realization of the charged-QD lasing scheme.
An effective approach for reducing Auger decay
rates is through smoothing the QD confinement
potential, which leads to suppression of the
intraband transition involved in the dissipation of
the electron–hole recombination energy ( 11 ). To
realize this effect practically, we enclose a small
emitting CdSe core within a thick CdxZn1-xSe
shell whereinxprogressively decreases in the
radial direction from 1 to 0 ( 8 , 9 )[fig.S1and
materials and methods ( 12 )]. These continuously
graded QDs (cg-QDs) are completed with a final
protective ZnSe0.5S0.5/ZnS layer for enhancing
dot stability, improving surface passivation, and
impeding defect-mediated QD discharging ( 13 ).
The photoluminescence (PL) wavelength of the
studied samples is 620 nm (inset of Fig. 2A) and
the PL quantum yield (QY) is ~55% in the solution
form. The single-exciton lifetime (tX)is~15ns,as
inferred from PL dynamics measured using weak
excitation (
tary text S2). The biexciton lifetime (tXX) ob-
tained from the fast PL component emerging
at high excitation powers is 1.5 ns (fig. S2, A and
B). The PL measurements of photochemically
doped samples (see below) indicate that the
lifetime of an exciton with one extra electron
(negative trion) is 4.5 ns (fig. S2A). On the basis
of this value and thene(ne+1)scalingofAuger
time constants ( 14 ), the lifetime of a doubly
negatively charged exciton is 2.5 ns. The PL QYs
of biexcitons and charged excitons withne=
1and2are40,60,and33%,respectively(sup-
plementary text S3). These values are considera-
bly higher than the corresponding QYs (2.5, 10,
and 3.5%, respectively) of standard CdSe QDs
with a similar band gap (supplementary text S3),
indicating strong suppression of Auger recom-
bination in the cg-QD samples. The suppressed
Auger decay also leads to extended optical-gain
lifetimes (fig. S3 and supplementary text S4).
Although this extension is observed for both
neutral and charged cg-QDs, the gain dynamics
inthesetwocasesarequalitativelydifferent.In
particular, in the case of charged dots, optical
gain persists in the long-time limit (Dt→∞),
whereas in neutral dots, it exists only during
a finite time window (Dt<1.2ns).Thisisan
important advantage of the charged-QD gain
approach, which can, in principle, simplify real-
ization of continuous-wave lasing.
To inject a controlled number of electrons into
cg-QDs, we use a modified version of a photo-
doping method with lithium triethylborohydride
(LiEt 3 BH) as a reducing agent ( 9 , 14 , 15 )(Fig.2B).
We apply this procedure to films of cg-QDs
cross-linked with bifunctional 1,8-diaminooctane
molecules for improving sample stability (fig. S4
and materials and methods S1). The PL QY of
filmsamplesis15to20%.Itislowerthan
that of the QD solution because of distortion
of the original surface passivation and dot-to-dot
energy transfer, which results in resampling de-
fect sites at each transfer step. In photodoping
experiments, a cross-linked cg-QD sample is im-
mersed into a solution of LiEt 3 BH in tetrahy-
drofuran (THF). QD charging is initiated by
Kozlovet al.,Science 365 , 672–675 (2019) 16 August 2019 2of4
Fig. 2. Photochemical doping of QDs.(A) PL dynamics of a cross-linked film of cg-QDs immersed
into a 0.2 M LiEt 3 BH/THF solution as a function of time of exposure to 400-nm light of a 100-kHz
laser with a per-pulse fluence (jp)of0.3mJcm−^2 (the measurement time per trace is 30 s). Inset: the cg-QD
absorption (black) and PL (red) spectra. (B) QD charging with an extra electron occurs by scavenging
a hole from a photoexcited dot by LiEt 3 BH acting as a sacrificial hole acceptor. Subsequent
photoexcitation of a charged dot leads to formation of a negative trion. (C) Average number of extra
electrons per QD (<ne>) as a function of photoexposure time [same conditions as in (A)]. Solid
symbols correspond to the traces shown in (A) using the same symbol styles. Inset: Normalization
procedure used to determine <ne>. The red and blue solid traces are“raw”PL dynamics for neutral
and charged QDs, respectively [same as color-matched traces in (A)]. The red dashed trace
shows the dynamics of neutral dots normalized to match the long-time tail of PL from the charged
dots. The normalization factorq 0 yields a relative fraction of neutral dots in the charged sample.Fig. 3. Lasing studies of a device made of cg-QDs coupled to a second-order DFB grating.
(A) Schematics of a cg-QD DFB laser.This device is immersed into a LiEt 3 BH solution in THF, which leads to in
situ QD photodoping (inset) induced by pump pulses used to excite lasing. (B) Scanning electron microscopy
image of a cg-QD layer (thicknessH) on top of a DFB grating (periodd, grove heighth,andgrovewidthw).
(C) Example of neutral-QD lasing using a cg-QD/DFB device immersed into neat THF without the reducing
agent (400-nm excitation with a femtosecond 1-kHz laser). The spectra of device surface emission for
progressively increasing excitation fluences indicate a sharp transition to single-mode lasing atjpof ~9mJcm−^2.RESEARCH | REPORT