A General Approach for Monolayer Adsorption of High Weight Loadings of Uniform Nanocrystals on Oxide Supports

Article abstract of DOI:10.1002/anie.202017238

Monodispersed metal and semiconductor nanocrystals have attracted great attention in fundamental and applied research due to their tunable size, morphology, and well-defined chemical composition. Utilizing these nanocrystals in a controllable way is highl

Full text of DOI:10.1002/anie.202017238

Angewandte
Research Articles
Chemie
How to cite: Angew. Chem. Int. Ed. 2021, 60, 7971–7979
International Edition: doi.org/10.1002/anie.202017238
German Edition: doi.org/10.1002/ange.202017238
Surface Functionalization
A General Approach for Monolayer Adsorption of High Weight
Loadings of Uniform Nanocrystals on Oxide Supports
Kun-Che Kao, An-Chih Yang, Weixin Huang, Chengshuang Zhou, Emmett D. Goodman,
Alexander Holm, Curtis W. Frank, and Matteo Cargnello*
Abstract: Monodispersed metal and semiconductor nano-
crystals have attracted great attention in fundamental and
applied research due to their tunable size, morphology, and
well-defined chemical composition. Utilizing these nanocrys-
tals in a controllable way is highly desirable especially when
using them as building blocks for the preparation of nano-
structured materials. Their deposition onto oxide materials
provide them with wide applicability in many areas, including
catalysis. However, so far deposition methods are limited and
do not provide control to achieve high particle loadings. This
study demonstrates a general approach for the deposition of
hydrophobic ligand-stabilized nanocrystals on hydrophilic
oxide supports without ligand-exchange. Surface functionali-
zation of the supports with primary amine groups either using
an organosilane ((3-aminopropyl)trimethoxysilane) or bond-
ing with aminoalcohols (3-amino-1,2-propanediol) were found
to significantly improve the interaction between nanocrystals
and supports achieving high loadings (> 10 wt.%). The
bonding method with aminoalcohols guarantees the opportu-
nity to remove the binding molecules thus allowing clean metal/
oxide materials to be obtained, which is of great importance in
the preparation of supported nanocrystals for heterogeneous
catalysis.
structure between NCs and oxide support are key parameters
that determine the observable catalytic properties and
stability of the catalysts.[5e, 7] Therefore, developing methods
to controllably and reliably deposit uniform NCs on support
surfaces is crucial for preparing advanced materials and
devices for many applications especially in catalysis[5e,7b] and
sensing.[8]
When impregnating NCs from aqueous solutions, surface
charges are the most common parameters used to tune the
electrostatic interaction between NCs and supports and
develop attractive force between them aimed at NC deposi-
tion.[9] Surface modification of supports with functional
groups such as amines, phosphates and carboxylates is
a typical way to match their Zeta potential to NCs at a certain
range of pH.[9c] However, a variety of uniform NCs are
produced in non-aqueous conditions capped with long chain
alkyl surfactants, and multiple steps of ligand exchange or
surface coating may be required to make NCs dispersible in
the aqueous phase, if possible at all. Although useful methods
have been developed for this purpose,[10] these modifications
might inevitably lead to issues such as low yield and particle
agglomeration. On the other hand, with a typical loading
method conducted in a non-aqueous solution, common
supports such as oxides are usually not well dispersible due
to their hydrophilic hydroxyl surfaces. Finding a solvent with
suitable dispersibility for both oxides and the non-polar
ligand-stabilized NCs is crucial to obtain well-distributed
NCs. In addition, minimizing residual surface ligands by
washing NCs with anti-solvents while avoiding destabilization
can be very subtle for successful NCs deposition. Finding the
appropriate conditions to deposit hydrophobic colloidal NCs
on common oxide surfaces has therefore proven to be
challenging, and a generalized and straightforward solution
has not been found yet.
Stucky and co-worker investigated the complex interac-
tions between colloidal Au NCs, oxide supports, and organic
solvents.[11] An aprotic, slightly polar solvent such as chloro-
form was used as a good solvent for dispersing oxide materials
while maintaining NCs stability and allowing uniform depo-
sition. The weak dipole/induced-dipole or charge/induced-
dipole interactions between the oxides and NCs were claimed
to drive deposition. Recently, de Jong and co-workers
demonstrated adsorption of iron/iron oxide NCs on carbon
nanotube supports at 2008C in octadecene showing that
diffusion and dynamic adsorption and desorption of NCs were
important factors to uniformly deposit large loadings of NCs
(2–30 wt.%) on a porous support.[12] The authors also
hypothesized a mechanism where the carbon nanotube
Introduction
Colloidal nanocrystals (NCs) with dimensions below
100 nm possess unique electronic, magnetic, optical, and
chemical properties distinct from their bulk counterparts
playing an important role in modern science and technology.[1]
Incorporating NCs into materials and devices allows a variety
of novel applications in sensing,[2] optoelectronic,[3] bio-
medicine,[4] and catalysis.[5] Nowadays, monodisperse NCs of
many metals and semiconductors are synthesized through
solution chemistry approaches.[6] To utilize these colloidal
NCs in a variety of applications, their controlled immobiliza-
tion on a surface is usually the important first step. For
example, in heterogeneous catalysis, the distribution of
adsorbed NCs and the nanoscale geometry and interfacial
[*] K.-C. Kao, A.-C. Yang, W. Huang, C. Zhou, E. D. Goodman, A. Holm,
C. W. Frank, M. Cargnello
Department of Chemical Engineering, Stanford University
Stanford, CA 94304 (USA)
E-mail: mcargnello@stanford.edu
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under:
https://doi.org/10.1002/anie.202017238.
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979
ꢀ 2021 Wiley-VCH GmbH
7971

Angewandte
Research Articles
Chemie
surface acts as a surface ligand to bind NCs and providing Results and Discussion
a driving force for lowering the NCs surface energy. While
these studies partially solve the issues of NC adsorption in
specific NC-support combinations, there is a lack of quanti-
tative measurements and a more detailed analysis of the
adsorption behavior. In this contribution, we report a gener-
alized method for the adsorption of a variety of hydrophobic,
ligand-stabilized colloidal NCs on oxide surfaces at high
weight loadings through the understanding of surface modi-
fication and binding configuration. Primary amines are
introduced on oxide support surfaces through condensation
of an organosilane (Scheme 1a) or using 3-amino-1,2-pro-
The uniform dispersion and deposition of colloidal NCs
on an oxide support was initially studied using previously
reported approaches. All the NCs synthesized in this work
(Figure S1) were stabilized by long chain alkyl surfactants
such as oleylamine and oleic acid. To obtain homogeneously
distributed NCs, a 10% by volume ethanol/toluene mixture
was selected as the solvent for NC adsorption. This mixture
was selected to obtain a compromise between dissolving the
hydrophobic NCs in toluene, and dispersing the oxide support
using ethanol to increase the mixture polarity. Comparing
with using pure toluene, the 10% ethanol/toluene mixture
shows more well-distributed NCs adsorption on oxide surfa-
ces (Figure S3). In fact, when using the Hansen solubility
parameter to compare this mixture with chloroform, their
parameters are very close (including the components of
dispersion, polarity, and hydrogen bonding). Chloroform has
been demonstrated to be a good solvent for depositing
dodecanethiol-stabilized Au NCs on oxides.[11] An initial
study was performed by measuring the actual NC loading
using monodisperse, 5.7 nm Pt NCs (Figure S1b) on selected
oxide supports (g-Al2O3, TiO2, SiO2,) and reporting it as
a function of targeted NC loading (Figure 1a–c). While g-
Al2O3 fully adsorbed Pt NCs up to at least 15 wt. % loading
(Figure 1a), TiO2 and SiO2 showed much lower adsorption
ability (Figure 1b,c). As Stucky and co-worker pointed out,
ethanol could interfere with the adsorption of NCs due to
hydrogen bonding between alcohol and hydroxyl groups on
oxides that is stronger than the interaction between NCs and
oxide surface.[11] Among different supports, the available
surface area and density of hydroxyl groups might cause
differences in adsorption capacity. We hypothesized that
chemical interactions between NC ligands and support oxides
were not optimal for NC adsorption. Therefore, we decided to
introduce functional groups on the surface of the oxide
materials to strengthen the interaction with the NCs and allow
deposition. The oxide supports were chemically modified with
amino groups (see FTIR spectra in Figure S4) using an
organosilane ((3-aminopropyl)trimethoxysilane, or APTS) as
shown in Scheme 1a. The NC adsorption amounts were
greatly improved after APTS modification for all the oxide
supports (Figure 1d–f), indicating the formation of stronger
interactions between NCs and amine-functionalized surfaces.
The functionalized APTS either can possibly disrupt the
hydrogen bonding between the protic solvent and oxides, or
the attractive interactions between functionalized amines and
NCs appear to be even stronger than the hydrogen bonding
because all APTS-modified supports could adsorb at least up
to 15 wt.% metal NCs even in the presence of ethanol. It is
noted that the adsorbed NCs on the APTS-modifed supports
could still desorb from the oxide surfaces once the nano-
composites are redispersed in pure ethanol, indicating that
the attractive force between the amine-functionalized oxides
and NCs could be mainly due to dipole/induced-dipole or
charge/induced-dipole interactions. It has been reported that
a small amine-containing molecule (dimethylamine) exchang-
es the surface ligands in oleylamine- or polyvinylpyrrolidone-
stabilized metal NCs even if the binding strength of amino
Scheme 1. Surface functionalization of amino groups on oxide sup-
ports using a) (3-aminopropyl)trimethoxysilane (APTS) or b) 3-amino-
1,2-propanediol (APD). Proposed adsorption behavior of NCs on
c) bare oxide and d) amine-functionalized oxide supports.
panediol (Scheme 1b). A strong affinity of the amino group to
the ligand-stabilized NCs was observed even in the presence
of a polar protic solvent, thus allowing strong adsorption of
the NCs to the oxide surface. The chemical interactions
between solvent, oxide surface, and NCs were elucidated in
a quantifiable way using adsorption isotherms. The function-
alized amino groups might partially exchange the surface
ligands attached on the NCs at the adsorption interfaces
leading to a stronger binding strength than that of bare oxide
surfaces (Scheme 1d).[13] In addition, the thermally remov-
able diol modification is proven to be an important alter-
native to the organosilane modification which guarantees the
preparation of a single-layer homogeneous supported NC
catalyst with a wide range of tunable NC weight loadings
without changing the surface property of the oxide supports
deriving from the residual silica layer. The catalysts made by
the presented method optimize the distribution of NCs on
oxide supports even at weight loadings as high as 10 wt.%,
displaying much higher catalytic activity than those prepared
by conventional impregnation methods that lead to sintered
metallic phases.
7972
www.angewandte.org
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979

Angewandte
Research Articles
Chemie
Figure 1. Loading amount of Pt NC (5.7 nm) on bare (a, b, c) and APTS-functionalized (d, e, f) g-Al2O3, TiO2, SiO2, respectively, in 10 vol.%
ethanol/toluene mixture and (g, h, i) in 10 vol.% ethanol/toluene mixture with added 0.01 vol.% of 1-oleylamine (circles: bare oxides; squares:
APTS-functionalized oxides). The insets in (g, h, i) show the pictures of vials containing the supernatant and NC-oxide supports following
centrifugation.
groups to the NCs surface is similar.[13] Therefore, we
considered the possibility of ligand-exchange at the interfaces
between the NCs and amine-functionalized supports, which
could in part contribute to the enhanced binding affinity and
adsorption capacity, as depicted in Scheme 1d. Through
optimization of the dispersity and adsorption capability of
the oxide supports by using a suitable solvent (10% ethanol/
toluene) and a simple surface functionalization (amino
groups), respectively, a monolayer of adsorbed NCs on
supports without obvious agglomeration even at the highest
loading point shown in Figure 1d–f was achieved (Figure S5).
Excess surfactant derived from the NC synthesis that was
not properly removed during purification may also represent
an important factor in determining the NC adsorption and the
reproducibility of the process. Figure 1g–i shows the adsorp-
tion of purified NCs in the ethanol/toluene mixture with
intentionally added 0.01 vol.% of 1-oleylamine, which is used
as the capping ligand for the Pt NCs. The adsorption capacity
of the bare oxides was heavily influenced by this low
concentration of free surfactant in solution. The loading
amount was lowered to less than 2.0 wt.% on any support,
including alumina, in the presence of oleylamine. One
possible explanation is that the amino group of oleylamine
molecules could physically adsorb on the oxide through
hydrogen bonding with the surface hydroxyl groups, leading
to the formation of a monolayer that varies the surface
properties and leads to repulsive interactions with the NC
capping ligands (Scheme 1c). In fact, it was observed that the
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979
ꢀ 2021 Wiley-VCH GmbH
www.angewandte.org
7973

Angewandte
Research Articles
Chemie
oxides are more dispersible in non-polar solvents when
oleylamine was introduced, indirectly confirming that oleyl-
amine binds to the surface. It is important to note that if NCs
are not thoroughly purified after synthesis, any residual
surfactant in the NC solution could then cause challenges in
adsorbing NCs onto oxide supports. Despite the presence of
extra surfactant, when the oxide surface is functionalized with
amino groups from APTS, the NCs still fully adsorbed on the
oxide surface within the tested loading range (Figure 1g–i).
This result suggests that surface functionalization impedes
oleylamine from binding to the oxide surface, and the NCs
therefore have stronger attractive interactions that allow their
deposition (Scheme 1d). Since the excess oleylamine could
still block particle deposition at high concentrations, this
observation further suggests that it is not the interaction
between organic groups on the support and organic ligands on
the surface of the particles that drive the deposition, but
instead dipole/induced-dipole or charge/induced-dipole in-
teractions.
Scheme 1b and two more molecules (Figure 2a) including
1,3-propanediamine (PDA) and N-(2-aminoethyl)-1,3-pro-
panediamine (AEPA) were tested for the functionalization of
TiO2 in ethanol. After adsorption and drying, the function-
alized supports were subjected to temperature-programmed
oxidation to quantify the amount of bonded amine-containing
molecules using CO2 formation as a proxy (Figure 2b). While
diamine (PDA)- and triamine (AEPA)-modified TiO2 showed
much less of the CO2 production (0.33 and 0.42 mmolgÀ1,
respectively) compared to that of APTS-modified TiO2
(0.70 mmolgÀ1), the aminodiol (APD)-modified TiO2
(0.71 mmolgÀ1) produced a similar amount of CO2 as
estimated from the integrated areas. The binding strength of
3-amino-1,2-propanediol to the oxide surface was clearly
stronger than other physisorbed amines. Indeed, catechol
compounds such as dopamine have been demonstrated to
strongly bind to oxide surfaces in aqueous solution.[16] We
chose the diol molecule instead of catechol compound to
avoid possible polymerization byproducts (for example with
dopamine[16b]).
The data in Figure 1 demonstrates that the functionaliza-
tion of oxide support surfaces with amino groups is therefore
a successful strategy to directly adsorb large and controlled
loadings of hydrophobic ligand-stabilized colloidal NCs from
solution. It is noted that APTS-functionalization is a very
common strategy used to prepare supported NCs in aqueous
solution as well,[14] but the driving force is often based on
electrostatic interaction which is different from the presented
method as discussed above. However, in both cases surface
modification using an organosilane such as APTS eventually
leaves a silicon layer on the surface which cannot be removed
with thermal treatments, thus changing the chemical proper-
ties of the support which are often crucial in some applica-
tions such as catalysis.[14a,15] To minimize the influence of
surface modification, we therefore investigated support
functionalization strategies with amine-containing molecules
that can be removed through thermal treatments following
deposition. 3-amino-1,2-propanediol (APD) which could
possibly condense on oxide hydroxyl groups as shown in
Adsorption isotherms of Pt NCs on the amine-function-
alized TiO2 supports were then measured and plotted in
Figure 2c. APTS-TiO2 and APD-TiO2 adsorbed a much
larger amount of NCs, up to three times higher compared to
the other samples, indicating that the density of amino groups
is essential to the loading of NCs. It is possible that the slightly
lower adsorption capacity of APD-TiO2 than that of APTS-
TiO2 is due to the partial re-dissolution of adsorbed APD.
Figure 2d shows the time-evolution of the NC adsorption
behavior on the amine-modified TiO2 targeting 5.0 wt.%
loading of Pt NCs. APTS-TiO2 showed very fast NC uptake,
reaching almost 100% adsorption in 5 minutes. APD-modi-
fied TiO2 only needed 60 minutes to achieve > 95%
adsorption whereas the samples of non-covalently modified
TiO2 needed much longer time to reach equilibrium. These
results highlight the importance of both the anchored amino
group density and the bonding strength of the amine-
containing molecules on oxide support because the residual
Figure 2. a) The structures of the four amine-containing molecules for surface functionalization. b) Temperature-programmed oxidation of the
amines molecules modified onto TiO2 and c) adsorption isotherms and d) time-course adsorption amount of Pt NC (5.7 nm) on the amine-
functionalized TiO2 supports targeted at 5.0 wt.%.
7974 www.angewandte.org
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979

Angewandte
Research Articles
Chemie
surfactants and protic ethanol could possibly disrupt the
individual NC is higher, leading to increased surface cover-
physisorbed amines and hinder NC adsorption.
ages up to ꢀ 90%. The surface coverage of NCs on the SiO2
spheres modified with mesoporous layers clearly surpassed
that on the bare SiO2 spheres (Figure S9), indicating once
again the crucial role played by the amino groups as surface
ligands to immobilize NCs (Scheme 1d). Figure 4a–d and
Figure S10 show examples of NC adsorption using APTS-
modified oxide supports including Pt NCs on amorphous SiO2
gel and various sizes of Pd NCs on SiO2 spheres with
mesoporous layers. Even at a high surface coverage, mono-
layer deposition of NCs without agglomeration was realized
by using the amine-functionalized oxides.
To evaluate the adsorption ability of amine-modified
supports more comprehensively and accurately, adsorption
isotherms for various metal NCs (Au, Pt, and Pd) capped with
oleylamine or oleic acid were measured for three representa-
tive samples, namely APTS-TiO2, APD-TiO2, and bare TiO2.
The data (Figure S6) were normalized in terms of surface
coverage, which was calculated as the ratio between the total
projected area of the adsorbed NCs over the available surface
area of the support calculated by N2 physisorption (see
Figure S7 for detailed calculations). Since all supports have
a similar specific surface area (Table S2), the difference in
adsorption capability is attributed to the surface functional-
ization. Interestingly, similar adsorption isotherms were
measured for the three tested noble metal NC samples
(Figure 3a–c). Considering the inhomogeneity of adsorption
sites and adsorption processes, the Langmuir-Freundlich
model was adopted to fit the isotherms. According to the
fitted isotherms (Figure 3, solid curves) and extracted fitting
parameters (Table S3), APTS- and APD-modified TiO2
present a thermodynamically stronger affinity to NCs and
higher saturation surface coverage than that of bare TiO2
(70.1%, 49.2% and 34.9%, respectively). Especially in the
ranges where NCs fully adsorbed at a target loading amount
(dashed lines), APTS-TiO2 and APD-TiO2 guarantee around
4-fold and 2-fold increase of the loading range, respectively,
compared with bare TiO2 (Figure 3a–c), suggesting that the
functionalization process with amines is indeed advantageous
to improve NC deposition. In general, NC surface coverage is
independent of oxide support specific surface area. For
example, APTS-modified SiO2 spheres (Figure S8), with
a lower available surface area (18 m2 gÀ1), achieved an overall
lower weight loading of Pd NC ( ꢀ 15.1 wt.%) compared to
APTS-modified TiO2, but a similar maximum surface cover-
age ( ꢀ 58%) is fitted with the same isotherm shown in
Figure 3a. Increasing the number of amino groups on the
surface increases surface coverage as well because more
binding sites are used for NC adsorption. We demonstrate this
hypothesis by creating a thin mesoporous silica layer on SiO2
spheres (Figure S9) such that the APTS binding site for each
The APD modification is a very general method to
achieve high NC loading and it is not limited to titania but it
can also be used for many other oxide supports. Systematic
deposition of a variety of NCs (Figure S1) using APD-
modified oxides in 10% ethanol/toluene mixture was dem-
onstrated using different loadings of NCs, different NC sizes,
different metals, and different oxide supports (Figure 4).
Importantly, in all cases, complete deposition was observed,
as proven by the colorless supernatant following NC depo-
sition (inset of each panel in Figure 4). Materials which have
the same NC size but varying loading amount could be easily
obtained by varying the initial NC concentration. This result
is difficult to achieve using conventional impregnation-
calcination methods especially at high weight loading. Fig-
ure S11 shows TEM images of Au (10 wt.%)/TiO2 prepared
by using a deposition-precipitation method and an incipient
wetness impregnation method where non-uniform deposition
and very large Au NCs were observed.[17] However, firstly, in
a tested loading range of Au NCs from low (1.0 wt.%) to high
(10.0 wt.%), homogeneous distribution of NCs on APD-
modified TiO2 was obtained without clustering (Figure 4e–h),
which is an important requirement for improved material
properties such as maximized surface area of active sites.
While the adsorption conditions remained the same, the
supernatants were obviously colored when bare TiO2 was
used (insets of Figure 4e–h). The APD-functionalized oxides
could indeed benefit from higher adsorption affinity in
preparing supported NCs (Figure 3 and Table S3). Secondly,
Figure 4i–l shows a series of Pd NCs successfully deposited on
Figure 3. Adsorption isotherms for Au NCs (4.5 nm, squares), Pt NCs (5.7 nm, circles), and Pd NCs (6.0 nm, triangles) on a) APTS-TiO2, b) APD-
TiO2, and c) bare TiO2 in 10 vol.% ethanol/toluene based on surface coverage. The solid lines are isotherms fitted using the Langmuir-Freundlich
model. The dashed lines indicate the surface coverage at which the target NCs loading were fully adsorbed.
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979
ꢀ 2021 Wiley-VCH GmbH
www.angewandte.org
7975

Angewandte
Research Articles
Chemie
Figure 4. TEM images of supported NCs. (a–d) extra-high coverage of NC adsorbed on APTS-functionalized supports. a) Pt NC (5.7 nm)/SiO2 gel,
and b) Pd NC (3.8 nm), c) Pd NC (6.0 nm), and d) Pd NC (8.3 nm) on SiO2 sphere coated with an APTS-modified mesoporous silica layer. e–
x) NCs adsorbed on APD-functionalized supports. e) Au NC (4.5 nm, 1.0 wt.%)/TiO2, f) Au NC (4.5 nm, 2.5 wt.%)/TiO2, g) Au NC (4.5 nm,
5.0 wt.%)/TiO2, h) Au NC (4.5 nm, 10.0 wt.%)/TiO2, i) Pd NC (3.8 nm, 2.0 wt.%)/g-Al2O3, j) Pd NC (6.0 nm, 2.0 wt.%)/g-Al2O3, k) Pd NC (8.3 nm,
2.0 wt.%)/g-Al2O3, l) Pd NC (13.7 nm, 2.0 wt.%)/g-Al2O3, m) Ni NC (9.2 nm, 2.0 wt.%)/SiO2 gel, n) Pt NC (2.4 nm, 2.0 wt.%)/SiO2 gel, o) Ru NC
(3.0 nm, 1.5 wt.%)/SiO2 gel, p) Ag NC (5.2 nm, 2.0 wt.%)/SiO2 gel, q) Pd NC (6.0 nm, 4.5 wt.%)/CeO2, r) Pd NC (6.0 nm, 4.5 wt.%)/TiO2, s) Pd
NC (6.0 nm, 4.5 wt.%)/NiO, t) Pd NC (6.0 nm, 4.5 wt.%)/g-Al2O3, u) Pd NC (6.0 nm, 4.5 wt.%)/b-zeolite, v) Pd NC (6.0 nm, 4.5 wt.%)/SiO2 gel,
w) Pd NC (6.0 nm, 4.5 wt.%)/V2O5, and x) Pd NC (6.0 nm, 4.5 wt.%)/CuO. Insets: pictures of centrifuged vials after adsorption of NC on the
amine-functionalized supports. The insets of (e)–(h) include both the APD-modified TiO2 (left) and bare TiO2 (right).
7976 www.angewandte.org
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979

Angewandte
Research Articles
Chemie
APD-modified g-Al2O3 (2.0 wt.%) with varying particle sizes
light-off curves of an APD-functionalized catalyst overlapped
ranging between 3.8 and 13.7 nm. Other metal NCs with
desired sizes and weight loading including Ni, Pt, Ru, and Ag
deposited on APD-modified amorphous SiO2 are shown in
Figure 4m–p. Finally, the deposition of 4.5 wt.% Pd NCs was
demonstrated on various kinds of commonly used oxides with
APD-modification including semiconducting, insulating, re-
ducible and non-reducible oxides and zeolites (Figure 4q–x).
Therefore, we show that this method can disperse particles of
different sizes, compositions, loadings, and surface ligands
onto several oxide supports functionalized with amino group
(Table S4). This robust NC deposition approach relies on (1)
the suitable solvent mixture (10 vol.% ethanol/toluene) in
which oxides achieve good dispersibility without destabilizing
the NCs; (2) strong binding of APD to oxide surfaces during
NC adsorption; and (3) strong affinity of amino groups on the
oxide surface for metal NCs. Furthermore, the method is
applicable to NCs with different ligand coatings.[18]
The use of APD rather than silanes for surface modifi-
cation is key to avoiding surface contamination of oxide
supports. This element is required in several applications and
particularly crucial in the preparation of heterogeneous
catalysts, where metal-support interactions determine catalyst
performance.[15b,19] Several catalysts made from bare, APTS-
modified, and APD-modified oxides were tested for three
different reactions (CO oxidation, Figure 5a; methane cata-
lytic combustion, Figure 5b; and hydrogen production from
glycerol photoreforming, Figure 5c) where metal-support
interactions are crucial to determine catalytic activity and
where silica-based supports are poorly active. The function-
alized ligands are removed after NC deposition using
a thermal treatment at 4508C for 30 minutes (Figure S4). In
CO oxidation (Figure 5a), support functionalized by APTS or
APD only barely influenced the catalytic activity of deposited
Pd NCs, suggesting that in both cases the functional groups do
not hinder catalytic activity. However, in catalytic methane
combustion, even a thin layer of Si from APTS drastically
decreases the activity of the deposited Pd NCs as shown by
the drastic increase in the light-off temperature by nearly
2008C, indicating that the interface between metal and oxide
support is very crucial (Figure 5b). The APD modification
does not change the support properties because the organics
are completely removed by a thermal treatment, and thus the
with the one where the bare support was used for deposition.
In photocatalysis, silica residue from thermally treated,
APTS-modified oxides results in an insulating layer which
decreases the electron transfer rate between titania and metal
NCs, leading to decreased hydrogen production rate for this
sample (Figure 5c). Again, with the APD-modified TiO2, the
hydrogen production rate was very similar to that of bare
TiO2. Therefore, APD modification allows to obtain a uniform
and high loading of metal NCs on many oxide supports
without compromising the properties and catalytic perfor-
mance of the final materials. In addition, the thermal stability
of the high weight loading catalysts prepared by this general
approach was evaluated on supported Au NCs (6.0 nm,
10.0 wt.%)/APD-TiO2 by analyzing the particle size distribu-
tion after treatments at 4508C in air between 0.5 and 24 h
(Figure S12). Although the particle size inevitably increased
(6.8 nm), the Au catalyst showed good thermal stability up to
24 h considering the low Tamman temperature of Au
( ꢀ 4008C).
The performance of catalysts made with a high weight
loading of colloidal NCs on oxide supports were further
studied to demonstrate the advantage of this method
compared to conventional impregnation approaches. In
conventional impregnation, high weight loadings unavoidable
lead to large supported particles and polydisperse size
distributions. As a proof of concept, the relationship between
NCs distribution and catalytic conversion as a function of
temperature at a high weight loading was demonstrated on
Pd(10 wt.%)/g-Al2O3 catalysts for methane catalytic combus-
tion prepared by the APD-assisted approach and three other
methods (Figure 6). In a first sample, Pd NCs were deposited
on alumina in the absence of ethanol and using pure hexanes;
the catalyst showed clustered Pd NCs due to the poor
dispersity of oxide support (Figure 6b and Figure S13a–c). In
a second sample, the NCs were deposited in an ethanol/
toluene mixture to increase alumina dispersion, but the low
adsorption capacity of bare g-Al2O3 could not allow full
uptake of 10 wt.% of Pd NCs and thus the Pd NCs were
deposited by evaporation of the solution. This option also
resulted in agglomeration of Pd NCs (Figure 6c and Fig-
ure S13d–f). These two catalysts showed poor catalytic
activity with full conversion of methane achieved at
Figure 5. Catalytic activities of NCs on bare (circles), APTS-functionalized (squares), and APD-functionalized (triangles) oxide supports. a) Pd
(6.0 nm, 0.5 wt.%)/CeO2 for CO oxidation, b) Pd (3.0 nm, 2.0 wt.%)/g-Al2O3 for methane catalytic combustion, and c) Pt (2.4 nm, 1.0 wt.%)/TiO2
for hydrogen production from glycerol photoreforming.
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979
ꢀ 2021 Wiley-VCH GmbH
www.angewandte.org
7977

Angewandte
Research Articles
Chemie
Figure 6. a) Catalytic conversion as a function of temperature and (b)–(e) the corresponding TEM images of several Pd(10 wt.%)/g-Al2O3
catalysts for methane catalytic combustion. The catalysts were prepared by adsorption of pre-synthesized Pd NCs (8.3 nm) in hexanes with bare
g-Al2O3 (b, squares), in 10 vol.% ethanol/toluene mixture with bare g-Al2O3 assisted by evaporation (c, circles), in 10 vol.% ethanol/toluene
mixture with APD-functionalized g-Al2O3 (d, triangles), and Pd(10 wt.%)/g-Al2O3 fabricated by a conventional impregnation method (e,
rhombuses).
ꢀ 5008C for the hexanes-prepared sample, and not even Conclusion
reaching full conversion in the case of the sample prepared
from toluene/ethanol. On the contrary, the catalyst prepared
using the APD-functionalized g-Al2O3 in ethanol/toluene
mixture showed Pd NCs that were uniformly distributed on
the support surface without obvious agglomeration after
removing the organic ligands (Figure 6d and Figure S13g–i).
This catalyst demonstrated outstanding performance with full
conversion of methane achieved at ꢀ 4258C under the
conditions of our tests that include steam in the feed. The
catalyst obtained by the APD-assisted approach also showed
much superior performance compared with a catalyst pre-
pared by a conventional impregnation method from PdII salt
(Figure 6e and Figure S13j–l). Despite the presence of Pd
particles of similar size to the APD sample, at this high weight
loading the conventional impregnation method again inevi-
tably produced sintered and large Pd particles (inset of
Figure 6e). As a result, its activity was not favorable and the
catalyst did not reach full methane conversion by 5008C.
Kinetic studies further exhibited a similar turnover frequency
among these catalysts indicating that the difference in
catalytic performance was mainly caused by the difference
in metal surface area (Figure S14). This result underlines that
it is important to control the NCs distribution on oxide
supports and that the performance of the catalyst at a high
metal loading was successfully optimized through controlling
the interfacial chemistry between NCs and oxide supports.
After 1 h of catalytic reaction at 4008C, the catalyst prepared
by the APD-functionalized support still showed uniformly
distributed NCs without obvious sintering (Figure S15c),
while sintered and large particles were observed in other
catalysts (Figure S15a, b, d). The developed APD-function-
alized strategy for the adsorption of hydrophobic ligand-
stabilized NCs is a reliable approach for the preparation of
supported catalysts with a wide loading range and high
stability and activity.
In summary, we demonstrated a general and robust
method for the deposition of hydrophobic ligand-stabilized
colloidal NCs on oxide support materials. Through function-
alization with a primary amine, we showed that amino groups
on the oxide surface are able to strongly bind metal NCs,
leading to uniform monolayer deposition at high weight
loadings. Adsorption isotherms clearly quantify the important
role played by amino groups in the monolayer deposition of
metal NCs on oxides. The development of removable
modification of oxide supports with 3-amino-1,2-propanediol
(APD) guarantees NC adsorption up to very high weight
loadings (10 wt.%) without affecting the oxide support
properties after a thermal treatment. The tunable loading
and homogeneously distributed NCs on oxide supports are
excellent materials in heterogeneous catalysis, showing no
degradation of catalytic properties after functionalization and
thus opening the opportunity to extend the use of colloidal
NCs in heterogeneous catalysts.
Acknowledgements
This work was supported by the TomKat Center at Stanford
University through a Seed Grant. This research was also
supported by the Taiwan Ministry of Science and Technology
through the Stanford-Taiwan LEAP Program administered
by NARLabs. Part of this work was performed at the Stanford
Nano Shared Facilities (SNSF), supported by the National
Science Foundations under award ECCS-1542152. M.C.
acknowledges support from the School of Engineering at
Stanford University and from a Terman Faculty Fellowship.
K.-C. K. thanks the National Applied Research Laboratories
of Taiwan for the visiting scholar fellowship.
7978 www.angewandte.org
ꢀ 2021 Wiley-VCH GmbH
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979

Angewandte
Research Articles
Chemie
Huang, E. D. Goodman, C. J. Wrasman, A. Holm, A. R. Riscoe,
Conflict of interest
J. A. Schwalbe, M. Cargnello, Nano Today 2019, 24, 15 –47; c) J.
Huang, R. Buonsanti, Chem. Mater. 2019, 31, 13 –25; d) E. D.
Goodman, A. C. Johnston-Peck, E. M. Dietze, C. J. Wrasman,
A. S. Hoffman, F. Abild-Pedersen, S. R. Bare, P. N. Plessow, M.
Cargnello, Nat. Catal. 2019, 2, 748 –755; e) K. Na, Q. Zhang,
G. A. Somorjai, J. Cluster Sci. 2014, 25, 83 –114.
The authors declare no conflict of interest.
Keywords: adsorption ·heterogeneous catalysis ·nanocrystals ·
oxide supports ·surface functionalization
[8] a) C. Tian, Y. Deng, D. Zhao, J. Fang, Adv. Opt. Mater. 2015, 3,
404 –411; b) Y.-W. Wang, K.-C. Kao, J.-K. Wang, C.-Y. Mou, J.
Phys. Chem. C 2016, 120, 24382 –24388.
[9] a) A. Dokoutchaev, J. T. James, S. C. Koene, S. Pathak, G. K. S.
Prakash, M. E. Thompson, Chem. Mater. 1999, 11, 2389 –2399;
b) J.-D. Grunwaldt, C. Kiener, C. Wçgerbauer, A. Baiker, J.
Catal. 1999, 181, 223 –232; c) K. Mori, A. Kumami, M. Tomo-
nari, H. Yamashita, J. Phys. Chem. C 2009, 113, 16850 –16854;
d) G. Marzun, C. Streich, S. Jendrzej, S. Barcikowski, P.
Wagener, Langmuir 2014, 30, 11928 –11936.
[10] a) A. Dong, X. Ye, J. Chen, Y. Kang, T. Gordon, J. M. Kikkawa,
C. B. Murray, J. Am. Chem. Soc. 2011, 133, 998 –1006; b) E. L.
Rosen, R. Buonsanti, A. Llordes, A. M. Sawvel, D. J. Milliron,
B. A. Helms, Angew. Chem. Int. Ed. 2012, 51, 684 –689; Angew.
Chem. 2012, 124, 708 –713.
[11] N. Zheng, G. D. Stucky, J. Am. Chem. Soc. 2006, 128, 14278 –
14280.
[12] M. Casavola, J. Hermannsdçrfer, N. de Jonge, A. I. Dugulan,
K. P. de Jong, Adv. Funct. Mater. 2015, 25, 5309 –5319.
[13] Q. Fan, K. Liu, Z. Liu, H. Liu, L. Zhang, P. Zhong, C. Gao, Part.
Part. Syst. Charact. 2017, 34, 1700075.
[14] a) S. L. Westcott, S. J. Oldenburg, T. R. Lee, N. J. Halas, Lang-
muir 1998, 14, 5396 –5401; b) Z.-J. Jiang, C.-Y. Liu, L.-W. Sun, J.
Phys. Chem. B 2005, 109, 1730 –1735; c) J. Kim, J. E. Lee, J. Lee,
Y. Jang, S.-W. Kim, K. An, J. H. Yu, T. Hyeon, Angew. Chem. Int.
Ed. 2006, 45, 4789 –4793; Angew. Chem. 2006, 118, 4907 –4911.
[15] a) M. Haruta, Catal. Today 1997, 36, 153 –166; b) M. Cargnello,
V. V. T. Doan-Nguyen, T. R. Gordon, R. E. Diaz, E. A. Stach,
R. J. Gorte, P. Fornasiero, C. B. Murray, Science 2013, 341, 771 –
773.
[16] a) C. Xu, K. Xu, H. Gu, R. Zheng, H. Liu, X. Zhang, Z. Guo, B.
Xu, J. Am. Chem. Soc. 2004, 126, 9938 –9939; b) H. Lee, S. M.
Dellatore, W. M. Miller, P. B. Messersmith, Science 2007, 318,
426 –430; c) M. S. Ata, Y. Liu, I. Zhitomirsky, RSC Adv. 2014, 4,
22716 –22732.
[1] a) M. V. Kovalenko, L. Manna, A. Cabot, Z. Hens, D. V. Talapin,
C. R. Kagan, V. I. Klimov, A. L. Rogach, P. Reiss, D. J. Milliron,
P. Guyot-Sionnnest, G. Konstantatos, W. J. Parak, T. Hyeon,
B. A. Korgel, C. B. Murray, W. Heiss, ACS Nano 2015, 9, 1012 –
1057; b) R. W. Murray, Chem. Rev. 2008, 108, 2688 –2720.
[2] a) S.-W. Hsu, A. L. Rodarte, M. Som, G. Arya, A. R. Tao, Chem.
Rev. 2018, 118, 3100 –3120; b) J.-F. Li, Y.-J. Zhang, S.-Y. Ding, R.
Panneerselvam, Z.-Q. Tian, Chem. Rev. 2017, 117, 5002 –5069;
c) K. Watanabe, D. Menzel, N. Nilius, H.-J. Freund, Chem. Rev.
2006, 106, 4301 –4320.
[3] a) I. Gur, N. A. Fromer, M. L. Geier, A. P. Alivisatos, Science
2005, 310, 462 –465; b) S. A. McDonald, G. Konstantatos, S.
Zhang, P. W. Cyr, E. J. D. Klem, L. Levina, E. H. Sargent, Nat.
Mater. 2005, 4, 138 –142; c) J. M. Luther, M. Law, M. C. Beard,
Q. Song, M. O. Reese, R. J. Ellingson, A. J. Nozik, Nano Lett.
2008, 8, 3488 –3492; d) A. M. Smith, S. Nie, Nat. Biotechnol.
2009, 27, 732 –733; e) S. Wu, G. Han, D. J. Milliron, S. Aloni, V.
Altoe, D. V. Talapin, B. E. Cohen, P. J. Schuck, Proc. Natl. Acad.
Sci. USA 2009, 106, 10917 –10921.
[4] a) N. L. Rosi, C. A. Mirkin, Chem. Rev. 2005, 105, 1547 –1562;
b) P. K. Jain, K. S. Lee, I. H. El-Sayed, M. A. El-Sayed, J. Phys.
Chem. B 2006, 110, 7238 –7248; c) S. E. Lohse, C. J. Murphy, J.
Am. Chem. Soc. 2012, 134, 15607 –15620.
[5] a) A. T. Bell, Science 2003, 299, 1688 –1691; b) R. Narayanan,
M. A. El-Sayed, J. Phys. Chem. B 2005, 109, 12663 –12676; c) K.
An, G. A. Somorjai, ChemCatChem 2012, 4, 1512 –1524; d) Y.
Kang, P. Yang, N. M. Markovic, V. R. Stamenkovic, Nano Today
2016, 11, 587 –600; e) M. Cargnello, Chem. Mater. 2019, 31, 576 –
596.
[6] a) C. B. Murray, D. J. Norris, M. G. Bawendi, J. Am. Chem. Soc.
1993, 115, 8706 –8715; b) S. Sun, C. B. Murray, D. Weller, L.
Folks, A. Moser, Science 2000, 287, 1989 –1992; c) J. Park, K. An,
Y. Hwang, J.-G. Park, H.-J. Noh, J.-Y. Kim, J.-H. Park, N.-M.
Hwang, T. Hyeon, Nat. Mater. 2004, 3, 891 –895; d) J. A. Dahl,
B. L. S. Maddux, J. E. Hutchison, Chem. Rev. 2007, 107, 2228 –
2269; e) J. Park, J. Joo, S. G. Kwon, Y. Jang, T. Hyeon, Angew.
Chem. Int. Ed. 2007, 46, 4630 –4660; Angew. Chem. 2007, 119,
4714 –4745; f) Y. Xia, Y. Xiong, B. Lim, S. E. Skrabalak, Angew.
Chem. Int. Ed. 2009, 48, 60 –103; Angew. Chem. 2009, 121, 62 –
108; g) J. Yang, E. Sargent, S. Kelley, J. Y. Ying, Nat. Mater. 2009,
8, 683 –689; h) C. B. Murray, S. Sun, H. Doyle, T. Betley, MRS
Bull. 2001, 26, 985 –991.
[17] L. Delannoy, N. El Hassan, A. Musi, N. N. Le To, J.-M. Krafft, C.
Louis, J. Phys. Chem. B 2006, 110, 22471 –22478.
[18] S. Y. Moon, B. Naik, C.-H. Jung, K. Qadir, J. Y. Park, Catal.
Today 2016, 265, 245 –253.
[19] T. W. van Deelen, C. Hernµndez Mejía, K. P. de Jong, Nat. Catal.
2019, 2, 955 –970.
Manuscript received: December 30, 2020
Accepted manuscript online: January 6, 2021
Version of record online: February 24, 2021
[7] a) A. Aitbekova, C. J. Wrasman, A. R. Riscoe, L. Y. Kunz, M.
Cargnello, Chin. J. Catal. 2020, 41, 998 –1005; b) P. Losch, W.
Angew. Chem. Int. Ed. 2021, 60, 7971 –7979
ꢀ 2021 Wiley-VCH GmbH
www.angewandte.org 7979