Colloidally synthesized nanocrystals
(NCs) are presently employed as artificial atoms for predictable design
of solid-state materials for a plethora of applications.
1,2
Tuning the electronic characteristics, foremost type, concentration
and mobility of charge carriers remains a formidable challenge for
bottom-up engineered nanostructures. As-synthesized NCs are usually
capped with long-chain organic ligands covalently attached to surface
NC atoms.
3
These ligands hamper electronic
transport in NC-based materials, making the removal of these electrically
insulating shells absolutely necessary. Organic ligands are often
replaced by strongly or weakly coordinating, shorter, organic or inorganic
ligands such as pyridine,
4
hydrazine,
5
ammonia,
6
thiols,
7
chalcogenidometalates,
8,9
thiocyanate,
10,11
tetrafluoroborate salts,
12
oxometalates,
13
halides,
14
halometalates
15
or chalcogenides,
16−18
metal ions,
19,20
etc. The capping of NCs with such ligands enhances electronic coupling
between adjacent NCs and allows for the modulation of nearly all practically
relevant electronic parameters.
21,22
Similar to bulk
semiconductors, intrinsic stoichiometry and extrinsic
impurities can be expected as primary players for controlling n-, p- or intrinsic
charge transport (Scheme 1
). The combined effect
of the NC core composition and of the chosen capping ligand can be
rationalized considering charge-orbital balance.
25
In the case of NCs, charge and bond counting must also
include covalently attached ligands and/or surface charges. In particular,
colloidal NCs of lead chalcogenides (PbX, X = S, Se, Te) contain a
fully stoichiometric core covered with an excess of Pb cations, acting
as adatoms for coordinating with X-type capping ligands such as carboxylates.
26
The fate of this additional quantity of Pb must
be considered for controlling and understanding the electronic properties
of the final solid material, as discussed later for various surface
treatments. Extrinsic dopants for substitutional doping can be introduced
via surface functionalization. Surface passivation with halide ions
(Cl–, Br– and I–) has been shown to result in NC solids with n-type
conductivity with adjustable charge carrier mobility and concentration.
23,27
This n-type doping effect from halide ions can
be rationalized based on the charge neutrality requirement: substitution
of one double-charged chalcogenide anion with a single-charged halide
ion and an electron, as illustrated in Scheme 1
.
Scheme 1
Atomic Depiction of Substitutional Electronic
Doping for Rendering
PbS n-Type
23
or p-Type (this work) with High Carrier Density
The similarity in corresponding
Shannon ionic radii
24
of cations (133 pm
for Pb2+ and 116 pm for Na+) and anions (184
pm for S2– and 181 pm for Cl–)
favors high doping levels.
Analogously, an
efficient p-type doping strategy
is to replace a double-charged Pb or Sn ion with a single-charged
cation, such as potassium or sodium, and a hole (Scheme 1
). For accomplishing this with
colloidal PbS NCs as starting building blocks, a two-step strategy
is detailed. First, an alkali metal containing inorganic capping ligand
(K2S, K2Te and Na2S) is attached
to the surface via a ligand-exchange reaction. Second, substitutional
doping is induced by thermal annealing. We then thoroughly characterize
the charge transport by electrical conductivity (σ), Hall-effect
and thermopower (Seebeck coefficient, S) measurements.
A variety of control experiments with other ligands, for differentiating
the effects of chalcogen and alkali metals, is presented. Experimental
results show that tunable p-type conductivity can
be accomplished with various NC–ligand combinations, either
chalcogen-matched (i.e., PbS-K2S) or mismatched (i.e.,
PbS-K2Te). Furthermore, fine-tuning of hole concentration
has been demonstrated with a mixture of ligands, wherein one contains
alkali metal (e.g., A2X) and the other contains only the
chalcogen (X dissolved in a dithiol/diamine mixture; denoted as X-complexes).
In the following, the details of the surface functionalization
and resulting electronic properties are presented for ∼11 nm
cubic PbS NCs for seven ligands and their mixtures, to illustrate
the rational chemical engineering of p-type conductivity
in nanostructured Pb chalcogenides (Figure 1
). All samples differ only in their surface
treatment, whereas the temperatures and procedures of thermal consolidation
are maintained very similarly (400–450 °C, see the Supporting Information (SI) for
further details
on all synthesis procedures and characterizations, including Figures S1–S16 and Tables
S1−S3).
Figure 1
(a) Summary of the used ligands for the surface
functionalization/treatment
of 11 nm PbS NCs and resulting transport properties at RT. (b) Temperature-dependent
electrical conductivities, σ. (c) Temperature dependence of
the Seebeck coefficients, S.
To understand the intrinsic nature of the PbS NC and its
relationship
to the electronic properties of the corresponding nanomaterial, taking
into account the purely inorganic part and the organic shell surrounding
it, two reference samples were prepared. In the first reference, both
inorganic core and organic shell were treated as a unit (OA-PbS).
In a typical experiment, several grams of PbS NCs were prepared according
to reported methods by reacting lead oleate with an oleylamine-based
sulfur precursor.
23
Purified NCs were capped
exclusively with long-chain oleate ligands, as confirmed by NMR measurements.
Prior to the consolidation, such organic ligands were thermally decomposed
by annealing the as-synthesized NCs at 450 °C under inert gas.
The obtained powder was consolidated by hot-pressing into ∼1
mm thick disk-shaped pellets, 10 mm in diameter (40 MPa, 420–440
°C, 4 min). Pellets obtained from oleate-capped PbS NCs (OA-PbS)
exhibit low densities (∼80%) attributed to decomposition and
removal of the capping ligands.
28
Consequently,
impurities of PbO and carbon, both quantities scaling with NC size
(surface-to-volume ratio), are typically observed in such samples.
These impurities accumulate at the grain boundaries.
29
For 11 nm PbS NCs, the amount of Pb-adatoms that are converted
into PbO is estimated to be ca. 7.7 at. % of the stoichiometric core
PbS (Tables S1 and S2). This material exhibits
rather low room temperature (RT) electrical conductivities (0.07 S
cm–1, Figure 1
b). A positive sign on the Seebeck coefficient (Figure 1
c) indicates a p-type conductivity. Measurements of the Hall hole concentration at
RT showed a relatively low doping level (p = 1016 cm–3). As the temperature increased,
thermally
activated electrons increasingly contributed to the conductivity,
as commonly observed for Pb chalcogenide solids,
6,23,30
eventually inverting the sign of the Seebeck
coefficient to negative at ca. 450 K.
On the other hand, to
evaluate the sole influence of the inorganic
part of NCs, we have studied the case of a ligand-free surface design.
Native organic ligands were removed using a 4 M solution of hydrazine
in anhydrous ethanol. Ethanol is known to desorb oleate from the NC
surface through the nucleophilic addition of ethoxide to the carboxylate
group, which activates the protonation of the oleate and consequently
its desorption.
31
In the presence of hydrazine,
the concentration of ethoxide nucleophiles increases promoting oleate
removal and generating desorbed oleic acid and oleic hydrazide.
31
Additionally, hydrazine was selected due to
its strong reducing character, which allows for producing nonoxidized,
fully inorganic compositions, without introducing new chemical elements
into consideration.
32,33
Solids produced from N2H4-treated PbS NCs exhibited high electrical conductivities
and a negative Seebeck coefficient in the whole temperature range.
Hall electron concentrations, measured at RT, were on the order of
2 × 1019 cm–3, much higher than
that of the OA-PbS-derived nanomaterial. This can be ascribed to the
conversion of divalent Pb-adatoms to Pb0, acting as n-type dopants.
34,35
Surface modification
with K2S, K2Te and Na2S was carried
out via a phase-transfer reaction, in which
the PbS NCs migrated from the nonpolar phase (hexane) to the polar
phase (N-methylformamide; MFA) upon displacement
of oleate ligands with the alkali metal chalcogenides (Figure 2
a, Figure S16). MFA was chosen due to its high dielectric constant, which
facilitates the electrostatic dissociation of the alkali counterions
and the adsorption of anionic ligands onto the NC surface. Consequently,
the steric stabilization of NC colloids is switched to electrostatic
stabilization (Figure S4). A byproduct
of alkali metal oleate was fully removed by several cycles of precipitation
and redispersion of NCs in MFA, using acetone as a nonsolvent to cause
precipitation. An analogous phase-transfer approach was utilized in
order to treat PbS NCs with molecular chalcogen complexes (X-complexes).
The latter were prepared by dissolving ∼4 wt % of elemental
chalcogens in ethanedithiol–ethylenediamine (1:4) mixtures.
36
In the subsequent ligand-exchange process, the
polar phase was composed of an MFA:ethylenediamine (1:1) mixture,
and acetone was replaced with acetonitrile as a nonsolvent. For a
controlled introduction of varying quantities of alkali metals onto
the NC surface, A2X ligands and X-complexes can be mixed
in desired ratios.
Figure 2
(a) Schematic of the ligand-exchange reaction at the surface
of
PbS NCs: oleate ions are replaced with Te2–; partial
S2–-to-Te2– anion exchange also
occurs. K-ions occupy Pb sites in the final solid. (b) HRTEM images
after the surface treatment with K2Te for 17 h, displaying
Moiré fringers and crystallographic maps showing the core–shell
PbS@PbTe structure. (c) Lower resolution TEM images after treatment
with K2Te (top) and Te-complexes (bottom) for 17 h.
The functionalization of initially
Pb-rich PbS NCs with S-complexes,
followed by thermal consolidation is assumed to yield PbS nanomaterial
with a stoichiometry closer to 1:1, similar to the reported effect
of the ammonium thiocyanate ligand.
23
The
oxidation state of the Pb-adatoms should rather follow the N2H4-scenario (n-type),
but the quantity
of the formed Pb0 should be much lower, if any. In accordance
with these considerations, this nanomaterial indeed shows a negative
Seebeck coefficient in the whole temperature range, with carrier concentrations
(n = 8 × 1017 cm–3) 25 times lower than in the case of N2H4-treatment,
but much higher than for OA-PbS. Similarly, bulk PbS had been shown
to acquire n-type conductivity due to a slight Pb
surplus, caused by the S loss upon prolonged annealing.
37
In striking contrast to a previous example,
solids derived from
A2S-treated (A = K, Na) PbS NCs (Figure 1
) exhibited clear p-type
behavior, apparent from the positive sign of the Seebeck coefficients
in the whole temperature range. The Hall hole concentrations in Na2S–PbS and K2S–PbS
nanomaterials,
measured at RT, were p = 9 × 1017 cm–3 and p = 2 × 1016 cm–3, respectively. This is
consistent
with the doping scenario depicted in Scheme 1
: Pb surplus at the surface is neutralized
by additional chalcogenide anions from the ligand, and some Pb ions
are substituted with shallow electron acceptors (A+) incorporated
into the cation sublattice. In other words, Pb2+ is substituted
by a localized K+ (or Na+) and a mobile hole
(h+). The highest achievable hole concentrations were ca.
9 × 1017 cm–3 (at RT). The difficulty
of reaching even higher hole concentrations in this bottom-up approach
is attributed to the deficiency of S, counteracting the p-doping
from alkali metals.
To overcome the doping limitation inherent
to S-based nanomaterial
we investigated Te-based ligands. The amount of chalcogen-based ligand
used in the ligand-exchange reaction is always much higher than that
required for binding each Pb-adatom at the NC surface. Hence, the
observation of core–shell PbS@PbTe NCs (Figure 2
) is unsurprising, caused by the anion-exchange
occurring when PbS NCs were subjected to Te-based ligands. High-resolution
transmission electron microscopy (HRTEM) micrographs of the produced
PbS@PbTe NCs revealed the presence of Moiré fringes characteristic
of the superposition of different crystal phases. The doublet points
marked by red and green in the power spectrum allowed for differentiation
between the core and the shell lattice. Both core and shell have identical
cubic rock-salt crystal structure (S.G.: Fm3m) and differ only in the lattice constant
(6.46 and 5.94
Å for PbTe and PbS, respectively). Low-magnification images revealed
that the formation of PbTe during ligand removal induced NC sintering
through the PbTe shell (Figure 2
c). Both the duration of the ligand exchange in solution (up
to 500 h) and thermal treatment (10 min, 210 °C) contribute to
the degree of the anion-exchange and crystallinity of the produced
PbTe shells. The resulting powder X-ray diffraction (XRD) pattern
indicates that the anion-exchange occurs rapidly in solution within
seconds (Figure 3
, S8). Longer reaction times lead to crystallization
of PbTe already at RT. PbTe content increases only slightly from 23%
after 0.5 h to 30% after 500 h (as determined from Rietveld refinement).
Figure 3
XRD patterns
of PbS NCs after the exchange of oleate with K2Te recorded
for different ligand-exchange reaction times before
(a) and after (b) annealing at 210 °C. Color code for atoms is
same as in Figure 2
.
Such Te-ion-exchanged PbS NCs,
using K2Te or Te-complexes,
were thermally consolidated by hot-pressing into solids of the approximate
composition K0.01Pb0.99S0.7Te0.3 (denoted as K2Te-PbS) and PbS0.7Te0.3 (i.e., Te-PbS).
K2Te-PbS nanocomposites exhibited
strong p-type behavior, with high electrical conductivities
over the whole studied temperature range, Hall hole concentrations
of approximately 3 × 1019 cm–3 at
RT, and a positive sign and value of 222 μV K–1 for the Seebeck coefficient (Figure
4
). In contrast, in the case of Te–PbS nanocomposites,
much lower electrical conductivities and low carrier concentrations
were obtained (p = 1016 cm–3), indicating a quasi-intrinsic behavior (very low doping
levels).
Seebeck coefficients were also much smaller (103 μV K–1) with a sign inversion at
approximately 470 K; a behavior associated
with bipolar effects. At RT, electronic transport was somewhat dominated
by holes, and as the temperature increased, the electrons become the
major carrier type. Carrier concentration could be further tuned in
the range of 1016–1019 cm–3 by combining K2Te- and Te-treated NCs in various ratios
(Figure 4
).
Figure 4
(a) Summary
of the results obtained with Te-based ligands and 11
nm PbS NCs at RT. (b) Temperature-dependent electrical conductivities
and (c) Seebeck coefficients.
The approach presented herein, which combines surface functionalization
and thermally induced substitutional doping, can be extended to other
semiconductor NCs. As an example, in a fully lead-free system, derived
in an analogous manner from colloidal SnSe NCs and K2Se
as capping ligands, a p-type electrical transport
with Seebeck coefficients of 320 μV K–1 and
high hole concentrations of 6 × 1017 cm–3 were obtained at RT (Figure 5
).
Figure 5
Temperature-dependent electrical conductivity (left) and Seebeck
coefficient (right) of SnSe nanocomposites obtained from organic-capped
and K2Se-treated SnSe NCs.
In summary, a bottom-up strategy to produce fully inorganic,
nanostructured
Pb chalcogenide solids with tunable p-type transport
using colloidal PbS NCs as initial building blocks has been presented.
In particular, surface functionalization was used as a platform to
modulate NC stoichiometry as well as to introduce controlled amounts
of dopants. Substitutional doping with K+ and Na+ ions, with hole concentrations adjustable
up to 3 × 1019 cm–3, was accomplished via ligand exchange
of the native organic surface molecules with alkali metal chalcogenides,
followed by thermal consolidation into densely packed solids. We envision
this strategy to be highly instrumental for both thin-film and bulk-like
solids, with possible applications in thermoelectrics and electronics.