Lili
Xu
,
Cheng
Huang
,
Mingchao
Luo
,
Wei
Qu
,
Han
Liu
,
Zhewei
Gu
,
Liumei
Jing
,
Guangsu
Huang
* and
Jing
Zheng
*
College of Polymer Science and Engineering, State Key Laboratory of Polymer Materials Engineering, Sichuan University, Chengdu 610065, China. E-mail: GuangsuHuang@scu.edu.cn; zhengjing@scu.edu.cn
First published on 24th September 2015
Phospholipids and proteins were separately removed as the main non-rubber components to individually study their effect on the structure and properties of the rubber. Fourier Transform Infrared Spectroscopy (FTIR) and 1H nuclear magnetic resonance spectroscopy (1H-NMR) were used to characterize the chemical structure and the residual non-rubber component. A rheology study and stress relaxation measurements were used to study the role that non-rubber components play in natural networks. A rheological study showed that natural rubber (NR) and deproteinized natural rubber (DPNR) exhibited similar dynamic modulus at 170 °C. The lack of superposition in van Gurp–Palmen (vGP) curves at different temperatures for NR and DPNR, together with the shape of vGP curves, proved that long chain branching was mainly constructed by phospholipids. Stress relaxation measurements at room temperature were fitted with the Maxwell model and showed that the NR relaxation curve underwent a quick decrease and then came to an equilibrium stress retention of 58%, which is about 3 times higher than that of DPNR, indicating that proteins in NR contributed to the network structure at room temperature. Combining molecular dynamic studies, the interaction of proteins and phospholipids in non-rubber networks was proposed.
As a natural material, the yield of natural rubber which is seriously restricted by land resources and natural conditions, has become the fatal bottleneck for its extensive application. Thus, the synthetic analogue polyisoprene as the best alternative, is modifed mainly by various molecular design methods,8–13 however the properties were disappointing.
It has been reported that non-rubbers,14 especially proteins and lipids, give natural rubber latex excellent properties which are unsurpassed by any synthetic latex.15 As a consequence, great industrial and scientific attention has been paid to the structures of non-rubbers and their importance on the structure of natural rubber to create the magical mechanical performance,16 such as the high strength of raw rubber (green strength), shape retention, high tensile strength, high crack growth resistance and minimal heat build up in the vulcanized state. Several attempts have also been made to study non-rubber components and investigate their corresponding structures in natural rubber. Y. Tanaka17,18 first proposed that cis-polyisoprenes obtained from H. brasiliensis and Parthenium argentatum consist of more than 5000 isoprene units and that there are trans-isoprene units per rubber chain. The mainstream view now is that NR molecules comprise 2 trans-isoprene units connected to a long chain of cis-isoprene units. Two terminal groups, referred to as α and ω, have been postulated to link with mono- and di-phosphate groups associated with the phospholipids by H-bonding at the α-terminal, whereas the ω-terminal is postulated to be a modified dimethylallyl group linked to the proteins by H-bonding.19 Then the non-rubbers can connect the linear polyisoprene chains in NR through functional terminals and generate a branching or network topology, which is designated a “naturally occurring network”.20 But since the Goodyear medal19 in 2001, no clear evidence of the structure of an end-linking network has been reported.
Although the real cross-linking structure and role of non-rubbers are not fully understood, it has been suggested that the cross-linking of NR formed by non-rubbers can be eliminated by deproteinization and transesterification. After the removal of proteins and phospholipids, linear rubber chains can be obtained.21
The effect of proteins and phospholipids on strain induced crystallization and the correlation with the tensile strength of NR have been extensively studied using X-ray diffraction and microscopic methods.22,23 However, spectroscopy methods are static and mostly accompanied by solvent, and X-ray characterization is indirect in “naturally occurring” network characterization. Rheological properties are highly sensitive to long chain branching because even a low degree of long chain branching has a significant effect on the viscoelasticity in polyethylenes23–28 or polypropylene.29 A van Gurp–Palmen curve, which was first proposed to judge the feasibility of Time–Temperature Superposition (TTS) principles and then evolved to branched structure studies independent of the chemical components of the macromolecule chain, is also very sensitive to the long chain branching of the polyisoprene melt in unvulcanized NR.30 Therefore, the rheological behavior of unvulcanized NR can be characterized using a dynamic spectrum, related to the molecular structure in the melt, which is strongly impacted on by branching or networking. However, to our knowledge, there is a lack of a systematic investigation on how non-rubber components affect the rheological behavior of rubber chains in melts and inversely testify the role of non-rubber components from rheology, especially the effect of long chain branching on rheological properties. While few rheological studies were conducted on NR, among which most focused on blended NR with plastic,31 silica,32,33 carbon black34 or another natural ingredient35 in a vulcanized regime, hardly any were on unvulcanized rubber. While for unvulcanized NR, the studies mainly focused on processability36 or nonlinear viscoelastic behavior.37 In this study, we conducted rheological studies on an unvulcanized NR melt, and introduced the widely used vGP curve to study the effect of non-rubber components on unvulcanized NR.
In addition, the stress relaxation behavior also provides important guidance in probing the network, and additionally the equilibrium relaxation modulus G∞ is proportional to the crosslinking density.28 The stress relaxation behavior of natural rubber is often investigated for vulcanized rubber,29–31 while scarcely for the unvulcanized if any. In this study, to investigate the network structure constructed by non-rubbers in unvulcanized NR, we introduced a stress relaxation test and analyzed the equilibrium relaxation modulus to evaluate the crosslinking density and the constraints on relaxation units.
In this work, an attempt was made to clarify the “naturally occurring network” through a molecular dynamics study, combining the chemical structure analysis. Fourier Transform Infrared Spectroscopy (FTIR), rheology and stress relaxation were used to analyze the composition and structure of the natural network. To individually study the effect of the non-rubber components, three models were prepared to individually study the effect of non-rubber components which are listed as follows, NR containing all of the non-rubber components, deproteined natural rubber (DPNR) containing phospholipids and fatty acids, transesterified deproteinized natural rubber (TEDPNR) containing only rubber chains.
Measurement of the nitrogen content of the deproteined natural rubber (DPNR) was made using a nitrogen elemental analyzer (CARLO ERBA 1106, Italian), and the nitrogen content was reduced from 0.5% to 0.07% after deproteinization.
Measurement of the phosphorus content of TEDPNR was made using a phosphorus elemental analyzer (IRIS 1000 ICP-AES, Thermo Electron Co. USA), and the phosphorus content was reduced from 224 ppm to 65 ppm after transesterification.
The molecular weight of all samples was determined using gel permeation chromatography (HLC-8320, Waters), and the number averaged molecular weight was about 150–160 kg mol−1 for NR and DPNR, 134 kg mol−1 for TEDPNR, while the MW of commercial linear isoprene rubber (IR) was 317 kg mol−1.
First, a wide H-bond peak at around 3400 cm−1 was absent in TEDPNR. For NR the wide peak comprised a relatively sharp peak at about 3300 cm−1 related to the N–H symmetric stretch of the H-bond, and the wide H-bond peak is mainly ascribed to the O–H symmetric stretch in the H-bond in non-rubbers, overlapping with a small amount of environmental water absorbed by non-rubbers.
Second, to pick out the non-rubbers, the assignment of the rubber chain peaks are listed below. vmax/cm−1 = 836 (trisubstituted olefin out-of-plane CH, wag), 1129 (–CH3, rock), 1300 (–CH2–, wag), 1376 (–CH3, symmetric deformation), 1450 (–CH2– symmetric and –CH3 asymmetric deformation), 1664 (C
C, stretch), 2720 (overtone of –CH2– umbrella), 2850 (–CH2– and –CH3, symmetric stretch), 2920 (–CH2–, asymmetric stretch), 2962 (–CH3, asymmetric stretch), and 3030 (olefin
CH– stretch).33,34
The recession of the characteristic protein peaks at 1663 and 1546 cm−1 (ref. 35) is observed in DPNR compared with NR, and the CO phospholipid peaks at 1738 cm−1 (ref. 36) in TEDPNR. The FTIR results, in combination with the elemental analysis in Section 2, proved that the intended samples were successfully prepared. As for the different intensity of the C
C stretch peak in the normalized infrared spectra, it can be interpreted as occurring due to various infrared activities with different non-rubber environments.
The shoulder peak at 1260 cm−1 in NR was shifted to 1248 cm−1 and became sharp after protein removal, continually shifting to about 1243 cm−1 and receding rapidly after phospholipid deprivation. It has been reported that the asymmetric O–P–O vibrational band is extremely sensitive to hydration, and the O–P–O asymmetric stretching of the hydrated phospholipid bilayer is usually observed at 1230 cm−1, whereas dried or anhydrous lipid always appears at a 30 cm−1 higher frequency.37 The shift of the O–P–O asymmetric stretching peak strongly suggests that the phospholipids in NR interacted with the protein via H-bonds, and in DPNR rubber they aggregate or link together via H-bonds.38
Combining the 1H-NMR and FTIR results, we can conclude that the intended samples were successfully prepared.
All measurements were conducted at a fixed strain, ranging from 2% to 4% at 170 °C where all samples display a linear viscoelastic performance, and the storage modulus G′ as a function of frequency is shown in Fig. 3. At such a high temperature, most H-bonding or other non-bonding interactions were destroyed at this temperature.
NR and DPNR show a higher modulus at low frequency than TEDPNR, indicating the existence of long-chain branching, crosslinking or phase-separation.40,41 However, the modulus of NR is a little higher than that of DPNR. DPNR displayed no apparent slope change at low frequency, while the modulus increased by as much as 2 orders of magnitude in contrast with TEDPNR.
To further investigate the modulus change, dynamic spectra at different temperatures were obtained for all of the samples, and van Gurp–Palmen plots were drawn using dynamic shear modulus values at three different temperatures (160, 170, and 190 °C), displayed in Fig. 4. Generally, very good superposition of the vGP curves was observed for classical viscoelastic polymers, while a small amount of long chains can significantly distort the vGP curve and thus cause a misalignment.42,43 The rubber models were confirmed to be thermo-rheological melts from the discrepancy of the vGP curves for NR and DPNR in Fig. 4, possibly with long-chain branching or phase-separation.
To judge the reason for the discrepancy in the vGP curves, a temperature sweep from 100 °C to 200 °C at a frequency of 1 Hz at 1% strain was conducted. As is known, phase-separation can cause a sharp uprush in the modulus of the temperature sweep.44,45 But here, the results show no uprush in the dynamic modulus in Fig. 5, indicating that no phase separation occurred in the testing temperature region. As is shown in Section 2, all samples were unvulcanized, so we can exclude crosslinking. After excluding phase-separation, the modulus shift of NR and DPNR in Fig. 3 is attributed to long chain branching, and a small amount of protein involved in forming long-chain branches.
In contrast to NR and DPNR, the overlap of the vGP curves at the various temperatures is good for TEDPNR, indicating that TEDPNR is a simple fluid. From the shape of the curve – only one minimum, spread monotonically to 90° in the phase angle – we can be conclude that TEDPNR was probably linear. Combining the good linearity of the Han map46 in Fig. 6, we can conclude that the curve for TEDPNR was linear at high temperature, in accordance with the dynamic spectrum.
The above rheological results showed that long-chain branching existed only in the NR and DPNR melt, indicating that phospholipids and a small amount of protein acted as the branching point in the long-chain branching construction. Furthermore, we can conclude that phospholipids and a small amount of protein connected with the rubber terminal point by chemical bonds or very strong interactions (which survive at 170 °C), then the non-rubbers coagulate through micellar interaction,47 and possibly formed the star structure in the melt.48
In this study, the stress relaxation curves of NR, DPNR and TEDPNR at 25 °C are displayed in Fig. 7, where all of the samples were quickly subjected to a strain of 20% to avoid the influence of strain induced crystallinity (SIC),49 and then the strain was maintained for 1200 s. For the sake of contrast, a relaxation curve of fairly linear IR with a 317 kg mol−1 average molecule weight (MW) was added. Generally, the stress of linear viscoelastic polymers quickly relaxed to a very small value, and even to zero.
To illustrate the effect of non-rubbers on the stress relaxation tendency, all of the instant stress values were normalized using the corresponding initial stress, as is displayed in Fig. 7. The relaxation curves of NR tended to form a plateau after about 300 s and the retention is more than 50%, the largest stress retention, indicating that protein mainly contributed to the natural network at room temperature. Unexpectedly, DPNR produced nearly the same behavior as IR, but relaxed slightly slower than linear IR. TEDPNR relaxed to zero during the testing period. It can be confirmed that the relaxation units in NR were obviously confined by a chemical or strong physical network. While in DPNR, in contrast to NR the network constraint was scarcely observed. However, in comparison with linear IR, we can conclude that the restriction derived from long chain branching. TEDPNR displayed typical relaxation behavior of linear molecular chains with fully relaxed stress.
To further confirm the above assumption, we adopted the Maxwell model41,50,51 to describe the stress relaxation of rubber, as shown in eqn (1):
![]() | (1) |
In eqn (1), σ(t) represents the stress in the relaxation, σe is the equilibrium stress, σi is the coefficient of the ith Maxwell model and τi is the relaxation time of the ith Maxwell model.
Elastomer relaxation followed a 7-element Maxwell model,52 as shown in eqn (2):
![]() | (2) |
Divided by the initial stress σ0 on both sides, eqn (2) may then be inverted to read:
![]() | (3) |
Let
![]() | (4) |
Then we fitted the experimental data from eqn (3) and the results are listed in Table 1.
Sample | A 0 | A 1 | T 1 | A 2 | t 2 | A 3 | t 3 | R 2 |
---|---|---|---|---|---|---|---|---|
NR | 0.58507 | 0.06667 | 9.87772 | 0.13743 | 57.85191 | 0.18798 | 364.7487 | 0.9876 |
DPNR | 0.15905 | 0.09697 | 12.62921 | 0.20228 | 82.65005 | 0.51581 | 544.9699 | 0.99534 |
IR | 0.06292 | 0.13534 | 17.76458 | 0.27497 | 96.31696 | 0.50153 | 561.9759 | 0.99674 |
TEDPNR | −0.09715 | 0.21816 | 16.88401 | 0.33616 | 115.2694 | 0.48798 | 740.3617 | 0.9829 |
From eqn (4), we can see that Ai represent the contribution of ti, and ti corresponds to the different relaxation unit in the samples. The values of A1, A2 and A3 increased in the same samples with increasing the magnitude of the relaxation times t1, t2 and t3. The consistency of Ai and ti proved that the larger relaxation unit contributed to larger stress relaxation.
Second, the crosslinking density was discussed through A0. As is shown in Table 1, NR displays the largest A0, quadruple that of DPNR, while the A0 of DPNR is also higher than that in linear IR, and TEDPNR shows a meaningless negative A0.
From eqn (4) we can see that A0 represents the stress retention rate, which is proportional to the equilibrium relaxation modulus G∞ and crosslinking density, regardless of being chemical or physical. Then we can conclude that NR possessed the largest cross-linking density, and the long chain branches in DPNR constructed a small amount of a strong entanglement network. From the crosslinking density value which is 3 times higher, we can also conclude that part of the network structure in NR is connected by interaction between proteins. Combined with the FTIR study, we can conclude that the interaction between proteins and phospholipids can also contribute to the natural network, in accordance with previous studies.53,54
For the linear IR, the super-high MW contributed to a strong entanglement network, but was still weaker than in DPNR. While in TEDPNR, the linear topology and lower MW results in the absence of a strong entanglement network.
Then the relaxation time was also considered. The relaxation time of all of the samples is displayed in the histogram in Fig. 8.
As is shown in Fig. 8, the relaxation times decrease in this order: NR < DPNR < IR < TEDPNR. In vulcanized rubber, the relaxation time decreases with increasing crosslinking density, which can be interpreted as the crosslinking point restriction.38 Similarly, in the unvulcanized rubber, the relaxation units were also constrained by the natural network, and then displayed a decrease in relaxation time. The results confirmed that the crosslinking density increased in the order: NR > DPNR > IR > TEDPNR, which agreed with the previous study of A0.
From the above results, we proposed the non-rubber in the structures of NR and DPNR shown in Scheme 1. In NR, the protein is connected in a ω-terminal and then coagulated with each other through H-bonds, part of which can also interact with phospholipids through H-bonds. Moreover, in the α-terminal, mono- or di-phosphate is bonded with the rubber chain, and then connected by a micelle cluster. While in DPNR, the phospholipids or protein fragments after the protease treatment can also interact with the ω-terminals by polar–polar interactions or H-bonds, together with the α-terminals being connected to phospholipids to form the network, whose density is much lower than that in NR.
This journal is © The Royal Society of Chemistry 2015 |