www.nature.

com/scientificreports

OPEN Analysis of mutual couplings in a

concentric circular ring plasmonic
optical antenna array
Received: 8 May 2017 Guiru Gu1, Lin Li2, Yingjie Zhang2, Thitikorn Kemsri2 & Xuejun Lu 2

Accepted: 10 August 2017

Published: xx xx xxxx In this paper, we report the analysis of a concentric circular ring plasmonic optical antenna (POA) array
using a simple lumped coupled circuit (LCC) model. The currents in the circular rings of the POA array
and their mutual couplings are analyzed using the LCC model. The results agree well with the numerical
®
simulation using CST’s Microwave Studio . The LCC model reveals the mutual couplings between the
antenna rings. It is found that the mutual couplings are not only between the adjacent antenna rings,
but also involve their second (2nd) nearest or farther neighbors. Since the near-fields of the optical
antennas are related to the currents in the optical antennas, the LCC model provides a useful tool for the
analysis of the near-field and their mutual interactions in the circular ring POA array.

Antennas are critical components in transmitting and receiving electromagnetic waves in the r.f., microwaves,
and millimeter-wave spectral regimes. Optical antennas1–11 are their counterparts in the optical spectral regimes
such as visible, near infrared (IR), and middle wave infrared (MWIR) and longwave infrared (LWIR). Optical
antennas and their applications in controlling light sensing and emission properties have been extensively stud-
ied1–5. Depending on the optical antenna structures, localized surface plasmonic resonance (LSPR) modes12, 13 can
be excited in optical antennas1–5, 14–16, which are referred to as plasmonic optical antennas (POAs). POA enhanced
quantum dot infrared photodetectors (QDIPs) have been reported with enhanced photocurrents and directional
antenna gains6, 17, 18. The analysis of the near-field of POAs and their roles in the plasmonic enhancement have
also been reported6.
In a POA array with multiple antenna elements, the near-fields of the POA array can be significantly affected
by the mutual couplings among the antenna elements. In addition, the requirement to satisfy the electro-
magnetic boundary conditions at multiple antenna elements also makes it complicated to analytically calcu-
late their near-fields and the couplings through their near-fields19. Numerical simulation methods such as the
finite-difference-time-domain (FDTD)20, the method of moments (MoM)21, the fast multipole method (FMM)22,
and the integral equation solver23, have been developed to numerically calculate the near-field distributions.
These numerical methods are very effective in numerically calculating the near-fields as well as performing the
design validations and obtaining optimal designs. However, the underlying physics, such as the LSPR modes
excitation, the near-fields and current distributions and their relationships with the interactions of the multiple
antenna elements, are generally hard to unveil from the numerical data.
Since the near-fields can be uniquely defined by the currents in optical antennas due to the uniqueness the-
orem24, 25, it is possible to determine the near-fields from the currents in the antennas. Therefore, the near-fields
and their couplings can be available once the current distributions and their couplings are obtained. Mutual
couplings in r.f. antenna arrays have been extensively studied using equivalent circuit analysis26. Terahertz (THz)
optical antennas have been modeled using the transmission line (TL) theory and the surface current can be cal-
culated using the TL theory27, 28. However, to our best knowledge, there is no report on the analytical modeling of
mutual couplings of POA arrays and their dependence on the POA array elements using a circuit model.
In this letter, we present a simple lumped coupled circuit (LCC) model to analyze a concentric circular ring
POA array. The current distributions of the array elements and their mutual couplings are analyzed using the LCC
model. The results are compared with the numerical simulation data using CST’s Microwave Studio . The two
methods show good agreement. The mutual couplings between the POA antenna elements are found to be not
®
just between adjacent antenna elements, but also involve 2nd nearest or farther antenna elements.

1
Department of Physics, Stonehill College, 320 Washington Street, Easton, MA, 02357, USA. 2Department of
Electrical and Computer Engineering, University of Massachusetts Lowell, One University Avenue, Lowell, MA,
01854, USA. Correspondence and requests for materials should be addressed to X.L. (email: [email protected])

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 1

www.nature.com/scientificreports/

Figure 1. Schematics of a single metallic (gold) circular ring POA. (a) Physical layout; (b) equivalent series RLC
circuit. The incident light is a surface normal plane wave traveling in the -z direction with the E-filed aligned in
the x-direction and the H-field in the y-direction. The magnitude of the E-field is set to 1 V/m.

Results and Discussion

Before modeling of the multiple element concentric circular ring POA array, we first analyze a single circular ring
POA and its equivalent lumped RLC circuit. Figure 1(a) shows the schematic layout of the circular ring POA. It
is a metallic (gold) ring on a GaAs substrate. The thickness of the antenna ring tm is 30 nanometers (nm), and the
thickness of GaAs substrate is td = 0.35 µm. The outer diameter and the width of the metallic ring are labeled as
d, and w, respectively. The incident light is a surface-normal plane wave propagating in the −z direction (i.e. top
illumination).
Figure 1(b) illustrates the equivalent lumped circuit. It is a simple series RLC circuit. The current is I(ω) in the
frequency domain.
®
The single POA was numerically simulated using CST’s Microwave Studio . The incident light is a surface
normal plane wave traveling in the -z direction (i.e. top illumination). The H-field is in the y-direction, and the
E-field is in the x-direction with a magnitude of 1 V/m throughout the paper. Since the circular rings are symmet-
ric, other linear polarizations give the same results. Open boundary conditions were used in all simulations. The
incident light excites LSPR modes in the circular ring POA and generates surface current in the ring.
The resonant condition of the fundamental LSPR mode can be written as:
πd
ksp = π,
2 (1)
where ksp is the wave vector of the surface plasmonic wave. At the resonance, the LSPR mode induced surface
current shows a sinusoidal-type distribution along the ring with a maximum current Im(ω) at the top (i.e. x = 0,
and y = d/2) and bottom (i.e. x = 0, and y = −d/2) of the ring. Figure 2 shows the simulated current distributions
(dots, crosses, and diamonds) in the three individual rings, Ring 1, Ring 2, and Ring 3 with the outer diameters of
1.2 µm, 1.4 µm, and 1.6 µm, respectively. The currents are at their corresponding resonant frequencies of 30 THz,
26 THz and 23 THz, respectively. The widths of the rings and thickness of the substrate are kept at w = 0.05 µm
and td = 0.35 µm, respectively. Under the plane wave incidence, the current distributions in the upper and lower
parts of each ring are symmetric. Therefore, only upper half ring (i.e. angle 0 ≤ φ ≤ 180°) is analyzed. The other
half is the same due to the symmetry.
The currents are also calculated using the sinusoid standing wave formula in a standard half-wave antenna,
i.e. |I| = |Imax|sin(φ). The calculated sinusoidal current distributions (solid lines) are also plotted in Fig. 2. The
current distributions follow the sinusoidal standing wave current distribution in a half-wave antenna.
Figures 3(a) and 3(b) show the simulated real parts Re(I) and imaginary parts Im(I) of the surface currents
Im(ω) of different circular rings (dots, crosses, and diamonds).
The surface current I(ω) in the equivalent lumped RLC circuit can be written as:
V0
I (ω ) = 1
,
R + jωL + jω C (2)
where ω is the angular frequency of the incident light, R is the resistance, L is the inductance, C is the capaci-
tance, and V0 is the induced voltage in the lumped circuit.

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 2

www.nature.com/scientificreports/

Figure 2. Current distributions in the individual rings at their corresponding resonant wavelengths. Dots,
crosses, and diamonds: the current values from the CST simulation; Solid lines: calculated current using
the |I| = |Imax|sin(φ) plots. The resonant current distributions follow the sinusoidal standing wave current
distribution in the half-wave antennas.

Figure 3. Surface currents in individual uncoupled circular rings with different diameters of 1.2 µm, 1.4 µm,
and 1.6 µm for Ring 1, Ring 2, and Ring 3, respectively: (a) Re (I); (b) Im(I). The solid curves are calculation data
using the lumped circuit models, and the points by the “o”, “+” and “◊” are numerical simulation data points
using CST’s Microwave Studio . ®

The resistances R are the Ohmic resistances of the individual rings:

π(R in + Rout )/2
R=ρ ,
2 Wtm , sk (3)
where ρ = 2.44 × 10 Ω-m is the resistivity of gold (Au) , and Rin and Rout are the inner and outer radius of the
−8 29

ring, respectively. tm,sk is the skin depth.

ρ
tm , sk = ,
πfµ (4)
where f is the frequency, and µ is the permeability of the metallic (gold) rings. Since the skin depth tm,sk is
frequency-dependent, the R also changes with the frequency. The 2 factor in Eq. (3) is due to the use of the Imax
in the power loss calculation. The calculated Ohmic resistances are 47.2 Ω, 47.5 Ω, and 51.2 Ω for Ring 1, Ring 2,
and Ring 3 at their resonant frequencies of 30 THz, 26 THz and 23 THz, respectively.
The voltages (V0) of the rings are calculated from the simulated E-field distribution along the rings, i.e.:
180 
V0 = ∫0 E ⋅ φˆRdφ,
(5)
The calculated voltage values are 2.0 × 10−7 V, 2.3 × 10−7 V, and 2.9 × 10−7 V, for Ring 1, Ring 2, and Ring 3,
respectively.
The inductance of the circular rings with the rectangular cross section can be calculated using Eq. (6) given
by Frederick W. Grover30:

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 3

www.nature.com/scientificreports/

Figure 4. (a) layout of the concentric circular POA array; (b) Equivalent circuit for each ring. The extra
inductors in each ring correspond to the couplings from the other rings.

R (Ω) V0 (10−7 V) L (10−13 H) C (10−17 F)

Ring 1 2 3 1 2 3 1 2 3 1 2 3
Calculated 47.2 47.5 51.2 2.0 2.3 2.9 6.0 7.8 9.7 5.3 5.0 5.2
Curve-fitting 47.2 47.5 51.2 2.1 2.5 2.9 6.0 7.5 8.2 4.7 5.0 5.8

Table 1. Comparison of calculated and curve-fitting values. Note that in the calculation, the circular ring is
treated as a perfect transmission line. No current distribution is considered across the width of the ring, whereas
in the CST numerical simulation the current nonuniformity is counted. This causes the differences between the
calcuted R, V0, L and C vlaues and those from the curve-fitting vlaues. As shown in Table 1, the values are quite
close.

 2R 
Lu = 0.002π 2 RK ′ ,
 t  (6)
m

where Lu is the inductance in microhenries (µH), R and tm are in centimeters (cm), and K’ is the is a dimensionless
factor that depends upon the (2 R/tm) ratio and the radius of the ring antennas. K’ values are given by Frederick
W. Grover30. The inductance L of the half ring is thus:
 2R  RK ′
L = 0.002π 2  ,
 t  2 (7)
m

The calculated inductances are 6.0 × 10 −13

H, 7.8 × 10 −13
H, and 9.7 × 10 −13
H, for Ring 1, Ring 2, and Ring
3, respectively.
The capacitance C can be obtained by:
T /2
Q ∫ Idt
C= = 0 ,
V0 V0 (8)
where T/2 is the half period of the incidence wave, and I is the surface current from the CST simulation. The cal-
culated capacitances are 5.3 × 10−17 F, 5.0 × 10−17 F, and 5.2 × 10−17 F, for Ring 1, Ring 2, and Ring 3, respectively.
The current distributions are also fitted using Eq. (2) and plotted in Fig. 3 (solid curves) together with those
from the CST simulation (dots, crosses, and diamonds). The comparison of the calculated R, V0, L and C values
and those from the curve-fitting are listed in Table 1.
After obtaining the parameters R, L, C for individual single rings, we then analyze the mutual couplings in a
concentric circular ring POA array. Figure 4(a) shows the schematic layout of a three-ring POA array. The outer
diameters of the rings are labeled as d1, d2, et al., and the widths of the rings are w1, w2, et al. Figure 4(b) shows
the equivalent circuits for the rings of the POA array. The extra inductors in the equivalent circuit of each ring
correspond to the mutual couplings from other rings.

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 4

www.nature.com/scientificreports/

Figure 5. Comparison of the simulated and calculated currents in a two coupled ring POA array: (a) Re(I); (b)
Im(I). The solid curves are the calculated currents using Eq. (11). The points by the “o” and “+” are numerical
simulation data points using CST’s Microwave Studio . ®

Coupled rings A B C

M12 (Calculated) 2.13 × 10−13 H 2.24 × 10−13 H 2.45 × 10−13 H

M12 (Curve-fitting) 2.07 × 10 −13
H 2.22 × 10 −13
H 2.47 × 10−13 H

Table 2. Mutual inductances M12 of different coupled rings.

In a matrix form the currents in coupled rings can be written as:

 1   jωM12 jωM1n 
 0 …0  0 … 
Z   Z1 Z1 
 I1   1   V10   I 
   1     jωM21 jωM2n   1 
 I2  =  0 … 0  V20  +  0 …   I2  ,
…  Z2   …   Z2 Z 2  … 
  … … … …  V   … … … …   I 
 In      n 
 0 1   n0   jωMn1 jωMn2 
 0 …   … 0 
 Zn   Zn Zn  (9)

−1
 −jωM12 −jωM1n   1 
 1 …   0 0 …
 Z1 Z1   Z1 
 I1       V10 
   −jωM21 −jωM2n   1  
 I2  =  1 …   0 … 0  V20  ,
…  Z 2 Z2   Z2 …
   … … … …  … … … …  V 
 In     
 −jωMn1 −jωMn2   0 1   n0 
 … 1   0 … 
 Zn Zn   Zn  (10)

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 5

www.nature.com/scientificreports/

Figure 6. Comparison of the simulated and calculated currents in a three-ring POA array: (a) Re(I); (b) Im(I).
The solid curves are the calculated currents using Eq. (11). The points by the “o”, “+” and “◊” are numerical
simulation data points using CST’s Microwave Studio . ®

−1
 I   Z1 −jωM12 … −jωM1n   V10 
 1    
 I2  = −jωM21 Z2 … −jωM2n  V20  ,
…  … … … …  …
    
 In  −jωMn1 −jωMn2 … Zn  Vn0  (11)
1
where Zn = R n + jωLn + is the total impedance of an individual uncoupled ring.
jωCn
Two different models are investigated to calculate the mutual induces. The first one is the mutual inductance
of circular coils31 using Nagaoka’s formula32:

M = µ0 Aa [4πq3/2 (1 + ε)], (12)

where A and a are the radii of the two circular coils and q and ε are the geometric parameters of the circular coils.
5 15
l l  l 
q= +   +   + … ,
2  2   2  (13)

1− k′
l= ,
1− k′ (14)

(A − a)2 + d 2
k′ = ,
(A + a)2 + d 2 (15)
where d is the distance between the two circular coils, and d = 0 for the concentric circular rings on the same
plane.

ε = 2q4 − 4q6 + 9q8 − 12q10 … , (16)

The mutual inductances calculated using the Nagaoka’s formula do not agree well with the simulation results.
32

This is possibly due to the plane-wave induced non-uniform current distributions.

The second method is the mutual inductances of two parallel wires given by Rosa33:

µ0   2leff  
M= (2leff )ln  − 1 + s  ,
  leff 
4π   s   (17)
where µ0 is the permittivity of vacuum, leff is the effective length of a wire, and s is the separation of the two wires.
The leff can be written as:
π Rout
leff = ,
2 2 (18)
where the factor 2 in the denominator is for the half-circle due to the plane-wave incidence induced symmetry,
and the factor 2 accounts for the sinusoidal current distribution.

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 6

www.nature.com/scientificreports/

Mutual inductances M12 M23 M13

Calculated 2.13 × 10−13 H 2.91 × 10−13 H 2.34×10−13 H
Curve-fitting 2.07 × 10−13 H 2.79 × 10−13 H 2.28×10−13 H

Table 3. Mutual inductances of different coupled rings.

Figures 5(a) and 5(b) show the simulated (points) real Re(I) and imaginary Im(I) parts of the surface currents
Im(ω) of two coupled circular rings compared with the calculated values using Eq. (5) (solid curves). The outer
diameters of Ring 1, and Ring 2 are, 1.2 µm, 1.4 µm, respectively. The widths of the rings and thickness of the
substrate are kept the same.
Table 2 lists the calculated mutual inductances M12 of different coupled rings using Eq. (17) compared with the
mutual inductances from the curve-fitting.
The results calculated from the parallel wire model using Eq. (17) agree well with the curve-fitting. This indi-
cates that under a plane wave illumination, the current distribution in a ring is similar to that in a wire transmis-
sion line.
Figures 6(a) and 6(b) show the calculated real and imaginary parts of currents (solid curves) in a three-ring
POA array from Eqs (10) and (11) compared with the numerical simulation (circles, crosses, and diamonds).
Table 3 lists the calculated mutual inductances using Eq. (17) compared with the values from the curve-fitting.
The calculations agree well with the numerical simulation. The mutual couplings from the 2nd nearest neigh-
bors (i.e. M13, M31) are on the same orders as the nearest couplings.

Conclusion
In conclusion, we develop an LCC model for the analysis of the mutual coupling in a concentric circular ring
POA array. The current distributions in the circular rings and their mutual couplings are analyzed using the LCC
model. The analytical calculations agree well the numerical simulation. The LCC model reveals the underlying
mutual couplings between the rings in the POA array. It is found that the mutual couplings from the 2nd nearest
neighbors are not negligible. The LCC model provides a useful tool for the analysis of the near-field and their
couplings in the circular ring POA array.

References
1. Novotny, L. & van Hulst, N. Antennas for light. Nat Photon 5, 83–90 (2011).
2. Bharadwaj, P., Deutsch, B. & Novotny, L. Optical Antennas. Adv. Opt. Photon. 1, 438–483, doi:10.1364/AOP.1.000438 (2009).
3. Krasnok, A. E. et al. Optical nanoantennas. Physics-Uspekhi 56, 539 (2013).
4. Kalkbrenner, T. et al. Optical Microscopy via Spectral Modifications of a Nanoantenna. Physical Review Letters 95, 200801 (2005).
5. Giannini, V., Fernández-Domínguez, A. I., Heck, S. C. & Maier, S. A. Plasmonic Nanoantennas: Fundamentals and Their Use in
Controlling the Radiative Properties of Nanoemitters. Chemical Reviews 111, 3888–3912, doi:10.1021/cr1002672 (2011).
6. Gu, G., Vaillancourt, J. & Lu, X. Analysis of near-field components of a plasmonic optical antenna and their contribution to quantum
dot infrared photodetector enhancement. Optics Express 22, 24970–24976, doi:10.1364/OE.22.024970 (2014).
7. Yu, N. et al. Light Propagation with Phase Discontinuities: Generalized Laws of Reflection and Refraction. Science 334, 333–337,
doi:10.1126/science.1210713 (2011).
8. Rui, G., Abeysinghe, D. C., Nelson, R. L. & Zhan, Q. Demonstration of beam steering via dipole-coupled plasmonic spiral antenna.
Sci. Rep. 3, doi:10.1038/srep02237 http://www.nature.com/srep/2013/130719/srep02237/abs/srep02237.html#supplementary-
information (2013).
9. Kosako, T., Kadoya, Y. & Hofmann, H. F. Directional control of light by a nano-optical Yagi-Uda antenna. Nat Photon 4, 312–315
(2010).
10. Taminiau, T. H., Stefani, F. D. & van Hulst, N. F. Enhanced directional excitation and emission of single emitters by a nano-optical
Yagi-Uda antenna. Opt. Express 16, 10858–10866, doi:10.1364/OE.16.010858 (2008).
11. Ramezani, M. et al. Hybrid Semiconductor Nanowire–Metallic Yagi-Uda Antennas. Nano Letters, doi:10.1021/acs.nanolett.5b00565
(2015).
12. Ritchie, R. H. Plasma Losses by Fast Electrons in Thin Films. Physical Review 106, 874–881 (1957).
13. Raether, H. Surface plasmons on smooth surfaces. (Springer, 1988).
14. Chang, C.-C. et al. A surface plasmon enhanced infrared photodetector based on InAs quantum dots. Nano letters 10, 1704–1709
(2010).
15. Kosako, T., Kadoya, Y. & Hofmann, H. F. Directional control of light by a nano-optical Yagi–Uda antenna. Nature Photonics 4,
312–315 (2010).
16. Middlebrook, C. T., Krenz, P. M., Lail, B. A. & Boreman, G. D. Infrared phased‐array antenna. Microwave and Optical Technology
Letters 50, 719–723 (2008).
17. Neda, M., Guiru, G. & Xuejun, L. A plasmonic dipole optical antenna coupled quantum dot infrared photodetector. Journal of
Physics D: Applied Physics 48, 475102 (2015).
18. Kemsri, T. et al. Angular-dependent photodetection enhancement by a metallic circular disk optical antenna. AIP Advances 7,
025013, doi:10.1063/1.4975705 (2017).
19. Zhang, S. et al. Anti-Hermitian Plasmon Coupling of an Array of Gold Thin-Film Antennas for Controlling Light at the Nanoscale.
Physical Review Letters 109, 193902 (2012).
20. Kane, Y. Numerical solution of initial boundary value problems involving maxwell’s equations in isotropic media. IEEE Transactions
on Antennas and Propagation 14, 302–307, doi:10.1109/TAP.1966.1138693 (1966).
21. Gibson, W. C. The method of moments in electromagnetics. Vol. 1 (Chapman & Hall/CRC London, UK, 2008).
22. Engheta, N., Murphy, W. D., Rokhlin, V. & Vassiliou, M. S. The fast multipole method (FMM) for electromagnetic scattering
problems. IEEE Transactions on Antennas and Propagation 40, 634–641, doi:10.1109/8.144597 (1992).
23. Bleszynski, E., Bleszynski, M. & Jaroszewicz, T. In Proceedings of IEEE Antennas and Propagation Society International Symposium
and URSI National Radio Science Meeting. vol. 411 416–419 (1994).
24. Jackson, J. D. Classical Electrodynamics. Third Edition edn, 411 (John Wiley & Sons, 1998).
25. Qingxin, C. & Changhong, L. The uniqueness theorem of electromagnetic fields in lossless regions. IEEE Transactions on Antennas
and Propagation 41, 245–246, doi:10.1109/8.214620 (1993).

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 7

www.nature.com/scientificreports/

26. Gupta, I. & Ksienski, A. Effect of mutual coupling on the performance of adaptive arrays. IEEE Transactions on Antennas and
Propagation 31, 785–791, doi:10.1109/TAP.1983.1143128 (1983).
27. Balanis, C. A. Antenna Theory Analysis and Design Third Edition edn, 137 (John Wiley & Son, 2005).
28. Wen, Q.-Y. et al. Transmission line model and fields analysis of metamaterial absorber in the terahertz band. Optics Express 17,
20256–20265, doi:10.1364/OE.17.020256 (2009).
29. Johnson, J. D. C. K. W. Physics (4th Edition). Vol. Volume 1 591 (Wiley, 1997).
30. Grover, F. W. & United, S. Tables for the calculation of the inductance of circular coils of rectangular cross section. (U.S. Dept. of
Commerce, Bureau of Standards: U.S. Govt. Print. Off. 1922).
31. Rosa, E. B. & Cohen, L. The mutual inductance of two circular coaxial coils of rectangular section. ([Govt. Print. Off.], 1906).
32. Nagaoka, H. The Inductance Coefficients of Solenoids. The Journal of the College of Science, Imperial University of Tokyo, Japan 27,
1–33 (1909).
33. Rosa, E. B. & Standards, U. S. N. B. o. The self and mutual inductances of linear conductors. (U.S. Dept. of Commerce and Labor,
Bureau of Standards 1908).

Acknowledgements
Dr. G. Gu’s work is supported by Stonehill College’s internal funding. Y. Zhang and L. Li are supported by the Air
Force Office of Scientific Research (AFOSR) under the contract No. FA9550-12-1-0176. Xuejun Lu is also a co-
founder of Applied NanoFemto Technologies LLC.

Author Contributions
X. Lu developed the concept and designed the simulation. G. Gu and X. Lu performed the numerical simulation.
X. Lu also performed analytical analysis and simulation. X. Lu, L. Li, Y. Zhang, and T. Kemsri checked the
simulation and reviewed the data. X. Lu and G. Gu drafted the manuscript. All authors reviewed and approved
the manuscript.

Additional Information
Competing Interests: The authors declare that they have no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

© The Author(s) 2017

Scientific REPOrTS | 7: 10996 | DOI:10.1038/s41598-017-10690-7 8