Heteroatom Codoped Graphene: The Importance of Nitrogen

Abstract

Although graphene has exceptional properties, they are not enough to solve the extensive list of pressing world problems. The substitutional doping of graphene using heteroatoms is one of the preferred methods to adjust the physicochemical properties of graphene. Much effort has been made to dope graphene using a single dopant. However, in recent years, substantial efforts have been made to dope graphene using two or more dopants. This review summarizes all the hard work done to synthesize, characterize, and develop new technologies using codoped, tridoped, and quaternary doped graphene. First, I discuss a simple question that has a complicated answer: When can an atom be considered a dopant? Then, I briefly discuss the single atom doped graphene as a starting point for this review’s primary objective: codoped or dual-doped graphene. I extend the discussion to include tridoped and quaternary doped graphene. I review most of the systems that have been synthesized or studied theoretically and the areas in which they have been used to develop new technologies. Finally, I discuss the challenges and prospects that will shape the future of this fascinating field. It will be shown that most of the graphene systems that have been reported involve the use of nitrogen, and much effort is needed to develop codoped graphene systems that do not rely on the stabilizing effects of nitrogen. I expect that this review will contribute to introducing more researchers to this fascinating field and enlarge the list of codoped graphene systems that have been synthesized.

1. Introduction

Humanity is facing a point of no return because if most of the planet’s population does not change their lifestyle, we will unchain the sixth extinction event on Earth. Chemistry alone cannot solve all the problems; it cannot stop wars or end racism or religious conflicts. However, chemistry can help humanity solve many pressing difficulties that urgently need to be circumvented, such as energy production, conversion, and storage, water purification, development of CO₂ capture systems, drug design, and food production, to mention just a few. To that end, discovering new materials with tailored properties has become an obsession. Graphene^1,2 is one of the rising stars in science. Nonetheless, despite its fantastic natural properties, the pristine form of graphene has limited use. Up to now there is not a flagship product derived from graphene as there is the transistor derived from silicon.

Several methods have been designed to modify graphene: it can be chemically functionalized,^3,4 external stress or electrical fields can be applied, and some of the carbon atoms of the graphene framework can be replaced by heteroatoms, a process that introduces an impurity in the system.⁵⁻¹² The latter will be the focus of this review. Although significant advances have been made in the preparation of single atom doped graphene, and the properties induced by elemental doping are fascinating,⁵⁻¹² a new research direction has been opened: the introduction of more than one dopant type in the graphene framework. This transformation is known as the codoping of graphene or sometimes as the dual doping of graphene when it exclusively refers to introducing two foreign atoms. Experimental research and theoretical research have demonstrated that the introduction of at least two dopants can dramatically alter the physicochemical properties of graphene. As an example, I mention the adsorption of lithium on mono- and codoped graphene with 2p and 3p elements.¹³ As we can appreciate in Table 1, the adsorption energies of Li on N and S doped graphene are −0.72 and −1.25 eV, respectively. Interestingly, when the N and S dopants are introduced simultaneously, the adsorption energy of Li becomes −2.99 eV, about 1 eV larger than the sum of the adsorption energies of Li on N and S doped graphene.¹³ This is not an isolated example of the dramatic effects that codoping can induce, as I will describe in this review.

Table 1. Interaction Energies, Magnetic Moments, Li–X Distances and Charges Determined for XY Dual Doped Graphene Interacting with a Single Lithium Atom, at the VDW-DF/DZP Level of Theory.

		IE (eV)
XY	position of Li	XY orthoa	XY parab	dLi–X (Å) XY ortho	Q (μB) XY ortho
AlB	below Al	–2.11	–2.52	2.75	1.0
AlN	below Al	–2.16	–2.03	2.66	1.0
AlO	below Al	–2.27	–2.23	2.83	0.0
SiB	below Si	–2.41	–2.59	2.62	0.62
SiN	below Si	–2.18	–2.14	2.70	0.0
SiO	below Si	–1.97	–1.90	2.77	0.0
PB	below P	–1.89	–1.45	2.67	0.06
PN	below P	–1.44	–1.41	2.65	0.19
PO	below P	–2.30	–1.34	2.73	0.0
SB	below S	–1.79	–1.87	2.81	0.06
SN	below S and bonded to N	–2.99	–1.35	2.77	0.0
SO	below S	–1.84	–1.38	2.74	0.0
B	below B-hex	–2.41		2.33	0.0
N	below N-hex	–0.72		2.34	0.0
graphene	over hex	–1.11		2.32	0.0

^aOrtho disposition of dopants.

^bPara disposition of dopants.

Several examples in the literature that evidence the advantage of introducing two, three, or even four different heteroatoms will be reported below. The present review is organized in the following way: I first discuss in section 2 what can be considered a dopant since there is no consensus in the literature, and then I briefly inform in section 3 which dopants have been introduced in the graphene framework. In section 4, I advocate commenting on the codoped graphene systems that have been studied, and finally, to conclude the review, in section 5 I indicate the challenges that need to be surmounted for the title field to mature. Finally, I apologize for all the articles that have not been included in this review. The literature about doped graphene contains over 1 million articles, thesis, books, and technical reports. Including all of them in a single review would be impossible. Nevertheless, I expect that the nearly 500 works included cover most of the examples available in the literature and contribute to developing this fascinating research area.

2. When can a heteroatom be considered a dopant?

In this review, heteroatom doping of graphene is considered to be attained when at least one carbon atom of graphene is replaced by another element. Although this definition seems straightforward, when reading the literature, it is possible to become confused. For example, several publications claim to have doped graphene with fluorine atoms.¹⁴⁻²¹ Considering that fluorine has a strong tendency to form only single bonds because it has seven valence electrons, it is not easy to imagine this element embedded in the graphene framework replacing a carbon atom, as it will bind only with one carbon atom. Vineesh et al.¹⁴ reported that N and F codoped graphene displays “enhanced electrocatalytic efficiency than the ‘N’ and ‘F’ individually doped graphene” for the oxygen reduction reaction (ORR). In Vineesh’s work,¹⁴ nitrogen was introduced by thermal treatment of fluorinated graphene in the presence of melamine. The amounts of N and F introduced were 2 and 4 atom %, respectively. XPS confirmed the presence of nitrogen in several forms—pyrrolic, pyridinic, and graphitic—while in the case of fluorine C–F and F–C–F bonds were confirmed. Along the same line, Jiang et al.¹⁵ prepared N and F codoped graphene by thermal treatment of graphene oxide/polyaniline composites and NH₄F. It was proposed as an efficient metal-free ORR catalyst in fuel cells. XPS confirmed the presence of N and F. Nitrogen was found in the same configuration discussed above. However, fluorine was proposed to be present in ionic and semi-ionic C–F bonds. Liu et al.¹⁶ also reported the preparation of N and F codoped graphene. The method was simple: they fluorinated N-doped reduced graphene oxide using XeF₂ at 180 °C. XPS confirmed C–F covalent bonding, and two more peaks corresponding to CF–CF_n and CF₂ were determined. The material exhibited interesting photoluminescence properties. Although the presented results represented a significant advance over the existing literature, it is hard to accept that the fluorine atoms are dopants. Instead, in my opinion it is fairer to consider fluorine a functional group covalently attached to graphene but not a member of the graphene framework because, once a fluorine atom is added, nothing more can be bonded.

Fluorine is not the only halogen reported to accompany nitrogen or phosphorus dopants. There are reports in the literature of nitrogen and halogen codoped graphene, with particular attention to iodine,²²⁻²⁴ and P–X codoped graphene with X = Cl, Br, and I.²⁵ In general, the amount of iodine is small, below 1%, and XPS is not conclusive on the nature of the iodine dopant. Wu et al.²² determined that, in PI codoped graphene, the high resolution of the XPS signal corresponded to I₃ and I₅. Thus it is unlikely that iodine was introduced in the graphene framework. Instead, it is more likely to be adsorbed between graphene layers. Nevertheless, it should be remembered that halogens heavier than fluorine have the capability to form multiple bonds and may be able to replace carbon atoms. Finally, it is essential to mention that, despite being adsorbed, halogens can effectively “dope” graphene because they are natural σ-donors and π-electron withdrawers.²⁵ Therefore, despite the nature of the halogen dopant, i.e., adsorbed or embedded, it works as a dopant, but adsorbed halogens will not be covered in this review.

Finally, it is interesting to mention the case of oxygen. First, it is crucial to consider that it is a common practice to obtain graphene sheets by reducing graphene oxide. In some cases, the reduction is not complete, and also during this process, dopants are introduced using a suitable gas. Considering that in some cases oxygen is not 100% eliminated, in practice, two dopants are present—the one whose introduction was pursued using a gas (for example) and oxygen—so many cases of single atom doping are truly codoped graphene systems. Zhan et al.²³ observed the O 1s peak when preparing the I and N codoped graphene nanocomposite described above. The codoping with iodine and nitrogen better facilitated oxygen bonding than the single atom doping of nitrogen or iodine.²³ Therefore, caution should be taken when doping is claimed because doping may have been achieved by functionalization, adsorption, or substitution. The latter process is the one that I discuss herein. I considered that an atom is a dopant if it is embedded in the graphene framework replacing one or more carbon atoms.

3. Brief summary of single-atom-doped graphene

When replacing a carbon atom with a dopant, it is not surprising to find that, as early as 2009, Panchakarla et al.²⁶ reported the preparation of boron²⁶⁻²⁸ and nitrogen^26,28−35 doped graphene. These two elements present atomic radii of 85 and 65 pm, respectively, very close to the value of carbon, 70 pm. Therefore, the graphene sheet will not suffer excessive stress if these elements are embedded in the graphene framework. The influence of precursors and synthesis conditions on the preparation of N-doped graphene have been established in the milestone investigation by Wang et al.³⁰ as early as 2014. Also, for a detailed review of the different strategies available to synthesize N-doped graphene, we refer the readers to the excellent review by Vesel et al.³⁵ In particular, Table 1 of the latter summarizes the postsynthesis of N-doped graphene while Table 2, of the same reference, provides an overview of the literature on the direct synthesis of N-doped graphene. A slightly different situation is observed for Be-doped graphene. Although this element has only four electrons, theoretical calculations have shown that the Be atom does not lie in the graphene but bulges out of the sheet.^36,37 The protrusion is significant; for example, in a 5 × 5 graphene sheet, the Be atom is located 0.78 Å above the carbon atoms. Although this system is stable, it is waiting to be synthesized. It is expected to have outstanding properties that may render it valid as an anode material for lithium ion batteries.³⁷ Moving down in the periodic table, Al-doped graphene was studied for more than 10 years by theoretical materials scientists.³⁸⁻⁴⁴ It was not until recently that Al-doped graphene was prepared.⁴⁵⁻⁴⁷ It was the last 3p element used to dope graphene. Ullah et al.⁴⁶ successfully prepared a large area Al-doped graphene by chemical vapor deposition. Zagler et al.⁴⁷ showed that the Al dopants were found in 3- and 4-fold coordinated configurations. Occasionally N dopants were found to be embedded in the graphene framework and bonded to Al, as predicted by us.^48,49

On the contrary, Si-doped graphene⁵⁰⁻⁵⁶ was prepared earlier. In 2012, Zhou et al.⁵⁰ reported that Si atoms could be found in mono- and divacancies in the graphene layer. When Si is bonded to three atoms, it prefers sp³ hybridization, but if it is bonded to four atoms, sp² d-like hybridization is adopted. As observed for Al, nitrogen was sometimes found to replace one of the carbon atoms bonded to the Si impurities, again in agreement with our findings.^48,49 Si-doped graphene is expected to have outstanding mechanical,⁵² electronic,⁵³⁻⁵⁵ and optical properties.^54,55 Also, I have demonstrated that, when it is present in the graphene layer, it can promote cycloaddition reactions,⁵⁶ which can be difficult to attain on perfect graphene.⁵⁷ The main effect of the Si dopant is to reduce the enormous deformation energy required to form a covalent bond with the diene or dienophile. It has the same effect as a functional group⁵⁸ present before the cycloaddition reaction or the SiC surface in which graphene can grow.⁵⁹

The doping of graphene with phosphorus⁶⁰⁻⁶⁹ has been developed at least since 2012 when Some et al.⁶⁰ reported the preparation of P-doped graphene, air-stable n-type field effect transistors. The system consisted of P-doped, double-layered graphene sheets. Subsequent reports by Li et al.⁶¹ and Zhang et al.⁶² also reported the preparation of P-doped graphene in 2013. This system has unique electronic and magnetic properties^{39,40,53,66,67} for designing metal-free ORR catalysts,^61,62 Li ion batteries,⁶² and NH₃ sensors.⁶³ It is important to notice that Susi et al.⁶⁴ were the first to obtain atomic resolution imaging and electron energy loss spectroscopy evidencing the presence of phosphorus atoms in the graphene lattice.

In 2009, I reported one of the first studies about sulfur-doped graphene and found that this material may have remarkable properties.^42,43,70 Three years later, Yang et al.⁷¹ reported the preparation of sulfur-doped graphene as an efficient metal-free cathode catalyst for ORR reactions. Similarly, Yang et al.⁷² also reported the preparation of S-doped graphene based on ultrathin graphene oxide for ORR reactions. Also, in 2012, Rao et al.⁷³ used liquid precursors to prepare S-doped graphene. The synthesis of the latter has been mastered during the last years, and many routes are available to obtain S-doped graphene, even large area sheets.⁷¹⁻⁷⁶

Going down in the periodic table, I studied graphene doped with the 4p elements Ga, Ge, As, and Se.⁷⁷ These nanomaterials exhibited outstanding electronic and magnetic properties and high reactivity near the dopant site, which may render them valuable catalysts. In 2018, Tripathi et al.⁷⁸ implanted germanium in a graphene framework. The authors claimed that ⁷⁴Ge⁺ was the heaviest impurity implanted into monolayer graphene at that time. This atom can substitute a single carbon atom, bonding to three neighbors, or be 4-fold coordinated in a divacancy. In their 2012 article about S-doped graphene, Yang et al.⁷¹ also claimed to have prepared Se-doped graphene by direct annealing of graphene oxide in the presence of diphenyl diselenide in argon. XPS studies confirmed the presence of Se atoms, and Se content did not change after sonication. Therefore, it was concluded that diphenyl diselenide was not adsorbed. However, further studies are necessary to have a more specific description of Se doping. In 2018, Meng et al.⁷⁹ prepared Se-doped graphene for high-efficiency triiodide reduction in dye-sensitized solar cells. SEM mappings revealed the presence of C, O, and Se. The distribution of Se was uniform, and it was concluded that Se atoms were incorporated into the graphene framework.

Turning our attention to the 3d metals, Toh et al.⁸⁰ prepared Mn-, Fe-, Co-, and Ni-doped graphene hybrids, which are helpful in electrocatalysis. Carnevali et al.⁸¹ doped epitaxial graphene by direct incorporation of Ni adatoms. Scandium-doped graphene was also reported in 2019 by Wen et al.,⁸² but nitrogen was necessary as a stabilizer. Again, as observed for Al and Si, the presence of Sc–N was confirmed by XPS. Robertson et al.⁸³ studied the dynamics of Fe atoms in graphene single and double vacancies, as can be appreciated in Figure 1, and as early as 2014, Zhao et al.⁸⁴ reported the synthesis of a free-standing, single-atom-thick Fe layer in a graphene pore. More recently, Mn-doped graphene⁸⁵ was prepared with a Mn concentration of 0.04 atom %. Magnesium was unambiguously located in a single vacancy, but the introduction in divacancies and other structures were observed also. Nevertheless, their structural identification is not clear.

Figure 1.

False color aberration-corrected TEM images of (a) an Fe substitutional defect in a graphene monovacancy (Fe@MV) and (b) an Fe interstitial defect occupying a divacancy (Fe@DV). (c) Smoothed aberration-corrected TEM image of the Fe@MV shown in (a). (d) DFT optimization of the Fe@MV structure (the inset shows a side view) and (e) a multislice TEM image simulation of the system. (f–h) Similar to (c)–(e) but for a DFT-optimized Fe@DV. Scale bars denote 0.5 nm. Reproduced from ref (83). Copyright 2013 American Chemical Society.

Heavier elements like gold were implanted by Trentino et al.⁸⁶ using a two-step implantation process. The new doping method utilizes a two-step low-energy ion implantation technique that, according to the authors, “overcomes the limitation posed by momentum conservation on the mass of the implanted species.” The gold atoms occupy double vacancy sites. Finally, Nb-doped graphene was prepared by Li et al.⁸⁷ The strategy employed to obtain Nb-self-doped graphene was to prepare it from 2D NbC, only by one step of incomplete chlorination. This proposal is similar to the procedure I suggested to prepare Si-doped graphene. The structure is shown in Figure 2. The partial annealing to the SiC layer may leave some Si atoms in a specific disposition of the dopants.⁸⁸ Finally, Sofer et al.⁸⁹ reported the preparation of U- and Th-doped graphene. However, XPS was not conclusive in the nature of the dopants since the energies of the U 4f peaks indicated the presence of the form UO₃, U₃O₈, or uranium carbide.

Figure 2.

Top (a) and side (b) views of the optimized unit cell for 4 × 4 siligraphene on 6H-SiC(0001). Carbon, silicon, and hydrogen atoms in SiC are colored in gray, light blue, and white, respectively. Carbon and silicon atoms in the siligraphene layer are colored pink and orange, respectively.

4. Codoped or dual-doped graphene

4.1. B and N Codoped Graphene

Boron and nitrogen are the most obvious choices to dope graphene because they bracket carbon in the periodic table. The fact that they have smaller and larger electronegativities than carbon makes them very appealing to develop metal-free catalysts, thanks to the charge imbalance induced by their presence. In 2013, Zheng et al.⁹⁰ reported the preparation of B and N codoped graphene with long-term stability and excellent activity for the ORR. The two-step method prevented the formation of catalytically inactive byproducts such as hexagonal boron nitride and induced a cooperative effect between the dopants. One year later, the solvothermal synthesis of B and N codoped graphene was achieved,⁹¹ and various theoretical studies were published highlighting the unique electronic properties of this system.⁹²⁻⁹⁸ As discussed in section 3, one of the main applications of doped graphene is as a catalyst for the ORR. B and N codoped graphene is not the exception, and multiple studies demonstrated that it could be an excellent catalyst for the ORR.⁹⁸⁻¹⁰⁸ Another essential use of B and N codoped graphene is in lithium batteries^109−112 and sodium ion batteries.¹¹³ Huang et al.¹⁰⁹ reported a high capacity of up to 909 mAh g^–1 and an excellent discharge capacity after 125 cycles. Three-dimensional B and N codoped graphene has been employed to construct symmetric and asymmetric supercapacitor electrodes.^114−116 The work by Kang et al.¹¹⁴ demonstrated that the specific capacitance of 283 F g^–1 at 1 A g^–1, in alkaline aqueous electrolyte, is more remarkable than those corresponding to N or B monodoped graphene.

In the catalysis area, it has been demonstrated that B and N codoped graphene can be an excellent catalyst for the electrooxidation of formic acid¹¹⁷ methanol¹¹⁸ and the reduction of nitroarenes¹¹⁹ and triiodide.¹²⁰ The development of graphene based sensors is another field in which B and N codoped graphene has excelled. It has been able to detect aflatoxin B1,¹²¹ mercury(II),^122,123 fluorine ions,¹²³ hydrogen peroxide,¹²⁴ cymoxanil,¹²⁵ and NO₂,¹²⁶ to mention just a few examples of molecules that can be monitored using B and N codoped graphene.

B and N codoped graphene quantum dots have fascinating optical properties.^127−130 They can be used in imaging and photothermal therapies.¹²⁸ In a related area, B and N codoped graphene was used in dye-sensitized solar cells,^129,130 as a counter electrode for iodine reduction.¹³⁰ Finally, it has been shown that codoping with B and N enhances the electromagnetic interference shielding (EMI)^131−134 to −42 dB (99.99% of attenuation) at a critical thickness of 1.2 mm.¹³³

4.2. O and N Codoped Graphene

I explained above that I considered an atom a dopant if embedded in the graphene framework replacing one or more carbon atoms. Nevertheless, I have included O and N codoped graphene in this review because plenty of works report preparations and applications of this system.^135−146 In 2014, Chen et al.¹³⁵ reported that N and O codoped carbon hydrogel could be used as substrate-free electrode for highly efficient oxygen evolution reaction.¹³⁵ The nitrogen atoms were introduced with ammonium hydroxide, but the oxygen atoms were residual oxygen impurities on graphene prepared by a chemical method. Oxygen was present in the forms carboxyl and epoxy, so it is questionable if oxygen is a dopant because COOH and COC groups can be present without replacing carbon atoms. I do not argue against the utility of these groups as catalytic centers for the oxygen evolution reaction, but they should be named functional groups instead of dopants, in my opinion. Again the discussion if oxygen can be considered a dopant emerges. Li et al.¹³⁸ assembled O and N codoped graphene on hierarchical carbon networks for all-solid-state flexible supercapacitors. XPS studies revealed that 11% of N was pyridinic and 85% was pyrrolic. However, oxygen was located at the edges as hydroxyl groups. Therefore, in this case, the claim that oxygen is a dopant is at least questionable. In our opinion, the material is N-doped graphene functionalized with hydroxyl groups. Supercapacitor electrodes were constructed using N- and O-doped graphene.^136−139 Lithium^140−143 and potassium^143−146 ion batteries based on N and O codoped graphene are available. Excellent potassium storage was achieved. For example, a reversible capacity of 464.9 mAh g^–1 at 0.05 A g^–1 was reported by Ruan et al.¹⁴⁶

4.3. S and N Codoped Graphene

Despite the fact that sulfur does not fit in the graphene plane and bulges out of the sheet,⁷⁰ the most studied codoped graphene system is sulfur and nitrogen codoped graphene. There are so many articles published about this system^147−360 that a review could be written. Including all the works dealing with this system would be an interminable task, so I apologize for the papers not being cited. In 2012, Liang et al.¹⁴⁷ reported the one-step synthesis of sulfur and nitrogen codoped mesoporous graphene for the ORR with a synergistically enhanced performance. The ORR activity was comparable to that of the best commercial Pt/C catalysts and better than the activity measured when only one of the dopants was present. The N and S elemental contents were 4.5 and 2.0 atom %, respectively. Nitrogen was present in the typically observed forms of pyridinic, pyrrolic, and graphitic, while sulfur was present in the thiophenic forms of C–S–C. These results are in partial agreement with our investigations on this system. By means of first principles calculations, I studied all codoped graphene systems with one 3p and one 2p element.^48,49,148 Our results revealed a notorious preference for the dopants to be located in specific positions. All XY codoped graphene systems (X = B, N, O; Y = Al, Si, P, S) preferred to replace contiguous C atoms (a CC bond) instead of being separated at large distances or in para/meta positions, as can be appreciated in Table 2; the structures are presented in Figure 3. There was only one exception to this empirical rule: Si and B codoped graphene, for which there was a slight preference for the atoms to adopt a para arrangement. The case of S and N codoped graphene is very particular because the atoms replace a CC bond but the dopants are not bonded. In effect, the nitrogen atoms adopt a pyridinic disposition and the sulfur atom protrudes out of the sheet, as shown in Figure 4. Therefore, nitrogen prefers a C–N–C environment and sulfur prefers a C–S–C environment. This particular dopant disposition may explain this system’s unusual properties. For example, adding O₂ to S and N codoped graphene is very exothermic, forming an SO₂ unit bonded to graphene.¹⁴⁸

Table 2. Relative Energies (eV) between the Five Configurations Studied of the Dopants and Heteroatom Bond Distances (Å) at the M06-L/6-31G* and VDW-DF/DZP Levels of Theory.

	ortho M06-L	meta M06-L	para M06-L	para VDW-DF	same latt M06-L	diff latt M06-L	XY distance (Å)a M06-L
AlB	0.0	0.57	0.10	–0.20	0.76	0.80	2.16
AlN	0.0	1.82	1.90	1.70	2.19	2.18	1.81
AlO	0.0	3.97	3.69	3.48	4.55	4.65	1.97
SiB	0.47	0.22	0.0	–0.60	0.19	0.26	1.88
SiN	0.0	1.34	1.34	1.19	1.59	1.54	1.80
SiO	0.0	3.57	2.91	2.76	3.76	3.64	2.02
PB	0.0	0.35	0.04	–0.05	0.60	0.67	1.80
PN	0.0	1.01	0.70	0.61	1.09	1.00	1.78
PO	0.0	3.13	2.71	2.51	3.27	3.17	2.48
SB	0.0	0.86	0.82	0.67	1.25	1.34	1.82
SN	0.0	1.56	1.32	1.18	1.46	1.43	2.58
SO	0.0	5.20	5.64	5.28	5.17	5.41	2.59

^aXY distance corresponds to the 2p and 3p dopants in an ortho arrangement (case a)).

Figure 3.

Substitution sites considered to study the structure of codoped graphene.

Figure 4.

Optimized 6 × 6 unit cell for sulfur and nitrogen dual-doped graphene.

Many methods are capable of producing S and N codoped graphene. It is possible to start from graphene oxide produced via Hummers’ method and use melamine and benzyl disulfide as N and S precursors. A different approach was proposed by Kicinski et al.,¹⁵⁵ who copolymerized S- and N-containing heterocyclic aldehydes, whereas Chen et al.¹⁵⁶ used methyl blue/montmorillonite composites. A green route to S and N codoped graphene is also available by using Chinese medical herbs, as proposed by Feng et al.¹⁵⁷ Along the same line, Wu et al.¹⁵⁸ reported that, assisted by supramolecular polymerization, petroleum coke can be converted into S and N codoped graphene. This approach is exciting because it starts from a byproduct of the oil refinery industry. In theory, creating a new synthetic route using a different heterocycle is possible. For example, Zhang et al.¹⁵⁹ self-polymerized polydopamide and then reacted the product with cysteine. These are six examples of the methods available to obtain S and N codoped graphene. More synthetic routes can be found in the references cited, but in our opinion, all of them offer advantages and disadvantages, such as the quality of the sheets prepared, their area, and the bonding nature of the dopants. The main problem with the synthetic procedures available is that there is no good structure–activity relationship.

The main application of the prepared S and N codoped graphene materials has been in perhaps the most critical reaction in the industry, the oxygen reduction reaction. Following the first work by Liang et al.¹⁴⁷ commented on above, more than 40 papers were published reporting the catalytic power of S and N codoped graphene for the ORR reaction.^{154,158−201} Most of these works evidence a performance similar to or superior to those of commercial Pt/C catalysts and superior stability. Nevertheless, more work is needed because, to the best of our knowledge, S and N codoped graphene is not the most used catalyst for oxygen reduction/evolution reactions. For a more detailed review of carbon based metal-free ORR catalysts, I refer the reader to the review by Ma et al.²⁰²

Another vital reaction that the graphene community has intensively studied is the hydrogen evolution reaction (HER).^203−212 Jiang et al.²⁰³ used an intelligent approach to synthesize S and N codoped graphene from a mixture of urea, glucose, and phosphoric acid. The high dopant content and porosity conferred a high catalytic activity in the HER reaction with an onset potential of 0.12 V and a Tafel slope of 79 meV/dec, values that are comparable with those of an average metallic catalyst. There have been some improvements in these values. For example, Guruprasad et al.²⁰⁴ reduced the Tafel slope to 47 meV/dec However, to the best of our knowledge, graphene based catalysts have not replaced the ones previously utilized in the industry. In this line, the photocatalytic production of hydrogen from water splitting is crucial for humankind. S and N codoped graphene has been combined with TiO₂ to improve the activity under visible light significantly. S and N graphene based materials are generally used as light absorbers combined with TiO₂ or similar materials.^208,209

The use of S and N codoped graphene as a catalyst is not limited to the ORR and HER reactions. It has been used with great success for catalytic phenol degradation,^212,213 nitrogen reduction,²¹⁴ methanol^213−217 and ethanol²¹⁸ electrooxidation, aerobic oxidation of alcohols²¹⁹ under visible light irradiation, and sonocatalytic decolorization of methylene blue.²²⁰ Another area of utmost importance is organic catalysis. Some works reported using S and N codoped graphene for Sonogashira²²¹ and Heck couplings.²²² Although Pd nanoparticles were still necessary, the presence of the graphene material contributed to a more effective use of the expensive Pd catalyst.

There are several examples in which S and N codoped graphene has been employed in photocatalysis.^223,224 For example, it can degrade rhodamine B^223,224 and methyl orange.²²⁵ In the former case, the activity is 3 and 10 times higher than those determined for N-doped graphene and P25 TiO₂, respectively.²²³ Efficient photocatalytic H₂O₂ production was achieved since the codoped graphene induced a better light absorption and promoted a more significant charge migration.²²⁶ Sulfur and nitrogen codoped graphene nanomaterials have broad visible absorption^223,224 and superb luminescence properties.^226−232 They can be used in bioimaging.^233,234 Also, they were reported to have a fluorescence quantum yield 9.3 times higher than that of undoped graphene. Thus, they are promising materials for developing light-emitting devices.²²⁸

The development of alkali metal ion batteries is another area that has been invaded by publications that use S and N codoped graphene.^235−269 In 2014, Ma et al.¹⁵³ produced S and N cocoped graphene using a chemical vapor deposition approach. The three-dimensional codoped graphene networks were utilized as anode materials for lithium ion batteries. The capacity was 3525 mAh/g at a current density of 50 mA/g, and the rate capability was as high as 870 mAh/g at 1000 mA/g, with with excellent cycling stability. Following Ma et al.’s¹⁵³ landmark investigation, several works continued this line of research.^235−261 In some cases, metal nanoparticles,^235,236 nanocables,²³⁷ and nanospheres²³⁸ are also combined with S and N doped graphene to improve performance. Although the numbers obtained are impressive, and are in agreement with our theoretical predictions,¹³ to the best of my knowledge, there is not a commercial lithium ion battery based on S and N codoped graphene. Therefore much work is needed in this area.

The excellent performance observed for lithium batteries prompted researchers to study if the effect was similar for heavier and cheaper alkali metals like Na and K. In effect, superb performances were observed for sodium^261−267 and potassium storage.^268,269 Recent theoretical calculations that I carried out support these results.¹⁵² I demonstrated that, as observed for lithium,¹³ S and N codoped graphene presents the strongest affinity toward Na and K adsorption.¹⁵² The adsorption energies of Na and K are gathered in Tables 3 and 4, respectively, and the structure is giving in Figure 5. The adsorption energies for Na and K were dramatically increased by the presence of the S and N dopants, another system for which I observed comparable adsorption energies was Si and B codoped graphene. Therefore, at least for Na and K, there seems to be another alternative to dope graphene and improve alkali metal storage. Metal–air batteries,^270−277 in particular Zn–air ones,^270−276 were also constructed using S and N codoped graphene. Geng et al.²⁷⁰ used S and N codoped graphene decorated with CoS nanoparticles as a cathode material for Zn–air batteries. The product could be charged and discharged for 50 cycles at 1.25 mA/cm², for 50 h with an almost constant discharge voltage of 1.23 V. In sharp contrast, the discharge voltage of the commercial Pt/C catalyst was reduced from 1.36 to 1.20 V. In general, the catalytic performances of the commercial catalyst and the S and N codoped graphene based one were similar, so it is a cheaper alternative. Ganesan et al.²⁷¹ also claimed that N and S codoped graphene/CoS₂ nanoparticles could be a suitable air cathode for Zn–air batteries. Chen et al.²⁷² went one step further and did not use nanoparticles to construct a cathode for Zn–air batteries. Instead, they used S and N codoped graphene microwires and obtained a superior performance. For more details on this topic, I refer the reader to refs (270−277). Finally, vanadium redox flow batteries were built using S and N codoped graphene by Li et al.²⁷⁸ and Daugherty et al.²⁷⁹ An improvement of 85.37% at a current density of 80 mA/cm² was reported by Li et al.²⁷⁸

Table 3. Sodium Adsorption Energies onto Codoped Graphene 5 × 5 and Circumcoronene.

dopant	VDW-DF/DZP G5×5a	M06-2X/6-311G* circumcoroneneb	PBE-D2/VASP G5×5	PBE-D2/6-311G* circumcoronene
AlB	–1.58	–1.81	–2.44	–2.10
AlN	–1.55	–1.98	–2.44	–2.05
AlO	–1.63	–1.64	–2.43	–1.99
SiB	–2.13	–2.90	–2.72	–2.94
SiN	–1.63	–2.11	–2.42	–2.09
SiO	–1.42	–1.34	–2.21	–1.62
PB	–1.47	–1.75	–2.36	–1.91
PN	–0.96	–1.15	–1.92	–1.37
PO	–1.71	–1.97	–2.51	–2.18
SB	–1.10	–1.96	–2.35	–2.03
SN	–2.00	–3.35	–2.91	–2.92
SO	–0.97	–1.30	–2.13	–1.49
G5×5	–0.67	–0.45	–1.08	–0.71

^aCalculations were performed using periodic conditions and a 5 × 5 unit cell of graphene.

^bCalculations were performed using a graphene flake (circumcoronene).

Table 4. Potassium Adsorption Energies onto Codoped Graphene 5 × 5 and Circumcoronene.

dopant	VDW-DF/DZP G5×5a	M06-2X/6-311G* circumcoroneneb	PBE-D2/VASP G5×5	PBE-D2/6-311G* circumcoronene
AlB	–1.81	–2.22	–2.55	–2.33
AlN	–1.75	–2.35	–2.51	–2.27
AlO	–1.78	–1.95	–2.47	–2.18
SiB	–2.42	–3.36	–3.06	–3.24
SiN	–1.84	–2.47	–2.51	–2.32
SiO	–1.64	–1.75	–2.31	–1.88
PB	–1.81	–2.20	–2.51	–1.93
PN	–1.40	–1.57	–2.05	–1.65
PO	–1.95	–2.32	–2.58	–1.39
SB	–1.08	–2.36	–2.47	–2.31
SN	–2.42	–3.64	–3.03	–3.05
SO	–1.20	–1.69	–2.22	–1.73
G5×5	–1.1	–0.96	–1.40	–1.10

^aCalculations were performed using periodic conditions and a 5 × 5 unit cell of graphene.

^bCalculations were performed using a graphene flake (circumcoronene).

Figure 5.

Optimized structure for codoped graphene with an ortho disposition of dopants and an alkali metal atom adsorbed below the 3p dopant.

Symmetric and asymmetric supercapacitor electrodes were created using S and N codoped graphene.^280−305 As early as 2015, Xing et al.²⁸¹ prepared three-dimensional S and N codoped graphene hydrogels using thiocarbohydrazide as a reducing and doping agent. Thanks to the highly porous architecture and presence of dopants, the materials exhibited a high specific capacitance of 141.1 F/g in KOH electrolyte. Their specific capacitance could be maintained at 71.3% even with a discharging current density of 10 A/g. Also, excellent electrochemical stability and reversibility were obtained, with about 90% of retention after 4000 cycles. Even more interesting results were obtained by Tran et al.,²⁸³ who prepared a S and N codoped hole defect graphene hydrogel. A maximum energy density of 14.8 Wh/kg and excellent cycle stability were reported. In general, the porous nature of the S and N based materials is crucial for the performance of the supercapacitor.^280−305 Comparing the performances of the electrodes prepared is not easy because different conditions are employed. Thus, I refer the reader to refs (280−305) gathered on this topic.

The quest for dye-sensitized solar cells with higher efficiency has also tested if the inclusion of S and N codoped graphene can be useful.^306−315 The advantages of codoping as compared with the monodoping when preparing dye-sensitized solar cells (DSSCs) were also confirmed by Luo et al.³⁰⁶ A power conversion efficiency (PCE) of 4.23% was obtained for sulfur reduced graphene oxide, while for sulfur and nitrogen reduced graphene oxide the value increased to 4.73%. In 2015, Xu et al.³⁰⁷ built bifacial DSSCs capable of utilizing incidental light from the rear and front sides that relied on transparent S and N codoped graphene as counter electrodes. A drastically enhanced power conversion efficiency was obtained for both front and rear illumination. Kannan et al.³¹¹ reported a PCE as high as 7.42% when S and N codoped graphene was utilized as a counter electrode. The latter value is comparable to the one determined for the Pt counter electrodes, namely 7.42%. The high catalytic activity of the codoped graphene electrode was attributed to the difference in electronegativity between S and N as well as to the structural distortions caused by the presence of the sulfur atom in the graphene framework. Theoretical investigations have also proven the utility of the codoping of graphene in this area. Liu et al.³¹² performed first principles calculations to show that the synergistic effect between the S and N dopants resulted in a much better charge transfer between the substrate and the adsorbed I₂ on the surface of S and N codoped graphene. Therefore, a much better efficiency would be obtained in DSSCs that employ codoped graphene.

S and N codoped graphene based sensors have been extensively studied in the literature.^316−347 S and N codoped graphene was used by Li et al.³¹⁶ as a support for CuO nanoparticles to promote electrocatalytic glucose oxidation. A low detection limit of 0.07 μM was reported by Li et al.³¹⁶ and a detection limit of 0.5 μM was reported by Tian et al.³¹⁸ using S and N codoped graphene prepared with a one-step and cost-effective microwave assisted solvothermal method. Masteri-Farahani et al.³¹⁹ developed a fluorescence based sensor to detect glucose. They combined boronic acid with the codoped graphene sheets to obtain a lower detection limit of 5.5 μM. The codoped graphene based new sensor is much cheaper because it reduces the use of the expensive boronic acid. In the biological area, S and N codoped graphene based sensors have been constructed for l-cysteine,³²⁰ hydroquinone/catechol,³²¹ cyanide,³²² an immunosensor for cardiac troponin I,³²³ okadaic acid,³²⁴ an immunosensor for cardiac troponin,³²⁵ ascorbic acid,^326,327 and dopamine,^327,328 to mention just a few examples. In the same vein, pesticides³²⁹ and toxic pollutants³³⁰ can be detected. The sensor array built by Zhu et al.³²⁹ was able to discriminate among five pesticides: lactofen, fluoroxypyr-meptyl, bensulfuron-methyl, fomesafen, and diafenthiuron. In this case, there was a practical application of the sensor array as it was validated by discriminating pesticides in soil. In food samples, ethion³³¹ and nitrites³³² are easily monitored using the fluorescence properties of S and N codoped graphene quantum dots. Martins et al.³³³ used glassy carbon electrodes modified with S and N codoped graphene quantum dots for monitoring multivitamins. The detection limits for vitamins B₂, B₆, and B₁₂ were 0.30, 30.1, and 0.32 nmol/L. The sensor was effectively implemented in the quantification of vitamins in classic and fruit based energy drinks. Simpler molecules like water³³⁴ and ammonia³³⁵ can be sensed using S and N graphene quantum dots. Jlassi et al.³³⁴ constructed a S and N graphene carbon dot based impedance sensor with good sensitivity for water and excellent response and recovery times of 15 s and 55 s, respectively. The detection of explosives is a very sensitive topic that needs constant updates. Mondal et al.³³⁶ observed a remarkable increase in fluorescence quenching effect using a micromolar solution of 2,4,6-trinitrophenol (TNP). The detection limit was as low as 19.05 ppb. Zhang et al.³³⁷ used an electrochemical sensing platform using S and N codoped graphene nanoribbons that showed a highly sensitive and selective response to trinitrotoluene (TNT) with a wide linear range from 0.0008 to 5.1 ppm and a detection limit as low as 0.1 ppb. Finally, S and N codoped graphene has been employed to detect a variety of metal and nonmetal atoms. For example, the fluorescence detection of Co(II) ions in water was reported by Boonta et al.,³³⁸ while the successful detection of Hg²⁺ ions is well documented.^339−342 In general, the detection is related to fluorescence quenching after interacting with the metal ion. Gu et al.³⁴⁰ demonstrated that the sensitivity of S and N graphene carbon dots was 4.23 times higher than that of N-doped counterparts. The sensor was sensitive enough to ensure that drinking water had less than 10 ppb Hg²⁺ ions. In addition to this, a report by Tian et al.³⁴⁴ revealed that Hg(II) ultraselectively separated from Pb(II) and Cu(II). A high removal ability above 99% was obtained. The list of elements that can be detected in quite long, but it includes Fe³⁺, with a low detection limit of 0.07 μM,^342,343 Pb(II)³⁴⁵ with a detection limit as low as 0.29 nM, iodide^341−346 with a detection limit of 4.23 nM, and Cu²⁺ and Ag⁺ with detection limits of 250 and 50 nM, respectively.³⁴⁷ Interestingly, the sensor constructed by Zhang et al.³⁴¹ was an off–on sensor that had the chemiluminescence quenched by iodide and recovered by mercury ions.

We have listed a few areas in which S and N codoped graphene has made an impact, but I limit them to maintain a balance in this review. There are others such as CO₂ capture,³⁴⁸ removal of organic dyes,^349−352 design of efficient electromagnetic wave absorbers,^353−356 piezocatalysis,^357,358 bioelectricity production,³⁵⁹ and synthesis of high performance polymer nanocomposites.³⁶⁰

4.4. P and N Codoped Graphene

It is not a surprise that another of the most studied cocoped graphene systems also contains nitrogen, but in this case it is chaperoned by a pnictogen: phosphorus.^361−420 In 2014, Zheng et al.³⁶¹ reported the synthesis of P and N codoped graphene and its use as nonmetal catalyst for the hydrogen evolution reaction. The performance was much better than those corresponding to monodoped graphene and comparable to some commercial catalysts. Two approaches were employed to synthesize this material. In the first approach, a single-step method, graphene oxide was annealed in the presence of melamine and triphenylphosphine, while in the second approach, which involved two steps, first P-doped graphene was obtained and the nitrogen was incorporated. K-edge NEXAFS and XPS studies indicated that nitrogen was present in pyridinic, graphitic, and pyrrolic forms whereas phosphorus was bonded to carbon or oxygen. The presence of P–N bonds was excluded. This result is in contrast to experimental studies that indicated the presence of AlN bonds in Al-doped graphene⁴⁷ and SiN bonds in Si-doped graphene.⁵¹ Also it is in disagreement with our investigations which indicated that the P and N dopants prefer to form PN in P and N codoped graphene. In effect, the ortho disposition of the dopants is 1.01 and 0.70 eV more stable than the meta and para ones, respectively.^48,49 A few months later, Xue et al.³⁶² prepared P and N codoped graphene with air-stable n-type characteristics. Phosphorus played a critical role in achieving improved charge donation. The material was synthesized over a Cu surface and using phosphonitrilic chlorine trimer as a P and N source. A different approach to prepare P and N codoped graphene was proposed by Li et al.³⁶³ The method was facile and cost-effective; they pyrolyzed a dried hydrogel composed of graphene oxide, polyaniline, and phytic acid. These are three examples of how P and N codoped graphene can be synthesized. Since there is not a standard method to obtain it, I refer the readers to the references included, as most of them report a unique method.^361−422 For example, Ananthanarayanan et al.³⁶⁴ used adenosine triphosphate and Guo et al.³⁶⁵ used hypophosphorous acid as a P source to dope graphene oxide.

In line with our discussion about S and N codoped graphene, the oxygen reduction reaction (ORR) and the oxygen evolution reaction (OER) were the subject of several investigations that tested the viability of using P and N codoped graphene as a catalyst.^{363,365−373} Li et al.’s³⁶³ study was one of the first to not only prepare the material but also to use P and N codoped graphene as a catalyst for ORR and OER reactions. The authors claimed that up to 2015 it was the best nonmetal bifunctional electrocatalyst reported for the latter reaction, with a potential gap of 0.71 V between the OER potential, at a current density of 10 mA/cm², and the ORR potential, at a current density of −3 mA/cm². A similar ORR and OER overpotential of 705 mV was reported by Chai et al.³⁶⁶ using P and N codoped graphene prepared via a one-pot hydrothermal method utilizing graphene oxide, ammonium dihydrogen phosphate, and cyanamide as precursors. The P sites were very active and could be oxidized, becoming nonactive for catalysts. However, it was pointed out by Chai et al.³⁶⁶ that if P is oxidized and bounded to an N codopant it stabilizes the graphitic N and increases the reactivity of the neighboring carbon atoms. For this reason the synthetic conditions were tuned to increase the number of P–N bonds, in line with our previous studies.^48,49 P and N codoped graphene can be synthesized using green chemistry. Cheng et al.³⁶⁷ used low-cost inorganic fertilizers and graphene oxide as precursors to obtain the said nanomaterial. The material presented super electrocatalytic activity and methanol tolerance. In particular, the electrochemical stability was better than those of commercial Pt/C catalysts. Molina-García et al.³⁶⁸ combined codoped graphene and perovskites to build a catalyst for the ORR reaction. Among the codoped graphenes tested, S/N, P/N, and B/N were employed. The lowest overpotential was obtained when perovskites were combined with P and N codoped graphene. Further experimental^369−372 and theoretical³⁷³ studies also corroborated the excellent performance of P and N codoped graphene as a catalyst for the ORR.

As mentioned above, the hydrogen evolution reaction has also been studied using P and N codoped graphene. However, it has received less attention than the ORR. Zheng et al.³⁶¹ demonstrated that the HER activity of P and N codoped graphene is comparable to those of metal catalysts such as gold, molybdenum, and Mo/Ni alloy. However, it is less effective than nanostructured MoS₂/WS₂. It was speculated that, via nanostructure engineering, the graphene based catalysts have the potential to replace all the commercial ones for the HER. Some years later, Hung et al.³⁷⁴ devised a procedure to dope graphene with P and N, paying particular attention to obtaining better crystallinity, conductivity, and elemental functionalities to obtain improved catalytic performance. The amount of P doping was as high as 6 atom %, and the electrode displayed excellent catalytic activity with an increase of 142% in sp² domain size and enormous lowering in overpotential and the Tafel slope, namely 78%. The efficiency of the catalyst was 25% better than that of the MoS_x one. An et al.³⁷⁵ prepared a stable catalyst composed of a Co/Ru alloy and P and N codoped graphene. The catalyst exhibited high activity, a low overpotential of 52 mV at 10 mA/cm², and a Tafel slope of 38 mV/dec.

P and N codoped graphene has been used for methanol electrooxidation. Chen et al.³⁷⁶ decorated the doped graphene sheets with Pd nanoparticles. The nanomaterial delivered a catalytic current density of 11.9 A/cm² in 1 M KOH with 1 M CH₃OH at −0.2 V after a test duration of 3600 s. The aerobic oxidation of alcohols was achieved using Co porphyrins supported on P and N codoped graphene.³⁷⁷ The catalyst converted 92% of the sample and exhibited a high selectivity of 86%, better than most reported photocatalysts. Among other processes that P and N codoped graphene can catalyze, I can highlight the work by Xi et al.,³⁷⁸ which reported the reduction of nitroarenes or the good performance of P and N codoped graphene as a cathode electrode catalyst in microbial fuel cells.³⁷⁹

In the biological field, P and N codoped graphene has been used for cellular imaging.^364,380,381 Ananthanarayanan et al.³⁶⁴ used P and N codoped graphene quantum dots that exhibited high photostability, strong two-photon upconversion, and small molecular weight for real-time tracking of transferring in live cells. Gong et al.³⁸⁰ also used the doped carbon dots for monitoring because of their exceptional fluorescent properties. The material was also utilized for doxorubicin delivery because the effectivity of the drug was much better when combined with the carbon material. Liu et al.³⁸¹ were able to use similar carbon dots for sensitive and selective detection of nitrite in live cells. Finally, Shumba and Nyokong³⁸² modified P and N codoped graphene with Co(II) phthalocyanine to detect H₂O₂; the sensitivity was 12 mA/M and the detection limit was 1.21 nM.

A topic that always is fueled by graphene doped with heteroatoms is that of rechargeable batteries. P and N codoped graphene was successfully used in lithium,^383−390 sodium,^391−394 potassium,³⁹⁵ and Zn–air batteries.^396−398 In 2015, Gu et al.³⁸³ prepared P and N codoped graphene using graphene oxide, triphenylphosphine, and ammonia with polytetrafluoroethylene as a binder. They constructed a membrane which was used as blocking layer to conductively confine polysulfides in the cathodes of lithium–sulfur batteries. The performance of the new battery was remarkable and significantly better than the ones prepared using monodoped graphene. The initial capacity was 1158.3 mAh/g at a current density of 1 C. The cycling stability was reasonable, decaying 0.09% per cycle. Wu et al.³⁸⁴ were able to suppress polysulfide dissolution by physical confinement and chemical interaction; the assembled Li–S batteries had an initial discharge capacity of 1446 mAh/g at a current density of 0.1 C. The capacity decay was extremely low: 0.034% per cycle. Other reports by Zhou et al.,³⁸⁵ Zeng et al.,³⁸⁶ and Zhang et al.³⁸⁷ also utilized P and N codoped graphene to solve the shuttle effect of polysulfides and volume expansion of sulfur that seriously affect performance. A slightly different approach was chosen by Muhammad et al.,³⁸⁸ who prepared P and N codoped graphene microspheres embedded with core–shell CoP@C and MoP@C nanoparticles. The interior carbon shell limited volume evolution and prevented nanoparticle aggregation. As a consequence, excellent lithium storage was achieved.

In an attempt to use sodium instead of lithium, Li et al.³⁹¹ constructed carbonaceous anodes for sodium batteries. P and N codoped graphene was obtained via low temperature phosphidation of NH₂ rich graphene precursor. A large reversible capacity of 330 mAh/g at 50 mA/g was measured. Studies by Wang et al.,³⁹² Qin et al.,³⁹³ and Wu et al.³⁹⁴ were also devoted to producing sodium battery anodes, with great performances. Potassium ion batteries were prepared by Gao et al.³⁹⁵ using P and N codoped graphene aerogels with a specific capacity of 507 mAh/g at 100 mA/g. Zn–air batteries were the focus of several studies that used P and N codoped graphene as an efficient electrocatalyst.^396−398 The 3D P and N codoped holey graphene foam presented excellent activity, showing a half-wave potential of 0.865 V in alkaline electrolytes. The design of supercapacitors is another field that experienced important innovations thanks to the use of P and N codoped graphene.^399−413 Wen et al.³⁹⁹ synthesized P and N codoped graphene monoliths by a facile hydrothermal method that used melamine phosphate as a single precursor. Excellent capacitive performance was obtained. Supercapacitors with an energy density of 8.2 Wh/kg were produced by Xia et al.⁴⁰⁰ when P and N codoped graphene with a high surface area and hierarchical pore structure was utilized. It was prepared by prefunctionalization of graphene and subsequent one-step ammonia phosphate activation. A high specific capacitance of 219 F/g at 0.25 A/g was obtained. Hierarchical porous P and N codoped graphene was also utilized by Zhao et al.;⁴⁰¹ the specific capacitance obtained was 204.4 F/g at 0.2 A/g. In this case, the codoped graphene was produced using nitrogen-containing biomass derived compounds in conjunction with phosphoric acid treatment. The specific capacity retained 97% after 2000 cycles. For more examples about this application of P and N codoped graphene, I refer the readers to refs (399−419). In all of them, the advantages of codoping are demonstrated. Finally, it has been postulated that P and N codoped graphene can be used as a flame retardant and smoke suppressant,^414−416 CO oxidator,⁴¹⁷ O₂ adsorbent,⁴¹⁸ electronic transport material,⁴¹⁹ and oxidation retardant of reduced graphene oxide.⁴²⁰

4.5. Codoped Graphene with Other Main Group Elements

The possibilities to construct codoped graphene systems are almost unlimited as there are plenty of possibilities. In this section I mention some of the systems prepared or studied theoretically. P and B codoped graphene was prepared this year by carbonizing in an Ar atmosphere a cellulose/phosphoric acid supramolecular collosol. Then sodium tetraborate decahydrate was used to adjust the B content, as indicated by Meng et al.⁴²¹ The nanomaterial presented an outstanding catalytic performance for benzyl alcohol oxidation. As early as 2015, I studied theoretically all 3p/2p codoped graphene systems, finding that P and B codoped graphene was very particular because it was one the few examples for which the stability of the ortho and para dispositions of the dopants were almost equally stable. In fact the ortho configuration, i.e., P and B replacing a CC bond, was more stable by 0.04 eV at the M06-L/6-31G* level.⁴⁸ The system is nonmagnetic and a semiconductor with a band gap of 0.2 eV at a 2 atom % doping. In a following article, I showed that when a B dopant is added to P-doped graphene, the effective masses of holes and electrons decrease.

Some studies reported the synthesis of P and O codoped graphene. I discussed above if oxygen can be considered a dopant. In this section I recall that any P-doped graphene produced from graphene oxide will have some amount of residual oxygen, which may be comparable to the levels of doping attained for P-doped graphene.⁶⁰⁻⁶⁹ Ma et al.⁴²² reported the preparation of P and O codoped graphene with exceptional properties as anode materials for potassium ion batteries. The authors prepared the codoped graphene by thermal annealing of graphene oxide with triphenylphosphine. The material had ultralong cycling stability and a capacity of 474 mAh/g at 50 mA/g.

In 2016, Yu et al.⁴²³ utilized P and S codoped hierarchically porous graphene aerogels for enhancing supercapacitor performance. The material was prepared by heating graphene oxide prepared by the modified Hummers method, thioglycoic acid, and pythic acid. The doping level was 5.8 and 4.6 atom % for S and P, respectively. XPS revealed the presence of S–C and P–C bonds. The specific capacitance was 438 F/g at 19 mV/s, greater than the ones measured for the monodoped counterparts. Patel et al.⁴²⁴ prepared P and S codoped graphitic carbons for aerobic oxidation reactions. Further studies indicated that P and S codoped graphene can be a good catalyst of the HER⁴²⁵ and methanol electrooxidation⁴²⁶ and improved capacitance.^427−429

Sulfur and boron doped graphene was synthesized and combined with Au@Pt nanorods. It was utilized as a immunosensing platform for the electrochemical determination of aflatoxin.⁴³⁰ The codoped graphene was prepared using a microwave-assisted hydrothermal approach. Boron trioxide and sodium sulfide were used as heteroatom sources. XPS indicated the presence of C–B and C–S bonds. Theoretical studies characterized SBe,⁴³¹ SiN,⁴³² NAl,⁴³³ and SiP⁴³⁴ codoped graphene, materials which presented outstanding electronic, NO₂ sensing, optical, and catalytic properties, respectively. I note that Al and N codoped graphene has not been synthesized, but when Al-doped graphene was reported, Al–N bonds were observed.^46,47 A facile synthesis for halogen (Cl, Br, and I) and nitrogen codoped graphene was reported by Liu et al.⁴³⁵ The material was utilized as advanced anodes for lithium ion batteries. The Cl and N codoped system presented a specific capacitance of 1200 mAh/g at 0.1 A/g.

Among the theoretical investigations I can highlight my study about graphene with two 3p elements,⁴³⁶ which revealed unexpected properties because the addition of a second dopant decreased the band gaps with respect to the monodoped systems. Ullah and co-workers^437−439 studied BeB, BeN,⁴⁴⁰ and BeO codoped graphene, which exhibited outstanding properties for alkali adsorption thanks to the presence of Be which makes the graphene sheet electron deficient. I studied codoped graphene with one 4p element and one 2p element, which exhibited remarkable properties when a perfect sheet was place above the codoped one because of the formation of interlayer bonds, in particular when Ga and Ge dopants were present.⁴⁴¹ In the same line, Safaei Ardakani et al.⁴⁴² studied theoretically SeX codoped graphene (X = Ga, P, and S), which exhibited outstanding electronic properties. N and Cl dopants were successfully incorporated into the graphene framework as revealed by the different characterization techniques employed.

Finally, again with the help of nitrogen, Se and N codoped graphene was prepared. Chen et al.⁴⁴³ synthesized codoped aerogels that showed a synergistically enhanced capacitive performance. The specific capacity was 302.9 F/g at 1 A/g. In my theoretical study about Se and N codoped graphene,⁴⁴¹ I found that this system has the larger gap among the 4p/2p codoped graphenes: 0.83 eV (spin up) and 0.84 eV (spin down) at the M06-L/6-31G* level for a 4 atom % level of doping. Excellent activity for iodine reduction reaction was calculated by Zhong et al.⁴⁴⁴ for Se and N codoped graphene.

4.6. Transition Metal and Nitrogen Codoped Graphene

Along this review I have highlighted the key role played by nitrogen to synthesize codoped graphene systems. Although in section 2 I mentioned that Mn, Fe, Co, Ni, Ir, and Au metal dopants have been embedded in the graphene framework,⁸⁰⁻⁸⁶ nitrogen has been intensively used to stabilize metals in the graphene sheet.^445−471 Among these systems Fe and N codoped graphene is one of the most studied ones,^445−453 probably because they can be used in single-atom catalysis. In 2015, Dong et al.⁴⁴⁵ demonstrated that the latter system is a very effective catalyst for the ORR. These authors exfoliated graphite using cyclopentadienyl iron, and then upon pyrolysis and ammonia activation, the material was converted into Fe and N codoped graphene. The new catalyst exhibited excellent methanol tolerance, superior to that of commercial Pt/C. Other studies by Zitolo et al.,⁴⁴⁶ Niu et al.,⁴⁴⁷ and Jiang et al.⁴⁴⁸ also supported the strong catalytic power of Fe and N codoped graphene. In particular the study by Zitolo et al.⁴⁴⁶ revealed the existence of porphyrin-like FeN₄C₁₂ moieties. Sibul el al.⁴⁴⁹ found that Fe and N codoped graphene is a superior catalyst for anion exchange membranes rather than proton exchange membranes for fuel cell applications. A maximum power density of 243 mW/cm² was obtained. Continuing with the use of Fe and N codoped graphene in catalysis, Zhang et al.⁴⁵⁰ employed density functional theory to show that by anchoring two Fe atoms with four nitrogen dopants it is possible to obtain an excellent catalyst for the oxygen reduction reaction. In a different area, Gao et al.⁴⁵¹ reported a fluorometric and colorimetric dual-mode sensor based on nitrogen and iron codoped graphene quantum dots for detection of Fe³⁺ ions in biological fluids and cellular imaging. Finally, there is evidence^452,453 indicating the usefulness of Fe and N codoped graphene in the development of lithium–sulfur batteries. Zhang et al. found that the codoped graphene can be used as an anchor material for sulfur in Li–S batteries. Li₂S and Li₂S₂ presented very low decomposition energies on its surface.

Cu and N codoped graphene has been produced by Ni et al.⁴⁵⁴ by the thermal conversion of Cu(II) 2,2′-bipyridine in the confined space of lamellar montmorillonites. The product presented excellent results in terms of ORR catalytic activity and methanol tolerance in alkaline media. Mn and N codoped graphene has been the subject of several investigations.^455−460 Zhu et al.⁴⁵⁵ employed density functional theory to show that this doped graphene is an excellent catalyst for the ORR, while Luo et al.⁴⁵⁶ employed first principles calculations to postulate that it can be a low-cost catalyst for CO oxidation at room temperature. Finally, Lee et al.⁴⁵⁷ prepared Co and N codoped graphene quantum dots used as bimodal resonance and fluorescence imaging nanoprobes.

In 2015, Li et al.⁴⁵⁸ performed a landmark theoretical investigation which proved that Co and N codoped graphene is a superior catalyst for the ORR and OER. It had a high selectivity for the four-electron-reduction pathway. These hypotheses were confirmed by experimental investigations.^459−461 Han et al.⁴⁵⁹ ultrasonicated g-C₃N₄, glucose, and Co(CH₃COOH)₂·4H₂O, and after several processes Co and N codoped carbon sheets were obtained. As usual for codoped carbons, the material had excellent ORR catalytic activity. The overpotential between ORR and OER reactions was 0.80 V at 10 mA/cm². A similar behavior was observed by Liu et al.,⁴⁶⁰ but in this case the Co and N codoped graphene was synthesized using a simple Mg(OH)₂ nanocasting method. In this case, the OER properties were similar to those corresponding to IrO₂. Finally, Du et al.⁴⁶¹ studied the use of Co and N codoped graphene as a single-atom catalyst for high sulfur content lithium–sulfur batteries, a result also supported by the work of Zhang et al.⁴⁶² It is important to note that the latter work studied not only Co and N codoped graphene as an anchor material in Li–S batteries but also the V, Cr, Mn, Fe, Co, Ni, and Cu counterparts. The Co–N–C coordination center served as a bifunctional electrocatalyst that facilitated the formation of Li₂S in discharge and its decomposition in the charge process. Finally, Ir and N codoped graphene which mimicked Ir porphyrins was synthesized by Xiao et al.,⁴⁶³ and it exhibited ORR catalytic activity significantly higher than that of Ir nanoparticles. These properties were attributed to the moderate adsorption energies of the intermediates. The idea of embedding metal atoms on N-doped graphene was extended to include more and different metals. Dual-metal and N-doped graphene has been predicted by theoretical calculations to have outstanding HER activity⁴⁶⁴ and catalytic effects for the CO₂ reduction reaction.⁴⁶²

Other combinations of metals were used to codope graphene include; for example, Fe and S codoped graphene quantum dots were synthesized by Kharangarh et al.⁴⁶⁶ using a facile one-pot hydrothermal method. The material had excellent electrochemical properties and improved electrical conductivity. The specific capacitance was 476.2 F/g, about 3.3 times higher than that corresponding to the undoped material. Gu et al.⁴⁶⁷ reported the preparation of Ni and Al codoped graphene by the reduction of graphene oxide. The product had an impressive hydrogen storage uptake of 5.7 wt %, at 473 K.

4.7. Triple-Atom and Higher Codoped Graphene

The obvious question that some scientists asked was that if two are good why not three or more dopants? I studied XBN tridoped graphene where X = Al, Si, P, or S.⁴⁶⁸ As expected, the XNB motif was preferred because nitrogen stabilizes most dopants. The exception was sulfur, which preferred an SBN motif. In line with my previous findings for two dopants, the theoretical calculations indicated that in general the dopants considered preferred to be bonded instead of separated. The tridoped sheets presented interesting electronic properties and high reactivity. For AlNB, PNB, and SNB the carbon atoms are more reactive than in their AlN, PN, and SN codoped counterparts. However, for SiNB the reactivity is lower than that of SiN dual-doped graphene. As a consequence I recommended that, in order to increase reactivity, Al, P, and S should be combined with BN motifs. There is experimental evidence available indicating that the use of three dopants is beneficial. Razmjooei et al.⁴⁶⁹ prepared P, S, and N triple-doped graphene finding that its ORR activity is 2 times higher than that of S and N codoped graphene and 5 times higher than that of single P-doped graphene. Three dopants were also used by Wang et al.⁴⁷⁰ to capture the harmful bisphenol A, but in this case the dopants utilized were P, S, and N. The same combination of dopants was selected by Wang et al.⁴⁷¹ to design a electrocatalyst of the ORR in alkaline medium. The material presented superior properties which indicated that it is a promising cathode catalyst for alkaline fuel cells. The good performance of P, S, and N triple-doped graphene for the ORR was also confirmed by the investigations of Dou et al.⁴⁷² Last year, Wang et al.⁴⁷³ also prepared P, S, and N codoped graphene and showed that it is an excellent metal-free bifunctional catalyst for superior electrocatalytic oxygen reaction in rechargeable Zn–air batteries. The dopants were introduced by an interesting modified ball-milling process. Zheng et al.⁴⁷⁴ also incorporated P, S, and N codoped graphene as cathodes for Zn–air batteries, but in this case the triple-doped graphene was derived from onium salts. Finally, Xu et al.⁴⁷⁵ reported that P, S, and N tridoped graphene quantum dots are a very interesting ion fluorescence probe. The material was prepared from inexpensive coal, and fluorescence was quenched by Pb²⁺.

Some reports are focused on N, F, and S tridoped graphene^476−479 and N, P, and S tridoped graphene.⁴⁸⁰ However, as I mentioned in the preceding sections, it is highly questionable whether fluorine is a dopant or not. Nevertheless, these materials were useful in catalysis,^476,477,480 solar cells,⁴⁷⁸ and photoluminescence.⁴⁷⁹

B, N, and P tridoped graphene has been synthesized at least in two reports that observed excellent catalytic properties for the ORR.^481,482 Lin et al.⁴⁸¹ synthesized the material using boron phosphate and a B/P source and ammonia as the N-dopant agent. First graphene oxide was combined with boron phosphate to introduce the B and P dopants. After the B and P codoped graphene aerogel was obtained, it was activated with NH₃ atmosphere to obtain the final tridoped graphene. P, N, and O tridoped graphene was reported by Zhao et al.⁴⁸³ and was proposed as a supercapacitor electrode and a metal-free catalyst for the oxygen reduction reaction. The specific capacitance was 426 F/g. Again the question of whether oxygen is a dopant is open because it is hard to evaluate if it is replacing a carbon atom or it is a functional group.

The high stabilization of metal dopants induced by graphene elicited several investigations of tridoped graphene containing two different metals and nitrogen. Hu et al.⁴⁸⁴ performed theoretical calculations which revealed that Fe, Co, and N tridoped graphene can be a superior catalyst of the ORR and OER. An extremely low overpotential of 0.22 V for both reactions was obtained. He and Santiago⁴⁸⁵ combined a variety of metal dimers on N-doped graphene, finding ultrahigh efficiency for the nitrogen reduction reaction. The landmark experimental and theoretical investigation by Zhou et al.⁴⁸⁶ proved that FeNi-N6 sites, where each metal is coordinated to four nitrogen atoms, dominate the catalytic activity of noble metal free catalysts. Excellent methane activation was indicated by the theoretical calculations performed by Wu et al.⁴⁸⁷ for tridoped graphene with two 3d metals and nitrogen.

Fe, S, and N tridoped graphene was obtained by Qiao et al.⁴⁸⁸ and Feng et al.⁴⁸⁹ In both works, it was fond that this tridoped graphene is an excellent catalyst of the ORR. Feng et al.⁴⁸⁹ obtained this material by using melamine and 2-aminotriazole. The addition of melamine increased the number of nitrogen atoms with pyridinic structure, and as a consequence the amount of FeN_x was augmented.

Finally, I mention the study by Molina-García el al.⁴⁹⁰ published in 2018 that achieved the inclusion of four dopants in the graphene sheet: boron, nitrogen, phosphorus, and sulfur. The level of doping was determined via XPS as 6.4, 6.1, 2.6, and 0.5% for B, N, P, and S, respectively. The presence of P–N, P–C, B–C, and S–C bonds was confirmed among many other types. There was improved performance of the quaternary-doped graphene as indicated by the effective number of electron transferred: 3.2.

5. Conclusions and future prospects

In this review, I have presented a thorough analysis of the multiple atom doped graphene systems that have been synthesized and studied theoretically. I first raised the question of when a foreign atom can be considered a dopant. Although I have discussed works that claim to have doped graphene with fluorine, I believe that this atom cannot be considered a dopant but is a functional group. The case of oxygen is more complicated because it may possibly replace a carbon atom in graphene’s framework. However, if oxygen is considered a dopant in all the chemical forms that it can adopt, most of the single-atom graphene systems that have been prepared from graphene oxide are likely to be considered as codoped graphene systems. The amount of oxygen present may be similar to that achieved for the introduced dopant. Also, the effect of this residual oxygen on the properties studied should be clarified. Will it be the same if oxygen is not present?

Regarding the codoped graphene systems synthesized, the list includes (but is not limited to) B/N, N/O, N/F, S/N, P/N, P/B, P/O, S/P, S/B, Si/N, Cl/N, Se/N, TM/N (where TM is a metal), Fe/S, and Al/Ni. By far, the most studied system is S and N codoped graphene, followed by P/N, B/N, and TM/N codoped graphene. From this list of codoped systems, it is crystal clear that nitrogen dominates the list not only in the number of systems but also in the list of articles published because it is present in the four most studied codoped graphene systems. Therefore, the question emerges: are we codoping graphene or doping N-doped graphene? The strong electron-withdrawing properties of nitrogen and its ability to form multiple bonds are crucial for introducing other heteroatoms. Also, nitrogen can be present in various forms—pyridinic, graphitic, and pyrrolic—that can adapt to the needs of the dopant. For example, in B/N doped graphene, it may prefer to be in graphitic form to pair boron. However, for sulfur it can adopt a pyridinic structure, allowing sulfur to be present in thiophenic form CSC (see Figure 4). In the case of transition metals, they can be trapped in an N4 environment, as in porphyrins, anchoring the metal to graphene’s surface and avoiding the undesired metal clustering. Therefore, one of the field’s critical challenges is synthesizing new codoped graphene systems that do not include nitrogen. Finally, the question of what the effect is of the residual oxygen in the codoped graphene systems prepared remains open.

Related to the latter point, it is impossible not to mention the article by Martin Pumera’s research group¹¹ asking if any “crap” that we put in graphene will enhance its catalytic activity. I believe this challenging paper has not received enough attention, and more studies are needed to give a final answer. In my opinion, one of the significant problems in the field is that the large number of methods available to synthesize codoped graphene makes it almost impossible to have a clear structure–property relationship. It is crucial to devise more promising approaches to synthesize codoped graphene systems with a specific disposition of the dopants. This may be attained, for example, by using adequate substrates as I proposed for siligraphene,⁸⁸ or with the aid of specific molecular precursors as in the work by Nguyen et al.⁴⁹¹ that reported the bottom-up synthesis of sulfur-doped graphene nanoribbons.

In this review, I have commented on how the multielemental doped graphene has been utilized in catalysis to develop new energy storage systems, sensing, protection against microwave radiation, piezoelectronics, and solar cells. Even though most of the works report outstanding properties, to the best of my knowledge, there is no widespread use of these nanomaterials in the industry. Maybe it is a bit soon because graphene was synthesized only 18 years ago and the time that doping has been investigated is even shorter. However, to achieve more maturity, this field needs an improved relationship with real-life applications so that society can realize how important graphene is. To that end, in my opinion, theory can be coupled with experiment to improve the synthesis protocols improving the structure–property relationship.

I expect that this review will contribute to introducing more researchers to this fascinating field and enlarge the list of codoped graphene systems that have been synthesized.

The author declares no competing financial interest.