Molecular electron doping to single-walled carbon nanotubes and molybdenum disulfide monolayers

Carrier doping is an essential way to inject holes and electrons to electronic materials, which modulates their transport properties. While the substitution of heteroatoms essentially modulates the band structure of most semiconducting materials, chemical (molecular) doping can achieve relatively reliable carrier concentration modulation, particularly for nanocarbons and two-dimensional semiconductors. Compared to p-type counterparts, the stabilization of n-type carbon materials has been a challenge not only for basic science but also for various electronic device applications. This Mini-Review describes rational concepts for, and the results of, a stable n-type doping technique mainly for carbon nanotubes using molecular reactions and interactions. The stable n-type carbon nanotubes with controlled carrier concentration are implemented in complementary circuits and thermoelectric energy harvesters. The molecular and supramolecular n-type doping is not limited for carbon nanotubes, but is utilized in the fabrication of conducting transition metal dichalcogenides such as a molybdenum disulphide (MoS2) monolayer.


Introduction
The carrier doping to nanomaterials including single-walled carbon nanotubes (SWCNTs) has been an active research field in the last 30 years. Particularly due to their large surface area, molecular adsorption has a strong influence on the electronic properties of nanocarbon materials. In the atmospheric environment, the adsorption of oxygen contributes to p-type conduction in nanocarbons such as SWCNTs and graphene [1]. The SWCNT films often exhibit p-type transport properties, and then switch to n-type when treated at high temperatures and under reduced pressure. For many applications, a methodology to fix the carrier polarity is required.
The carrier doping of is enabled by various injection ways inclduing heteroatom insertion, and chemical/ electrochemcal oxidation/reduction. Nanocarbons as well as organic materials and two-dimensional transition metal dichalcogenides (TMDCs) are known to be doped by intermolecular charge transfer, called as molecular doping. The adsorption of carrier donors, such as metals and chemical molecules, is the first step in the molecular doping process. Intercalation of alkali metals between graphite interlayers is a well-known phenomenon within this concept. It is well known that the introduction of impurity atoms dopes materials, but the controlled introduction of impurity atoms into some robust materials, such as SWCNTs, has yet to be established. In this context, it is anticipated that molecular doping will be more adaptable to the injection of carriers into sturdy low-dimensional materials, ranging from nanocarbons to transition metal dichalcogenide nanosheets.
Within the molecular doping regime, we would like to emphasize that the efficiency of carrier injection relies on the conduction band (otherwise, lowest unoccupied molecular orbital, LUMO level) while the stability of doped states depends on charge neutrality. Additionally, molecular doping is dependent on the surface area of Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. electronic materials, where large specific surface area is preferential. In this context, low dimensional materials such as SWCNTs and 2D materials potentially show significant affinity with molecular doping.
The doping of SWCNTs has long been studied for electronic and energy applications since their discovery (figure 1). Especially, controlling the threshold voltage of both p-and n-type transistors, comprising complementary integrated circuits, is required to develop low-voltage and low-power flexible electronics [2]. In order to achieve this, tuning the electronic type and the carrier neutral point is necessary, which is enabled by precise electronic doping. Blocking oppossite carriers is also important for suppressing off-current [3].
The thermoelectric properties of SWCNTs have been actively investigated from the viewpoint of energy harvesting as well as condensed matter physics, and the transport mechanisms of single-wire and network thin films have already been reported [12]. In the networks, the temperature gradient is localized at the nanotube-nanotube contact, and the Seebeck effect at the contact is dependent on the electronic structure of the CNT [13]. Depending on the chiral angle of the graphene sheet, CNTs can be classified as semiconducting or metallic, and semiconducting CNT-enriched films show the huge Seebeck coefficient of up to about 2 mV K −1 [8]. In general, peak power outputs are optimized by tuning the carrier concentration; in this context, continuous tuning from p-type to n-type by electrochemical doping of CNTs has been examined [14].
One more requirement is the module design for efficient thermoelectric power generation, where p-type and n-type components are inplemented in series. Since the dawn of molecular electronics including carbon and organic electronics, however, n-type doping has been a significant challenge, due to air-instability along with difficulty in charge injection [15]. From the viewpoint of static chemical bonding theory, the positively and negatively charged carriers on CNTs can be interpreted as carbocations (positively charged carbons) and carboanions (negatively charged carbons) delocalized in graphene sheets. This is based on the exchange of charge through sp 2 bonds, called π-conjugation. From a static point of view (figure 2), upon oxidation, carbocations are generated so that the structural arrangement of backbones does not change. On the other hand, carboanions are highly reactive to electron-withdrawing (oxidizing) agents such as oxygen, carbon dioxide, and water in atmosphere, since the reactive conformation could form upon electron injection (chemical reduction). The issue to be solved is the stabilization of such negatively charged carbons.
Considering above, the development of design principles for rational chemical doping is highly helpful for future electronic application based on SWCNTs and related materials. This Mini-Review introduces recent approaches to the n-type doping of SWCNTs and related nanomaterials such as transition metal dichalcogenides (TMDCs), particularly for field effect transistors and thermelectric power generation applications. We first show that various dopants were proposed to n-dope SWCNTs and TMDCs, where guiding principals for both doping are quite similar. Then, we would like to mention that recent tremendous efforts have addressed the fine tuning of doped states and improved their stability for practical applications.

Chemical doping to form n-type SWCNTs
A plenty of investigations have been devoted for realizing the n-type molecular doping of SWCNTs. Representative charge transfer type compounds which successfuly generate n-type SWCNTs are shown in figure 3.

Electron transfer-type dopants
When the Lewis base compounds, especially with high highest-occupied-molecular-orbital (HOMO) levels, are adsorbed onto SWCNTs, partial electron transfer, and charge separation could occur. Recent representative n-dopants are listed in table 1. The addition of strong reducing agents (e.g. alkali metals such as potassium (K)) enables the n-type doping of SWCNTs [16], which is analogous to the charge transfer-mediated intercalation into graphite, so called graphite intercalation compounds (GICs) [23]. Due to air-instability, unfortunately, this dopant is now substituted to organic base and reducing agents. Chronologically, polyethylene imine was used to efficiently convert p-type SWCNTs to their n-type form. This dopant was first examined for the construction of field effect transistors with individual SWCNTs [17]. This polymeric dopant has widely been applied for various  investigations for the last two decades, and has been recognized as a standard method. Successful n-type dopants, furthermore, include reduced benzyl viologen, and metallocene with high HOMO levels. Although these materials are highly reactive in air, they offer excellent electron doping ability due to their high reduction potentials (shallow HOMO levels). Various moderate and tuneable reducing agents based on amines and phosphines were proposed to produce air-stable n-type SWCNTs [20][21][22], promoting many subsequent studies on composite materials. The encapsulation of cobaltocene was reported to reliably promote n-type doping [24]. 1, 2, 4, 5-Tetrakis(tetramethylguanidino)benzene (ttmgb) showed two electron transfer to SWCNT transistors, leading to the extraction of pure n-type transport, rather than ambipolar [3]. Even if the dopants' HOMO levels are moderate, dopant encapsulation might improve n-type doping efficiency [25,26]. Furthermore, weaker electron donors were used to achieve n-type characteristics from SWCNT films, including neutral surfactants (e.g. Pluronic PEG-type polymers, and polyvinylpyrrolidone (PVP)) [27], and simple electron-donating polymers (e.g. poly(vinyl alcohol) and poly(vinyl acetate)) [28].

Hydrogen, hydride, and radical transfer-type dopants
The chemical reduction resulting in n-type doping is not limited in the electron transfer from high HOMO compounds (table 2). Hydrogen and/or hydride (H -) transfer reagents are mostly successful for the reliable n-type doping of SWCNTs. This reduction motif can be used widely in the electron transport chain in biological systems (e.g. nicotinamide adenine dinucleotide NAD + /NADH redox couple for eukaryote). This reaction system was promptly useful for the n-type doping of SWCNTs [29]. Furthermore, the derivatives of dimethyldihydro-benzimidazole (DMBI) were found to efficiently produce air-stable SWCNTs and also organic semiconductors/conductors. One of representative forms is 4-(2, 3-Dihydro-1, 3-dimethyl-1H-benzimidazol-2-yl)-N,N-dimethylbenzenamine, known as N-DMBI. The chemical reaction mechanism with N-DMBI is still under investigation [30][31][32], while its doping ability can be widely used.
Organometallic dimer dopants were developed to efficiently dope nanocarbons as well as organic semiconductors, similar to DMBI derivatives. They also launch hydrogen and/or hydride. Similar to these dopants, various hydride reagents including triarylmethane derivatives are expected to work as efficient n-type dopants for SWCNTs. Recently, highly reactive species such as organic radicals have been used for efficient n-type doping. Fujigaya and colleagues utilized in situ generated pyridine (py)-boryl radicals to n-dope SWCNTs [38].

A guiding principles for stabilization
When a charge is acquired by doping, a substance with the opposite polarity is required for charge compensation. In the case of n-type SWCNTs, for example, an external positive ion (cation) is introduced to the negative charge on the SWCNT ( figure 4). When alkali metals such as sodium and potassium are added to neutral CNTs, as noted before, electron transfer occurs quickly and the CNTs exhibit n-type transport. In this process, the metal ions that lose electrons are thought to form an ionic pair with the negatively charged CNTs.
On the other hand, this n-type material is known to be easily oxidized by atmospheric species (oxygen, water, etc). This instability is attributed partly to the compatibility between the charges. In general, small ions are known to form ion pairs through Coulomb interactions (e.g. NaCl). Relatively large cations such as cesium ion and tetramethylammonium form stable complexes with relatively large anions such as iodine ion. In this process, the large ion pairs are stabilized by Coulomb interaction as well as dipole-dipole interaction, that is, the shape symmetry effect. This phenomenon is known as the hard-soft acid-base (HSAB) rule which is an empirical  rule for ionic interactions in solution [39]. This concept is widely accepted in supramolecular science, and is considered to be applied to the chemical doping of SWCNTs. Investigating the introduction of various cations, Nonoguchi, Kawai, and colleagues found that extremely delocalized supramolecular cations such as crown ether-metal ion complexes ( figure 5(b)) stabilized n-type SWCNTs [40].
Representatively, SWCNTs were soaked in ethanol solution containing sodium borohydride (NaBH 4 ) and 15-crown-5-ether, which are typical reducing salts, prior to thorough drying. The carrier polarity can be determined from the sign of the Seebeck coefficient. The raw SWCNT films gave a positive value indicates that they are p-type, and the crown ether and NaBH 4 alone are slightly doped. The addition of crown ether or NaBH 4 alone led to slight doping, but the materials gradually returned to the initial state under atmospheric conditions. On the other hand, the CNT film treated with both reagents showed extremely stable n-type characteristics under air ( figure 5(d)). This result strongly supports the working hypothesis described above.

Anion-induced electron transfer
Cationic metal ions such as sodium ions and potasium ions form stable complexes with crown ethers. At the same time, the anions are detached from the cations and destabilized by the charge shielding of the crown ether shells around cations, where these free anions are called the naked anions. The naked anions are reported to be extremely electron-donating [41]. Crown ethers can selectively take up specific metal ion species depending on the size of their internal space [42]. The combinations of metal salts and crown ethers with different ring sizes and various metal ions were examined for this application in the electron doping [40]. The n-type doping of SWCNTs were examined with ethanol solutions of 0.1 mol L −1 crown ether complex and onium complex are shown in figures 5(e) and (f). First, the negative Seebeck coefficient was not observed when sodium hydroxide or crown ether were applied to SWCNT films alone, but n-type materials were obtained very reproducibly when both were coexposed. Similarly, even halides and alkoxides salts, which are generally inert, were found to induce n-type doping in the presence of crown ether. Some organic onium salts based on tetramethylammonium and tetrabutylammonium with hydroxides were also found to produce n-type SWCNTs, although their stability remained a problem. The similar effect was reported for the electron doping of fullerenes for organic solar cells [43]. Such anion-induced electron transfer may be caused by the activation of anions according to the HSAB rule in addition to charge shielding. The Seebeck coefficient of CNTs with 15-crown-5 and various sodium salts showed that the efficiency of n-type doping varies with the size of the anion species ( figure 5(f)). In particular, the efficiency of n-type doping is relatively high for small anions such as hydroxide ions (133 pm) and chloride ions (17.2 pm), which are small anions, showed relatively high n-type doping efficiency. In other words, large anions such as triflate (307 pm) can form relatively stable ion-pairs (complexes) with metal ions encapsulated in the crown, while small anions cannot form stable ion pairs, and are expected to induce anion-induced electron transfer.
Thermal stability was also examined for thermoelectrics application based on SWCNT sheets. When potassium hydroxide and benzo-18-crown were used as dopants, no significant degradation of any thermoelectric properties was observed for more than 700h (about one month), at 100°C. A systematic dopant survey revealed supramolecular dopants with an aromatic structure show significant thermal stability, compared to small crown ether complexes.

Monitoring doping progress
Tuning the carrier concentration by doping is essential for optimizing thermoelectric conversion as well as tuning the threshold voltage in field effect transistors/thin film transistors (TFTs). Tracking doping progress is important, and various ways for this have been proposed including electron spin resonance, x-ray photoelectron spectroscopy. Ultraviolet photoelectron yield spectroscopy was used to examine the Fermi level of doped Table 3. Representative molecular doping to TMDCs.

Dopants
Type of doping References reduced benzyl viologen two-step reduction [44] cesium carbonate (Cs 2 CO 3 ) a strong base widely used also in organic electronics [45] triphenyl phosphine stable and reversible [46] polyethylene imine moderate electron donor [47] crown ether complexes anion induced electron transfer high air-stability [48] SWCNTs, which correslates well to the transport properties [22]. Absorption spectroscopy of infrared interband and far-red plasmon resonance mode offers prompt and quantitative doping information [6].
More practically, doping progress can be monitored by, and is applied in, transfer curves in TFTs. Transfer curves, which indicates the switching performance, showed tuneability for both p-type and n-type doping ( figure 6) [2]. The threshold voltages could shift gradually to positive and negative sides upon p-type and n-type doping, respectively. Based on this tuneability, Ohno and colleagues developed complementary integrated circuits achieving operation at a low supply voltage of 0.5 V with fairly high gain and noise margins.

Molecularly and supramolecularly doped n-type transition-metal dichalcogenide (TMDC) monolayers
Most molecular dopants developed for nanocarbons and organic electronics are readily useful for the doping of TMDC thin layers and monolayers (table 3). Representatively, molybdenum disulfide (MoS 2 ), known as a n-type semiconductor, was doped by organic reductants such as reduced benzyl viologen [44], leading to a substantial increase in current density in its field effect transistors. For the similar purpose, various chemical n-dopants were proposed for tuning the electronic properties of MoS 2 , including polyethylene imine [47], and triphenyl phosphine [46]. Electron doping with ammonia (NH 3 ) was applied for the development of a MoS 2 -based NH 3 detector, achieving the high sensitivity of ca. 300 ppb, and the quantitativity at room temperature [49].
In addition to electronic device applications, the doping is also important for elucidating optical response. Trion is a charged exciton (a electron-hole pair), and has attracted significant attention particularly for semiconductor quantum dots and SWCNTs. The tuneable chemical n-doping by cesium carbonate (Cs 2 CO 3 ) revealed that, at room temperature, MoS 2 monolayers were found to show trion photoluminescence (PL) at a region slightly low-shifted from the exciton photoluminescence [45]. Since heavy doping would spoil a photoliminescence quantum yield, the controllability of chemical doping is important for this puropose. Inversely, controled p-type doping can dramatically enhance PL derived from the exciton [50].
We would like to show a representative approach to the stable doping of MoS 2 monolayer and its full characterization. Miyata and colleagues have recently reported that MoS 2 monolayer doped with the compexes of potassium hydroxide and crown ether shows air-stable, metallic conduction [48]. They found a dramatic increase in current density in MoS 2 field effect transistors upon the supramolecular doping. Treatments with increased dopants led to a substantial increase in the off-current and in metallic gate dependence. Eventually, the on-current reached ∼100 μA for the sample treated with the 100 mM dopant solution, which represents an increase of approximately two orders of magnitude compared with the on-current of the untreated sample. Notably, the doped MoS 2 FETs were highly stable in ambient air. The transfer curves remained nearly unchanged even after 24 days of air exposure ( figure 7(e)).
The n-doped states of TMDCs can be tracked quantitatively by the measurements of optical properties that reflect electronic structures. Figure 7(f) shows the Raman spectra of monolayer MoS 2 , recorded before and after the doping treatments with different concentrations of the dopants (0.1-100 mM). The two characteristic Raman peaks denoted by E′ and A′ 1 are attributed to the in-plane and out-of-plane vibration modes of monolayer MoS 2 , respectively [51]. The E′ mode (383 cm −1 ) varied depending very little on the dopant concentration, while the A′ 1 mode was downshifted by 6 cm −1 . This downshift is consistent with the earlier literature of electron doped MoS 2 with reduced benzyl viologen [44], suggesting that an increase in the electron concentration enhanced the electron-phonon interaction.
The photoluminescence (PL) spectroscopy revealed doping-enabled spectral change. PL measurements showed the emission peak from A exciton at 1.80 eV for the undoped MoS 2 (figure 7(g)). This PL peak was substantially suppressed by the n-doping treatments and was completely quenched at above a 1 mM crown-KOH concentration with a downshift of the peak to 1.72 eV. These PL shift and quenching can be explained by an increase of the emission from negatively charged trions as a result of the stable electron doping and the suppression of neutral exciton formation [52]. These optical responses are consistent with the transport measurement results for electron-doped monolayer MoS 2 .

Outlook
Improving the air-stability of n-type SWCNTs and related materials has significantly advanced the understanding of physical properties as well as the applications in electronic devices and power generators. Behind such electronics applications, rational chemistry is required for tuning the electronic properties of nanocarbons and 2D semiconductors. We would like to emphasize that most chemical doping described here can be easily reproduced by anyone with commercially available dopants, and SWCNTs. The present doping procedure can also be applied to other nanocarbons (e.g. graphene, fullerene) and inorganic materials with large surface area [53]. Along with n-type doping, counterparts for p-type doping are also important for the development of reliable electronic devices. Such chemical doping has recently been utilized for elucidating the emerging optical properties of TMDCs [54,55]. We hope that, in near future, the versatile molecular doping technology further promotes a variety of developments of emerging nano-molecular devices and their derived physical properties.