Printer Friendly

Correlation Between Inhomogeneity in Polyacrylonitrile Spinning Dopes and Carbon Fiber Tensile Strength.


High performance carbon fibers are being increasingly used in structural composite applications due to their high specific tensile strength and specific tensile modulus as compared with conventional materials such as steel. Carbon fiber composites find applications in several sectors such as aerospace, nuclear engineering, transportation, sporting goods and biomedical engineering [1,2]. The latest civilian and military aircrafts such as the Dreamliner and French Rafael would not have become reality without carbon fibers [1], Presently, the state-of-the-art carbon fibers certified by Federal Aviation Administration for aerospace applications such as IM7 possess tensile strength and tensile modulus of 5.6 and 276 GPa, respectively [3], There have been reports of commercial carbon fibers with various combinations of high tensile strength and high tensile modulus, such as T1100 G carbon fiber from Toray and 1M10 carbon fiber from Hexcel with tensile strength and tensile modulus combinations of 7.0, 324 [4], 7.0, and 310 GPa, respectively [5]. We have reported successful manufacture of carbon fibers at laboratory scale, via processing a 100 filament polyacrylonitrile (PAN) precursor tow through a continuous carbonization line, with tensile strength and tensile modulus of up to 5.8 and 375 GPa, respectively [6] Although there has been considerable enhancement in the tensile strength of carbon fibers from about 3.5 GPa in T300 type of carbon fibers to up to 7 GPa in IM10 carbon fiber over last few decades, the highest tensile strength achieved is <10% of the theoretical strength of carboncarbon bond [7,8]. PAN is the predominant precursor for high tensile strength carbon fibers. The key parameters to obtain high tensile properties in carbon fibers are (a) to maximize molecular alignment along the fiber axis and (b) to minimize defect structures arising from surface defects, polymeric chain entanglements, voids and chain ends [9]. The structural defects in PAN fibers can be reduced significantly by minimizing the inhomogeneity in solutions due to polymeric chain entanglements, gelation, microbubbles and contaminants. We have reported use of dynamic shear rheology (DSR) [10] and DLS [11] to characterize inhomogeneity in the PAN solutions. The assessment of inhomogeneity of polymer solutions using DSR [12-15] and DLS [16] is well established. In DSR, the inhomogeneity of polymer solutions or melts is correlated to the slope of log G' (storage modulus) versus log G" (loss modulus) [12-15], In DLS, the size of the polymeric chain entanglements is derived from the diffusion coefficients of the slow relaxation modes of polymer solutions [16].

Herein, we report the use of mean of the count rate (MCR) from DLS to detect microscale inhomogeneity in PAN solution caused by microscale contaminants such as foreign particles. The nanoscale inhomogeneity representing the size of polymeric chain entanglements in the spinning dope was also measured. We further show the correlation between inhomogeneity in PAN solution detected using DLS with the tensile strength of carbon fibers. Finally, we demonstrate for the first time a surfactant assisted purification method using Triton X-100 and bis(2-ethylhexyl) sodium sulfosuccinate, to remove contaminants in PAN powder, and to reduce inhomogeneity in the spinning dope.


Polyacrylonitrile-co-methacrylic acid polymer with a viscosity average molecular weight of 513,000 g/mol (MAA content of 4 wt%) was purchased from Japan Exlan Co. Ltd. (Osaka, Japan). Chemicals, dimethyl formamide (DMF) (HPLC grade), dimethyl acetamide (DMAc) (HPLC grade), Triton X 100 (TX 100; 98%) and bis(2-ethylhexyl) sodium sulfosuccinate (AOT, 98%) were purchased from Sigma Aldrich.

As received PAN polymer was dried under vacuum at 75[degrees]C. PAN solutions were prepared with solids concentration in the range of 10-10.5 g/dL of solvent in two kinds of reactors, a 6 L glass reactor and 1 gallon Parr reactor. Two types of 1 gallon vessels were used in Parr reactor, Parr VI and Parr V2. Difference between these two vessels was that V2 had glass window to observe the solution whereas vessel VI was without glass window. Details of solution preparation conditions are provided in Table 1. Distilled DMF or DMAc was used as the solvent in PAN solution preparation. PAN-1 solution was prepared using DMAc as the solvent whereas PAN-2 to PAN-8 solutions were prepared using DMF as the solvent. Solution preparation and fiber manufacture was carried out in class 1000 cleanroom to minimize the incorporation of airborne particles in the solution and fiber during processing. The maximum stirring speed used was 150 and 700 rpm in the glass and Parr reactors, respectively. Dilute PAN solutions at 0.1 and 1 wt% were prepared in a class 1000 cleanroom by slowly adding the appropriate quantity of pre-dried PAN polymer into DMF and the mixture was heated to 70[degrees]C while stirring. One liter glass reactor was used to prepare the dilute solutions. Solutions were stirred using a mechanical stirrer set at 150 rpm.

Fiber spinning was carried out using the equipment supplied by Hills Inc, FL. The spinneret containing 100 holes of 200-[micro]m capillary diameter was used in all the spinning trials except for solution PAN-2, in which case a spinneret with 100 holes of 150-[micro]m diameter capillary

was used. 100% methanol maintained at -50[degrees] C was used as the gelation medium in all the trials. Stabilization and carbonization of the precursor fiber was carried out in class 1000 cleanroom. Stabilization was carried out under tension in the oxidative environment at temperatures in the range of 180-290[degrees]C. Low temperature carbonization was carried out in nitrogen atmosphere at temperature up to 1,000[degrees]C. High temperature carbonization was carried out in nitrogen atmosphere at temperature of 1450[degrees]C. Further details of fiber spinning, fiber drawing and carbonization are provided in our earlier publications [6,10], The scanning electron microscopic (SEM) images of the representative carbon fibers is shown in Supporting Information Fig. S1.

Rheological behavior of spinning dopes was studied using dynamic frequency sweep on parallel plate rheometer (ARES, TA Instruments, Co.) using 50 mm plates and a gap of 1 mm at room temperature. Upon sample loading, a thin layer of silicone oil was applied at the exposed surface of solution to prevent solvent evaporation and gelation by moisture absorption. The DLS measurements for PAN polymer solutions were carried out at 25[degrees] C using a multi-angle DLS Brookhaven instrument with BI-200SM/TurboCorr scattering system, and with excitation laser of wavelength 514 nm. The microscale inhomogeneity was the mean of the photon count rate (MCR) of PAN solutions in DLS. In DLS measurements, the time fluctuation of the scattered light intensity, Is(t), of a dynamic system is recorded, from which the time correlation functions, g1([tau]), of the scattered electric field is accordingly determined [17] and used for extracting the diffusion coefficients of the polymer entanglements from the slow relaxation modes in PAN solutions using the well-known CONTIN method [18,19]. The nanoscale inhomogeneity (hydrodynamic diameter) was calculated from this diffusion coefficient by applying Stokes-Einstein relation (Eq. 1) as reported elsewhere [11].

d = kT/3[pi][eta]D (1)

where, d is the hydrodynamic diameter, k is Boltzmann's constant, T is absolute temperature, [eta] is zero shear viscosity, and D is the diffusion coefficient.

The diffusion coefficient can be determined by dynamic light scattering (DLS) measurement and is given by Eq. 2.

D = [GAMMA]/[q.sup.2] (2)

where [GAMMA] is the delay rate of the relaxation mode at a scattering vector q. q is related to the length scale that can be probed [20] and is inversely proportional to the wavelength of the laser and directly proportional to the sine of the one half of the scattering angle.

An optical microscope from Leica was used to obtain optical micrographs of solutions. Zeiss Ultra 60 FE-SEM was used to obtain micrographs and to analyze elemental composition via energy-dispersive X-ray spectroscopy. Differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) of the solutions was performed using DSC Q100 and TGA Q500 from TA instruments, respectively. Details of evaluation of carbon fiber diameter and tensile properties are provided in our earlier publication [6], Tensile testing was carried out on single fibers using Favimat at gauge length of 25.4 mm and at strain rate of 0.1 %/s. Average of 50 tests is reported.

Surfactant assisted purification of as-received PAN powder was carried out. A 90 g of as-received PAN powder was dispersed in 100 mL of aqueous micellar solution containing 2 wt% TX-100. For a period of 48 h, the mixture was stirred and purged with nitrogen gas to generate the foam to remove lighter materials. Foam generated was removed and water lost through foaming was compensated every ~8 h. The dispersion of PAN powder in deionized (DI) water in the presence of TX-100 was centrifuged. The residue was redispersed in ~1 L of DI water. The process was iterated 20 times, until no visible foam was seen on shaking the dispersion. The PAN powder was dried in an oven, at 45[degrees]C in vacuum for 2 days, then at 70 and 90[degrees]C on third and fourth day, respectively. The dried PAN powder was redispersed in 100 mL of hexane containing 2 wt% AOT reverse micelles. The PAN dispersion in hexane was centrifuged and hexane was decanted. The AOT was washed in the same manner as detailed for TX 100 using DI water. The recovered PAN powder was dried in the manner similar to that of the PAN powder recovered after purifying with TX-100.


The particles larger than the wavelength of the laser light that come in the path of the laser can cause large fluctuations in the overall scattering intensity. The fluctuations in the intensity of the scattered laser light after passing through unfiltered 0.1 wt% PAN solution, measured as photon count rate is shown in Fig. la. The large fluctuations in the scattered light, are due to the presence of inhomogeneities such as micro-bubbles, PAN gel, and foreign particles, and so on. The latter mentioned systems (PAN gel and foreign particles) henceforth will be collectively referred to as contaminants. The presence of contaminants in unfiltered PAN solution was observed in optical micrograph and SEM micrograph (see Supporting Information Fig. S2). The 0.1 wt% PAN solution after filtration through 1 pm PTFE membrane (Fig. lb) showed substantially lower fluctuations, indicating size of the contaminants is of the order of microns.

The MCR has been used to characterize micelles in the solution [21], But in the case of micelles, (a) the viscosity of the micellar solution is similar to that of the water, (b) the size of micelles in solution is usually a few nanometers, and (c) micellar solutions are expected to be free from contaminants. However, for polymer solutions suitable for fiber spinning, the viscosity is usually several orders of magnitude higher than that of the solvent [16], and the contaminants are few micrometers in size. Therefore, it is necessary to first ascertain the conditions under which MCR can be used to characterize the inhomogeneity in the solutions. After extensive experiments (see section on MCR as a tool to characterize inhomogeneity in Supporting Information Fig. S3), we have ascertained that MCR is proportional to inhomogeneity for solutions with similar viscosities.

Average carbon fiber diameters were in the range of 5.2 (0.5) to 5.6 (0.4) [micro]m for all the samples except for sample 2, of which the average carbon fiber diameter was 4.5 (0.2) [micro]m (The numbers in parentheses are the standard deviations of fiber diameters. The plot of microscale and nanoscale inhomogeneity for various spinning dopes measured as MCR and size of nanoscale inhomogeneity, respectively, along with their corresponding strength of carbon fibers is shown in Fig. 2. For the sake of clarity, the data in Fig. 2 are replotted as plot of tensile strength of carbon fibers vs MCR and the size of polymeric entanglements in Fig. 3a and b, respectively. Figure 2 clearly demonstrates that, PAN-1 and PAN2 show relatively higher inhomogeneity in comparison to other spinning dopes (PAN-3 to PAN-8). PAN-1, and PAN-2 solutions were found contaminated by the graphite bushing of the stirrer assembly. Therefore, the lowest tensile strength of carbon fibers from PAN-1 and PAN-2 can be attributed to higher inhomogeneity in the corresponding solutions. For most of the spinning dopes under study, the nanoscale inhomogeneity varied in the range of 5 [+ or -] 4 nm. There appears to be no direct correlation (Fig. 3b) between the nanoscale inhomogeneity of the spinning dopes and the corresponding strength of carbon fibers. The nanoscale inhomogeneity in all the spinning dopes except that of PAN-2, may be considered to be almost similar between these solutions because (a) the light scattered by small nanoscale particles could be overshadowed by very large particles (detected in MCR), which could lead to errors in the accurate estimation of the nanoscale entanglements [17], (b) small errors in the measurement of viscosity would propagate errors in estimation of diffusion coefficient and consequently on the size of the nanoscale entanglements, and (c) the polymeric entanglements need not be spherical, although they are assumed to be spherical, to fit in the Stokes-Einstein relation. It is reported that the nanoscale defects will limit the tensile strength of high strength carbon fibers [6]. Nanoscale inhomogeneity of the size of around 0.5-2 nm was observed in the high-resolution transmission electron microscope images of carbon fibers with tensile strength of 5.5-5.8 GPa [6]. This range of tensile strength is higher than the highest tensile strength reported in this paper.

PAN-6 solution exhibited least heterogeneity but the tensile strength was not the highest. Probable reason lies in the fact that this solution was prepared in glass reactor while other solutions were prepared in Parr reactor. High shear mixing that can be achieved in Parr reactor cannot be achieved in glass reactor. Thus there exist competing contributions namely lower impurities in solutions prepared in glass reactor (PAN-6) and high shear mixing in the solutions prepared in Parr reactor (other solutions) that may be the probable reasons behind similar tensile strength in the carbon fibers prepared from precursor fibers that were spun from the solutions in glass and Parr reactor. The nanoscale inhomogeneity is similar between the samples PAN^l, 5, 7, and 8. However, the microscale inhomogeneity is different between theses samples. At 95% confidence interval, the tensile strength of samples PAN-4, PAN-5 and PAN-7 is statistically similar (p value > 0.05) while that of PAN-8 is different than that of PAN-4 and PAN-5 (p value < 0.05). Impurity levels in the solution (MCR) is one of the several factors that govern the tensile strength of carbon fibers, other factors being precursor fiber properties, strains in stabilization and carbonization, time and temperature of stabilization and carbonization. The data (Figs. 2 and 3) demonstrate that microscale inhomogeneity in the spinning dope (MCR) is a contributor to the decrease in the strength of carbon fibers in the range of 3-4.5 GPa. Considering the fact that carbon fiber strength is highly defect sensitive, and several other factors are present in the carbon fiber manufacture besides MCR of solutions that govern the defects in the carbon fibers, correlation coefficient of 0.59 does suggest that MCR of solutions may be one of the several factors affecting carbon fiber tensile strength. Although systematic surface analysis of carbon fibers from each of the experiments reported in this manuscript was not carried out, SEM characterization of carbon fibers carried out during the research program showed similar morphologies of carbon fibers obtained from various trials and non-existence of defects such as voids. Supporting Information Fig. SI shows the SEM image of cross section of carbon fibers of PAN-7 and Supporting Information Fig. S2 shows the SEM micrograph of surface of carbon fibers. The impurities seen on the carbon fiber surfaces are believed to be that of the burnt residues of silicone spin finish used in the precursor fiber manufacture.

In view of the correlation observed between tensile strength of carbon fibers and microscale inhomogeneity caused by contaminants in the spinning dope, contaminants from PAN powder were removed using surfactant based process. TX 100 was chosen as surfactant to form micelles. TX 100 is a nonionic surfactant that forms micelles at very low concentration (cmc-critical micelle concentration of 0.24 mM in water) [22], AOT was chosen as surfactant to form reverse micelles. The cmc of AOT in hexane is 0.05 mM [23]. Since PAN powder is insoluble in both water and hexane, lighter contaminants like dust particles are expected to be removed by frothing or by trapping them in micelles or reverse micelles. The plot of MCR vs time for 0.1 wt% solution before and after filtration through 0.2 [micro]m membrane and after treating PAN powder with surfactants is shown in Fig. 4. It is evident from Fig. 4 that the fluctuation in the MCR decreased substantially for PAN solution after treatment with surfactants as compared with that corresponding to as received powder. In fact, the fluctuations in the MCR in 0.2-[micro]m filtered PAN solution are comparable to that in the surfactant treated but unfiltered PAN solution. Thus, while filtration of solutions with spinnable concentrations is not possible through small size filters such as 1 [micro]m, surfactant-assisted purification is a possible approach to reduce the inhomogeneity in the PAN solutions. In order to manufacture high strength carbon fibers, it is necessary to make sure that the process of surfactant treatment does not leave its chemical residues in the purified polymer. The fourier transform infrared (FTIR), TGA, and DSC studies on PAN powder (Supporting Information Fig. S5), before and after treatment with surfactants overlay on top of each other, and suggest complete removal of surfactant from PAN powder.


The presence of microscale and nanoscale inhomogeneity in the PAN solution decreases the tensile strength of carbon fibers. We show the use of the MCR from DLS as a tool to assess microscale inhomogeneity in spinning dope. It is established that MCR can be used as a tool to assess and compare the quality of spinning solutions with similar viscosities. The nanoscale inhomogeneity of various spinning dopes was estimated using slow mode in the correlation function from DLS measurements. The presence of nanoscale and microscale inhomogeneity in various spinning solutions was correlated with the tensile strength of carbon fibers. It is found that the increase in microscale inhomogeneity in spinning dope contributed more to the reduction in the carbon fiber tensile strength than the contribution due to nanoscale inhomogeneity in the case of carbon fibers with tensile strength in the range of 3-4.5 GPa. The contaminants causing microscale inhomogeneity were removed by surfactant treatment. The PAN treated with surfactants was characterized using FTIR, DSC, and TGA. Noticeable decrease in the MCR from solutions prepared using purified PAN indicates the practical approach that can be adopted to reduce microscale inhomogeneity in the PAN solutions.


[1.] Donnet, J.-B., Carbon Fibers, 3rd ed., rev. and expanded. (1998), CRC Press, New York.

[2.] A.A. Ogale, M. Zhang, and J. Jin, J. Appl. Polym. Sci., 133 43794 (2016).

[3.] IM7 carbon fiber, media/raw/IM7_HexTow_DataSheet.pdf.

[4.] T1100G carbon fiber, torayca_t1100g.pdf.

[5.] 1M10 carbon fiber, media/raw/IM10_HexTow_DataSheet.pdf.

[6.] H.G. Chae, B.A. Newcomb, P.V. Gulgunje, Y. Liu, K.K. Gupta, M.G. Kamath, K.M. Lyons, S. Ghoshal, C. Pramanik, L. Giannuzzi, K. [section]ahin, I. Chasiotis, and S. Kumar, Carbon, 93 81 (2015).

[7.] T. Dumitrica, M. Hua, and B.I. Yakobson, Proc. Natl. Acad. Sci. U.S.A., 103(16) 6105 (2015).

[8.] A. Kelly and M.N. Strong Solids. Oxford University Press, New York (1986).

[9.] H.G. Chae and S. Kumar, Science, 319(5865) 908 (2008).

[10.] B.A. Newcomb, P.V. Gulgunje, Y. Liu, K. Gupta, M.G. Kamath, C. Pramanik, S. Ghoshal, H.G. Chae, and S. Kumar, Polym. Eng. Sci., 56(3) 361 (2016).

[11.] V. Raghavan, M. Gupta, Z. Xiao, H. Gi Chae, T. Liu, and S. Kumar, Polym. Eng. Sci., 55(6) 1403 (2015).

[12.] W.S. Lyoo, B.C. Kim, and W.S. Ha, J. Polym. Sci., 37(7) 1259 (1997).

[13.] E.J. Lee, N.H. Kim, K.S. Dan, and B.C. Kim, J. Polym. Sci. Part B: Polym. Phys, 42(8) 1451 (2004).

[14.] H.G. Chae, B.C. Kim, S.S. Im, and Y.K. Han, Polym. Eng. Sci., 41(7) 1133 (2001).

[15.] P.J. Yoon and C.D. Han, Macromolecules, 33(6) 2171 (2000).

[16.] J.F. Li, T. Ngai, and C. Wu, Polym. J., 42(8) 609 (2010).

[17.] Berne, B.J. and R. Pecora, Dynamic Light Scattering: With Applications to Chemistry, Biology, and Physics. New York: John Wiley and Sons, Inc. (1976).

[18.] S.W. Provencher, Comput. Phys. Commun., 27(3) 229 (1982).

[19.] S.W. Provencher, Macromol. Chem. Phys., 180(1) 201 (1979).

[20.] W. Brown. Dynamic Light Scattering: The Method and Some Application. Oxford: Clarendon Press, (1993).

[21.] H. Lu, L. Yu, Q. Liu, and J. du, Polym. Chem., 4(12) 3448 (2013).

[22.] A. Chattopadhyay and E. London, Anal. Biochem., 139(2) 408 (1984).

[23.] N.V. Bukar', S.Y. Krasnobaeva, E.V. Dyushenko, G.V. Avramenko, K.I. Kienskaya, Russ. J. Appl. Chem., 81(2), 276 (2008).

Vijay Raghavan, Prabhakar V. Gulgunje, Kishor K. Gupta, Manjeshwar G. Kamath, Yaodong Liu, Chandrani Pramanik, Bradley A. Newcomb, Han Gi Chae, Satish Kumar (iD)

School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA

Additional Supporting Information may be found in the online version of this article.

Correspondence to: S. Kumar; e-mail:

Present address: Han Gi Chae; School of Materials Science and Engineering, Ulsan National Institute of Science and Technology, 50 UNIST-gil, Ulsan, South Korea 44919.

Contract grant sponsor: Air Force Office of Scientific Research; contract grant number: FA9550-14-1-0194. contract grant sponsor: Army Research Office; contract grant number: W911NF-10-1-0098.

DOI 10.1002/pen.24947

Published online in Wiley Online Library (

Caption: FIG. 1. Photon count rate vs time of 0.1 wt% PAN solution in DMF (a) unfiltered and (b) filtered through membrane filter with pore size of 1 [micro]m. [Color figure can be viewed at]

Caption: FIG. 2. Histograms of MCR (green diagonal lines) and size of nanoscale inhomogeneity) (blue horizontal lines) for various spinning dopes overlaid on top of each other, plot of spinning dopes and corresponding tensile strength of carbon fibers (pink squares). Error bars are the standard error of the mean for the measured property of each sample. [Color figure can be viewed at]

Caption: FIG. 3. Tensile strength of carbon fibers vs (a) MCR and (b) nanoscale inhomogeneity. Error bars are the standard error of the mean for tensile strength. The equation of the trend line fitting for data points in Fig. 3a is y = [-0.2 [+ or -] 0.01 (*)] + [4.89 [+ or -] 0.4]; R = 0.59.

Caption: FIG. 4. Count rate vs time for 0.1 wt% PAN solution (a) before filtration, (b) after filtration using 0.2 [micro]m PTFE filter, and (c) unfiltered after purifying impurities using surfactants.
TABLE 1. PAN solution preparation conditions

Sample code   Solution concentration (g/dL)       Reactor

PAN-1 (a)                  10.0                   Parr VI
PAN-2                      10.5                   Pan-VI
PAN-3                      10.5                   Parr V2
PAN-4                      10.5                   Parr V2
PAN-5                      10.5                   Parr V2
PAN-6                      10.5                glass reactor
PAN-7                      10.5                   Parr V2
PAN-8                      10.5                   Parr V2

(a) Solution was collected using 10-pm filter at discharge
point of Parr reactor.
COPYRIGHT 2019 Society of Plastics Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Raghavan, Vijay; Gulgunje, Prabhakar V.; Gupta, Kishor K.; Kamath, Manjeshwar G.; Liu, Yaodong; Pram
Publication:Polymer Engineering and Science
Article Type:Report
Date:Mar 1, 2019
Previous Article:Effects of Temperature and Comonomer Content on Poly ([epsilon]-Caprolactam-Co-[epsilon]-Caprolactone) Copolymers Properties: An Evaluation of...
Next Article:Effect of Surface Roughness and Reciprocating Time on the Tribological Properties of the Polyimide Composites.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters