MW carbonisation was applied for PAN PF, currently the most common precursor material for CF production. In order to find the optimum carbonisation methodology PAN fibres were carbonise via MW heating utilising the full power mode and controlling the MW power in several steps.
PAN precursor fibres were coated with 1-5c (cycles) of susceptor, and the layer was applied before or after stabilisation. SEM images, mechanical properties, and carbonization profiles of CFs produced via full power MW carbonisation from PAN precursor are shown in Fig. 2. Cross-sections of CFs show characteristics of brittle fracture (Fig. 2A). In this case Oval pores/defects of up to about 1.5 µm in the longest (radial) direction are observed in all samples. Pores are found along the circumference of the fibre’s cross-section and sometimes elongated radially. Similar features were reported previously by Barnet and Norr in a structural model of high modulus PAN-based CF [17, 18]. It was proposed that a radial structure of disordered crystalline webs and large voids is surrounded by highly ordered crystalline skin/sheath. Formation of such a structure was attributed to radial stresses and differential shrinkage during carbonisation and cooling of the fibre. These defects/voids visible in fibres are most likely caused by the high heating rates observed during MW carbonisation. Development of a skin-core structure could lead to stress between the two zones during shrinkage. It was previously reported that incomplete stabilisation could result in different orientations of inner and outer zones of the cross-section [19]. Also, fast carbonisation leads to evolution of large amounts of decomposition products over short periods of time instead of slow diffusion of volatiles to the outer surface. This promotes the creation of relatively large voids. In general, voids in cross-sections are slightly smaller in CFs coated before stabilisation, due to lower Tmax temperatures. Pores observed in CF produced via conventional carbonisation are much smaller and scarce, which suggests that the rates of heating or cooling have a predominant effect on their formation.
Mechanical properties of CFs produced via high power MW carbonisation from PAN precursor are shown in Fig. 2 (B, C). Average Tensile strength (TS), Young’s moduli (YM) and strain of samples coated before stabilisation are in the range of 0.84–0.93 GPa, 75.5-160.7 GPa and 0.6–1.25%, respectively. Average TS, YM and strain of samples coated after stabilisation are in the range of 0.41–0.83 GPa, 76.5-131.75 GPa and 0.64–0.84%, respectively. The properties were benchmarked off control sample, carbonised in conventional tube furnace (Conv). Measured TS of 1.42 GPa, YM of 114.8 GPa and strain of 1.24% are close to values previously reported for PAN carbonised with no tension control [20]. TS is generally slightly lower compared to the control, which can be attributed to premature fracture due to pores observed through SEM. Defects formed by the thermal decomposition products are known to have detrimental effect on mechanical properties during carbonisation at high heating rates [18]. However, overlapping of the error bars indicates that it was possible to produce CF with properties comparable to control CF (Conv).. These promising results are supported by YM values which match those of the control. Microwaved fibres generally ruptured at strain values lower than that of the control, but an overlap of the error bars is also visible in this case. Coating before stabilisation produced CFs with higher TS, YM and strain. This can be explained by milder processing conditions (lower Tmax) during carbonisation. Also, highest values of TS and YM in both groups correspond to median values of Tmax, which indicates improvement of mechanical properties when overheating, and therefore porosity, is limited.
Heating profiles for full power MW carbonisation of PAN precursor are shown in Fig. 2 (D, E). For clarity purposes, only the first 10 min of heating is presented. The samples experience extremely fast heating to temperatures above 800°C in the first 30 s. Recorded maximum temperatures (Tmax) are in the range of 847–956°C for fibres coated before stabilisation and 911–1076°C for fibres coated after stabilisation. In general, the Tmax increases with the increasing number of coating cycles and is higher for fibres coated after stabilisation. Coatings applied before stabilisation undergoe degradation/melting during the stabilisation step, which results in a decrease of the MW heating response. On the contrary, coatings applied after stabilisation respond efficiently to electromagnetic waves, most likely through increased electrical conductivity. It is also more fragile, more readily peeled off by the volatile matter i.e. crosslinking condensation, dehydrogenation and denitrogenation reactions [18, 22], during carbonisation which presumably is the cause of the abrupt changes of recorded temperatures during MW heating, see Figure S2. [21], Alternatively, some of the most abrupt peaks could be attributed to plasma formation [23].
Carbonisation conditions and properties of CFs produced via full power MW carbonisation from PAN precursor are summarised in Table S2 and compared to conventional carbonisation at 1000°C. Conversion rates are calculated from TGA curves and are in the range of 86–91% for CFs coated before stabilisation and 95–98% for CF coated after stabilisation (Figure S3). This is in agreement with higher Tmax temperatures recorded for samples coated after stabilisation, which results in fuller conversion to CF. The calculated graphitic crystallite size (La) is in the range of 6.42–9.23 nm for CFs coated before stabilisation, 5.12–11.43 nm for CFs coated after stabilisation and 9.61 nm for sample carbonised conventionally (Conv). Crystallite size is generally larger at higher Tmax. CFs that reached temperatures close to or exceeding 1000°C display crystallite sizes comparable with conventional carbonisation at 1000°C. This is attributed to increased crystallite growth as the temperature or the duration of heat treatment is increased [24]. Generally, heat treatment at higher temperatures increases La and this improves mechanical properties. Results show that the carbon phase of MW heated CF is within ranhe to produce mechanically viable CF however there is disagreement between La and mechanical data (especially TS) for the MW heated CF and this strongly suggests that the elimination of defects is the main factor to optimise the performance of CFs produced by this new MW method.
For the case of the MW heating under progressive power leves. an equivalent batch of samples were prepared with the objective to prevent defects and improve mechanical properties of the CFs produced via MW carbonisation. Heating profiles of an equivalent set of PAN samples carbonised at progressive power levels are shown in Fig. 3A, Fig. 3B, and Figure S4. Noticeably lower heating rates are observed during the initial stage at lowest power, with temperatures not exceeding 600°C in the first minutes of carbonisation. Increasing the MW power increases the rate of heating significantly and heats the samples to temperatures above 600°C. The final stage at full power (230V, 700W) results in highest heating rates and Tmax temperatures in the range of 690–827°C for fibres coated before stabilisation and 695–974°C for fibres coated after stabilisation.
SEM images of CFs produced via progressive MW carbonisation from PAN precursor are shown in Fig. 3C. Cross-sections show characteristics of brittle fracture. There were a very limited number of oval pores/defects in the fibres utilising this carbonisation protocol. In addition, the morphology of the pores found was shorter, narrower and sometimes needle-shaped. This can be explained by lower heating rates during initial stages of heating and also lower Tmax compared to counterparts carbonised at full power only, which limits radial residual stress in the fibre and development of cavities. Furthermore, milder carbonisation conditions lead to slower evolution of volatiles during MW heating and restrict damage of the produced CF. Generally, it is possible to find similar amounts of defect-free cross-sections within both sample groups.
Mechanical properties of CFs produced via progressive MW carbonisation from PAN precursor are shown in Fig. 3D. Average TS, YM and strain of samples coated before stabilisation are in the range of 0.42–0.88 GPa, 60.73-82 GPa and 0.56–1.26%, respectively. Average TS, YM and strain of samples coated after stabilisation are in the range of 0.23–1.18 GPa, 43–100 GPa and 0.34–1.31%, respectively. The results indicates that for this heating protocol the optimum samples were achieved for the fibres coated with 5 cycles and carbonised after stabilisation which reached values very closed to the control sample carbonised utilising conventional heating.
Carbonisation conditions and properties of CFs produced via progressive MW carbonisation from PAN precursor are summarised in Table S3 and compared to conventional carbonisation at 1000°C. Conversion rate evaluated from TGA is in the range of 80.9–86.6% for fibres coated before stabilisation and 81-91.1% for fibres coated after stabilisation (Figure S5). Conversion rate is lower compared to fibres carbonised at full power only. Incomplete carbonisation is in agreement with lower Tmax temperatures observed during progressive heating. The calculated graphitic crystallite size (La) is in the range of 4.65–6.75 nm for CFs coated before stabilisation and 4.25–7.78 nm for CFs coated after stabilisation. The values are lower than La of their counterparts, which is in agreement with lower Tmax and results in lower degree of orientation in CFs.
A blend of lignin and polyurethane (50:50 - lignin/TPU) was recently developed by the and identified as a promising precursor for CF production [25]. Melt-spun lignin/TPU precursor fibres were coated with MW susceptor to enable achievement of carbonisation temperatures during MW heating and conversion of sustainable lignin/TPU precursor into CF.
Firstly, full power MW carbonisation was used to convert lignin-based fibres coated after stabilisation. Heating profiles of full power MW carbonisation of lignin-based precursor are shown in Fig. 4A. Most of the samples experienced extremely fast heating to temperatures close to their Tmax, in the first 15 s of MW heating. This was followed by a sudden drop to 300–500°C and subsequent rise to a stable temperature between 400–950°C. Minor signal disruptions are observed (especially at higher temperatures), which are attributed to carbonisation reactions and evolution of volatiles. Recorded Tmax are in the range of 500–1044°C. Rates and maximum temperatures are observed to be more dependent on the number of the susceptor coating cycles compared to MW heating of PAN. This can be attributed to larger diameter of lignin-based fibres (~ 100 vs ~ 10 µm). This leads to smaller surface area covered by MW susceptor coating and smaller number of individual fibres mounted for MW carbonisation. Consequently, there is less MW susceptor present during heating of lignin/TPU compared to PAN carbonisation, which results in enhancement of differences between cycles of susceptor coating and should enable easier optimisation of the MW heating.
SEM images of CFs produced via full power MW carbonisation from lignin-based precursor are shown in Fig. 4B. Cross-sections show brittle fracture of the fibres with highly porous structures observed for samples with 2 or more layers of susceptor coating. Porous cross-sections present a structure of interconnected pores (mostly oval pores with diameter of up to about 20 µm in the longer dimension and some long cavities and cracks with lengths of up to about 50 µm) surrounded by a sheath/skin with thickness of up to 3 µm. The skin-core structure could be predefined during stabilisation [26, 27] and this is enhanced for fibres with larger diameters due to incomplete stabilisation [28]. Smooth and homogeneous cross-sections of samples 1c and Conv (conventional carbonisation at 1000°C) suggest the following: (i) the porosity is induced at temperatures above Tmax (1c) = 500°C; (ii) it is most likely caused by extremely fast heating (up to 200°Cs− 1 vs 10°Cmin− 1 for Conv). Most likely the formation of these pores in lignin-based CFs can be attributed to the release of volatiles during carbonisation [29, 30], gasification of oxygen [31]. High heating rate leads to fast release of volatiles and highly porous morphology. In fact, the evolution of volatile solvents from the fibre can be used to prepare porous fibres via electrospinning [32]. Open pores and waviness of the outer surface are also observed for samples with two or more layers of susceptor coating. This is due to the porosity induced during carbonisation, which manifests itself in the form of open pores and distorted shapes of produced CFs. This is confirmed by previous reports, where prolonged heating at higher temperatures leads to destruction of walls between pores and their appearance on the surface [33]. Samples 1c and Conv remain straight after the heat treatment. CFs presented here are too brittle for testing of mechanical properties due to porosity observed through SEM.
Carbonisation conditions and properties of CFs produced via full power MW carbonisation from lignin based precursor are summarised in Table S4, and compared to conventional carbonisation at 1000°C. Conversion rate evaluated from TGA is in the range of 84.7–87% for samples 2-5c (Fig. 4C). The low yield of sample 1c (55%) can be attributed to the lowest Tmax reached by the fibre during MW heating (500°C), which was not sufficient for carbonisation of the fibre. The lack of porosity on SEM images of this sample confirms that it was not exposed to conditions, experienced by the rest of the group. This also suggests that the pores observed in other samples were mostly formed at temperatures above 500°C. The calculated graphitic crystallite size (La) for samples 2-5c is in the range of 5.32–6.24 nm compared to 6.71 nm for sample Conv (conventional at 1000°C). This is in consensus with Tmax temperatures and analogous to findings reported for PAN, where most of the samples have smaller La compared to Conv, due to Tmax<1000°C. Although it is possible to calculate La value (6.58 nm) for sample 1c, it is only theoretical. The sample is not fully carbonised, as suggested by TGA and low intensity of Raman spectrum. Furthermore, the method used here for La calculation applies only for crystallite size > 2 nm [14, 24].
Progressive MW carbonisation was used to convert a set of lignin-based fibres coated with the MW susceptor before and after stabilisation, in order to find optimum conditions and improve mechanical performance of CFs produced from this precursor. Heating profiles of lignin based precursor samples carbonised at progressive power levels are shown in Fig. 5A, Fig. 5B, and Figure S6. Noticeably lower heating rates are observed during the initial stage at lowest power. After 5 minutes of heating, most of the samples reach temperature between 300–450°C, which is usually higher for samples coated after stabilisation. This correlates with improved MW absorption observed previously for the susceptor layer prepared via this sequence. Increasing the MW power increases the rate of heating significantly. It is observed that at a certain point during initial and intermediate stages of heating, sample temperature suddenly increases from about 400°C to about 600–800°C at which it tends to stabilise again. This sudden rise is recorded earlier in the process for samples with thicker susceptor coating. Reduction of heating rate before this transition is probably the key to process optimisation, as suggested by similar practice for conventional carbonisation of other hardwood lignin fibres [27]. After this, relatively stable reading of temperature (sometimes preceded by a sharp peak instead of round shoulder) is recorded and the temperature scales with MW power until the final stage at 230V, 700W. Recorded Tmax are in the range of 1020–1259°C for fibres coated before stabilisation and 1000–1346°C for fibres coated after stabilisation. Highest temperatures are recorded for samples after 4 cycles of susceptor coating (4c). Wider range and higher Tmax is observed for samples coated after stabilisation, which could be explained by improved MW absorption of these samples.
SEM images and mechanical properties of CFs produced via progressive MW carbonisation from lignin-based precursor are shown in Fig. 5C. Cross-sections of CFs present evidence of brittle fracture. Most samples are defect free with a smooth and homogeneous morphology, which is almost identical to the control (Conv) and very similar to the one reported previously for the same precursor [25] or other lignin-based precursors [34]. This is a clear improvement over high-power MW carbonisation of lignin based fibre and over progressive MW carbonisation of PAN fibre. The main reason for this remarkable result is undoubtedly reduction of heating rates during initial carbonisation stage, especially in the range of 400–800°C which allows more time for carbonisation reactions, slower evolution of volatile products and this improves morphology of the carbon phase. This is achieved more easily in MW carbonisation of lignin-based precursor, most likely due to larger diameter (almost an order of magnitude compared to PAN) and the resulting lower MW susceptor to fibre volume ratio which allows fibres retain their overall shape.
Mechanical properties of CFs produced via progressive MW carbonisation from lignin based precursor are shown in Fig. 5D. Properties are compared to conventional carbonisation (Conv) at 1000, 1400 1600 and 2000°C. Average TS, YM and strain of samples coated before stabilisation are in the range of 0.13–0.31 GPa, 51–81 GPa and 0.23–0.5%, respectively. Average TS, YM and strain of samples coated after stabilisation are in the range of 0.08–0.33 GPa, 34.67–64.67 GPa and 0.21–0.57%, respectively. TS of fibres carbonised in MW exceeded performance of 0.25 ± 0.05 GPa for conventional carbonisation (Conv) at 1000°C. Even a single layer of susceptor (1c) is sufficient to produce CFs in MW which are comparable with this result. Furthermore, most of the values overlap with result of 0.36 ± 0.06 GPa for conventional carbonisation (Conv) at 1400°C, with closest values reported for 5 cycles of susceptor coating. This improvement can be attributed to reduction of heating rates and temperatures during initial heating stage and is in good agreement with improved morphology revealed by SEM. Similarly, mechanical properties of CFs produced in microwave assisted plasma were improved by addition of low temperature pre-carbonisation step [35]. Values of YM match those of samples carbonised conventionally (Conv) at temperatures of 1400°C or higher, which implies that an improvement of the YM could be achieved by optimisation of MW carbonisation through power, time or temperature variation. Strain at failure averaged at values of conventional carbonisation at 1400–1600°C, and in some cases was found superior than sample Conv. carbonised at 2000°C.
This is an excellent result illustrating that, if proper conditions are used, susceptor assisted MW carbonisation proposed in this work can be successfully applied for rapid carbonisation of CFs with mechanical performance comparable with conventional carbonisation. Reduction of high heating rates observed during intermediate stages of MW carbonisation will eliminate pores from all cross-sections. This will improve reliability of the process and increase the average mechanical properties. Further improvement of mechanical performance of the presented lignin based CFs could be achieved by reduction of fibre diameter [18, 22, 25], as illustrated by reported TS of 1.1 ± 0.1 GPa for the same precursor with a final diameter of 25 ± 3 µm. On the other hand, this will require closer control of susceptor coating thickness, MW power and other process conditions in order to tune MW heating.
Carbonisation conditions and properties of CFs produced via progressive MW carbonisation from lignin/TPU precursor are summarised in Table S5, and compared to conventional carbonisation at 1000, 1400 1600 and 2000°C. Conversion rate evaluated from TGA is in the range of 91.9–94.4% for fibres coated before stabilisation and 78.8–98.6% for fibres coated after stabilisation (Figure S7). Significant improvement in conversion over full power carbonisation can be explained by smaller diameter of PF and is in agreement with Tmax temperatures > 1000°C recorded for all samples. Highest conversion rates are observed for coating after stabilisation, which is in good agreement with previous findings. The calculated graphitic crystallite size (La) is in the range of 5.12–7.12 nm for CFs coated before stabilisation, 5.02–6.73 nm for CFs coated after stabilisation and 6.71 nm for conventional carbonisation at 1000°C. This is an improvement over high-power carbonisation and is in agreement with general trends observed for Tmax. Calculated values are higher than La size reported for the same precursor, but this is due to different calculation methods being employed [25].
Experimental and modelled MW heating profiles are compared in Fig. 6. Experimental MW heating of PAN samples 1–5c is compared to modelled MW heating using corresponding susceptor thickness. It is observed that modelled rates are two orders of magnitude higher from experimental heating rates and that modelled maximum temperatures (Tmax) (1020–1320°C) are similar than those recorded experimentally (910–1060°C). In addition, the model represent the trend observed experimentally according to the coating thickness. More coating cycles increase the susceptor coating that is translated in higher temperature profiles, which exactly what is predicted in the model. The differences in the heating rates are mainly attributed to the assumption of the susceptor coating is entirely composed of MWCNTs (which is not true because the MWCNTs have some voids between them and are coved by the stabilisers) and that no heat transfer occurs within the coating itself. The peak intensity of the electric field is expected to act on each fibre in the bundle. None of the reactions taking place during the conversion of PF to CF are taken into account in the model, and this leads to predicted heating rates which are higher and the absence of peaks/disruptions, as seen in the experimental data. The modelled heat loss is a function of an assumed value for emissivity (ε).
As the carbon phase develops within the fibre, it changes from MW transparent to MW absorbing. This increases its contribution to heat generation.
A Tmax closest to experimental data is obtained for a modelled susceptor thickness of about 100 nm. This is therefore used for modelling temperature gradients within the PF cross-section. The modelled results for a 10 µm diameter PF cross-section heated via a 100 nm layer of susceptor in an electromagnetic field of a microwave are shown in Fig. 7. Four points in time are chosen to visualise temperature distributions at 100, 400, 700 and 1000°C. A maximum edge-to-core temperature difference of 1.5°C is observed at the start of heating and this decreases over time of 43 ms to a difference of 0.02°C at 1000°C. At this point fibre reaches a stable Tmax, as shown in Fig. 6B.
Increasing of the precursor diameter to 70 µm decreases the rate of MW heating (Figure S1). In this case, the edge-to-core difference at the start of the heating reaches nearly 9.9°C and decreases over time of 295 ms to a difference of 0.17°C at 1000°C. This confirms that increases in diameter result in larger temperature gradients during heating. Interestingly, a predicted Tmax close to 1000°C is observed regardless of the fibre diameter. The model does not take into account absorption of MWs by the precursor material itself and therefore the modelled diameter has little effect once the temperature reaches the maximum where equilibrium between the heat generated in the susceptor and the heat lost through radiation/convection is achieved.
Life cycle analysis of MW heating was performed to quantify the environmental and cost benefits achieved via MW carbonisation (the LCA methodology is described in the SI file). The energy demand and environmental impact for production of 1 kg of CF from PAN and lignin-based precursor via conventional and MW heating are compared in Fig. 8.
Using MW technology for CF production shows great potential to decrease the cumulative energy demand for both PAN and lignin-based fibres. It is predicted that application of MW heating reduces energy demand of PAN based and lignin-blend based CFs production by 70.2% and 66.8% respectively. Results also shows that the climate impact of the CFs production using a conventional furnace is significantly higher for PAN-based fibres compared to lignin- based fibres, despite lower carbon yield of the latter material. This is attributed to lower climate impact of the PF production. Furthermore, lignin’s inherent properties in theory reduce the energy consumption in the CF production step due to high level of aromaticity [36]. MW technology also reduces CO2 emission of PAN based CFs production to 27.6% of its conventional process value. Application of lignin blend reduces the climate impact of PF production by up to 69.5%.