Highly efficient doping of carbon nanotube films with chloroauric acid by dip-coating

Single-walled carbon nanotube (SWCNT) based transparent and conductive films (TCFs) are one of the most prospective materials for novel flexible and stretchable electronic devices. Development of reproducible and scalable doping procedure is the key step towards the widespread implementation of SWCNT TCFs. Here, we thoroughly investigate a dip-coating technique for SWCNT doping as a promising approach for the practical manufacturing of SWCNT films with high performance. We examine the effect of dip-coating parameters on optical and electrical properties of the films using HAuCl 4 solution in isopropyl alcohol (IPA) and in situ investigate doping effects. This method appeared to easily fine-tune the optoelectronic parameters of SWCNT films and achieve a record sheet resistance value of 36 Ohm/sq. at the 90% transmittance in the middle of visible spectral range by increasing a work function value from 4.8 (for pristine SWCNTs) to 6.0 eV. The proposed approach allows efficient, uniform, and reproducible fabrication of highly conductive and transparent SWCNT films and opens an avenue for precise tailoring of SWCNT Fermi level for optoelectronic devices.

Thin films of single-walled carbon nanotubes (SWCNTs) [13,14,15,16] are promising materials for the ITO replacement and development of novel electronics beyond the existing one based on brittle ceramics. SWCNT films possess remarkable electrical, [17] optical [18] and mechanical properties [19]. Typically, SWCNT thin films can be prepared either from liquid or gas phases. The former approach firstly requires obtaining stable dispersions of nanotubes in a surfactant solution, [20,21,22] and then SWCNT film deposition either by vacuum filtration, [23] printing, [24] spin-, [20] spray-, [25] rod-, [26] dip-coating, [27] or Langmuir-Blodgett technique [28]. However, the dispersing process deteriorates the resulting properties of SWCNT films, since ultrasonication introduces defects and shortens the nanotubes, while the presence of surfactants increases the contact resistance between the tubes [29,30]. In contrast, an aerosol (floating-catalyst) chemical vapor deposition (CVD) method [31,32] allows dry film fabrication without the intermediate detrimental stages, keeping intrinsic SWCNT properties unaltered that makes the films superior for electronic component fabrication.
During the synthesis, both semiconducting and metallic SWCNTs are simultaneously produced [13]. Therefore, the formed Schottky barriers at the tube-to-tube interfaces significantly degrade the conductivity of SWCNT films [33]. To enhance the electrical performance of the SWCNT films, usually, adsorption doping is applied [14]. As a result, the Fermi level in the SWCNTs is shifted either in the valence or conduction band, which leads to a decrease in the energy barrier between nanotubes and increase in the charge carrier concentration [34]. Among various dopants (HSO 3 Cl, [ [42]) applied for the SWCNT doping, HAuCl 4 is the most promising to achieve the lowest sheet resistance [43] with relatively long-term stability [44].
SWCNT films are usually doped by drop-casting [45], spin- [41] or spray-coating [46], in which a dopant is adsorbed on the surface of the nanotubes. Drop-casting is the simplest method; however, it lacks uniformity, scalability, and reproducibility. The spray-coating can be easily scaled-up, but the reproducibility of the dopant layer is a large challenge. The spin-coating method provides precise control over the coating quality and thickness of a dopant layer, however, dopant losses during the processing make this method economically unfavorable.
A dip-coating is an approach for a thin layer deposition onto a substrate surface. Implementation of the dip-coating technique to SWCNT technology has been extensively studied: several research groups utilized it for fabrication of thin CNT films, [13,27,35,47,48,49] CNTs/ AgNWs hybrids, [50] or CNT/polymer composites [51] from CNT dispersion. By this technique, SWCNT films with the sheet resistance of 100 Ohm/sq. were deposited from SWCNT/chlorosulfonic acid dispersion [35]. Besides, the dip-coating technique was utilized to functionalize SWCNTs for electrochemical applications [52]. Even though dipcoating was proven to be an effective technique for a thin layer deposition, it has not been yet examined for the doping of SWCNTs.
Here, we have performed the dip-coating technique for uniform and reproducible doping of SWCNT films. We demonstrate that by adjusting the process parameters such as withdrawal speed and dopant concentration, SWCNT films can be efficiently doped. Fine parameter adjustment allowed us to achieve SWCNT films with the record low sheet resistance of 36 Ohm/sq. at the 90% transmittance in the middle of the visible spectral range which corresponds to the Fermi level shift of 1.2 eV. Our study boosts SWCNT-based transparent and conductive film technology and opens an avenue for optoelectronic applications, requiring high conductivity and precise control of the work function.

SWCNT film preparation and doping procedure
High-quality SWCNTs with mean diameter of 2.1 nm were synthesized by an aerosol (floating catalyst) chemical vapor deposition method and collected on filter for subsequent dry transfer onto quartz substrates as described elsewhere [53]. For our studies, we used SWCNT films with the transmittance of 80-85% (at a wavelength of 550 nm).

Sample characterization
Optical properties of SWCNT films were investigated using Perki-nElmer LAMBDA 1050 UV-Vis-NIR spectrophotometer in a wide spectral range from 175 to 2750 nm. The quartz substrate with the doped SWCNT films was wiped off from the backside by isopropyl alcohol before the measurements to remove the chloroauric acid.
Thermo Scientific DXRxi Raman imaging microscope and diodepumped solid-state laser operating at 532 nm was used to study the quality and doping degree of the SWCNTs.
Ultraviolet photoelectron spectroscopy (UPS) measurements were carried out by using Riken Keiki AC-3 photoelectron spectrometer operating in air. The spectra were measured in the energy range of 4.2-6.5 eV with the resolution of 0.05 eV. The spectra were measured 3 times at 5 different points and averaged (Fig. S1a). The work function of pristine SWCNTs is 4.8 eV.
The sheet resistance of the samples was measured by a linear fourprobe method using Jandel RM3000 Test Unit that served as a current source and a digital multimeter. The resistance was measured 3 times at 5 different points and averaged (Fig. S1a). For a comparison of the optoelectrical properties of the SWCNT films, we utilized an equivalent sheet resistance at T = 90% (at 550 nm) calculated according to the following equation: [54] where A and R s are the absorbance (at 550 nm) and sheet resistance of the films. Investigations of the film morphology were carried out by means of FEI Tecnai G2 F20 transmission electron microscope (TEM) at an acceleration voltage of 100 kV. Samples were prepared by immersing gold TEM grids with deposited SWCNT films into HAuCl 4 solutions by dipcoating technique, followed by the subsequent solvent evaporation in air.
During the doping process, we in situ monitored the change in the SWCNT film resistance. For that purpose, we sputtered four 100 nm thick gold contacts with a width of 2.0 mm onto a quartz glass with the surface size of 2.0 × 3.0 cm 2 and thickness of 3 mm, as shown in Fig. S2. Subsequently, four copper wires were attached to the gold contacts using silver epoxy. Then a SWCNT film with a size of 0.8 × 2 cm 2 was transferred onto the fabricated substrate so that it touches all the four contacts. Finally, the substrate with a deposited SWCNT film was partially dipped into a dopant solution, while keeping the silver epoxy contacts outside. The current was applied to the outer two contacts, while the potential drop was measured between the middle contacts spaced by 8 mm. The initial and final sheet resistance of the SWCNT films were verified by a linear four-probe measurements using Jandel RM3000 tool.

Results and discussion
UV-vis-NIR spectroscopy was used to analyze the electronic structure of SWCNT films before and after chemical doping HAuCl 4 . (Fig. 1,  Fig. S3). Doping of carbon nanotubes consists of charge transfer between HAuCl 4 and the nanotubes, where the difference in the work function causes electron depletion in the vicinity of Van Hove singularities in semiconducting (S 11 , S 22 ) and metallic (M 11 ) SWCNTs and concomitant reduction of Au 3+ to Au o [38]. The stepwise change of the dopant concentration from 0.01 to 60 mM results in a gradual change in the SWCNT electronic structure, that is observed by disappearance of absorption peaks corresponding to the optical transition between Van Hove singularities in semiconducting (S 11 , S 22 ) and metallic (M 11 ) SWCNTs. At a concentration of 3.75 mM and above, the low energy transitions S 11 , S 22 and M 11 fully disappear, while a new peak attributed to intersubband plasmon [55] arises at 1435 nm (Fig. 1).
Meanwhile, due to the doping by 3.75 mM HAuCl 4 , the 3-fold decrease in the sheet resistance (Fig. 2a) and 12 cm −1 blue-shift in the Raman G-mode (Fig. 2b) are observed as a result of complete disappearing of the low energy transitions S 11 , S 22 , M 11 . The steady increase in the work function up to 1 eV (Fig. 2c) fits well the evolution of optical absorption peaks associated with the transitions between different levels of Van Hove singularity in SWCNTs (Fig. 3). The consequent growth of the dopant concentration leads to an increase in the Raman shift (Fig. 2b) with a maximum value of 26 cm −1 for a 60 mM solution and a work function of 6.07 eV (Fig. 2c). Thus, an increase of a p-type dopant's concentration leads to a shift in the Fermi level downward and the higher work function, accordingly [34]. Although the 60 mM concentration results in the highest shift of the Raman G-mode and work function, the equivalent sheet resistance is not beneficial when compared to a 30 mM solution. Most likely, this can be explained by formation of a thick dopant layer at the highest HAuCl 4 concentration which lowers the film transmittance in the visible range (Fig. 1). Taking into account that the adsorption doping is associated with a trade-off between conductivity and transmittance, we identified 30 mM HAuCl 4 solution in IPA as an optimal concentration for the highest optoelectronic performance (9-fold decrease in the equivalent sheet resistance and Raman shift of 25 cm −1 ), which was utilized in our further experiments.
We carried out a control experiment to determine the role of the IPA (Fig. 4a,d) and observed effects in resistance change similar to previously reported by Shu Li et al. [56] In the pure solvent, a resistance initially drops due to the bundling process. After 1-minute immersion, an infiltration of IPA molecules results in a gradual increase of the SWCNT film resistance. Over the next 3 h, the equilibrium in the system is practically established. Subsequent withdrawal of the film from IPA leads to a sharp drop in the resistance to the value lower than that of the pristine film as a result of the film densification by surface tension of the solvent during its evaporation. Besides, due to the densification process by IPA, insignificant changes in UV-spectra and Raman G-mode blueshift are observed (Fig. S4).
In situ measurements of the SWCNT film resistance during the dipcoating revealed slight resistance decrease immediately after the withdrawal of SWCNTs from 1 mM HAuCl 4 solution (Fig. 4b) and a 2-fold resistance drop in the case of 30 mM HAuCl 4 (Fig. 4c). Further solvent evaporation results in resistance decrease due to the film densification. The final trend of the film resistance change strongly depends on the dopant concentration. It can give either a slight rise at a low concentration (1 mM, Fig. 4b) or a significant drop of the film resistance at high concentrations (30 mM , Fig. 4c).
TEM observation of the films after the doping confirms lower doping capacity of the 1 mM HAuCl 4 solution (Fig. 4b,e) with formation of small-size Au 0 clusters when compared to the 30 mM HAuCl 4 (Fig. 4c,f).
Previously, it was shown that withdrawal speed is a key factor determining the thickness of the deposited film during a dip-coating process [57]. In fact, there are 3 deposition regimes, so-called: capillary, intermediate, and viscous drag regimes [57]. The capillary regime corresponds to a very low withdrawal speeds (lower than 6 mm/min), resulting in thick film, and it is generally used with non-volatile solvents (e.g. water). In contrast, viscous drag regime (higher than 60 mm/min) is used while working with volatile solvents (e.g. ethanol) and, as a result, intermediate thickness is formed [57]. The intermediate regime associated with both intermixed regimes is implied for ultra-thin films. Dipcoating of SWCNT films in HAuCl 4 solutions was performed with withdrawal speed range of 50-400 mm/min corresponding to both intermediate and viscous drag regimes. To examine the withdrawal speed influence on the doping degree, the immersion time was fixed at one minute.
Our results demonstrate that increasing the withdrawal speed from 50 to 400 mm/min gives additional raise to a doping degree, as can be seen from the intersubband plasmon peak (optical feature observed in highly-doped nanotubes) [55] blue-shift (Fig. 5a) and a gradual increase in Raman G-mode shift from 21 cm −1 to 24 cm −1 (Fig. 5b). At 300 mm/ min, the lowest equivalent sheet resistance value of 36 Ohm/sq. is achieved as a result of high doping efficiency. At the maximal withdrawal speed of 400 mm/min, the equivalent sheet resistance increases due to the formation of the thick dopant layer on the nanotubes surface, lowering SWCNT film transmittance.
Long-term resistance measurements of the dip-coated SWCNT films (Fig. 6a,b) show the following trend: the film resistance decreases within 1 min by 60% immediately after the immersion into the dopant solution and then remains constant. After the withdrawal of the film from the solution, the solvent evaporation, accompanied by film densification and additional two-fold resistance decrease takes place. Complete solvent evaporation at ambient conditions occurs during ~ 7 h and subsequently leads to a rise in the film resistance because of the dopant decomposition. Over the following week, the equivalent sheet resistance almost doubles and then practically does not change up to 90 days of monitoring (Fig. 6b).
The optical spectra of the doped SWCNT films highlight the evolution of the film resistance (Fig. 7). Right after the doping, two distinct peaks at 227 and 316 nm from [AuCl 4 ] − appear [58]. Over time, these two peaks disappear and plasmon peak arising from gold nanoparticles broadens and red-shifts (Fig. 7b). This indicates gold nanoparticle growth as a result of HAuCl 4 decomposition and, as a consequence, gold particle aggregation on the surface of the doped SWCNT film that is    supported by SEM and TEM data ( Figure S5).
The evolution of optical absorption after doping is confirmed by position and intensity of intersubband plasmon peak [55] (Fig. 7c).
There are two plasmon peaks in SWCNTs. The π-plasmon is a result of collective excitations of the π-band electrons excited by light with parallel polarization to nanotube axis, which is a well-known feature of most graphitic materials. While intersubband plasmon is optical feature observed only in highly-doped nanotube films (usually after the -S 22 transition disappears) in response to the perpendicularly polarized light. It corresponds to electron transition within single conduction (ntype doping) or valence (p-type) band. The calculation of intersubband plasmon dispersion [55] at different Fermi energies (doping levels) reveals that the plasmon energy blue-shifts with Fermi energy rise. The maximum intersubband plasmon peak intensity is observed on the first day after doping, which may indicate a high doping level; the subsequent red-shifts after one week indicates a decrease in the doping level of SWCNT film, which correlates well with the film resistance trend (Fig. 6b). Since the tetrachloroauric (III) acid has a major contribution to the films' conductivity, "freezing" of the system by creating of a protective layer for the period up to 8 h (Fig. 6a) can preserve the lowest value of the films' resistance.
The uniformity of samples after doping was confirmed by measurements of the sheet resistance, transmittance, and G-peak position in Raman spectra and evaluation of work function in 5 different points (Fig. S1). It was demonstrated that the dip-coating technique allows to obtain fairly uniform distribution of the measured parameters.   The development of the dip-coating doping technique solves the challenge of large-scale transparent and conductive film fabrication. Indeed, the most popular and straightforward doping approaches, such as drop casting, spin-coating, and aerosol coating, primarily lack scalability that makes them incompatible with effective and high-speed rollto-roll technology. At the same time, the dip-coating procedure could be easily embedded in the continuous roll-to-roll process by immersing roller modules into the reservoir with a dopant.

Conclusions
The novel approach for efficient, uniform and reproducible doping of SWCNT films by dip-coating technique was proposed. The electrical and optical properties of the SWCNT films can be controlled by dip-coating's parameters such as the concentration of the dopant solution and the withdrawal speed. Moreover, the influence of the HAuCl 4 concentration in the range from 0.01 to 60 mM on the equivalent sheet resistance of SWCNT films was investigated. Dip-coating into 30 mM HAuCl 4 solution and 300 mm/min withdrawal speed leads to uniformly doped SWCNT' films with the state-of-the-art equivalent sheet resistance value of 36 Ohm/sq. and high work function value of 6 eV, which corresponds to a Fermi level shift of 1.2 eV. Since the method we utilized is scalable, uniform, reproducible, and compatible with flexible/stretchable substrates, we believe it can become a key technology for fabrication of highly conductive and transparent flexible and stretchable films for future electronic applications.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.