Conducting Interface for Efficient Growth of Vertically Aligned Carbon Nanotubes: Towards Nano-Engineered Carbon Composite

Vertically aligned carbon nanotubes (VACNT) are manufactured nanomaterials with excellent properties and great potential for numerous applications. Recently, research has intensified toward achieving VACNT synthesis on different planar and non-planar substrates of various natures, mainly dependent on the user-defined application. Indeed, VACNT growth has to be adjusted and optimized according to the substrate nature and shape to reach the requirements for the application envisaged. To date, different substrates have been decorated with VACNT, involving the use of diffusion barrier layers (DBLs) that are often insulating, such as SiO2 or Al2O3. These commonly used DBLs limit the conducting and other vital physico-chemical properties of the final nanomaterial composite. One interesting route to improve the contact resistance of VACNT on a substrate surface and the deficient composite properties is the development of semi-/conducting interlayers. The present review summarizes different methods and techniques for the deposition of suitable conducting interfaces and controlled growth of VACNT on diverse flat and 3-D fibrous substrates. Apart from exhibiting a catalytic efficiency, the DBL can generate a conducting and adhesive interface involving performance enhancements in VACNT composites. The abilities of different conducting interlayers are compared for VACNT growth and subsequent composite properties. A conducting interface is also emphasized for the synthesis of VACNT on carbonaceous substrates in order to produce cost-effective and high-performance nano-engineered carbon composites.


Importance of Carbon Nanotubes (CNT) in the Development of Innovative Materials
The properties of its components govern the performance of any composite material. Different assemblies of CNT, seen as hollow cylindrical carbon structures of sp 2 carbon, are indeed becoming part of the composite world within many industrial areas due to their remarkable properties. As previously reported [1][2][3], these manufactured carbon nanostructures are characterized and evaluated to exhibit extreme thermal and electrical conduction, mechanical performance, and many other extraordinary properties in various applications [4,5].
Once observed to possess highly interesting properties, today's research focuses on the effective production of CNT with different properties and arrangements, thus, expanding their application portfolio. The effective production of CNT is of high importance to develop industrial-grade, ready-to-use, CNT materials and overpass many challenges concerning production cost, energy consumption, environmental sustainability, material concept, and performance [6][7][8][9]. One of many, the VACNT, also known as CNT forests, arrays, or carpets, 3D substrates (CF tow, cloth, felt, paper, or other woven or non-woven fabrics) are advantageous since they possess exceptionally high mechanical strength and well-defined anisotropic thermal and electrical properties [53]. In addition, the fast development of 3D nano-to macro-scale composites by the growth of VACNT on fibrous carbon substrate presents even greater interest among the research and industry communities. Compared to various planar substrates, most fibrous carbons are flexible and porous matrixes with improved specific surface area, desired to prepare hierarchical composites in many applications. Respectively, the importance of various CF-CNT and CF-VACNT hybrids have been reported in scientific publications and patents [54,55].
However, not all substrates are suitable to support the required growth of VACNT. For example, Fe, Ni, or Co catalyst used for the direct growth of VACNT on Si is suppressed due to metal silicide or metal silicate formation at the synthesis temperature, i.e., the catalyst fails its primal activity [56]. Thus, the selection of the substrate nature mainly depends on the foreseen use and its compatibility to support the growth of VACNT with designed properties. Hence, some substrates/templates can only be used once covered with a suitable diffusion barrier layer (DBL), also known as interface, buffer, or even adhesion layer, interlayer or underlayer, and even catalyst support layer that promotes the growth of VACNT. Depending on its properties, one DBL should improve the catalyst activity (without alloying with the catalyst) and suppress its diffusion within the substrate at temperatures matching the nanoparticles formation and VACNT growth processes. Additionally, one DBL can promote adhesion and conductivity between the neighboring interfaces (substrate/DBL/catalyst/VACNT).

DBL towards Enhanced Growth of VACNT on Diverse Flat and 3-D Fibrous Substrates
Although beneficial for the homogeneous deposition of catalyst nanoparticles/growth of mm scale VACNT, most of the DBL used to date (such as Al 2 O 3 , SiO 2 , or MgO) [33,35,57,58] are generally insulators that limit the conducting contact between the VACNT and the substrate [59][60][61]. Unlike insulating DBL materials, semi-/-conducting DBLs show higher activity with the catalyst layer, such as interdiffusion and alloy formation that further affects the dewetting of the catalyst layer into a dense network of nanoparticles [62]. Therefore, different research groups are working on developing improved semi-/-conducting DBL for the growth of lengthy, dense, and tailored VACNT. Both semi-conducting and conducting DBL are suitable for final targeted applications compared to the generally insulating equivalents. Hence, in the following review, we will focus on both conducting and semiconducting material (hereunder mentioned as 'conducting') candidates for DBL.
To the best of our knowledge, it is critical to obtain an up-to-date comprehensive review that highlights and summarizes the development and use of conducting DBL for VACNT growth. Within the increasing demand for the development and fast application of high-performing composites, this review details the available methods to achieve the growth of VACNT in conducting DBLs. Special attention is given to the growth of VACNT on carbonaceous substrates and the essential use of suitable conducting DBL. Therefore, the review highlights the importance of the initial materials selection, substrate preparation, DBL deposition, and CVD procedures under which the growth of VACNT is performed, Research teams have deeply investigated the use of Al DBL, with different postdeposition treatments, for the growth of VACNT [46,51,[64][65][66][68][69][70][71][72].
As early as 2003, Ng et al. [64] presented Al DBL with Ni and Ni-Fe catalyst layer as an effective solution for the growth of short multi-walled VACNT. Burt et al. (2009) [65] investigated the deposition of Al DBL on the surface of native or introduced SiO2 as an Research teams have deeply investigated the use of Al DBL, with different postdeposition treatments, for the growth of VACNT [46,51,[64][65][66][68][69][70][71][72].
As early as 2003, Ng et al. [64] presented Al DBL with Ni and Ni-Fe catalyst layer as an effective solution for the growth of short multi-walled VACNT. Burt et al. (2009) [65] investigated the deposition of Al DBL on the surface of native or introduced SiO 2 as an effective way to grow single-wall VACNT with the use of Ni catalyst (detailed in Table 1). Modified morphologies of Al DBL were achieved using Si substrates with native or introduced SiO 2 top-layer and the change of substrate temperature during the PVD process, other conditions kept identical ( Figure 2, Group I). The CVD performance resulted in the growth of SWCNT, ranging from very short not-aligned to longer VACNT ( Figure 2, Group II). The observed differences in the CNT properties were explained by the different DBL surface roughness and the level of Al oxidation (annealing process in the presence of oxide substrate) on the native SiO 2 or the thick SiO 2 (formed with performed wet thermal oxidation). No analyses were performed to determine the properties and the potential use of the resulting composites.  [ 66] Nanomaterials 2022, 12, 2300 7 of 49 in the growth of SWCNT, ranging from very short not-aligned to longer VACNT ( Figure  2, Group II). The observed differences in the CNT properties were explained by the different DBL surface roughness and the level of Al oxidation (annealing process in the presence of oxide substrate) on the native SiO2 or the thick SiO2 (formed with performed wet thermal oxidation). No analyses were performed to determine the properties and the potential use of the resulting composites. Correlated to Burt et al. (2009) [65], Patole et al. (2013) [68] modified the grain size, i.e., the surface roughness of an Al DBL with the change of the deposition rate instead of the change of substrate temperature during PVD. Thus, the authors have revealed the DBL structure, morphology, and rearrangement (during the annealing and CNT synthesis process) to influence the growth of VACNT with modified CNT properties, such as CNT alignment, height, and diameter (details given in Table 1). Overall, the results present the beneficial use of large grains Al DBL topography (with enhanced rugosity) to support the formation of catalyst nanoparticles and allow higher carbon supply for the growth of small diameter, highly aligned, and long CNT. Even though not mentioned, the presence/formation of Al2O3 in the Al DBL could not be evaded due to the presence of native SiO2 and the post-PVD air transfer of the samples before the CVD process. The enhanced presence of Al2O3 is well defined by other authors [46,65,66]. For example, Correlated to Burt et al. (2009) [65], Patole et al. (2013) [68] modified the grain size, i.e., the surface roughness of an Al DBL with the change of the deposition rate instead of the change of substrate temperature during PVD. Thus, the authors have revealed the DBL structure, morphology, and rearrangement (during the annealing and CNT synthesis process) to influence the growth of VACNT with modified CNT properties, such as CNT alignment, height, and diameter (details given in Table 1). Overall, the results present the beneficial use of large grains Al DBL topography (with enhanced rugosity) to support the formation of catalyst nanoparticles and allow higher carbon supply for the growth of small diameter, highly aligned, and long CNT. Even though not mentioned, the presence/formation of Al 2 O 3 in the Al DBL could not be evaded due to the presence of native SiO 2 and the post-PVD air transfer of the samples before the CVD process. The enhanced presence of Al 2 O 3 is well defined by other authors [46,65,66]. For example, Choi et al. (2010) [66] considered the use of Al DBL for the formation of high-density Al 2 O 3 and its effect on the formation of stable Fe 2 O 3 catalytic nanoparticles for the growth of mm scale VACNT ( Figure 3). The latter was achieved with the PVD of the DBL on a thick SiO 2 surface and the use of CVD annealing (20 min)/CNT synthesis at a high temperature of 780 • C (details given in Table 1). The study generally presented a controllable way to optimize vertical alignment, growth rate, and density of VACNT by adjusting the DBL and catalyst layer thickness. The conductivity of the above-presented DBLs [65,66,68] should be heavily reduced compared to a pure Al film due to the enhanced oxidation of the Al for improved growth of VACNT. However, multiple studies confirm the necessity of a well-balanced compromise between the conductivity of a DBL (metal purity) and the vital presence of an DBL on a thick SiO2 surface and the use of CVD annealing (20 min)/CNT synthesis at a high temperature of 780 °C (details given in Table 1). The study generally presented a controllable way to optimize vertical alignment, growth rate, and density of VACNT by adjusting the DBL and catalyst layer thickness. The conductivity of the above-presented DBLs [65,66,68] should be heavily reduced compared to a pure Al film due to the enhanced oxidation of the Al for improved growth of VACNT. However, multiple studies confirm the necessity of a well-balanced compromise between the conductivity of a DBL (metal purity) and the vital presence of an oxide phase to anchor the active catalyst nanoparticles and endure the tunable growth of VACNT. To advance conducting Al DBL in application composites, other teams have used different substrates and modified methods to decrease the unavoidable occurrence of native aluminium oxide [51,67,72,73,91]. To lower the contact resistance, Dijon et al. (2010) [67] have deposited Cu-Al (99.5%) alloy on Si wafer as DBL for the VACNT growth. The presence of Cu should have influenced the lower oxidation of the Al DBL surface. Additionally, the authors have de-oxidised (wet-etch method) the DBL before thin Fe catalyst film deposition. Thus, they observed a bottom contact resistance decrease of a factor of 100 when compared to the use of idem DBL without the deoxidation step. However, they also concluded that the contact force of the AFM tip probe on the surface of the CNT carpet, i.e., change of contact area, has also influenced differences in the measured resistance (detailed in Table 2). Shah et al. (2013) [70] used low resistivity n-type silicon (100) wafers with a 300 nm thick Al DBL to grow short VACNT. Field emission studies present the improved current density of 20 mA cm −2 at a field of 3.5 V µm −1 with VACNT-Al DBL samples compared to 0.41 mA cm −2 of the directly grown VACNT sample. The reduction in contact resistance is attributed to Al DBL that formed thin Al2O3 film (offering an electron-tunneling effect) compared to the thicker native SiO2 layer on the lone Si substrate. Zhang et al. (2015) [51] performed the deposition of different thicknesses of Al DBL on Cu substrate to grow VACNT. Additionally, they investigated the interfacial adhesion properties of the complete composite (details given in Table 2). The authors performed floating catalyst CVD with a constant injection of active Fe nanoparticles on the DBL's top-most surface. Moreover, thicker DBL was beneficial for the growth of the highest evidenced VACNT. The interfacial adhesion To advance conducting Al DBL in application composites, other teams have used different substrates and modified methods to decrease the unavoidable occurrence of native aluminium oxide [51,67,72,73,91]. To lower the contact resistance, Dijon et al. (2010) [67] have deposited Cu-Al (99.5%) alloy on Si wafer as DBL for the VACNT growth. The presence of Cu should have influenced the lower oxidation of the Al DBL surface. Additionally, the authors have de-oxidised (wet-etch method) the DBL before thin Fe catalyst film deposition. Thus, they observed a bottom contact resistance decrease of a factor of 100 when compared to the use of idem DBL without the de-oxidation step. However, they also concluded that the contact force of the AFM tip probe on the surface of the CNT carpet, i.e., change of contact area, has also influenced differences in the measured resistance (detailed in Table 2). Shah et al. (2013) [70] used low resistivity n-type silicon (100) wafers with a 300 nm thick Al DBL to grow short VACNT. Field emission studies present the improved current density of 20 mA cm −2 at a field of 3.5 V µm −1 with VACNT-Al DBL samples compared to 0.41 mA cm −2 of the directly grown VACNT sample. The reduction in contact resistance is attributed to Al DBL that formed thin Al 2 O 3 film (offering an electron-tunneling effect) compared to the thicker native SiO 2 layer on the lone Si substrate. Zhang et al. (2015) [51] performed the deposition of different thicknesses of Al DBL on Cu substrate to grow VACNT. Additionally, they investigated the interfacial adhesion properties of the complete composite (details given in Table 2). The authors performed floating catalyst CVD with a constant injection of active Fe nanoparticles on the DBL's top-most surface. Moreover, thicker DBL was beneficial for the growth of the highest evidenced VACNT. The interfacial adhesion strength was observed to increase with the Al DBL thickness (from around 0.2 to 0.4 MPa for 10 to 30 nm Al DBL composites, respectively) using a lab-modified pull-off/compression test apparatus. However, the adhesion properties of the composite with Al DBL were much lower compared to the same with Al 2 O 3 DBL. Thus, other studies (discussed later on) present the use of bi-metal (including Al) DBL to improve the overall composite mechanical properties. Zhong et al. (2016) [73] demonstrate the growth of high quality/density SWCNT forests on commercially available Cu foils via cold-wall CVD (details given in Table 2). The authors confirm the presence of thin native AlOx film on the Al DBL (due to air transfer and annealing process) that did not suppress the examined Ohmic contact between the SWCNT and the Cu substrate. The electrical properties of the SWCNT forests-Cu composite were investigated with a two-point probe station, yielding a very low overall resistivity of 60-80 Ohms. The electrical properties of the examined SWCNT forests were presented as highly promising for applications such as supercapacitors, batteries, and thermal interfaces. Pitkänen et al. (2015) [72] produced a library of conducting substrates with an Al interface effectively used for the growth of VACNT (details given in Table 2). A range of 5-10-20 nm was presented as effective DBL thickness for the growth of VACNT on all examined substrates. However, applying a 10 nm thick Al interface was found to support the longest VACNT on most of the studied substrates ( Figure 1 within [72]). According to the electrical properties measurements, using an Al DBL (with the presence of thin native Al 2 O 3 top-film) between the Fe catalyst and a conducting substrate did not increase the overall composite resistivity, compared to the same without the Al DBL. The latter indicates the possible use of the studied composites for the production of CNT-based electrodes, useful in many applications. For example, the total electrical resistance of VACNT-10nm Al DBL-Inconel ® 600 composite was found to be very low and below 10 Ω. Whereas, specific capacitances were measured to be 8.1 ± 0.6, 7.9 ± 0.4, and 10.0 ± 0.4 F g −1 for VACNT-10nm Al DBL on Inconel ® 600, Cu, and stainless steel substrate, respectively. Overall, the use of appropriate Al DBL on conducting substrate can be seen as a simple and tuneable solution to grow VACNT for various applications. Li et al. (2017) [91] proposed the use of multilayer Ni/Al/Ni catalyst/DBL/catalyst (sandwiched DBL) solution for the growth of VACNT via plasma-enhanced CVD (PECVD) on TiSi conducting substrate, at low temperature of 650 • C (details given in Table 2). The use of the Al interlayer improved the fine roughness of the top catalyst layer (prevent Ostwald ripening effect) and controlled the sub-surface diffusion of the catalyst in order to produce the growth of VACNT. On the other hand, the control sample with lone Ni catalyst did not form VACNT, all other conditions kept identical ( Figure 4). The performed successive 2 nm Ni/1 nm Al/2 nm Ni PVD prevent the oxidation of the Al interlayer (up to the formation of Ni catalyst nanoparticles) that reinforced the Ohmic contact between the substrate and the VACNT, with a total resistance of 540 Ω. Moreover, the use of the presented or similar promising multilayers, with sandwiched Al DBL, was not tested with pure-thermal CVD to validate the growth of VACNT without the plasma-enhanced alignment effect on the CNT.   PVD deposited Ta metallic layer was presented by Nessim et al. (2009) [62] (2010) [75] for the growth of 2 µm tall VACNT via fast-heat thermal CVD. The technique allowed preheating the reducing carrier gas and the carbon precursor in a hot zone before the growth zone to avoid thermal degradation of the used substrate and improve catalyst stabilization/activity, leading to successful VACNT growth (details found in Table 3). The same authors measured the electrical resistance of 2 µm-tall VACNT grown on Fe (5 nm)/Ta (30 nm) catalyst/DBL with the use of AFM, equipped with a metal-coated (Pt-Ir) conducting tip. The measured resistance, at around 35 kΩ, was defined as a good indication of good electrical contact between the CNT root and the metallic DBL. However, the previous study of Wang et al. [76] (and references within) has already shown a growth  Other authors have explored alternative routes to directly use high-surface-energy (higher than the one of Al 2 O 3 ), conducting metals or metal alloys/silicides/nitrides/silicidenitrides to achieve controlled growth of VACNT. The use of conducting DBLs is evidence of difficulty stabilizing metallic nanoparticles and the tendency of the catalyst to unfavorably alloy with and/or diffuse within the DBL at the conditions for VACNT growth. If the growth of VACNT is achieved on conducting DBL, the same are examined as less aligned, shorter/thicker CNT that form lower density carpets than an Al 2 O 3 DBL. Therefore, the ultimate solution is still to be explored for different conducting DBL-catalyst combinations that meet the presented drawbacks.
PVD deposited Ta metallic layer was presented by Nessim et al. (2009) [62] (2010) [75] for the growth of 2 µm tall VACNT via fast-heat thermal CVD. The technique allowed preheating the reducing carrier gas and the carbon precursor in a hot zone before the growth zone to avoid thermal degradation of the used substrate and improve catalyst stabilization/activity, leading to successful VACNT growth (details found in Table 3). The same authors measured the electrical resistance of 2 µm-tall VACNT grown on Fe (5 nm)/Ta (30 nm) catalyst/DBL with the use of AFM, equipped with a metal-coated (Pt-Ir) conducting tip. The measured resistance, at around 35 kΩ, was defined as a good indication of good electrical contact between the CNT root and the metallic DBL. However, the previous study of Wang et al. [76] (and references within) has already shown a growth of VACNT of greater length (8-10 µm) on supports with Ta DBL via simple catalytic CVD. Moreover, the use of Ta as a DBL is observed as an insufficiently tunable method for enhanced growth of VACNT of high density and length. Thus, Ta as conducting DBL is explored through the use of TaN ( Figures 5 and 6). Our research group has already been working on developing TaNx DBL for the growth of VACNT via simple AACCVD on Si substrate. TaNx thin DBL have been deposited on Si via conventional [77] or high-power pulse [78] magnetron sputtering (details given in Table 3). The modification of the N 2 flow during the PVD in reactive plasma N 2 /Ar with Ta target resulted in the deposition of DBLs with different N/Ta atomic ratios and structure that was observed to be altered after the annealing process [78]. The latter was evident to directly impact the catalyst settlement/nanoparticles formation/activity, i.e., the growth and properties of VACNT ( Figure 6). Thus, neat and high-quality MWCNT forests were grown, with their diameter and length decreasing as the nitrogen content increased in the TaNx layers. Moreover, preformed material characterizations confirmed an exchange of the N atoms on the DBL surface with O atoms, found as residue in the CVD reactor. The presence of incorporated O atoms on the DBL surface proportionally increased with N atoms, yet always found to be 6 times lower. The oxidation of the TaNx, results in the formation of Ta 2 O 5 in unstable under-stoichiometric α-Ta(N) phases, or a ternary segment of FeTa 2 O 6 (during the CNT growth) at the stoichiometric nanocrystalline TaN phase, leading to CNT growth with different properties (shorter and larger or longer and thinner, respectively). Yet, the formation of oxide or ternary surface phases should significantly influence the electrical conductivity of the DBL and the overall composite.    [112] have explored the influence of alloy composition, alloy concentration, and alloy thickness on the CNT growth and the impact of the CVD gas phase composition, temperature, and pressure. They detail the PVD of alloys with    [112] have explored the influence of alloy composition, alloy concentration, and alloy thickness on the CNT growth and the impact of the CVD gas phase composition, temperature, and pressure. They detail the PVD of alloys with  [112] have explored the influence of alloy composition, alloy concentration, and alloy thickness on the CNT growth and the impact of the CVD gas phase composition, temperature, and pressure. They detail the PVD of alloys with catalytic metal (Ct, such as Ni, Co, Fe, Pd) and transition metal (Me, such as Ti, V, Cr, Zr, Nb, Mo, Ta, W, Re) in an inert or a mixture of argon and nitrogen atmosphere for the achievement of conducting DBLs for the growth of VACNT. As previously published [113], the same team achieved a successful growth of well-developed VACNT only on Co-Zr-N-(O) alloy with specified properties of the DBL and CVD process of CNT growth (given in Table 3). Thus, they achieved growth of multi-layered VACNT with the rate of~250 nm s −1 and a maximal CNT length of around 35 µm (Figure 7). Notably, the reduction in the DBL alloy during the annealing process was observed to decrease the presence and density of the top-surface active catalyst and suppress the growth of VACNT. The latter is connected with the reduced presence of native oxygen during the DBL restructuration at higher temperatures and in reactive NH 3 atmosphere. The authors present the PVD deposition/CVD synthesis conditions with mutual importance for having both well-structured catalyst-DBL metalalloy and successful VACNT growth process. Moreover, they confirm the complexity of the overall process due to the large number of factors that clearly have an immense impact on the CVD growth mechanism and demands further investigations towards full understanding.  [113], the same team achieved a successful growth of well-developed VACNT only on Co-Zr-N-(O) alloy with specified properties of the DBL and CVD process of CNT growth (given in Table 3). Thus, they achieved growth of multi-layered VACNT with the rate of ~250 nm s −1 and a maximal CNT length of around 35 µm ( Figure 7). Notably, the reduction in the DBL alloy during the annealing process was observed to decrease the presence and density of the top-surface active catalyst and suppress the growth of VACNT. The latter is connected with the reduced presence of native oxygen during the DBL restructuration at higher temperatures and in reactive NH3 atmosphere. The authors present the PVD deposition/CVD synthesis conditions with mutual importance for having both well-structured catalyst-DBL metal-alloy and successful VACNT growth process. Moreover, they confirm the complexity of the overall process due to the large number of factors that clearly have an immense impact on the CVD growth mechanism and demands further investigations towards full understanding. Alternatively, titanium nitride, titanium silicide nitride, or aluminium silicide with attractive electronic properties are presented as conducting DBLs for the growth of VACNT for applications such as supercapacitors and micro-electronic devices/sensors/interconnectors. The use of conducting TiN DBL can be traced even as far as 2004. de los Arcos et al. (2004) [81] have grown VACNT by pure-thermal CVD using DBL such as Al2O3, TiN, and TiO2 with Fe catalyst. VACNT growth was presented only on Al2O3 and TiN with a growth rate of 5 and 2 µm min −1 . Compared to the Al2O3 DBL, TiN did not support the growth of dense and well-aligned low-diameter CNT, yet it was recognized as promising to conduct DBL for further development. Subsequently, Garcia-Céspedes et al. (2009) [83] performed the deposition of TiN DBL on Cu substrate for the growth of VACNT via ferrocene injection (catalytic) CVD (details given in Table 4). The constant feed of active catalyst nanoparticles on the DBL surface resulted in areas of VACNT on the covered Cu faces, with a length between 20 and 30 µm and a dispersed diameter range between 10 and 100 nm ( Figure 8). Hence, the thicker CNT were aligned, whereas the thinner ones were generally randomly orientated. The VACNT presented bamboo-like multi-wall morphology, with wall defects and a high presence of incorporated catalyst particles. Apart from the effect of the DBL, the low CNT carpet homogeny is presumably impacted by an asymmetrical dispersion of the Fe catalyst. However, the  Table 4). The constant feed of active catalyst nanoparticles on the DBL surface resulted in areas of VACNT on the covered Cu faces, with a length between 20 and 30 µm and a dispersed diameter range between 10 and 100 nm ( Figure 8). Hence, the thicker CNT were aligned, whereas the thinner ones were generally randomly orientated. The VACNT presented bamboo-like multi-wall morphology, with wall defects and a high presence of incorporated catalyst particles. Apart from the effect of the DBL, the low CNT carpet homogeny is presumably impacted by an asymmetrical dispersion of the Fe catalyst. However, the authors point out a null diffusion of the Fe catalyst into the TiN DBL by performing composition depth profiling via Auger electron spectroscopy and secondary ion mass spectrometry trace detection on the surface of the coated Cu substrates. Finally, the catalyst stability on the established TiN DBL roughness could have played an important role concerning the presented CNT properties. authors point out a null diffusion of the Fe catalyst into the TiN DBL by performing composition depth profiling via Auger electron spectroscopy and secondary ion mass spectrometry trace detection on the surface of the coated Cu substrates. Finally, the catalyst stability on the established TiN DBL roughness could have played an important role concerning the presented CNT properties.  [85] reported TiN properties to promote catalyst stability/activity at a high number density and suppress subsurface diffusion to achieve VACNT growth on conducting DBL. The TiN DBL/Fe catalyst substrate composite supported VACNT growth, though, with much shorter and lower quality CNT due to the low activity and shorter catalyst lifetime than Al2O3 DBL ( Figure 9). The observed CNT growth on TiN DBL was attributed to enhanced Ostwald ripening processes in the early growth phases and due to the observed subsurface diffusion of the catalyst. Moreover, the catalyst structural promotion during annealing and CNT growth processes is speculated to modify the interactions between the catalyst nanoparticles and the carbon precursor gas molecules. Finally, the authors do not exclude different results if other thermal CVD techniques are used instead of the cooled wall, hot filament CVD. In this context, Campo et al. (2019) [88] presented improved growth of denser and higher SWCNT forests via hot wall alcohol catalytic (AC-) CVD, compared to the previous results [85]. To increase the catalyst density and effectiveness, the authors used a dip-coating DBL deposition method of Mo-Co acetate precursor solutions (information on TiN DBL/catalyst deposition, CVD growth of VACNT, and the morphology of the latter are detailed in Table 4). Thus, the role of Mo is to suppress the coalescence of Co nanoparticles during the calcination process, the latter being the real catalyst for the growth of VACNT. Moreover, the growth of VACNT was not observed over some TiN DBL zones due to its low thickness (~200 nm) (Figure 10a) [88]. As observed previously [114], the authors evidenced VACNT absence when the TiN was neighboring a layer of SiO2, instead of sputtered Si, due to the higher surface free energy (Defined by the thermodynamics of surface states, in equilibrium, a solid composite tends to organize to the lowest surface free energy, thus, the catalyst favors a surface with a higher surface free energy, resulting in a stable composite system [88]) of SiO2 compared to TiN (64.7 and 40.6 mJ m −2 , respectively). Therefore, both studies confirm that the properties of the used DBL and other surface layers in the vicinity (with uneven properties) can influence the preferential deposition/stabilization of the catalyst, i.e., the growth of VACNT.  [85] reported TiN properties to promote catalyst stability/activity at a high number density and suppress subsurface diffusion to achieve VACNT growth on conducting DBL. The TiN DBL/Fe catalyst substrate composite supported VACNT growth, though, with much shorter and lower quality CNT due to the low activity and shorter catalyst lifetime than Al 2 O 3 DBL ( Figure 9). The observed CNT growth on TiN DBL was attributed to enhanced Ostwald ripening processes in the early growth phases and due to the observed subsurface diffusion of the catalyst. Moreover, the catalyst structural promotion during annealing and CNT growth processes is speculated to modify the interactions between the catalyst nanoparticles and the carbon precursor gas molecules. Finally, the authors do not exclude different results if other thermal CVD techniques are used instead of the cooled wall, hot filament CVD. In this context, Campo et al. (2019) [88] presented improved growth of denser and higher SWCNT forests via hot wall alcohol catalytic (AC-) CVD, compared to the previous results [85]. To increase the catalyst density and effectiveness, the authors used a dip-coating DBL deposition method of Mo-Co acetate precursor solutions (information on TiN DBL/catalyst deposition, CVD growth of VACNT, and the morphology of the latter are detailed in Table 4). Thus, the role of Mo is to suppress the coalescence of Co nanoparticles during the calcination process, the latter being the real catalyst for the growth of VACNT. Moreover, the growth of VACNT was not observed over some TiN DBL zones due to its low thickness (~200 nm) ( Figure 10a) [88]. As observed previously [114], the authors evidenced VACNT absence when the TiN was neighboring a layer of SiO 2 , instead of sputtered Si, due to the higher surface free energy (Defined by the thermodynamics of surface states, in equilibrium, a solid composite tends to organize to the lowest surface free energy, thus, the catalyst favors a surface with a higher surface free energy, resulting in a stable composite system [88]) of SiO 2 compared to TiN (64.7 and 40.6 mJ m −2 , respectively). Therefore, both studies confirm that the properties of the used DBL and other surface layers in the vicinity (with uneven properties) can influence the preferential deposition/stabilization of the catalyst, i.e., the growth of VACNT. A comparative study for pure TiN DBL deposited via PVD or ALD technique is detailed in Section 2.3, highlighting the advantageous use of ALD over PVD towards more effective catalyst stability and growth of VACNT [86]. Nanomaterials 2022, 12, x FOR PEER REVIEW 13 of 54  A comparative study for pure TiN DBL deposited via PVD or ALD technique is detailed in Section 2.3, highlighting the advantageous use of ALD over PVD towards more effective catalyst stability and growth of VACNT [86].
Differently, Yang et al. (2014) [89] have achieved robust growth of VACNT, with improved CNT density and length on TiSiN DBL, as compared to TiN of the identical thickness (Figure 11a,b). Due to a lack of grain boundaries, thin TiSiN films beneficially supported a prolonged catalytic activity of homogeneously-sized nanoparticles, kept  A comparative study for pure TiN DBL deposited via PVD or ALD technique is detailed in Section 2.3, highlighting the advantageous use of ALD over PVD towards more effective catalyst stability and growth of VACNT [86].
Differently, Yang et al. (2014) [89] have achieved robust growth of VACNT, with improved CNT density and length on TiSiN DBL, as compared to TiN of the identical thickness (Figure 11a,b). Due to a lack of grain boundaries, thin TiSiN films beneficially supported a prolonged catalytic activity of homogeneously-sized nanoparticles, kept  [89] have achieved robust growth of VACNT, with improved CNT density and length on TiSiN DBL, as compared to TiN of the identical thickness (Figure 11a,b). Due to a lack of grain boundaries, thin TiSiN films beneficially supported a prolonged catalytic activity of homogeneously-sized nanoparticles, kept immobilized throughout the CVD process. Consequently, the authors produced denser VACNT of low diameter CNT using 1 nm Fe catalyst film on TiSiN, compared to the thicker 5 nm catalyst film on TiN (details provided in Table 4). The TiSiN was also proven as an effective DBL to support Fe catalyst/VACNT even at a temperature of 800 • C (Figure 11d,f). Contrary, TiN DBL did not yield VACNT due to the enhanced Fe diffusion into the TiN bulk at the identical CVD conditions (Figure 11c,e). Moreover, TiSiN DBL substrate with low surface energy (less than 0.1 eV nm −2 , compared to~3.06 eV nm −2 for TiN [90]) disabled both subsurface diffusion and Ostwald ripening processes, allowing catalyst de-wetting into high-density nanoparticles (Figure 11c,d). The latter favored forest growth with area densities of the order~10 12 CNT cm −2 , compared to~10 10 CNT cm −2 on TiN DBL, all other parameters kept identical. In both cases, Ohmic contact was examined between the substrate and the VACNT with resistance values of 0.8 and 3.4 kΩ for TiSiN and TiN, respectively. Yuan et al. (2015) [90] have also presented the effectiveness of matching TiSiN DBL with the use of ultra-thin Fe catalyst film and Al top-layer, with an analyzed thickness of 0.4 and 0.1 nm, respectively (details provided in Table 5). Furthermore, Al top-layer was evidenced to suppress the Fe catalyst surface (shallow) diffusion. Compared to their previous study results [89], the latter presented a superior areal density of thinner CNT with double/triple walls ( Figure 12). Moreover, the irregular (post-annealing) ultrathin Al top-layer (with a native presence of alumina) was observed not to affect the composite resistance. Thus, the TiSiN is presented as an effective conducting DBL with matching or enhanced performance compared to some insulating DBL [115] as of CNT properties and areal density [89,90].
immobilized throughout the CVD process. Consequently, the authors produced denser VACNT of low diameter CNT using 1 nm Fe catalyst film on TiSiN, compared to the thicker 5 nm catalyst film on TiN (details provided in Table 4). The TiSiN was also proven as an effective DBL to support Fe catalyst / VACNT even at a temperature of 800 °C (Figure 11d,f). Contrary, TiN DBL did not yield VACNT due to the enhanced Fe diffusion into the TiN bulk at the identical CVD conditions (Figure 11c,e). Moreover, TiSiN DBL substrate with low surface energy (less than 0.1 eV nm −2 , compared to ~3.06 eV nm −2 for TiN [90]) disabled both subsurface diffusion and Ostwald ripening processes, allowing catalyst de-wetting into high-density nanoparticles (Figure 11c,d). The latter favored forest growth with area densities of the order ~10 12 CNT cm −2 , compared to ~10 10 CNT cm −2 on TiN DBL, all other parameters kept identical. In both cases, Ohmic contact was examined between the substrate and the VACNT with resistance values of 0.8 and 3.4 kΩ for TiSiN and TiN, respectively. Yuan et al. (2015) [90] have also presented the effectiveness of matching TiSiN DBL with the use of ultra-thin Fe catalyst film and Al toplayer, with an analyzed thickness of 0.4 and 0.1 nm, respectively (details provided in Table 5). Furthermore, Al top-layer was evidenced to suppress the Fe catalyst surface (shallow) diffusion. Compared to their previous study results [89], the latter presented a superior areal density of thinner CNT with double/triple walls ( Figure 12). Moreover, the irregular (post-annealing) ultrathin Al top-layer (with a native presence of alumina) was observed not to affect the composite resistance. Thus, the TiSiN is presented as an effective conducting DBL with matching or enhanced performance compared to some insulating DBL [115] as of CNT properties and areal density [89,90].   [90], the potential use of different conducting DBL can be further improved by applied surface modifications, such as metal topping layer, that enhance the growth of VACNT with enhanced properties. Thus, well-defined VACNT have also been grown on bi-layer DBL, using Al top-layer. Moreover, researchers have investigated the use of Al DBL before introducing the catalyst or simultaneous   [90], the potential use of different conducting DBL can be further improved by applied surface modifications, such as metal topping layer, that enhance the growth of VACNT with enhanced properties. Thus, well-defined VACNT have also been grown on bi-layer DBL, using Al top-layer. Moreover, researchers have investigated the use of Al DBL before introducing the catalyst or simultaneous sputtering of both catalyst and Al on metal/metal nitride DBL [34,60,67,71,74,87]. Thus, a thin capping layer of Al has been evidenced as a crucial part of one conducting DBL structure toward tunable growth of VACNT on different substrates. Multi-layer DBL is a beneficial solution to improve interlayer adhesion, suppress catalyst bulk diffusion and improve VACNT growth and properties, and at the same time retain good Ohmic contact between each composite component. Tas (2020) [60] and Ahmad et al. (2015) [87] introduced (additionally sputter-deposited) a thin Al top-layer (~10 nm) over the TiN before the annealing and VACNT growth processes. The Al topping layer was introduced to hinder the catalyst agglomeration during annealing, i.e., acting as an anchor for the formed catalyst nanoparticles. Moreover, the post-formation of Al 2 O 3 from the Al top-layer is not excluded in both studies during the substrate annealing process. Though, the electric resistance of the presented combination is found to be as low as 10 Ω (as for as grown VACNT network probe), even after the annealing process and the possible formation of native Al 2 O 3 [60]. Thus, TiN/Al solution is presented as an improved version of TiN alone with minimized interfacial resistance (TiN can be easily oxidized, adding resistance [79]) and improved catalyst stability/activity, reflected in tunable growth of high quality/length VACNT. A photo-thermal (PT)-CVD technique was used in both studies to grow VACNT. Figure 13 [60] and Figure 14 [87] compare and represent the length, and length and diameter, respectively, of VACNT grown on TiN DBL of 100 nm thickness. For [60], details of TiN DBL/Al top-layer/catalyst PVD deposition, CVD growth of VACNT, and the morphology of the latter are given in Table 5. The use of good electrically-conducting TiN as thermal insulation (8.92 W m −1 K −1 at 1000 • C) is highly suitable in PTCVD for protecting the substrate from severe thermal degradation. Thus, the substrates used in the cited studies are reported to stay at a very low temperature between 415-455 • C [60] and 350-440 • C [87] during the PTCVD growth of VACNT. Imperatively, TiN DBL and Al top-layer demonstrate good optical reflectance of infrared (IR) radiation (wavelength of 700-1000 nm) that further enhances the growth of VACNT on heated up, both from direct and reflected IR radiation. Vollebregt et al. (2014) [71] used TiN DBL with Al/Co top-layer as a DBL top-cover and catalyst layer. The Co catalyst was beneficially used to grow VACNT at temperatures as low as 350 • C due to the need for much lower activation energy than for the Fe catalyst (details given in Table 5). Favorable, the presence of Al further enhanced the Co nanoparticle stability that resulted in VACNT. For example, the top layer of 3 nm alloy Co-Al (28%) supported VACNT of identical density compared to the 5 nm lone Co catalyst layer.
Bi-metal Mo/Al was presented as more promising in conducting DBL than other bi-metal solutions [34]. In addition, the use of unexplored Nb instead of Mo was observed as a possible alternative since both metals have similar oxidation resistance/mechanical properties at high temperatures and high melting points [116]. Mo alone did not present growth of CNT due to the strong alloying with the Fe catalyst and the loss of catalytic activity. The electrical resistance of the Mo DBL after the post-CNT growth process was examined to be lower than the same before the CVD process. In contrast, the use of a sole Al layer resulted in the growth of VACNT with high CNT areal density. Yet, the post-synthesis resistance of the Al layer was observed to be many orders of magnitude higher than the initial one. A possible explanation of the latter is the additional formation of a thick insulating oxide layer, hypothesized to be suppressed in the presence of Mo under-/-layer. The bi-metal Mo/Al DBL presented low contact resistance of 26 Ω (details in Table 5 [71] used TiN DBL with Al/Co top-layer as a DBL top-cover and catalyst layer. The Co catalyst was beneficially used to grow VACNT at temperatures as low as 350 °C due to the need for much lower activation energy than for the Fe catalyst (details given in Table 5). Favorable, the presence of Al further enhanced the Co nanoparticle stability that resulted in VACNT. For example, the top layer of 3 nm alloy   [71] used TiN DBL with Al/Co top-layer as a DBL top-cover and catalyst layer. The Co catalyst was beneficially used to grow VACNT at temperatures as low as 350 °C due to the need for much lower activation energy than for the Fe catalyst (details given in Table 5). Favorable, the presence of Al further enhanced the Co nanoparticle stability that resulted in VACNT. For example, the top layer of 3 nm alloy Overall, the use of PVD Al DBL has been investigated through different studies showing the effect of the DBL surface morphology and structure, and the level of Al surface oxidation over the process of VACNT growth and their properties, or the properties of the final composite. In general, the vital presence of an oxide top-cover has been confirmed to anchor the active catalyst nanoparticles and enhance the growth of VACNT, regardless of its possible negative effect on the overall composite conductivity. Moreover, research teams have used different substrates and modified methods to decrease the unavoidable presence of native aluminium oxide to lower the presence of contact resistance.
Apart from the well-examined use of lone Al DBL, other authors have explored different conducting metals or metal alloys/silicides/nitrides/silicide-nitrides to achieve controlled VACNT growth. Finally, progress is made to use PVD conducting DBLs to boost conducting properties and simultaneously surface-stabilize catalytic nanoparticles with prolonged catalytic activity.

Via Atomic Layer Deposition (ALD)
Contrary to the frontal coverage with PVD, atomic layer deposition (ALD) could achieve dense and atomic-scale precision films on planar and non-planar substrates with screened surfaces. Recently, several studies have discussed the differences between an ALD and PVD (electron-beam or DC magnetron sputtering) techniques for the deposition of DBL and their later use for the growth of VACNT on flat and 3-D structures [86,93,117]. However, not many have presented the use of ALD for the deposition of DBL with improved conductivity properties.
Contrary to the ineffective use of PVD TiO 2 as DBL [81],  presented the use of ALD for the deposition of metal oxide TiO 2 as DBL for the growth of VACNT [93]. Unlike the growth of long and dense VACNT carpets on Al 2 O 3 and SiO 2 , the TiO 2 DBL supported only short and low-density multi-walled CNT carpets with marginal alignment (Figure 15) due to severe sub-surface diffusion of the Fe catalyst nanoparticles or the high surface activation energy. Moreover, not all Fe nanoparticle catalysts achieved a CNT nucleation and growth due to the high surface activation energy on TiO 2 DBL compared to SiO 2 and Al 2 O 3 . The Ostwald ripening effect should have also influenced the density of the VACNT, yet to a lower order. The latter can be correlated with the presence of CNT of similar morphology on the tested DBLs. Multi-walled VACNT have been grown on ALD ZnO films only with PECVD [118] in the presence of NH 3 [94], whereas the same was not achieved when pure-thermal and PECVD was performed in the presence of H 2 atmosphere [93,94] (Figure 16). Yuan et al. (2021) [94] have presented the unsuccessful growth of VACNT with pure-thermal and PECVD in the H 2 atmosphere only due to the high activation energy for the nucleation and initial growth of VACNT. The authors show the unmodified presence of Fe nanoparticles at different durations of the annealing process via morphology analysis of the catalyst-decorated DBL. The latter analysis and the CNT growth tests in different atmospheres strongly suggest the surface activation energy as the only barrier toward successful nucleation and initial growth of VACNT on ALD deposited and conducting ZnO films. The activation energy at 60.7 kJ mol −1 on the ALD ZnO film (Figure 17e) was found to be much higher when compared with the same on the ALD Al 2 O 3 DBL, at about 39.1 kJ mol −1 . Thus, the authors concluded the improved chemical reactivity of the carbon source (C 2 H 2 ) in the presence of NH 2 radicals than H radicals for the growth of CNT on Fe catalyst. Additionally, the VACNT growth is improved when synthesis temperature increases from 700 to 800 and is strongly reduced at 900 • C, related to enhanced Ostwald ripening and/or nanoparticles diffusion effects of the Fe catalyst (Figure 17a-d).
As for the use of transition metal nitride DBLs, TiN is a candidate of high interest with a low resistivity (~20 µΩcm) that should support ohmic contacts with grown CNT. Esconjauregui et al. (2014) studied the ALD deposition and use of TiN thin film with-/out a native oxide layer and different thicknesses of the catalyst top-layer for the growth of VACNT via pure-thermal CVD [86]. Additionally, the authors compare the effect of ALD and PVD deposition on the film structure/morphology and its ability to support VACNT growth. The authors show that the growth of VACNT on TiN DBL depends simultaneously on the DBLs initial/post-annealing structure/morphology (as of deposition technique), the presence of surface native oxide (the surface free energy), and the Fe catalyst thickness (studied in the range from 0.5 to 5.0 nm). Differences have been observed in the nanoparticle size and density distributions, showing the dense presence of smaller size Fe nanoparticles on the ALD compared to the PVD deposited DBL (Figure 18g-h). The latter is attributed to the difference in the film crystallinity, i.e., highly crystalline TiN tends to restructure (cruck and surface rearrange) during a CVD process. Thus, a higher TiN crystallinity generates faster deactivation/diffusion of the Fe catalyst nuclei. ALD TiN film with a low-crystalline structure tends to preserve its microstructure even after the annealing process, which is beneficial for supporting dense Fe nanoparticle decoration (Figure 18a). The same was not evidenced with the highly crystalline PVD TiN (with an identical thickness of Fe top-layer as on ALD TiN, Figure 18b,f). PVD TiN needed thicker Fe film (as of 5 nm) to retain a sufficient number of Fe nanoparticles for the growth of VACNT (Figure 18c,e). In addition, the differences in PVD chamber pressure have been shown to influence the film microstructure, with more efficient DBL produced at a higher pressure of 10 −1 mbar (Figure 18c,e). Another crucial parameter in DBL quality is the presence of a thin top layer of native oxide before the evaporation of the Fe catalyst layer (Figure 18c,d). The top-surface presence of TiO 2 is attributed to the air exposure during the transfer process, analyzed/quantified via surface chemistry analysis. Lower-surface-energy TiO 2 promoted the surface interactions while suppressing the bulk diffusion/interactions of the catalyst with the DBL. For densely-packed VACNT, an equilibrium balance should be found between the growth of VACNT and enhanced bulk diffusion/interactions (the native oxide is reduced and/or desorbed), both increasing at temperatures above 500 • C. [86,93,117]. However, not many have presented the use of ALD for the deposition of DBL with improved conductivity properties.
Contrary to the ineffective use of PVD TiO2 as DBL [81], Li et al. (2019) presented the use of ALD for the deposition of metal oxide TiO2 as DBL for the growth of VACNT [93]. Unlike the growth of long and dense VACNT carpets on Al2O3 and SiO2, the TiO2 DBL supported only short and low-density multi-walled CNT carpets with marginal alignment (Figure 15) due to severe sub-surface diffusion of the Fe catalyst nanoparticles or the high surface activation energy. Moreover, not all Fe nanoparticle catalysts achieved a CNT nucleation and growth due to the high surface activation energy on TiO2 DBL compared to SiO2 and Al2O3. The Ostwald ripening effect should have also influenced the density of the VACNT, yet to a lower order. The latter can be correlated with the presence of CNT of similar morphology on the tested DBLs. Multi-walled VACNT have been grown on ALD ZnO films only with PECVD [118] in the presence of NH3 [94], whereas the same was not achieved when pure-thermal and PECVD was performed in the presence of H2 atmosphere [93,94] (Figure 16). Yuan et al. (2021) [94] have presented the unsuccessful growth of VACNT with pure-thermal and PECVD in the H2 atmosphere only due to the high activation energy for the nucleation and initial growth of VACNT. The authors show the unmodified presence of Fe nanoparticles at different durations of the annealing process via morphology analysis of the catalyst-decorated DBL. The latter analysis and the CNT growth tests in different atmospheres strongly suggest the surface activation energy as the only barrier toward successful nucleation and initial growth of VACNT on ALD deposited and conducting ZnO films. The activation energy at 60.7 kJ mol −1 on the ALD ZnO film (Figure 17e) was found to be much higher when compared with the same on the ALD Al2O3 DBL, at about 39.1 kJ mol −1 . Thus, the authors concluded the improved chemical reactivity of the carbon source (C2H2) in the presence of NH2 radicals than H radicals for the growth of CNT on Fe catalyst. Additionally, the VACNT growth is improved when synthesis temperature increases from 700 to 800 and is strongly reduced at 900 °C, related to enhanced Ostwald ripening and/or nanoparticles diffusion effects of the Fe catalyst (Figure 17a-d).     Details of the DBL deposition and VACNT synthesis/properties are summarized in Table 6 for all the above-mentioned studies using the ALD technique. Finally, the use of ALD for the deposition of conducting DBL is an exciting approach to the growth of VACNT. Compared to other techniques, the presented advantages of ALD are the ability to perform conformal deposition of thin films on both planar and non-planar substrates with high precision and beneficial microstructure for the growth of VACNT. The ALD is still considered a highly time-consuming technique, even for the deposition of nanometric scale thin films. However, research is ongoing to develop alternative spatial ALD and its adjustment on continuous (roll to roll) manufacturing lines.

Growth of VACNT on Carbon Substrates, with a Focus on CF Substrates
In this subsection, the review summarizes different approaches reported for the use of un-/modified carbon substrate or the same with predeposited DBL (via CVD or PVD) for the growth of VACNT. Even though few studies have examined the growth of VACNT on a carbonaceous substrate without the use of interface [5,32,48,49], promoted growth of tunable VACNT on CF is today achievable through the use of suitable DBL [4,36,43,44,47,[95][96][97][98][99][100]. The cited and other references therein show that covering a CF substrate with an intermediate DBL can radically improve the growth of VACNT and prevent the possible degradation of the fiber structural properties.
The absence of an up-to-date review of CF-VACNT materials, as for CF-carbon nanotube (CNT) composites [54], inspired us to dedicate a special section of this review to highlight the development of CF composites through the growth of VACNT.

Growth of VACNT on Carbon Substrates with-/-Out Surface Modifications
The ideal solution to obtain a VACNT without contact resistance and well adhered to the surface of a carbon substrate (by establishing carbon-carbon covalent bonds) is to directly deposit the catalyst nanoparticles without additional interlayers [5]. Unfortunately, the carbon substrate must own specific/defined structural properties to support the growth of VACNT. The structural properties of chemically stable carbon substrate are essential to suppress severe carbon-metal reactions (as metal interdiffusion) and suppress Ostwald ripening processes (retain well-dispersed nanocatalyst top layer) [5,32]. To the best of our knowledge, only a few studies examine the growth of the VACNT on commercial CF materials with specified structural properties. Despite today limited growth of the VACNT on uncovered commercial CF, the predeposition of a DBL is still observed as a critical synthesis point to achieve tunable growth of VACNT with remarkable properties.
The low number of studies concerning the direct growth of VACNT on carbon is directly related to the difficulty of controlling the homogeneous dispersion of catalyst nanoparticles on the surface of carbonaceous substrates [32]. The major drawback of the catalytic growth of VACNT on carbon is the interaction between a carbon substrate and catalytic nanoparticles, resulting in the subsurface diffusion and/or poisoning of the latter. On the other hand, the following studies encourage future research on the potential growth of VACNT on CF, the latter with improved graphitic structure.
Cartwright et al. (2014) [32] showed that it is possible to retain a well-dispersed deposit of catalyst nanoparticles on the surface of a carbon substrate of defined sp 2 :sp 3 ratio content (Figure 19 Group A). They also present the growth of VACNT of around 200 µm (Figure 19 Group B) via the base growth mechanism, indicating the strong adhesion of the Fe catalyst on the carbon substrate. Moreover, the authors present the boundaries for the direct growth of VACNT on different carbon substrates by varying the sp 2 :sp 3 hybridization ratio. Thus, the authors performed CVD synthesis of VACNT on amorphous carbon (a-C, 85:15 sp 2 :sp 3 ), tetrahedral amorphous carbon (ta-C, 30:70 sp 2 :sp 3 ), highly oriented pyrolytic graphite (HOPG, 100:0 sp 2 :sp 3 ), and CVD diamond substrate (0:100 sp 2 :sp 3 ) (Figure 19 Group B). The results show the successful formation of catalyst nanoparticles and growth of VACNT on the ratio of 85:15 sp 2 :sp 3 (a-C). After annealing at 750 • C in Ar and at atmospheric pressure, the predeposited Fe nanolayer on the a-C restructures into a highly populated zone (~3.2 × 10 10 ) of nanoparticles, ranging from 10 to 22 nm (details for the substrate, catalyst deposition and CVD growth of VACNT are given in Table 7). On the contrary, the process yields low density, large Fe catalyst nanoparticles on the HOPG substrate, idem for the other substrates with the increased (or complete) sp 3 carbon structure [32]. As for the HOPG, homogeneous dispersion of Fe nanoparticles is not observed only due to its completely flat surface, with only a few edge irregularities of the stacked graphitic planes [5]. Contrarily, the developed surface roughness of the ta-C and CVD diamond substrates indicate the enhanced diffusion of the Fe catalyst during the annealing process, suppressing the growth of dense, thin, long, and aligned CNT. The detailed morphological post-annealing changes are also correlated with the facility of excited Fe catalyst (or other catalysts) to decompose a single σ bond (with energy from 284-368 kJ mol −1 ) in sp 3 carbon when compared to the double σ and π bond in sp 2 carbon (with energy around 615 kJ mol −1 ). Thus, a substrate rich in sp 2 carbon and, to some extent, irregular surface with adequate active sites possess a sufficiently inert surface and welldefined morphology to form a homogenous network of active Fe catalyst nanoparticles and further growth of VACNT. The same authors [32] highlight the importance of the catalyst pretreatment process, demonstrating that a reducing atmosphere (using H 2 or NH 3 ) can cause a catalytic carbon hydrogenation process and diffusion of the Fe nanoparticles beneath substrate surface.
Yoneda et al. (2019) [5] performed a slightly different approach yet again emphasizing the importance of the rich graphitic structure and substrate surface roughness for the direct growth of VACNT on carbon materials (substrate preparation and growth/properties of VACNT can be found in Table 7). Instead of using pure HOPG with a flat surface (examined in [32]), they performed oxidative plasma treatment to produce densely protruded surface (Figure 1 within [5]), graphitic defects, and surface functional groups. The same authors successfully developed VACNT through the deposition of a dense network of Co or Fe catalysts on the modified HOPG substrate (Figure 4 within [5]). Moreover, the modified substrate morphology/chemistry suppresses the catalyst nanoparticle agglomeration and influences the formation of carbon-metal bonds (correlated with the presence of C=O functional groups) [5]. No diffusion of the catalyst nanoparticles was observed since the surface modification of the HOPG substrate did not modify the sp 2 carbon nature of the substrate sub-surface. The final VACNT/CF composite presented specific capacitance observed to be higher than the same of the examined pristine HOPG and therein referenced composites (VACNT on TiN or Al DBL). Compared to referenced VACNT composites, the higher performance was observed due to much thinner CNT (10 nm of CNT thickness compared to 20 nm on TiN DBL), higher CNT density, and enhanced electrochemical active surface area. No information was given about the cycling performance and stability of the examined composite.
bond in sp 2 carbon (with energy around 615 kJ mol −1 ). Thus, a substrate rich in sp 2 ca bon and, to some extent, irregular surface with adequate active sites possess a sufficien ly inert surface and well-defined morphology to form a homogenous network of acti Fe catalyst nanoparticles and further growth of VACNT. The same authors [32] highlig the importance of the catalyst pretreatment process, demonstrating that a reducing mosphere (using H2 or NH3) can cause a catalytic carbon hydrogenation process and d fusion of the Fe nanoparticles beneath substrate surface.  [5] performed a slightly different approach yet again emphas ing the importance of the rich graphitic structure and substrate surface roughness for direct growth of VACNT on carbon materials (substrate preparation a growth/properties of VACNT can be found in Table 7). Instead of using pure HO with a flat surface (examined in [32]), they performed oxidative plasma treatment produce densely protruded surface (Figure 1 within [5]), graphitic defects, and surfa functional groups. The same authors successfully developed VACNT through the dep sition of a dense network of Co or Fe catalysts on the modified HOPG substrate (Figu 4 within [5]). Moreover, the modified substrate morphology/chemistry suppresses catalyst nanoparticle agglomeration and influences the formation of carbon-metal bon (correlated with the presence of C=O functional groups) [5]. No diffusion of the catal  Even though performed via PECVD, i.e., the CVD process assists in the directional growth of CNT due to the attraction of the catalyst nanoparticles under plasma electrostatic forces [118], some research groups have successfully developed VACNT on graphite foil [48] or activated reduced graphene oxide (a-rGO) paper [50]. Ryu et al. (2014) [48] reported growth of VACNT on conducting and flexible commercial graphite foil, to be used as an all-carbon electrode under mechanical stress (conditions of substrate preparation and PECVD conditions for the growth of VACNT are provided in Table 7). The graphite foil surface was not modified before the PVD deposition of the Ni catalyst layer that, after annealing, resulted in the formation of nanoparticles. Different from the extremely flat surface of HOPG, the graphite foil composed of irregularly stacked graphene layers allows the homogeneous deposition of catalyst nanoparticles ( Figure 20). Overall, the study presents a possible way to directly grow VACNT on a carbon substrate with the use of Ni catalyst (Figure 21). When compared to the use of graphitic foil alone, the allcarbon electrode (VACNT grown directly on a flexible graphite foil) demonstrated a higher electrochemical active surface area and higher electrocatalytic activity. In addition, the same authors performed flexibility tests to show that the examined all-carbon electrodes were bendable and non-breakable. The achieved strong and direct covalent bond of the CNT with the graphite substrate is defined to support the structural integrity during mechanical bending.
Nanomaterials 2022, 12, x FOR PEER REVIEW 31 of 54 graphite foil surface was not modified before the PVD deposition of the Ni catalyst layer that, after annealing, resulted in the formation of nanoparticles. Different from the extremely flat surface of HOPG, the graphite foil composed of irregularly stacked graphene layers allows the homogeneous deposition of catalyst nanoparticles ( Figure 20). Overall, the study presents a possible way to directly grow VACNT on a carbon substrate with the use of Ni catalyst ( Figure 21). When compared to the use of graphitic foil alone, the all-carbon electrode (VACNT grown directly on a flexible graphite foil) demonstrated a higher electrochemical active surface area and higher electrocatalytic activity. In addition, the same authors performed flexibility tests to show that the examined all-carbon electrodes were bendable and non-breakable. The achieved strong and direct covalent bond of the CNT with the graphite substrate is defined to support the structural integrity during mechanical bending.   [49] have presented high-performance advanced EMI shielding and thermal management multifunctional multilayer composite based on the VACNT-graphene paper (GP)/polydimethylsilane (PDMS). The authors detail the application performance of the cited composite due to its mechanical and conductivity properties and multi-layer structure (details given in Table 7). Whereas little is given for the CVD growth process of VACNT onto spruce GP and the structural characteristics of the GP substrate or the achieved VACNT (summarized in Table 7). GP is generally synthesized from a graphene oxide film, derived via the Hummer method, and additional film preparation processes [119]. The final result is a GP with a dense graphene flakes struc- graphite foil surface was not modified before the PVD deposition of the Ni catalyst layer that, after annealing, resulted in the formation of nanoparticles. Different from the extremely flat surface of HOPG, the graphite foil composed of irregularly stacked graphene layers allows the homogeneous deposition of catalyst nanoparticles ( Figure 20). Overall, the study presents a possible way to directly grow VACNT on a carbon substrate with the use of Ni catalyst ( Figure 21). When compared to the use of graphitic foil alone, the all-carbon electrode (VACNT grown directly on a flexible graphite foil) demonstrated a higher electrochemical active surface area and higher electrocatalytic activity. In addition, the same authors performed flexibility tests to show that the examined all-carbon electrodes were bendable and non-breakable. The achieved strong and direct covalent bond of the CNT with the graphite substrate is defined to support the structural integrity during mechanical bending.   [49] have presented high-performance advanced EMI shielding and thermal management multifunctional multilayer composite based on the VACNT-graphene paper (GP)/polydimethylsilane (PDMS). The authors detail the application performance of the cited composite due to its mechanical and conductivity properties and multi-layer structure (details given in Table 7). Whereas little is given for the CVD growth process of VACNT onto spruce GP and the structural characteristics of the GP substrate or the achieved VACNT (summarized in Table 7). GP is generally synthesized from a graphene oxide film, derived via the Hummer method, and additional film  [49] have presented high-performance advanced EMI shielding and thermal management multifunctional multilayer composite based on the VACNT-graphene paper (GP)/polydimethylsilane (PDMS). The authors detail the applica-tion performance of the cited composite due to its mechanical and conductivity properties and multi-layer structure (details given in Table 7). Whereas little is given for the CVD growth process of VACNT onto spruce GP and the structural characteristics of the GP substrate or the achieved VACNT (summarized in Table 7). GP is generally synthesized from a graphene oxide film, derived via the Hummer method, and additional film preparation processes [119]. The final result is a GP with a dense graphene flakes structure, closely stacked along the cross-section direction, presenting an essential wrinkle-like surface morphology. A well-prepared GP always presents a low defect level and good crystalline structure with crystallite size in the range of µm-scale and stacking order of graphene flakes similar to that of HOPG. As therein presented [49], the structural properties of the used GP have been sufficient to support the deposition of a dense network of Fe catalyst nanoparticles and the simultaneous growth of VACNT. Figure 22 presents SEM images of 20 µm height VACNT grown on two sides of a single layer of GP. Additionally, the TEM image (Figure 22c) details the solid interfacial adhesion of the bamboo-like CNT and GP sheets following an ultrasonic treatment. Once more, this study and the ones presented in this subsection clearly show the importance of an sp 2 -rich carbon substrate in combination with the irregular surface morphology, along with other methods and material parameters, to achieve VACNT without the use of a DBL. good crystalline structure with crystallite size in the range of µm-scale and stacking order of graphene flakes similar to that of HOPG. As therein presented [49], the structural properties of the used GP have been sufficient to support the deposition of a dense network of Fe catalyst nanoparticles and the simultaneous growth of VACNT. Figure 22 presents SEM images of 20 µm height VACNT grown on two sides of a single layer of GP. Additionally, the TEM image (Figure 22c) details the solid interfacial adhesion of the bamboo-like CNT and GP sheets following an ultrasonic treatment. Once more, this study and the ones presented in this subsection clearly show the importance of an sp 2rich carbon substrate in combination with the irregular surface morphology, along with other methods and material parameters, to achieve VACNT without the use of a DBL. Finally, several studies showed the growth of radially aligned CNT or VACNT on desized or/and surface-modified CF via CVD [120][121][122][123][124][125]. Li et al. (2015) [122] performed adsorption of polymeric functional coating Poly(styrene-alt-[dipotassium maleate]) (K-PSMA) on sized CF to introduce surface potassium ions available for Fe catalyst (0.05 M Fe(NO3)3٠9H2O in isopropanol) adhesion and avoid the interaction/diffusion of the later with-/-in the CF structure. This concept of 'direct' CNT synthesis on CF resulted only in homogenous growth of low density short radially aligned CNT on the CF circumference, the later aligned in dense unidirectional filament yarn (Figure 23). Contrary to previous studies (cited therein), the authors performed longitudinal tensile and interface shear strength testing to show that the mechanical properties of the used CF was retained and further improved with the synthesis of radially aligned CNT. Another point influencing the suppressed thermal, but also, catalytic degradation of the CF structure is the perfor- Finally, several studies showed the growth of radially aligned CNT or VACNT on desized or/and surface-modified CF via CVD [120][121][122][123][124][125]. Li et al. (2015) [122] performed adsorption of polymeric functional coating Poly(styrene-alt-[dipotassium maleate]) (K-PSMA) on sized CF to introduce surface potassium ions available for Fe catalyst (0.05 M Fe(NO 3 ) 3 ·9H 2 O in isopropanol) adhesion and avoid the interaction/diffusion of the later with-/-in the CF structure. This concept of 'direct' CNT synthesis on CF resulted only in homogenous growth of low density short radially aligned CNT on the CF circumference, the later aligned in dense unidirectional filament yarn ( Figure 23). Contrary to previous studies (cited therein), the authors performed longitudinal tensile and interface shear strength testing to show that the mechanical properties of the used CF was retained and further improved with the synthesis of radially aligned CNT. Another point influencing the suppressed thermal, but also, catalytic degradation of the CF structure is the performance of the CVD growth of CNT at temperature as low as 480 • C. Contrary to the production of radially aligned CNT (Figure 23) [122], Rahmanian et al. (2013) [120] and Zhang et al. (2017) [123] performed growth of VACNT ribbons, evidenced in double or multiple rows, on the CF circumference (as shown in Figures 24c and 25b, respectively). To directly synthesize VACNT via CVD [120] or PECVD [123], both groups performed chemical activation pre-treatment to enhance the anchoring of catalyst nanoparticles on CF. The CFs were oxidized through immersion in 65% HNO 3 [120] or 2M HNO 3 [123] and subsequently impregnated with dense Fe catalyst coating (0.0001M Fe(NO 3 ) 3 ·9H 2 O in acetone [120] or Fe(NO 3 ) 3 in ethanol [123]). At an elevated temperature of 250 • C, a calcination process followed the coating process. Moreover, the mild oxidation process created carbon-oxygen functional groups (carbonyl and carboxylic groups), previously observed to enhance the formation of carbon-metal bonds [5], but also introduced improved surface roughness by etching-out present amorphous carbon zones (Figure 24a,b). The developed surface roughness further influenced the homogeneous deposition of catalyst nanoparticles for the growth of VACNT (Figure 24c). Thus, performed TEM analyses show the presence of multi-wall CNT (MWCNT), with an outer diameter of 20 nm and an inner diameter of 5 nm (Figures 24d and 25d). Additionally, the developed specific surface area is considered advantageous for using the final composites in electro-chemical applications. Details of CF modification, impregnation of catalyst and CVD growth of VACNT are given in Table 8. Noteworthy, Russello et al. (2018) [125] present direct growth of VACNT on spread CF tow tapes via pure-thermal CVD (details of catalyst deposition and CVD growth of VACNT are given in Table 8). Again, 0.2 M Fe(NO 3 ) 3 ·9H 2 O in isopropanol was adsorbed and dried on the surface of the CF prior to the CVD process, yet without the pre-adsorption of any polymeric functional coating and/the CVD process was performed at elevated temperature of 700 • C. The study presents the growth of short, VACNT (2-10 µm) with high density that resulted in the formation of VACNT ribbons ( Figure 26). Moreover, the CNT growth process was stopped before the catalyst inactivation, thus, prolonged CVD process might result in VACNT with superior height, but also, influence structural degradation of the CF substrate. The 'direct' growth of 1-2 µm tall VACNT on CF yielded to composites with through-thickness electric conductivity of 0.11-0.22 S cm −1 , respectively [125].  [123] performed growth of VACNT ribbons, evidenced in double or multiple rows, on the CF circumference (as shown in Figure 24c and Figure 25b, respectively). To directly synthesize VACNT via CVD [120] or PECVD [123], both groups performed chemical activation pre-treatment to enhance the anchoring of catalyst nanoparticles on CF. The CFs were oxidized through immersion in 65% HNO3 [120] or 2M HNO3 [123] and subsequently impregnated with dense Fe catalyst coating (0.0001M Fe(NO3)3٠9H2O in acetone [120] or Fe(NO3)3 in ethanol [123]). At an elevated temperature of 250 °C, a calcination process followed the coating process. Moreover, the mild oxidation process created carbon-oxygen functional groups (carbonyl and carboxylic groups), previously observed to enhance the formation of carbon-metal bonds [5], but also introduced improved surface roughness by etching-out present amorphous carbon zones (Figure 24a,b). The developed surface roughness further influenced the homogeneous deposition of catalyst nanoparticles for the growth of VACNT (Figure 24c). Thus, performed TEM analyses show the presence of multi-wall CNT (MWCNT), with an outer diameter of 20 nm and an inner diameter of 5 nm (Figures 24d and 25d). Additionally, the developed specific surface area is considered advantageous for using the final composites in electro-chemical applications. Details of CF modification, impregnation of catalyst and CVD growth of VACNT are given in Table 8. Noteworthy, Russello et al. (2018) [125] present direct growth of VACNT on spread CF tow tapes via pure-thermal CVD (details of catalyst deposition and CVD growth of VACNT are given in Table 8). Again, 0.2 M Fe(NO3)3٠9H2O in isopropanol was adsorbed and dried on the surface of the CF prior to the CVD process, yet without the pre-adsorption of any polymeric functional coating and/the CVD process was performed at elevated temperature of 700 °C. The study presents the growth of short, VACNT (2-10 µm) with high density that resulted in the formation of VACNT ribbons ( Figure 26). Moreover, the CNT growth process was stopped before the catalyst inactivation, thus, prolonged CVD process might result in VACNT with superior height, but also, influence structural degradation of the CF substrate. The 'direct' growth of 1-2 µm tall VACNT on CF yielded to composites with through-thickness electric conductivity of 0.11-0.22 S cm −1 , respectively [125].        Though achievable, the growth of VACNT with the suggested methods is highly restrictive and cannot be considered viable for tuneable growth of VACNT on CF due to: 1. The properties of some of the examined carbonaceous materials (ex., highly graphitized materials with a precise ratio of 85:15 sp 2 :sp 3 hybridization [32], oxidative plasma-treated HOPG [48], or GP [49]) do not reflect the structural properties of commercially used carbon substrates (as CF, mostly ex-PAN [126][127][128]) available for large-scale industrial use and applications. 2. The composite properties stability, which depends on the controllable growth of VACNT (CNT length/diameter/density), is challenging to achieve when direct growth is performed on CF substrates. 3. Presented pre-treatments, such as mild oxidation/activation, of the CF before the growth of VACNT should influence the CF's structural and physical properties.
Accordingly, this subsection is only a critical review of the 'direct' growth of VACNT on carbon substrates without the presence of intermediate DBL. Yet, the derived information does not justify the direct use of evidenced methods for the controlled growth of highly dense and long VACNT on CF. Nonetheless, the recovered information can be seen as an excellent bibliographic backup for advanced direct VACNT growth on carbon substrates within extended time framings. Within the following subsection, different methods are thus deeply explored for the growth of VACNT on CF with suitable conducting DBL and the development of new composites. Though achievable, the growth of VACNT with the suggested methods is highly restrictive and cannot be considered viable for tuneable growth of VACNT on CF due to: 1.
The properties of some of the examined carbonaceous materials (ex., highly graphitized materials with a precise ratio of 85:15 sp 2 :sp 3 hybridization [32], oxidative plasma-treated HOPG [48], or GP [49]) do not reflect the structural properties of commercially used carbon substrates (as CF, mostly ex-PAN [126][127][128]) available for large-scale industrial use and applications.

2.
The composite properties stability, which depends on the controllable growth of VACNT (CNT length/diameter/density), is challenging to achieve when direct growth is performed on CF substrates.

3.
Presented pre-treatments, such as mild oxidation/activation, of the CF before the growth of VACNT should influence the CF's structural and physical properties.
Accordingly, this subsection is only a critical review of the 'direct' growth of VACNT on carbon substrates without the presence of intermediate DBL. Yet, the derived information does not justify the direct use of evidenced methods for the controlled growth of highly dense and long VACNT on CF. Nonetheless, the recovered information can be seen as an excellent bibliographic backup for advanced direct VACNT growth on carbon substrates within extended time framings. Within the following subsection, different methods are thus deeply explored for the growth of VACNT on CF with suitable conducting DBL and the development of new composites.

Via Chemical Vapor Deposition (CVD)
In the last decade, radially aligned CNT has been grown on pyrolytic carbon (PyC) DBL deposited on CF materials by CVD [45,129,130]. The predeposition of PyC protects the CF surface properties by suppressing the interactions with the catalyst and sub-layer diffusion of the latter. The PyC is a promising DBL due to its compatibility with commercial CF, derived from PAN or pitch. Moreover, this kind of DBL resulted in the advanced growth of straight CNT with higher purity and length when compared to the direct growth of CNT over CF ( Figure 27). The examined CVD protocols and composite properties are summarized in Table 9. The radially aligned CNT on PyC DBL positively affected the final composites' oxidation resistance and mechanical properties. The mechanical properties improvements result from the effective transfer of the mechanical stress from the matrix to the CFs via the intermediate presence of CNT and the PyC [45,130]. Compared to neat C/C composites, calefactive oxidation tests show superior oxidation resistance of C/C composites with grown aligned CNT [129]. However, the low density and high diameter CNT obtained through this approach still demand further research before reaching an advanced technological readiness level compared to VACNT obtained on metal oxide, insulating, DBL [36,43,44,[95][96][97][98][99]. Moreover, the high-temperature CVD deposition of the PyC (above 1000 • C) is a disadvantage for the use of the presented DBL that should influence the final cost and commercialization of potential CF-VACNT composites.
properties are summarized in Table 9. The radially aligned CNT on PyC DBL positively affected the final composites' oxidation resistance and mechanical properties. The mechanical properties improvements result from the effective transfer of the mechanical stress from the matrix to the CFs via the intermediate presence of CNT and the PyC [45,130]. Compared to neat C/C composites, calefactive oxidation tests show superior oxidation resistance of C/C composites with grown aligned CNT [129]. However, the low density and high diameter CNT obtained through this approach still demand further research before reaching an advanced technological readiness level compared to VACNT obtained on metal oxide, insulating, DBL [36,43,44,[95][96][97][98][99]. Moreover, the hightemperature CVD deposition of the PyC (above 1000 °C) is a disadvantage for the use of the presented DBL that should influence the final cost and commercialization of potential CF-VACNT composites.    2 Ω) [4] Hahm et al. (2012) [92] have also investigated the use of CVD for both synthesis of a graphitic substrate and the growth of VACNT. An anodized aluminium oxide template with nanocups morphology was also used to critically enlarge the specific surface area of the graphitic substrate, i.e., the final composite. Following the CVD of the graphitic substrate, PVD of a 1.5 nm Fe catalyst layer was performed before a second CVD process to grow VACNT (Details given in Table 9). Resultant, dense carpets of single- (Figure 28c), double-or multi-walled CNT with 5-10 µm length were grown on the graphitic substrate to produce a 3-D hybrid composite, tested as a supercapacitor (Figure 28(a, a1-3)). The thickness of the synthesized graphitic substrate (the graphitic nanocups) was not mentioned, and the use of Raman spectra presented a high degree of structural disorder, based on the strong intensity of the D band (1364 cm −1 ) (Figure 28b). Contradictory to the latter, Cartwright et al. (2014) [32] confirmed the use of substrate rich in sp 3 carbon as inadequate for forming a homogenous network of active Fe catalyst nanoparticles and further growth of VACNT. Though not stated, the anodized aluminium oxide template (present during the PVD of Fe catalyst and the CVD processes) might have affected the observed Fe catalyst surface stabilization and growth of VACNT. Finally, the VACNT−carbon nanocups-based supercapacitor exhibited 50% higher specific capacitance (0.6 mF/cm 2 ) than the reference carbon nanocups one. The increase in the specific capacitance was attributed to the larger surface area and hybrid 3-D morphology of the VACNT−carbon nanocups-based supercapacitor. Cyclic stability tests showed stable capacitance even up to 10,000 charge/discharge cycles. More important, the electrochemical impedance spectra measurements presented a low equivalent series resistance of 23 Ω between the VACNT−carbon nanocups electrode and gold current collector.  [92] have also investigated the use of CVD for both synthesis of a graphitic substrate and the growth of VACNT. An anodized aluminium oxide template with nanocups morphology was also used to critically enlarge the specific surface area of the graphitic substrate, i.e., the final composite. Following the CVD of the graphitic substrate, PVD of a 1.5 nm Fe catalyst layer was performed before a second CVD process to grow VACNT (Details given in Table 9). Resultant, dense carpets of single- (Figure  28c), double-or multi-walled CNT with 5-10 µm length were grown on the graphitic substrate to produce a 3-D hybrid composite, tested as a supercapacitor (Figure 28(a, a1-3)). The thickness of the synthesized graphitic substrate (the graphitic nanocups) was not mentioned, and the use of Raman spectra presented a high degree of structural disorder, based on the strong intensity of the D band (1364 cm −1 ) (Figure 28b). Contradictory to the latter, Cartwright et al. (2014) [32] confirmed the use of substrate rich in sp 3 carbon as inadequate for forming a homogenous network of active Fe catalyst nanoparticles and further growth of VACNT. Though not stated, the anodized aluminium oxide template (present during the PVD of Fe catalyst and the CVD processes) might have affected the observed Fe catalyst surface stabilization and growth of VACNT. Finally, the VACNT−carbon nanocups-based supercapacitor exhibited 50% higher specific capacitance (0.6 mF/cm 2 ) than the reference carbon nanocups one. The increase in the specific capacitance was attributed to the larger surface area and hybrid 3-D morphology of the VACNT−carbon nanocups-based supercapacitor. Cyclic stability tests showed stable capacitance even up to 10,000 charge/discharge cycles. More important, the electrochemical impedance spectra measurements presented a low equivalent series resistance of 23 Ω between the VACNT−carbon nanocups electrode and gold current collector.  The CVD deposition of hybrid metal oxide/nitride, graphitic, and other promising conducting DBL has yet to be investigated for the growth of VACNT on carbonaceous substrates. To the best of our knowledge, no other studies were identified to use CVD to achieve deposition of conducting DBL and the growth of VACNT on carbon materials.

Via Physical Vapor Deposition (PVD)
PVD has been used to deposition DBL further used in at least two methods for the CVD growth of VACNTs on CF.
Though not a real conducting interface between a CF and VACNT, yet, playing the role of DBL for the preservation of surface-active Fe catalyst nanoparticles, researchers have demonstrated the use of appropriate capping top layer via the 'Odako' method [57,[131][132][133][134][135][136]. The 'Odako' VACNT growth of VACNT is performed with a distinctive CVD technique for ultra-fast detachment and fragmentation of the extremely thin capping layer with incorporated catalyst nanoparticles. This method is also discussed to allow a direct CNT covalent interface with a carbon substrate [131] (Figure 29). Usually, an alumina capping layer (of a few nanometers) is preferred when a Fe catalyst is used to grow VACNT. The Fe catalyst is immobilized on the alumina layer due to interfacial reactions yet kept active and uncovered to develop the formation and growth of CNT [115]. The CVD deposition of hybrid metal oxide/nitride, graphitic, and other promising conducting DBL has yet to be investigated for the growth of VACNT on carbonaceous substrates. To the best of our knowledge, no other studies were identified to use CVD to achieve deposition of conducting DBL and the growth of VACNT on carbon materials.

Via Physical Vapor Deposition (PVD)
PVD has been used to deposition DBL further used in at least two methods for the CVD growth of VACNTs on CF.
Though not a real conducting interface between a CF and VACNT, yet, playing the role of DBL for the preservation of surface-active Fe catalyst nanoparticles, researchers have demonstrated the use of appropriate capping top layer via the 'Odako' method [57,[131][132][133][134][135][136]. The 'Odako' VACNT growth of VACNT is performed with a distinctive CVD technique for ultra-fast detachment and fragmentation of the extremely thin capping layer with incorporated catalyst nanoparticles. This method is also discussed to allow a direct CNT covalent interface with a carbon substrate [131] (Figure 29). Usually, an alumina capping layer (of a few nanometers) is preferred when a Fe catalyst is used to grow VACNT. The Fe catalyst is immobilized on the alumina layer due to interfacial reactions yet kept active and uncovered to develop the formation and growth of CNT [115]. The presented 'Odako' method is advantageous compared to others, except for the direct growth of VACNT on carbon substrate, since in both ways, the VACNT are anchored on a carbon substrate via covalent carbon-carbon bonds. The latter is vital for optimal adhesion and electrical contact between the carbon substrate and the VACNT. Using suitable electrolyte and electrochemical test conditions, 'Odako' achieved VACNT/carbon (graphite, graphene) composites have been examined as excellent capacitors [57,132,133,136]. Although highly innovative and beneficial, there are a few criteria for the performance of this method, such as: • Use of special engineering approaches to perform precise deposition of a few nanometer layers of alumina with specific physical density and oxidation state; • To grow VACNT on the catalyst lower layer, a particular joule heating or similar CVD system should be used to stress crack the alumina top layer into small fragments under the sharp increase in temperature at the rate of 300 °C min −1 . Thus, deviations in the capping layer thickness and the CVD heating rate can disturb the activation of the Fe catalyst and the VACNT growth; • The performance of the 'Odako' method demands an additional plasma etching process to remove the catalyst-alumina layer standing on the top of the VACNT.
All the above-numbered are some points of the 'Odako' method that restrains and limits the immediate use of this technique on an industrial scale. In addition, one should further examine the commercial viability and the cost of this production method of car- The presented 'Odako' method is advantageous compared to others, except for the direct growth of VACNT on carbon substrate, since in both ways, the VACNT are anchored on a carbon substrate via covalent carbon-carbon bonds. The latter is vital for optimal adhesion and electrical contact between the carbon substrate and the VACNT. Using suitable electrolyte and electrochemical test conditions, 'Odako' achieved VACNT/carbon (graphite, graphene) composites have been examined as excellent capacitors [57,132,133,136]. Although highly innovative and beneficial, there are a few criteria for the performance of this method, such as:

•
Use of special engineering approaches to perform precise deposition of a few nanometer layers of alumina with specific physical density and oxidation state; • To grow VACNT on the catalyst lower layer, a particular joule heating or similar CVD system should be used to stress crack the alumina top layer into small fragments under the sharp increase in temperature at the rate of 300 • C min −1 . Thus, deviations in the capping layer thickness and the CVD heating rate can disturb the activation of the Fe catalyst and the VACNT growth; • The performance of the 'Odako' method demands an additional plasma etching process to remove the catalyst-alumina layer standing on the top of the VACNT.
All the above-numbered are some points of the 'Odako' method that restrains and limits the immediate use of this technique on an industrial scale. In addition, one should further examine the commercial viability and the cost of this production method of carbon substrate-VACNT hybrids. The recovered information should be further compared with the composite production using conducting DBL (between the CF and the VACNT) with an advanced technological readiness level. Inline, the 'Odako' method is herein referenced as future long-term solutions to be developed for commercial growth of VACNT on CF/carbon substrates.
Contrary to the mentioned 'Odako' method, PVD of conducting DBL is widely investigated for further growth of VACNT via conventional pure thermal CVD, seen as an easy-to-industrialize process with advanced technological readiness level. PVD predeposition of 20 nm Ni (as both DBL and catalyst) has been shown to support fuzzy CNT on CF via thermal CVD at temperatures as low as 550 • C [22]. The results present Ni as a DBL/catalyst for the development of a fuzzy CNT network that boosted the mechanical stiffness and further improved the electrical conductivity of the final composites. However, VACNT on CF have not been reported with the examined PVD deposited Ni DBL/catalyst and CVD process. Another pioneering study [137] presents the growth of short VACNT on CF with PVD pre-deposition of Ti DBL and again Ni catalyst top-layer. Moreover, the study endorses the formerly observed failure of lone Ni DBL/catalyst to produce VACNT without the presence of 5 nm Ti under-layer due to the extensive reaction between the CF surface and the catalyst. Unfortunately, the effect of the Ti DBL cannot be clearly justified due to the applied electric field during the performed plasma-enhanced CVD that influences the directional growth of VACNT (as seen in [48,50]). Fang et al. (2017) have also investigated the use of 70 nm Ti DBL to support 12 nm Ni catalyst and the subsequent growth of VACNT via pure-thermal CVD at a high temperature of 850 • C [80]. As a result, only fuzzy CNT, of 100 nm mean diameter and 2-3 µm length, were observed on the surface-modified CFs even after a CVD duration of 30 min. The latter confirms the inability of a Ti DBL to support the formation of a few nanometer metal nanoparticles and retain an active Ni catalyst to achieve VACNT. Finally, the growth of VACNT on CF cannot be effectively performed only with Ni DBL/catalyst due to the metal diffusion within the CF structure. In addition, the use of Ti DBL/Ni catalyst solution was examined effective only via the plasma-enhanced CVD method. Thus, the latter was not achieved (or should be reevaluated) via pure-thermal CVD.
While no CNT were observed on bear CFs, bundles of aligned multi-walled carbon nanotubes (MWCNT) were synthesized by floating catalyst CVD on CF paper with predeposited 10 nm Al DBL [46]. Once again, the exposed presence of native alumina was presented as a crucial factor for stabilizing the Fe catalyst nanoparticles and the improved growth of high-density VACNT. Yet, the thickness of 10 nm of Al DBL was not observed as sufficient to form a continuous and dense network of Fe nanoparticles for self-sustaining VACNT bundles that collapsed with the increase in the CNT length (observed at around 20 µm). Another recent study shows the beneficial effect of monometallic Al DBL on the growth of VACNT on CF substrates. Pozegic et al. (2016) [4] presented CVD growth of VACNT with improved density, length, and alignment on CFs than previously reported study [46]. The VACNT properties are improved mainly due to 35 nm Al DBL that minimized the interaction between the substrate and the catalyst. Previously, Pozegic et al. (2014) [138] confirmed the incompetence of PVD deposited Fe catalyst alone to grow VACNT via the same CVD process, producing only scattered fuzzy CNT on the CF surface. Figure 30 presents the scheme of used photo-thermal CVD, morphological details of the CF prior/post VACNT growth, and the variation of the CF cloth material passing from one preparation phase to the next one. Interesting to note is the PDV deposition of Al DBL without desizing the CF surface polymer layer that has not been previously reported and did not influence the effective growth of VACNT. Once more, the growth zone of VACNT was imposed from the PVD of the Al DBL, only present on the outer CF cloth surface. Detailed information regarding the synthesis conditions and the CNT properties can be seen in Table 9. As previously mentioned, Al DBL formed a native oxide top-layer during the air transfer and catalyst annealing process. The latter confirmed by numerous studies to improve the formation and anchoring of Fe catalyst nanoparticles for promoted growth of VACNT [60]. The formation of this top-layer oxide can further increase the electrical resistance of an Al DBL layer by severe order of magnitude [34], thus influencing its initial role as conducting DBL. However, Pozegic et al. (2016) [4] reported improved overall composite conductivity (electrical and thermal) by a few hundred percent due to the low (but sufficient for the purpose) thickness of the DBL layer, which retains its conducting purpose. In addition, the beneficial use of photo-thermal CVD (PTCVD) is shown as an easily scalable/controlled optical top-heating method for the growth of VACNT. Thus, the use of PTCVD in combination with appropriate conducting DBL, with good optical reflectance of infra-red/ultra-violet radiation, is presented to boost the thermal decomposition of the carbon precursor on the irradiated catalyst nanoparticles and simultaneously protect the thermal degradation of the used substrate. The latter confirmed by numerous studies to improve the formation and anchoring of Fe catalyst nanoparticles for promoted growth of VACNT [60]. The formation of this toplayer oxide can further increase the electrical resistance of an Al DBL layer by severe order of magnitude [34], thus influencing its initial role as conducting DBL. However, Pozegic et al. (2016) [4] reported improved overall composite conductivity (electrical and thermal) by a few hundred percent due to the low (but sufficient for the purpose) thickness of the DBL layer, which retains its conducting purpose. In addition, the beneficial use of photo-thermal CVD (PTCVD) is shown as an easily scalable/controlled optical top-heating method for the growth of VACNT. Thus, the use of PTCVD in combination with appropriate conducting DBL, with good optical reflectance of infra-red/ultra-violet radiation, is presented to boost the thermal decomposition of the carbon precursor on the irradiated catalyst nanoparticles and simultaneously protect the thermal degradation of the used substrate. Furthermore, a more recent study by Fontana et al. (2020) [47] describes an effective way to grow dense VACNT on CF gas diffusion layers (GDL) via hot-filament CVD at temperatures as low as 500 °C ( Figure 31). The key to the successful growth of VACNT was the use of PVD deposited diffusion barrier bi-layer of Ti/Al metal alloy with a Fe catalyst top-layer. Primarily, the Ti under-layer was stated (yet not tested in the study) to promote the DBL adhesion with the CF surface. Secondly, Ti alloy with Al (forming titanium aluminide-TiAl) was stipulated to suppress the strong surface oxidation of the Al layer that preserved its electrical conductivity properties. Moreover, the authors validate the importance of the Al upper layer for the successful annealing of the Fe catalyst layer, i.e., the formation of a dense network of active Fe catalyst nanoparticles for the growth of VACNT. In addition, the morphology of the produced VACNT was solely correlated with the properties of the PVD deposited DBL/catalyst and the CVD process, Furthermore, a more recent study by Fontana et al. (2020) [47] describes an effective way to grow dense VACNT on CF gas diffusion layers (GDL) via hot-filament CVD at temperatures as low as 500 • C ( Figure 31). The key to the successful growth of VACNT was the use of PVD deposited diffusion barrier bi-layer of Ti/Al metal alloy with a Fe catalyst top-layer. Primarily, the Ti under-layer was stated (yet not tested in the study) to promote the DBL adhesion with the CF surface. Secondly, Ti alloy with Al (forming titanium aluminide-TiAl) was stipulated to suppress the strong surface oxidation of the Al layer that preserved its electrical conductivity properties. Moreover, the authors validate the importance of the Al upper layer for the successful annealing of the Fe catalyst layer, i.e., the formation of a dense network of active Fe catalyst nanoparticles for the growth of VACNT. In addition, the morphology of the produced VACNT was solely correlated with the properties of the PVD deposited DBL/catalyst and the CVD process, excluding any influence from the CF surface morphology and/or chemistry. Due to the highly directive PVD deposition of the DBL/catalyst coating, the growth of the VACNT was also oriented and confirmed on the PVD exposed surface of the GDL (Figure 32). In contrast, the same was not observed on the opposite side and the substrate interior. Details of the PVD deposition and the CVD protocol for the growth of the VACNT can be found in Table 9. Finally, the authors presented VACNT modified GDL as a cathode to improve the overall fuel cell performances up to 1.7 A/cm 2 (current density improvement up to 30% at 0.75 V or 50% efficiency for a fuel cell) in comparison with up-to-date best state-of-the-art GDL. Nanomaterials 2022, 12, x FOR PEER REVIEW 45 of 54 excluding any influence from the CF surface morphology and/or chemistry. Due to the highly directive PVD deposition of the DBL/catalyst coating, the growth of the VACNT was also oriented and confirmed on the PVD exposed surface of the GDL (Figure 32). In contrast, the same was not observed on the opposite side and the substrate interior. Details of the PVD deposition and the CVD protocol for the growth of the VACNT can be found in Table 9. Finally, the authors presented VACNT modified GDL as a cathode to improve the overall fuel cell performances up to 1.7 A/cm² (current density improvement up to 30% at 0.75V or 50% efficiency for a fuel cell) in comparison with up-to-date best state-of-the-art GDL.   excluding any influence from the CF surface morphology and/or chemistry. Due to the highly directive PVD deposition of the DBL/catalyst coating, the growth of the VACNT was also oriented and confirmed on the PVD exposed surface of the GDL (Figure 32). In contrast, the same was not observed on the opposite side and the substrate interior. Details of the PVD deposition and the CVD protocol for the growth of the VACNT can be found in Table 9. Finally, the authors presented VACNT modified GDL as a cathode to improve the overall fuel cell performances up to 1.7 A/cm² (current density improvement up to 30% at 0.75V or 50% efficiency for a fuel cell) in comparison with up-to-date best state-of-the-art GDL.

Conclusions
Future developments require rapid technological advancements in various material applications. The use/growth of VACNT on different substrates through the predeposition of conducting DBL is a promising approach, with an intense scientific exploration towards the improvement of existing and the development of new composites. As a recent branch of carbon composites, CF-VACNT exhibits excellent potential to go beyond existing limits. Three-dimensional flexible CF-VACNT composites are being extensively explored, owing to their unique morphological and physico-chemical properties. Recent progress on the introduction of DBL with improved conducting properties should meet the requirements needed for advanced construction composites, state-of-the-art wearable sensors or power devices, and many other applications.
In this review, we summarized and discussed various deposition/synthesis methods and techniques used to deposit conducting DBL and grow VACNT on flat or 3-D substrates of different natures, with a focus on fibrous carbons. The parameters/conditions for each of these methods and techniques and the nature/properties of the deposited DBL/grown VACNTs have been shown to affect the final composite and its performances compared to rivals or reference materials. The post-deposition evolution of a conducting DBL, CNT catalyst diffusion/deactivation, or the process condition restrictions have been generally determined as common bottleneck points towards robust fabrication technology.
Overall, the conducting DBL plays a significant role in both the growth of tunable VACNT and the final physico-chemical properties of the composite. Thus, the conducting coupling of a concerned substrate and VACNT with appropriate DBL is a challenging topic under examination, aiming at the cost-effective mass-production of novel high-performant materials.

Funding:
The review was conducted during BK post-doctoral fellowship, as part of the NawaShell2 project, supported by BPI-France (Concours Mondial de l'Innovation) and coordinated by NawaTechnologies.