobolsky found in their investigations.

Loss shear modulus scans

Figure 4 compares the loss shear modulus, G", of unaged

HNBR and samples aged in ATF at 150 [degrees] C for the listed time

periods. All of the data are extracted from a temperature

sweep scan taken at 0.1 rad/sec.

[FIGURE 4, GRAPH OMITTED]

Again, starting the review of data in the low temperature

regime for the unaged and 3,500 hours samples (figure 4), it

appears both have their b-relaxation occurring at-120 [degrees] C

to - 70 [degrees] C. This relaxation is probably occurring in the

polyethytlene or butadiene segments of the unaged HNBR (ref. 2]. The

b-relaxation in the sample aged 3,500 hours is probably

occurring in the polyethylene segments, since we suspect that

all of the butadiene segments have been reacted after this

amount of test time. None of the other samples were

analyzed below -75 [degrees] C since our primary area of focus was on

[Tg.sub.1] and above.

The a-relaxation, [Tg.sub.1], of the unaged sample occurs at

35 [degrees] C. After aging up to 3,500 hours, [Tg.sub.1] is essentially the

same, or shifted very slightly to higher temperatures

([is less than] 1 [degrees] C).

The primary difference between unaged and aged scans is the

breadth of the transition region.

Between the unaged sample and the sample aged for 750

hours, the transition region has already begun to widen

(figure 4). After 1,500 and 2,500 hours, there is the clear

development of a second transition, earlier designated as [Tg.sub.2]. This

was seen in Eisenberg's work, thereby we are proposing that

this second peak is due to an increase in ion concentration

(i.e. isocyanide or isonitrile ions), but still below the critical

ion concentration.

Continuing aging out to 3,500 hours, the intensity of this

peak diminishes and the loss modulus curve takes on a

different profile, an inflection point develops (figure 5a and b).

Eisenberg found the same type behavior in his loss shear

modulus versus temperature scans, although he was working

with a styrene-based polymer. His research proposed that,

below a critical ion concentration, a second peak in the loss

modulus curve would exist (what we earlier referenced as

[Tg.sub.2]). Above the critical ion concentration, no distinct second

peak in the loss modulus curve would be observed; instead

an inflection point would appear (figure 4).

[FIGURES 5A & 5B, GRAPHS OMITTED]

Comparison of our loss shear modulus scans to those of

Eisenberg's would indicate the possibility that at up to 2,500

hours, the ionic species formed is below the critical ion

concentration. Continuing the aging to 3,500 hours, the critical

ion concentration is exceeded, thus changing the profile of

the loss modulus curve.

Discussions

This initial investigation into the aging of HNBR in

nucleophilic and basic environments has made it apparent that the

nitrile moiety is subject to attack in these environments.

Chemical literature has shown this not to be a discovery, but

maybe an overdue review. Jerry March's "Advanced organic

chemistry -- reactions, mechanisms and structure," although

extremely compartmentalized, is a good treatise to

investigate this mechanism.

Further research has already been completed in this area,

but was not presented. We have investigated the effects of

acrylonitrile content, filler types and levels. Nothing so far

has been found to indicate the attack is not occurring at the

cyano group.

O-rings have been obtained that experienced a substantial

volume swell (15-20%) and increase in shear modulus (from

Shore A to Shore D scale). This precludes the fact that the

hardening could have occurred first. The swelling occurring

first would be within reason, with the formation of some type

of ionic species. This would lead to a significant uptake of

water, due to a potential (zeta) gradient. Eisenberg et al.,

found the same trend with the addition of ionic species to his

styrene-based ionomers.

Conclusions

* Using dynamic mechanical analysis, the chemical

changes of HNBR in the presence of nucleophilic and basic

species was characterized to verify the findings from the

analytical study.

* Results from our dynamic mechanical characterization

of HNBR aged in ATF, in combination with prior research

efforts by Eisenberg et al., and Tobolsky et al., we are

proposing the hypothesis that the species initially formed

through aging is "ionic" in nature. This could be due to the

formation of ioscyanide and/or isonitrile ions, before the

eventual formation of an imide crosslink.

* Data indicate that up to 3,500 hours in ATF, there is the

development of a first plateau modulus, [G.sub.N], which we are

proposing is due to the formation of an ionic structure; and

there is an increase in the equilibrium plateau modulus, [G.sub.E],

which we are proposing is the formation of an imide

crosslink.

References

[1.] March, Jerry, "Advanced organic chemistry - reactions,

mechanisms and structure," 3rd Edition, Wiley-Interscience,

New York (1985).

[2.] McCrum, N.G., Read, B.E. and Williams, G., "Anelastic

and dielectric effects in polymeric solids," Dover

Publications Inc., New York (1991).

[3.] Osswald, Tim A. and Menges, George, "Materials science 

of polymers for engineers," Hanser Publishers. Munich

(1996).

[4.] Ugi, Ivar, "Isonitrile chemistry," Academic Press, New

York (1971).

[5.] Ward, Thomas and Tobolsky, Arthur, Journal of Applied

Polymer Science, Vol. 11. pp. 2403-2415 (1967).

[6.] Eisenberg, A. and Navratil, M., "Polymer letters," Vol. 10,

pp. 537-542 (1972).

[7.] Eisenberg, A. and Navratil, M., "Ion clustering and

viscoelastic relaxation in styrene-based ionomers II: Effect of

ion concentration," Macromolecules, Vol. 6, No. 4,

July-August 1973.

[8.] Eisenberg, A. and Navratil, M., "Ion clustering and

viscoelastic relaxation in styrene-based ionomers III: Effect of

counterions, carboxylic groups and plasticizers,"

Macromolecules, Vol. 7, No. 1, January-February 1974.

[9.] Eisenberg, A. and Navratil, M., "Ion clustering and

viscoelastic relaxation in styrene-based ionomers IV: X-ray and

dynamic mechanical studies," Macromolecules, Vol. 7,

No. 1, January-February 1974.

[10.] Yeo, S.C. and Eisenberg, A., "Physical properties and

supermolecular structure of perfluorinated ion-containing

(Nation) polymers," Journal of Applied Polymer Science,

Vol. 21, pp. 875-898, 1977.

[11.] Holmes, G.A., "The effect of molecular architecture on

the mechanical properties of epoxy resin systems," Master's

Thesis - Texas A&M University - 1992.

[12.] Bell, J.P., Journal of Polymer Science, A2 (6), 417, 1970.

[13.] Flory, P.J., Principles of Polymer Chemistry Cornell

University Press, Ithaca, New York, 1953.

[14.] Treloar, L.R.G., The Physics of Rubber Elasticity,, Oxford

University Press, New York 1958.

[15.] McCrum, N.G., Buckley, C.P. and Bucknall, C.B.,

Principals of Polymer Engineering, Oxford University Press,

New York 1988.

[16.] Tripathy, B.S. and Smith, K.C., Meeting of the ACS 

Rubber Division, Cleveland, OH, Oct. 21-24, 1997, Paper

No. 123.