, Volume 61, Issue 7–8, pp 717–725 | Cite as

Creep Deformation of Carbon-Based Cathode Materials for Low-Temperature Aluminum Electrolysis

  • Wei Wang
  • Weijie Chen
  • Wanduo Gu

The uniaxial-compression creep behavior of semi-graphitic carbon products was investigated using modified Rapoport equipment in a K3AlF6–Na3AlF6–AlF3 and a Na3AlF6–AlF3 system. The stress exponent is low for the K3AlF6–Na3AlF6–AlF3 system in the steady-state creep stage. With an increase in graphitization degree and grain size, the interlayer space and porosity of the tested samples decrease after aluminum electrolysis. A low temperature can suppress carbon-cathode damage. Based on these stress exponents and a microstructural investigation using transmission electron microscopy, it is proposed that dislocation glide is the dominant creep mechanism for the carbon cathode during aluminum electrolysis in the steady-state creep stage.


creep carbon cathode stress exponent aluminum electrolysis potassium cryolite 

Aluminum is produced by the Hall–Heroult electrolysis process at ~950°C. Despite progress made by using this conventional bath, it has many shortcomings, such as a high cost, serious environmental pollution, unstable properties, and a high electrolysis temperature. Potassium cryolite, which is a low-temperature-electrolyte system, could solve these problems to some extent, and it has attracted significant research attention [1, 2]. Because potassium salts corrode electrodes nine times more than sodium salts, potassium salts are considered inhibitive in the aluminum-electrolysis industry [3]. To date, research has been lacking on the elpasolite system. As a critical cell component, the running of a cathode directly affects the energy consumption, the cell life, and the spent pot-lining emissions. An increase in the directionality frequency distribution of the hole angle after creep deformation is one of the reasons for the deterioration of cathode life. Creep deformation of the cathode during aluminum electrolysis is related closely to cathode failure and the service life of aluminum reduction cells [4, 5]. Therefore, it is necessary to study the creep resistance of carbon-based cathode materials in the K3AlF6–Na3AlF6–AlF3 and Na3AlF6–AlF3 systems.

Expansion and creep of cathode materials because of the penetration of alkaline elements and intercalation with carbon are considered to be a major chemical reason for cathode deterioration in aluminum reduction cells [6]. Of the three typical industrial cathode blocks (semi-graphitic, full graphitic, and graphitized), a higher extent of material graphitization yields a better resistance to the penetration of alkaline elements [7, 8]. Potassium is important in expansion and penetration compared with other alkaline elements [9]. Recently, c reep deformation has received increased attention because of its effect on the deterioration of cathode materials. During aluminum electrolysis, compression tests on cathode carbons indicate that the graphitized material displays a good resistance to creep deformation because of the reduced amount of Na and melt penetration [10, 11, 12]. Much experimental research has been conducted to extend the life of individual cells, whereas few studies on the effect of potassium on the creep resistance of cathode carbon blocks have been published.

The aim of this work was to investigate the creep properties of the semi-graphitic cathode samples in a K3AlF6–Na3AlF6–AlF3 and a Na3AlF6–AlF3 system. Operative creep mechanisms were suggested based on stress exponents and microstructural analysis results.


Materials. Cryolite (200 g cryolitic electrolyte, industrial grade with a cryolite ratio of 2.0) that contained 5 wt.% CaF2 and 8 wt.% Al2O3 (analytical reagent) was used for each run. When the cryolite ratio (molecular ratio of NaF/AlF3) was 2.5, the KR(m(K3AlF6)/(m(K3AlF6) + m(Na3AlF6)) was 0.2. All chemicals were dried at 400°C for 4 h before testing aluminum electrolysis.

All test samples were machined to a cylinder (25 mm diameter and 50 mm length) for use as a cathode sample in laboratory aluminum electrolysis. The material consisted of 65% electrocalcined anthracite and 35% graphite and was baked with a coal tar pitch binder to ~1200°C.

Experimental procedure. The experimental setup is shown in Fig. 1. An external load (1) was applied through a loading frame (3) and a loading extension rod (6) to the test sample (9). The measuring extension pins (4) were fixed into the cathode sample, whereas it could move independently of the load through a hole in the middle of the frame (3). The electrolysis cell was heated with argon to purge the furnace tube. The bath temperature was measured by a calibrated Pt–Pt/Rh thermocouple, with an accuracy of ±0.2 K at 1223 K (950°C). A constant current was provided by a MPS302 DC power supply and the current density at the cathode was 0.5 A/cm2. A silicon nitride sheet was placed between a graphite crucible (the anode) and the cylindrical sample (the cathode) in the experimental cell for electrical isolation. The cathode sample was immersed in the melt at a ~40 mm length and was subjected to a creep measurement that occurred 2 h after starting the electrolysis. The external load was provided by a constant-pressure system, which could maintain the pressure at a given value for a period of time. In each experiment, when the temperature was reached, the arrangement was left to stabilize for at least two hours to ensure that there was no apparatus creep. The sample displacement was measured by an LVD transducer (with a range of 10 mm and a resolution of 1 μm) on the top of the furnace. In general, only the axial strain is important; the axial strain (ε) of the cathode can be calculated from
Fig. 1.

A modified Rapoport system for creep measurement: 1) load; 2) LVD sensor; 3) loading frame; 4) measuring extension pin; 5) air gate; 6) loading extension rod; 7) graphite crucible; 8) electrolyte; 9) cathode sample; 10) silicon nitride sheet; 11) electrical heating element; 12) anode bar; 13) air gate.

$$ \upvarepsilon =\frac{h_t-{h}_0}{h_0}, $$

where h 0 is the original height of the cathode sample, and h t is the height of the sample for the corresponding electrolysis time.

Sample characterization. After the experiment, a 4-mm-thick specimen was cut from the bottom 10 mm of the cathode sample for x-ray diffraction (XRD) analysis. A measurement of the interlayer spacing and the crystallite size in the cathode carbons was carried out by XRD (D8ADVANCE) in a step scan mode and Cu K α radiation (1.5406 Å in wavelength). Microstructural analyses were carried out by scanning electron microscopy (SEM, JSM–5610LV) and an energy-dispersive-spectrometry (EDS) analyzer. The interlayer spacing (d 002) and the crystallite size in the cathode carbons can be calculated from the Bragg and Scherrer equations [13]:

$$ 2d\;\sin \uptheta =\uplambda; $$
$$ {L}_c=\frac{K\uplambda}{\upbeta\;\cos\;\uptheta}, $$

where λ is the x-ray wavelength, β is the peak width in radians, and θ is the incident beam angle.


Effect of electrolyte on creep behavior. Figure 2 shows the variation in creep strains with testing time for a semi-graphitic cathode sample at a cryolite ratio of 2.5 and a compressive loading of 2, 4, 5, and 6 MPa in a Na3AlF6–AlF3 system. Figure 3 shows similar curves with a compressive loading of 2, 4, 5, and 6 MPa and a KR ratio of 0.2.
Fig. 2.

Creep strains with time for semi-graphitic cathode sample in Na3AlF6–AlF3 system for different loads at 950°C.

Fig. 3.

Creep strains with time for semi-graphitic cathode sample in K3AlF6–Na3AlF6–AlF3 system for different loads at 925°C.

In Figs. 2 and 3, the values of the creep strains in the Na3AlF6–K3AlF6–AlF3 system were ~0.02% higher than those in the Na3AlF6–AlF3 system under the same load. This trend was similar to that for the alkali and melt penetration into the carbon layers by intercalation in the material micropores, which could be the major cause for the carbon-structure deterioration, including the creep [14]. The potassium salt corroded the electrode to a greater extent than the sodium salt because of its larger atomic radius [15, 16]. The corrosion of potassium salt on the carbon cathode was not as high as described in previous papers [3] because of two main reasons. First, the lower bath temperature makes it difficult for carbon-atom diffusion. Second, the difference in deposition potential between the alkaline elements and Al will increase for a lower bath temperature and result in less precipitation of K and Na (see Figs. 4 and 5).
Fig. 4.

EDS patterns of center of sample cross-section based on Na3AlF6–AlF3 system.

Fig. 5.

EDS patterns of center of sample cross-section based on K3AlF6–Na3AlF6–AlF3 system.

Analysis of steady-state creep. The creep curves can be divided into three stages of transient, stable, and accelerating creep stages, which is consistent with literature [17]. In the first stage or within the first 15 min, the curve records an increase in the concerned variable with time, with a decreasing rate, followed by a steady-state region in which the strain increases linearly with time. With an increase in applied stress at a constant temperature, the steady-state region occurs at shorter periods of time and creep occurs more easily. Because sample fracture does not occur, it is not possible to record the third stage of the curve compared with what happens in an ordinary creep test. In general, the minimum creep rate (\( \overset{\cdot }{\upvarepsilon} \)) of the semi-graphitic cathode samples is described with an applied stress σ by the following power-law equation [3]:

$$ \overset{\cdot }{\upvarepsilon}=A{\upsigma}^n\kern0.5em \exp \left(\frac{-Q}{RT}\right); $$
$$ \ln \kern0.5em \overset{\cdot }{\upvarepsilon}=\ln \kern0.5em A\kern0.5em +n\ln \upsigma -\frac{Q}{RT}, $$
where \( \overset{\cdot }{\upvarepsilon} \) is the steady-state creep rate, n and A are material-related constants, Q is the apparent activation energy for creep, T is the absolute temperature, and R is the gas constant. This implies that if σ is plotted against \( \overset{\cdot }{\upvarepsilon} \) on a double logarithmic scale, a straight line with a stress exponent n is obtained. From the creep curves, the steady-state creep rate can be determined, and plots of the logarithmic creep rate against the logarithmic stress and the reciprocal of temperature are shown in Fig. 6. The results agree well with Eq. (5). The stress exponent values that can be calculated from the slopes of the best-fit line to the variation of \( \ln \overset{\cdot }{\upvarepsilon} \) and lnσ at a constant temperature are 0.6797 and 0.6485 for the Na3AlF6–AlF3 and Na3AlF6–K3AlF6–AlF3 systems, respectively. A slight decrease in stress exponent was observed in the Na3AlF6–K3AlF6–AlF3 system, which decreased the high-temperature resistance performance of the carbon cathode. This outcome could be related to the extensive corrosion of potassium salt on the carbon cathode and the instability of the intercalation compounds that occurred during the high-temperature creep process [18].
Fig. 6.

Relation between creep rate and stress during steady-state creep.

A creep mechanism could be determined from microstructural investigation using TEM and by calculating the stress exponent, which is used to determine the creep properties of the material. Figure 7 shows that the dislocation glide is the dominant mechanism for the carbon cathode during aluminum electrolysis. Creep of graphite is caused by an increase in edge dislocations that are distributed near the tips of cracks. By forcing the crack faces apart, dislocations at the crack tip are characterized in that their Burgers vectors are parallel to the c-axis. The dislocations form a wedge of material, the wedge lengthens and thickens with time, and a time-dependent strain results [19].
Fig. 7.

Dark-field images of dislocations for the carbon cathode.

Microstructures analysis. Figure 8 shows the SEM images of the cathode samples before electrolysis compared with those after 1 h of aluminum electrolysis under an external load of 6 MPa (see Fig. 9).
Fig. 8.

SEM images of cathode sample before electrolysis.

Fig. 9.

SEM images of the cathode sample after creep deformation in Na3AlF6–K3AlF6 system.

These images show that the applied stress reduced the cathode porosity, and that the porous structure became long and narrow. Through cracks are visible in the specimen microstructure after creep under a compressive stress of 6 MPa, as shown in Fig. 9. The bonding action between the electrons in potassium or sodium orbits and π-electrons in carbon yields a corresponding reaction and form a graphite−alkali metal intercalation compound [CxM(K,Na)], which leads to an enlargement of interlayer spacing. The destructive power of potassium was larger, and the cathode undergoes more extensive corrosion. This, in turn, could have a detrimental effect on the creep behavior of the carbon cathode, because cracks may accelerate creep behavior. Cathode failure is visible in macroscopic view. Although the increasing stress reduced the pore size and the porosity of the cathode samples, the denser cathode samples and the greater creep resistance resulted in the small strain change [20]. This was proven to be correct from the data in Figs. 2 and 3.

Figure 10 shows the XRD analysis of the original cathode and cathodes after aluminum electrolysis in different electrolyte systems. The 2θ value of the original cathode is 26.30, which indicates a characteristic peak of the laminated-structure graphite. The 2θ value of the cathode sample at the characteristic peaks is largest in the K3AlF6–Na3AlF6–AlF3 system, and exceeds that in the Na3AlF6–AlF3 system. Thus, the cathode sample interlayer spacing (d 002) is the smallest according to Eq. (2). It is expected to decrease d 002 with an increase in degree of graphitization [21]. The extent of graphitization is the probability for the nearest-neighbor pairs of layers to have a graphitic relationship, namely, the volume fraction of the graphitic structure that has a three-dimensional stacking order, in all carbon hexagonal net layers.
Fig. 10.

XRD analysis of cathode samples before and after creep deformation in different electrolyte systems.

To compare the relationship between the microstructure and graphitization degree, the interlayer spacing (d 002) and the crystallite size of the tested samples were calculated, as presented in Table 1.
Table 1.

Structural Parameters and Porosity of Cathode Samples Before and After Creep Deformation in Different Electrolyte Systems


\( \overline{d} \) 002/nm

L c/nm

Apparent porosity (%)

Before electrolysis












Table 1 shows that the intercalation of alkali metals into the carbon cathode increases L c and this value increases with the atomic diameter of the alkali metals. As a layered material, graphite has a hexagonal structure in a planar condensed-ring system with a weak bonding between graphene sheets [22, 23]. As K and Na diffused and reacted with the cathode to form intercalation compounds during aluminum electrolysis, the mutual bonding force between the carbon particles was reduced. The penetration of alkali metals and the bath caused movement, rearrangement, cleavage and crosslinking in the carbon. The hexagonal layer facilitated the diffusion of defects under stress. This led to a decrease the d 002 value, and the corresponding diffraction peak moved forward to the site of the larger angle. It increased from 26.30 (\( \overline{d} \) 002 = 0.3386 nm) to 26.35 (\( \overline{d} \) 002 = 0.3380 nm) after aluminum electrolysis in the Na3AlF6–AlF3 system. With increase in graphitization degree and grain size, the interlayer spacing and compression strength of the sample decreased further after aluminum electrolysis in the K3AlF6–Na3AlF6–AlF3 system, which reduced the cathode life [24, 25].

Conclusions. In low-temperature aluminum electrolysis systems, K and Na atom penetration into the cathode blocks leads to a lower stress exponent in the steady-state creep stage. A dominant mechanism for the carbon cathode is climb-controlled dislocation creep in the steady-state creep stage.

Low-temperature aluminum electrolysis could decrease the material deterioration of cathodes because of a reduced penetration of electrolyte into the cathode. Long and narrow cracks occurred during creep deformation, which led to a deterioration in cathode carbon material and a possible subsequent weakening of the cathode.

Under an external load, penetration of alkali metals results in a higher creep strain amplitude, a decrease in interlayer spacing and an increase in grain size, which accelerates the deterioration of carbon cathodes under low-temperature aluminum electrolysis.



Financial support from the Collaborative Innovation Center of Nonferrous Metals Henan Province (15A450001) is gratefully acknowledged.


  1. 1.
    H. W. Yan, J. H. Yang, W. X. LI, et al., “Alumina solubility in KF–NaF–AlF3-based low-temperature electrolyte,” Metall. Mater. Trans. B, 42, No. 5, 1065–1070 (2011).Google Scholar
  2. 2.
    J. Frazer and J. Thonstad, “Alumina solubility and diffusion coefficient of the dissolved alumina species in low-temperature fluoride electrolytes,” Metall. Mater. Trans. B, 41, No. 3, 543–548 (2010).CrossRefGoogle Scholar
  3. 3.
    M. Yuan, S. Kan, Q. Meng, et al., “Effect of LiF and Al2O3 on the physical and chemical characters in elpasolite containing aluminium electrolyte system,” Rare Met. , 30, No. 1, 595–598 (2011).CrossRefGoogle Scholar
  4. 4.
    A. Zolochevsky, J. G. Hop, T. Foosnaes, et al., “Rapoport–Samoilenko test for cathode carbon materials-II. Swelling with external pressure and effect of creep,” Carbon, 43, No. 6, 1222–1230 (2005).CrossRefGoogle Scholar
  5. 5.
    T. Chen, J. L. Xue, and X. Li, “High-temperature creep deformation and change in porous structure of graphite cathode in aluminum electrolysis process,” Light Met., 123–130 (2014).Google Scholar
  6. 6.
    G. V. Arkhipov and Y. S. Gorlanov, “Development of technology for producing a wettable coating on a carbon cathode by electrodeposition,” Light Met., 1367–1372 (2012).Google Scholar
  7. 7.
    D. Picard, M. Fafard, G. Soucy, et al. “Room temperature long-term creep/ relaxation behaviour of carbon cathode material,” Mater. Sci. Eng., A, 496, No. 1, 366–375 (2008).CrossRefGoogle Scholar
  8. 8.
    A. Zolochevsky, J. G. Hop, G. Servant, et al. “Rapoport–Samoilenko test for cathode carbon materials: I. Experimental results and constitutive modelling,” Carbon, 41, No. 3, 497–505 (2003).CrossRefGoogle Scholar
  9. 9.
    Z. Fang, X. L. Wu, J. Yu, et al., “Penetrative and migratory behavior of alkali metal in different binder based TiB2-C composite cathodes,” Trans. Nonferr. Met. Soc, 24, No. 4, 1220–1230 (2014).CrossRefGoogle Scholar
  10. 10.
    T. Gruber, T. W. Zerda, and M. Gerspacher, “Raman studies of heat-treated carbon blacks,” Carbon, 32, No. 7, 1377–1382 (1994).Google Scholar
  11. 11.
    J. Li, Y. Y. Wu, Y. Q. Lai, et al., “Simulation of thermal and sodium expansion stress in aluminum reduction cells,” J. Cent. South Univ., 15, 198–203 (2008).CrossRefGoogle Scholar
  12. 12.
    W. Wang, J. L. Xue, J. Q. Feng, et al., “Creep behaviors of industrial graphitic and graphitized cathodes during modified Rapoport tests,” Light Met., 1053–1057 (2011).Google Scholar
  13. 13.
    Y. H. Zhang, N. X. Feng, J. P. Peng, et al., “Effect of KF additions in Na3AlF6–Al2O3 electrolytes on expansion of cathode blocks,” Light Met., 681–684 (2012).Google Scholar
  14. 14.
    J. L. Xue, Q. S. Liu, and B. S. Li, “Creep deformation in TiB2/C composite cathode materials for aluminum electrolysis,” Light Met., 1023–1027 (2008).Google Scholar
  15. 15.
    X. Li, J. L. Xue, J. Zhu, et al., “Analysis of porous structures of graphitic cathode materials and the correlation to penetrated sodium,” Light Met., 1319–1324 (2012).Google Scholar
  16. 16.
    J. L. Xue, L. C. Wu, W. Wang, et al., “Characterization of sodium expansion of industrial graphitic and graphitized cathodes,” Light Met., 849–853 (2010).Google Scholar
  17. 17.
    Q. Y. Lin, T. Q. Li, Z. J. Liu, et al., “High-resolution TEM observations of isolated rhombohedral crystallites in graphite blocks,” Carbon, 50, No. 6, 2369–2371 (2012).Google Scholar
  18. 18.
    F. Kabirian and R. Mahmudi, “Impression creep behavior of a cast AZ91 magnesium alloy,” Mater. Trans. A, 40, No. 1, 116–127 (2009).CrossRefGoogle Scholar
  19. 19.
    E. G. Zukas and W. V. Green, “High temperature creep of a poco graphite,” Carbon, 10, No. 3, 519–522(1972).Google Scholar
  20. 20.
    N. Luiggi and M. Gómez, “Rhombohedral graphite: Comparative study of the electronic properties,” J. Mol. Struct.: Theochem, 897, No. 1, 118–128 (2009).Google Scholar
  21. 21.
    P. Y. Brisson, H. Darmstadt, M. Fafard, et al., “x-Ray photoelectron spectroscopy study of sodium reactions in carbon cathode blocks of aluminium oxide reduction cells,” Carbon, 44, No. 8, 1438–1447 (2006).CrossRefGoogle Scholar
  22. 22.
    D. G. Liu, Z. H. Yang, W. X. Li, et al., “Electrochemical intercalation of potassium into graphite in KF melt,” Electrochim. Acta, 55, No. 3, 1013–1018 (2010).CrossRefGoogle Scholar
  23. 23.
    X. G. Lü, Q. Y. Li, Y. Q. Lai, et al., “Digital characterization and mathematic model of sodium penetration into cathode material for aluminum electrolysis,” J. Cent. South. Univ., 16, 96–100 (2009).CrossRefGoogle Scholar
  24. 24.
    M. B. Rapoport and V. N. Samoilenko, “Deformation of cathode blocks in aluminium baths during process of electrolysis,” Tsvet. Met., 30, No. 2, 44–51(1957).Google Scholar
  25. 25.
    A. Zolochevsky, J. G. Hop, T. Foosnaes, et al., “Surface exchange of sodium, anisotropy of diffusion and diffusional creep in carbon cathode materials,” Light Met., 745–750 (2005).Google Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2017

Authors and Affiliations

  • Wei Wang
    • 1
    • 2
  • Weijie Chen
    • 1
  • Wanduo Gu
    • 2
  1. 1.College of Materials Science and EngineeringHenan University of Science and TechnologyLuoyangChina
  2. 2.Collaborative Innovation Center of Nonferrous Metals Henan ProvinceLuoyangChina

Personalised recommendations