18 Effect of long-term hydrogen exposure on the mechanical properties of polymers.pdf

International Journal of Pressure Vessels and Piping 89 (2012) 203 e 209

Contents lists available at SciVerse ScienceDirect

International Journal of Pressure Vessels and Piping

journal homepage: www.elsevier.com/locate/ijpvp

Effect of long-term hydrogen exposure on the mechanical properties of polymers

used for pipes and tested in pressurized hydrogen

Sylvie Castagnet * , Jean-Claude Grandidier, Mathieu Comyn, Guillaume Benoît

Institut P

(UPR CNRS 3346), CNRS-ENSMA-Université de Poitiers, Département de Physique et Mécanique des Matériaux, ENSMA, 1 Avenue Clément Ader, BP40109, 86961

Futuroscope cedex, France

’

article info

abstract

Article history:

Received 14 July 2010

Received in revised form

8 November 2011

Accepted 9 November 2011

The in

uence of long-term exposure to hydrogen on the mechanical properties of polymers needs to be

characterized for a reliable design of storage or transport facilities. However, mechanical tests in

hydrogen atmosphere have been rarely reported. In the present study, two possible effects of hydrogen

on tensile properties have been investigated in two polymers currently used for gas transport i.e.

polyethylene (PE) and polyamide 11 (PA11): the mechanics-diffusion coupling and the in

uence of long-

term exposure to hydrogen. Tensile tests in hydrogen atmosphere (30 bars) and atmospheric air at room

temperature were compared, in the as-received materials as well as after aging in various conditions

(pressure, temperature and duration). Results showed that the in

Keywords:

Polyethylene

Polyamide 11

Tension

Aging

Diffusion

Differential scanning calorimetry

uence of hydrogen was prevalent

neither on the tensile behavior nor on microstructure changes. This suggested that the design of

hydrogen-dedicated parts could be based on data obtained in atmospheric air, even for long-term use.

1. Introduction

polymer modi

es the macromolecular mobility and subsequently

the mechanical properties among which stiffness. Plasticization is

the best known example, whatever on purpose, by incorporating

a plasticizer into the polymer formulation, or as a spontaneous

process, for instance by water sorption into hydrophilic polymers

like polyamides [7 e 9] . Therefore, gas transport properties have

been widely investigated in various gas/polymer systems, among

which hydrogen [10] into PE or polyamide 11 (PA11) [11 e 14] .

Secondly, physical aging is bound to occur during long-term use,

mainly in glassy amorphous polymers but also in semi-crystalline

ones due to the high degree of constraint sustained by the amor-

phous phase connected to the crystalline skeleton [15,16] . Effects of

physical aging on the gas permeation properties were reported in

various gases [17 e 19] , showing a decrease of permeability but an

increase of selectivity. Thirdly, degradation processes may arise

from the exposure to chemically active gas or liquids. A dense

literature, beyond the scope of this paper, has been published about

the chemical and physical mechanisms associated with long-term

exposure to nocive environment like UV [19] or gamma [20] radi-

ation, but also to fuels [21] , weak acids like carbon dioxide [22] ,

chlorinated water [23] and oxygen [24,25] . Regarding the two

polymers studied in the present paper, most studies addressed

water hydrolysis [22] and oxidative induction times (OIT) [26] , with

a special interest paid on the kinetics and in

Some polymers like polyethylene (PE) have been used for gas

piping for a long time. In such applications, polymers are exposed

to methane and methane-based mixtures. For a few years,

hydrogen is being considered as a possible energetic vector, alter-

nately to fossil energies. A great deal of effort is being expended in

storage and networking issues for which materials properties

characterization is crucial. In metals, hydrogen embrittlement

processes are known to affect the strength and structural integrity

[1 e 3] . Several models coupling hydrogen diffusion and mechanics

have been proposed to analyze and predict hydrogen-induced

cracking [4 e 6] . The design of parts like pipes requires quantita-

tive data about the in

uence of pressurized hydrogen on the

mechanical properties on one side, and about the in

uence of long-

term exposure on the other side. However, such effects have not

been investigated a lot in polymers so far.

In a general way, three main phenomena are bound to be

addressed in polymers when used in a gas or liquid environment

for very long durations. Firstly, in the same way as temperature,

pressure or time, the diffusion of gas or liquid molecules into the

33 549 498 238.

E-mail addresses: sylvie.castagnet@ensma.fr , sylvie.castagnet@cnrs.pprime.fr

(S. Castagnet).

Corresponding author. Tel.:

33 549 498 226; fax:

uence factors of anti-

oxidants loss [25,26] .

0308-0161/$ e see front matter

doi: 10.1016/j.ijpvp.2011.11.008

204

S. Castagnet et al. / International Journal of Pressure Vessels and Piping 89 (2012) 203 e 209

uence

directly the mechanical properties, as a consequence of molecular

mobility and/or crystalline microstructure changes. This was often

tracked from Dynamic Mechanical Analysis (DMA) experiments,

however more in order to detect relaxation transitions [7,9 e 11 ]

than to quantify mechanical moduli. Except for a few works

based on

Each of the above depicted phenomena is bound to in

experiments, PE and PA11 samples were maintained in a 100% H 2

atmosphere during various times (from 1 to 13 months) at three

different temperatures (20, 50 and 80 C) and two differential

pressures between the sheet sides (5 and 2 MPa). Aging conditions

are listed in Table 1 . PA11 samples have been dried before the

permeation experiments but maintained at ambient humidity

between the end of permeation measurement and mechanical

testing.

exural loading or impact [8,21] , most studies were in

uniaxial tension [19,20] and widely focused on static properties and

ultimate properties (elongation at break, fracture mode) [8] . Most

of the time, mechanical tests were performed ex-situ, after aging or

exposure to gas atmosphere, and very rarely in liquid or gas directly

[27 e 30] . In particular, and despite the current interest for hydrogen

as a future energy, no data are available to evaluate the coupling

between mechanical behavior and hydrogen diffusion into ther-

moplastics, and the in

2.2. Differential scanning calorimetry

Standard DSC was used to characterize the microstructure.

Samples (around 10 mg) were sealed in aluminum pans and

continuously heated at 10 C/min under nitrogen sweep. The

percent crystallinity X DSC was calculated from Equation (1) :

uence on long-term exposure. However, due

to the drastic safety standard associated with hydrogen transport,

the design of hydrogen-exposed parts like pipes cannot be foreseen

without any guarantee about the possible in

H m

X DSC

100

(1)

uence of hydrogen on

both mechanical behavior and long-term aging.

This is the aim of the present paper. A tensile test machine was

H 0 m

where

H m is the energy required to melt the crystalline phase,

determined from the endothermic melting peak area, and

H 0 the

enthalpy of melting of the pure crystalline polymer (290 J/g for PE

[32] and 226.4 J/g for PA11 [33] ). The melting temperature was

determined as the maximum of the melting peak.

tted with a pressurized hydrogen chamber to address (i) the

possible coupling between hydrogen diffusion and tensile behavior

and (ii) the in

uence of long-term exposure to hydrogen, on the

mechanical response of semi-crystalline polymers. The work was

based on two polymers already used for natural gas transport: a PE

and a PA11. The pressure level (30 bars) was selected in accordance

with gas networking standards. Microstructure changes were

tracked by standard Differential Scanning Calorimetry (DSC).

2.3. Mechanical tests under pressure

tted with a pressurized-

hydrogen chamber displaying pressures from the ambient up to

40 MPa. The gas chamber could be

A tensile testing machine has been

2. Experimental

lled with nitrogen or hydrogen.

All tests were carried out at 3 MPa, in accordance with gas distri-

bution standard, and at

2.1. Materials

temperature, after equilibration

of the sample to the surrounding air temperature before the test.

However, due to safety standard linked to hydrogen explosibility,

the testing machine was located in a well-aerated building and thus

exposed to outdoor temperature variations from one day to

another. Once stabilized, the temperature variation during the

experiment was

“

ambient

”

Two semi-crystalline polymers currently used for gas transport,

a PE 100 and a PA11, have been tested in the as-received form

(1 mm-thick extruded sheets) and after long-term aging in a pres-

surized hydrogen atmosphere (samples named

the following). As determined by DSC (following the protocol

described below), PE exhibited a crystallinity ratio of 57% and

a melting temperature of 130 C. The alpha-c transition corre-

sponding to the onset of mobility of conformational defects in the

crystalline lamellae, usually located at T

“

hydrogen-aged

”

0.1 C.

Still due to safety reasons, the volume of the hydrogen chamber

was small (inner diameter 150 mm; deepness 100 mm) and

moreover reduced by the grips volume. Then, short enough

samples had to be tested to keep a suf

60 e 80 C, could not be

observed in DSC thermograms. In as-received PA11, the crystallinity

ratio was 22%, the melting temperature was 189 C and the glass

transition temperature was 49 C.

Before testing, as-received materials were conditioned at

ambient humidity. Tensile tests were performed in dumbbell

specimens machined in the extrusion direction with the geometry

summarized in Table 1 .

cient displacement range:

23 mm for the shortest specimens (i.e. a maximal conventional

strain of 125%) and 7 mm for the longest ones (i.e. a maximal

conventional strain of 15%).

The axial force F, axial displacement of the cylinder d, temper-

ature T and pressure P were monitored during the test. The axial

force was recorded from a 20 kN load cell located between the

cylinders. The load cell accuracy was 0.4% of the full range, i.e.

samples were machined

with a hollow punch from 1-mm thick extruded sheets used in

another part of the research program to perform hydrogen

permeation experiments [31] . During these prior permeation

“

Hydrogen-aged

”

10N. Tensile tests were carried out at a constant displacement

speed of the cylinder, i.e. at a constant conventional strain-rate. The

Cauchy stress

and the logarithmic strain

were calculated from F,

Table 1

Geometry and aging history (temperature, pressure and duration) of samples.

S. Castagnet et al. / International Journal of Pressure Vessels and Piping 89 (2012) 203 e 209

205

d, the initial gauge length of the sample l 0 and the initial cross

section area S 0 , assuming a constant volume deformation.

The Young

where D is the diffusion coef

cient of hydrogen into PE or PA11

and e is the sheet thickness. Coef

cients S g and D were deduced

from permeation tests performed in another part of the research

program [35] .S g values for PE at 40 C and PA11 at 60 C were

respectively 3.6.10 7 and 4.29.10 6 cm 3 of gas (STP).cm 3 of pol-

ymer.Pa 1 . Values for D were 1.68 10 10 and 4.54 10 11 m 2 .s 1 for

PE at 40 CandPA11at60 C respectively. 250 terms were

retained for the calculation. Finally, the estimated time for

hydrogen saturation of the entire specimen at 3 MPa was 1 h for

both materials.

Upon loading/unloading of the sample, the temperature stabi-

lized after about 40 min at a constant value (

’

s modulus was calculated from a linear interpolation

of the stress-strain curves between 0 and 1%/2% for PA11/PE

respectively. The yield stress was de

ned as the maximal force

point in softening curves and as the onset of the stress-hardening

stage in consolidating curves.

Samples were tightened between grooved grips and maintained

at nil force before testing. To avoid any dangerous mixture between

hydrogen and oxygen from the ambient air after closure of the

chamber, three successive pressurization/depressurization purging

cycles were performed

0.1 C). The ambient

temperature in the chamber was not regulated and varied from one

day to another between 16 and 29 C. PA11, which glass transition

temperature T g is around 40 C, was mostly bound to be sensitive to

such

rst, by introducing nitrogen up to 1 MPa.

Simultaneous temperature raises and decreases (by a few degrees)

were logically yielded. Then, the chamber was

lled with hydrogen

at a constant pressure rate of 0.6 MPa/min up to 3 MPa. The

temperature raise was rather quick at the beginning and progres-

sively equilibrated by the water circulating around the chamber

and the load cell.

A constant pressure stage was imposed before testing, in order

to equilibrate temperature (about 20 min long) and to complete

hydrogen sorption.

The time needed to reach hydrogen saturation of the sample

was estimated from a one-dimensional calculation through an

uctuations.

A constant cylinder displacement was imposed, corresponding

to a constant conventional strain-rate of 8.6 10 3 s 1 for as-received

samples and 1.6 10 2 s 1 for aged samples. The temperature kept

stable all along the mechanical test (

0.1 C). The chamber was

depressurized at a constant pressure rate of 0.6 MPa/min, yielding

a temperature decrease. For safety reasons again, the residual

hydrogen was removed by three nitrogen purging cycles at 1 MPa

before opening the chamber.

Similar tests have been carried out also in pressurized nitrogen

and air at atmospheric pressure. The experimental protocol for

nitrogen tests was very close to that applied for hydrogen, except

for the

nite sheet of constant thickness e, i.e. by assuming that the other

sample dimensions were much higher than the thickness and by

neglecting diffusion through the lateral faces. Therefore, the

calculated saturation time resulted overestimated. The hydrogen

concentration at the top and bottom surfaces of the sheet C

H2 was

nal purging step which was needless. In this way, the

mechanical history was the same for experiments under hydrogen

and nitrogen. Sorption calculations with diffusion and solubility

coef

’

deduced from a Henry

s law given by Equation (2) :

C N H2

P ext S g

(2)

cients for nitrogen into PE [14] showed that the saturation

regime was also reached after the 1-h step de

ned for the hydrogen

where P ext stands for the pressure of the surrounding gas and S g

represents the solubility, assumed to be temperature independent.

Concentration is expressed as a volume ratio between the sorbed

gas and the polymer (given in cm 3 (STP).cm 3 where STP stands for

Standard conditions of Temperature (STP) (273K) and Pressure

(0.1013 MPa). The initial hydrogen concentration within the sheet

was assumed to be zero. The current global concentration C(t)at

any time t was obtained from the analytical solution in Equation

(3) , based on Fick

protocol.

The reproducibility was estimated as a standard deviation

from the mean value by repeating the same experiment between

3 and 6 times for each material and loading condition. The

reproducibility of experiments in ambient air (including the

experimental device precision, the dimensional variability of

samples and material heterogeneity) was estimated from tests

performed in as-received PE.

The

axial

force

scattering

’

s theory and proposed by Crank [34] .

measurement over 6 tests was

3.1%. A higher scattering was

expected for PA11 due to the room temperature and moisture

content variability; it was discussed in the following. The

experimental error increased for pressure experiments due to

sliding effects between cylinders and o-rings: the axial force

e 2 !

2 exp

2 N

2 Dt

Þ¼

(3)

Fig. 1. In

’

uence of the room temperature variability on the (a) Young

s modulus and (b) yield stress in as-received PA11.

206

S. Castagnet et al. / International Journal of Pressure Vessels and Piping 89 (2012) 203 e 209

Fig. 2. Stress-strain curves obtained in PE in atmospheric air and 3 MPa hydrogen, after various aging histories (see Table 1 ).

scattering reached

10% for PE (4 tests). A more precise

measurement of the mechanical load under hydrogen pressure

would require larger specimens not really compatible with the

small volume of the chamber.

3.2.

uence of long term aging in pressurized hydrogen

uence of aging parameters like

temperature or hydrogen pressure on the tensile properties,

coupled or not to hydrogen diffusion. To this aim, series of samples

aged for several months in the conditions listed in Table 1 have

been tested both in atmospheric air and pressurized hydrogen

(3 MPa). Aging temperatures were 20 Cor80 C (meaning below

and above T g for PA11 respectively; and below and above T

The goal was to evaluate the in

3. Results and discussion

3.1. Coupling between hydrogen diffusion and tensile properties in

as-received materials

a c for PE

respectively), except for a few tests performed at an intermediate

temperature of 50 C.

Fig. 2 plots the tensile stress-strain curves recorded for aged PE

samples. Let us focus

rst series of tensile experiments were performed in as-

received materials at ambient air, and under 3 MPa of hydrogen

and nitrogen.

Concerning PE, differences between curves obtained in various

environments could not be dissociated from the experimental

scatter and no signi

uence of hydrogen diffusion on

the tensile behavior, by comparing tests performed in atmospheric

air and 3 MPa hydrogen for each aging history. After annealing

above T

rst on the in

c (V1 and V3), ambient air curves were slightly above

hydrogen curves, unlike after annealing below T

cant pressure effect could be evidenced. The

Young ’ s modulus was 954 74 MPa (over 7 tests) in atmospheric

air, 972

c (V2, V5 and V6).

However, differences were very

ne, consistently with the scatter

depicted above in as-received materials. DSC thermograms in PE

samples after aging for 13 months at 20 C (V2), 50 C (V5) and

80 C (V3) were compared in Fig. 3 . The corresponding crystallinity

ratio was 56.8, 57 and 60.5% respectively and the melting temper-

ature equaled 130.2, 131.1 and 131 C respectively. Unsurprisingly,

annealing at increasing temperature led to better crystallized

materials but consequences on the stress-strain behavior were

minor than the experimental scatter. Any effect of the aging pres-

sure could have aroused from the comparison of V2 (2 MPa) and V6

57 MPa (over 4 tests) under 3 MPa of hydrogen and

978

37 MPa (over 5 tests) under 3 MPa of nitrogen. The yield

stress was respectively 27

1 MPa, 26.6

1.6 MPa and

27.8

1.1 MPa. It means that, if existing, the hydrogen in

uence on

the tensile behavior of PE at 18 C did not exceed 10%.

Data obtained in PA11 with the same sample geometry were

even more scattered. The same average analysis as conducted

above in PE led to a Young

’

s modulus of 845

115 MPa (over 10

tests) in atmospheric air, 905

130 MPa (over 5 tests) under 3 MPa

of hydrogen and 915

90 MPa (over 6 tests) under 3 MPa of

nitrogen. The yield stress was respectively 34.3

5 MPa,

36.7

4.3 MPa. Like for any polymer close to the

glass transition temperature T g , the mechanical properties of PA11

were expected to vary with room temperature (i.e. around 30 C

below T g ). Moreover, polyamides are known to be hydrophilic and

mechanically dependent on the water weight content. None of

these two factors were regulated in the present study and the latter

one could not even be accessed. However,

5.9 MPa and 37

uence of

temperature effect could be examined, as proposed in Fig. 1 by

plotting the Young

the in

y against the

ambient temperature measured for each test. Both E and

’

s modulus E and the yield stress

expectedly dropped with the temperature raise and evolutions

appeared reasonably correlated to the testing temperature.

clearly evidenced a major in

uence of the testing temperature on

the tensile stress scattering, compared with any pressure or envi-

ronment effect. In the same way as previously established for PE,

coupling between hydrogen diffusion and tensile behavior

appeared to be negligible.

rst heating run; 10 C/min) in PE samples aged for 13

months at 20 C (V2), 50 C (V5) and 80 C (V3).

Fig. 3. DSC thermograms (

S. Castagnet et al. / International Journal of Pressure Vessels and Piping 89 (2012) 203 e 209

207

Fig. 4. Stress-strain curves in PA11 tested in atmospheric air and 3 MPa hydrogen after aging at various temperatures, durations, pressures and atmospheres.

(0.5 MPa) at 20 C but the two series were too close together

considering the tests

pressure (0.5 MPa) and after 13 months, aging at 80 C resulted in

a stiffer behavior than aging at 20 C. Such a result was pointed out

for PE. In the same way as for PE under 2 MPa, and despite variable

aging times, all V2 (20 C; 13 months) and V1 (80 C; 9 months)

curves laid within the experimental scatter.

In many ways, trends did not appear very convincing regarding

the large scatter previously pointed out in as-received specimens.

Furthermore, correlation to the ambient temperature was worth

being examined here again, as displayed in Fig. 5 . Both E and

reproducibility. Concerning the aging

temperature in

uence, the stress level after aging at 0.5 MPa and

80 C (V3) was rather clearly higher than after aging at 20 C (V6).

However, this difference was no more observed after aging at

2 MPa: all curves overlapped over a limited scatter range regardless

the aging temperature of 20 C (V2), 50 C (V5) or 80 C (V1). These

results supported the weak in

uence of hydrogen diffusion

coupling and aging processes.

The same series of tests in PA11 were presented in Fig. 4 .

Considering the wide scatter pointed out in as-received PA11, no

evidence was brought from the comparison of tests performed in air

or hydrogen after the same aging protocol. This emphasized the

minor coupling effect previously pointed out in as-received mate-

rials. The aging time in

s y

monotonically decreased for increasing room temperature

regardless aging conditions. Here again, the principal in

uence

dealt with ambient temperature variability. Fig. 6 a showed the

corresponding

rst heating run of DSC experiments carried out

after one month at 80 C in air (atmospheric pressure), vacuum and

hydrogen (3 MPa). Aging in vacuum aimed at separating the

oxidation effect bound to affect PA11 when annealed in air at high

temperature. Firstly, the glass transition could not be detected very

precisely, due to the overlapping endothermic peak yielded by the

physical aging occurring between the end of the aging protocol and

the DSC experiment. It seated at about 45 C for all samples.

Secondly, another endothermic peak could be distinguished at

around 90 e 95 C: it is linked to the annealing at 80 C [8] . Finally,

the melting peak of primary crystals was not signi

uence could be addressed by comparing

1-month (V4) and 9-months (V1) aging at 80 C and 2 MPa of

hydrogen: the scatter between the two curves was not signi

cant

again. The in

uence of hydrogen pressure during aging was also

negligible, as deduced from the comparison between 2MPa (V2) and

0.5MPa (V6) experiments at 20 C. The V2 curveswere slightly above

the V6 ones but the difference results less noticeable than previously

observed for PE. Let us consider

nally the aging temperature effect.

Unlike in PE, only two series of curves (V3 andV6) could be compared

to keep constant the aging pressure and time. Under the same aging

ed.

The melting temperature after aging in vacuum (190.4 C) was

cantly modi

Fig. 5. Relationship between (a) Young

’

s modulus, (b) yield stress and room temperature in PA11 samples after various aging protocols in hydrogen (temperature, pressure,

duration).

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: