Novel insights into L-cysteine adsorption on transition metal doped graphene: influences of the dopant and the vacancy

Abstract

Exploring potential applications of transition metal (TM) doped graphene in biomolecular adsorption is of fundamental relevance to the area of nanobiotechnology. Herein, we investigated L-cysteine adsorption on first-row transition metal (Sc–Zn) doped single-vacancy and double-vacancy graphenes (MSVs and MDVs) using density functional theory calculations. Three types of upright adsorption configurations, via unprotonated S-end, O-end and N-end functional groups, were considered. All the MSVs chemically adsorb L-cysteine with no regular variation tendency. MDVs show decreasing chemisorption from V to Co, followed by emergence of physisorption from Ni to Zn. L-Cysteine adsorption on MDVs is weaker than that on MSVs, starting from Mn to Zn. Both the TM dopant and the vacancy type contribute to adsorption tendency. In addition, site-specific chemisorption is revealed. The magnetic behaviour of the adsystems is also interesting. In particular, FeSV, ZnSV and NiSV become magnetic after all three end-type adductions. L-Cysteine adsorption induced distribution of the increasing number of 3d electrons and TM–C interactions could account for the magnetism mechanism. Interesting magnetization patterns of MSVs and MDVs occur in most magnetic chemisorbed systems, exhibiting different mirror symmetries. This study could facilitate applications of TM doped graphenes in biosensing, biomolecule immobilization, magnetic bioseparation and other fields in bionanotechnology.

---

1. Introduction

There is growing attention to the development of nanobiotechnology, in which one necessary aspect is to find and utilize nanomaterials that are able to mediate the related processes, including adsorption of biomolecules in the appropriate positions, structures and orientations on specific scaffolds of interest.

Graphene materials possess fascinating physical and chemical properties as well as facile modifiability of the flexible structure. In the previous decades, more and more research efforts have been made to extend the application potential of graphene in nanobiotechnology. Biosensing,^1,2 drug/gene delivery,^3–5 biocatalysis^6,7 and antibacterial materials^8,9 are only a few out of many potential applications that have been brought about by exploiting the two-dimensional, monolayer, and sp²-bonded carbon network. The attachment of biomolecules on graphenes, in appropriate positions, structures and orientations, plays a fundamental role in the synergistic interactions between the biomolecules and graphenes. These interactions are of vital importance in various bioapplications, including adjusting detection sensitivity and stability of biosensors, facilitating biomolecules assembly with selectivity and affinity, tuning catalysis activity, and achieving anti-bacteria function. Theoretical and experimental studies have mainly focused on the adsorption of biomolecules on pristine graphene,^9–14 graphene oxides,^8,15–18 non-metals (mainly B,^19,20 N,^19–21 S^22,23) doped graphenes and metal nanoparticle decorated graphenes,^24–27 with the development of experimental availability of these graphene materials.^28–32

Recently, several studies on the fabrication of transition metal (TM) embedded graphene materials with single-atom precision, using electron beam irradiation and subsequent magnetic sputtering or evaporation of respective metals, have been reported.^33,34 Owing to the high capability of in situ observation and the high capability of tunable processing conditions, TM embedded graphenes with different numbers of vacancies have been fabricated and observed. As uncovered experimentally^33,34 and theoretically,^35–39 both the metal dopant and the vacancy show significant modulation on geometrical, electronic and magnetic features of TM embedded graphenes. On the other hand, it has been demonstrated that doping TM atoms into graphene could effectively tune the molecular adsorption^40–43 and molecular catalysis.⁴⁴ With respect to the interactions between biomolecules and TM doped graphene related study⁴⁵ is very rare. Therefore, intensive work to explore TM doping effects on the interactions between graphene and biomolecules is of critical relevance and highly desirable. To achieve this, we selected a series of first-row TM atoms (from Sc to Zn) doped single-vacancy and double-vacancy graphenes for investigation of biomolecular adsorption. To assure consistent comparability with our previous study,⁴⁵ L-cysteine was still chosen as the probe biomolecule.

In this study, the adsorption of L-cysteine (L-cys) on the single-vacancy and the double-vacancy graphenes doped with the first-row transition metals, namely, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu and Zn, was investigated using density functional theory (DFT) calculations. The main objective was to unveil the effects of the type of transition metal and type of vacancy on the adsorption behavior and magnetism of the L-cys/graphene materials. Proper explanations for adsorption stability and magnetism have also been proposed. Our results could provide a useful reference for application of TM-doped graphenes in biological fields.

2. Modelling and calculation details

2.1 Model building

In the present calculations, transition metal (TM = Sc–Zn) embedded graphenes were built on the model of an optimized 6 × 6 intrinsic graphene sheet, with the lattice parameters being 14.76, 14.76 and 20.00 Å in the a, b and c directions, respectively. By replacing the C atom by a transition metal atom at the centre of the intrinsic graphene supercell, the corresponding TM embedded single-vacancy graphene (MSV, M = Sc–Zn) was obtained. By replacing two C atoms by a transition metal atom at the centre of the intrinsic graphene supercell, the corresponding TM embedded double-vacancy graphene (MDV, M = Sc–Zn) was obtained. The doping concentrations are 1.38% and 1.41% for MSV and MDV, respectively. In addition, the MG (M = Sc–Zn) term is introduced, for simplicity to denote a corresponding MSV and MDV with the same TM dopant.

With respect to the docking geometry of a single L-cys on the MGs, three types of upright adsorption configurations (Fig. 1), i.e., through the unprotonated S-end, O-end or N-end functional group, were considered. The S, O or N atom in L-cys was initially located directly on top of the dopant in MG (M = Sc–Zn) at a distance of 3 Å. The upright adsorption configurations of L-cys, with one head group attached to graphene and the other two extending away towards the vacuum, are of great significance. Upright L-cys could act as an anchoring site, either polymerizing through the free amino or carboxyl groups or incorporating metal ions through any of the three free functional groups.^46,47 For convenience, the S-end, O-end and N-end L-cys/graphene adducts were labelled as S-MSV, O-MSV, N-MSV and S-MDV, O-MDV, N-MDV for MSV and MDV, respectively. Optimized adsorption configurations were obtained through full relaxation.

Fig. 1 Structure illustration of L-cysteine (a), unprotonated S-end (b), O-end (c) and N-end (d) L-cys adsorption on MGs before DFT calculations.

2.2 Calculation parameters

Dispersion-corrected density functional theory (DFT-D) calculations were carried out using the DMol³ package in Materials Studio.⁴⁸ The Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional of the GGA-type was employed, with long-range dispersion correction via Grimme's scheme.⁴⁹ The DFT-D approach was applicable to the present systems, as validated in our previous study.⁴⁵ The double numerical basis set augmented with polarization function (DNP) was chosen as the basis set. The DFT semicore pseudopotentials (DSPP) core treatment was applied.

Structural optimization was carried out on all systems without any constraint, with convergence tolerances of 1.0 × 10⁻⁵ hartree on energy, 1.0 × 10⁻³ hartree Å⁻¹ on gradient and 1.0 × 10⁻⁵ Å on displacement. The SCF convergence criterion of 1.0 × 10⁻⁶ hartree on energy was achieved with respect to the 6 × 6 × 1 k-point sampling over the Brillouin zone. Magnetic moments were computed using the Mulliken analysis method. The isolated L-cys molecule, MG sheet and L-cys/MG adsystems were calculated, without any constraint, under the same parameters to ensure the comparability of the results.

2.3 Binding energy

The binding energy, E_b, of the substitutional transition metal atom in the single-vacancy or double-vacancy graphene was calculated³⁵ as

E_b = E_MSV − E_TM − E_SV (1)

E_b = E_MDV − E_TM − E_DV (2)

where E_MSV and E_MDV denote the total energy of the optimized MSV and MDV, E_TM the total energy of an isolated TM atom, and E_SV and E_DV the total energy of the reconstructed single-vacancy and double-vacancy graphene, respectively.

To evaluate the adsorption stability of L-cys on MG systems, the binding energy E_b was calculated^50,51 as follows:

E_b = E_tot + 1/2E_H2 − E_MG − E_L-cys (3)

where E_tot denotes the total energy of the optimized L-cys/MG adduct, E_MG the total energy of MG, including MSV and MDV, E_L-cys the total energy of an L-cys molecule, and E_H2 the total energy of a hydrogen molecule. Unprotonated cysteine is not the most stable species in the gas phase. We assume that the hydrogen atoms that dissociated from the cysteine formed molecules. For that reason, we consider the total energy E_H2 of the gas-phase H₂ molecule.^50,51 A more negative value of E_b means a more stable structure and a stronger binding energy.

3. Results and discussion

In the following sections, geometrical optimizations of MSVs and MDVs (M = Sc–Zn) were first performed, providing initial structures and magnetic characteristics for subsequent investigation of L-cys adsorption. Then, the adsorption stability of L-cys on MSVs and MDVs was evaluated by discussing E_b values, charge density distributions and geometrical structures. An adsorption mechanism was proposed. Finally, the magnetism of the L-cys/MG adducts was presented, by analyses of spin density distribution and Mulliken spin population, followed by analyses of the magnetism mechanism. Magnetization symmetry was also exhibited.

3.1 The main properties of MGs

Typical optimized configurations of MSVs and MDVs are shown in Fig. 2. The binding energies of the MSV series (Fig. 3(a)) for TM doping are in the range from −12 to −2 eV. The change behaviour is non-monotonic. Ti [3d²4s²] presents a stronger local binding energy, with the four valence electrons saturated.³⁵ Weaker local binding energies among the 3d transition metals are found for Cr and Mn. For Cu and Zn, with almost full d-shells, much weaker binding energies are observed. The E_b values of the MDV series show a similar non-monotonic trend from Sc to Zn. These non-monotonic behaviours are related to the simultaneous energy down-shift and compression of the 3d shell of the metal as we increase the atomic number.³⁹

Fig. 2 Typical atomic configurations of relaxed MSV: side view (a), top view (b) and MDV: side view (c), top view (d). The h denotes the elevation of the TM dopant, relative to the average height of carbon atoms.

Fig. 3 (a) The binding energy E_b, (b) elevation h of the TM dopant, (c) average TM–C bond length and (d) magnetic moment (MM) of MSVs and MDVs. L_COV in (c) indicates the sum of the covalent radii of TM and C.

With respect to specific geometrical changes, the TM atom of MSVs is displaced outwards from the graphene surface owing to a larger atomic radius relative to carbon (Fig. 2(a)). The elevation trend of the dopant is shown in Fig. 3(b). Neighbouring carbon atoms are also protruded outwards to facilitate the formation of three transition metal–carbon (TM–C) bonds. The average TM–C bond lengths (Fig. 3(c)) are all below the sums of the covalent radii⁵² of TM (from Sc to Zn) and C, which is consistent with strong chemical binding. Moreover, the average TM–C bond length decreases from Sc to Fe as the atomic size increases, and then it goes up as the bonding becomes weaker. In most cases, the transition metal atom forms three equivalent TM–C bonds with neighbouring carbon atoms (Fig. 2(b)) for TM = Sc–Ni. CuSV and ZnSV suffer Jahn–Teller-like distortion, with the lengths of the three Cu–C bonds and the three Zn–C bonds as 1.87, 1.89, 1.87 Å and 1.89, 2.02, 1.89 Å, respectively. The behaviour of the MDVs is simple. All the MDVs form the “cross” configuration (Fig. 2(d)), with the TM dopant equally bonded to four neighbouring carbon atoms. Due to the combined effect of atomic size and bonding strength, the TM dopant is also elevated and the covalent TM–C bond length is also changed (Fig. 3(b and c)). For CuG and ZnG, the introduction of bigger vacancies leads to stronger binding ability, due to the Jahn–Teller-like effect in the single-vacancy case.

The magnetic properties are presented in Fig. 3(d) and Table S1.† The MGs (M = V, Cr, and Mn) and CoDV are magnetic with dominant 3d magnetism. MSVs (M = Co, Cu) are magnetic with large contribution of carbon atoms in graphene. These calculated results of MSVs are in good accordance with previous findings.^35,39 Mild discrepancies in the MDV systems appear under our calculation scheme. FeDV shows no magnetism, which is different from the magnetic moments of ∼3.2 μ_B in ref. 35.

3.2 Adsorption stability of the L-cys/MG adducts

3.2.1 Adsorption stability of the L-cys/MSV adducts. The binding energies of the L-cys/MSV adducts (Fig. 4) are all below −0.75 eV, suggesting that all types of unprotonated L-cys are chemically adsorbed on the MSV substrate. No general variation trend appears in the values of E_b along the TM series from Sc to Zn. Fig. S1 of ESI† shows the typical charge density graphs. Large overlap in charge density is observed, indicating existence of stable adsorptions. Comparison of the E_b values of the L-cys/MSV adducts reveals that different types of MSVs exhibit distinct adsorption affinities towards each functional group (Fig. S1, ESI†). Cr–Co and Zn doped single-vacancy graphenes show the best affinity towards the S-end L-cys radical, whereas Sc, Ti, V, Ni and Cu doped single-vacancy graphenes show the best adsorption stability towards the O-end radical. The N-end adduction is always the weakest adsorption type for all systems.

Fig. 4 E_b of L-cys/MSVs and L-cys/MDVs with different end-types of L-cys adsorbed.

The MSV substrates endure strain relaxation to contribute to the strong binding (Fig. 5(a–d)). On one hand, the positions of the TM dopant (Fig. 6(a)) and neighbouring carbon atoms change obviously. The lengths of the average TM–C bonds vary (Fig. 6(b)) for different types of transition metals. Concerning ZnSV, the average length of the Zn–C bond (2.28–2.30 Å) is surprisingly much higher than the sum of the covalent radii (1.98 Å). Further investigation on geometrical structures reveals intensified Jahn–Teller distortion. Two Zn–C bonds are broken and only one Zn–C binding remains to adsorb L-cys (Fig. 5(d)). Moreover, the Zn dopant is largely elevated out. In contrast, most of the bond lengths and bond angles of L-cys are not obviously changed after adsorption, except for the angles related to adsorbed functional groups, which are greatly changed.

Fig. 5 Typical adsorption geometries of L-cys on MGs to illustrate S-end adsorption with a single TM–S bond (a), O-end adsorption with two TM–O bonds (TM–OH and TM–O=) (b), N-end adsorption with a single TM–N bond (c), special adsorption structures of the S–ZnSV (d), S–CuDV (e) adducts, and O-end adsorption with a single TM–O bond (O in –OH) (f) on MDV.

Fig. 6 (a) The elevation h of the TM dopant and (b) the average TM–C bond length in MSVs. L_COV indicates the sum of the covalent radii of TM and C.

For all L-cys/MSV adducts through the S-end chemisorption, the covalent TM–S bond forms (Fig. 5(a)), with the free –COOH and –NH₂ groups extending outwards. The bonding distance between L-cys and the graphene substrate (G–L) monotonously decreases from ScSV to CoSV and then goes up to ZnSV (Fig. 7(a)). For O-end adsorption, two-site adsorption modes (Fig. 5(b)) exist for all the types of MSVs (Table S2†). The –COOH group becomes very close to the graphene surface, with the C–O bond shortened to 1.28 Å and the C–O= bond lengthened to 1.28 Å. Both O atoms bind the metal dopant. The two TM–O bonds are nearly equal in length, and the average value varies in the range of 2.00–2.20 Å with the change of the type of TM dopant (Fig. 7(b)). A large steric hindrance, arising from the two TM–O bonds, exists to maintain the upright adsorption configuration robust. It is noted that starting from CrSV, the average bond length is larger than the sum of the covalent radii of the corresponding linking atoms, which is due to the combined effects of the bi-O (denoting two-site O-end adsorption) adsorption mode and the decreasing adsorption ability. Regarding N-end adduction, one-site adsorption through the formation of M–N covalent bond exists for most MSVs, with the free –COOH and –SH extending outwards (Fig. 5(c)). The bonding distance between L-cys and the graphene substrate basically monotonously decreases from ScSV to CuSV, with the exception of Fe (Fig. 7(c)).

Fig. 7 The G–L bonding length through the (a) S-, (b) O- and (c) N-end docking. L′_COV indicates the sum of the covalent radii of TM and linking atom in L-cys.

3.2.2 Adsorption stability of the L-cys/MDV adducts. With respect to L-cys docking on MDVs, the E_b values (Fig. 4) reveal an interesting transition from negative to positive along the TM series from Sc to Zn. For L-cys docking on MDV (M = Sc–Co), all related E_b values are less than zero, showing strong chemisorption towards the three types of L-cys radicals. Moreover, NiDV shows strong chemisorption towards the S-end L-cys (−0.82 eV), weaker chemisorption towards the O-end L-cys (−0.20 eV) and physisorption towards the N-end L-cys (0.12 eV). Regarding CuDV, only the S-end immobilization is chemically stable (−0.22 eV), whereas the other two types of L-cys radicals are physisorbed, with positive values of E_b. ZnDV shows physical adsorption towards all three types of L-cys radicals.

The weakest binding energies (∼1 eV) for O-end and N-end adsorption among the metals studied herein appear for Cu with filled d electronic shells and half-filled s electronic shells. However, for Zn, with filled d electronic shells and filled s electronic shells, the O-end and N-end physisorptions are stronger than those of Cu. The observation may be ascribed to the distinct activity of Cu relative to Zn under specific configurations. With respect to S–CuDV, one of the four carbon atoms around the Cu dopant is protruding and a S–C covalent bond (1.87 Å) is formed (Fig. 5(e)). The finding of an S–C bond in S–CuDV also suggests that neighbouring carbon atoms should be active⁵³ in molecular adsorption. More investigation on the chemical activity of neighbouring carbon atoms around the TM dopant is being undertaken.

In the case of adsorption on MDVs (M = Sc–Cu), charge density distributions (Fig. S2, ESI†) are similar to those of the MSVs cases. Much overlap appears for all adsorption configurations, showing stable adsorption for both chemisorption and physisorption. Compared to our results in ref. 45, no overlap exists for inert adsorption of L-cys on pristine graphene, thus the overlap herein indicates more reactivity for TM doped double-vacancy graphenes. This also suggests sensing potential of NiDV, CuDV and ZnDV, taking advantage of the relatively stable physisorption. Moreover, the interesting findings would provide a meaningful theoretical reference in practical applications about selecting appropriate dopants for physisorption or chemisorption towards adsorbates.

The geometrical distortions are gentler compared to the MSVs cases. The graphene substrate undergoes mild geometrical distortions. Metals are elevated and neighbouring carbon atoms are mildly protruded by the L-cys (Fig. 8). Structural changes in L-cys are not obvious.

Fig. 8 (a) The elevation h of the TM dopant and (b) the average TM–C bond length in MDVs. L_COV indicates the sum of the covalent radii of TM and C.

For the S-end adduction, most adsystems show one-site adsorption, with a single TM–S binding (Fig. 5(a)), while two-site adsorption occurs on ScDV with one Sc–O= bond formation in addition to the Sc–S bond (Fig. S3, ESI†). The TM–S binding distance is shorter than that of the corresponding MSV cases from TiDV to NiDV, while it is larger for ScDV, CuDV and ZnDV (Fig. 7(a)).

For the O-end adduction, the bi-O adsorption mode (Fig. 5(b)) is favoured by the Sc, Ti, Mn, Fe, Co, and Ni doped double-vacancy graphenes, whereas VDV and CrDV exhibit mono-O (binding through O in OH) adsorption (Fig. 5(f)). Owing to the single site bearing the whole of the chemisorption, the formed TM–O bond is much shorter than that under the bi-O mode (Fig. 7(b)) compared to other types of MDVs. The physisorbed systems of CuDV and ZnDV also exhibit mono-O adsorption, with the Cu–O and Zn–O bonds greatly larger than the sum of the covalent radii. The TM–O binding distance is shorter than that of the corresponding MSV cases from ScDV to CrDV but larger than that from MnDV to ZnDV (Fig. 7(b)).

With respect to the N-end adduction, all adsystems show one-site adsorption with merely TM–N binding (Fig. 5(c)). N–NiDV, which is on the border between chemisorption and physisorption, has an E_b value of 0.12 eV. The N–Ni bond is 1.84 Å, a little longer than the theoretical covalent bond, which is consistent with the weak adsorption. The TM–N binding distance is shorter than that of the corresponding MSV cases from ScDV to CoDV but larger than that from NiDV to ZnDV (Fig. 7(c)).

3.2.3 Site-specific adsorption phenomena. Herein, site-specific chemisorption (Table 1) appears by analyses of the most stable adsorption configurations. In the situation of the single-vacancy graphenes, the S-end L-cys docking on Cr–Co and Zn doped graphenes is stronger than the –COOH or –NH₂ docking, whereas the O-end L-cys shows the best stability on Sc, Ti, V, Ni and Cu doped single-vacancy graphenes. Under the situation of double-vacancy graphenes, the S-end L-cys docking on V–Zn doped graphenes is stronger than the –COOH or –NH₂ docking, whereas the O-end L-cys shows the best stability on Sc and Ti doped double-vacancy graphenes, i.e., for CrG–CoG and ZnG, S-end immobilization shows the most affinity regardless of the vacancy type; for ScG and TiG, O-end immobilization shows the most affinity regardless of the vacancy type; and VG, NiG, CuG's adsorption affinity is easily affected by the vacancy. Moreover, the capability of docking the N-end L-cys to the graphene surface is weakest in all L-cys/graphene systems. It is consistent with the fact that exploiting the amino group to bind L-cys to the metal surface is plausibly hindered in most cases wherein L-cys is the anchoring group of proteins to surfaces, because the NH₂ fragment is engaged in peptide bonds along the primary chain.⁵¹

Table 1 The most stable adsorption on MSVs and MDVs

	Sc	Ti	V	Cr	Mn	Fe	Co	Ni	Cu	Zn
MSV	O	O	O	S	S	S	S	O	O	S
MDV	O	O	S	S	S	S	S	S	S	S

---

Therefore, site-specific chemisorptions, including site-type-specific and site-number-specific chemisorptions, could be realized by tuning the type of metal dopant and type of vacancy in the graphene substrate. The interesting findings could provide references for nanobiomedicine fields, nanocrystal growth, decoration of graphenes and biocatalysis.

3.2.4 Adsorption mechanism analyses. The adsorption mechanism could be understood from the aspects as follows: (i) the TM dopants, which possess different numbers of outer-shell 3d electrons and show different hybridization preferences or (ii) the vacancy types, which contribute different TM–C interactions, including different hybridization environments and different bond lengths between TM atoms and neighbouring carbon atoms.

The role of the TM dopant could be firstly deduced from L-cys adsorptions on MDVs, which show a decreasing tendency from V to Zn. The starting minima of VDV result from formation of saturated bonds using all five valence electrons of V (3d³4s²). Along with the increase of the number of 3d electrons starting from V, redundancy in electrons appears. The 3d shell becomes gradually occupied and moves to lower energies. The hybridization near the Fermi energy (E_F) is reduced (Fig. S4, ESI†). Moreover, few changes happen in the space nearby. This causes increase in electron repulsion and thus causes decrease in binding interactions and even the occurrence of physisorptions.

Another role of the TM dopant is to influence TM–C interactions, illustrated together with the role of vacancy types. The TM–C interactions involve the TM–C bonding distance and the TM's hybridization environment preferences. For TM = Sc–Cr, although shorter TM–C bonding distances occur in the case of MSVs than of MDVs (Fig. S5, ESI†), different preferences to the SV or DV coordination environment lead to no obvious variation tendency of E_b values. Thus, for TM = Sc–Cr, the contribution of TM's preferences to hybridization environments is larger than the interactions represented by the bonding distance. For TM = Mn–Ni, L-cys chemisorption on MDVs is generally weaker, accompanied with longer TM–C bonding distances (Fig. S5†), compared to the corresponding chemisorption on MSVs. This also means weaker roles of coordination environments preferences compared to the role of the bonding distance, relative to TM = Sc–Cr. For TM = Cu–Zn, coordination environments play a key role, because the average TM–C bonding distance (Fig. S5†) of the MSVs are comparable (for Cu) to or much larger (for Zn) than that of the MDVs. Moreover, in the case of the single-vacancy environment, chemisorptions occur with intensified Jahn–Teller distortion in the hybridization environment, even including two broken Zn–C bonds. In the case of the double-vacancy environment, the metals maintain their four TM–C bonds and show physisorptions towards L-cys.

It is noted that adsorption on MDVs shows a regular variation tendency from VDV, whereas adsorption on MSVs shows no regular variation tendency from VSV. This is attributed to different TM–C interactions. In MSVs, strong TM–C interactions, contributed by the short TM–C bonding length coupled with hybridization preferences, outperform the role of TM's outer-shell 3d electron redundancy. The complicated TM–C interactions lead to no regular variation trend in the E_b values along the TM series from Sc to Zn.

3.3 Magnetism analyses

3.3.1 Magnetic properties of L-cys/MSV adducts. The magnetic moments of the L-cys/MSV adsystems, along the TM series from Sc to Zn, are presented in Fig. 9. For Sc and Ti, no change in magnetism appears for the L-cys/ScSV and L-cys/TiSV adducts after any of the three types of end-group adsorption. The original spin polarization of L-cys radicals are quenched to zero, which is attributed to loss of unpaired electrons under strong chemical interactions.

Fig. 9 The magnetic moment (MM) of the (a) MSV and L-cys/MSV adsystems and (b) MDV and L-cys/MDV adsystems.

For VSV, CrSV and MnSV, in which the magnetic moments predominantly come from spin polarization of 3d orbitals, L-cys adsorption induced magnetic variation is illustrated as follows. The disappearance of magnetism of VSV happens after L-cys immobilization regardless of the docking site. The initial spin moment of 2.00 μ_B of CrSV is decreased to 1.77 μ_B, 1.78 μ_B and 1.37 μ_B for the S, O, and N-end adduction, respectively. The spin density distribution (O–CrSV as an example in Fig. 10(a)) shows that most of the magnetization still locates on the TM dopant, arising from spin polarization of 3d orbitals. The carbon atoms present ferromagnetic coupling within separate sublattices and antiferromagnetic coupling between sublattices, suggesting sublattice symmetry breaking. More reduction occurs in the MnSV systems. The magnetic moments reduce from 2.99 μ_B to 1.98 μ_B, 2.00 μ_B and 2.00 μ_B. Both 3d magnetism of Mn and 2p_z spin polarization of the neighbouring carbon atoms contribute to the magnetization. The linking atoms S, N, and O in O= or OH also show rather mild magnetism.

Fig. 10 Schematic side views of spin density distribution of O–CrSV (a), O–FeSV (b), S–NiSV (c) and N–ZnSV (d). The isovalue is 0.01 e ų. The dark blue and light yellow surfaces correspond to majority and minority spins, respectively.

Interesting changes happen in the L-cys/FeSV adsystems. For FeSV, the S–FeSV and N–FeSV adducts remain non-magnetic, whereas the O–FeSV adduct becomes magnetized, with a magnetic moment of 1.00 μ_B per unit cell. The magnetic moment mainly locates on Fe (1.11 μ_B), indicating awakening of the 3d magnetism (Fig. 10(b)) of Fe by the O-end adsorption. The spin moments of O atoms in O= and OH are −0.01 μ_B and −0.01 μ_B, respectively. The other atoms in L-cys exhibit no spin moment.

We put the analyses of the late transition metals Co, Ni, Cu and Zn doped single-vacancy graphenes together, because the magnetic moment oscillates between 0 and 1.00 μ_B. After L-cys adsorption, CoSV shows completely quenching off in magnetism from 1.00 μ_B to 0 μ_B. In the L-cys/NiSV adsystems, completely switching on of magnetism up to 1.00 μ_B, 0.88 μ_B and 1.00 μ_B per unit cell exists (Fig. 10(c)), with 0.11 μ_B, 0.03 μ_B and 0.13 μ_B coming from Ni. The linking atoms S, O, N are 0.07 μ_B, 0.00 μ_B, and 0.01 μ_B, respectively, whereas other atoms in L-cys contribute no spin polarization. In the CuSV adsystems, L-cys immobilization quenches the magnetism to zero for all three types of adsorption. For ZnSV, N-end adsorption induces switching on of the magnetism up to 0.95 μ_B (Zn: 0.01 μ_B) per unit cell (Fig. 10(d)), whereas the S- and O-end adsorptions lead to no influence on magnetism. For the abovementioned magnetic adsystems, the spin moment mainly comes from graphenic carbon atoms owing to spin polarization of 2p_z orbitals, vividly depicted by the spin density distribution graphs in Fig. 10(c and d). The linking atoms S, O in O= or O in OH and N also exhibit magnetism.

3.3.2 Magnetic properties of L-cys/MDV adducts. With respect to L-cys docking on MDVs, interesting transition from chemisorption to physisorption happens along the TM series from Sc to Zn. Therefore, we first discuss the cases of MDVs (M = Sc–Co), which show chemisorption towards all three types of L-cys radicals. ScDV and TiDV remains non-magnetic whether L-cys is adsorbed or not. The magnetic VDV system changes into non-magnetic after L-cys docking, regardless of the end type. Reduction in magnetism happens after L-cys docking on CrDV and MnDV, with most of the spin moment still populated on Cr and Mn (Fig. 11(a)), respectively. FeDV remains non-magnetic for all three types of adsorption. Regarding magnetic CoDV, possessing a magnetic moment of 1.00 μ_B, the magnetic moment decreases drastically and nearly diminishes showing 0.19 μ_B, 0.12 μ_B and 0.08 μ_B for the S-end, O-end and N-end dockings, respectively. The magnetic moment mainly comes from negligible 3d spin polarization of Co (Fig. 11(b)).

Fig. 11 Schematic side views of spin density distribution of the O–MnDV (a), S–CoDV (b), N–CuDV (c) and N–ZnDV (d) adducts. The isovalue is 0.01 e ų. The dark blue and light yellow surfaces correspond to majority and minority spins, respectively.

Furthermore, we go to magnetic properties of the L-cys/MDV (M = Ni, Cu, Zn) adsystems. Regarding NiDV, the S-end and O-end chemical adductions could not change the non-magnetic feature. The magnetic moments of the L-cys radicals are quenched to zero due to loss of unpaired electrons induced by chemisorption. For the N-end adduction, although it shows physical adsorption, the system is still non-magnetic, with disappearance of spin polarization of L-cys. Concerning CuDV, the S-end and O-end chemical adductions induce no magnetism. The N-end physical adduction retains the magnetism of L-cys radicals (Fig. 11(c)), with a magnetic moment of 0.87 μ_B per unit cell. With respect to the non-magnetic ZnDV, the N-end adduction (Fig. 11(d)) induces a magnetic moment of 0.71 μ_B per unit cell, whereas the other two adsorption modes have no influence on magnetism.

3.3.3 Magnetism mechanism analyses. In brief, the variation in magnetism could be understood from L-cysteine adsorption induced distribution of the increasing 3d electron and TM–C interactions: (i) non-bonding 3d electrons for L-cys/MG (M = V, Cr, Mn) adsystems; and (ii) TM–C interactions, influenced by more 3d electrons and the vacancy, for L-cys/MG (M = Co, Ni, Cu and Zn) adsystems.

It is instructive to start the analysis with the VSV complex. The V atom has five valence electrons. One can assume that three of them go into the V–C covalent bonds and one goes into the V–L-cys covalent bond, depicted with double dots in Fig. 12(a), while the fifth (depicted as the circle) replaces the electron of the missing carbon atom. This yields four σ bonds and one π (slightly out-of-plane) bond. Five valences are saturated, the spin is zero, and the binding is strong. In this case, the E_b is the largest among all TM atoms. Cr and Mn have 1 and 2 extra electrons, respectively, which go to separate nonbonding orbitals [red arrows in Fig. 12(a)] and obtain ∼1 and ∼2 μ_B, in agreement with the simple counting rules. With respect to ScSV and TiSV, exhibiting full occupation of bonding orbitals, no change in magnetism happens.

Fig. 12 Schematic of the electronic structure of L-cys/MnSV (a) and L-cys/MnDV (b) complexes. Herein, the numbers “1/3” symbolically represent ligand contribution to the approximate π bond, because one π electron is shared among three π bonds.

For CoSV, NiSV, CuSV and ZnSV adsystems, the TM atom has more than 5 electrons so that some of the nonbonding orbitals are filled. The Coulomb interaction obtains a shift of the impurity states into the conduction band, wherein TM–C hybridization with a large component of graphenic carbon atoms occurs. For L-cys/CoSV and L-cys/CuSV, non-bonding 3d orbitals and TM–C hybridization levels are filled, in agreement with even number of electrons. For L-cys/NiSV and L-cys/ZnSV, the structural strain induced by covalent bonding of L-cys influences TM–C hybridization, inducing splitting of levels and switching on of magnetism. This is similar to curvature induced magnetism on Ni doped carbon nanomaterials^54,55 and other types of metal doped graphenes.⁵⁶ To the best of our knowledge, easy determination and control of the local curvature at the location of the dopant Ni in experiments remains a challenge. Herein, our interesting finding demonstrates that controlling cys adsorption on NiSV or ZnSV through different end groups could effectively tune the magnetism and magnetization distribution of the systems, presenting a facile and easy method in experiments. Fe and Zn doped single-vacancy graphenes could also work in specific end-type magnetic applications.

For MDV adsystems, there are five local σ bonds (Fig. 12(b)). The π TM–C interaction is weaker, due to a larger separation between TM and carbon atoms. The π electrons on the dangling bond atoms have opposite spins, deduced from the naked DV graphene, just as they are on different sublattices. In this case, V is still the strongest bonding. L-cys/CrDV and L-cys/MnDV show 3d magnetisms of 1 and 2 μ_B, attributed to 1 and 2 extra electrons, respectively. For L-cys/CoDV, the initial 3d magnetism of CoDV disappears due to formation of a covalent bond with L-cys. For NiDV, CuDV and ZnDV, weak TM–C interactions induce weak chemisorptions and physisorptions towards L-cys. Only physisorption systems of N–CuDV and N–ZnDV show retainment of L-cys radical magnetism.

With respect to FeG systems, the magnetism is sensitive to the complicated competition between intra-atomic interactions (coulomb repulsion and exchange within the 3d levels) and TM–C hybridization. The magnetism could be switched on owing to sensitive perturbation in O–FeSV. The other systems, L-cys/FeSV and L-cys/FeDV, are non-magnetic.

3.3.4 Magnetization symmetry of MG in L-cys/MG adducts. Determination of a regular magnetization pattern on a substrate could provide a good way to design a substrate with a predefined or controlled magnetization or to provide a substrate with appropriate active sites of catalysis. Regarding spin density distributions on magnetic MSV (M = V, Cr, Mn, Fe, Co and Cu), they show a C_3v similar symmetry (Fig. 13(a)). As for spin density distributions on magnetic MDV (M = V, Cr, Mn and Co), they initially show C_2v symmetry (Fig. 13(e)). The C_2v symmetry contains two mirror planes. One mirror plane is across the respective geometrical centres of the two 5-member rings and the metal dopant. The other mirror plane is across the respective geometrical centres of the two 7-member rings and the metal dopant.

Fig. 13 Schematic top views of spin density distribution on MG in MnSV (a), S–MnSV (b), O–MnSV (c), N–MnSV (d) and MnDV (e), S–MnDV (f), O–MnDV (g), N–MnDV (h). The isovalue is 0.01 e ų. The L-cys atoms in (b–d) and (f–h) were removed to illustrate spin density distribution on the MG surface. The light yellow and dark blue surfaces correspond to majority and minority spins, respectively. The dashed line indicates the mirror plane.

The spin density distribution on MSVs exhibits interesting patterns under different adsorption configurations (Fig. 13(b–d) and S6–S8, ESI†). The spin polarization distributions on CrSV in the O–CrSV adduct, MnSV in the O–MnSV and N–MnSV adducts, FeSV in the O–FeSV adduct and NiSV in the O–NiSV and N–NiSV adducts show the same mirror symmetry. The mirror plane is along one armchair chain containing the metal dopant, which is perpendicular to one lattice vector (Fig. 13(c and d)). In the N–CrSV, S–MnSV, S–NiSV, and N–ZnSV adducts, the spin polarization on CrSV, MnSV, NiSV and ZnSV shows mirror symmetry as well. However, this mirror plane is along the longest armchair chain containing the metal dopant in the unit cell (Fig. 13(b)). Magnetization of CrSV in the S–CrSV adduct is nearly symmetric (Fig. S7(b), ESI†). It is noted that we have unveiled switching on of magnetism and magnetization symmetry in L-cys/NiSV adducts in our previous study,⁴⁵ wherein Ni was initially doped inside the plane of graphene. The same phenomena found herein further demonstrate the stability of such magnetization and magnetization symmetry.

Furthermore, we go to magnetism symmetry of the L-cys/MnDV and L-cys/CrDV systems, presenting chemisorption and showing strong magnetism. Only the O–MnDV and O–CrDV adducts retain the initial C_2v symmetry, accompanied with two mirror planes (Fig. 13(g) and S7(g) of ESI†). For the N–MnDV adduct, the mirror plane is across the respective geometrical centres of the two 7-member rings and the metal dopant (Fig. 13(h)). Magnetization of the doped graphene sheet in the S–CrDV and N–CrDV adducts are almost symmetric (Fig. S7(f and h), ESI†). The S–MnDV adduct shows no symmetry of magnetization (Fig. 13(f)). Spin values by the Mulliken spin population method are illustrated in Fig. S8 of ESI† for MnG in L-cys/MnG adsystems.

4. Conclusions

To advance application potentials of TM doped graphene in the growing field of bionanotechnology, our study investigates influences of the TM type and vacancy type on the interactions between L-cys and TM doped graphenes, including TM (Sc–Zn) doped single-vacancy and double-vacancy graphenes, using the DFT-D approach. This study could provide atomic-scale insights into applications of TM doped graphenes in biosensing, biomolecule immobilization, magnetic bio-separation and other bionanotechnological fields. The main conclusions are summarized as follows.

(i) TM doping atoms could significantly change the chemical reactivity of graphene and lead to strong physical or chemical interactions towards L-cys. Moreover, site-specific adsorption by L-cys is exhibited.

(ii) The binding energies for L-cys adsorption on MSVs show no obvious variation tendency, while a decreasing tendency from V to Zn emerges for adsorption on MDVs, with emergence of physisorption for the Ni, Cu and Zn doped double-vacancy graphenes. From Mn to Zn, L-cys adsorption on the MDVs is weaker than the corresponding adsorption on MSVs. The adsorption mechanism could be understood through combined analyses of the metal dopants and the vacancy types.

(iii) The L-cys/MG adducts show interesting magnetic properties after L-cys chemisorption. Weakened spin polarization appears on CrG, MnG and CoDV. NiSV could be tuned to be magnetic regardless of the end sites, whereas FeSV and ZnSV could be tuned to be magnetically selective towards the O-end and the N-end adduction, respectively. The magnetism is influenced by distribution of the increasing number of 3d electrons and TM–C interactions after L-cys adsorption. Most of the magnetization of the chemisorbed systems show mirror symmetry.

On the basis of the abovementioned conclusions, specific applications of TM doped graphenes could be expected. Cu, Ni and Zn doped double-vacancy graphenes may act as good sensor platforms for L-cysteine. MSVs (M = Sc–Zn) and MDVs (M = Sc–Co) could act as good immobilization platforms for L-cysteine. Through tuning the transition metal dopant and the vacancy type in doped graphene, site-specific immobilization could be realized. In addition, if magnetic bio-separation or delivery is required, Mn, Cr doped graphenes and Ni doped single-vacancy graphenes could make the process facile. Fe and Zn doped single-vacancy graphenes could work in specific end-type magnetic applications.

Acknowledgements

This study has been supported by the National Natural Science Foundation of China (Grant No. 51432008, U1435202, 51572223), the “111” Project (Grant No. B08040), the Doctorate Foundation of Northwestern Polytechnical University (Grant No. CX201431). The Centre for High Performance Computing of Northwestern Polytechnical University in Xi'an, China, wherein the computational work was carried out, is gratefully acknowledged.

References

1. M. S. Artiles, C. S. Rout and T. S. Fisher, Adv. Drug Delivery Rev., 2011, 63, 1352–1360.
2. G. Yang, C. Zhu, D. Du, J. Zhu and Y. Lin, Nanoscale, 2015, 7, 14217–14231.
3. K. Yang, L. Feng, X. Shi and Z. Liu, Chem. Soc. Rev., 2013, 42, 530–547.
4. M. Nurunnabi, K. Parvez, M. Nfiujjaman, V. Revuri, H. A. Khan, X. Feng and Y. Lee, RSC Adv., 2015, 5, 42141–42161.
5. L. Tang, Y. Wang and J. Li, Chem. Soc. Rev., 2015, 44, 6954–6980.
6. A. Karimi, A. Othman, A. Uzunoglu, L. Stanciu and S. Andreescu, Nanoscale, 2015, 7, 6909–6923.
7. E. Lam and J. H. T. Luong, ACS Catal., 2014, 4, 3393–3410.
8. W. Hu, C. Peng, W. Luo, M. Lv, X. Li, D. Li, Q. Huang and C. Fan, ACS Nano, 2010, 4, 4317–4323.
9. Y. Tu, M. Lv, P. Xiu, T. Huynh, M. Zhang, M. Castelli, Z. Liu, Q. Huang, C. Fan, H. Fang and R. Zhou, Nat. Nanotechnol., 2013, 8, 594–601.
10. Z. Kong, W. Zheng, Q. Wang, F. Xi, L. Liang and J.-W. Shen, J. Mater. Chem. B, 2015, 3, 4814–4820.
11. W. Qin, X. Li, W.-W. Bian, X.-J. Fan and J.-Y. Qi, Biomaterials, 2010, 31, 1007–1016.
12. J. Guo, X. Yao, L. Ning, Q. Wang and H. Liu, RSC Adv., 2014, 4, 9953–9962.
13. H. Vovusha and B. Sanyal, RSC Adv., 2015, 5, 67427–67434.
14. J. Katoch, S. N. Kim, Z. F. Kuang, B. L. Farmer, R. R. Nalk, S. A. Tatulian and M. Ishigami, Nano Lett., 2012, 12, 2342–2346.
15. H. Sun, A. Zhao, N. Gao, K. Li, J. Ren and X. Qu, Angew. Chem., Int. Ed., 2015, 54, 7176–7180.
16. H. Vovusha, S. Sanyal and B. Sanyal, J. Phys. Chem. Lett., 2013, 4, 3710–3718.
17. H. Lei, L. Mi, X. Zhou, J. Chen, J. Hu, S. Guo and Y. Zhang, Nanoscale, 2011, 3, 3888–3892.
18. L. Sun, N. Hu, J. Peng, L. Y. Chen and J. Weng, Adv. Funct. Mater., 2014, 24, 6905–6913.
19. S. K. Mudedla, K. Balamurugan and V. Subramanian, J. Phys. Chem. C, 2014, 118, 16165–16174.
20. J. Ortiz-Medina, F. Lopez-Urias, H. Terrones, F. J. Rodriguez-Macias, M. Endo and M. Terrones, J. Phys. Chem. C, 2015, 119, 13972–13978.
21. J. Guo, T. Zhang, C. Hu and L. Fu, Nanoscale, 2015, 7, 1290–1295.
22. Y. Tian, Y. Liu, W. Wang, X. Zhang and W. Peng, Electrochim. Acta, 2015, 156, 244–251.
23. M. Li, C. Liu, H. Zhao, H. An, H. Cao, Y. Zhang and Z. Fan, Carbon, 2015, 86, 197–206.
24. J. Lu, D. Cui, H. Li, Y. Zhang and S. Liu, Electrochim. Acta, 2015, 165, 36–44.
25. S. Liu, J. Zhang, W. Tu, J. Bao and Z. Dai, Nanoscale, 2014, 6, 2419–2425.
26. R. Aswathi, M. M. Ali, A. Shukla and K. Y. Sandhya, RSC Adv., 2015, 5, 32027–32033.
27. W. Xiao, R. Zeng, L. Cheng, J. W. Wang, L. J. Jiang and L. G. Wang, RSC Adv., 2015, 5, 61861–61867.
28. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666–669.
29. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitkii, Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, ACS Nano, 2010, 4, 4806–4814.
30. L. S. Panchokarla, K. S. Subrahmanyam, S. K. Saha, A. Govindaraj, H. R. Krishnamurthy, U. V. Waghmare and C. N. R. Rao, Adv. Mater., 2009, 21, 4726–4730.
31. H. L. Poh, P. Simek, Z. Sofer and M. Pumera, ACS Nano, 2013, 7, 5262–5272.
32. M. Khan, M. N. Tahir, S. F. Adil, H. U. Khan, M. R. H. Siddiqui, A. A. Al-warthan and W. Tremel, J. Mater. Chem. B, 2015, 3, 18753–18808.
33. A. W. Robertson, B. Montanari, K. He, J. Kim, C. S. Allen, Y. A. Wu, J. Olivier, J. Neethling, N. Harrison, A. I. Kirkland and J. H. Warner, Nano Lett., 2013, 13, 1468–1475.
34. H. Wang, Q. Wang, Y. Cheng, K. Li, Y. Yao, Q. Zhang, C. Dong, P. Wang, U. Schwingensschlögl, W. Yang and X. X. Zhang, Nano Lett., 2012, 12, 141–144.
35. A. V. Krasheninnikov, P. O. Lehtinen, A. S. Foster, P. Pyykkö and R. M. Nieminen, Phys. Rev. Lett., 2009, 102, 126807.
36. L. Tsetseris, B. Wang and S. T. Pantelides, Phys. Rev. B: Condens. Matter Mater. Phys., 2014, 89, 035411.
37. C. B. Crook, C. Constantin, T. Ahmed, J.-X. Zhu, A. V. Balatsky and J. T. Haraldsen, Sci. Rep., 2015, 5, 12322.
38. H.-X. Luan, C.-W. Zhang, S.-S. Li, R.-W. Zhang and P.-J. Wang, RSC Adv., 2013, 3, 26261–26265.
39. E. J. G. Santos, A. Ayuela and D. Sánchez-Portal, New J. Phys., 2010, 12, 053012.
40. H. Zhang, X. Luo, X. Lin, X. Lu and Y. Leng, Int. J. Hydrogen Energy, 2013, 38, 14269–14275.
41. Z. Teng, Q. Xue, M. Shan, Z. Jiao, X. Zhou, C. Ling and Z. Yan, J. Phys. Chem. C, 2012, 116, 19918–19924.
42. S. A. Tawfik, X. Y. Cui, S. P. Ringer and C. Stampfl, RSC Adv., 2015, 5, 50975–50982.
43. E. H. Song, J. M. Yan, J. S. Lian and Q. Jiang, J. Phys. Chem. C, 2012, 116, 20342–20348.
44. S. Impeng, P. Khongpracha, J. Sirijaraensre, B. Jansang, M. Ehara and J. Limtrakul, RSC Adv., 2015, 5, 97918–97927.
45. H. Luo, H. Li, Q. Fu, Y. Chu, X. Cao, C. Sun, X. Yuan and L. Liu, Nanotechnology, 2013, 24, 495702.
46. A.-C. Liu, D. Chen, C.-C. Lin, H.-H. Chou and C. Chen, Anal. Chem., 1999, 71, 1549–1552.
47. A. Kühnle, T. R. Lindroth, M. Chunack and F. Besenbacher, Langmuir, 2006, 22, 2156–2160.
48. B. Delley, J. Chem. Phys., 2000, 113, 7756–7764.
49. S. Grimme, J. Comput. Chem., 2006, 27, 1787–1799.
50. J. L. Fajin, J. R. Gomes and M. N. Cordeiro, Langmuir, 2013, 29, 8856–8864.
51. B. Höffling, F. Ortmann, K. Hannewald and F. Bechstedt, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 81, 045407.
52. B. Cordero, V. Gómez, A. E. Platero-Prats, M. Revés, J. Echeverria, E. Cremades, F. Barragán and S. Alvarez, Dalton Trans., 2008, 21, 2832–2838.
53. L. Zhang and Z. Xia, J. Phys. Chem. C, 2011, 115, 11170–11176.
54. E. J. G. Santos, A. Ayuela and D. Sánchez-Portal, J. Phys. Chem. C, 2012, 116, 1174–1178.
55. Y.-L. Mao, X.-H. Yan and Y. Xiao, Nanotechnology, 2005, 16, 3092–3096.
56. B. Huang, J. Yu and S.-H. Wei, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 84, 075415.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra25599f

---