Ultra-Long Carbon Nanotubes-Based Flexible Transparent Heaters

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

---- The manuscript claims that ultra-long CNTs (UL-CNTs) are relatively underexplored compared to SWCNT systems. Could the authors provide a more direct quantitative comparison with other references 
----- Since cost-effectiveness is a major driver for using MWCNTs/UL-CNTs, could the authors briefly comment on the relative cost of their final $R_s/T$ film (considering the ink, coating process, and post-treatment) compared to commercially available $\mathbf{ITO}$-coated flexible PET? 
-------Surfactant Residue: The choice of Solsperse 46000 is justified by its superior dispersion stability. However, surfactants can leave insulating residues, which negatively affect conductivity. Did the authors quantify the amount of residual surfactant left on the film surface after the drying and post-treatment steps?
----- Need to explain more about Figure 9.
---- the references are good and new 

Author Response

Point-by-point response

Reviewer 1 Comments:

1. The manuscript claims that ultra-long CNTs (UL-CNTs) are relatively underexplored compared to SWCNT systems. Could the authors provide a more direct quantitative comparison with other references

We thank the reviewer for this comment. It should be noted that SWCNT has been explored intensively in many fabrication techniques such as CVD¹, growth², stamping³, etc. Since our work is focused on ink formulation and wet deposition, here we chose to present a quantitative comparison of our ULCNT-based transparent heaters vs. to SWCNT and MWCNT transparent conducting films (TCFs). Hecht et al. ³ reported on SWCNT transparent conductive films with Rₛ ≈ 80 Ω·sq⁻¹ at 80% T₅₅₀ after nitric-acid treatment. Jia et al.⁴reported on SWCNT heaters with Rₛ ≈ 95 Ω·sq⁻¹ at 72% T₅₅₀ that reach ~100 °C within 60 s under 35 V. Kim et al. ⁵ reported MWCNT/PI bilayer heaters with sheet resistances decreasing from ~2.0 × 10⁵ to ~3.9 × 10³ Ω·sq⁻¹ as the MWCNT thickness increased, while the transmittance at 550 nm decreased from ~78% to ~52%.

In our work, a single flow-coated ULCNT layer on PET after flash photonic heating (FPH) achieves Rₛ ≈ 1.2 kΩ·sq⁻¹ at ~85% T₅₅₀, while multilayer stacks reduce Rₛ further at the expense of transparency. Thus, in terms of Rₛ/T, our films do not compete with state-of-the-art SWCNT, but instead we propose better results compared to MWCNT-based transparent heaters. Our main contributions are (i) demonstrating that ultra-long MWCNTs can be formulated in a purely aqueous ink, (ii) scalable flow coating on multiple flexible substrates, and (iii) improved performance after post-treatments (HNO₃ fumes and FPH).

 

The following text is added to the introduction Page 2:  

Hecht et al. ³ reported on SWCNT transparent conductive films with Rₛ ≈ 80 Ω·sq⁻¹ at 80% T₅₅₀ after nitric-acid treatment. Jia et al.⁴ reported on SWCNT heaters with Rₛ ≈ 95 Ω·sq⁻¹ at 72% T₅₅₀ that reach ~100 °C within 60 s under 35 V. Kim et al. ⁵ reported MWCNT/PI bilayer heaters with sheet resistances decreasing from ~2.0 × 10⁵ to ~3.9 × 10³ Ω·sq⁻¹ as the MWCNT thickness increased, while the transmittance at 550 nm decreased from ~78% to ~52%. While SWCNT-based transparent electrodes have been studied extensively across CVD¹, growth², stamping³ transfer, and wet-coating routes, their limited scalability continue to impede widespread adoption in flexible heating applications.

In addition, the following part will be added to the results and discussion Page 13: 

In this work, a single flow-coated ULCNT layer on PET after flash photonic heating (FPH) achieves Rₛ ≈ 1.2 kΩ·sq⁻¹ at ~85% T₅₅₀, while multilayer stacks reduce Rₛ further at the expense of transparency. Thus, in terms of Rₛ/T, our films still not as good as the costly SWCNT (R_s ≈ 80−150 Ω·sq⁻¹at ), but instead we propose better results compared to MWCNT (~2.0 × 10⁵ to ~3.9 × 10³ Ω·sq⁻¹ at  ~78% to ~52) transparent heaters. Our main contributions are (i) demonstrating that ultra-long MWCNTs can be formulated in a purely aqueous ink, (ii) scalable flow coating on multiple substrates including flexible, and (iii) improved performance after post-treatments (HNO₃ fumes and FPH).

 

2. Since cost-effectiveness is a major driver for using MWCNTs/UL-CNTs, could the authors briefly comment on the relative cost of their final $R_s/T$ film (considering the ink, coating process, and post-treatment) compared to commercially available $\mathbf{ITO}$-coated flexible PET?

 

We thank the reviewer for this comment. Commercially available flexible ITO-PET with  and typically costs 253 USD/m² (www.msesupplies.com), 55 USD/m² (www.aliexpress.com). In this work, the  deposition of the CNT was done by flow-coating at ambient conditions, which is clearly much cheaper than the sputtering processes (in vacuum) that is used to fabricate the ITO-PET. As for the material price, the cost of the ULCNT (TorTech CNTM1.5) used in this work is 3000 USD/kg, the amount of CNT in heaters is therefore calculated and compared to commercial ITO-coated PET. For the 0.2 wt% ULCNT ink the areal CNT is ~0.4  g/m² (as derived from the coated area and ink concentration). This corresponds to a CNT cost of:

 

As can be seen, not only that the process is cheaper, but also the material cost is very low. As for SWCNT, no commercially available TCFs were found, but the price of raw SWCNT is 10 time higher than ULCNT^6,7. This underlines that our approach is more cost-effective than ITO-PET and SWCNT TCFs. As for the FPH, the main cost goes for the capex, which is still lower than the sophisticated sputtering systems, and also the opex is much lower. 

 

The following text is added to the results and discussion Page 13:

 

Finally, we compared the material cost of the ULCNT heaters to that of commercial ITO-PET films. The cost of the ULCNT (TorTech CNTM1.5) used in this work is 3000 USD/kg, the amount of CNT in heaters is therefore calculated and compared to commercial ITO-coated PET. For the 0.2 wt% ULCNT ink the areal CNT is ~0.4  g/m² for a single coating pass (k-bar #3) resulting in a CNT material cost of approximately 1.2 USD/m². In contrast, flexible ITO-PET with T₅₅₀ > 85% typically costs 55–253 USD/m² depending on supplier and sheet resistance grade. Thus, even without considering processing advantages, the material cost of ULCNT-based heaters is at least two orders of magnitude lower than that of commercial ITO substrates, reinforcing the economic motivation for CNT-based transparent heaters.

 

 

3. Surfactant Residue: The choice of Solsperse 46000 is justified by its superior dispersion stability. However, surfactants can leave insulating residues, which negatively affect conductivity. Did the authors quantify the amount of residual surfactant left on the film surface after the drying and post-treatment steps?

 

We thank the reviewer for the comment. It should be noted that Solsperse is a good dispersing agent but it is also an insulator. Before the post processing step, we did not perform any washing steps and therefore the Solsperse amount in the layer should have not been changed (weight ratio of CNT:surfactant  3:1). In case of nitric acid, we made washing after the acid treatment, but amount of the surfactant was not analyzed, and indeed has to be done on the future work. In case of FPH it might be that some of the Solsperse is partially decomposed and therefore we used XPS to identify the C 1s and O 1s regions and to qualitatively track chemical changes induced by the flash photonic heating (FPH) process. As shown in Figure 9 and Table 1, the untreated CNT film exhibits a large fraction of oxygen-containing carbon species (C–O and C=O), indicative of Solsperse residues and mild CNT surface oxidation during dispersion. After FPH, the relative contributions of the C–O and C=O components decrease, while the hydrocarbon/sp²-like C–H/C=C component increases, and the O 1s signal intensity is reduced. These spectral changes probably related to a partial decomposition and/or removal of surfactant-related organic species. Importantly, these XPS trends correlate with a substantial decrease in sheet resistance from the as-dried value of ~7.5 kΩ·sq⁻¹ to ~1.2 kΩ·sq⁻¹ for a single layer at ~85% T₅₅₀ after FPH. We explicitly acknowledge in the revised manuscript that a more rigorous, quantitative determination of residual surfactant is beyond the scope of the present study and is an interesting direction for future work.

 

The following text is added to the results and discussion Page 13:

It should be noted that Solsperse is a good dispersing agent but it is also an insulator. Before the post processing step, we did not perform any washing steps and therefore the Solsperse amount in the layer should have not been changed with a weight ratio of 3:1 CNT:Solsperse. Following HNO₃ vapor exposure, the films were rinsed with IPA, which likely removed a portion of unbound or weakly adsorbed Solsperse, whereas FPH-treated films were not rinsed, and thus any removal of the surfactant arises from photothermal decomposition. Consistent with this, XPS measurements reveal a clear decrease in oxygenated carbon species (C–O and C=O) and a relative increase in sp²-like C–C/C–H signals after FPH, indicating partial decomposition of the surfactant. This chemical evolution correlates with the substantial improvement in conductivity—from ~7.5 kΩ·sq⁻¹ in the as-dried state to ~1.2 kΩ·sq⁻¹ after FPH. Although these XPS trends qualitatively confirm changes in surface composition, a full quantitative determination of residual surfactant content would require further investigation.

 

4. Need to explain more about Figure 9.

We thank the reviewer for this comment. We have changed the description of the figure (now it is Figure 10), Page 13: Figure 10. High-resolution XPS spectra of CNT films before and after flash photonic heating (FPH).  (a) The C 1s region (≈284–290 eV) for the FPH-treated film (top) and the as-coated film (bottom); the black line is the measured signal with the red, blue, green, and purple curves correspond to the fitted C 1s (CH), C 1s (C–C), C 1s (C=O), and C 1s (COOH) signals, respectively. (b) shows the O 1s region (530–540 eV) for the FPH-treated film (top) and the as-coated film (bottom);  the black line is the measured signal with the red and blue curves are the O 1s (I) and O 1s (II) signals, respectively.

 

1. Fortunato, E., Ginley, D., Hosono, H. & Paine, D. C. Transparent conducting oxides for photovoltaics. MRS Bull 32, 242–247 (2007).
2. Hu, L., Hecht, D. S. & Gruner, G. Percolation in Transparent and Conducting Carbon Nanotube Networks. Nano Lett 4, 2513–2517 (2004).
3. Hecht, D. S. et al. High conductivity transparent carbon nanotube films deposited from superacid. Nanotechnology 22, (2011).
4. Jia, S. L. et al. Carbon nanotube-based flexible electrothermal film heaters with a high heating rate. R Soc Open Sci 5, (2018).
5. Kim, Y. J., Yu, S. J. & Jeong, Y. G. Carbon nanotube/polyimide bilayer thin films with high structural stability, optical transparency, and electric heating performance. RSC Adv 6, 30106–30114 (2016).
6. Carbon Nanotube Price: A Comparison Between Asia and Europe - info@graphenerich.com. https://graphenerich.com/carbon-nanotube-price-a-comparison-between-asia-and-europe/.
7. Carbon Nanotubes Price Guide in 2025 - Dazhan Nanomaterial. https://cnanotube.com/carbon-nanotubes-price-guide-in-2025/.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The study aims to develop a scalable, low-cost, and environmentally friendly method for fabricating transparent conductive films (TCFs) and flexible heaters using ultra-long multi-walled carbon nanotubes (UL-MWCNTs). The authors seek to overcome the limitations of Indium Tin Oxide (brittleness, cost) and Silver Nanowires (oxidation, instability) by utilizing a water-based flow-coating process. This manuscript can be published after addressing the following comments:

1. The title and text heavily emphasize the use of "Ultra-Long" CNTs (>100 um). However, the SEM images provided (Figure 3) show a dense network where individual tube lengths are difficult to verify. Can the authors provide length distribution data (histogram) or lower-magnification microscopy (TEM or SEM) of the dispersed nanotubes before coating to substantiate the ">100 um" claim?
2. If the tubes shorten significantly during the probe sonication process (2.5 hours at 85% amplitude is quite aggressive ), the "ultra-long" advantage may be diminished. A comparison of tube length before and after sonication would strengthen the paper.
3. The sheet resistance reported (1.2k at ~85% T) is functional for heaters but is relatively high compared to ITO (<50 ) or high-quality AgNW networks. How does the Figure of Merit (FoM) for these films compare to other recent CNT-based transparent heaters?
4. Have the authors performed long-term "ON/OFF" cycling tests (e.g., 100+ thermal cycles) to check for hysteresis in resistance or degradation of the CNT-polymer interface?
5. The XPS data (Figure 9, Table 1) shows a reduction in oxygen groups and an increase in sp² carbon. Does the Flash Photonic Heating cause any embedding of the CNTs into the polymer substrate (PET/PC)?

< !-- notionvc: 9f5e9317-4d2e-4552-84de-99c476a85958 -->

Author Response

Reviewer 2 comments:

1. The title and text heavily emphasize the use of "Ultra-Long" CNTs (>100 um). However, the SEM images provided (Figure 3) show a dense network where individual tube lengths are difficult to verify. Can the authors provide length distribution data (histogram) or lower-magnification microscopy (TEM or SEM) of the dispersed nanotubes before coating to substantiate the ">100 um" claim?

We thank the reviewer for this comment.  For this purpose, we diluted the film in a ratio of 1:1000 with deionized (DI) water and drop-casted it onto TEM grids to image individual CNTs. We added the TEM images on the revised manuscript; as seen, the CNTs formed tangled networks, while the length of the CNT is beyond the ~20 microns, while observed with the lowest magnification. We were unable to reliably trace the entire contour of the individual tubes, and therefore, could not present any histogram. For the original CNT it was impossible to make such evaluation because the CNTs were all embedded within a mat.

 

The following text and figure are added to the results and discussion Page 4:

 

To further examine the length of the ULCNTs in the dispersion, the ink was diluted in a ratio of 1:1000 in deionized (DI) water and drop-casted it onto TEM grids. As shown in the TEM image in Figure 3c and 3d, the ULCNTs form extended and entangled networks, with many segments exceeding 20 µm within a single field of view. Due to their extremely high aspect ratio and the tendency of ultra-long CNTs to coil and overlap, it was not possible to trace complete individual tubes or to construct a statistically meaningful length histogram. Nevertheless, the observable uninterrupted segments support that the nanotubes retain lengths well above the micron-scale regime after dispersion and sonication. For the as-received CNT mats, which consist of densely interwoven fibers, isolating individual CNTs for direct measurement was not feasible. These observations are consistent with the known morphology of ultra-long CNTs provided by the manufacturer and support their role in reducing junction density within the percolating network.

 

And in Figure 3 caption, page 5:

 

TEM images at magnifications of (c) 9,500X and (d) 100,000X, of the diluted ink (1:1000) in deionized (DI) water showing the ultra-long CNT.

 

2. If the tubes shorten significantly during the probe sonication process (2.5 hours at 85% amplitude is quite aggressive ), the "ultra-long" advantage may be diminished. A comparison of tube length before and after sonication would strengthen the paper.

 

We thank the reviewer for the comment. As described in the reply to comment 2, it was impossible to get a reliable number describing the average length.   

It should be noted that the exact same composition of ink but with a MWCNT from a different source (metallic multiwall CNTs from nanocyl with a length of only 1.5 micron) resulted in non condutive films.

 

3. The sheet resistance reported (1.2k at ~85% T) is functional for heaters but is relatively high compared to ITO (<50 ) or high-quality AgNW networks. How does the Figure of Merit (FoM) for these films compare to other recentCNT-based transparent heaters?

We thank the reviewer for this comment. following this comment, we calculated the FoM based on Haacke Figure of Merit¹. The calculated FoM for our system based on T ≈ 0.85 and Rₛ ≈ 1200 Ω·sq⁻¹ is which is comparable to heater made of SWCNT by Jia et al.² with a value of . When compared to MWCNT, the FoM value by Kim et al. ³ is 2.0 × 10⁻⁵ Ω⁻¹,  which is 8 times lower than ours.

 

The calculations are presented here:

where is the transmittance at 550 nm (as a fraction) and is the sheet resistance in Ω·sq⁻¹.

 

 

 

 

Taking the considiration of the cost of ULCNT, our approach definitly brings a significant cost advantage (3,000 $/Kg for ULCNT vs. >15,000 $/Kg for SWCNT [https://shop.nanografi.com/carbon-nanotubes/single-walled-carbon-nanotubes-purity-96-dia-1-0-nm/]).

 

The following text is added to the results and discussion Page 8:

 

To benchmark the optoelectronic performance of the ULCNT films, the Haacke Figure of Merit (Φ_TC = T¹⁰/Rₛ) was calculated and compared to recent CNT-based transparent heaters. For the present system (T ≈ 0.85, Rₛ≈ 1200 Ω·sq⁻¹), we obtain Φ_TC ≈ 1.6 × 10⁻⁴ Ω⁻¹. This value is on the same order of magnitude as that reported for SWCNT-based heaters by Jia et al.² (Φ_TC ≈ 4.0 × 10⁻⁴ Ω⁻¹), despite their much lower sheet resistance (Rₛ ≈ 95 Ω·sq⁻¹ at T ≈ 0.72). In contrast, multi-walled CNT films reported by Kim et al. ³ exhibit Φ_TC ≈ 2.0 × 10⁻⁵ Ω⁻¹, roughly an order of magnitude lower than our ULCNT system. Thus, although SWCNT networks still deliver the highest FoM values, the ULCNT heaters demonstrate substantially better optoelectronic performance than typical MWCNT-based transparent heaters while offering a significant material-cost advantage.

 

 

4. Have the authors performed long-term "ON/OFF" cycling tests (e.g., 100+ thermal cycles) to check for hysteresis in resistance or degradation of the CNT-polymer interface?

We thank the reveiwer for the comment and following that we have preformed the ON/OFF cycling test at 100 V with 1 min On, 1 min Off. This was repeated for 125 cycles, lasting for 15000 seconds. As seen in the figure below the heating appears to have no deterioration in the performance. In addition, the heater was left “on” for six hours and the temperature remained constant.

 

 

The following text and figure are added to the results and discussion Page 9:

 

To evaluate long-term operational stability, we performed ON/OFF thermal cycling of a single-layer ULCNT heater at 100 V using 1 min ON / 1 min OFF intervals for a total of 125 cycles (15,000 s). As shown in Figure 7, the maximum temperature reached during each On cycle remained consistent throughout the experiment, with no observable drift or hysteresis. Similarly, the baseline temperature during OFF cycles remained stable, indicating that neither joule heating nor repeated thermal expansion caused degradation of the CNT network or its interface with the PET substrate. In addition, the heater was left On for six hours and the temperature remained constant. These results demonstrate excellent cycling durability, outperforming brittle ITO-based heaters and aligning with the robust mechanical behavior observed in CNT networks.

 

 

Figure 7: Cyclic heating of a single-layer UL-CNT PET heater under 1 min ON / 1 min OFF operation at 100 V for 125 cycles (total duration 15,000 s).

 

5. The XPS data (Figure 9, Table 1) shows a reduction in oxygen groups and an increase in sp² carbon. Does the Flash Photonic Heating cause any embedding of the CNTs into the polymer substrate (PET/PC)?

We thank the reviewer for this comment. Top-view SEM images before and after FPH did not show any significant visual change. In view of the reduction in oxygen groups and an increase in sp² carbon could be that the CNT became embedded within the substrate or it could be that the FPH induced chemical reduction ^4,5 this should be thoroughly investigated in a future study.

 

The following text and figure is added to the results and discussion Page 12:

 

Top-view SEM images before and after FPH did not show any significant visual change. In view of the reduction in oxygen groups and an increase in sp² carbon could be that the CNT became embedded within the substrate or it could be that the FPH induced chemical reduction ^4,5, this should be thoroughly investigated in a future study.

1. Haacke, G. New figure of merit for transparent conductors. J Appl Phys 47, 4086–4089 (1976).
2. Jia, S. L. et al. Carbon nanotube-based flexible electrothermal film heaters with a high heating rate. R Soc Open Sci 5, (2018).
3. Kim, Y. J., Yu, S. J. & Jeong, Y. G. Carbon nanotube/polyimide bilayer thin films with high structural stability, optical transparency, and electric heating performance. RSC Adv 6, 30106–30114 (2016).
4. Zhou, Y. & Azumi, R. Carbon nanotube based transparent conductive films: progress, challenges, and perspectives. Sci Technol Adv Mater 17, 493 (2016).
5. Jung, S. et al. Microwave flash annealing for stability of chemically doped single-walled carbon nanotube films on plastic substrates. Nanoscale 6, 2971–2977 (2014).

 

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript can be published now