Nanoscale

COMMUNICATION

Cite this: Nanoscale, 2017, 9, 7016

Received 6th March 2017,Accepted 3rd May 2017

DOI: 10.1039/c7nr01601h

rsc.li/nanoscale

The origin of low workfunctions in OH terminatedMXenes†

Hassan A. Tahini, Xin Tan and Sean C. Smith *

The workfunction is an important parameter that governs several

electronic phenomena occurring at the surfaces and interfaces of

materials. Here, we study MXenes, which are two dimensional metal

carbides and nitrides. The workfunction is strongly dependent on

the terminating functional groups which induce surface dipoles and

Fermi level shifts. Here, we establish a correlation between the

workfunction and the adsorbate’s 2p band centres. Focusing on the

OH terminated MXenes which have intrinsically low workfunctions,

we show that a rigid relation between the 2p band centres and

workfunctions exists which resembles a volcano plot. This imposes

a limit on the lowest possible workfunctions of ∼1.2 eV and sets an

optimum value of the 2p band centres at which this low workfunc-

tion can occur which we determined to be ∼−5.45 eV relative to the

Fermi level. We demonstrate that neither strain modulation nor

doping can achieve workfunctions lower than this.

Introduction

Understanding and controlling a material’s intrinsic propertiesis of great importance in materials science as it advances boththe fundamental understanding of numerous mechanismsand the enhancement of current devices and applications. Onebasic parameter that is normally invoked to describe severalelectronic phenomena taking place at surfaces and interfacesis the workfunction (WF), which is simply the energy cost toextract an electron from a solid through its surface to the externalvacuum.1,2 In the field of nanoelectronics, the WF determinesthe alignment between metallic contacts and is therefore usedto gauge and assess the possibility of developing detrimentalSchottky barriers.3,4 Low WFs are particularly useful in ther-mionic devices where it is necessary to release electrons froma surface at the lowest energy cost. However, the workfunctionis a sensitive property which is highly dependent on the

atomic details of the surface under examination. Experimentalstudies on the same material can vary by as much as 1 eVor more, since surface orientation and – equally importantly –

terminations and reconstructions all play a significant role inthe observed WF.1 Selecting materials that are less susceptibleto surface orientations and with well-defined terminations aretherefore highly desirable. Such might be the case with 2Dmaterials5 where isolated atomic layers should in principlehave unambiguous workfunctions. Of particular interest is theMXene class of compounds, which have gained considerableattention since they were first fabricated in 2012 and havedemonstrated significant potential in numerous energy relatedapplications.6–16 MXenes are derived from their correspondingMAX phase17–19 (where M = transition metal, A = group 13–16element, X = C or N) by etching the A ions using HF or amixture of HCl and LiF,20,21 leading to materials with thegeneral formula Mn+1Xn.

Methodology

The calculations were performed based on density functionaltheory as implemented in the Vienna ab initio SimulationPackage (VASP).22 Core and valence electrons were treatedusing the projector augmented wave method.23,24 Exchange–correlation interactions were treated using the Perdew–Burke–Ernzerhof (PBE) functional.25 A plane wave cutoff energy of500 eV was used to describe the wavefunctions. Atomic positionsand lattice vectors in the x–y plane were fully optimized untilforces were less than 0.01 eV Å−1, while the tolerance on theself-consistent field loop energies was set to 1 × 10−6 eV. TheBrillouin zone was sampled using 17 × 17 × 1 Γ-centred mesh.The slabs were separated by a 20 Å vacuum region to avoidspurious interactions between periodic images. For all calcu-lations here we use a 1 × 1 slab except when defects are intro-duced to the MXene host where a 2 × 2 slab is used. Furthercalculations employing the HSE06 hybrid functional were per-formed. For these we used the default mixing of the PBEexchange–correlation (75%) and the Hartree–Fock exchange

†Electronic supplementary information (ESI) available. See DOI: 10.1039/c7nr01601h

Integrated Materials Design Centre (IMDC), School of Chemical Engineering,

UNSWAustralia, Sydney, NSW 2052, Australia. E-mail: sean.smith@unsw.edu.au

7016 | Nanoscale, 2017, 9, 7016–7020 This journal is © The Royal Society of Chemistry 2017

Publ

ishe

d on

05

May

201

7. D

ownl

oade

d by

Nor

th D

akot

a St

ate

Uni

vers

ity o

n 8/

27/2

018

7:41

:50

PM.

View Article OnlineView Journal | View Issue

(25%). Charge partitioning was performed using the Badermethod implemented by Henkelman et al.26

Results and discussion

Fig. 1(a) shows a structural model of the MXenes used in thisstudy. Here, we focus on the simple possible MXene structureswith the formula M2X, where X is sandwiched between the twometal layers. Due to the synthesis process, the MXenes arealways saturated with functional groups. The common onespresent in solution are F, O and OH and thus most studieshave considered these terminations. The possible sitesassumed by these terminations are shown in the top view ofFig. 1(a) and the preferred site of the terminating speciesdepends on the MXene and the functional group itself. TheMXenes considered here are Sc2X, Ti2X, V2X, Cr2X, Zr2X, Nb2X,Mo2X, Hf2X, and Ta2X (X = C or N). These were terminatedwith F, O and OH by filling the ontop, hcp, and fcc sites separ-ately which gives a total of 180 possible structures. We findthat the fcc site is the most favorable site for T to occupywhich is in agreement with previous studies.13,27,28 The aim inthis study is to extract trends and descriptors that can explainand guide the design of materials with a desired workfunction.This property is formally defined as the minimum energyneeded to extract an electron from the bulk of material (i.e.from its Fermi level, EF) to a region far away from its surfacedenoted as the vacuum level (Evac) and is given by: WF (Φ) =Vvac − EF. Because the shortcomings of density functionaltheory in describing the electronic level in solids and becauseof the large number of calculations involved here it is impor-tant to assess the suitability of PBE which is a fast andefficient functional relative to a more accurate but highlydemanding functional such as HSE06. The calculated WFsusing both PBE and HSE06 are shown in Fig. 1(b). One can seethat the deviation between PBE and HSE06 is not significantqualitatively and, in many cases, quantitatively the same. Theabove method is also consistent with experimental findings,being able to capture qualitative trends and in many cases areasonable quantitative agreement with experiments (see ESIFig. S1†).29,30 We note that in some cases the MXenes are pre-

dicted to be semiconducting and therefore the WF is depen-dent on the position of the Fermi level and is no longer anintrinsic property of the material.28 In such cases we definethe Fermi level to be at the top of the valence band. This leadsto a stronger deviation between the two functionals given thatPBE severely underestimates band gaps.

For the functionalized MXenes, modifications to the work-function relative to the bare MXene can be associated with theinduced surface dipoles caused by the functional groups aswell as shifts in the Fermi level of the material due to elec-tronic redistribution.1,27,28,31 According to Leung et al. this canbe expressed as ΔWF = −(e/ε0)ΔP, where ΔP is the change in thetotal dipole moment.32 The total dipole moment P can bebroken down to include contributions from the adsorbates (pa),the relaxed (p0) and unrelaxed (ps) bare surface and the chargeredistribution upon the formation of the functionalized struc-ture (Δp). In each case the xy-planar averaged charge density,ρðzÞ ¼ Ð

dxÐdyρðx; y; zÞ=S (S is the surface area) is used to

obtain the dipole moment via pðzÞ ¼ Ð z1z0zρðzÞdz. We note that

while this definition can lead to an arbitrary value of p(z) basedon the choice of the limits in the above integral, it is conven-tional to set z0 to coincide with the centre of the slab and z1 setto be deep in the vacuum region.27,31,32 This relation was shownto be robust and holds for a large number of materials includ-ing MXenes of different thicknesses.27,28,32 Fig. 2 shows the be-havior of ΔWF with respect to the surface dipole moment. Theexpected linear relation holds, but we also noticed some scatterof the data points, in particular for some OH terminatedMXenes which show stronger deviation from linearity. This hasbeen observed before and attributed to the complexity of theenergy and charge landscape induced by the OH group.27,28

Obtaining a descriptor that directly correlates with theworkfunction can be a useful tool to design and screen formaterials with a particular functionality. In the case ofMXenes, we see that the workfunction is dependent on thetype of functional group present which leads us to probe theirelectronic structure further. One such electronic descriptorthat has gained lots of attention in the field of electrocatalysisis the oxygen 2p-band centre,33–35 which was used to assessthe activity of a surface in binding reaction intermediatesbased on the relative position of the p-bands with respect tothe Fermi level. Recently, this was used to describe the work-

Fig. 1 (a) Top and side views of the M2XT2 structural models used inthis study. The ontop, hcp and fcc sites are represented by the orange,blue and purple, respectively. (b) The workfunctions of all the stableMXenes with F, O and OH terminations using PBE and HSE06 calculations,which shows a reasonable agreement between the two functionals.

Fig. 2 Changes in the workfunction upon functionalization withrespect to the surface dipole moment.

Nanoscale Communication

This journal is © The Royal Society of Chemistry 2017 Nanoscale, 2017, 9, 7016–7020 | 7017

Publ

ishe

d on

05

May

201

7. D

ownl

oade

d by

Nor

th D

akot

a St

ate

Uni

vers

ity o

n 8/

27/2

018

7:41

:50

PM.

View Article Online

function of a range of perovskites, and a direct correlationbetween the two was demonstrated.36 We apply a similarapproach by calculating the p-band centres (Tp) of F and O inthe F, O and OH terminated carbide and nitride MXeneswhich is evaluated using:

TpðEÞ ¼ÐE�DTpðEÞdEÐDTpðEÞdE

� EF

where DTp(E) is the projected density of states of the p-orbitals.

To clearly display the trend in our results, Fig. 3(a)–(c) separ-ately show the F, O and OH terminated carbide and nitrideMXenes. A positive correlation emerges for M2XF2 and M2XO2

while a negative correlation appears for the M2X(OH)2 com-pounds. We also performed similar analysis using N and C 2p-band centres but no useful trends emerged. These correlationspoint to the importance of the band shifts on the magnitude ofthe workfunction. The band position is a direct result of thechemical interaction between the metal ions and the functionalgroups and can be partially understood based on the interactingspecies’ electronegativities and their degree of hybridization.

To further investigate the influence of chemical interactionon the T2p-band centres and how this affects workfunctions weapplied strain to a selected number of OH terminated MXeneswhich exhibited the lowest workfunctions under strain freeconditions. Fig. 4(a) shows that for the subset of MXenes con-sidered here workfunctions generally decrease (increase) withtensile (compressive) biaxial strain. The two exceptions areZr2C(OH)2 and Sc2C(OH)2 for which the former shows a weakdependence on strain while the latter exhibits a nearly lineardependence with workfunctions increasing (decreasing) withtensile (compressive) strain. Sc2C(OH)2 which is a semi-conductor under strain free conditions with a band gap ofabout 0.41 eV (0.44 eV in ref. 12) becomes metallic underincreasing compressive strain. Under tensile strain its bandgap increases reaching 0.72 eV at 4% strain. At −4% strain itsworkfunction reaches 1.26 eV which is very low consideringother materials used in thermionic devices where a low work-function is a key factor such as alkali ion coated materials withvalues ranging between 1–1.4 eV,37–39 which makes OH termi-nated MXene possible candidates as thermionic electron emit-ters. Previous studies have demonstrated that OH terminated

MXenes possess favorable formation energies and strong OHbonds to make them durable under high temperature con-ditions.27,28 Under real experimental conditions where mul-tiple terminations exist and particularly where O terminationsare preferred, a scheme has been proposed based on applyinga positive electrochemical potential, U, to convert the surfaceto OH termination.28

Fig. 4(b) shows the variations in the workfunctions as afunction of O2p from which we can see that for each compounda correlation emerges for these two quantities. This trend isvery clear for Sc2C(OH)2, Cr2C(OH)2 and V2C(OH)2. Again, wenote that Sc2C(OH)2’s workfunction behaves differently fromother compounds in that its slope is positive whereas othercompounds exhibit a negative one. The emergence of negativeand positive sloped correlations indicates that there are twobranches that depend on T2p where one corresponds tomaterials with low band centres and the other corresponds tomaterials with high band centres. This implies that the work-function cannot be changed to arbitrarily smaller values.Rather, if the two branches were extrapolated we will find thatthere is an optimum value of T2p corresponding to the lowestpossible workfunction. In this regard, Fig. 4(b) can be viewedas an inverted volcano plot. The concept of volcano plots iswidely used in catalysis and emerges directly from the Sabatierprinciple. To probe this idea further, we consider what willhappen if we introduce dopants with varying physical pro-perties such as ionic radii and electronegativities. To this endwe doped the M sites of the MXenes shown in Fig. 4 withalkali and alkaline earth metals (Li, Na, Be, Mg, Ca, Sr, Ba).Here, we substituted 2 and 4 sites of the 8 M sites in a (2 × 2)MXene supercell with dopants. This allowed us to generate alarge number of configurations with different workfunctions

Fig. 3 The workfunctions of carbide and nitride based MXenes (shownin red and blue, respectively) as a function of the p-band centre of theterminating species. (a) F terminated, (b) O terminated and (c) OH termi-nated MXenes. The workfunctions display a positive correlation with thep-band centre for the F and O terminations, whereas a negative corre-lation emerges for the OH terminated ones.

Fig. 4 Strain modulated workfunctions for MXenes with low intrinsicworkfunction under strain free conditions are shown in (a). Generally,the workfunctions drop with tensile strain with the exception of Sc2C(OH)2. (b) Linear dependence of the workfunctions on the O 2p-bandcentre emerges for each compound.

Communication Nanoscale

7018 | Nanoscale, 2017, 9, 7016–7020 This journal is © The Royal Society of Chemistry 2017

Publ

ishe

d on

05

May

201

7. D

ownl

oade

d by

Nor

th D

akot

a St

ate

Uni

vers

ity o

n 8/

27/2

018

7:41

:50

PM.

View Article Online

and O 2p-band centres as shown in Fig. 5. We emphasize herethat we are trying to demonstrate a clear relation betweenp-band centres and the workfunction and that some of thesecompositions might not be thermodynamically feasible giventhat the substitution energies sometimes exceed 1 eV (on theother hand, some compositions gave negative substitutionenergies as well, full details are in the ESI†). A distinct volcanoplot is clearly visible with the 2p-band centres spanning from−6.5 eV to −3.5 eV relative to the Fermi level. The lowest work-function achievable is about 1.2 eV within the set of 6 OH ter-minated MXenes considered here.

Looking at Fig. 3 (b) and (c) we notice that O2p for O andOH terminated MXenes for both carbides and nitrides lie ontwo separate legs of the volcano. The different behavior of theO and OH terminated MXenes stems from the changes causedby H on the way the O 2p and the M d orbitals hybridize witheach other. Bader analysis reveals that the charges on O and X(C or N) are least affected upon hydrogenation, whereas thecharges on M are changed dramatically and become less posi-tive with OH terminations (see the ESI† for more details). Asthe hybridization increases between O and M due to the largeroverlap between the O 2p and M d orbitals O2p is shifted closerto the Fermi level leading to an increase in the workfunctions(which corresponds to the high O2p leg of the volcano plot).36

The presence of H reduces the O–M hybridization shifting the2p band centres away from the Fermi level resulting in a lowerWF (which corresponds to the low O2p leg of the volcano plot).

Conclusions

In this work we investigated the workfunctions of F, O and OHterminated MXenes. We assessed PBE against the more accu-rate HSE06 functional and verified its suitability to describe

the workfunctions of a wide range of configurations. Thechanges in the workfunction due to functionalization aremainly due to changes in the surface dipole moment inducedby the terminating species and these two properties are line-arly correlated. A powerful descriptor emerges from the 2pband centres of the terminating F or O ion, which correlateswith the computed workfunctions. This correlation holdswhen describing a number of strained OH terminated MXeneswhere we obtain positive and negative sloped correlations indi-cating that there is a limit to how much the workfunction canbe lowered. We confirmed this volcano type behaviour by con-sidering MXenes alloyed with alkali and alkaline earth metalswhich showed a wide range of 2p band centres for which theworkfunctions were distributed on the two legs of the volcanowith the lowest achievable workfunction being 1.2 eV. Thechanges in the workfunctions are a result of the hybridizationof the metal d orbitals and O 2p orbitals which is reflected inthe position of the O 2p-band centre which provides a directtool to design materials with specific workfunctions.

Acknowledgements

This research was undertaken with the assistance ofUNSW Australia SPF01 funding (SCS). We acknowledgegenerous allocations of supercomputing time at the PawseySupercomputing Centre via the Australian NationalComputational Merit Allocation Scheme (NCMAS project fr2)and the Energy and Resources Merit Allocation Scheme of thePawsey Supercomputing Centre (project pawsey0111).

References

1 A. Kahn, Mater. Horiz., 2016, 3, 7–10.2 G. Hua and D. Li, Phys. Chem. Chem. Phys., 2016, 18, 4753–

4759.3 M. T. Greiner and Z.-H. Lu, NPG Asia Mater., 2013, 5, e55.4 Y. Jiao, A. Hellman, Y. Fang, S. Gao and M. Käll, Sci. Rep.,

2015, 5, 11374.5 Q. Tang, Z. Zhou and Z. Chen, Wiley Interdiscip. Rev.:

Comput. Mol. Sci.., 2015, 5, 360–379.6 Q. Hu, D. Sun, Q. Wu, H. Wang, L. Wang, B. Liu, A. Zhou

and J. He, J. Phys. Chem. A, 2013, 117, 14253–14260.7 D. Er, J. Li, M. Naguib, Y. Gogotsi and V. B. Shenoy, ACS

Appl. Mater. Interfaces, 2014, 6, 11173–11179.8 J. Zhu, A. Chroneos and U. Schwingenschlögl, Phys. Status

Solidi RRL, 2015, 9, 726–729.9 Z. Guo, J. Zhou, L. Zhu and Z. Sun, J. Mater. Chem. A, 2016,

4, 11446–11452.10 C. Ling, L. Shi, Y. Ouyang, Q. Chen and J. Wang, Adv. Sci.,

2016, 3, 1600180.11 H. Pan, Sci. Rep., 2016, 6, 32531.12 S. Kumar and U. Schwingenschlögl, Phys. Rev. B: Condens.

Matter, 2016, 94, 35405.

Fig. 5 Alloying MXenes with group 1 and 2 metals (Li, Na, Be, Mg, Ca,Sr, Br) can further reduce the workfunctions in some cases. A volcanoplot shows that the workfunctions cannot be reduced to an arbitraryvalue but rather scale with the O 2p-band centres with workfunctionsnot dropping below ∼1.2 eV at on optimum O2p ∼ −5.45 eV vs. EF.Green, light blue, red and grey spheres represent M sites, dopants,oxygen and hydrogen, respectively.

Nanoscale Communication

This journal is © The Royal Society of Chemistry 2017 Nanoscale, 2017, 9, 7016–7020 | 7019

Publ

ishe

d on

05

May

201

7. D

ownl

oade

d by

Nor

th D

akot

a St

ate

Uni

vers

ity o

n 8/

27/2

018

7:41

:50

PM.

View Article Online

13 Z. W. Seh, K. D. Fredrickson, B. Anasori, J. Kibsgaard,A. L. Strickler, M. R. Lukatskaya, Y. Gogotsi, T. F. Jaramilloand A. Vojvodic, ACS Energy Lett., 2016, 1, 589–594.

14 J.-C. Lei, X. Zhang and Z. Zhou, Front. Phys., 2015, 10,276–286.

15 Q. Tang, Z. Zhou and P. Shen, J. Am. Chem. Soc., 2012, 134,16909–16916.

16 X. Zhang, J. Lei, D. Wu, X. Zhao, Y. Jing and Z. Zhou,J. Mater. Chem. A, 2016, 4, 4871–4876.

17 M. W. Barsoum, Prog. Solid State Chem., 2000, 28, 201–281.

18 Z. M. Sun, Int. Mater. Rev., 2011, 56, 143–166.19 Y. Mo, P. Rulis and W. Y. Ching, Phys. Rev. B: Condens.

Matter, 2012, 86, 165122.20 M. Naguib, M. Kurtoglu, V. Presser, J. Lu, J. Niu, M. Heon,

L. Hultman, Y. Gogotsi and M. W. Barsoum, Adv. Mater.,2011, 23, 4248–4253.

21 M. Naguib, O. Mashtalir, J. Carle, V. Presser, J. Lu,L. Hultman, Y. Gogotsi and M. W. Barsoum, ACS Nano,2012, 6, 1322–1331.

22 G. Kresse and J. Furthmüller, Comput. Mater. Sci., 1996, 6,15–50.

23 P. E. Blöchl, Phys. Rev. B: Condens. Matter, 1994, 50, 17953–17979.

24 G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter,1999, 59, 1758–1775.

25 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett.,1996, 77, 3865–3868.

26 G. Henkelman, A. Arnaldsson and H. Jónsson, Comput.Mater. Sci., 2006, 36, 354–360.

27 M. Khazaei, M. Arai, T. Sasaki, A. Ranjbar, Y. Liang andS. Yunoki, Phys. Rev. B: Condens. Matter, 2015, 92, 75411.

28 Y. Liu, H. Xiao and W. A. Goddard, J. Am. Chem. Soc., 2016,138, 15853–15856.

29 K. Kobayashi, Surf. Sci., 2001, 493, 665–670.30 P. A. Khomyakov, G. Giovannetti, P. C. Rusu, G. Brocks,

J. van den Brink and P. J. Kelly, Phys. Rev. B: Condens.Matter, 2009, 79, 195425.

31 W. Zhang, C. Cheng, P. Fang, B. Tang, J. Zhang, G. Huang,X. Cong, B. Zhang, X. Ji and L. Miao, Phys. Chem. Chem.Phys., 2016, 18, 4376–4384.

32 T. C. Leung, C. L. Kao, W. S. Su, Y. J. Feng and C. T. Chan,Phys. Rev. B: Condens. Matter, 2003, 68, 195408.

33 A. Grimaud, K. J. May, C. E. Carlton, Y.-L. Lee, M. Risch,W. T. Hong, J. Zhou and Y. Shao-Horn, Nat. Commun.,2013, 4, 2439.

34 Y.-L. Lee, J. Kleis, J. Rossmeisl, Y. Shao-Horn andD. Morgan, Energy Environ. Sci., 2011, 4, 3966.

35 B. Han, M. Risch, Y.-L. Lee, C. Ling, H. Jia and Y. Shao-Horn, Phys. Chem. Chem. Phys., 2015, 17, 22576–22580.

36 R. Jacobs, J. Booske and D. Morgan, Adv. Funct. Mater.,2016, 26, 5471–5482.

37 J.-L. Desplat and C. A. Papageorgopoulos, Surf. Sci., 1980,92, 97–118.

38 A. C. Papageorgopoulos and M. Kamaratos, J. Phys.:Condens. Matter, 2000, 12, 9281–9291.

39 H. Yuan, S. Chang, I. Bargatin, N. C. Wang, D. C. Riley,H. Wang, J. W. Schwede, J. Provine, E. Pop, Z.-X. Shen,P. A. Pianetta, N. A. Melosh and R. T. Howe, Nano Lett.,2015, 15, 6475–6480.

Communication Nanoscale

7020 | Nanoscale, 2017, 9, 7016–7020 This journal is © The Royal Society of Chemistry 2017

Publ

ishe

d on

05

May

201

7. D

ownl

oade

d by

Nor

th D

akot

a St

ate

Uni

vers

ity o

n 8/

27/2

018

7:41

:50

PM.

View Article Online