Computational Materials Science 77 (2013) 281–285

Contents lists available at SciVerse ScienceDirect

Computational Materials Science

journal homepage: www.elsevier.com/locate/commatsci

Effects of thickness on the structural, electronic, and optical properties

of MgF2 thin films: The first-principles study q
Li-ping Wang a,b, Pei-de Han c, Zhu-xia Zhang a, Cai-li Zhang c, Bing-she Xu a,⇑
a
Key Laboratory of Interface Science and Engineering in Advanced Materials of Taiyuai University of Technology, Ministry of Education, Taiyuan 030024, PR China
b
Department of Physics and Optoelectronics, Taiyuan University of Technology, Taiyuan 030024, PR China
c
College of Materials Science and Engineering, Taiyuan University of Technology, Taiyuan 030024, PR China

a r t i c l e i n f o a b s t r a c t

Article history: First-principles calculations were performed on MgF2 (0 0 1) thin films and their structural, electronic,
Received 11 March 2013 and optical properties were investigated. Results showed that the films are energetically stabilized when
Received in revised form 10 April 2013 the layer number is increased to 25, corresponding to a thickness of about 4 nm. The thin films exhibited
Accepted 12 April 2013
narrower band gaps compared with their bulk counterparts; gaps increased with increasing thickness of
the thin films. Analysis of the optical properties of the films showed that their refractive index is signif-
icantly decreased and their extinction coefficient is slightly increased compared with the bulk. With
Keywords:
increasing thin film thickness, the refractive index decreased slightly and the extinction coefficient
MgF2 (0 0 1) thin films
Surface energy
remained unchanged. These results indicate that MgF2 thin films of a certain thickness range have excel-
Electronic properties lent performance as anti-reflection and high-reflection coatings at visible wavelengths.
Optical properties Ó 2013 The Authors. Published by Elsevier B.V. All rights reserved.
Thickness dependency
Density functional theory (DFT)

1. Introduction structures and properties of nanoscale materials are often quite

Optical thin films have recently attracted significant interest
because of their potential applications in electronics, optics,
and photonics. Coatings for these applications are required to
have efficient optical properties, such as very low residual
reflection loss, wide band width, and high efficiency of trans-
mission. As one of the most important optical film materials
currently available, MgF2 possesses a low refractive index, large
band gap (about 10.8 eV), and good transparency over a wide
range of wavelengths. MgF2 has been used for solar cells, UV la-
ser applications, and organic light emitting diodes [1,2]. Over
the past few years, multilayers consisting of MgF2 and other
materials, such as MgF2/LaF3 [3,4], MgF2/TiO2 [5], MgF2/ZnS
[6], and MgF2/Si [7], have been widely used as high-reflection
and anti-reflection coatings in many devices. MgF2 single thin
films, the most widely used single-layer anti-reflection coating
material, have been applied in optical lenses, cameras, and tele-
scopes, among other applications [8]. The characteristic size of
such films is known to be as low as the nanoscale level. The
q gations on MgF2 single thin films have been made. Thus, many
This is an open-access article distributed under the terms of the Creative
Commons Attribution-NonCommercial-No Derivative Works License, which per-
mits non-commercial use, distribution, and reproduction in any medium, provided
the original author and source are credited.
⇑ Corresponding author. Tel.: +86 351 6018843; fax: +86 351 6010311.
E-mail address: wangliping@tyut.edu.cn (B.-s. Xu).
different from those of the bulk [9,10]. For example, boron-
doped diamond (0 0 1) and (1 1 1) thin films have higher super-
conductivity than boron-doped bulk diamond [11]. Even Sn-
doped SiO2 thin films exhibit better optical properties than their
bulk counterparts [12]. Due to the lack of valid testing methods
for ultrathin thin films, related studies are usually performed
only theoretically to understand their structure, properties, and
mechanism. For instance, a theoretical study on ZnO thin films
indicated that the wurtzite film transforms into a graphitic
structure when the thickness is below 18 layers [13]. Further-
more, graphitic ultrathin ZnO (0 0 0 1) thin films exhibit wider
band gaps compared with bulk films and the band gap de-
creases with increased nanofilms thickness [14]. A study on
(1 1 1) diamond films revealed a layer number (n) of 12 as a
threshold for realizing thin (1 1 1) diamond films. Analysis of
electronic band structures also revealed the transition from
semimetal to wide band semiconductor with increasing n [15].
Whilst many theoretical studies that investigate the geometric
structure, electronic structure, stability, and properties of MgF2
surfaces [16–19] and bulk [20] have been reported, few investi-
questions about MgF2 single films remain unanswered, such as
whether or not and how the structure and stability of MgF2 sin-
gle thin films depend on the film thickness, differences between
the performance of the bulk and film materials, as well as the

0927-0256/$ - see front matter Ó 2013 The Authors. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.commatsci.2013.04.031
282 L.-p. Wang et al. / Computational Materials Science 77 (2013) 281–285

dependence of the electronic structure and optical properties on Table 1

the film thickness, among others. Therefore, efforts are made in Surface energies for different MgF2 (0 0 1) thin films.

the current study to present a detailed theoretical discussion on Atomic layer Atom Surface energy
MgF2 single thin films using the plane wave pseudo-potential number (n) number (J/m2)
method. The structural, electronic, and optical properties of 5 15 3.15
MgF2 films are analyzed and the effects of film thickness on 9 27 2.32
the structure and performance of the films are addressed. 13 39 2.31
17 51 2.29
21 63 2.28
25 75 2.20
2. Computational details
29 87 2.20
33 99 2.20
MgF2 has a rutile structure where Mg2+ ions are surrounded by
six F ions, and each F ion is further surrounded by three Mg2+
ions [21]. A low-index surface is usually relatively stable. The
(0 0 1) surface, which is the simplest surface of MgF2, was selected the lowest possible cutoff energy for the plane-wave basis set.
for construction of the film models. Details on the geometry of a Norm-conserving pseudo-potentials [26] were used in the
typical MgF2 (0 0 1) film are given in Fig. 1. The film was cut into calculations for optical properties. To compensate for LDA underes-
a 1  1 (0 0 1) slab from the optimized
0
bulk structure. A vacuum timation of the band gap, the calculated optical spectra were
layer with a thickness of 12 Å A along the [0 0 1] direction was in- shifted by the scissors operator, which is a rigid shift of the conduc-
cluded in each model. The layer was thick enough to eliminate tion band (CB) relative to the valence band (VB) to match the
the effect of the periodic boundary conditions in the [0 0 1] direc- experimental band gaps of MgF2. Calculations for the MgF2 bulk
tion. To maintain symmetry, the bottom and top layers of the films structure and band gap were in good agreement with the results
were designed to be identical. To determine a reasonable of previous studies [20,21], indicating that the calculations made
nanofilms model within our computational resources, a systematic in this study are reliable.
approach was used. Different film structures were constructed by
controlling the number of the atomic layer. The smallest MgF2 3. Results and discussion
(0 0 1) thin film contained five 1  1 (0 0 1) atomic layers. Based
on this film, four more atomic layers are added to build a new case. 3.1. The structure of MgF2 (0 0 1) thin films
Thus, eight cases were established altogether, consisting of 5, 9, 13,
17, 21, 25, 29, and 33 layers; the largest model possessed 99 atoms. Surface stability is usually determined by the surface energy,
In all calculations, the five central layers were fixed at their bulk- and surfaces with smaller surface energies are generally more sta-
like sites whereas the surface and subsurface layers on either side ble than those with high energies. In this study, the surface energy
of the slab were allowed to relax without any symmetry is defined as follows [27]:
constraints.
All calculations were performed with the program package Ca-
ESurf ¼ ðEslab  NMg EMg  NF EF Þ=2A ð1Þ
step [22] of Accelrys Materials Studio. Local density approximation where Eslab is the total energy of MgF2 (0 0 1) slab, NMg and NF are
(LDA) [23] was employed to determine the exchange and correla- the respective numbers of Mg and F atoms per computational cell,
tion between electrons. The local CA–PZ functional [24] was used and EMg and EF are the reference bulk total energies per atom of
to treat electron–ion interactions. In our calculations, the energy Mg and F, respectively. A is the area of the surface. All related cases
cutoff for a plane-wave expansion was 500 eV. Grids of converged well and the results of the surface energies are shown in
5  5  1 k-point, including the C-point, were used for sampling Table 1. The surfaces energies of all of the thin films were positive,
irreducible Brillouin zones. The crystal relaxation was halted when indicating that all of the structures are energetically stable. More-
the charge difference was less than 0.001 e between two steps as a over, with increasing atomic layer number n from 5 to 25, the sur-
convergence tolerance condition. During electronic property calcu- face energy decreased gradually and the stability of the thin films
lations, Vanderbilt-type ultrasoft pseudo-potentials [25] were
used. These potentials allow calculations to be performed with

Fig. 1. Side view of the MgF2 (0 0 1) thin films: (a) 9-layer, (b) 17-layer, and (c) 25- Fig. 2. Distances between layer relaxations of the 25-layer (0 0 1) MgF2 thin film
layer thin film. given in percentage, Ddi,i+1.
 L.-p. Wang et al. / Computational Materials Science 77 (2013) 281–285 283

Fig. 3. Total and partial DOS of MgF2 (a) bulk and (b) (0 0 1) thin film.

Fig. 4. Electron density difference between (a) middle layers and (b) top layers of the 25-layer MgF2 (0 0 1) thin film (The white portion shows 0 electron density difference,
the blue area shows for low density distribution, and the red area shows high density distribution). (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)

increased. Films consisting n = 25, 29, and 33 had the same mini- creases. At n = 25, 29, and 33, the thickness of the thin films was
mum surface energy (2.2 J/m2), which implies that the stability no 3.7, 4.3, and 4.9 nm, respectively. Therefore, we can conclude that
longer increases past a certain threshold even if the value of n in- the MgF2 (0 0 1) thin films can exist stably at about 4 nm thickness,
284 L.-p. Wang et al. / Computational Materials Science 77 (2013) 281–285

i.e., 4 nm is the threshold for realizing the thinnest MgF2 (0 0 1)
films. The 25-layer slab was chosen as the film model in subsequent
calculations.
To illustrate the structure of the thin films more clearly, the
variations in distance between adjacent atomic layers parallel to
the MgF2 (0 0 1) surface during relaxation were also determined.
Here, Ddi,i+1 is given in terms of percent with respect to the unre-
laxed interlayer spacing, d0, i.e., Ddi,i+1 = 100(di,i+1  d0)/d0, where
d0 is the bulk interlayer distance. For the 25-layer film under con-
sideration, the value of i ranged from 1 to 25, where 1 is defined as
the top layer. As seen in Fig. 2, the interlayer spacing between the
two outermost layers had a significant contraction of about 4.1%,
implying a strong interaction, whereas the distance between the
second and third layers slightly increased. From the third layer on-
wards, the distance continued to decrease until the eleventh layer.
No change was found in the three middle layers (layers 11, 12, and
13) with respect to the bulk. These layers are fixed in their bulk
sites during optimization. Due to symmetry, the distances between
layers 14 and 25 had the exact same changes.
3.2. Electronic properties
distributions compared with the middle layers. In Fig. 4b, the area
close to the vacuum of the top Mg layer is deep blue, indicating evi-
dent electron transfer. By contrast, the white region near the F
atoms of the second layer implies electron transfer to F 2s and F
2p orbitals, whereas for those F of the middle layers electron trans-
fer only to 2p orbital. Furthermore, the charge distributions of sev-
eral top layers were less localized than those of middle layers,
signifying that the atom interactions of top layers appear certain
covalent, while that of those middle layers appear ionic only. These
differences become negligible as n increases to 6, which is consis-
tent with the conclusion that the band gap increases with increas-
ing thickness.
3.3. Optical properties
The optical properties were determined with a scissors operator
correction of 3.96 eV. The optical properties of matter can be
described by the transverse dielectric function, eðxÞ ¼ e1 ðxÞþ
e2 ðxÞ, which is mainly a function of the electronic structure. The
imaginary part, e2 ðxÞ, can be calculated from the momentum
matrix elements between the occupied and unoccupied wave

Using the relaxed film structure obtained from optimization,

electronic structure calculations were performed on the film sys-
tem and the results were compared with their corresponding bulk
systems. The results of the density of states (DOS) of the MgF2 bulk
and 25-layer (0 0 1) nanofilms are plotted in Fig. 3; the Fermi en-
ergy is 0 eV on the energy axis. For the bulk MgF2 structure, two
sets of DOS can be observed near the Fermi level, as seen in
Fig. 3a. The CB between 6.7 and 22.5 eV corresponded mainly to
Mg 2p states and was composed of F 2p states. The VB corre-
sponded predominantly to F 2p characters and only slightly to
Mg 3s from 4.7 eV to 0 eV. The theoretical value of the band
gap of the bulk was 6.84 eV, much smaller than the experimental
result of 10.8 eV [28] but still in close agreement with other theo-
retical values [20]. The difference between the theoretical and
experimental results may be attributed to inherent limitations in
all density functional theory (DFT) simulations. The DOS of the
MgF2 (0 0 1) film are shown in Fig. 3b. Compared with the bulk,
the VB of the nanofilm consisted of F 2p and Mg 3s states, whereas
its CB mainly consisted of F2s and Mg 2p states. However, the band
gap of the 25-layer (0 0 1) nanofilm was 5.91 eV and was narrowed
by about 0.93 eV, very similar to other ionic fluorites [29]. The
band gap reduction is mainly attributed to the downward shift in
CB by about 1 eV. In addition, the energy state of the film was split
and surface states appeared due to the dangling bonds of the top
layer. In addition to the change in its location, the width of the
CB of the nanofilms was reduced by about 10 eV relative to the
bulk, which indicates that the electrons of the surface exhibit
strong locality and larger effective mass. The band gaps of the
29- and 33-layer thin films were 5.94 and 5.95 eV, respectively,
implying a slight increase in band gap with increasing film thick-
ness. The increment in film DOS in Fig. 3 is attributed to the in-
crease in number of atoms.
To determine the origin of the changes in electronic structure of
the MgF2 (0 0 1) film relative to the bulk, the charge density differ-
ence distribution is plotted on the best-fit slice, which is perpen-
dicular to the (0 0 1) surface. Calculation results of the middle
atomic layers of the 25-layer film and the top layer are shown in
Fig. 4a and b, respectively. The atom positions are labelled in the
plots. The color map legend of the slice ranges from 0.1 e to
0.15 e, where the excluded above 0.15 e is represented by white.
The electron distributions of the middle and top layers appeared
approximately spherical, indicating that, similar to the bulk [20], Fig. 5. Calculated optical properties of MgF2 (0 0 1) thin films: (a) refractive index
MgF2 (0 0 1) films feature ionic bonds. Comparison of Fig. 4a and and (b) extinction coefficient calculated according to a scissors operator correction
b shows that the top layers have slightly different charge of 3.96 eV.
 L.-p. Wang et al. / Computational Materials Science 77 (2013) 281–285
functions. The real part, e1 ðxÞ, can be evaluated from the imagi-
nary part by the famous Kramer-Kronig relationship. The corre-
sponding refractive index and extinction coefficient can be
derived from e1 ðxÞ and e2 ðxÞ, given by [30]:
pffiffiffi
n ¼ 1= 2½ðe21 þ e22 Þ1=2 þ e1 1=2 ð2Þ
pffiffiffi
k ¼ 1= 2½ðe21 þ e22 Þ1=2  e1 1=2 ð3Þ
The calculated optical properties of MgF2 (0 0 1) thin films,
including bulk experimental and theoretical data, are shown in
Fig. 5. To discuss the effects of film thickness on the optical prop-
erties of the films, 25-, 29-, and 33-layer models were considered.
As seen from Fig. 5a, the calculated MgF2 bulk refractive index was
in good agreement with the experimental value [31], indicating
reliable calculations. However, the theoretical peak was less in-
tense than the experimental result. This discrepancy may be attrib-
uted to the inadequacy of the exchange–correlation functional in
DFT theory. Compared with the bulk, the refractive indices of the
three MgF2 (0 0 1) thin films were evidently lower by about 0.3 in
almost all wavelength regions, which is consistent with experi-
mentally reported results [31]. The refractive index of the 4 nm ul-
tra-thin film was about 1.05 at 2.255 eV (550 nm), much lower
than that of the 230 nm film, which is 1.379 at 550 nm [8]. The
lower refractive index of the MgF2 ultra-thin nanofilms confirms
that the material has promising applications in anti-reflection
coatings and high-reflective systems. The refractive index of the
film was found to decrease slightly with the increase of the film
thickness. Fig. 5b shows the dependence of the extinction coeffi-
cient of the MgF2 thin films on photon energy. The MgF2 thin films
have good transmittance in a wide region from 2 eV to 8 eV, which
is similar to the bulk material. At regions above 8 eV, the extinction
coefficient of the film slightly increased, implying slight absorp-
tion. However, unlike its real part, the imaginary part of the refrac-
tive index showed almost no dependence on film thickness from
2 eV to 8 eV. A slight decrease can only be observed with the
increasing film thickness above 8 eV. These results indicate that
MgF2 ultra-thin thin films with a certain range of thickness have
favorable performance as anti-reflection and high-reflection coat-
ings at visible wavelengths due to their lower refractive index.
All of the results are in excellent agreement with the results of pre-
vious studies [18,31–33].
4. Conclusions
In summary, the structural, electronic, and optical properties of
MgF2 (0 0 1) thin films were investigated using first-principles cal-
culations. The results showed that MgF2 (0 0 1) films of 25, 29, and
33 layers have the same minimum surface energy, which indicates
that the MgF2 (0 0 1) films can exist stably when the thickness is
greater than 4 nm. Electronic structural analysis of the 25-layer
nanofilms revealed that atoms of the top 6 layers interact by cova-
lent bonding, indicating a charge distribution different from the
bulk. The band gaps of the thin films were narrower compared
with the bulk and they increased slightly with increasing film
285
thickness. Finally, calculations of the optical properties clearly
showed that the MgF2 (0 0 1) nanofilms have a significantly lower
refractive index and slightly higher extinction coefficient than
the bulk material. With increasing film thickness, the refractive in-
dex slightly decreased, whereas the extinction coefficient re-
mained almost unchanged. MgF2 thin films with a certain range
of thickness have improved performance as anti-reflection and
high-reflection coatings at visible wavelengths due to their low
refractive index.
Acknowledgements
This research was supported by the National Natural Science
Foundation of China (Grant No. 51002102), and Youth Foundation
of Taiyuan University of Technology (No. 2012L037).
References
[1] F. Perales, D. Soto, Thin Solid Films 518 (2010) 4221–4224.
[2] S.F. Chen, H.Y. Shi, F. Cheng, et al., Org. Electron. 9 (2012). <http://dx.doi.org/
10.1016/j.orgel.2012.09.038>.
[3] G.H. Liu, Q.L. Xiao, Y.X. Jin, et al., Vacuum 84 (2010) 778–781.
[4] H. Yu, L.Q. Wu, C.F. Liu, et al., Opt. Laser Technol. 44 (2012) 810–814.
[5] A. Bittner, A. Schmitt, R. Jahn, P. Lömann, Thin Solid Films 520 (2012) 1880–
1884.
[6] S.L. Diedenhofen, G. Grzela, E. Haverkamp, et al., Sol. Energy Mater. Sol. Cells
101 (2012) 308–314.
[7] H.H. Guan, P.D. Han, Y.P. Li, et al., Opt. Commun. (2012), http://dx.doi.org/
10.1016/j.optcom.2012.01.080.
[8] S. Korkmaz, S. Elmas, N. Ekem, S. Pat, M.Z. Balbağ, Opt. Commun. 285 (2012)
2373–2376.
[9] J.R. Jameson, W. Harrison, P.B. Griffin, J. Appl. Phys. 92 (2002) 4431–4440.
[10] Qing Peng, Xu Zhang, Gang Lu, Comput. Mater. Sci. (2010) 769–774.
[11] Y. Yan, J. Gong, Z.G. Zong, Thin Solid Films 518 (2010) 4989–4996.
[12] D.P. Jiang, H.Z. Jia, H.C. Lu, J. Light Electron Opt. (2012), http://dx.doi.org/
10.1016/j.ijleo.2011.11.033.
[13] C.L. Freeman, F. Claeyssens, N.L. Allan, et al., Phys. Rev. Lett. 96 (2006) 066102.
[14] J. Kang, Y. Zhang, Y.H. Wen, et al., Phys. Lett. A 374 (2010) 1054–1058.
[15] H.D. Li, J. Li, Z.G. Wang, et al., J. Chem. Phys. 550 (2012) 130–133.
[16] A.F. Vassilyeva, R.I. Eglitis, E.A. Kotomin, et al., Physica B 405 (2010) 2125–
2127.
[17] A.F. Vassilyeva, R.I. Eglitis, E.A. Kotomin, et al., Cent. Eur. J. Phys. 9 (2011) 515–
518.
[18] L.L. Zhang, P.D. Han, C.L. Zhang, et al., Acta. Phys-Chim. Sin. 27 (2011) 1609–
1614.
[19] A.F. Fix, F.U. Abuova, R.I. Eglitis, et al., Phys. Scripta 86 (2012) 035304.
[20] K. Ramesh Babu, Ch. Bheema Lingam, S. Auluck, et al., J. Solid State Chem. 184
(2011) 343–350.
[21] M. Wojciechowska, M. Zielinski, M. Pietrowski, J. Fluorine Chem. 120 (2003) 1–
11.
[22] M. Segall, P. Lindan, M. Probert, et al., J. Phys.: Condens. Matter 14 (2002)
2717–2744.
[23] J.E. Jaffe, J.A. Snyder, Z. Lin, et al., Phys. Rev. B 62 (2000) 1660–1665.
[24] J.P. Perdew, A. Zunger, Phys. Rev. B 23 (1981) 5048–5079.
[25] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892–7895.
[26] J.S. Lin, A. Qteish, M.C. Payne, et al., Phys. Rev. B 47 (1993) 4174–4180.
[27] D.F. Li, H.Y. Xiao, X.T. Zu, et al., Chin. Phys. B19 (2010) 517–525.
[28] G. Vidal Valat, J.P. Vidal, C.M.E. Zeyen, et al., Acta Cryst. B35 (1979) 1584–1590.
[29] H. Shi, R.I. Eglitis, G. Borstel, J. Phys.: Condens. Matter 18 (2006) 8367–8381.
[30] X.C. Shen, The Spectrum and Optical Property of Semi Conductor, second ed.,
Science Press, Beijing, China, 1992.
[31] M.W. Williams, R.A. Macrae, E.T. Arakawa, J. Appl. Phys. 38 (1967) 1701–1705.
[32] H. Krüger, E. Kemnitz, A. Hertwig, et al., Thin Solid Films 516 (2008) 4175–
4177.
[33] F. Perales, J.M. Herrero, D. Jaque, Opt. Mater. 29 (2007) 783–787.