Design Principles for Trap-Free CsPbX3 Nanocrystals: Enumerating and Eliminating Surface Halide Vacancies with Softer Lewis Bases

Article abstract of DOI:10.1021/jacs.8b11035

We introduce a general surface passivation mechanism for cesium lead halide perovskite materials (CsPbX₃, X = Cl, Br, I) that is supported by a combined experimental and theoretical study of the nanocrystal surface chemistry. A variety of spectroscopic methods are employed together with ab initio calculations to identify surface halide vacancies as the predominant source of charge trapping. The number of surface traps per nanocrystal is quantified by ¹H NMR spectroscopy, and that number is consistent with a simple trapping model in which surface halide vacancies create deleterious under-coordinated lead atoms. These halide vacancies exhibit trapping behavior that differs among CsPbCl₃, CsPbBr₃, and CsPbI₃. Ab initio calculations suggest that introduction of anionic X-type ligands can produce trap-free band gaps by altering the energetics of lead-based defect levels. General rules for selecting effective passivating ligand pairs are introduced by considering established principles of coordination chemistry. Introducing softer, anionic, X-type Lewis bases that target under-coordinated lead atoms results in absolute quantum yields approaching unity and monoexponential luminescence decay kinetics, thereby indicating full trap passivation. This work provides a systematic framework for preparing highly luminescent CsPbX₃ nanocrystals with variable compositions and dimensionalities, thereby improving the fundamental understanding of these materials and informing future synthetic and post-synthetic efforts toward trap-free CsPbX₃ nanocrystals.

Full text of DOI:10.1021/jacs.8b11035

Subscriber access provided by TULANE UNIVERSITY
Article
3
Design Principles for Trap-Free CsPbX Nanocrystals: Enumerating
and Eliminating Surface Halide Vacancies with Softer Lewis Bases
David P. Nenon, Kimo Pressler, Jun Kang, Brent A. Koscher, Jacob H. Olshansky,
Wojciech T. Osowiecki, Matthew A. Koc, Lin-Wang Wang, and A. Paul Alivisatos
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.8b11035 • Publication Date (Web): 02 Dec 2018
Downloaded from http://pubs.acs.org on December 2, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Design Principles for Trap-Free CsPbX Nanocrystals:
3
Enumerating and Eliminating Surface Halide Vacancies
with Softer Lewis Bases
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
David P. Nenon,†,⊥ Kimo Pressler, Jun Kang, Brent A. Koscher, Jacob H. Olshansky,†,⊥
†
⊥
†,⊥
⊥
Wojciech T. Osowiecki,†,⊥ Matthew A. Koc,†,⊥ Lin-Wang Wang, and A. Paul
Alivisatos*,†,⊥,§,#
†
Department of Chemistry,
#
Department of Materials Science and Engineering, University of
California, Berkeley, Berkeley, California 94720, United States
⊥
Material Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California
9
4720, United States
§
Kavli Energy NanoScience Institute, Berkeley, California 94720, United States
*
Author to whom correspondence should be addressed
ABSTRACT
We introduce a general surface passivation mechanism for cesium lead halide perovskite
materials (CsPbX , X = Cl, Br, I) that is supported by a combined experimental and theoretical
3
study of the nanocrystal surface chemistry. A variety of spectroscopic methods are employed
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 2 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
together with ab initio calculations to identify surface halide vacancies as the predominant source
1
of charge trapping. The number of surface traps per nanocrystal is quantified by H NMR
spectroscopy, and that number is consistent with a simple trapping model in which surface halide
vacancies create deleterious under-coordinated lead atoms. These halide vacancies exhibit trapping
behavior that differs between CsPbCl , CsPbBr , and CsPbI . Ab initio calculations suggest that
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
3
3
introduction of anionic X-type ligands can produce trap-free band gaps by altering the energetics
of lead-based defect levels. General rules for selecting effective passivating ligand pairs are
introduced by considering established principles of coordination chemistry. Introducing softer,
anionic, X-type Lewis bases that target under-coordinated lead atoms results in absolute quantum
yields approaching unity and monoexponential luminescence decay kinetics, thereby indicating
full trap passivation. This work provides a systematic framework for preparing highly luminescent
CsPbX nanocrystals with variable compositions and dimensionalities, thereby improving
3
fundamental understanding of these materials and informing future synthetic and post-synthetic
efforts towards trap-free CsPbX nanocrystals.
3
INTRODUCTION
Essentially trap-free nanocrystals represent optimal building blocks for optoelectronic
technologies, as these materials possess high luminescence efficiencies,1-3 narrow fluorescence
4
-9
linewidths, excellent photostability, and tunable absorption and emission. Inorganic cesium lead
halide (CsPbX , X = Cl, Br, I) perovskite nanocrystals offer an appealing alternative to prototypical
3
metal chalcogenide nanocrystals, as they represent an important class of ionic, defect-tolerant
semiconductors with tunable emission in the visible spectrum.10-18 The defect tolerance of CsPbX3
nanocrystals is a major enabling factor in their bright photoluminescence without a passivating
shell.19-21 However, the quantum yields of these nanocrystals do not yet regularly approach unity.
ACS Paragon Plus Environment

Page 3 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
The utility of devices based on CsPbX nanocrystals is currently limited by incomplete
3
understanding of their surface chemistry and the underlying charge trapping pathways that prevent
near-unity photoluminescence quantum yield. Addressing these challenges not only has the
potential to greatly enhance the utility of this new class of nanocrystals, but also to increase
understanding of charge trapping and ligand binding for highly ionic semiconductor surfaces.
Termination of an extended inorganic solid crystal inherently creates under-coordinated
atoms, making surface traps a ubiquitous challenge for nanocrystals.22-31 The pursuit of trap-free
nanocrystals is thus intimately intertwined with identification and passivation of surface defects.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Nanocrystal Ligand Binding Motifs
Typically, passivation is achieved by
According to the Covalent Bond Classification Method
employing an epitaxial shell of a larger
band gap material1, 32 or through the use of
capping ligands.33-36 Passivating ligands
are a particularly promising option, as a
larger band gap shell introduces a
permanent tunneling barrier that hinders
3
7
charge extraction. Ligands can bind to nanocrystal surfaces in a variety of motifs, well described
3
8
by the Covalent Bond Classification. Scheme 1 depicts L-, X-, and Z-type ligands, which are
classified by the number of electrons (2, 1, and 0, respectively) that the neutral ligand donates to
the metal when forming a bond to the surface.
Nanocrystal surfaces are difficult to characterize due to both their spatial and temporal
heterogeneity.23, 39, 40 This problem is exacerbated by weak bonding on CsPbX nanocrystal
3
surfaces, which creates a highly dynamic ligand shell. Initial reports on CsPbBr nanocrystals
3
1
0
synthesized by the standard oleylamine/oleic acid protocol have shown that a highly dynamic
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
equilibrium exists between free and bound oleylammonium bromide ligand pairs while oleate
species do not bind to the surface.41-43 Surface cesium can be replaced by the oleylammonium head
4
2
group in a substitution mechanism. Alternatively, oleylammonium can bind surface halides
through an addition mechanism. The literature on surface passivation of lead halide perovskite
materials is growing rapidly, with a wide variety of chemical species having been shown to
improve photoluminescence.3, 44-57 Here, we seek to unify these findings through a molecular-level
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
understanding of surface trap passivation in CsPbX nanocrystals; this requires a combined
3
experimental and theoretical investigation of their surface chemistry.
In this article, we demonstrate a systematic route to highly luminescent CsPbX3
nanocrystals with variable compositions and morphologies. We explore the relationship between
surface chemistry and optoelectronic performance by performing a wide range of ligand
exchanges. X-ray photoelectron spectroscopy, absolute photoluminescence quantum yield
(PLQY) experiments, and ab initio calculations are used to elucidate the nature of the organic-
inorganic interface and its role in charge trapping. The number of surface defects per nanocrystal
1
is quantified by H NMR spectroscopy and is consistent with a simple model of charge trapping.
Surface halide vacancies are identified as the predominant source of charge trapping, with an effect
that varies significantly based on CsPbX halide composition. Different degrees of defect tolerance
3
between the halide compositions helps to explain the relatively poor PLQY of CsPbCl in addition
3
to the high PLQY of CsPbBr and CsPbI . Guided by ab initio calculations, we develop a general
3
3
scheme for ligand passivation of deleterious under-coordinated lead atoms that is grounded in
established principles of coordination chemistry. This analysis is used to prepare essentially trap-
free CsPbBr and CsPbI nanocrystals along with brighter CsPbBr Cl nanocrystals. We propose
3
3
3-x
x
a general passivation mechanism whereby new ligands bind surface lead by substituting into
ACS Paragon Plus Environment

Page 5 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
surface halide vacancies, thus raising the energy of lead 6p levels to where they are no longer
present in the band gap. This mechanism is able to unify existing literature reports, thereby
informing synthetic and post-synthetic design principles for trap-free CsPbX nanocrystals.
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
MATERIALS AND EXPERIMENTAL METHODS
Chemicals. Cesium carbonate (99.9%), 1-octadecene (90%), oleic acid (90%), oleylamine (70%),
lead(II) bromide (99.999%), lead(II) chloride (99.999%), lead(II) iodide (99.999%), lead(II) oxide
(
99.999%), anhydrous toluene (99.8%), anhydrous ethyl acetate (99.8%), anhydrous hexanes
>99%), anhydrous tetrahydrofuran (>99%), didodecyldimethylammonium bromide (98%),
(
mesitylene (>99.8%), benzoic acid (>99.5%), benzylamine (>99%), decylamine (95%), myristic
acid (>99%), hexylphosphonic acid (95%), 2-ethyl-1-hexylamine (98%), fluoroacetic acid (95%),
difluoroacetic acid (98%), trifluoroacetic acid (99%), methanesulfonic acid (>99%), tri-n-
octylphosphine (97%), butyric acid (>99%), lead(II) nitrate (>99%), lead(II) hydroxide (99%),
cesium acetate (>99.99%), toluene-d (>99%), nickel(II) chloride hexahydrate (>97%), and zinc
8
dust (>98%) were purchased from Sigma-Aldrich and used as received. Ethyliododifluoroacetate
(96%) was purchased from Matrix Scientific and used as received. 2,2-difluorononanoic acid was
synthesized from ethyliododifluoroacetate and 1-octadecene following a previously reported
procedure.58, 59
Nanocrystal Synthesis and Isolation. CsPbBr nanocrystals were synthesized following a
3
1
0
previously reported procedure, with slight modifications. Typically, anhydrous ODE (5 mL) and
PbBr (0.188 mmol) were loaded into a 3-neck flask inside of a glovebox, and the flask was then
2
transferred to a Schlenk line and dried/degassed under vacuum at 110 °C for 15-20 minutes. Dried
oleic acid (0.5 mL) and oleylamine (0.6 mL) were then injected under dry argon gas and the
temperature was raised to 140 °C to complex the PbBr salt. Following complete dissolution of
2
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
PbBr , which we found to occur within 15-20 minutes, the temperature was raised to 155 °C in
2
preparation for the injection of warm (~100 °C) cesium oleate solution (0.5 mL, 0.125 M).
Following injection, the mixture was stirred for 5 s and then cooled using an ice-water bath.
Following the synthesis, 5-10 mL of anhydrous hexanes was added to the crude solution, which
was then centrifuged at 4000 rpm for 3 min to remove excessively large nanocrystals and other
unwanted byproducts. The supernatant was observed to be transparent and intensely green in color.
The solubility of the nanocrystals depended on the relative proportions of ODE, a poor solvent for
these nanocrystals, and hexanes, a better solvent. Hexanes was removed in a step-wise fashion by
flushing the surface of the solution with a stream of argon or nitrogen, and different sizes of
nanocrystals were isolated from the reaction mixture. Isolated nanocrystals were resuspended in
anhydrous hexanes or toluene and stored in a glovebox.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Ligand Exchange Method. Stock solutions of nanocrystals (~1-2 M) were precipitated with
addition of an antisolvent (typically anhydrous ethyl acetate, 2 the volume of original nanocrystal
solution), separated from the supernatant via centrifugation (10,000 rpm for 6 minutes), then
resuspended in a dilute (~1-10 mM) mixture of new ligand pairs in anhydrous hexanes or toluene.
Repeating this precipitation-resuspension process a total of 3 times was found to yield a complete
exchange in all cases. Performing this process in neat solvent rather than a dilute ligand pair
solution rapidly degrades the nanocrystals.
Optical Spectroscopy. All optical measurements were performed on dilute samples of
nanocrystals in hexanes or toluene. For typical CsPbBr nanocrystals, concentrations on the order
3
of 1-10 nM were found to have suitable optical densities. Absorption spectra were collected on a
Shimadzu UV-3600 spectrophotometer with 0.5 nm increments using the slowest scan speed.
Photoluminescence emission spectra were collected on a Horiba Jobin Yvon TRIAX 320
ACS Paragon Plus Environment

Page 7 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Fluorolog. Time-resolved fluorescence lifetimes were collected on a Picoquant Fluotime 300 with
PMA 175 detector and an LDH-P-C-405 diode laser (excitation wavelength of 407.7 nm).
Absolute quantum yields were determined optically using a custom integrating sphere
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
spectrofluorometer which is described elsewhere.60
Powder X-ray diffraction (XRD). Diffraction patterns of drop-cast nanocrystal samples were
obtained using a Bruker D-8 GADDS diffractometer equipped with a Co K source and a Bruker
Vantec 500 detector. Patterns were typically collected by merging frames centered at 2 = 30, 50,
and 70 degrees where each frame was collected for 40 minutes.
X-ray Photoelectron Spectroscopy (XPS). XPS spectra of CsPbBr nanocrystals drop-cast on a
3
Si wafer were collected using Thermo Scientific K-Alpha Plus X-ray photoelectron spectrometer.
The spectra were acquired with monochromatized Al Kα radiation and 400 µm beam size. Cs, Pb,
and Br were quantified by fitting GL(30) peak shapes and calibrating the C 1s edge to 284.8 eV.
NMR spectroscopy. NMR spectra of micromolar concentrations of CsPbBr nanocrystals in
3
toluene-d were measured on a Bruker 700 MHz spectrometer with an inverse cryoprobe for
8
1
greatly enhanced H sensitivity. Quantitative measurements were collected after tuning the probe
and determining the exact 90 radio frequency pulse. After a spectrum was collected, concentration
was determined by integrating the peak(s) of interest against an internal standard (mesitylene) of
known concentration. 2D NOESY experiments were performed using a gradient-enhanced
NOESY pulse sequence, with a mixing time of 500 ms and a delay time of 7 s. Spectra were
typically collected overnight (10-12 hour acquisition times) to ensure high-quality data.
Computational Methods. Density functional theory calculations were performed using the
6
1
Vienna ab initio simulation package (VASP). The core−valence interaction was described by the
62
projector−augmented wave (PAW) method. The cutoff for planewave expansion was set to 300
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
eV. Structures were relaxed until the force on each atom was smaller than 0.05 eV/Å. For the
structural relaxation of systems with halide vacancies, the screened Coulomb hybrid functional of
6
3
Heyd−Scuseria−Ernzerhof (HSE) was adopted, and for the relaxation of other systems, the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
64
generalized gradient approximation of Perdew−Burke−Ernzerhof (GGA−PBE) was used. The
electronic structures for all systems considered were calculated at the HSE level after relaxation.
The spin−orbit coupling (SOC) correction was also included in both structural relaxation and
electronic structure calculations. The surface slab model was constructed based on orthorhombic
CsPbX . It contains 11 atomic layers in a 2×2 supercell with 216 atoms. A vacuum layer larger
3
than 12 Å was used to avoid interaction between periodic images. The Brillouin zone was sampled
by the  point.
RESULTS AND DISCUSSION
On the Atomistic Nature of the CsPbX Inorganic Surface. Following synthesis and isolation,
3
colloidal CsPbBr nanocrystals exhibit cube-like morphology with typical size distributions 10%
3
in edge length as measured (N=75 particles) by transmission electron microscopy (TEM) (See
Supporting Information). Photoluminescence quantum yield (PLQY) values range from 60–90%,
60
determined optically using an integrating sphere. Although these nanocrystals are relatively
bright, sub-unity PLQY values demonstrate the presence of energetic losses due to deleterious
charge trapping and nonradiative recombination, thereby motivating an investigation into the
origins of these processes. The underlying trapping pathways that prevent near-unity PLQY in
CsPbX nanocrystals remain a subject of debate, which hinders rational improvements of
3
optoelectronic performance.
For metal chalcogenide nanocrystals, dangling bonds at the surface introduce localized
mid-gap states that can trap charges.22-24, 65, 66 However, ab initio calculations of pristine (100)
ACS Paragon Plus Environment

Page 9 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
20, 42, 67, 68
CsPbX surfaces terminated by CsX facets
yield trap-free, fully delocalized electronic
3
structures without the need for passivating ligands (Figure S1).20, 69 This suggests that surface
dangling bond states of low-index CsX terminated CsPbX nanocrystals lie outside of the band
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
gap and thus do not affect photoluminescence efficiency. Therefore, we hypothesized that as-
synthesized surfaces are not pristine, but rather contain local point defects that are a likely cause
of sub-unity PLQY in CsPbX nanocrystals. Unfortunately, the atomistic nature of the CsPbX
3
3
nanocrystal surface remains poorly understood. As such, quantitative X-ray photoelectron
spectroscopy (XPS) – a surface-sensitive technique – was used to elucidate the surface
stoichiometry of four different CsPbBr samples. Based on the inelastic mean free path, the first
3
2.0 nm of material was probed in this experiment. For these samples, each with PLQY of 60-65%,
an average surface stoichiometry of Cs:Pb:Br of 0.780.04:1:2.830.02 was determined (Figure
S2, Table S1). Deviations from the expected stoichiometry (Table S2-S3) support the hypothesis
regarding the presence of surface point defects, namely vacancies in these systems.19, 21, 70
However, the number of vacancies determined by XPS represents an upper bound, as the low
cohesive energy of lead halide perovskite materials can lead to the loss of surface atoms when
samples are transferred from the solvent phase to the vacuum, and exposed to high-energy X-rays
in the ultra-high vacuum environment.71, 72 In contrast, ambient techniques such as 1H NMR
spectroscopy can offer quantitative insight into surface structure without inducing sample
evolution and/or degradation, as will be demonstrated later in this work.
Sub-stoichiometry of surface cesium is well-documented in the literature, and has been
explained by a substitution mechanism where oleylammonium ions substitute into cesium
vacancies.42, 73, 74 A Pb:Br ratio that deviates from the expected stoichiometry is also well-
documented in the literature, with the common interpretation being a lead-rich surface.3, 52
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
However, this result can be equivalently interpreted as a bromide-deficient surface, an
interpretation that shapes the way we think about charge trapping in these materials. One can assess
the validity of each interpretation by considering existing ab initio calculations of APbX surface
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
defects. For a lead-rich surface, extra lead atoms on CsX facets would need to occupy sites such
as adatoms, antisites, and interstitials, which have prohibitively high (>2 eV) formation energies.21,
6
9, 70
Additionally, a lead-rich surface introduces defect levels that are too deep (defined here as
>
5k T) within the band gap to be consistent with 80-90% PLQY for as-synthesized CsPbBr3
b
7
0
nanocrystals. In contrast, a bromide-deficient surface would present bromide vacancies as the
dominant defect type, which introduces shallow (defined here as <5k T) defect levels.75-77 The
b
presence of surface bromide vacancies is also consistent with labile oleylammonium bromide
ligand pairs, bromide-poor synthetic conditions, and high anionic conductivities.11, 12, 78, 79
Therefore, for both electronic and synthetic reasons, we believe that the 1:2.830.02 Pb:Br surface
ratio can be accounted for with bromide vacancies rather than excess lead atoms.
The Role of Surface Halide Vacancies in Charge Trapping. Non-stoichiometric nanocrystal
surfaces can harm optoelectronic performance by creating localized trap states.66, 80-82 Guided by
non-stoichiometric XPS results, halide-deficient surfaces were investigated as the potential source
of charge trapping. We sought a method for controlling the introduction of halide vacancies on the
surface of CsPbX nanocrystals. Given the unstable nature of lead halide perovskite materials, it
3
was particularly important to devise a method that can remove surface halides without altering or
degrading the material in any other way. This was achieved by directly exploiting the weak
bonding of oleylammonium halide ligand pairs, whose chemical equilibrium is described by eq 1.
V represents a surface halide vacancy.
X
푅푁퐻 + 푋 + 퐶푠푃푏푋 ∙ 푉 ↔퐶푠푃푏푋 ∙ 푅푁퐻 푋 -
-
+
(1)
3
3
푋
3
3
ACS Paragon Plus Environment

Page 11 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
It is unfavorable for oleylammonium ions to dissociate as discrete solvated ions in low polarity
solvents such as toluene, therefore they are most likely removed from the surface with a counterion
8
3
41-43
to preserve charge neutrality. With oleate ions absent from the surface,
halide ions are the
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
only available counterions present. A simple dilution experiment offers a very mild and
controllable way to introduce surface halide vacancies as the ligand pair equilibrium shifts towards
free species. As such, monitoring the PLQY over a range of concentrations offers direct insight
into the PLQY dependency on the halide vacancy concentration. Although excessive dilution can
induce rapid degradation of CsPbX3
nanocrystals, this was only found to occur for
nanocrystal concentrations <0.005 M, which
is far below the concentrations utilized here.
These experiments demonstrate a
composition-dependent relationship between
surface halide vacancy concentration and
PLQY. Increasing the concentration of surface
Figure 1. Absolute PLQY as a function of nanocrystal
3 3 3
concentration for CsPbCl , CsPbBr , and CsPbI .
Nanocrystal concentration and surface halide vacancy
concentration are inversely related due to weak binding
halide vacancies is observed to have a
of oleylammonium halide ligand pairs. This allows the negligible effect on CsPbI , a small to
3
relationship between PLQY and halide vacancy
concentration to be investigated. Surface halide
vacancies are observed to have a negligible effect on
moderate effect on CsPbBr , and a significant
3
CsPbI
3
nanocrystals, a moderate effect on CsPbBr
3
nanocrystals, and a significant effect on CsPbCl
3
effect on CsPbCl , as shown in Figure 1.
3
nanocrystals. The data can be fit with a simple trapping
model, which offers insight into the relative defect
tolerance of each material by allowing the
Although CsPbI nanocrystal photophysics
3
r
determination of k /knr,trap.
1
appear unchanged upon dilution, H NMR
spectroscopy experiments were used to confirm that dilution does in fact result in significant ligand
1
pair desorption. The oleylammonium iodide H NMR linewidth decreased significantly as the
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
sample was diluted (Figure S3). The linewidth of a ligand pair in fast exchange is a population-
8
4
weighted average of free and bound signals, as described by eq 2:
푁푓푟푒푒
푁푡표푡푎푙
푁푏표푢푛푑
휆표푏푠
=
휆푓푟푒푒
+
휆푏표푢푛푑
(2)
푁푡표푡푎푙
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
where  is the linewidth and N is the number of ligand pairs per nanocrystal. Ntotal, free,
and bound are unchanged by dilution, therefore a narrowing of the observed linewidth must be
caused by a decrease in Nbound and a concomitant increase in N . This directly demonstrates that
free
the equilibrium in eq 1 is shifting towards free species in these experiments.
The plot of PLQY vs. nanocrystal concentration was analyzed within the context of a
simple trapping model, eq 3-4.
푁푡푟푎푝
푃퐿푄푌 = 1 ―
,
(3)
(4)
푘푟
+
푁푡푟푎푝
푘푛푟,푡푟푎푝
푁푚푎푥
푅푁퐻 + 푋 -
푁푡푟푎푝
=
1
+퐾푒푞
[
]
3
where k is the radiative rate, k
is the effective nonradiative rate per trap, Ntrap is the
r
nr,trap
number of surface traps per nanocrystal, Nmax is the maximum number of surface traps per
nanocrystal, K is the free vs. bound equilibrium constant for oleylammonium halide ligand pairs,
eq
+
–
and [RNH X ] is the concentration of oleylammonium halide ligands pairs free in solution. The
3
determination of all variables other than k and k
(Figure S4) leaves k /k
as the sole fitting
r
nr,trap
r
nr,trap
parameter.
ACS Paragon Plus Environment

Page 13 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Defect tolerance, which we define here as arising through very large values of k /knr,trap,
r
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 14 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
varies significantly between the different halide compositions (Figure 1). k /k
= 9,500 allows
r
nr,trap
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 15 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
CsPbI nanocrystals to maintain near-unity PLQY despite the presence of several hundred surface
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 16 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
halide vacancies. Within the lead halide perovskite family, the iodide materials appear to be the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 17 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
closest to defect impervious. k /k
= 390 for CsPbBr nanocrystals indicates these materials are
3
r
nr,trap
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 18 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
also relatively tolerant of defects, although the effects of surface halide vacancies clearly cannot
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 19 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
be ignored in this system. k /k
= 53 for CsPbCl nanocrystals is the most analogous to defect-
r
nr,trap
3
intolerant metal chalcogenide nanocrystals, where even a small number (<5% of the surface) of
traps will yield PLQY values <20%. As a final comment in support of halide-deficient surfaces, it
should be noted that if under-coordinated lead in a form other than that created by halide vacancies
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
was the predominant source of charge trapping, no variation in k /knr,trap and PLQY between
r
Figure 2. Electronic structure and charge density calculations at the HSE+SOC level of theory for (A) CsPbCl
surface chloride vacancy (CsPbCl +VCl), (B) CsPbBr with a surface bromide vacancy (CsPbBr +VBr), and (C) CsPbI
with a surface iodide vacancy (CsPbI +V ). The top surface of the crystal slab is halide-deficient while the bottom
3
with a
3
3
3
3
3
I
1
8
surface is pristine. Absolute CBM energies are set according to electrochemically measured values, and relative
energies are given by the calculations. Cs, Pb, Cl, Br, and I atoms are shown as blue-green, gray, green, orange, and
purple, respectively.
different halide compositions would be expected. Ab initio calculations have shown that these
defects form deep mid-gap states that would affect all halide compositions equally.70
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 20 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Ab Initio Calculations of Halide-Deficient CsPbX Surfaces. Ab initio calculations were used
3
to further investigate halide-deficient surfaces as the suspected source of charge trapping. A single
halide vacancy (V , V , or V ) was created on the surface of each crystal slab to model a halide-
Cl
Br
I
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
deficient surface. Native oleylammonium surface ligands are excluded because pristine CsPbX3
crystals terminated by CsX facets exhibit fully delocalized, trap-free band structures without the
need for passivating ligands (Figure S1). As such, differences in electronic structure between
pristine and halide-deficient surfaces can be directly attributed to surface halide vacancies.
CsPbCl +V is readily recognized as a three-level system, with a highly localized trap
3
Cl
state deep within the band gap in addition to the fully delocalized valence band maximum (VBM)
and conduction band minimum (CBM), as shown in Figure 2A. The mid-gap state exhibits
significant atomic character; ~80% of this state is comprised of 6p orbitals of the lead atom that is
left under-coordinated by the surface chloride vacancy. CsPbBr +V and CsPbI +V also appear
3
Br
3
I
as three-level systems (Figure 2B-C). However, the nature of the mid-gap defect levels differs
significantly from the defect level in CsPbCl +V . Rather than being highly localized and deep
3
Cl
within the band gap, these states are shallow and only weakly localized towards the halide-
deficient surface. In these two systems, the under-coordinated lead atoms only contribute ~3% to
the defect states. It should be noted that the HSE+SOC level of theory is required to observe the
8
5, 86
shallow levels in CsPbBr +V and CsPbI +V .
Calculations with GGA–PBE+SOC, which
3
Br
3
I
8
7
underestimated the band gap by ~1 eV, produced nominally trap-free systems (Figure S5).
However, sub-unity experimental PLQY values indicate that these systems are not trap-free.
It is particularly insightful to interpret these calculations of halide-deficient CsPbX3
surfaces in light of experimental results. The calculated position of these defect levels relative to
the CBM is highly consistent with the k /k
ratios determined in Figure 1. Nanocrystals can
r
nr,trap
ACS Paragon Plus Environment

Page 21 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
88
sustain a high charge transfer rate at high driving force, and with the CsPbCl +V defect state
3
Cl
located directly in the middle of the band gap, efficient trapping of charges would be expected.89
Additionally, the highly localized nature of this mid-gap state should result in significant
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
nonradiative recombination, thereby yielding a relatively low k /k
as well as relatively poor
r
nr,trap
PLQY values for CsPbCl nanocrystals. In contrast to highly localized charges in CsPbCl +V ,
3
3
Cl
these calculations suggest that electrons in halide-deficient CsPbBr and CsPbI are only weakly
3
3
localized while holes remain fully delocalized, and thus trap-assisted nonradiative recombination
9
0-92
is therefore expected to be relatively ineffective in CsPbBr and CsPbI .
This allows k /k
3
3
r
nr,trap
1
0, 93
to become quite large in these two systems, even though k is decreased relative to CsPbCl .
r
3
We sought to use ab initio calculations not only to investigate the predominant source of
charge trapping, but also to identify a potential surface passivation mechanism. Oleate was
investigated as a passivating ligand, as it has the potential to bind the under-coordinated lead atoms
–
that have been implicated in charge trapping. X-type CH COO moieties, truncated computational
3
models for oleate, were substituted into surface halide vacancies of the structures studied in Figure
2
. These structures produce trap-free band gaps with fully delocalized VBM and CBM states for
each halide composition, indicating that anionic surface ligands can effectively alter the energetics
of lead-based defect levels, removing them from within the band gap (Figure 3). However, several
reports have demonstrated that oleate is not present on the surface of as-synthesized CsPbBr3
nanocrystals,41-43 which we ascribe to a hard-soft mismatch between hard carboxylates and soft
lead binding sites.23, 94 This motivates a search for new, softer ligands that can bind deleterious
under-coordinated lead.
A Versatile Ligand Exchange and Purification Method. Investigations into the effects of new
ligand shells would be greatly enhanced by a versatile ligand exchange method.65, 95 However, to
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 22 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the best of our knowledge, no versatile colloidal ligand exchange method has yet been reported for
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 23 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
CsPbX nanocrystals. As such, precise manipulations of CsPbX nanocrystal surfaces and entire
3
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 24 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ligand shells are not readily available. Here, we introduce a colloidal ligand exchange method that
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 25 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
can exchange native oleylammonium halide ligand pairs for a mixture of alkylammonium halide
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 26 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and alkylammonium-alkylcarboxylate/alkylphosphonate ligand pairs (Figure S6). This is
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. Charge density calculations at the HSE+SOC level of theory for (A) CsPbCl
3
with a surface chloride vacancy
+VBr), (C) CsPbI with a surface iodide vacancy
COO X-type passivating ligand, (E) CsPbBr +VBr from (B) with
COO X-type passivating ligand. VBM and
(
(
CsPbCl
CsPbI +V
COO X-type passivating ligand, (F) CsPbI
3
+VCl), (B) CsPbBr
3
with a surface bromide vacancy (CsPbBr
3
3
–
3
I
), (D) CsPbCl +VCl from (A) with a CH
3
3
3
–
–
a CH
3
3
I 3
+V from (C) with a CH
CBM states are unaffected by surface halide vacancies in (A), (B), and (C), and thus are excluded. In all cases, X-type
passivation of halide-deficient surfaces yields a trap-free band gap with fully delocalized VBM and CBM states. Cs,
Pb, Cl, Br, I, O, C, and H atoms are shown as blue-green, gray, green, orange, purple, red, brown, and white,
respectively.
ACS Paragon Plus Environment

Page 27 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
accomplished by directly exploiting the weak bonding between native ligand pairs and the CsPbX3
surface (Figure S7). Stock solutions of nanocrystals (~1-2 M) are precipitated with the addition
of antisolvent, separated from the supernatant via centrifugation, and resuspended in ~1-10 mM
solutions of new ligand pairs. Given the ionicity of perovskite nanocrystals, the use of molecular
combinations that can form ion pairs is highly beneficial. Performing this exchange process on
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Figure 4. (A) H NMR spectra for each step of a ligand exchange from native (blue) unsaturated ligand pairs
(oleylammonium and oleate) to new (red) saturated ligand pairs (decylammonium and myristate). Resonances
denoted by X are impurities. Concentration of CsPbBr nanocrystals: 1.6 M in toluene-d . (B) The 4-6 ppm region
3
8
of interest from (A). This region contains resonances from native oleyl vinyl protons (5), 1-octadecene (ODE), and
unknown impurities (X). All native species are reduced to <0.5% of their original concentration, demonstrating a
complete ligand exchange.
CsPbBr nanocrystals with a dilute solution of alkylamine or alkylcarboxylic/alkylphosphonic acid
3
alone, but not both species, yielded a weakly fluorescent, off-white precipitate. X-ray diffraction
determined the precipitate to be comprised primarily of CsBr and Cs PbBr phases (Figure S8).
4
6
To gain insight into the step-by-step evolution of this method, an exchange from as-
synthesized, unsaturated ligand pairs (oleylammonium and oleate) to saturated ligand pairs
1
(
decylammonium and myristate) was studied via quantitative H NMR for CsPbBr nanocrystals
3
(Figure 4, Figure S9). Unsaturated ligand pairs and 1-octadecene (ODE) both have alkene
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 28 of 55
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
resonances in the 4-6 ppm range, which are useful for quantitative studies since they are located
downfield from the many overlapping alkyl resonances in the 0-3 ppm region. Exchanging for
saturated ligand pairs, which have no alkene resonances, allows the region of interest to be studied
for the removal of alkene-containing species, all of which should be removed to constitute a
complete ligand exchange and purification (Figure 4B).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The exchange is observed to reach >99% exchange after three precipitation and
resuspension steps, i.e. three exchange cycles. Appearance of decylammonium and myristate
resonances and disappearance of alkene resonances confirms that native unsaturated ligand pairs
are exchanged for non-native saturated ligand pairs (Figure 4). All native organic species being
reduced to <0.5% of their original concentration confirms a complete ligand exchange. This is
explicitly demonstrated by exchanging to ligand pairs such as benzylammonium-oleate,
oleylammonium-benzoate, and oleylammonium-difluoroacetate, which have spectrally isolated
resonances that grow in downfield of the overlapping alkyl region as the exchange progresses
(Figure S10-S12). Each of these resonances shows a broadened peak that is characteristic of
interaction with nanocrystal surfaces.37, 65, 88, 96
This exchange method also serves to purify nanocrystals of ODE, impurities, and synthesis
byproducts such as lead oleate (Figure S13). As such, a resuspension solution of dilute oleylamine
and oleic acid can be used to purify as-synthesized nanocrystals without compromising high PLQY
or colloidal stability. X-ray diffraction, absorbance, photoluminescence, TEM, and integrating
sphere PLQY measurements confirm that the inorganic nanocrystalline core remains unchanged,
indicating that purification induces no macroscopic changes to the ensemble of particles (Figures
S14-S16).
ACS Paragon Plus Environment

Page 29 of 55
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Softer Lewis Bases Bind to Nanocrystal Surface. With control over entire ligand shells
established, new passivating ligand pairs were investigated. In light of the proposed hard-soft
mismatch between hard oleate and soft lead, softer carboxylates were targeted by exploring tail
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
Figure 5. H NOESY NMR spectra of CsPbBr
3
nanocrystal samples exchanged to ligand pairs of oleylammonium
and (A) benzoate, (B) difluoroacetate, and (C) hexylphosphonate. All ligand pairs feature negative (black) NOE
signals rather than positive (red) NOE signals, thereby corroborating their interaction with the nanocrystal surface.
A small amount of red coloring is also present, but this is due to t1 noise rather than a positive NOE signal.
groups that decrease electron density on the binding head (Figure S17, Table S5). This can be
accomplished through resonance and/or induction.97, 98 Following a ligand exchange to
–
–
oleylammonium–R-COO pairs, where R-COO is benzoate, fluoroacetate, and difluoroacetate,
Nuclear Overhauser Effect (NOESY) NMR spectroscopy confirms that these softer carboxylates
bind to the nanocrystal together with oleylammonium (Figure 5A-5B). Each ligand pair features
negative (black) cross peaks in the presence of nanocrystals, in contrast to positive (red) cross
peaks when no nanocrystals are present (Figure S18).41, 99 A change in the sign of the cross peak
demonstrates that the tumbling frequency of these ligand pairs decreases significantly in the
presence of nanocrystals, thereby confirming their interaction with the nanocrystal surface.100 By
utilizing the tail group to modulate electron density on the binding group, one can selectively
ACS Paragon Plus Environment