pubs.acs.

org/JPCC Article

Large Exciton-Driven Linear and Nonlinear Optical Processes in

Monolayers of Nitrogen Arsenide and Nitrogen Antimonide
Miroslav Kolos, Rekha Verma, František Karlický,* and Sitangshu Bhattacharya*
Cite This: J. Phys. Chem. C 2022, 126, 14931−14959 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: We demonstrate that atomically thin nitrogen-based binary

group V wide band gap indirect semiconductors (β-NX (X = As, Sb))
embody 3-fold valley degenerated band structures and strong linear and
nonlinear optical activities. Contrary to NAs, the fundamental optical
Downloaded via 223.189.238.161 on September 26, 2023 at 10:47:07 (UTC).

absorption in NSb is quite resilient toward the temperature variations

between 0 and 450 K. In the absence of lattice vibrations, exciton in NSb
is, however, less strongly bound (1.30 eV) compared to the tighter case in
NAs (1.62 eV), leading to a more delocalized Mott−Wannier-type texture.
The inhomogeneous excitonic line widths for both monolayers within
these temperatures are within the 100−400 meV range. The most
significant nonlinear second harmonic optical coefficients (2ω reso-
nances) in NAs and NSb monolayers are obtained as ∼636 and 270 pm/V
at 3.2 eV (387 nm) and 2.6 eV (477 nm), respectively. We also present a
detailed analysis of in-plane biaxial strain on these structures and found
that tensile strain dynamically stabilizes structures with no loss in absorbance spectra (and does not influence exciton binding energy
in NAs). The nonlinear coefficient also significantly improves at the two most important 810 and 1560 nm wavelengths compared to
the cases offered by the monolayer transitional metal dichalcogenides. Our analyses originate from a fully ab initio G0W0+Bethe-
Salpeter excited-state theory. The temperature-dependent linear absorption spectra are evaluated by including the electron−phonon
self-energies, whereas the nonlinear spectra are treated using the modern theory of polarization within the same perturbative
approach. Our reference findings provide important advanced characteristics for applications of two-dimensional β-NX (X = As, Sb)
materials and should motivate further theoretical and experimental investigations of these interesting materials.

■ INTRODUCTION
Excitons are the essential ingredients responsible for optical
excitation in materials. Fundamentally pictured as a quasi-
particle in the framework of many-body interactions, an
exciton is an electron−hole pair bound together by a
Coulombic force of attraction. The pair can be generated
when the electron is kicked out by, for example, photons from
the ground-state orbital. How long this pair will hold together
depends on its binding energy. The exciton so formed
resembles an atomic hydrogen, and likewise, distinct optical
spectra are therefore expected. Similar to the hydrogen
electronic energy quantization, the exciton Coulombic energy
lowers the width of the direct band gap of the material. In
principle, the one-electron picture obtained from the standard
generalized gradient approximation (GGA) density functional
theory (DFT) does not account for such interactions. Instead,
excitons are a two-body dynamically correlated system, and
thus, to understand their energy and lifetime, many-body
methods must be used, such as the Bethe−Salpeter equation
(BSE) over the single-shot GW corrections.1−4 The solution of
the former can directly access the information about the
optical absorption spectrum and the bound-state electron−
hole energies, while the latter fixes the missing energy
© 2022 The Authors. Published by
American Chemical Society
14931
eigenvalues that the standard DFT rigorously struggles for.
Nowadays, these methods have become a standard routine task
to understand the excitonic-driven optical properties in bulk to
atomically thin layers,5−12 including comparison with the
quantum Monte Carlo calculations.13
When the intensity of the light becomes very high, the
semiconductor optical properties tune to the nonlinear regime.
One can then observe frequency overtones like second, third,
and other higher harmonics.14 For example, suppose the crystal
structure is such that both the inversion and time-reversal
symmetries are broken. In that case, one can then show that
the induced time-varying electric dipoles do not cancel out on
an average, instead of leading to a minimum detected second
harmonic signal as a nonlinear one.15 In order to capture these
nonlinear dynamics at an atomic scale in a purely ab initio way,
Received: May 30, 2022
Revised: August 5, 2022
Published: August 25, 2022
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
a time-dependent DFT (TD-DFT) is therefore needed. This
TD-DFT formalism was initially developed by Runge and
Gross,16 who numerically integrated the time-dependent
Kohn−Sham equation directly in the real-time domain. The
accuracy of such calculations, however, depends on the level of
approximation used in the exchange-correlation kernel of the
time-dependent Hamiltonian.3,17 The kernel hierarchy starts
from an independent particle approximation (only the Kohn−
Sham eigenvalues) to a time-dependent Hartree (arising due to
local field effects resulting in system inhomogeneities) to an
adiabatic local density approximation (quasi-statically space
and time-varying density) and to the most advanced time-
dependent screen exchange (electron−hole correlation is
screen exchanged). This last kernel particularly contains the
excitonic dynamics and is also known as the time-dependent
BSE (TD-BSE) or the real-time BSE.18 The modern theory of
polarization developed by Kingsmith and Vanderbilt 19
prescribes a way to obtain the coupling between the time-
varying electric field and the Bloch electrons in an extended
system that is required in the time-dependent Hamiltonian.
The Berry’s phase change around the complete Brillouin zone
(BZ) leads to the macroscopic time-varying polarization, and
therefore, the TD-DFT equation of motion with such various
kernel hierarchy can be solved. The nonlinear coefficients can
then be obtained by imposing a suitable cutoff in the
polarization Fourier series.20
There is, however, a fundamental part missing which needs
to be addressed. The BSE so developed was accounted only for
the frozen atom condition; i.e., atoms were assumed to be fixed
in space or, in other words, in the absence of lattice vibrations.
As a result, the pole of the dielectric function (i.e., it is
imaginary part) is required to be manually tuned so that the
broadening in the absorption spectrum matches with that of
the experimental spectrum. The role of this broadening carries
significant importance since it directly confirms Heisenberg’s
uncertainty principle. The absence of the broadening would
simply mean that the exciton lives forever in that state,
characterized by a sharp delta-like spectrum. One should note
that, in principle, this thermal broadening should enter as an
exciton−phonon coupling matrix in the BSE. However, the
current theoretical strategy addresses this formidable challenge
by adopting its entry as an electron−phonon coupling matrix
into the BSE Hamiltonian.21 This first-cut approximation
approach quintessentially removes the manual tuning of the
broadening parameter but is also quite successful in justifying
the temperature-dependent absorption spectra, the underlying
excitonic binding energies, and lifetimes in a variety of
dimensional systems like monolayer MoS2,22 WSe2,23 h-BN
(both bulk24,25 and monolayer26), monolayer β-phase NP,27
bulk zinc-blende GaN,28 carbon-nanotube,29 etc., demonstrat-
ing a remarkable accuracy with experimental observations. The
temperature-dependent analyses of the optical properties are
therefore seems more plausible on one hand; however, on the
other hand, it makes the entire computation an expensive affair
with rigorous cascaded process-flow. Due to these trade-offs,
such calculations are rarely reported in the literature.
Two quantities are greatly required in order to observe a
strong exciton-driven optical properties in semiconductors: a
weak dielectric screening and crystalline noncentrosymmetry
(i.e., lack of inversion symmetry). Whether or not spin−orbit
coupling (SoC) will originate depends on the availability of an
additional crystalline time-reversal symmetry. The latter leads
to the coupling between spin and momentum in the valleys
14932
pubs.acs.org/JPCC Article
and defines the momentum forbidden dark and bright states.
For example, the monolayer transitional metal dichalcogenide
(TMDC) WSe2 possesses a time-reversal symmetry and thus
has a large spin−orbit splitting in its valence band at K point.
The bright (dark) exciton is thus formed when an intravalley
(intervalley) electron pairs with a hole having a parallel
(antiparallel) spin.8,30 In recent years, there have been many
benchmark strategic experiments to push the binding energies
and nonlinear coefficients to their limits (for example, see
Table 4 below). The excitons in two-dimensional semi-
conductors possess rather larger binding energies and oscillator
strengths due to an in-plane quantum confinement than their
bulk structures. In the latter case, the electric field is properly
screened out by the surrounding charges, while in monolayers,
it is not. Therefore, an intense Coulombic interaction develops,
leading to a strongly bounded pair with enhanced optical
absorption. There is, however, another justification as to why
such an atomic layer should absorb a large amount of light.
Castro Neto et al.31 showed that the strong peaks in the optical
conductivity could also be justified from the ground-state
electronic band structure. They observed that the peaks
correspond to those points in the BZ where the transition
bands run parallel to each other. This is known as band-
nesting. It was recently demonstrated that this band-nesting
could also justify the nonlinear second harmonic generation
(SHG) signals in uniaxially tensile strained TMDCs.32
Nevertheless, shifting the focus from the traditional planar
monolayers, Taheri et al.33 demonstrated that group V
monolayers based on nitrogen (NX, where X = P, As, and
Sb) in their β-phase exhibit thermodynamically stable buckled
structures. Using a purely ab initio technique (along with a
temperature dependent BSE calculation), it was recently
shown27 that this β-phase nitrogen phosphide (NP) monolayer
can exhibit a very high excitonic binding energy and strong
absorbance spectra in the visible region. The nonlinear
coefficients SHG along with the third harmonic generation
(THG) were also found to be very high compared to the
traditional TMDs and monolayer transitional metal mono-
chalcogenides (TMMCs). In this work, we extend the
roadmap of exciton-driven optical activities in monolayers of
β-phase nitrogen arsenide (NAs) and nitrogen antimony
(NSb). We ask the following: (1) How an electron taking part
in the formation of an exciton relaxes from the conduction
band in the presence of lattice vibrations, (2) how temperature
controls the excitonic nonradiative line widths via exciton−
phonon coupling, (3) can band-nesting in these cases justify
the strong absorption, (4) are the excitons of the Mott−
Wannier or Frenkel type, (5) what are the SHG coefficients,
and (6) can an in-plane biaxial strain improve the binding
energies and SHG coefficients? We respond to these questions
in a purely ab initio way. We use the many-body perturbation
theory at the level of BSE to justify the excitonic-driven linear
spectra. All temperature corrections are computed using
density functional perturbation theory (DFPT) on the top of
the ground-state DFT eigenvalues. The nonlinear SHG
coefficient is computed by solving the TD-BSE in the real-
time domain. Berry’s phase polarization is used to compute the
couplings between Bloch electrons and the time-dependent
electric field. In a hope to improve the optical properties, we
extend our analyses in the presence of an in-plane biaxial strain
also, and we report the optimized thermodynamic stability,
energy gaps, binding energies, absorption and SHG spectra for
both compressive and tensile strains. All strain-dependent
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Table 1. Electronic Ab Initio Energy Comparisons in β-NX (X = P, As, Sb)a,b

Material (ML) a (Å) z (Å) Indirect gap (eV) Direct gap (eV) Δ0 (meV) ref
β-NP 2.73 0.87 1.79 (PBE), 4.15 (G0W0) 4.95 (G0W0, M) Kolos et al.27
2.78 1.48 (PBE), 2.64 (HSE06) Yu et al.43
2.73 0.86 Taheri et al.33
β-NAs 2.79 1.00 2.36 (PBE, K Γ ↔ M) 3.06 (PBE, M Γ) 44 (PBE, Γ) this work
4.51 (G0W0, K Γ ↔ M) 5.00 (G0W0, M Γ) this work
3.03 1.70 (PBE), 2.76 (HSE06) Yu et al.43
2.98 0.97 Taheri et al.33
β-NSb 3.00 1.13 2.06 (PBE, K Γ ↔ M) 2.52 (PBE, K) 57 (PBE, Γ) this work
3.70 (G0W0, Γ ↔ K) 4.80 (G0W0, K) this work
3.58 (PBE, Γ) this work
4.45 (G0W0, Γ) this work
2.88 (PBE, M Γ) this work
4.77 (G0W0, M Γ) this work
3.33 1.35 (PBE), 2.26 (HSE06) Yu et al.43
3.28 1.02 Taheri et al.33
a
Blank spaces denotes values not reported. ba and z are the in-plane lattice parameter and buckling heights, respectively. Δ0 is the spin-orbit
splitting energy. K Γ and M Γ symbols mean specific k-points of gap location on the K-Γ and M-Γ paths, respectively.

calculations follow the same extensive process flow as Table 3. DFT and Zero-Point Gap (Using the Dynamical
discussed above. Our results on the electronic and optical HAC Theory) in Various Materials Including This Work
properties of β-phase NAs and NSb are summarized in Tables (ZB = zinc-blende)
1, 2, and 6. A sizable comparison with other monolayers and
Materials DFT (eV) Zero-point gap (meV)

Table 2. Ab Initio Lattice Vibration and Formation Energy Bulk BP 0.53 (HSE06)62 20 (PBE)63
Comparisons in β-NX (X = P, As, Sb)a ML WSe2 1.26 (PBE)23 31 (PBE)23
trans-polyacetylene 0.6345 (LDA)60 40 (LDA)60
ωLO − Formation ML MoS2 1.69 (LDA)64 75 (LDA)22
Material ωTO|Γ ωZO|Γ energy (eV/ dEgidT
(ML) (cm−1) (cm−1) atom) (meVK−1) ref ML β-NSba 2.06 (PBE) 80 (PBE)
ML β-NAsa 2.36 (PBE) 105 (PBE)
β-NP 566 672 6.67 Kolos et
al.27 Si 0.46 (LDA)65 123 (LDA)66
480 570 6.23 Yu et ML β-NP 1.79 (PBE)27 200 (PBE)27
al.43 SiC 2.54 (LDA)67 223 (LDA)66
500, 534 600 Taheri et Bulk-ZB BN 6.4 ± 0.5c,68 366 (LDA)36
al.33 Polyethylene 5.78 (LDA)60 280 (LDA)60
β-NAs 566 672 10.7 −0.18 this work Diamond 5.5 (LDA)69 622 (LDA)59
480 570 5.58 Yu et Bulk-ZB LiF 10.9b,70 718 (LDA)36
al.43
a
500, 534 600 Taheri et This work. bWith Hartree−Fock corrections. cExperimental value.
al.33
β-NSb 523 596 9.3 −0.47 this work
460 470 5.31 Yu et down the Coulombic interactions due to periodic images along
al.43 the out-of-plane direction, a vacuum-slab-vacuum structure
480, 500 517 Taheri et was created with a vacuum separation extending to 20 Å on
al.33 either side. Figure 1 below demonstrates the schematic view of
a
ωLO and ωTO are the zone-centre longitudinal and transverse optical such a monolayer. The optimized in-plane and out-of-plane
in-plane frequencies, while ωZO is the optical out-of plane frequency. lattice constants were found to be (2.79, 1.00)Å for NAs and
dEig (3.00, 1.13)Å for NSb, respectively. These optimizations were
shows the temperature dependent indirect band-gap.
dT
done by using the open-source Quantum Espresso DFT
package.34 Fully relativistic norm-conserving pseudopotentials
bulks are provided in Tables 3, 4, and 5. In addition, we also were first generated35 which also included nonlinear core-
put the convergence and supporting figures in the Supporting corrections in N (core states: [He], valence states 2s and 2p),
Information. What follows, in the Methodologies section, we As (core states [Ar], valence states 3d, 4s and 4p), and Sb
provide a lucid picture of our computational details for the (core states [Kr], valence states 4d, 5s and 5p). A Perdew−
ground-state, excited-state, and nonlinear analysis. In the Burke−Ernzerhof (PBE) exchange-correlation functional was
Results and Discussion, we justify our results together with used rather than a local density approximation (LDA) since
summarizing our outcomes.

■
METHODOLOGIES
Ground-State Calculations. β-Phase NAs and NSb
monolayers are thermodynamically stable in their non-co-
planar structures. A unit cell composed of two atoms was thus
made, leading to a C3v 3m point group 3-fold symmetry. To cut
14933
the latter is known to underestimate the electron−phonon
couplings by 30%.36 A kinetic cutoff energy of 80 Ry for all the
atomic species was found to be sufficient (see Figure S1 in the
Supporting Information) to achieve energy convergences. A
plane-wave basis set was then used, and the cell was finally
allowed to undergo energy minimization at Γ-centered 12 × 12
× 1 Monkhorst−Pack grid constraining all forces and energy
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Table 4. Comparison of Experimental Quasi-Particle Direct Band-Gap, Fundamental Exciton Binding Energy (BE), Line
Width (LW), and Nonlinear SHG Coefficient (at Excitation Wavelength in nm) in Single and Few Atomic Layers, including
This Work (rt = room temperature 300 K)a
Material Substrate rt Band-gap (eV) rt BE (eV) rt LW (meV) |χ(2)|(λnm) (pm/V)
72 72 72
BP (4L) PDMS 0.84 0.139 20
BP (29L) SiO2/Si
ML WSe2 SiO2/Si(e) 2.0273 0.3773 5074 16.5 (1560)75
Sapphire 1.89b,76 0.240b,76
Au 1.75b,76 0.140b,76
SiO2/Si(e) 2.63(4−300 K)77 0.887(4−300 K)77
hBN encap (comm)78 3.9−4.2(4 K)
32−34(292 K)
Quartz(e) ∼13 (160 K)79
Si/SiO2/(e) 1 × 104 (816)80
ML MoS2 Fused silica(e) 2.17 ± 0.181
0.31 ± 0.04 81
160 (810)82
SiO2/Si(e) 1 × 105 (810)83
Sapphireb 2.1176 0.24076
Si/SiO2(c) 684
Si/SiO2(c) 5 × 103 (816)83
amorphous quartz(e) 30−100 (680−1080)85
Au 1.9b,76 0.090b,76
hBN/Fused silica(e) 2.47 ± 0.0886 0.44 ± 0.08f,86
isolatedc 4422
HOPG(c) 2.15 ± 0.06 (77 K) 87
0.2 (77 K) 87

HOPG(c) 2.15 ± 0.1 (77 K)88 ∼0.3 (77 K)88

hBN encap (comm)78 2.0−4.5 (4 K)
44−46 (292 K)
hBN encap (CPT)78 3.9−5.0 (4 K)
44−49 (292 K)
ML MoSe2 Si/SiO2(e) 0.1 (5 K)89 4090 37(1560)75
91
6H-SiC(0001)(m) 1.58 (40 K)
hBN encap (VPT)78 2.4−4.9 (4 K)
34−36 (292 K)
hBN encap(m) 6.6 (4 K)92
Si/SiO2(c) 50 (1200−1800)93
ML MoTe2 Si/SiO2(e) 0.540 (10 K)94
ML WS2 suspended and Si/SiO2(c)b 9 × 103(832)95
SiO2/Si(e) 2.14 ± 0.04 (5 K) 0.32 ± 0.04 (5 K)
Si/SiO2(e) 16.2 (1560)75
SiO2(c) 2.31−2.5396 0.26−0.4896
SiO2/Si(e) 2.7397 0.71 ± 0.0197
SiO2/Si(e) 3.01 (4−300 K)77 0.929 (4−300 K)77
Fused silica(e) 2.33 ± 0.0586 0.32 ± 0.0586
Fused quartz(e) 2.38 ± 0.0681 0.360 ± 0.06081
Si/SiO2(e) 2490
hBN encap (comm)78 4.3−4.8 (4 K)
23−25 (292 K)
Sapphire(c) 5.96 (10 K)98
ML InSe hBN(e) 2.9 (100 K)99
hBN encap(e) 223 ± 138 (810)100
at 300 K
ML GaSe Si/SiO2(c) 2.4 × 103101
ML hBN Fused silica(e) 10 (810)82
Graphite(m) 6.1102 1.9102 32 (10 K)−38 (300 K)102
isolatedc26 7.36 1.81 97 (0 K)−260 (300 K)
hBN (110 ± 2L) SiO2(e) 42 (810)103
ML β-NPc27 isolated 4.92 2e
151 800 (364)
ML β-NAscd isolated 5.21 (M Γ) 1.63e 278 636 (387)
52 (810)
14 (1560)
ML β-NSbcd isolated 4.45 (Γ) 258 270 (477)
4.92 (M Γ) 1.74e 31 (810)
21 (1560)

14934 https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Table 4. continued

a
Symbols: c = chemical vapor deposition, e = mechanical exfoliation, m = molecular beam epitaxy, L = layer, PDMS = polydimethylsiloxane, rt =
room temperature (300 K). b= temperature not mentioned. c= ab initio. d=this work, comm = commercial, encap = encapsulation, CV(P)T =
chemical vapor(pressure) transport and blank space = not reported. Data at other than 300 K are explicitly mentioned. eThe rt-binding energies are
obtained after adding the respective G0W0 corrections to the temperature dependent PBE direct gaps. fObtained by analyzing the B-excitonic
transition state.

Table 5. Experimental Excitonic Line Width Parameters in Monolayer NAs and NSb together with Conventional TMDCsa
Material (ML) Substrate Γ0(0K) (meV) γAC (μeV/K) γOP (meV) ωOP (meV) temp range (K)
90
WS2 SiO2/Si(e) 9.2 28 6.5 20 0−300
MoSe2 SiO2/Si(e)90 4.3 91 15.6 30 0−300
WSe2 Sapphire(c)106 1.6 ± 0.3 60 5−50
Quartz(e)79 7.3 4 39 31 40
MoS2 hBN/ML MoS2/hBN78 4 ± 0.2 70 ± 5 42.6 ± 1.2 24.2 4−300
quartz/suspended107 4.5 45 5−40
β-NP†27 isolated 151 33 90 74.4 0−1000
β-NAs†§ isolated 105 0.4 36 71.9 0−450
β-NSb†§ isolated 129 0.18 52 76.9 0−450
a
The notations are c = chemical vapor deposition, e = mechanical exfoliation, † = ab initio, §=this work and blank space = not mentioned.

Table 6. Optical Phonon Mode Comparisons (in cm−1) at Γ in β-NX (X = As, Sb) at Various Strain Levelsa
Strain Material Modes 2% 4% 6% 8% 10%
NAs E 519.4 476.6 436.8 399.5 365.5
Tensile A1 640.4 612.1 585.9 561.9 539.9
NSb E 477.6 435.4 396.5 360.3 326.5
A1 563.3 533.9 507.2 482.5 459.9
NAs E 617.2 672.9 733.1 799.3 871.8
Compressive A1 706.9 745.0 786.6 832.9 884.1
NSb E 573.4 627.4 685.9 749.8 815.9
A1 632.0 671.6 715.1 763.4 818.6
a
Both E and A1 modes are infra-red and Raman active in all cases.

Using the same energy threshold, the dynamical matrices and

Figure 1. Optimized in-plane and out-of-plane crystal structure of β-
phase NX (X = As, N). Both these structures possess a trigonal
symmetry and belong to space group 156 and point group 3m.
cut-offs below 10−5 Ry/Bohr and 10−5 Ry, respectively. The
self-consistent charge densities and states were computed using
a two-spinor wave function and switching on the noncollinear
spin−orbit interactions.
Electron−Phonon Coupling Calculations. The lattice
vibration computations were calculated using the extended
PHonon package within the same DFT code. For both the
structures, a regular dense phonon q-grid 18 × 18 × 1 was
chosen at a rigid self-consistent energy threshold 10−16 Ry and
single iteration mixing factor of 0.7 Ry. In order to compute
the electron−phonon couplings, the entire irreducible BZ was
first randomly sampled to a fine 200 phonon momenta q-grids.
14935
the perturbed potentials on these grids were then computed by
mapping the self-consistent charge densities on the regular
grids. A non self-consistent calculation was then carried out on
these random grids, which led to the electron−phonon
corrected electronic states along the BZ. Using this last step,
the electron−phonon matrix elements were finally evaluated
after constructing the initial states.
Linear Spectra G0W0+BSE Calculations. The excited-
state corrections were computed using the MBPT open-source
code package Yambo.37 The G0W0 corrections were applied to
the five electronic bands on either side of the valence band
maximum, which we also find to be most crucial in capturing
the optical transitions. A total of 200 bands (lowest 20
occupied bands and highest 180 unoccupied bands) were used
to sum up the irreducible polarization response function. Local
field effects were introduced in this linear response sum by
switching on the random-phase approximation (RPA) kernel.
An energy cutoff of about 10 Ry was found sufficient to
converge this sum describing the system inhomogenity. The
microscopic dielectric function is then constructed by
convoluting the polarization function with the bare Coulomb
potential. The dynamic screening is then computed by
convoluting the inverse of the microscopic function again
with the bare Coulomb potential. A final convolution between
this Coulombic screening and the noninteracting Green’s
function (G0) leads to the full frequency G0W0 self-energy. We
note here that most open-source code splits this self-energy
into a pure exchange term and a correlational term for
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
Figure 2. Electronic valence band dispersion of NAs (a, b) and NSb (c, d) monolayers showing the effect of spin−orbit splitting. In case of NAs,
the energy differences in the absence of SoC results in Γ3v − Γ1v = 61 meV. In the presence of SoC, the spin−orbit splitting is Δ0 = Γ5v − Γ4v = 44
meV. The strength of SoC shifts Γ1v to Γcf4v resulting in Γ1v − Γcf4v = −55 meV, Γ4v − Γcf4v = 70 meV. The spin-splitting along the M−Γ direction in
the top valence band is Δ5,6 = 33 meV. In case of NSb the respective energy differences are Γ3v − Γ1v = 318 meV, Δ0 = 57 meV, Γ1v − Γcf4v = −104
meV, Γ4v − Γcf4v = 298 meV, and Δ5,6 = 74 meV.
computational convenience. The former is static, while the
latter is dynamic in frequency and is thus a complex quantity.
The inverse microscopic dielectric function inside the dynamic
screening term demands some comments. This inverse
function is plagued with numerous poles near the real axis,
making the screening convolution integral very expensive to
compute. To implement this practically, a plasmon-pole model
is often used, which mimics the inverse dielectric function by
approximating it at the most prominent pole, which is the
plasmonic frequency. The plasmon-pole model by Godby and
Needs38 is used in the package, which uses one pole at zero
frequency while the other at the plasmon frequency. The latter
is to be chosen such that the dielectric function converges.
This correlational self-energy is then updated for each energy
starting from the noninteracting Kohn−Sham eigenvalue and
put back again into the nonlinear quasi-particle equation.
However, such an iteration would create redundancy in
computing the self-energy. Thus, the nonlinear equation is
linearized using Taylor’s series up to the first order under the
assumption that the renormalized energies are not far from the
Kohn−Sham mean-fields. If the noninteracting Green’s
function and the dynamic screening are also updated at each
iteration process, then the computation scheme becomes GW.
If the dynamic screening is updated only once, then it is a
G0W0 process or a single-shot GW scheme. Coulombic
divergences occurring in all self-energies are fixed by
integrating them over the irreducible BZ space by assuming
first that the density matrices are the smooth function of
momenta. The diverging quantities are then numerically
integrated using a Monte Carlo technique that creates
numerous randomly small BZs about each momenta vector.
We, therefore, used 107 random points distributed all over the
irreducible BZ with a cutoff of about 3 Ry to converge the
integrals. In addition to these, a Coulomb truncation (similar
to the DFT case) is applied to cut down the periodic
interactions between the repeated monolayer images.
The absorption spectra are obtained by solving the time-
independent BSE. The same cut-offs are used to build up the
exchange electron−hole attractive and repulsive kernels in the
BSE matrix. An important note is that electronic transitions are
very sensitive to BZ samplings. A dense sampling would then
14936
pubs.acs.org/JPCC Article
be more meaningful to capture optical transitions properly
along the high symmetry routes. For this purpose, we fine
sample the entire BZ into 48 × 48 × 1 on a shifted grid. First,
we obtain the kernels for the independent-particle cases on
these grids. Next, we obtain the interacting BSE kernels on the
unshifted 12 × 12 × 1-course grids. The BSE kernels on the
fine grids are now interpolated using a Wannier interpolator
which maps the fine and the course grids. This methodology39
makes the expensive fine-grid BSE computation relatively
quick without losing accuracy. A manual broadening of 0.1 eV
is applied to construct a Lorentzian shape spectra together with
an in-plane perturbing electric field. The quasi-particle energy
corrections and the static screening are then added from the
previous G0W0 calculation. In the presence of lattice vibrations,
instead, the corrections corresponding to the electron−phonon
interactions on the energy bands are implemented. The G0W0
gaps are now opened up using a scissor operator that provides
a rigid shift to the bands. No external broadening is required as
the exciton line widths are now computed using the electron−
phonon matrix elements. We go beyond the standard Tamm−
Dancoff approximation40 by including both the resonant and
antiresonant electron−hole matrix elements in the BSE
Hamiltonian. The BSE matrix is then diagonalized and solved
to obtain the poles corresponding to the transition energies.
Nonlinear SHG Calculations. The presence of a finite
external electric field breaks down the crystalline symmetry.
We chose an in-plane electric field and removed the symmetry
of the crystal (the initial 12 × 12 × 1 course-grid). We then
map this to a dense symmetry removed 48 × 48 × 1 shifted
grid in the presence of the same electric field. The time-
dependent BSE was then solved at these fine grids. The linear
coefficient was extracted by applying a delta-like pulse. A
monochromatic sinusoidal wave was used to obtain the higher-
order nonlinear coefficients. The time-dependent polarization
was obtained from the geometric Berry’s curvature. This TD-
BSE was finally solved using the Crank−Nicholson algorithm41
at a time step of 0.01 fs. The sudden switching-on of the
electric field results in spurious noise in the polarization signal
during the initial time. We, therefore, add a dephasing time of
about 8 fs, which is approximate to a spectral broadening of
about 0.17 eV. We then run this simulation for up to about 7-
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 3. Bare electronic dispersion of NAs monolayer showing the partial density of states projected on the former. Plots (a) nitrogen 3PJ=(1/2), Ms
= −(1/2) and (b) arsenic 4PJ=(3/2), Ms = (1/2) show the spin−orbital occupancy in valence and conduction bands, respectively. The arrows in (b)
demonstrate how the electron that takes part in optical transition gets scattered due to the thermal broadening of the energy bands.

Figure 4. Bare electronic dispersion of NSb monolayer showing the partial density of states projected on the former. Plots (a) nitrogen 3PJ=(1/2), Ms
= −(1/2) and (b) antimony 4PJ=(3/2), Ms = (1/2) show the spin−orbital occupancy in valence and conduction bands, respectively. The arrows in
(b) demonstrate how the electron that takes part in optical transition gets scattered due to the thermal broadening of the energy bands. Note that
the optical transitions occurs between Γ and M instead at K. This happens because of the modification of the gaps when the GW corrections are
applied.

time constants ∼55−60 fs until a clean signal was obtained.
Such a dephasing also mimics the inadvertent addition of the
experimental broadening due to thermal effect, defects, etc.
Once this clean polarization signal is obtained, the nonlinear
coefficients can be extracted by truncating the Fourier series.
energy bands. The splittings observed in these two materials
are nonmagnetic Zeeman-type (shift in the energy) across the
BZ, except in the region around the high symmetry M point,
where it can be observed as crossing of the bands along the
momentum k-space, resulting in a Rashba-type splitting.42 The

■ RESULTS AND DISCUSSION

Ground-State DFT Dispersion. Figures 2−4 demonstrate
the ground-state electronic dispersion of β-NAs and β-NSb. As
both of these materials contain heavier atoms, relativistic spin−
orbit calculations were performed, resulting in splitting of the
14937
valence and conduction splittings are even small in NAs
compared to NSb, signifying a smaller SoC. In general, in the
presence of a crystal field only, the C3v point group symmetry
splits up in energy at the Γ point of the BZ into a 2-fold high in
energy degenerate level (Γ3v) and a lower in energy
nondegenerate level (Γ1v). When the SoC is switched on,
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 5. Lattice vibration spectra, phonon density of states, and electron−phonon Eliashberg function of (a)−(c) NAs and (d)−(f) NSb
monolayers, respectively. The individual Eliashberg functions at the ground-state conduction maxima and valence minima for both the NAs and
NSb are calculated, justifying why the indirect gap should decrease with increasing temperature.

this double group level Γ3v further splits into three energy
levels Γ4v, Γ5v, and Γ6v. Group theoretic calculations tell us that
Γ5v and Γ6v are degenerate and higher in energy levels, while
Γ4v is nondegenerate and lower in energy level. Consequently,
Γ5v and Γ6v are called heavy-hole and act as a Kramer’s doublet.
The lower state Γ4v is a light-hole. The additional presence of
SoC shifts the state Γ1v to a lower level Γcf4v and is called the
Crystal-field split-off hole.
This is indicated in Figure 2, which demonstrates the
valence dispersion in both the monolayers, both in the absence
and presence of an SoC. We find that all the aforementioned
splittings are actually seen in the case of NSb. The crystal-field
splitting (in the absence of SoC) and the spin−orbit energy
splittings (Δ0) are found to be 318 and 57 meV, respectively.
The shift of energy from Γ1v (in the absence of SoC) to Γcf4v (in
14938
the presence of SoC) is obtained as 104 meV. In the case of
NAs, the corresponding crystal-field and Δ0 splittings are 61
and 44 meV, respectively. The shift of Γ1v to Γcf4v is an energy
difference of 55 meV. Therefore, we see that because of small
spin−orbit energy splittings in NAs the use of spin−orbit
coupling is not essential.
Figures 3 and 4 demonstrate the full dispersion with orbital
occupancy. Interestingly, both of them exhibit valley
degenerated indirect band-gaps. The location of the valence
band maxima (VBM) for these materials are along the path
K Γ of the BZ, while the valley degenerated VBM are along
Γ M. Similarly, the conduction band minima (CBM) are at
the M point of the BZ, whereas the valley degenerated CBMs
are at the K point. The location of direct gap for NAs is along
M Γ, while in NSb it is at the K point. The local density-of-
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
Figure 6. Ground-state (blue) and G0W0 (green) electronic band structure in (a) NAs and (b) NSb monolayers. The horizontal dashed lines are
with respect to the DFT top and G0W0 bottom of the valence and conduction bands, respectively. All the vertical arrows represent direct quasi-
particle gaps with most of the electronic transition weights.
states (DOS) projected on the respective electronic dispersion
are also demonstrated in the same figures. We observed that
the valence bands in both the materials are mainly occupied by
the P orbital of nitrogen (3PJ=(1/2), Ms=−(1/2)), whereas the
conduction bands are filled with the P orbital of the second
sublattice atom (4PJ=(3/2), Ms=(1/2)), i.e., As and Sb atoms,
respectively. To get more insight, we have shown the
contribution of all the local DOS for all atoms of both the
materials in the Supporting Information in Figures S2 and S3.
Comparing Figures S2 and S3 with Figure 2, we see that in
both materials there are the partial occupancy at points Γcf4v,
Γ4v, Γ5v, and Γ6v coming from 2PJ=(3/2), Ms=(1/2), and
3PJ=(1/2), Ms = −(1/2) of the nitrogen atom.
Electronic and Thermal Stability. We now discuss the
electrothermal stability of these isolated monolayers. The
lattice dispersions for both materials are shown in Figure 5a,d.
We note here that the C3v 3m group materials are polar. In
their bulk structures, the coupling between the optical
longitudinal (LO) modes and the polarization charge density
leads to a finite splitting between them at the Γ point;
however, in the case of 2D systems, this splitting disappears,
and the two modes become degenerated. Sohier et al.44
demonstrated this disappearance by appropriately calculating
the screened Coulomb interactions, which eliminated the
sudden discontinuities in the slope of the LO mode at the BZ
center. We find that the LO and the optical transverse (TO)
modes are, in fact, quite stable and also degenerate at Γ. All the
optical modes at Γ are also both infrared and Raman active.
However, along the off-Γ direction (near M), particularly the
in-plane longitudinal acoustic and optical branch frequencies
are infected with complicated torsional motions (see
phonons.mp4 in the Supporting Information for a schematic
visualization of the vibrational modes in these materials). We
have also computed the binding energies of these two isolated
monolayers in their free-standing form. Our calculations show
14939
pubs.acs.org/JPCC Article
that these energies for β-NAs and β-NSb could be as large as
−10.7 and −9.4 eV/atom, respectively, when compared to the
isolated atoms. One crucial observation is that there are small
out-of-plane acoustic (ZA) soft-modes located along M Γ and
K Γ directions. In fact, the presence of these soft modes
simply indicates that the exfoliation mechanism may not lead
to a standalone monolayer system. Other preparations such as
direct-grow on a suitable substrate may stabilize these
monolayers (see also stabilization of β-NAs and β-NSb by
the strain in the appropriate section below). A similar situation
is also observed during the stabilization of monolayer β-NP,
which falls under the same family.27 Also, buckled hexagonal
structures of silicene and germanene, which show negative
phonon frequencies,45 were experimentally prepared on
substrate of Ag46 and Al.47 We finally note that several
buckled V−V binary materials were experimentally prepared
recently.48 Nevertheless, in Tables 1 and 2, we have
summarized the electronic and thermal gaps for both of
these materials. The discussions about the phonon DOS and
the electron−phonon Eliashberg function are outlined in the
zero-point energy gap section below.
Excited-State G0W0+BSE Spectra. To gain more accurate
values of the electronic band gap of a material, it is crucial to
use many-body methods, such as GW. Corrected energies from
the dynamically screened electron−electron correlation fix
these gaps, which is beyond the scope of the standard DFT. In
Figure 6 we have shown the comparison between the bare and
the G0W0 band structures. We observe that the inclusion of the
exited-state corrections leads to a significant change in the
eigen-energies near the Γ point of BZ. This process results in
changing the position of NSb indirect and direct band-gaps.
The NSb indirect band gap VBM is now shifted from K Γ to
Γ, whereas the direct band gap is shifted from K to Γ.
However, in the case of NAs, the position of band-gaps does
not change. Interestingly, both these materials now show a
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 7. Absorption spectra and joint density of states in (a) NAs and (b) NSb monolayers. Solid line corresponds to the G0W0+BSE calculations
in the absence of lattice vibration. The effect of lattice vibrations on the spectra changes the widths and positions and is shown for temperatures 0 K
(dashed curve) up to 450 K at a step of 50 K. The vertical dashed lines are the direct quasi-particle gaps. The dots are the guide to eye showing the
shifts of the peak positions.

Figure 8. |∇k(EC − EV) |and EC − EV in (a) NAs and (b) NSb monolayer along the BZ route. EC and EV represent the lowest conduction and
highest valence band, respectively.

three times valley degeneracy in their respective valence band.
We report here a large indirect as well as direct G0W0 gaps for
both the materials. The important optical transitions for NAs,
as we shall see below, is mainly located along the route M Γ
and with a gap of about 5.00 eV. For NSb, the transitions are
distributed at Γ and M Γ (labeled by I) with a gap of about
4.45 and 4.77 eV, respectively. In summary, we have presented
all of the exited-state band-gaps in Table 1. The additional
energy convergences are demonstrated in Figures S4 and S5 of
the Supporting Information.
The two-body Bethe−Salpeter equation of motion com-
pletely describes the excitonic character. In fact, it is the pole of
the two-body Green’s function that gives direct access to the
absorption spectra. In Figure 7, we present the imaginary part
of the dielectric function (which is proportional to the
absorption spectra) for both the materials. Our calculations
show that the NAs spectrum is dominated by two prominent
peaks, one at 3.38 eV and the other at 5.37 eV. The former is
the bound exciton B1, housed below the onset of the
continuum, while the latter is the resonant exciton R1 and is
located above the continuum. Instead, the NSb spectra consist
14940
of only one dominant bound excitonic peak (B1) at about 3.15
eV. The exciton binding energies for these two materials are
obtained as 1.62 and 1.30 eV, respectively. We note here that
the effect of spin-splittings in both valence and conduction
bands (relatively small in both cases) does not show any
additional splittings in the absorption spectra. This could be
the reason since an additional manual Lorentzian spectral
broadening of about 0.1 eV (which is reasonable when actual
experiments are performed) is applied, which might have
overshadowed the spectra splittings. In the same figure, we
have plotted the joint density of states (JDOS) to understand
the presence of dark excitons. We carefully observe that below
the bound excitons in both materials, the JDOS changes the
slope. This indicates that there are possibilities of finding dark
excitons. We analyzed the character of these dark excitons and
noted that in the case of NAs, there are degenerate dark
excitons located at 3.27 (double) and 3.25 (triple) eV. There is
also a singlet positioned at 3.34 and 3.24 eV. All these are
located below the main peak at 3.38 eV. In the case of NSb, we
found that there are two double degenerate dark excitons
located at 2.92, 2.96, and 3.07 eV, while the singlets are at 3.12,
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
2.94, and 2.89 eV. In addition, we also found one triply
degenerate exciton (two bright and one dark) at 3.05 eV and
one double degenerate bright exciton at 3.08 eV. These bright
excitons at 3.05 and 3.08 eV, however, did not reflect on the
corresponding absorption spectra. It was, however, suspected
that these are due to the spin-splittings, but our careful
analyses did not reveal such possibilities. Additionally, we have
also shown how the spectra would be underestimated in case
the excitonic theory is absent. The equation of motion when
contains only the Hartree kernel, the spectra are governed
through the one-body equation of motion and the peak
relocates itself above the continuum. It is also important to
note that Carvalho et al.31 presented a lucid picture in order to
understand why an atomically thin layer should absorb light
strongly. Their theory, using a variety of crystal structures,
demonstrated that the optical transitions actually come from
those areas of the BZ where the conduction and valence bands
run parallel to each other. At these critical points, the JDOS
suffers discontinuities and hence contributes to a very large
optical conductivity. This is known as band-nesting and
provides a quick visual inspection about the origin of the
spectrum peaks. For various monolayer TMDCs, they show
that this presence of the band-nesting might be the reason for
obtaining large conductivities. In order to see whether the
band-nesting is present inβ-NAs and β-NSb, we have plotted in
Figure 8 the gradients of the differences between the top
valence and bottom conduction band along the BZ. We
observe that between Γ and M, there are no critical points
where the band runs parallel to each other. This implies that
the JDOS do not suffer discontinuity along this route, and
therefore, the strong absorption observed in Figure 7 are
mainly due to the excitonic formation. This particular
information is displayed in Figure 9 where we have
Figure 9. Fundamental exciton formation in (a) NAs and (b) NSb
monolayers. The electronic transitions representing exciton weights
are projected onto the ground-state electronic dispersion and in both
cases are shown by the yellow shade. The inset exhibits the
corresponding excitonic weights in the BZ of the two cases.
superimposed the bound excitonic weights onto the bare
dispersion diagram to understand better. The yellow shades
demonstrate the vertical electronic transition from the direct
gaps that contribute to the exciton formation and,
consequently, the absorption spectra. In NAs, this is along
M Γ route, whereas in the case of NSb, it is contributed
mainly due to transitions at Γ and along M Γ route. The
states of the valence and conduction bands poses no difference
14941
pubs.acs.org/JPCC Article
in angular momentum ΔL = 0; however, the valence and
conduction bands in both materials have an opposite quantum
number of spin Ms (with the difference ΔMs = 1) and
implicates the nonforbidden transitions in the intermediate
coupling. We can now access the information about the inset
pictures of the respective excitonic weights in the reciprocal
space. Comparing Figure 9 with Figures 3 and 4 we see that in
both the materials the optical transition comes from the
valence occupancy with spin Ms = −(1/2) to the conduction
band with Ms = (1/2). This means that the development of the
first bright exciton in these materials is because of the coupling
between antiparallel spins within the intravalley. This is in stark
contrast with the contemporary monolayer TMDCs, where an
opposite situation prevails.8,30 In Figure 10 we map these first
Figure 10. Isosurface of the first bound exciton with fixed hole (the
boundary value of the surface is a tenth of the highest value) over the
real space lattice of both NAs ((a) and (c) at 3.38 eV) and NSb ((b)
and (d) at 3.15 eV). Subplots (a) and (b) are seen from the top,
whereas (c) and (d) are from the side view.
bright bound excitonic isosurfaces onto the real-space lattice of
the respective materials. In order to evaluate this texture, we
fixed the position of the hole on the top of the nitrogen atom
(which contributes mainly to the top of the valence bands, see
the Supporting Information in Figures S2 and S3) separated by
about 1 Å within the unit cell. The extent to which isosurface
spreads gives valuable information on whether the exciton is a
Mott−Wannier (more delocalized) or Frenkel (more local-
ized) type. We observe that the first bright exciton is in the
NAs case more localized than in the NSb case (Figure 10),
which agrees with stronger exciton binding energy in the first
NAs exciton (1.62 eV) than in the first NSb exciton (1.30 eV).
A careful analysis further reveals that these doubly degenerate
bright excitons in fact preserve the crystalline 3-fold symmetry.
De-excitation can be overwhelmed by means of the scattering
of electrons, accounted for, in the structural model, by the
thermal broadening of the bands. In panels (b) of Figure 3 and
4 we show possible patterns for the scattering of the electrons,
where the scattering event arrows (represented by II and II′)
were resolved by studying the life times of all states and so the
de-excitation process can be expected from the bottom of the
conduction band. We discuss this in the next section. The
energy convergences for the BSE spectra are demonstrated in
Figures S6−S8 in the Supporting Information.
Zero-Point Energy-Gaps. All the aforementioned dis-
cussions are presented under the condition when the influence
of lattice vibrations is absent. However, in reality, experiments
are performed at finite temperatures, and therefore, one should
look onto the temperature-dependent optical properties. In
1950, Fan49 first proposed the coupled electronic and lattice
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 11. Spectral functions dependencies on temperature in NAs and NSb monolayer. (a), (b) Spectral function NAs monolayer at the direct
valence and conduction state (see Figure 3b). (c), (d) Same at the indirect valence and conduction state, respectively. The spectra in (e)−(l)
depict the cases for the NSb monolayer (see Figure 4b at high-symmetry points). The vertical dashed lines represents the band edges in the absence
of electron−phonon couplings.

vibration self-energy using the first-order perturbation theory.
Later, based on the crystalline X-ray diffraction on nonpolar
diamond and ZnS type crystals, Antončı ́k50 modified the
Debye−Waller (DW) second-order perturbation theory by
adding empirical factors to it. In a chronological series of
papers, Keffer,51 Walter,52 and Kasowski53 used the DW
factors to estimate temperature-dependent band-gaps of PbTe,
GaAs, and NiS, respectively. However, it was not until 1976
when Allen and Heine54 made the DW term a transitional
invariant and thus were able to unify the Fan theory with the
DW. This approach allows them to expand the DW self-energy
as a product of the first-order derivatives of the self-consistent
potential. Following, Cardona and Allen,55−57 in a series of
papers, developed the electron−phonon renormalization
theory. Together with Heine, this is now known as the
Heine, Allen, and Cardona (HAC) theory. They argued for the
otherwise distinct electronic state. Below we address how the
bare states |nk⟩ and energies ϵnk of β-NAs and β-NSb
monolayers are affected in the presence of this electron−
phonon coupling.
Following eqs S.4 and S.6 in the Supporting Information of
Kolos et al.,27 we first compute the Fan (∑Fan
nk (ω,T)) and DW
(∑DW
nk (T)) self-energies. The former is the dynamic one and is
a complex quantity, while the later is a static and purely real.
The temperature dependencies on these two self-energies
comes from their respective Bose occupation numbers. Once
these self-energies are computed, the interacting electronic
Green’s function (Gnk(ω,T)) can then be constructed from
1
Gnk (ω , T ) = Fan DW
ω − ϵn0k − ∑nk (ω , T ) − ∑nk (T ) (1)
in which ϵ0n,k is the bare electronic energy. The dressed

ÄÅ ÉÑ
fundamental fact that even at zero temperature, due to the

ÅÅ ÑÑ
electronic energies are the outcome of real part of the pole

ÅÅ Ñ
inherent quantum mechanical atomic spatial uncertainties, the

ÅÅω − ϵnk − ℜ ∑ (ω , T ) − ∑ (T )ÑÑÑ − iℑ ∑ (ω , T )

ÅÅ ÑÑ
couplings between electron and atomic vibrations can modify Fan DW Fan

ÅÇ ÑÖ
0
the eigenenergies of the former. This electronic energy
correction at zero Kelvin is the zero-point renormalization nk nk nk
(ZPR) or motion (ZPM). However, because of the presence of =0 (2)
an immense number of computational complexities of these
self-energies, the scientific community abandoned this topic in whereas the imaginary portion gives the electronic lifetime in
Fan
the early days in an anticipation that these corrections might that state. If the pole of ℜ ∑nk (ω , T ) and ϵ0n,k are not far
be extremely small when compared to the dynamic electron− from each other, then the real portion of eq 2 can be linearized
electron couplings. Eventually, the importance of ZPR by using the Taylor’s expansion up to first order followed and
resurfaced when it was observed that these corrections led to assigning a weight factor Znk. The dressed energy uncertainties

ÄÅ DW ÉÑ
a giant ZPR in diamond58,59 (in range of 615−622 meV) and
ÅÅ ÑÑ
(now called as quasi-particle energy spreading) now become

ÅÅ Ñ
LiF36 (718 meV). Marini21 and co-workers28,60,61 demon-
Δϵnk = ϵnk + ZnkÅÅ∑ (T ) + ℜ ∑ (ϵnk , T )ÑÑÑ
ÅÅ ÑÑ
Fan
strated the computational strategies of the ZPR in a purely ab
ÅÇ nk ÑÖ
0 0
initio way within the framework of DFPT. The inclusion of
these self-energies adds a finite energy spreading of the nk (3)

14942 https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
ij yz
where the dimensionless weight factor
j z
Znk jj = 1 − ∂ω ∑nk (ω , T ) z is a value between 0
ω =ϵnk z
−1
k {
∂ Fan
and 1. The spectral function at each state and band can also be
1 ℑ ∑nk (ϵn0k , T )
A n k (ω , T ) =
π [ω − ϵ0 − ℜ ∑Fan (ϵ0 , T ) − ∑DW (T )]2 + [ℑ ∑Fan (ϵ0 , T )]2
nk nk nk
whose peak is centered at Δϵnk. A Zn,k → 0 implies a very
strong quasi-particle (electron−phonon) interaction leading to
a flattened (distorted Lorentzian) Ank(ω,T) spectra. The well-
defined electronic state now spreads more in energy and,
therefore, as per Heisenberg’s uncertainty rule, acquires a
shorter lifetime. This is indicated through a quick scattering of
the electron to a new accessible state. Instead, a Zn,k → 1
means that the interaction is weak and is characterized by a
sharp Lorentzian peak. The energy spreading then tends to a
well-defined state, allowing the electron there to live longer.
Figure 11 shows exactly this scenario. We have shown the
spectral function for three different temperatures, 0, 300, and
450 K. These are computed at the direct and indirect (valence
and conduction) points of the BZ (see Figures 3 and 4). The
direct points are those which contribute to the optical
absorption, whereas the indirect points are included to
broaden the view of the temperature-dependent electronic
gaps. We observe that the spectral function in all the cases
diminishes and become more asymmetric with an increase in
temperature, signifying a strong coupling. For example, analysis
on NAs indirect valence band shows that the ℜZnk drops to
almost 70% at 450 K. Effectively, Znk physically represents the
quasi-particle charge distribution and is the area under the
spectral function Ank(ω,T) . The energy distance between the
valence and conduction spectral function peaks is the
renormalized band gap. There are some other interesting
activities at the direct gap point of NAs. We see that the
indirect gap actually reduces as the temperature increases and
is characterized by a right (left) movement of the valence
(conduction) spectral function peak. The deviation of the
peaks at 0 K from their respective frozen atom (DFT) band
edges (see the vertical lines in Figure 11) gives the ZPR. This
trend is observed generally in most semiconductors. Instead,
we found that the direct gap between between M and Γ
increases. We show these results in the same figure for NSb
also, in which the indirect gap (from M to Γ through K and
direct gap (labeled as I in Figures 4 and 6b) decreases, but the
direct gap at Γ increases. To account for the decreasing band
gap trends, we used the Varshni’s empirical model71
αT 2
Eg (T ) = Eg (0K) −
(T + β ) (5)
in which α and β are the numerical fits and Eg(0K) is the band
gap at 0 K. We see that in both materials, the decrease of the
indirect gap with temperature is rather slow. However, at the
direct gap of NAs, the rate is quite large. The net deviations of
the VBM and CBM at 0 K with respect to the frozen atom
edges (i.e., the ZPR) for NAs and NSb are found to be 105 and
80 meV, respectively. The reason for a low ZPR of the latter is
because of the heavier atomic mass of Sb, which indicates that
heavy atoms vibrate less. In Tables 2 and 3, we have
summarized the values of the temperature rate of change of
14943
pubs.acs.org/JPCC Article
1
constructed from A nk (ω , T ) = π |Gnk (ω , T )| leading to a
Lorentzian-like distribution function
Fan
nk nk nk (4)
indirect gaps and the ZPR respectively. We have also listed the
room temperature optical gaps in Table 4. The necessary
spectral function convergence with respect to the number of
electronic bands is shown in Figure S9 of the Supporting
Information. In Figure S10, we have additionally shown the full
electronic dispersion at 0 K.
Another interesting observation from Figure 11a−d can be
seen for NAs valence and conduction energy line widths at the
same temperatures. We highlight those cases of electronic
transitions which contribute to the formation of excitons. One
can see that at a particular temperature the line width of the
conduction band at the direct point is more than that
compared to the corresponding indirect point. This means
that, after a transition (vertical) to the conduction band, an
electron quickly gets scattered to the corresponding indirect
state. This mechanism path (I−II) is shown in Figure 4b by an
orange color arrow. This is not the case with NSb. There are
two major pathways one at Γ (labeled in magenta as I′−II′)
and the other between Γ and M (labeled in orange as I−II).
The optical gap at Γ is more than the direct gap between Γ and
M. However, since the line width of the conduction band at
the direct point (between Γ and M) is comparatively large
than that at Γ (see Figure 11f,j) it is more likely that the
electron from pathway I will scatter along the pathway II at M.
This leaves the excited electron at Γ to scatter into state K
(pathway II′). This is also plausible since both the scattered
electrons cannot be in the same state with same quantum
numbers. We now justify the temperature dependencies of the
gaps within the purturbative theory. To understand this, we
ÄÅ ÉÑ
compute the Eliashberg function63,104
ÅÅ |g qν |2 Ñ
ÅÅ 1 Λ nn ′ k ÑÑÑ
g Fnk(ω) = ∑ ÅÅÅ ÑÑδ
n′qν Å 2 ϵnk − ϵn ′ k ÑÑÑ
qν
ÅÇÅ nk ÑÖ
2 nn ′ k
−
ϵ − ϵn ′ k − q
(ω − ωqν) (6)
within the adiabatic approximation, i.e., |ϵnk − ϵn′k−q| ≫ ωqν, in
which the “primes” denotes the scattered states. The
numerator of the first term inside the parentheses stems
from the first-order electron−phonon matrix elements of the
Fan self-energy, while the numerator in the second term is due
to the second-order DW self-energy. The sum is over the
empty electronic states and for each phonon branch and
momenta. The delta function simply states that g2Fnk(ω) is
proportional to the phonon DOS. We evaluate g2Fnk(ω) at the
VBM and CBM and obtained their difference g2Fc(ω) −
g2Fv(ω) for both NAs and NSb. This is shown in Figure 5c,f,
respectively. We observe that the difference is negative. Such
behavior are commonly found in most semiconductors where
the g2Fc(ω) is negative while g2Fv(ω) is positive. This
eventually cancels out any increment in the gap with
temperature, and therefore, the indirect gaps reduces.
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 12. (a) NAs and (b) NSb excitonic oscillator strengths as a function of temperature for all exciton with strengths more than 20%.

Figure 13. Linear optical properties of NAs (left column) and NSb (right column) in ultraviolet and visible regions for frozen atom approach and
300 K: the refractive index n, the extinction coefficient k, and the absorbance A as a function of wavelength.

However, this does not seems to be true at the direct gap
points in NAs (between M and Γ) and NSb (at Γ), which
shows an increasing trend. We discuss this situation below
during the exciton−phonon interaction.
Temperature-Dependent Absorption Spectra and
Excitonic Line Widths. Marini21 demonstrated how the
quasi-particle energy spreading (Δϵnk) due to the electron−
phonon couplings can be introduced in a fully ab initio way in
the development of excitonic formalism through the BSE. This
careful study has indicated that the energy corrections from the
dynamical HAC theory in the development of excitons cannot
be avoided if one aims to justify the optical properties properly.
Figure 7 shows the optical absorption spectra in the
temperature regime 0−450 K at a step of 50 K. All spectra
are computed using the BS kernel. As polaronic excitation
energies are involved now, we go beyond the Tamm-Dancoff
approximation105 in order to correctly account for the
electron−hole pairs and antipairs in Fock-space BS Hamil-
14944
tonian. The sharp distinction seems to appear in the 0 K
spectrum when compared with the corresponding frozen-atom
case. We see that the effect of the ZPM drops the intensity
significantly, revealing the true spectral broadening. In the case
of NAs, we see that the bound B1 and the resonant R1 excitonic
peaks blue shift in energy, indicating that the direct gap
increases with temperature. At 0 K, the shift of B1 is only about
20 meV. This is not the case with NSb. Instead, we see that the
effect of ZPM has clearly red-shifted the 0 K B1 peak by a giant
430 meV compared to its frozen-atom condition. We also
observe that up to 350 K, there is not much change in the
energy position of this peak, suggesting that the fundamental
excitonic energy in NSb is a very slow function of temperature
compared to NAs. The two NSb peaks above 4 eV broaden to
a single peak with a large width as the temperature increases
further. Figure 12 shows interesting characters about the
excitonic oscillator strengths in the range 0−450 K. For better
clarity, we have plotted only those excitonic energies which
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
have more than 20% oscillator strengths. Excitons with >20%
are shown in this figure with filled circles of varying colors. We
now focus on the first pair of excitons of NAs at 0 K, located
between 3 and 3.5 eV. We see that the first exciton of this pair
is dark. However, as the temperature rises, the oscillator
strength increases and makes it bright. Interestingly, the other
exciton adjacent on it, it is initially bright and starts losing its
strength as the temperature increases. As these two excitons
are very close to each other, we understand this effect through
an exchange of oscillator strength. In the case of NSb, we see
that the first exciton (between 2 and 3 eV) is always dark,
whereas the second one always remains bright. However, the
exchange of oscillator strengths starts between the third and
fourth excitons when they come close to each other above 200
K. Globally, we observe that most excitons in the two
monolayers are dark (except a few resonant excitons between
5.5 and 6.5 eV in NAs and a few bound excitons between 4 and
4.5 eV in NSb). We also observe the same blue and redshifts of
the fundamental bright excitonic energies in NAs and NSb,
respectively. In Figure 13, we also provide information on the
influence of temperature on the refractive index n, the
extinction coefficient k, and the absorbance spectra of both
materials. We observe that the refractive index converges to
1.63 and 1.81 for NAs and NSb, respectively. This figure also
shows a strong absorbance peak in the visible light region for
NSb, on the other hand, the absorbance peaks for NAs case are
mostly in the UV region. Temperature effects on these
parameters are mostly on the broadening of the peaks.
Particularly in the influence of temperature on the absorbance
spectra is a visible lowering of the maximal values from 37%
(NAs) and 26% (NSb) to values between 15 and 20%.
In order to understand which phonon mode couples to the
exciton during its development, we have computed the
exciton−phonon coupling function, which is similar to the
previously described generalized Eliashberg function.21 These
are shown in subplots (a) of Figures 14 and 15. We evaluated
this coupling function for both the valence and conduction
state at the direct gap (between M and Γ, see Figure 3b).
Figure 14. (a) Exciton−phonon coupling (black solid line) as a
function of phonon frequency, showing the most prominent mode
that builds the NAs fundamental bright exciton. The shaded areas
represent the coupling function of valence and conduction bands at
the direct point (see Figure 3b). (b) The inhomogeneous line width
of this fundamental exciton. The symbols are the ab initio data, while
the dashed line is the phenomenological excitonic width equation.
14945
pubs.acs.org/JPCC Article
Figure 15. (a) Exciton−phonon coupling (black solid line) as a
function of phonon frequency, showing the most prominent mode
that builds the NSb fundamental bright exciton. The shaded areas
represents the coupling function of valence and conduction bands at
the direct-point Γ (see Figure 4b). (b) Inhomogeneous line width of
this fundamental exciton. The symbols are the ab initio data while the
dashed line is the phenomenological excitonic width equation.
Comparing with the phonon dispersion and DOS in Figure
5a,b, we see that in NAs, the most prominent exciton−phonon
interactions are located around the optical frequencies near
580 and 680 cm−1 respectively. These reflect an in-plane E and
out-of-plane A1 infrared modes at the phonon zone center. In
the acoustic region, the interactions are mainly from the mode
near 270 cm−1 and are infected mostly with an in-plane
torsional motion (see Supporting Information for a schematic
visualization of the vibrational modes in these materials). We
also note that near lower frequency, the coupling function
crosses the zero line and becomes positive. This is an
anomalous behavior that justifies the increase of the direct
gap with temperature (see Figure 16a). Similarly, in NSb, we
see that the main exciton−phonon interaction is located
around 560 and 620 cm−1 of the optical frequency,
respectively. These correspond mostly to the in-plane and
out-of-plane vibrations. In the lower acoustic region at about
50 cm−1, we note a prominent interaction from the in-plane
mode. There is, however, a small encroachment of the coupling
function (around 220 cm−1) in the upper positive region. This
again leads to an anomalous behavior and justifies the increase
of the direct gap with temperature at Γ (see Figures 4b and
16b). The temperature-dependent excitonic inhomogeneous
line widths are shown in Figures 14b and 15b. We write here
the phenomenological nonradiative line width equation
Γ(T ) = Γ0(0K) + γacT + ÅÄ ÑÉÑ
ÅÅ
γop
ÅÅexp op − 1ÑÑÑ
ÅÅÇ ÑÑÖ
ℏω
( )
kBT (7)
in which Γ0(0K) is the 0K broadening. The constants γac and
γop are the exciton-acoustic and exciton-optical interaction
strengths, respectively, ωop is the optical frequency and kB is
the Boltzmann’s constant. We fit eq 7 to our result shown in
Figures 14b and 15b to report that the interaction strengths γac
and γop are 0.4 μeV K−1 and 36 meV for NAs at ωop = 580
cm−1 and 0.18 μeV K−1 and 52 meV for NSb at ωop = 620
cm−1, respectively. We obtained that Γ0(0K) in NAs and NSb
are 105 and 129 meV. At room temperature, the line widths
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
Figure 16. Temperature dependence of the DFT gaps in the presence of electron−phonon couplings in (a) NAs and (b) NSb monolayers. The
direct gap in (a) corresponds to transition I (in M Γ, see Figure 3(b)), while the indirect gap is between K Γ and M points of the BZ. The direct
gaps in (b) correspond to I and I′ transitions (at M Γ and Γ, respectively, see Figure 4(b)), while the indirect gap is between K Γ and M points
of the BZ.
are in the range of 250−280 meV. The residual width at 0 K
can be obtained by taking the slope of Γ(T) at the bending
knee down to 0 K. This corresponds to almost 94 and 110
meV for NAs and NSb, respectively. Effectively, these are large
widths when compared to only 1−10 meV found in
conventional TMDCs monolayers on various substrates (see
Table 5) and therefore suggest that the excitonic motions in
NAs and NSb are diffusive rather than ballistic.
Nonlinear SHG Spectra. We now present the optical
nonlinear second harmonic coefficients generated from these
monolayers. In the presence of an intense light, the induced
macroscopic polarization in the frequency domain is related to
the external time-dependent electric field ,(t ) as15
7i(ω) = ε0{χij(1) (ω),j(ω) + χijk(2) (−ω; ω1, ω2),j(ω1), k(ω2) + ...}
(8)
where the index i denotes the polarization direction and j, k,
etc. are the directions of the electric field. The first term in the
parentheses holds true for linear dielectrics, whereas the next
term onward represents nonlinearity. The linear response
function ℑχij(1) (ω) is proportional to ℑεM at low intensity. At
large intensities and under the condition ω1 = ω2 = ω,
χijk
(2)
(−ω;ω1,ω2) denotes the second harmonic coefficient.
Furthermore, these tensorial components under the C3v
(3m) crystalline symmetry converges as χ(2) bbb = −χbba = −χaab
(2) (2)
= −χaba in which a and b are the in-plane crystal axes.
(2)
It should be noted that the optical nonlinear responses are
not measured directly. Instead, light from a pump with a
tunable power is incident on the sample, and the
corresponding reflected average power is then measured. If
the sample is monolayer, the response χ(2) ijk (−2ω;ω,ω) is then
14946
pubs.acs.org/JPCC Article
calculated by equating this power to either one of the following
two empirical models namely, “bulk” and “sheet” models.85
Clark et al.84 compared these two models and showed them to
(2) n2D(ω) n2D(2ω) (2)
be related as χbulk = 32πF χsheet in which n2D is
(ns(ω) + 1)3
the refractive index of the monolayer at frequencies ω and 2ω
and ns is the refractive index of the underlying substrate. The
factor F (which appears in the bulk model) is the so-called
“scaling factor” in the reflectance and transmittance signals and
stems due to the large mismatch of the refractive index
between the sample-substrate and sample-air. This often gives
rise to a severe overestimation of the SHG coefficient. For
example, when the MoS2 experimental intensity counts are
used in the bulk model, the SHG is estimated to be on the
order of 103 (CVD)−105 (exfoliated) pm/V.83 The sheet
model instead predicts an order of 102 pm/V82,108 only.
Similarly, in the case of monolayer TMMCs, the sheet model
results an SHG of the order 102 pm/V109 which is
overestimated by 103 when the bulk model is used.101 In
fact, the prediction of SHG coefficient using purely first-
principle calculations, however, indicates that the sheet model
is more accurate.75,109,110
Nevertheless, following Attacallite et al.,20 the nonlinear
optical responses can be predicted by integrating the time
dependent Schrödinger’s equation of motion (EOM)
ı
d
iℏ |υnk ⟩ = [/k + i ,(t ) ·∂ k]|υnk ⟩
dt (9)
for the time-dependent occupied periodic Bloch electronic
states |υnk⟩. The second term in the parentheses represents the
coupling between the time-dependent external electric field
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 17. (a) Convergence demonstration of the absorption spectra using the approach of nonlinear real-time BSE and static BSE for NAs
monolayer. The shaded region (yellow) demonstrates that in the presence of a delta-like field with low intensity (500 KWcm−2 in this case), the
linear response (black) can be obtained from the solution of the time-dependent BSE. A scissor is added in the exciton Hamiltonian to mimic the
G0W0 gap in both cases. The time-dependent induced polarization along the crystalline a-axis is also shown due to (b) delta-like field and (c) a
quasi-monochromatic field along the b-axis, respectively.

Figure 18. (a) Convergence demonstration of the absorption spectra using the approach of nonlinear real-time BSE and static BSE for NSb
monolayer. The shaded region (yellow) demonstrates that in the presence of a delta-like field with low intensity (500 KW cm−2 in this case), the
linear response (black) can be obtained from the solution of the time-dependent BSE. A scissor is added in the exciton Hamiltonian to mimic the
G0W0 gap in both the cases. The time-dependent induced polarization along the crystalline a-axis are also shown due to (b) delta-like field and (c)
a quasi-monochromatic field along the b-axis respectively.

and the Bloch electrons. Here the coupling ∂̃k is the k-
derivative operator due to the Born-von Kármán periodic
boundary conditions. The states |υnk⟩ under such conditions
remain invariant under unitary rotation.67,111 /k is the model
Hamiltonian where a number of hierarchy can be used to
approximate the electron−hole correlation. In case of a single-
particle Kohn−Sham states (i.e., no correlation between the
electron and hole), /k is simply / KS
k with bare energy eigen-
value ϵ0nk. This is an independent-particle picture. If the
dynamic electron−electron correlation is added (i.e., the G0W0
correction), the picture stands then as an independent quasi-
particle approximation. Excitonic picture is captured when the
14947
electron−hole correlation is used. The self-consistent Hartree
potential VH (as functional of time-dependent electronic
density, ρ(r, t)) and static screen exchange (SEX) can then
be convoluted with density fluctuation as20
/k = / KS
k + ∑ Δn|υnk0 ⟩⟨υnk0 | + VH[Δρ(r , t )] + ∑ [Δγ ]
nk SEX
(10)
in which Δρ(r,t) = ρ(r,t) − ρ(r,0) is the incremental density
and Δγ(r,r′,t) = γ(r,r′,t) − γ(r,r′,0) is the density fluctuation
matrix. The second term in the above equation represents a
scissor operator that mimics the electron−electron self-energy
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
Figure 19. Phase-delay between the induced polarization and the applied field in monolayers of (a) NAs and (b) NSb respectively. The real and
imaginary parts of the linear spectra is obtained from the real-time approach. All the analyses are obtained within the TD-BSE level of theory. Note
the convergence of the delay to π at low frequencies.
correction through Δn = ϵGn 0W0 − ϵ0n. Equation 10 when used as
a kernel in eq 9 is the TD-BSE. In-fact it was also shown that in
the presence of an impulse-like field with weak intensity, eq 9
converges to the solution of the time independent (static)
BSE.20 Since the position operator in the reciprocal space of
periodic structures appearing in the coupling EOM above is
not defined properly, a Berry’s geometric phase is therefore
used to construct an in-plane polarization
ij 2π yzzz
j
detSjjjk, k + z
Nk zzz
Nk − 1
jj
eg |a|
k {
7=− s ∑ ℑlog ∏
2π Ω Nk⊥
k⊥ k (11)
once |υnk⟩ are known. This was prescribed by Resta and 112
Vanderbilt19 in their modern theory of polarization, and later
i 2π y
spin degeneracy, Sjjjk, k + N zzz is the overlap matrix between
used by Attaccalite and co-workers.18,20,109,110,113 Here gs is the
k {
k switching on of the electric field results in spurious initial
the states |υnk⟩ and |υmk + 2π / Nk ⟩, and Nk∥ and Nk⊥ are the
respective in and out-of plane k-points to the polarization
direction and Ω is the cell volume. Equation 9 is then solved
numerically using the Crank−Nicolson algorithm41
I − i(△t /2)/k(t )
|υnk (t + Δt )⟩ = |υnk (t )⟩
I + i(Δt /2)/k(t )
at a time step of Δt where I is the identity element.
First, in order to probe the response at all frequencies, we
choose an in-plane δ-like impulse electric field in our model
Hamiltonian. We then show that at weak field intensities, the
linear response function ℑχij(1) (ω) obtained from the solution
of eq 9 converges to the ℑϵM spectra (i.e., the static BSE). For
this, we used a low (500 KW cm−2) intensity. To compute the
self-consistent VH and the SEX convolution integrals, the same
cutoff of 10 Ry was used (similar to BSE). The EOM was then
integrated at a time step of 10 as Figures 17 and 18
demonstrates this scenario for NAs and NSb monolayers,
respectively. We see from the subplots (a) in both cases that
the linear coefficient ℑχij(1) (ω) is in excellent agreement to the
respective absorption spectra ℑϵM . The subplots (b) show the
typical corresponding behavior of the induced in-plane
pubs.acs.org/JPCC Article
macroscopic polarization over time. An important information
can be obtained if one plots the phasor diagram of the
response χ(1)
ij (ω). The complex phase delay ϕ between 7 and
, can be evaluated from χ(1) aa (ω) = |χaa (ω)|exp(iϕ). This is
(1)
shown in Figure 19. The boundary ϕ = π/2 corresponds to the
in-phase oscillation of the in-plane macroscopic polarization
∂P
current − ∂t with , .20 We see that ϕ closely follows ℜχaa(1) (ω).
At low frequencies where the absorption is minimum, ϕ → π,
signifying plasmonic oscillations. The delay of the first bound
B1 absorption peak appears to be greater than π/2 and thus
signatures clearly a localized excitation. Instead, at the peak
near 5.4 eV (R1 in case of NAs), the delay decreases below π/2
indicating a delocalized exciton.
The subplot (c) in Figures 17 and 18 shows the in-plane
macroscopic polarization over time when the in-plane electric
field is harmonic. Careful observation reveals that the sudden
spikes in the polarization spectra. To filter out these spikes and
get a clean polarization signal, a dephasing time of about 8 fs,
equivalent to a damping of 0.2 eV, is applied. This
phenomenological damping also imitates the presence of any
experimental ambient conditions like damping due to thermal
broadening, defects, etc. We observed a clean signal after about
30 fs (about 4 time constants). The simulation was allowed to
(12) run for a total of 55 fs. Within this time window, the extraction
of χ(2)
aab(−2ω;ω,ω) was done at equally spaced 100 frequencies
between 0.5 and 4 eV. To get some meaningful insights from
the SHG spectra, we first note that the nonlinear responses are
computed using various choices of gauge. These responses, for
example, χaab (2)
(−2ω;ω,ω), are plagued with unphysical
divergences of the form ω−1 and ω−2 as their leading orders.
Sipe and Aversa,114 however, demonstrated that these
divergences could be avoided if one uses a length gauge.
Instead, if the gauge choice is velocity, then at low frequencies
(ω → 0), such divergences can be removed either by
separating the intra- and interband transition processes or by
applying a sum over the transition band rule. The length gauge,
although rules out this problem but could only be applied to an
isolated system where the boundaries are aperiodic, like in
clusters and molecules. Sipe114−118 along with Ghahramani119
showed that these divergences can be overcome in a periodic
system and can also be independent of any choice of gauge.
14948 https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
Figure 20. (a) Absolute SHG coefficient in NAs monolayer within the time-dependent (TD) BSE (TD-BSE) level of theory exhibiting the peaks
A1, and A′1 due to the bound and resonant excitons B1 and R′1 respectively. The x axis is the laser frequency. (b) The absorption spectra at ω and ω/
2 under the same TD-BSE level of theory. The vertical dashed line exhibits the ω/2 gap. One can see that the ℑχaab
gap. (c), (d) All respective cases in the NSb monolayer.
The sum-over transition bands demonstrated that the
aab(−2ω;ω,ω) resonance occurs at ω and 2ω. However, this
χ(2)
is only possible when the electron−hole interaction is at their
lowest hierarchy, i.e., at the independent particle approx-
imation. In such a case, it becomes easy to separate the intra
and interband transition processes. However, in the presence
of local fields and excitons, such separation cannot be done
very easily and therefore to understand the ω and 2ω
resonances, its stands reasonable to compare the
χ(2)
aab(−2ω;ω,ω) spectra with ℑε(ω) and ℑε(ω /2). Such a
comparison, although a qualitative approach, provides great
details about the nonlinear responses.18,26,27,67,109,120,121 Figure
20 represents such a comparison between |χ(2) aab(ω) |and the
absorption spectra ℑε(ω) and ℑε(ω/2) of NAs and NSb. We
use the real-time BSE kernel in the Hamiltonian to evaluate the
SHG spectra. In the case of NAs, the first peak A1 at 1.68 eV is
in excellent agreement with the peak B1 of the ℑε(ω/2)
spectrum. By comparing ω/2 spectra, we observe that A1
actually stems due to a 2ω resonance. Similarly, the shoulder
peak A1′ at 2.81 eV stems out due to the resonant exciton A1′.
This is also a 2ω resonance. Comparing with the 1ω resonance
spectra, we see that its contribution starts beyond 3.3 eV. In
addition, we see that the SHG spectra are dominated by a
largest peak at 3.20 eV, which is due to the various resonant
excitonic process. As ω → 0, χ(2) aab(0) → 16 pm/V. At this
point, we note here that the losses inside a silica fiber optic are
minimum at a wavelength of 810 and 1560 nm. These two
lossless windows are therefore favored much to send optical
signals. We find that at 810 and 1560 nm, χ(2)aab(ω) → 56 and 17
pm/V, respectively. In the case of NSb, we observe that the
first 2ω SHG peaks at 1.32 eV and are due to the B1 exciton.
This is followed by a broad spectrum in the region 2−3 eV,
where we see that A1′ is spread over large energies. This is
14949
pubs.acs.org/JPCC Article
(2)
goes to zero below half of the
again a 2ω resonance and is due to the resonant exciton R1′.
The corresponding SHGs at 810 and 1560 nm are close to
each other at 27 and 31 pm/V, respectively. The SHG
coefficients of these lossless windows come out to be at par
with those obtained from monolayer TMDCs, TMMCs, and h-
BN (extracted after using the sheet model). The maximum
SHG in each case occurs at 3.2 eV or 387.5 nm (636 pm/V)
for NAs and 2.6 eV or 476.9 nm (270 pm/V) for NSb. These
are near the elemental nitrogen (337.1 nm) and Argon (476.5
nm) gas lasers. In addition, we have also shown the real and
imaginary part of the SHG spectra for a complete under-
standing of how these are coupled together through the
Kramers and Kronig relation.15
Optical Coefficients in the Presence of an in-Plane
Biaxial Strain. We now focus our attention on the
improvement of these optical coefficients by applying a
suitable in-plane strain. Recently, there have been some
benchmark experiments which show that it is possible to strain
the traditional monolayers122 both uniaxially32,123,124 and
biaxially125,126 and are supported by various ab initio
studies.127−131 In principle, any application of strain on a
two-dimensional system tends to alter the lattice constant and
primarily changes the electronic overlap wave functions,
thereby tuning the electron and phonon eigenenergies.
Heuristically, a uniaxial strain can be applied mechanically by
depositing the monolayer onto a flexible substrate and bending
the latter. This procedure may produce a homogeneous strain
but breaks down the initial crystalline symmetry. A biaxial
strain instead can be applied by depositing the monolayer onto
a substrate with a smaller thermal expansion coefficient. Upon
heating or cooling, a homogeneous biaxial tensile or
compressive strain can respectively be achieved, which also
preserves the initial crystalline symmetry. Since the thermal
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
expansion or cooling is a time-bounded interplay, thus the
biaxial strain becomes a time-dependent effect. This is contrary
to the uniaxial case, where the time-dependency could be made
very slow depending on the substrate stiffness toward bending.
As a result, the biaxial strain may find usefulness in dedicated
optical devices where real-time dynamics are primarily
important, like optical sensors, etc.
In the following, we apply an in-plane biaxial strain σij by
altering the equilibrium in-plane lattice constant. Here i and j
denote the usual in-plane directions. While carrying out an ab
initio computation, this in-plane strain is evaluated from the
real-space translation lattice vector ⎯→
⎯a ′ = (1 + σ ) ⎯→
ij
⎯
ij a ij . This
also preserves the initial crystalline symmetry, and thus the
optical nonlinear coefficients are expected to remain as
nonvanishing quantities. We note here that since the
traditional monolayers are reported to withstand up to a strain
as large as 11% while in their suspended conditions;132
therefore, up to a ± 10% strain is used here. All convergence
energy thresholds used in the strainless conditions are found to
be sufficient in the above-mentioned strain range. Table 6
shows the dynamic stability of the optical phonons at Γ of the
BZ. We observe that the influence of a tensile strain up to
+10% actually reduces the optical mode frequencies in both
the monolayers without splitting ωLO − ωTO at Γ. However,
the difference between the optical phonon-gap frequency, i.e.,
between ωLO − ωTO and ωZO, increases with strain. In the case
of a compressive strain, these scenarios behave in just opposite
ways. The mode frequency increases together with the
reduction of the optical phonon gap. The acoustic modes,
however, show an interesting behavior. It appears that the
effect of tensile strain removes the imaginary frequencies in the
phonon dispersion of both the monolayers. A tensile strain
beyond +4% starts to recover the well-known quadratic
behavior of the out-of-plane acoustic mode, indicating better
dynamic stability. For example, see the phonon dispersion at
+8% tensile strain demonstrated in Figure 21. We found that
the compressive strain tends to put more imaginary frequencies
in the out-of-plane acoustic mode, thereby destabilizing both
the monolayers (see Figures S11 and S12 in the Supporting
Information for a complete phonon dispersion in the presence
of tensile and compressive strains). Such a tendency indicates
Figure 21. Phonon dispersion in monolayers NAs (left) and NSb
(right) for biaxial strain of 8%.
14950
pubs.acs.org/JPCC Article
that a biaxial tensile strain might prove to thermally stabilize
these monolayers.
Figure 22 shows the G0W0 electronic gap variations at
different strain levels. The strain impact on the excited-state
energies are clear from subplots (a) and (b) exhibiting a
decrease in the direct band gap as the monolayer stretches.
However, the gap tends to increase with compression also until
it starts to decrease beyond −8% for NAs and −4% for NSb. A
careful analysis indicates that in the case of NAs (NSb), the
direct gap between M and Γ shifts to K below −8% (−4%)
strain. The indirect gap, however, varies slowly at compressive
strains for NAs, but significantly reduces upon stretching. A
similar trend is seen in the case of NSb, which indicates that
during compressive strain, the dynamic correlation is less
sensed toward the energy differences between indirect gap
points.
The influence of strain on the absorption spectra and JDOS
are shown in Figures 23 and 24. In the subplots, we have also
included the RPA spectra for each strain level for comparison.
We see that in the case of NAs, the tensile strain tends to
redshift the first bound exciton. This exciton in the presence of
strain (B*) is now also reduced in intensity compared to its
strain-less counterpart B1. The excitonic peak, however,
continues to red-shift until +8%. At 10%, we see some
emergence of shoulder peaks on the left of B* indicating
electronic transitions from different parts of the BZ (see Figure
26). Additionally, we see the effect of the decrease in the
corresponding G0W0 gaps, which anticipate the decreasing
trends in the binding energies (shown in Figure 25).
Surprisingly, the resonant peak near 5.37 eV gets more
intense, indicating it to become a more bright exciton. In
effect, we see that the tensile strain (between 0% to 10%) tends
to reduce the optical transparency (on-set of 10% of the
fundamental excitonic peak position) region by about 0.9 eV.
Instead, the compressive strain shows a completely opposite
behavior. The fundamental exciton B* now blue shifts and
tends to become more bright with increasing strain, whereas
the resonant peak gradually dies off. Between −8% to −10%
strain, we rather see a dramatic movement. The peak B* at
−8% has now emerged as two prominent peaks (see Figure
23j) specifying the on-set of optical transitions from other BZ
points (see Figure 27 below). We, therefore, see that the
compressive strain (between 0% and −10%) tends to increase
the optical transparency by about 0.4 eV in NAs. In the case of
NSb, we observe similar trends except for the emergence of the
secondary peak C* at higher tensile strain. At compressive
strains, we see that the peak B* gradually diminishes, and while
many new secondary peaks A*, C*, and D* start to emanate.
These are possible due to various optical transitions occurring
at different BZ points. In effect, the optical transparencies
reduce by about 1 eV in the case of tensile, while increase with
minor modulations at higher negative strains (see Figure 24i,j).
The fundamental bound excitonic energies are shown in Figure
25a, where we observe the movements of the peaks with strain
in both monolayers. The binding energies in Figure 25b rather
show a slow variation in NAs (between +2 and +8% strain).
This also indicates that these energies in NAs remain almost
immune to tensile strain. A possible justification for this
behavior is because of the simultaneous change in the G0W0
gap and the excitonic energies that try to keep the difference
constant. This does not seem to be true for NSb at larger
tensile strain where the binding energies reduce sharply below
50% of its value at the 0 strain. It is worth mentioning here that
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 22. Quasi-particle G0W0 gaps (in eV) in biaxially strained (a) NAs and (b) NSb monolayers.

Figure 23. Absorption spectra and joint density of states of monolayer NAs under biaxial (a)−(e) tensile and (f)−(j) compressive strains,
respectively. The vertical dotted lines are the most important electronic transitions with maximum relative excitonic intensity, while the dashed
lines are the direct quasi-particle gaps.

the trend of red and blue shifts of the fundamental peaks with
tensile and compressive biaxial strains, respectively, was also
found similar to both theoretically and experimentally for
monolayer MoS2.126 The influence of the biaxial strain on the
position of the first exciton weights in the BZ is shown in
Figures 26 and 27. In both materials, the effect of tensile strain
is the gradual shifting of the first bound exciton weight from
the region between M Γ to the Γ point of BZ. The
compressive strain is resulting in a shift from M Γ to the K
point in NSb. This shifting behavior is mainly due to changes
14951
in band structures induced by strain. To explain this, one may,
however, look at the GW band structures, which are shown
below in Figures 28 and 29. The most significant influence of
tensile biaxial strain on the band structures of both materials is
in the region of Γ point of the BZ, where the gap is
significantly lowering, resulting in nearly another valley
degeneracy of conduction band at Γ under +10% biaxial
strain. The influence of the tensile biaxial strain on the valence
bands and the compressive strain on both bands is less
significant. However, this interplay of changes in valence and
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 24. Absorption spectra and joint density of states of monolayer NSb under biaxial (a)-(e) tensile and (f)-(j) compressive strains,
respectively. The vertical dotted lines are the most important electronic transitions with maximum relative excitonic intensity, while the dashed
lines are the direct quasi-particle gaps.

Figure 25. (a) First exciton energies and (b) excitonic binding energies in biaxially strained NAs and NSb monolayers.

conduction bands is the leading cause of shifting the excitonic
weights of the first exciton.
Variations in nonlinear SHG coefficients in the presence of
the aforementioned strains are shown in Figures 30 and 31 for
NAs and NSb, respectively. We see that the influence of a
tensile strain is to again redshift the SHG spectra in both cases,
while the compressive strain tends to blueshift. This behavior is
similar to the corresponding linear spectra shown in Figures 23
and 24. In the case of NAs, we find from Figure 30a that at
these 810 and 1560 nm loss-less windows, a strain to +10%
improves |χ(2)aab(ω)| to about 88% (∼105 pm/V) and 135%
(∼40 pm/V) compared to the zero-strain conditions,
respectively. In addition, we also see some adversity in which
the tensile strain degrades the brightness of the most
prominent peak, which slows down from 3.2 eV (zero strain)
to about 2.7 eV (+10% strain). In the case of compressive
strain, as shown in Figure 30b, we find that the most
prominent peak in the region 3.0−4.0 eV reduces from −8 to
−10% which could possibly be because of the change in
position of the optical transitions in the BZ (see Figure 26)
above). The spectra of NSb show more dramatic results. We
see from Figure 31a that the strain to +10% improves |χ(2)
aab(ω)|

14952 https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 26. Excitonic weights showing the most important electronic transitions along the high symmetry points of the BZ of monolayer NAs under
biaxial (a)−(e) tensile and (f), (g) compressive strains respectively. The horizontal lines indicate zero strain band edges.

Figure 27. Excitonic weights showing the most important electronic transitions along the high symmetry points of the BZ of monolayer NSb under
biaxial (a)−(e) tensile and (f), (g) compressive strains respectively. The horizontal lines indicate zero strain band edges.

to about 193 pm/V (∼615%) and 79 pm/V (∼155%) at 810
and 1560 nm, respectively. The results shown here, particularly
in the lossless windows, seem to be better than most traditional
14953
TMDCs.131 The influence of the strain on the absorption
spectra, the extinction coefficient k, and the refractive index n is
shown in Figures S13 and S14.
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 28. Ground-state (gray) and G0W0 (red) band structure of monolayer NAs under various biaxial (a)−(e) tensile and (f), (g) compressive
strains respectively.

Figure 29. Ground-state (gray) and G0W0 (red) band structure of monolayer NSb under various biaxial (a)−(e) tensile and (f), (g) compressive
strains respectively.

We finally note that all the important resonances discussed
here are 2ω resonances based on the qualitative arguments. A
more general theory on Sipe’s117 sum-over states might shine
important guidelines on excitonic transitions. Moreover, the
discussion of the temperature-dependent spectra in this work
14954
avoided the effects from lattice thermal expansion (LTE). It is
well-known that the band-gaps also depend on the LTE,133
however, for most semiconductors, the effects from LTE are
minor toward the electronic eigenvalues,63 and therefore, they
have been excluded.
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Figure 30. Improvements in second harmonic coefficients in monolayer NAs under biaxial (a) tensile and (b) compressive strains respectively. The
dotted and dashed lines are the two important wavelengths 1560 and 810 nm, respectively.

Figure 31. Improvements in second harmonic coefficients in monolayer NSb under biaxial (a) tensile and (b) compressive strains, respectively.
The dotted and dashed lines are the two important wavelengths 1560 and 810 nm, respectively.

■ CONCLUSIONS
We present a detailed study of the fundamental properties of
two new monolayer materials, nitrogen arsenide NAs, and
nitrogen antimony NSb in their β-phases. We used a
combination of advanced many-body techniques to study
electron−electron, electron−hole, electron−phonon, and even
exciton−phonon interactions, leading to unique descriptions
and deciphering of extraordinary properties of materials under
investigation. We provided deep insights into the effects of
temperature and biaxial strain on the electronic structure,
accurate optical absorption spectra, and nonlinear optics.
Both wide-gap semiconductors, NAs and NSb (indirect
quasiparticle gap ∼4.5 and ∼3.7 eV, respectively), embody
strong exciton binding energies, ∼1.6 and ∼1.3 eV,
respectively. While we observed giant zero-point exciton
binding energy renormalization of 430 meV in NSb (red-
shift), we detected almost no renormalization in NAs. In
opposite, finite-temperature optical absorption in NSb is quite
14955
resilient to the temperature variations between 0 and 450 K
(with the dominant bound excitonic peak around 2.8 eV only
slightly red-shifting). In NAs, however, we showed stronger
blue-shifting and broadening of spectra region around
dominant bound excitonic 3.4 eV peak. The other important
NAs spectra region around dominant 5.4 eV resonant exciton
was almost flattened by higher temperatures. The maximum
nonlinear second harmonic optical coefficient is large: 636
pm/V at 3.2 eV (387.5 nm) for NAs and 270 pm/V at 2.6 eV
(476.9 nm) for NSb. We further proved that even a small
tensile biaxial strain stabilizes considered materials with no loss
in absorbance spectra; therefore, the practical preparation of
these materials on support or included in heterostructures is
possible. Generally, NAs and NSb seem as promising
semiconductors with potential for use in optoelectronic
applications.
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
■
ASSOCIATED CONTENT
* Supporting Information
sı
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.jpcc.2c03708.
Kinetic energy cutoff convergences, bare electronic
orbital dispersion, single-shot GW band gap conver-
gence plots, absorption spectra convergence plots,
spectral function plots, zero Kelvin electronic dispersion,
strain effect on phonons, and strain effects on refractive
index, extinction coefficient and absorbance (PDF)
visualization of phonon modes of both materials (ZIP)
■
AUTHOR INFORMATION
Corresponding Authors
František Karlický − Department of Physics, Faculty of
Science, University of Ostrava, 701 03 Ostrava, Czech
Republic; orcid.org/0000-0002-2623-3991;
Email: frantisek.karlicky@osu.cz
Sitangshu Bhattacharya − Electronic Structure Theory Group,
Department of Electronics and Communication Engineering,
Indian Institute of Information Technology, Allahabad, Uttar
Pradesh 211015, India; orcid.org/0000-0002-4442-
5225; Email: sitangshu@iiita.ac.in
Authors
Miroslav Kolos − Department of Physics, Faculty of Science,
University of Ostrava, 701 03 Ostrava, Czech Republic;
orcid.org/0000-0003-3821-7814
Rekha Verma − Nanoscale Electro-thermal Laboratory, Indian
Institute of Information Technology, Allahabad, Uttar
Pradesh 211015, India; orcid.org/0000-0003-1636-2092
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.jpcc.2c03708
Author Contributions
The manuscript was written through the contributions of all
authors. All authors have given their approval to the final
version of the manuscript.
Notes
The authors declare no competing financial interest.
■ ACKNOWLEDGMENTS
The present work is carried out under the India-Czech
Bilateral Scientific and Technological Cooperation (Indian No.
DST/INT/Czech/P-14/2019, Czech InterAction program No.
LTAIN19138). Part of the work is also supported under
Science and Engineering Research Board (SERB)-MATRICS
MTR/2021/000017, India. Portion of the calculations were
performed at IT4 Innovations National Supercomputing
Center through the e-INFRA CZ (ID:90140). The authors
acknowledge the Central Computation Facility at Indian
Institute of Information Technology-Allahabad for all the
major computations.
■ REFERENCES
(1) Aryasetiawan, F.; Gunnarsson, O. The GW method. Rep. Prog.
Phys. 1998, 61, 237.
(2) Rohlfing, M.; Louie, S. G. Electron-hole excitations and optical
spectra from first principles. Phys. Rev. B 2000, 62, 4927.
(3) Onida, G.; Reining, L.; Rubio, A. Electronic excitations: density-
functional versus many-body Green’s-function approaches. Rev. Mod.
Phys. 2002, 74, 601.
14956
pubs.acs.org/JPCC Article
(4) Karlický, F.; Turoň, J. Fluorographane C2FH: Stable and wide
band gap insulator with huge excitonic effect. Carbon 2018, 135,
134−144.
(5) Arnaud, B.; Lebègue, S.; Rabiller, P.; Alouani, M. Huge Excitonic
Effects in Layered Hexagonal Boron Nitride. Phys. Rev. Lett. 2006, 96,
026402.
(6) Ramasubramaniam, A. Large excitonic effects in monolayers of
molybdenum and tungsten dichalcogenides. Phys. Rev. B 2012, 86,
115409.
(7) Wang, X.; Jones, A. M.; Seyler, K. L.; Tran, V.; Jia, Y.; Zhao, H.;
Wang, H.; Yang, L.; Xu, X.; Xia, F. Highly anisotropic and robust
excitons in monolayer black phosphorus. Nat. Nanotechnol. 2015, 10,
517.
(8) Palummo, M.; Bernardi, M.; Grossman, J. C. Exciton Radiative
Lifetimes in Two-Dimensional Transition Metal Dichalcogenides.
Nano Lett. 2015, 15, 2794.
(9) Echeverry, J. P.; Urbaszek, B.; Amand, T.; Marie, X.; Gerber, I.
C. Splitting between bright and dark excitons in transition metal
dichalcogenide monolayers. Phys. Rev. B 2016, 93, 121107R.
(10) Wang, G.; Chernikov, A.; Glazov, M. M.; Heinz, T. F.; Marie,
X.; Amand, T.; Urbaszek, B. Colloquium: Excitons in atomically thin
transition metal dichalcogenides. Phys. Rev. B 2018, 90, 021001.
(11) Biswas, T.; Singh, A. K. Excitonic effects in absorption spectra
of carbon dioxide reduction photocatalysts. Npj Comput. Mater. 2021,
7, 189.
(12) Rasmussen, A.; Deilmann, T.; Thygesen, K. S. Towards fully
automatized GW band structure calculations: What we can learn from
60.000 self-energy evaluations. Npj Comput. Mater. 2021, 7, 22.
(13) Dubecký, M.; Karlický, F.; Minárik, S.; Mitas, L. Fundamental
gap of fluorographene by many-body GW and fixed-node diffusion
Monte Carlo methods. J. Chem. Phys. 2020, 153, 184706.
(14) Shen, Y. R. Surface properties probed by second-harmonic and
sum-frequency generation. Nature 1989, 337, 519.
(15) Boyd, R. W. Nonlinear Opt., 3rd ed.; Academic Press: New
York, 2008.
(16) Runge, E.; Gross, E. K. U. Density-Functional Theory for
Time-Dependent Systems. Phys. Rev. Lett. 1984, 52, 997.
(17) Ketolainen, T.; Macháčová, N.; Karlický, F. Optical Gaps and
Excitonic Properties of 2D Materials by Hybrid Time-Dependent
Density Functional Theory: Evidences for Monolayers and Prospects
for van der Waals Heterostructures. J. Chem. Theory Comput. 2020,
16, 5876−5883.
(18) Attaccalite, C.; Grüning, M.; Marini, A. Real-time approach to
the optical properties of solids and nanostructures: Time-dependent
Bethe-Salpeter equation. Phys. Rev. B 2011, 84, 245110.
(19) King-Smith, R. D.; Vanderbilt, D. Theory of polarization of
crystalline solids. Phys. Rev. B 1993, 47, 1651.
(20) Attaccalite, C.; Grüning, M. Nonlinear optics from an ab-initio
approach by means of the dynamical Berry phase: Application to
second- and third-harmonic generation in semiconductors. Phys. Rev.
B 2013, 88, 235113.
(21) Marini, A. Ab initio finite-temperature excitons. Phys. Rev. Lett.
2008, 101, 106405.
(22) Molina-Sànchez, A.; Palummo, M.; Marini, A.; Wirtz, L.
Temperature-dependent excitonic effects in the optical properties of
single-layer MoS2. Phys. Rev. B 2016, 93, 155435.
(23) Mishra, H.; Bose, A.; Dhar, A.; Bhattacharya, S. Exciton-
phonon coupling and band-gap renormalization in monolayer WSe2.
Phys. Rev. B 2018, 98, 045143.
(24) Cannuccia, E.; Monserrat, B.; Attaccalite, C. Theory of
phonon-assisted luminescence in solids: application to hexagonal
boron nitride. Phys. Rev. B 2019, 99, 081109.
(25) Shen, T.; Zhang, X.-W.; Shang, H.; Zhang, M.-Y.; Wang, X.;
Wang, E.-G.; Jiang, H.; Li, X.-Z. Influence of high-energy local orbitals
and electron-phonon interactions on the band gaps and optical
absorption spectra of hexagonal boron nitride. Phys. Rev. B 2020, 102,
045117.
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
(26) Mishra, H.; Bhattacharya, S. Giant exciton-phonon coupling
and zero-point renormalization in hexagonal monolayer boron nitride.
Phys. Rev. B 2019, 99, 165201.
(27) Kolos, M.; Cigarini, L.; Verma, R.; Karlický, F.; Bhattacharya, S.
Giant Linear and Nonlinear Excitonic Responses in an Atomically
Thin Indirect Semiconductor Nitrogen Phosphide. J. Phys. Chem. C
2021, 125, 12738−12757.
(28) Kawai, H.; Yamashita, K.; Cannuccia, E.; Marini, A. Electron-
electron and electron-phonon correlation effects on the finite-
temperature electronic and optical properties of zinc-blende GaN.
Phys. Rev. B 2014, 89, 085202.
(29) Cannuccia, E.; Marini, A. Ab-initio study of the effects induced
by the electron-phonon scattering in carbon based nanostructures.
arXiv 2013, V.
(30) Mueller, T.; Malic, E. Exciton physics and device application of
two-dimensional transition metal dichalcogenide semiconductors. npj
2D Mater. Appl. 2018, 2, 29.
(31) Carvalho, A.; Ribeiro, R. M.; Castro Neto, A. H. Band nesting
and the optical response of two-dimensional semiconducting
transition metal dichalcogenides. Phys. Rev. B 2013, 88, 115205.
(32) Mennel, L.; Smejkal, V.; Linhart, L.; Burgdörfer, J.; Libisch, F.;
Mueller, T. Band nesting in two-dimensional crystals: an exceptionally
sensitive probe of strain. Nano Lett. 2020, 20, 4242.
(33) Taheri, A.; Da Silva, C.; Amon, C. H. Phonon thermal transport
in β−NX (X = P, As, Sb) monolayers: A first-principles study of the
interplay between harmonic and anharmonic phonon properties. Phys.
Rev. B 2019, 99, 235425.
(34) Giannozzi, P.; Andreussi, O.; Brumme, T.; Bunau, O.; Nardelli,
M. B.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Cococcioni,
M.; et al. Advanced capabilities for materials modelling with Quantum
ESPRESSO. J. Phys.: Condens. Matter 2017, 29, 465901.
(35) Hamann, D. Optimized norm-conserving Vanderbilt pseudo-
potentials. Phys. Rev. B 2013, 88, 085117.
(36) Antonius, G.; Poncé, S.; Lantagne-Hurtubise, E.; Auclair, G.;
Gonze, X.; Côté, M. Dynamical and anharmonic effects on the
electron-phonon coupling and the zero-point renormalization of the
electronic structure. Phys. Rev. B 2015, 92, 085137.
(37) Sangalli, D.; Ferretti, A.; Miranda, H.; Attaccalite, C.; Marri, I.;
Cannuccia, E.; Melo, P.; Marsili, M.; Paleari, F.; Marrazzo, A.; et al.
Many-body perturbation theory calculations using the yambo code. J.
Phys.: Condens. Matter 2019, 31, 325902.
(38) Godby, R. W.; Needs, R. J. Metal-insulator transition in Kohn-
Sham theory and quasiparticle theory. Phys. Rev. Lett. 1989, 62, 1169.
(39) Kammerlander, D.; Botti, S.; Marques, M. A. L.; Marini, A.;
Attaccalite, C. Speeding up the solution of the Bethe-Salpeter
equation by a double-grid method and Wannier interpolation. Phys.
Rev. B 2012, 86, 125203.
(40) Cannuccia, E. Giant polaronic effects in polymers: breakdown
of the quasiparticle picture. Ph.D. thesis, Rome Tor Vergata
University, 2011.
(41) Crank, J.; Nicolson, P. A practical method for numerical
evaluation of solutions of partial differential equations of the heat-
conduction type. Math. Proc. Camb. Philos. Soc. 1947, 43, 50.
(42) Mera Acosta, C.; Fazzio, A.; Dalpian, G. M. Zeeman-type spin
splitting in nonmagnetic three-dimensional compounds. npj Quantum
Materials 2019, 4, 41.
(43) Yu, W.; Niu, C.-Y.; Zhu, Z.; Wang, X.; Zhang, W.-B. Atomically
thin binary V−V compound semiconductor: a first-principles study. J.
Mater. Chem. C 2016, 4, 6581.
(44) Sohier, T. D. P.; Gibertini, M.; Calandra, M.; Mauri, F.;
Marzari, N. Breakdown of optical phonons’ splitting in two-
dimensional materials. Nano Lett. 2017, 17, 3758.
(45) Roome, N. J.; Carey, J. D. Beyond Graphene: Stable Elemental
Monolayers of Silicene and Germanene. ACS Appl. Mater. Interfaces
2014, 6, 7743−7750.
(46) Chen, M. X.; Weinert, M. Revealing the Substrate Origin of the
Linear Dispersion of Silicene/Ag(111). Nano Lett. 2014, 14, 5189−
5193.
14957
pubs.acs.org/JPCC Article
(47) Derivaz, M.; Dentel, D.; Stephan, R.; Hanf, M.-C.; Mehdaoui,
A.; Sonnet, P.; Pirri, C. Continuous Germanene Layer on Al(111).
Nano Lett. 2015, 15, 2510−2516.
(48) Guo, S.; Zhang, Y.; Ge, Y.; Zhang, S.; Zeng, H.; Zhang, H. 2D
V-V Binary Materials: Status and Challenges. Adv. Mater. 2019, 31,
1902352.
(49) Fan, H. Y. Temperature Dependence of the Energy Gap in
Monatomic Semiconductors. Phys. Rev. 1950, 78, 808.
(50) Antončík, E. On the theory of temperature shift of the
absorption curve in non-polar crystals. Czech. J. Phys. 1955, 5, 449−
461.
(51) Keffer, C.; Hayes, T. M.; Bienenstock, A. PbTe Debye-Waller
Factors and Band-Gap Temperature Dependence. Phys. Rev. Lett.
1968, 21, 1676.
(52) Walter, J. P.; Zucca, R. R. L.; Cohen, M. L.; Shen, Y. R.
Temperature Dependence of the Wavelength-Modulation Spectra of
GaAs. Phys. Rev. Lett. 1970, 24, 102.
(53) Kasowski, R. V. Band Structure of NiS as Calculated Using a
Simplified Linear-Combination-of-Muffin-Tin-Orbitals Method. Phys.
Rev. B 1973, 8, 1378.
(54) Allen, P. B.; Heine, V. Theory of the temperature dependence
of electronic band structures. J. Phys. C 1976, 9, 2305.
(55) Allen, P. B.; Cardona, M. Theory of the temperature
dependence of the direct gap of germanium. Phys. Rev. B 1981, 23,
1495.
(56) Allen, P. B.; Cardona, M. Temperature dependence of the
direct gap of Si and Ge. Phys. Rev. B 1983, 27, 4760.
(57) Lautenschlager, P.; Allen, P. B.; Cardona, M. Temperature
dependence of band gaps in Si and Ge. Phys. Rev. B 1985, 31, 2163.
(58) Giustino, F.; Louie, S. G.; Cohen, M. L. Electron-phonon
renormalization of the direct band gap of diamond. Phys. Rev. Lett.
2010, 105, 265501.
(59) Antonius, G.; Poncé, S.; Boulanger, P.; Côté, M.; Gonze, X.
Many-body effects on the zero-point renormalization of the band
structure. Phys. Rev. Lett. 2014, 112, 215501.
(60) Cannuccia, E.; Marini, A. Zero point motion effect on the
electronic properties of diamond, trans-polyacetylene and poly-
ethylene. Eur. Phys.J. B 2012, 85, 320.
(61) Cannuccia, E.; Marini, A. Effect of the quantum zero-point
atomic motion on the optical and electronic properties of diamond
and trans-polyacetylene. Phys. Rev. lett. 2011, 107, 255501.
(62) Qiao, J.; Kong, X.; Hu, Z.-X.; Yang, F.; Ji, W. High-mobility
transport anisotropy and linear dichroism in few-layer black
phosphorus. Nat. Commun. 2014, 5, 4475.
(63) Villegas, C. E.; Rocha, A. R.; Marini, A. Anomalous temperature
dependence of the band-gap in black phosphorus. Nano Lett. 2016,
16, 5095.
(64) Molina-Sánchez, A.; Sangalli, D.; Hummer, K.; Marini, A.;
Wirtz, L. Effect of spin-orbit interaction on the optical spectra of
single-layer, double-layer, and bulk MoS2. Phys. Rev. B 2013, 88,
045412.
(65) Rohlfing, M.; Krüger, P.; Pollmann, J. Efficient scheme for GW
quasiparticle band-structure calculations with applications to bulk Si
and to the Si(001)-(2 × 1) surface. Phys. Rev. B 1995, 52, 1905.
(66) Monserrat, B.; Needs, R. J. Comparing electron-phonon
coupling strength in diamond, silicon, and silicon carbide: First-
principles study. Phys. Rev. B 2014, 89, 214304.
(67) Attaccalite, C.; Nguer, A.; Cannuccia, E.; Grüning, M. Strong
second harmonic generation in SiC, ZnO, GaN two-dimensional
hexagonal crystals from first-principles many-body calculations. Phys.
Chem. Chem. Phys. 2015, 17, 9533.
(68) Chrenko, R. M. Ultraviolet and infrared spectra of cubic boron
nitride. Solid State Commun. 1974, 14, 511.
(69) Hybertsen, M. S.; Louie, S. G. Electron correlation in
semiconductors and insulators: Band gaps and quasiparticle energies.
Phys. Rev. B 1986, 34, 5390.
(70) Chaney, R. C.; Lafon, E. E.; Lin, C. C. Energy band structure of
lithium fluoride crystals by the method of tight binding. Phys. Rev. B
1971, 4, 2734.
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C
(71) Varshni, Y. Temperature dependence of the energy gap in
semiconductors. Physica 1967, 34, 149−154.
(72) Zhang, G.; Chaves, A.; Huang, S.; Wang, F.; Xing, Q.; Low, T.;
Yan, H. Determination of layer-dependent exciton binding energies in
few-layer black phosphorus. Sci. Adv. 2018, 4, 9977.
(73) He, K.; Kumar, N.; Zhao, L.; Wang, Z.; Mak, K. F.; Zhao, H.;
Shan, J. Tightly Bound excitons in monolayer WSe2. Phys. Rev. Lett.
2014, 113, 026803.
(74) Arora, A.; Koperski, M.; Nogajewski, K.; Marcus, J.; Faugerasa,
C.; Potemski, M. Excitonic resonances in thin films of WSe2: from
monolayer to bulk material. Nanoscale 2015, 7, 10421.
(75) Autere, A.; Jussila, H.; Marini, A.; Saavedra, J. R. M.; Dai, Y.;
Säynätjoki, A.; Karvonen, L.; Yang, H.; Amirsolaimani, B.; Norwood,
R. A.; et al. Optical harmonic generation in monolayer group-VI
transition metal dichalcogenides. Phys. Rev. B 2018, 98, 115426.
(76) Park, S.; Mutz, N.; Schultz, T.; Blumstengel, S.; Han, A.; Aljarb,
A.; Li, L.-J.; List-Kratochvil, E. J. W.; Amsalem, P.; et al. Direct
determination of monolayer MoS2 and WSe2 exciton binding energies
on insulating and metallic substrate. 2D Mater. 2018, 5, 025003.
(77) Hanbicki, A.T.; Currie, M.; Kioseoglou, G.; Friedman, A.L.;
Jonker, B.T. Measurement of high exciton binding energy in the
monolayer transition-metal dichalcogenides WS2and WSe2. Solid State
Commun. 2015, 203, 16.
(78) Cadiz, F.; Courtade, E.; Robert, C.; Wang, G.; Shen, Y.; Cai,
H.; Taniguchi, T.; Watanabe, K.; Carrere, H.; Lagarde, D.; et al.
Excitonic linewidth approaching the homogeneous limit in MoS2-
based van der Waals heterostructures. Phys. Rev. X 2017, 7, 021026.
(79) Miyauchi, Y.; Konabe, S.; Wang, F.; Zhang, W.; Hwang, A.;
Hasegawa, Y.; Zhou, L.; Mouri, S.; Toh, M.; Eda, G.; Matsuda, K.
Evidence for line width and carrier screening effects on excitonic
valley relaxation in 2D semiconductors. Nat. Commun. 2018, 9, 2598.
(80) Ribeiro-Soares, J.; Janisch, C.; Liu, Z.; Elías, A. L.; Dresselhaus,
M. S.; Terrones, M.; Cançado, L. G.; Jorio, A. Second harmonic
generation in WSe2. 2D Mater. 2015, 2, 045015.
(81) Rigosi, A. F.; Hill, H. M.; Rim, K. T.; Flynn, G. W.; Heinz, T. F.
Electronic band gaps and exciton binding energies in monolayer
MoxW1‑xS2 transition metal dichalcogenide alloys probed by scanning
tunneling and optical spectroscopy. Phys. Rev. B 2016, 94, 075440.
(82) Li, Y.; Rao, Y.; Mak, K. F.; You, Y.; Wang, S.; Dean, C. R.;
Heinz, T. F. Probing symmetry properties of few layer MoS2 and h-
BN by optical second-harmonic generation. Nano Lett. 2013, 13,
3329.
(83) Kumar, N.; Najmaei, S.; Cui, Q.; Ceballos, F.; Ajayan, P. M.;
Lou, J.; Zhao, H. Second harmonic microscopy of monolayer MoS2.
Phys. Rev. B 2013, 87, 161403R.
(84) Clark, D. J.; Senthilkumar, V.; Le, C. T.; Weerawarne, D. L.;
Shim, B.; Jang, J. I.; Shim, J. H.; Cho, J.; Sim, Y.; Seong, M.-J.; et al.
Strong optical nonlinearity of CVD-grown MoS2 monolayer as probed
by wavelength-dependent second-harmonic generation. Phys. Rev.B
2014, 90, 121409R.
(85) Malard, L. M.; Alencar, T. V.; Barboza, A. P. M.; Mak, K. F.; de
Paula, A. M. Observation of intense second harmonic generation from
MoS2 atomic crystals. Phys. Rev. B 2013, 87, 201401R.
(86) Hill, H. M.; Rigosi, A. F.; Roquelet, C.; Chernikov, A.;
Berkelbach, T. C.; Reichman, D. R.; Hybertsen, M. S.; Brus, L. E.;
Heinz, T. F. Observation of excitonic Rydberg states in monolayer
MoS2and WS2 by photoluminescence excitation spectroscopy. Nano
Lett. 2015, 15, 2992.
(87) Zhang, C.; Johnson, A.; Hsu, C.-L.; Li, L.-J.; Shih, C.-K. Direct
imaging of band profile in single layer MoS2 on Graphite:
quasiparticle energy gap, metallic edge states, and edge band bending.
Nano Lett. 2014, 14, 2443.
(88) Chiu, M.-H.; Zhang, C.; Shiu, H.-W.; Chuu, C.-P.; Chen, C.-
H.; Chang, C.-Y. S.; Chen, C.-H.; Chou, M.-Y.; Shih, C.-K.; Li, L.-J.
Determination of band alignment in the single-layer MoS2/WSe2
heterojunction. Nat. Commun. 2015, 6, 7666.
(89) Arora, A.; Nogajewski, K.; Molas, M.; Koperskiab, M.;
Potemski, M. Exciton band structure in layered MoSe2: from a
monolayer to the bulk limit. Nanoscale 2015, 7, 20769.
14958
pubs.acs.org/JPCC Article
(90) Selig, M.; Berghäuser, G.; Raja, A.; Nagler, P.; Schüller, C.;
Heinz, T. F.; Korn, T.; Chernikov, A.; Malic, E.; Knorr, A. Excitonic
linewidth and coherence lifetime in monolayer transition metal
dichalcogenides. Nat. Commun. 2016, 7, 13279.
(91) Zhang, Y.; Chang, T.-R.; Zhou, B.; Cui, Y.-T.; Yan, H.; Liu, Z.;
Schmitt, F.; Lee, J.; Moore, R.; Chen, Y.; et al. Direct observation of
the transition from indirect to direct bandgap in atomically thin
epitaxial MoSe2. Nat. Nanotechnol. 2014, 9, 111.
(92) Pacuski, W.; Grzeszczyk, M.; Nogajewski, K.; Bogucki, A.;
Oreszczuk, K.; Kucharek, J.; Połczyńska, K. E.; Seredyński, B.; Rodek,
A.; Bożek, R.; et al. Narrow excitonic lines and large-scale
homogeneity of transition-metal dichalcogenide monolayers grown
by molecular beam epitaxy on hexagonal boron nitride. Nano Lett.
2020, 20, 3058.
(93) Le, C. T.; Clark, D. J.; Ullah, F.; Senthilkumar, V.; Jang, J. I.;
Sim, Y.; Seong, M.-J.; Chung, K.-H.; Park, H.; Kim, Y. S. Nonlinear
optical characteristics of monolayer MoSe2. Ann. Phys. 2016, 528, 551.
(94) Robert, C.; Picard, R.; Lagarde, D.; Wang, G.; Echeverry, J. P.;
Cadiz, F.; Renucci, P.; Högele, A.; Amand, T.; Marie, X.; et al.
Excitonic properties of semiconducting monolayer and bilayer
MoTe2. Phys. Rev. B 2016, 94, 155425.
(95) Janisch, C.; Wang, Y.; Ma, D.; Mehta, N.; Elías, A. L.; Perea-
López, N.; Terrones, M.; Crespi, V.; Liu, Z. Extraordinary second
harmonic generation in tungsten disulfide monolayers. Sci. Rep. 2015,
4, 5530.
(96) Stier, A. V.; M, K.; McCreary; Jonker, B. T.; Kono, J.; Crooker,
S. A. Exciton diamagnetic shifts and valley Zeeman effects in
monolayer WS2 and MoS2 to 65 T. Nat. Commun. 2016, 7, 10643.
(97) Zhu, B.; Chen, X.; Cui, X. Exciton binding energy of monolayer
WS2. Sci. Rep. 2015, 5, 9281.
(98) Guo, L.; Chen, C.-A.; Zhang, Z.; Monahan, D. M.; Lee, Y.-H.;
Fleming, G. R. Lineshape characterization of excitons in monolayer
WS2 by two-dimensional electronic spectroscopy. Nanoscale Adv.
2020, 2, 2333.
(99) Hamer, M. J.; Zultak, J.; Tyurnina, A. V.; Zólyomi, V.; Terry,
D.; Barinov, A.; Garner, A.; Donoghue, J.; Rooney, A. P.; Kandyba, V.;
et al. Indirect to direct gap crossover in two-dimensional InSe
revealed by angle-resolved photoemission spectroscopy. ACS Nano
2019, 13, 2136.
(100) Leisgang, N.; Roch, J. G.; Froehlicher, G.; Hamer, M.; Terry,
D.; Gorbachev, R.; Warburton, R. J. Optical second harmonic
generation in encapsulated single-layer InSe editors-pick. AIP Adv.
2018, 8, 105120.
(101) Zhou, X.; Cheng, J.; Zhou, Y.; Cao, T.; Hong, H.; Liao, Z.;
Wu, S.; Peng, H.; Liu, K.; Yu, D. Strong second-harmonic generation
in atomic layered GaSe. J. Am. Chem. Soc. 2015, 137, 7994.
(102) Elias, C.; Valvin, P.; Pelini, T.; Summerfield, A.; Mellor, C.;
Cheng, T.; Eaves, L.; Foxon, C.; Beton, P.; Novikov, S.; et al. Direct
band-gap crossover in epitaxial monolayer boron nitride. Nat.
Commun. 2019, 10, 2639.
(103) Kim, S.; Fröch, J. E.; Gardner, A.; Li, C.; Aharonovich, I.;
Solntsev, A. S. Second-harmonic generation in multilayer hexagonal
boron nitride flakes. Opt. Lett. 2019, 44, 5792.
(104) Mahan, G. D. Many-Particle Physics, 3rd ed.; Springer
International ed.: New York, 2014.
(105) Grüning, M.; Marini, A.; Gonze, X. Exciton-plasmon States in
nanoscale materials: breakdown of the Tamm-Dancoff approximation.
Nano Lett. 2009, 9, 2820−2824.
(106) Moody, G.; Dass, C. K.; Hao, K.; Chen, C.-H.; Li, L.-J.; Singh,
A.; Tran, K.; Clark, G.; Xu, X.; Berghaüser, G.; et al. Intrinsic
homogeneous linewidth and broadening mechanisms of excitons in
monolayer transition metal dichalcogenides. Nat. Commun. 2015, 6,
8315.
(107) Dey, P.; Paul, J.; Wang, Z.; Stevens, C. E.; Liu, C.; Romero, A.
H.; Shan, J.; Hilton, D. J.; Karaiskaj, D. Optical coherence in atomic-
monolayer transition-metal dichalcogenides limited by electron-
phonon interactions. Phys. Rev. Lett. 2016, 116, 127402.
(108) Hsu, W.-T.; Zhao, Z.-A.; Li, L.-J.; Chen, C.-H.; Chiu, M.-H.;
Chang, P.-S.; Chou, Y.-C.; Chang, W.-H. Second Harmonic
https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959
The Journal of Physical Chemistry C pubs.acs.org/JPCC Article

Generation from Artificially Stacked Transition Metal Dichalcogenide (130) Khatibi, Z.; Feierabend, M.; Selig, M.; Brem, S.; Linderälv, C.;
Twisted Bilayers. ACS Nano 2014, 8, 2951−2958. Erhart, P.; Malic, E. Impact of strain on the excitonic linewidth in
(109) Attaccalite, C.; Palummo, M.; Cannuccia, E.; Grüning, M. transition metal dichalcogenides. 2D Mater. 2019, 6, 015015.
Second-harmonic generation in single-layer monochalcogenides: A (131) Beach, K.; Lucking, M. C.; Terrones, H. Strain dependence of
response from first-principles real-time simulations. Phys. Rev. Mat second harmonic generation in transition metal dichalcogenide
2019, 3, 074003. monolayers and the fine structure of the C exciton. Phys. Rev. B
(110) Grüning, M.; Sangalli, D.; Attaccalite, C. Dielectrics in a time- 2020, 101, 155431.
dependent electric field: A real-time approach based on density- (132) Bertolazzi, S.; Brivio, J.; Kis, A. Stretching and breaking of
polarization functional theory. Phys. Rev. B 2016, 94, 035149. ultrathin MoS2. ACS Nano 2011, 5, 9703.
(111) Souza, I.; Iniguez, J. I.; Vanderbilt, D. Dynamics of Berry- (133) Mounet, N.; Marzari, N. First-principles determination of the
phase polarization in time-dependent electric fields. Phys. Rev. B 2004, structural, vibrational and thermodynamic properties of diamond,
graphite, and derivatives. Phys. Rev. B 2005, 71, 205214.
69, 085106.
(112) Resta, R. Macroscopic polarization in crystalline dielectrics:
the geometric phase approach. Rev. Mod. Phys. 1994, 66, 899.
(113) Attaccalite, C.; Cannuccia, E.; Grüning, M. Excitonic effects in
third-harmonic generation: The case of carbon nanotubes and
nanoribbons. Phys. Rev. B 2017, 95, 125403.
(114) Aversa, C.; Sipe, J. E. Nonlinear optical susceptibilities of
semiconductors: Results with a length-gauge analysis. Phys. Rev. B
1995, 52, 14636.
(115) Moss, D. J.; Ghahramani, E.; Sipe, J. E.; van Driel, H. M.
Band-structure calculation of dispersion and anisotropy in χ →(3) for
third-harmonic generation in Si, Ge, and GaAs. Phys. Rev. B 1990, 41,
1542.
(116) Hughes, J. L. P.; Sipe, J. E. Calculation of second-order optical
response in semiconductors. Phys. Rev. B 1996, 53, 10751.
(117) Sipe, J. E.; Shkrebtii, A. I. Second-order optical response in
semiconductors. Phys. Rev. B 2000, 61, 5337.
(118) Virk, K. S.; Sipe, J. E. Semiconductor optics in length gauge: A
general numerical approach. Phys. Rev. B 2007, 76, 035213.
(119) Sipe, J. E.; Ghahramani, E. Nonlinear optical response of
semiconductors in the independent-particle approximation. Phys. Rev.
B 1993, 48, 11705.
(120) Takimoto, Y.; Vila, F. D.; Rehri, J. J. Real-time time-
dependent density functional theory approach for frequency-depend-
ent nonlinear optical response in photonic molecules. J. Chem. Phys. Recommended by ACS
2007, 127, 154114.
(121) Narsimha Rao, E.; Vaitheeswaran, G.; Reshak, A. H.; Auluck, Lone Pair Electrons with Weak Nuclear Binding Inducing
S. Role of spin-orbit interaction on the nonlinear optical response of Sensitive Nonlinear Optical Responses in Phosphorus
CsPbCO3F using DFT. Phys. hem. Chem. Phys. 2017, 19, 31255. Clusters
(122) Peng, Z.; Chen, X.; Fan, Y.; Srolovitz, D. J.; Le, D. Strain
Quanjie Zhong.
engineering of 2D semiconductors and graphene: from strain fields to
JULY 07, 2023
band-structure tuning and photonic applications. Nat. Light Sci. Appl. THE JOURNAL OF PHYSICAL CHEMISTRY LETTERS READ
2020, 9, 1.
(123) Aslan, O. B.; Datye, I. M.; Mleczko, M. J.; Cheung, K. S.;
Full Temporal Characterization of Ultrabroadband Few-
Krylyuk, S.; Bruma, A.; Kalish, I.; Davydov, A. V.; Pop, E.; Heinz, T.
Cycle Laser Pulses Using Atomically Thin WS2
F. Probing the optical properties and strain-tuning of ultrathin
Mo1−xWxTe2. Nano Lett. 2018, 18, 2485. Óscar Pérez-Benito, Rosa Weigand, et al.
(124) Aslan, O. B.; Deng, M.; Heinz, T. F. Strain tuning of excitons JUNE 15, 2023
ACS PHOTONICS READ
in monolayer WSe2. Phys. Rev. B 2018, 98, 115308.
(125) Ahn, G. H.; Amani, M.; Rasool, H.; Lien, D.-H.; Mastandrea,
J. P.; J, W. A., III; Dubey, M.; Chrzan, D. C.; Minor, A. M.; Javey, A. Strain Engineered Electrically Pumped SiGeSn Microring
Strain-engineered growth of two-dimensional materials. Nat. Lasers on Si
Commun. 2017, 8, 608. Bahareh Marzban, Dan Buca, et al.
(126) Frisenda, R.; Drüppel, M.; Schmidt, R.; de Vasconcellos, S. DECEMBER 16, 2022
M.; de Lara, D. P.; Bratschitsch, R.; Rohlfing, M.; Castellanos-Gomez, ACS PHOTONICS READ
A. Biaxial strain tuning of the optical properties of single-layer
transition metal dichalcogenides. npj 2D Mater. Appl. 2017, 10, 1. Ultrafast Thermionic Electron Injection Effects on Exciton
(127) Villegas, C. E. P.; Rocha, A. R. Elucidating the optical Formation Dynamics at a van der Waals
properties of novel heterolayered materials based on MoTe2 InN for Semiconductor/Metal Interface
photovoltaic applications. J. Phys. Chem. C 2015, 119, 11886. Kilian R. Keller, Nicolò Maccaferri, et al.
(128) Rhim, S. H.; Kim, Y. S.; Freeman, A. J. Strain-induced giant JULY 20, 2022
second-harmonic generation in monolayered 2H-MoX2 (X = S, Se, ACS PHOTONICS READ
Te). Appl. Phys. Lett. 2015, 107, 241908.
(129) Zollner, K.; Junior, P. E. F.; Fabian, J. Strain-tunable orbital,
Get More Suggestions >
spin-orbit, and optical properties of monolayer transition-metal
dichalcogenides. Phys. Rev. B 2019, 100, 195126.

14959 https://doi.org/10.1021/acs.jpcc.2c03708
J. Phys. Chem. C 2022, 126, 14931−14959