Brightened emission of dark trions in transition-metal dichalcogenide monolayers
V. Jindal1*, K. Mourzidis1*, A. Balocchi1, C. Robert1, P. Li2, D. Van Tuan2, L. Lombez1, D.
Lagarde1, P. Renucci1, T. Taniguchi3, K. Watanabe4, H. Dery2,5† and X. Marie1†
1
Université de Toulouse, INSA-CNRS-UPS, LPCNO, 135 Av. Rangueil, 31077 Toulouse, France
2
Department of Electrical and Computer Engineering, University of Rochester, Rochester, New
York 14627, USA
3
International Center for Materials Nanoarchitectonics, National Institute for Materials Science,
1-1 Namiki, Tsukuba 305-00044, Japan
4
Research Center for Functional Materials, National Institute for Materials Science, 1-1 Namiki,
Tsukuba 305-00044, Japan
5
Department of Physics and Astronomy, University of Rochester, Rochester, New York 14627,
USA
The optical emission spectra of semiconducting transition-metal dichalcogenide monolayers
highlight fascinating recombination processes of charged excitons (trions). When charge tunable
WSe2 monolayers are moderately doped with electrons, a strong luminescence peak emerges just
below the well-understood spectral lines associated with the recombination of negatively charged
bright and dark trions. Despite previous investigations, its origin remains elusive. Here, we
demonstrate that this luminescence peak is the result of electron-electron assisted recombination
that brightens the dark trion emission. Supporting evidence for this second-order recombination
process comes from identifying the equivalent brightened emission of positively charged dark
trions when the monolayer is electrostatically doped with holes. Remarkably, the discovered hole-
hole assisted luminescence peak emerges in the near infrared, about 500 meV below the well-
studied spectral region of excitons and trions. In addition to identifying new recombination
channels of these excitonic complexes, our findings accurately determine the spin-split energies of
the conduction and valence bands. Both of which play crucial roles in understanding the optical
properties of WSe2 based homo- and hetero-structures.
*These authors contributed equally to this work.
†
hdery@[Link] , marie@[Link]
The optical properties of atomically thin semiconductors based on transition-metal dichalcogenides
(TMDs) are dominated by robust exciton complexes which give rise to rich physics associated with various
interactions and correlation effects1–3. Charge-tunable WSe2 monolayers are possibly the most studied
model system due to their relatively high quality. When electrostatically doped with electrons, the
luminescence spectra of the monolayer exhibit well identified peaks corresponding to the optical
recombination of bright and dark negatively charged excitons (trions) 4–7 . Surprisingly, between slight and
moderate densities of resident electrons, the most intense luminescence line does not correspond to these
direct optical recombination channels: an intense luminescence component lying ~10 meV below the dark
trion has been clearly identified by many groups in charge tunable WSe2 monolayer devices 4,8–11 . Two
fundamental questions arise: which exciton complex is associated with this optical recombination? And
what is the nature of the recombination process? Several different interpretations were proposed. It was
suggested that the line could result from the trion’s fine structure4, the recombination of double charged
trion X2-,9 the coupling between excitons and short-range intervalley plasmons12 or the consequence of the
interaction between dark and bright trions13,14.
Here, we elucidate the origin of this luminescence line and demonstrate that its relatively intense emission
is the result of a second-order transition that involves dark trions. A resonant effect associated with the
electron-electron interaction yields radiative recombination along with spin-conserving intervalley
transition of the left behind electron to the upper conduction band (the electron that originally resides in the
same valley of the hole). As a result of energy conservation and translation symmetry, the emitted photon
is shifted to lower energy compared to the dark trion by a value equal to the conduction band (CB) spin-
orbit splitting Dc. The interpretation is further strengthened by the fact that we observe and interpret the
same type of second-order recombination channel for the positively charged trion when the monolayer is
electrostatically doped with holes. It results in a near infrared emission line, as the emission is now down-
shifted by the valence band (VB) spin-orbit splitting Dv, which is much larger than the CB one. These
investigations yield accurate determinations of both CB and VB spin-orbit splitting energies. We find Dc=
12.0 ± 0.5 meV and Dv=478 ±1 meV. Previous experimental determinations of the valence band spin-orbit
(SO) splitting were either based on Angle Resolved Photoemission Spectroscopy (ARPES) measurements
with a spectral resolution of few tens of meV or reflectivity/transmission experiments which include
contributions of both Dc and Dv as well as the binding energy of the exciton15,16 .
Results
We investigated high-quality charge tunable WSe2 devices as sketched in Fig. 1a. Details on sample
fabrication can be found in the Methods section17,18. A voltage bias applied between the monolayer and the
back gate allows us to electrostatically dope the monolayer. The doping densities in either the electron- or
hole-doped regimes are such that a change of 1 Volt in voltage bias corresponds to a typical change of
0.8x1011 cm−2 in charge density19. We deliberately remain in the low-density regime to avoid collective
phenomena, renormalization effects, and the emergence of hexcitons20 . In this regime, the three-particle
trion picture is an adequate description21,22. Continuous-wave polarization-dependent micro-
photoluminescence (PL) experiments are performed in a closed cycle cryostat (T = 4 K), as explained in
the Methods section.
Brightened emission of the negatively charged dark trion. Figure 1b displays the PL color plot as a
function of bias voltage and photon energy. In the charge-neutral regime (V ~ - 0.5 volts), indicated by the
horizontal dashed line, the recombination of bright (X0) and dark (D0) excitons dominate the PL
spectrum23,24. When electrons are added to the monolayer (positive voltages), the PL intensity of the neutral
exciton vanishes and we observe the well-known recombination of intravalley (singlet) 𝑋!" and intervalley
(triplet) 𝑋#" negatively charged trions, composed of two electrons and one hole 7,25 . As shown in Fig. 2a,
the triplet trion is composed of a photogenerated electron-hole pair, made of an electron in the top CB and
a missing electron in the topmost VB of the same K+ valley, and the pair is bound to a resident electron in
the bottom CB of the time-reversed valley at K-. For the singlet trion 𝑋!" , the three particles reside in the
same valley, as shown in Fig. 2b. The emission of the dark trion (D−) is also detected at lower energy. As
sketched in Fig. 2c, D− is composed of two electrons in the lowest conduction bands at K+ and K- , and a
hole at the top of the valence band. Its radiative recombination for in-plane polarized light is forbidden.
However it is allowed for z-polarized light10,11 ; the z-axis is perpendicular to the monolayer plane. The
clear observation of D− in the PL spectrum, as shown in Fig.1, is the consequence of using high numerical
aperture objective (NA=0.82) which allows us to detect the z-polarized luminescence component23.
Importantly, we observe a bright line, labelled as 𝐷" , lying 12 meV below D−. This intense line was
$
observed by many groups, but its origin remained unclear 4,8,10,11. It was shown that this line is co-polarized
with the laser, following circularly-polarized laser excitation and that its g-factor is about -4.5, i.e. close to
the g-factor of bright trions or excitons24. In a work devoted to the general effects of exciton-electron and
trion-electron interactions, it was first proposed that this transition could result from the brightened emission
of the dark trion D− 26. The corresponding second-order recombination mechanism is summarized in
Fig.2d. The light is generated from recombination of the indirect exciton component composed of a
conduction-band electron in K- and a hole in K+. To respect translation symmetry, the second electron of
the trion ends-up in the opposite top valley (i.e., Coulomb induced intervalley spin-conserving transition).
As a result of energy conservation, the light emission is shifted to lower energy compared to the
D− transition by an amount equal to the conduction band spin-obit splitting Dc. The measured energy
difference between D− and 𝐷" in Fig. 1(a) yields a direct measurement of the conduction band spin orbit
$
splitting; we find Dc ~ 12 meV, in agreement with recent determinations19,27. As the 𝐷$" transition involves
electronic transitions to the upper conduction band states, this explains that the associated g-factor is that
of the bright trion (and not of the dark excitonic species)24 . The schematics in Fig. 2d also illustrate why
the emission of 𝐷$" is co-polarized with the laser following circularly polarized excitation, as observed
experimentally14. We emphasize that contrary to the direct recombination process of bright or dark trions,
the non-recombining electron in the dark trion is not merely a spectator during the 𝐷$" emission. Though
this 𝐷$" emission results from a second-order process, its intensity is rather significative thanks to the
resonance effect involving states in two nearby conduction bands. The interpretation of the 𝐷$" transition is
strongly reinforced by the observation of an equivalent emission, linked to the same type of Coulomb
induced second-order mechanism of positively charged trions.
Brightened emission of the positively charged dark trion. Figure 1b shows that in the hole-doped regime
(negative voltages), the spectrum is dominated by the recombination of the bright positive trion 𝑋 % at 1.688
eV and the dark one 𝐷% at 1.654 eV, as already observed by many groups8,9,28. Remarkably, when holes are
injected in the monolayer, a new line shows up in the near-infrared region of the spectrum - beyond 1 µm
wavelength. Figure 3a displays the PL spectrum in this region, which clearly evidences a new line, labelled
𝐷$% at 1.175 eV; this line is absent in the neutral regime and its intensity rises with the voltage increase.
Figure 3b shows that its intensity is commensurate with that of the dark trion 𝐷% . This peak was not detected
before, probably because its emission lies at much lower energies compared to those associated with other
exciton species. The inset of Fig. 3b shows that the intensity of 𝐷$% rises with the excitation laser power,
showing no sign of saturation, ruling out a possible interpretation based on impurity recombination.
The energy difference between the resonances 𝐷$% and 𝐷% is 478 meV (Fig. 3c). Notably, it is very close
to the valence band spin-orbit splitting which has been calculated29 and measured in ARPES ( 485±10
meV) 15 . This is a strong argument to interpret the emission of 𝐷$% as the brightened emission of the dark
trion 𝐷% , following a second-order recombination mechanism like the one described above for the
negatively charged dark trion 𝐷" . The process is sketched in Fig. 3d, and it starts with a dark trion
𝐷% comprising two holes in the K+ and K- top valleys of the valence band and one electron at the bottom
of the conduction band. Owing to the hole-hole interaction, the 𝐷$% near-infrared recombination is facilitated
by radiative recombination of the momentum-indirect electron-hole component of the trion. Energy and
wavevector conservation rules are fulfilled thanks to the fact that the second hole ends up in the lower-
energy valence band. As a consequence, the energy difference between the 𝐷% and 𝐷$% lines can be used to
extract the valence band spin-orbit splitting Dv .
The interpretation of the 𝐷$% line is further confirmed by the measurement of the polarized luminescence
following circularly polarized excitation.
The circularly polarized luminescence of 𝐷$% is measured following circularly polarized laser excitation
with helicity σ+. Figure 4a shows that the PL intensity counter-polarized with respect to the laser has a
stronger intensity than the co-polarized PL component (𝐼 '% < 𝐼 '" ). The resulted PL polarization degree is
) !" ") !#
𝑃( = )!" %)!# ≈ −0.25. This negative PL polarization can be well explained by the scenario summarized in
Fig. 4b. The σ+ laser excitation yields the initial photogeneration of a bright positively charge trion X+
composed of an electron in the K+ valley of the upper conduction band and two holes in K+ and K- valley
respectively. The energy relaxation of the electron to the lowest CB leads to the formation of the dark trion
𝐷% . This can occur through a spin conserving process implying inter-valley K3 phonon or an intra-valley
relaxation based on G5 zone-center phonon24,30. As was shown in refs.18,24 , the inter-valley process is
dominant, meaning that the largest population of dark trions D+ following σ+ laser excitation has electrons
lying in the K- valley. Following hole-hole interaction, the second-order transition will yield a radiative
recombination in the K- valley, as shown in Fig. 4d. Consequently, a negative PL circular polarization of
the 𝐷$% line is expected, as we observe experimentally.
Discussion
The spectral positions of the brightened emission from dark trions and their polarization support the
physical picture of charge-charge assisted recombination. One way to reassure this hypothesis is to compare
the amplitudes of the emission lines 𝐷$" and 𝐷$% . By carefully taking into account the different detection
efficiencies of both lines (spectrometer and detector response functions), the ratio between their integrated
luminescence intensities is 𝐼*$# /𝐼*$" ≈ 70 when the excitation laser power is 50 µW, and the electrons or
holes densities are 3.6𝑥1011 cm-2 corresponding to gate voltages of +4 V and -5 V respectively (the neutral
regime corresponds to V ~ -0.5 volts). This ratio can be compared with the calculated ratio between the
brightened-emission recombination rates of negative (R-) and positive (R+) dark trions. Using second-
order perturbation, the recombination rates ratio is
+# , - . /
+"
= , %,' 3-( .* 3 , (1)
(,) % +
where m(,1 /m2,3 is the ratio between the effective masses of electrons in the top conduction band and holes
in the bottom valence band, accounting for the density of states in the valleys of the left-behind charge
particles. 𝑉4(6) is the matrix element denoting the electron-electron (hole-hole) intervalley Coulomb
interaction. Note that the ratio excludes the dipole-transition matrix element because the emission process
involves recombination of similar electron and hole states in either positive or negative dark trions.
Employing the calculated first-principal values of effective masses m(,1 = 0.29 and m2,3 = 0.54 29 , and
the measured SO energy splittings, Dv=478 meV and Dc=12 meV, the calculated ratio 𝑅" /𝑅% ~
850.(𝑉4 /𝑉6 )2. We can think of two possible ways to reconcile this result with the measured value 𝐼*$# /𝐼*$" ≈
70. The first one is that central-cell corrections of the shortwave Coulomb interaction render the hole-hole
matrix element stronger, such that 𝑉6 ~ 3.5𝑉4 . While detailed calculations of central-cell corrections are
beyond the scope of this paper, we notice that this behavior is reminiscent of the ratio between the zone-
edge phonon replica emissions of dark trions24,26. In more detail, similar to the intensity ratio between 𝐼*$#
and 𝐼*$" , the emission amplitude of the K1 phonon replica is also not three orders of magnitude smaller than
that of the K3 phonon replica, indicative of a stronger intervalley transition in the valence band (assisted by
the K1 mode) compared with the conduction band (assisted by the K3 mode). The second possible reason
to reconcile the measured and calculated ratios, 𝐼*$# /𝐼*$" versus 𝑅" /𝑅% , is that the bottom valence-band
valleys at ±K are nearly degenerate with the valence band valley at G, as suggested by recent ARPES
experiments15,16,31. The intervalley transition from the top valley of K to the valley at G respects the same
translation symmetry conditions, implying that their contributions should be added if these valleys are
indeed energy degenerate. The experimental result can then be explained by the much larger effective mass
(and ensuing density of states) of the G valley.
Another difference between the emission lines 𝐷$" and 𝐷$% is that the linewidth of 𝐷$% is about ten times
larger than that of 𝐷$" (Fig. 3c). The measured full width at half maximum of 𝐷$% is ~24 meV, showing no
dependence on the holes density in the studied range between 0 and ~ 5. 1088 𝑐𝑚"/ . We attribute the large
broadening to the ultrafast lifetime of the free hole in the bottom valence band (Fig. 3d). The ultrafast
lifetime could stem from efficient intervalley hole scattering between the nearly degenarate valence band
valley at G and the valleys at ±K 15,16,31. Such efficient scattering has no parallel in the conduction band.
Similarly, spin-conserving intervalley transition mediated by K1 phonon can lead to ultrafast transition of
the hole from the bottom valley to the top opposite valley, whereas equivalent transition with K3 mode in
the conduction band is hampered by phonon bottleneck (Dc > EK3). All in all, the strikingly different
linewidths of 𝐷$" and 𝐷$% can be reasoned by a relatively long lifetime of a free electron in the top conduction
band compared with that of a free hole in the bottom valence band.
Finally, the accurate determination of SO splitting energy in both conduction and valence bands provides
valuable information on the difference between binding energies of A and B neutral excitons. The A exciton
has been investigated in great detail both experimentally and theoretically in WSe2 monolayer2,32 . However,
less is known about the B exciton which comprises an electron in the bottom conduction band and a hole
in the bottom valence band (whereas the A exciton comprises an electron in the top conduction band and
hole in the top valence band). The photon energy that resonates with the A(B) exciton can be written as
9($)
𝐸 9($) = 𝐸: + ∆((2) − 𝐸3 , where 𝐸: is the free carrier gap involving the bands edges (bottom
9($)
conduction band and top valence band), and 𝐸3 is the binding energy of A(B) exciton. Thus, the energy
difference between B and A excitons, measured in reflectivity or transmission experiments is
𝐸 $ − 𝐸 9 = ∆2 − ∆( − A𝐸3$ − 𝐸39 B. (2)
Though measurements of 𝐸 $ − 𝐸 9 have been commonly used to roughly estimate the valence band SO
splitting ∆2 2,6,33, Eq. (2) highlights that it also depends on the small SO splitting in the conduction band.
For hBN encapsulated WSe2 monolayer, differential reflectivity experiments yield 𝐸 $ − 𝐸 9 = 430 𝑚𝑒𝑉 34.
This energy difference is clearly smaller than the valence band SO splitting ∆2 ~ 478 meV, which can be
determined independently from the measured energy difference between the resonances 𝐷% and 𝐷$% in the
PL [Link] ∆2 and ∆( measured here, we find that 𝐸3$ − 𝐸39 ≈ 36 meV. The larger binding energy
of the B exciton can be attributed to the larger effective masses of its electron and hole components35 .
Indeed, if we use k.p theory to evaluate the effective masses in the corresponding valleys, the masses are
typically smaller than that of the electron mass in vaccum (m0) by a factor that is commensurate with p2/Eg,
where p and Eg are the in-plane dipole transition and energy gap between the corresponding conduction and
valence states. The effective masses of the electron and hole of the A exciton are expected to be smaller
because of the smaller energy gap in this case. Using the calculated effective masses from density-
functional theory, the masses of the electron and hole components of the A exciton are 0.29m0 and 0.36m0,
respectively, and those of the B exciton are 0.4m0 and 0.54m0 29. The resulting reduced masses of the A and
B excitons are 0.16m0 and 0.23m0, thereby supporting the larger binding energy of the B exciton.
Acknowledgements : This work was supported by the Agence Nationale de la Recherche under the
program ESR/EquipEx+ (Grant No. ANR-21-ESRE- 0025) and ANR projects ATOEMS, IXTASE and
PEPR SPIN. Work at the University of Rochester was supported by the DOE Basic Energy Sciences,
Division of Materials Sciences and Engineering, under Award No. DE-SC0014349.
Methods
Sample fabrication. We have fabricated a van der Waals heterostructure made of an exfoliated ML-WSe2
embedded in high quality hBN crystals using a dry stamping technique in the inert atmosphere of a glove
box36,37. The layers are deterministically transferred on top of a SiO2/Si substrate with Ti/Au electrodes
patterned by photolithography. Flux-grown WSe2 bulk crystals are purchased from 2D semiconductors. We
use few layers of graphene exfoliated from a Highly Ordered Pyrolytic Graphite (HOPG) bulk crystal for
the back gate and to contact the ML-WSe2.
Experimental set-up. Low temperature photoluminescence measurements were performed in a home built
micro-spectroscopy set-up built around a closed-cycle, low vibration cryostat (T=4 K). The sample is
excited by a He-Ne laser (632.8 nm). The PL signal is dispersed by a monochromator and detected by a
silicon based charged-coupled device camera38. Unless otherwise stated, the excitation power is typically
∼50 μW, focused to a spot size of ∼1 μm diameter.
References
1. Xu, X., Yao, W., Xiao, D. & Heinz, T. F. Spin and pseudospins in layered transition
metal dichalcogenides. Nature Phys 10, 343–350 (2014).
2. Wang, G. et al. Colloquium: Excitons in atomically thin transition metal dichalcogenides.
Rev. Mod. Phys. 90, 021001 (2018).
3. Mak, K. F. & Shan, J. Semiconductor moiré materials. Nat. Nanotechnol. 17, 686–695
(2022).
4. Jones, A. M. et al. Optical generation of excitonic valley coherence in monolayer WSe2.
Nature Nanotech 8, 634–638 (2013).
5. Mak, K. F. et al. Tightly bound trions in monolayer MoS2. Nature Mater 12, 207–211
(2013).
6. Wang, G. et al. Valley dynamics probed through charged and neutral exciton emission in
monolayer WSe 2. Phys. Rev. B 90, 075413 (2014).
7. Courtade, E. et al. Charged excitons in monolayer ${\mathrm{WSe}}_{2}$: Experiment
and theory. Phys. Rev. B 96, 085302 (2017).
8. Ye, Z. et al. Efficient generation of neutral and charged biexcitons in encapsulated WSe2
monolayers. Nat Commun 9, 3718 (2018).
9. Barbone, M. et al. Charge-tuneable biexciton complexes in monolayer WSe2. Nat
Commun 9, 3721 (2018).
10. Liu, E. et al. Gate Tunable Dark Trions in Monolayer WSe 2. Phys. Rev. Lett. 123,
027401 (2019).
11. Li, Z. et al. Direct Observation of Gate-Tunable Dark Trions in Monolayer WSe 2. Nano
Lett. 19, 6886–6893 (2019).
12. Van Tuan, D., Scharf, B., Žutić, I. & Dery, H. Marrying Excitons and Plasmons in
Monolayer Transition-Metal Dichalcogenides. Phys. Rev. X 7, 041040 (2017).
13. Danovich, M., Zólyomi, V. & Fal’ko, V. I. Dark trions and biexcitons in WS2 and WSe2
made bright by e-e scattering. Sci Rep 7, 45998 (2017).
14. Tu, J.-S., Borghardt, S., Grützmacher, D. & Kardynał, B. E. Experimental observation of
a negative grey trion in an electron-rich WSe 2 monolayer. J. Phys.: Condens. Matter 31, 415701
(2019).
15. Nguyen, P. V. et al. Visualizing electrostatic gating effects in two-dimensional
heterostructures. Nature 572, 220–223 (2019).
16. Zribi, J. et al. Strong interlayer hybridization in the aligned SnS2/WSe2 hetero-bilayer
structure. npj 2D Mater Appl 3, 27 (2019).
17. Robert, C. et al. Measurement of Conduction and Valence Bands g -Factors in a
Transition Metal Dichalcogenide Monolayer. Phys. Rev. Lett. 126, 067403 (2021).
18. Robert, C. et al. Spin/valley pumping of resident electrons in WSe2 and WS2
monolayers. Nat Commun 12, 5455 (2021).
19. Ren, L. et al. Measurement of the conduction band spin-orbit splitting in WSe 2 and WS
2 monolayers. Phys. Rev. B 107, 245407 (2023).
20. Van Tuan, D., Shi, S.-F., Xu, X., Crooker, S. A. & Dery, H. Hexcitons and oxcitons in
monolayer WSe$_2$. Phys. Rev. Lett. 129, 076801 (2022).
21. Sidler, M. et al. Fermi polaron-polaritons in charge-tunable atomically thin
semiconductors. Nature Phys 13, 255–261 (2017).
22. Glazov, M. M. Optical properties of charged excitons in two-dimensional
semiconductors. J. Chem. Phys. 153, 034703 (2020).
23. Wang, G. et al. In-Plane Propagation of Light in Transition Metal Dichalcogenide
Monolayers: Optical Selection Rules. Phys. Rev. Lett. 119, 047401 (2017).
24. He, M. et al. Valley phonons and exciton complexes in a monolayer semiconductor. Nat
Commun 11, 618 (2020).
25. Jones, A. M. et al. Excitonic luminescence upconversion in a two-dimensional
semiconductor. Nature Phys 12, 323–327 (2016).
26. Yang, M. et al. Relaxation and darkening of excitonic complexes in electrostatically
doped monolayer WSe 2 : Roles of exciton-electron and trion-electron interactions. Phys. Rev. B
105, 085302 (2022).
27. Kapuściński, P. et al. Rydberg series of dark excitons and the conduction band spin-orbit
splitting in monolayer WSe2. Commun Phys 4, 186 (2021).
28. Wang, Z., Shan, J. & Mak, K. F. Valley- and spin-polarized Landau levels in monolayer
WSe2. Nature Nanotech 12, 144–149 (2017).
29. Kormányos, A. et al. k · p theory for two-dimensional transition metal dichalcogenide
semiconductors. 2D Mater. 2, 022001 (2015).
30. Song, Y. & Dery, H. Transport Theory of Monolayer Transition-Metal Dichalcogenides
through Symmetry. Phys. Rev. Lett. 111, 026601 (2013).
31. Yeh, P.-C. et al. Layer-dependent electronic structure of an atomically heavy two-
dimensional dichalcogenide. Phys. Rev. B 91, 041407 (2015).
32. Zipfel, J. et al. Spatial extent of the excited exciton states in WS 2 monolayers from
diamagnetic shifts. Phys. Rev. B 98, 075438 (2018).
33. Koperski, M. et al. Optical properties of atomically thin transition metal dichalcogenides:
observations and puzzles. Nanophotonics 6, 1289–1308 (2017).
34. Manca, M. et al. Enabling valley selective exciton scattering in monolayer WSe2 through
upconversion. Nat Commun 8, 14927 (2017).
35. Stier, A. V., McCreary, K. M., Jonker, B. T., Kono, J. & Crooker, S. A. Exciton
diamagnetic shifts and valley Zeeman effects in monolayer WS2 and MoS2 to 65 Tesla. Nat
Commun 7, 10643 (2016).
36. Castellanos-Gomez, A. et al. Deterministic transfer of two-dimensional materials by all-
dry viscoelastic stamping. 2D Mater. 1, 011002 (2014).
37. Cadiz, F. et al. Excitonic Linewidth Approaching the Homogeneous Limit in MoS 2 -
Based van der Waals Heterostructures. Phys. Rev. X 7, 021026 (2017).
38. Shree, S., Paradisanos, I., Marie, X., Robert, C. & Urbaszek, B. Guide to optical
spectroscopy of layered semiconductors. Nat Rev Phys 3, 39–54 (2020).
Figures
Fig.1 Photoluminescence as a function of the gate voltage. a Sketch of the charge adjustable
WSe2 monolayer device. b Photoluminescence intensity as a function of the gate voltage
displaying the recombination of the different exciton complexes; the excitation energy is 1.96 eV.
The brightened negatively charged exciton resulting from a second-order recombination process
is labelled 𝐷!" .
Fig. 2 Negatively charged exciton recombination channels. a and b Schematic of the bright
triplet (𝑋#" ) and singlet (𝑋$" ) trion emission. c sketch of the dark (𝐷" ) trion radiative
recombination which is z-polarized. d Illustration of the second order transition starting from the
dark trion (𝐷" ) and yielding the radiative emission of the brightened dark trion (𝐷!" ) line : the
light is generated from recombination of the indirect exciton component composed of the
conduction band electron in K- and and hole in K+ valley. As a result of energy and wavevector
conservation, the second electron of the trion ends-up in the opposite top valley (Coulomb induced
inter-valley spin-conserving transition) and the process yields a light emission shifted to lower
energy compared to the 𝐷" transition by an amount equal to the conduction band spin-obit splitting
Dc. In contrast to the direct optical recombination of bright or dark trion (as in a, b or c), the second
electron here is no more a spectator.
Fig. 3 Brightened emission of the positively charged dark trion. a PL intensity of the 𝐷!% line
as a function of the gate voltage; note the energy range which corresponds to emission wavelengths
beyond 1 µm. b Normalized PL intensity of the dark 𝐷% and brightened dark 𝐷!% trions as a
function of the gate voltage ; the inset displays the excitation power dependence of the 𝐷!% line
for a voltage V=-5 volts. c Photoluminescence spectrum displayed on a large energy range showing
that the energy difference between 𝐷% and 𝐷!% corresponds to the valence band spin-orbit splitting
energy: Dv= 478 meV. d Schematic of the second order transition corresponding to the radiative
emission of the positively charged brightened dark trion 𝐷!% . The dark trion 𝐷% is composed of 2
holes lying in K+ and K- valleys at the top of the valence band (A) and one electron lying at the
bottom of the conduction band. The 𝐷!% near-infrared recombination results from the hole-hole
interaction which allows the radiative recombination of the momentum indirect electron-hole
component of the trion. The second hole ends up in the lowest energy valence band (B). The energy
difference between the 𝐷% and 𝐷!% lines is simply the valence band spin-orbit splitting Dv as a
consequence of energy and wavevector conservation in the recombination process.
Fig. 4 Circularly-polarized emission of the positively charged brightened dark trion. a Right
(s+) and left (s-) circularly-polarized luminescence of the 𝐷!% line following s+ polarized laser
excitation; the corresponding circular polarization Pc is also displayed (note its negative value). b
Simplified scheme yielding a negative PL circular polarization of the 𝐷!% line as a consequence of
the second-order transition (see text).
