The azide-alkyne cycloaddition catalysed by transition metal oxide nanoparticles

Article abstract of DOI:10.1039/c9nj04690a

Colloidal nanoparticles of Earth-abundant, first-row transition metal oxides and sulfide, namely magnetite (Fe₃O₄), manganese and cobalt ferrite, (MnFe₂O₄, CoFe₂O₄), manganese(ii) oxide (Mn

Full text of DOI:10.1039/c9nj04690a

NJC
View Article Online
View Journal | View Issue
PAPER
The azide–alkyne cycloaddition catalysed by
transition metal oxide nanoparticles†
Cite this: New J. Chem., 2019,
43, 18049
b
c
Giorgio Molteni,*a Anna M. Ferretti,
Mario Italo Trioni,c Fausto Cargnoni
b
and Alessandro Ponti
*
Colloidal nanoparticles of Earth-abundant, first-row transition metal oxides and sulfide, namely magnetite
(Fe3O4), manganese and cobalt ferrite, (MnFe2O4, CoFe2O4), manganese(II) oxide (MnO) and sulfide (a-MnS),
were used as catalysts in the cycloaddition between azides and methyl propiolate. The presence of these
nanoparticles allowed us to carry out the cycloadditions under milder conditions and with a regioselectivity
comparable to the classic ‘‘metal-free’’ thermal processes. Ferrite nanoparticles gave higher conversion than
MnO and a-MnS nanoparticles. The feasibility of the cycloaddition onto 1,2-disubstituted acetylenes was also
proved. Ferrite nanocatalysts could be magnetically recovered and reused without significant loss of catalytic
activity. Density functional theory (DFT) calculations support a mechanistic hypothesis that attributes the
increased cycloaddition rate to the adsorption of the azide onto to the nanocatalyst surface.
Received 12th September 2019,
Accepted 28th October 2019
DOI: 10.1039/c9nj04690a
rsc.li/njc
prominent position at the interface between chemistry, biology,
and materials science.15,16 As examples, some derivatives of 1,2,3-
1. Introduction
Catalysis by inorganic nanoparticles (NPs) is a very rapidly triazoles are effective drugs against epilepsy17 and diabetes,18 and
growing field1 since catalytically active NPs may combine the 1,2,3-triazolium-containing ionic liquids containing mixed polymeric
advantages and overcome the shortcomings of the homogeneous materials are of great interest as immobilised catalysts.19
and heterogeneous approaches to catalysis.2 NP catalysts are
In 2006, some of us introduced the use of colloidal Cu@Cu2O|
robust, easily recoverable and recyclable, like heterogeneous CuO NPs as effective catalysts for the cycloaddition of organic
catalysts. Colloidal NPs – as opposed to supported NPs – are well azides and terminal alkynes to 1,2,3-triazoles.20 Several examples
dispersed within the reaction medium, thus increasing the followed,21–23 which involved colloidal Cu,24,25 Fe@Cu,26
available catalytic surface and improving the diffusion of reactants Cu2O,27,28 CuO,29 and CuFe2O430,31 NPs, in some cases extending
and products, though they cannot achieve the unimpeded mass this approach to aqueous solvents.24,26,27,29,31 In these reactions,
transfer and high accessibility of active sites typical of homo- the 4-substituted triazole is formed and internal alkynes do not
geneous catalysts.
react, as in the homogeneous copper-catalysed ‘click’ reaction.
The choice route to the 1,2,3-triazole ring is provided by the The NP-catalysed mechanism was shown to be similar32 to that of
azide–alkyne cycloaddition.3 Although such a reaction dates to the homogeneous reaction.33 Ag2O NPs also catalyse this
the end of the 19th century,4 its mechanism and synthetic cycloaddition but the mechanism seems different as electron-
applications were disclosed by Huisgen in the early 1960s.5–7 poor azides gave a mixture of the 4- and 5-substituted isomers.34
Forty years later, the concept of ‘‘click’’ reaction allowed for the
We then turned our attention to cheaper metals as possible
first time the fully regioselective cycloaddition to 4-substituted- nanocatalysts for the azide–alkyne cycloaddition. We selected
1,2,3-triazoles.8,9 Quite recently, novel homogeneous metallorganic in particular iron as an abundant, inexpensive, environmentally
catalysts accomplished the regioselective synthesis of the friendly and biocompatible metal and focused on its oxide
5-substituted, complementary isomer.10–14 Due to its robustness, Fe3O4 (magnetite) for its robustness, stability and magnetization,
it is fair to say that the azide–alkyne cycloaddition has assumed a which makes magnetite nanocatalysts easy to recover by an
external magnet.35 Colloidal iron oxide NPs catalyse a number of
organic reactions, mainly oxidations1,35 but also polymerization,36
a
`
Dipartimento di Chimica, Universita degli Studi di Milano, via C. Golgi 19, 20133
Milano, Italy. E-mail: giorgio.molteni@unimi.it
isomerization,37,38 hydrogenation,39 alkylation and alkenylation,1
and reactions involving the oxidative activation of the C–H bond.35
We have recently shown that magnetite NPs are good catalysts
for the cycloaddition of nitrilimines to alkenes, alkynes, and
activated nitriles.40 To widen the scope of our investigation
b Laboratorio di Nanotecnologie, Istituto di Scienze e Tecnologie Molecolari (ISTM),
Consiglio Nazionale delle Ricerche, via G. Fantoli 16/15, 20138 Milano, Italy.
E-mail: alessandro.ponti@istm.cnr.it
c Istituto di Scienze e Tecnologie Molecolari (ISTM), Consiglio Nazionale delle
Ricerche, via C. Golgi 19, 20133 Milano, Italy
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9nj04690a about azide–alkyne cycloaddition, we also considered the iron
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18049

View Article Online
Paper
NJC
neighbours manganese and cobalt, despite their slightly higher a-MnS nanoparticles is as follows. In a 25 ml two-necked
impact on health and environment. This, in addition to magnetite round-bottom flask, equipped with condenser and thermo-
NPs, we considered NPs made of either mixed ferrite (MnFe2O4, couple, Mn2(CO)10 (104 mg, 0.267 mmol), stearic acid (StAc, 448
CoFe2O4) and manganese(II) oxide (MnO) and sulfide (a-MnS). The mg, 1.57 mmol), and elemental sulfur (34 mg, 1.1 mmol) were
Mn and Co ferrite NPs have found use as catalysts for organic dissolved in 1.1 ml of octadec-1-ene (ODE) under stirring and
reactions.41,42 Here we show that colloidal magnetite NPs – along argon atmosphere. The solution was heated to 320 1C (heating
with Mn and Co ferrite, Mn(II) oxide and sulfide NPs – are effective rate 10 1C minꢀ1) and aged at that temperature for 1 h. After
catalysts for the cycloaddition between organic azides and both cooling at RT, the solid reaction crude was dissolved in about
terminal and internal alkynes.
3 ml of toluene and added with 12 ml of ethanol. The nano-
particles were collected by centrifugation (3400g, 10 min). Next,
the nanoparticles were washed with ethanol (5%) and acetone
(ꢁ2) and collected by centrifugation (3400g, 10 min). Extensive
washing is required to eliminate the excess stearic acid. The
resulting StAc-coated MnO nanoparticles were dispersed in
2. Experimental section
Synthesis of iron oxide (Fe3O4) nanoparticles
Fe3O4 nanoparticles were synthesized by modification of a
reported procedure.43 In a 50 ml three-necked round-bottom
flask, equipped with condenser, thermocouple, and rubber
septum, oleic acid (OlAc, 2.4 ml, 7.5 mmol) was dissolved in
10 ml of octadec-1-ene (ODE) under stirring and argon atmo-
sphere. The solution was heated to 105 1C and degassed three
times by vacuum-argon cycles. After 40 min. Fe(CO)5 (330 ml,
2.5 mmol) was injected through the rubber septum. The reaction
mixture was heated to 320 1C (heating rate 15 1C minꢀ1) and aged
at that temperature for 3 h. After cooling at RT, the reaction crude
was precipitated with acetone and the nanoparticles were collected
by centrifugation (3400g, 10 min). The nanoparticles were repeatedly
washed with acetone and collected by centrifugation (3400g,
10 min). The resulting OlAc-coated NPs were dispersed in
n-hexane at a concentration of 1.21 gMn l
ꢀ1. To obtain a-MnS
nanoparticles, the above procedure was followed dissolving
Mn2(CO)10 (107 mg, 0.273 mmol), stearic acid (StAc, 298 mg,
1.05 mmol), and elemental sulfur (68 mg, 2.1 mmol) in 1.1 ml
of ODE. The resulting StAc-coated a-MnS nanoparticles were
dispersed in n-hexane at a concentration of 1.05 gMn lꢀ1
.
Cycloaddition between azides 1 and methyl propiolate in the
presence of CuI. General procedure
A solution of azide 1 (0.5 mmol) and methyl propiolate (84 mg,
1.0 mmol) in dry toluene (2.0 ml) was added with CuI (190 mg,
1.0 mmol). The heterogeneous mixture was stirred at 20 1C and
the reaction was monitored with periodic TLC (hexane/ethyl
acetate 7 : 3). After 9 h (4a, Rf = 0.35), 18 h (4b, Rf = 0.13) or 6 h
(4c, Rf = 0.28), the undissolved material was filtered over a Celite
pad and washed with acetone (3 ꢁ 2 ml). The collected solvents
were evaporated under reduced pressure and the residue was
crystallized with diisopropyl ether affording 4-substituted-1,2,3-
triazoles 4a–c.
toluene at a concentration of 3.22 gFe
was 58% with respect to Fe(CO)5.
l
ꢀ1. The isolated yield
Synthesis of mixed ferrite (MFe2O4, M = Mn, Co) nanoparticles
MFe2O4 nanoparticles were synthesized following a reported
procedure44 except for the use of 1,2-hexadecanediol instead of
1,2-tetradecanediol. The molar ratio M(acac)2 : Fe(acac)3 was
1: 2 (acac = acetylacetonate). In a 100 ml three-necked round-
bottom flask, equipped with condenser, thermocouple, and
rubber septum, oleic acid (OlAc, 1.0 ml, 3.15 mmol), oleylamine
(1.0 ml, 3.04 mmol), Fe(acac)3 (0.350 g, 0.990 mmol), and
M(acac)2 (0.495 mmol, M = Mn, Co) were dissolved in 25 ml of
dibenzylether under stirring and argon atmosphere. The reaction
mixture was rapidly heated to 120 1C and held at that tempera-
ture for 30 min under vacuum and 30 min under argon. Next, it
was heated to 210 1C (heating rate: 8 1C minꢀ1) and held at that
temperature for 2 h under argon. Finally, the reaction mixture
was heated to 300 1C (heating rate: 3 1C minꢀ1) and aged at that
temperature for 1 h. After the high temperature synthesis, the
nanoparticles were washed 5 times with acetone/ethanol mixture
and collected by centrifugation (3400g, 10 min). The resulting
nanoparticles were dispersed in toluene at a concentration
2.16 gFe lꢀ1 (MnFe2O4) and 2.31 gFe lꢀ1 (CoFe2O4). The isolated
yield was about 30% with respect to Fe(acac)3.
Uncatalysed cycloaddition between azides 1 and methyl
propiolate 2. General procedure
A solution of azide 1 (0.5 mmol) and methyl propiolate (67 mg,
0.8 mmol) in dry toluene (2.0 ml) was stirred at the temperatures
and the times listed in the Tables 3 and 4. The reactions were
monitored by TLC (hexane/ethyl acetate 7 : 3). The solvent was
removed under reduced pressure and the residue was chromato-
graphed on a silica gel column with hexane/ethyl acetate 7:3.
In the case of the reaction carried out at 20 1C (Table 3, entry 1),
unchanged 1a was eluted first (83 mg, 55%). Further elution gave
5-substituted-1,2,3-triazole 5a (Rf = 0.46) and 4-substituted-1,2,3-
triazole 4a (44 mg, 23%).
In the case of the reaction carried out at 75 1C (Table 4, entry 1),
5b (Rf = 0.37) was eluted first, followed by 4b (67 mg, 64%).
In the case of the reaction carried out at 90 1C (Table 4, entry 7),
5c (Rf = 0.56) was eluted first, followed by 4c (75 mg, 74%).
Cycloaddition between azides 1 and methyl propiolate 2 in the
presence of Fe3O4 nanoparticles
Synthesis of manganese(II) oxide (MnO) and manganese(II)
sulfide (a-MnS) nanoparticles
A solution of azide 1 (0.5 mmol), methyl propiolate (67 mg,
MnO and a-MnS nanoparticles were synthesized adapting 0.8 mmol) and Fe3O4 nanoparticles (1 ml, 58 mmol of Fe) in
reported procedures.45 The actual procedure for MnO and toluene (1.0 ml) was stirred at the temperatures and the times
18050 | New J. Chem., 2019, 43, 18049--18061 This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019

View Article Online
NJC
Paper
listed in the Tables 3 and 4. The reaction crude was filtered over Uncatalysed cycloaddition between azide 1a and DMAD 3
a Celite pad and washed with ethyl acetate (3 ꢁ 3 ml). The
A solution of azide 1a (0.15 g, 0.5 mmol) and DMAD 3 (0.14 g,
collected solvents were evaporated under reduced pressure.
1.0 mmol) in toluene (2.0 ml) was stirred at the temperatures and the
In the case of the reaction carried out at 20 1C (Table 3,
entry 4), the residue was chromatographed on a silica gel
(hexane/ethyl acetate 7 : 3). Evaporation of the solvent gave a residue
column with hexane/ethyl acetate 7 : 3. Unchanged 1a was
times listed in the Table 5. The reactions were monitored by TLC
which was crystallized with diisopropyl ether affording pure 6 (Rf =
0.6), reaction at 20 1C: 0.16 g, 72%; reaction at 95 1C: 0.22 mg, 99%.
eluted first (60 mg, 40%). Further elution gave a mixture of
1,2,3-triazoles 4a and 5a (110 mg, 57%) in the ratio 65 : 35 as
1
determined by H NMR.
Cycloaddition between azide 1a and DMAD 3 in the presence of
Fe3O4 nanoparticles
In the case of the reactions carried out at 45 1C (Table 3,
entry 5 and Table 4, entries 2 and 8), 75 1C (Table 3, entry 6) and
95 1C (Table 3, entry 7) the ratio 4 : 5 was determined by 1H NMR.
A solution of azide 1a (0.15 g, 0.5 mmol), DMAD (0.14 g,
1.0 mmol) and Fe3O4 nanoparticles (0.25 ml, 14 mmol of Fe) in
toluene (1.75 ml) was stirred at the temperatures and the times
listed in the Table 5. Evaporation of the solvent gave a residue
which was crystallized with diisopropyl ether affording pure 6
(reaction at 20 1C: 0.22 g, 99%, reaction at 95 1C: 0.22 mg, 99%).
Cycloaddition between azides 1 and methyl propiolate 2 in the
presence of CoFe2O4 nanoparticles
A solution of azide 1 (0.5 mmol), methyl propiolate (67 mg,
0.8 mmol) and CoFe2O4 nanoparticles (0.66 ml, 27 mmol of Fe,
10 mmol of Co) in toluene (1.34 ml) was stirred at 45 1C for the
times listed in Table 3, entry 8 and Table 4, entries 3 and 9. The
reaction crude was filtered over a Celite pad and washed with
ethyl acetate (3 ꢁ 3 ml). The collected solvents were evaporated
under reduced pressure.
Cycloaddition between azide 1a and DMAD 3 in the presence of
MnFe2O4 nanoparticles
A solution of azide 1a (0.15 g, 0.5 mmol), DMAD (0.14 g,
1.0 mmol) and MnFe2O4 nanoparticles (0.44 ml, 17 mmol of Fe,
6 mmol of Mn) in toluene (1.56 ml) was stirred at 20 1C for 20 h.
Evaporation of the solvent gave a residue which was crystallized
1
The ratio 4 : 5 was determined by H NMR of the residue.
Cycloaddition between azides 1 and methyl propiolate 2 in the with diisopropyl ether affording pure 6 (0.20 g, 91%).
presence of MnFe2O4 nanoparticles
Recycling of the Fe3O4 NP catalyst in the cycloaddition between
azide 1c and methyl propiolate 2
A solution of azide 1 (0.5 mmol), methyl propiolate (67 mg,
0.8 mmol) and MnFe2O4 nanoparticles (0.45 ml, 17 mmol of Fe,
The recycling experiments were carried out using dichloro-
methane as a solvent because it is easier to keep the reaction
temperature constant and to isolate products and NPs.
First run. In a 100 ml cylindrical reaction funnel, phenyl-
azide 1c (0.50 g, 4.2 mmol) and methyl propiolate 2 (0.38 g,
4.5 mmol) were dissolved in dry dichloromethane (16.3 ml). Fe3O4
NPs (47 mg, 0.84 mmol of Fe) dispersed in chloroform (4.7 ml)
were added dropwise in 2 min. The mixture was submitted to
vigorous magnetic stirring for 6 h at 40 1C. The undissolved
material was recovered with an external magnet, washed with
dichloromethane (8 ml) and recovered again with an external
magnet. The mother solution was washed with water (3 ꢁ 25 ml),
dried over sodium sulfate and evaporated under reduced pressure.
The crude was crystallised with diisopropyl ether giving a mixture of
the isomeric triazoles 4a + 5a (0.76 g, 89%).
Subsequent runs (2nd–5th). The recovered NPs were dispersed in
dry dichloromethane (21 ml) and amounts of 1c and 2 were added
as in the first run. After 6 h at 40 1C, the isomeric triazoles were
isolated and the NPs recovered and washed as in the first run.
The recycling experiments where the NP mass after each run was
measured were carried out as above except that the recovered NPs
were dried with a rotary pump (0.05 mmHg) for 1 h and weighted
before dispersion in dichloromethane for the next cycle.
6 mmol of Mn) in toluene (1.55 ml) was stirred at 45 1C for the
times listed in Table 3, entry 9 and Table 4, entries 4 and 10.
The reaction crude was filtered over a Celite pad and washed
with ethyl acetate (3 ꢁ 3 ml). The collected solvents were
evaporated under reduced pressure.
1
The ratio 4 : 5 was determined by H NMR of the residue.
Cycloaddition between azides 1 and methyl propiolate 2 in the
presence of MnO nanoparticles
A solution of azide 1 (0.5 mmol), methyl propiolate (67 mg,
0.8 mmol) and MnO nanoparticles (0.45 ml, 10 mmol of Mn) in
toluene (1.55 ml) was stirred at 45 1C for the times listed in
Table 3, entry 10 and Table 4, entries 5 and 11. The reaction
crude was filtered over a Celite pad and washed with ethyl
acetate (3 ꢁ 3 ml). The collected solvents were evaporated
under reduced pressure.
1
The ratio 4 : 5 was determined by H NMR of the residue.
Cycloaddition between azides 1 and methyl propiolate 2 in the
presence of MnS nanoparticles
A solution of azide 1 (0.5 mmol), methyl propiolate (67 mg,
0.8 mmol) and MnS nanoparticles (0.45 ml, 8 mmol of Mn) in
toluene (1.55 ml) was stirred at 45 1C for the times listed in
Table 4, entry 11 and Table 4, entries 6 and 12. The reaction
crude was filtered over a Celite pad and washed with ethyl
Characterization
acetate (3 ꢁ 3 ml). The collected solvents were evaporated Transmission electron microscopy (TEM) images and electron
under reduced pressure.
The ratio 4 : 5 was determined by H NMR of the residue.
diffraction (ED) patterns were collected using a Zeiss LIBRA
200FE microscope. The TEM specimen was prepared by evaporating
1
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18051

View Article Online
Paper
NJC
in air a drop of diluted NP dispersion on a carbon coated copper complexes,48 the LANL2 pseudopotential, and the LANL2DZ
grid. The size distribution of the iron oxide cores was obtained basis set as implemented in the Gaussian09 Suite.49 All geo-
analysing TEM images by the software PEBBLES.51 PEBBLES is metries were fully optimized and harmonic analysis showed
freely available from the authors (http://pebbles.istm.cnr.it). FT-IR that they were true energy minima (no imaginary frequency) or
spectra were collected using a Thermo Nicolet NEXUS 670 FTIR transition states (one imaginary frequency), as required.
spectrometer. The specimen for FTIR was prepared by grinding and
pelleting dry NPs with KBr (NP : KBr 1 : 100 w/w). 1H NMR (300 MHz)
and 13C NMR (75 MHz) spectra were taken with a Bruker Fourier 3. Results and discussion
300 spectrometer (in CDCl3 solutions at room temperature).
Chemical shifts are given as parts per million from tetramethyl-
silane. Coupling constants (J) values are given in hertz and are
3.1 Nanoparticles synthesis and characterization
Transition metal oxides (Fe3O4, MnFe2O4, CoFe2O4, MnO) and
a-MnS NPs were synthesized by the solvothermal decomposition
quoted to ꢂ0.1 Hz consistently with NMR machine accuracy.
of the appropriate precursors as detailed in the Experimental
Element analyses were carried out by a PerkinElmer 2400 series
section. All NPs used in this study consist of a nanocrystal coated
II CHNS/O Analyzer.
with a layer of fatty acid and are colloidally stable in apolar
solvents. The main features of the NPs are collected in Table 1.
The concentration of Fe (ferrites) and that of Mn (MnO and MnS)
Computational methods
Periodic Density Functional (DFT) computations were carried in the NP dispersions were obtained by spectrophotometry50 and
out by the SIESTA 4.0 suite of programs.46 The surface of a magnetite chemical analysis (see ESI†), respectively.
nanoparticle was simulated as a two-dimensional supercell reprodu-
Transmission electron microscopy (TEM) images of as-
cing the stoichiometry of the bulk compound. In the case of the synthesized NPs can be found in Fig. 1(a–e) (wide field TEM
(100) slab the cell contains 62 atoms plus one 1c molecule. In both images can be found in the ESI,† Fig. S1). The size distribution
systems the 1c moiety is adsorbed onto an under-coordinated of the NPs (see ESI,† Fig. S2) was obtained by analysing TEM
surface Fe atom. The surface unit cells was designed such that the images by the software PEBBLES.51 Magnetite NPs are spherical
overall stoichiometry reproduced the oxygen to iron ratio of the bulk and have a very small size dispersion. Manganese (spherical)
material. When necessary, surface atoms were saturated with –H and and cobalt (irregular) ferrite NPs are slightly smaller and have
–OH groups to reproduce their formal oxidation state. DFT compu- larger size dispersion. Both are slightly deficient in the divalent
tations were conducted with the PBE exchange and correlation metal, as evidenced by electron energy loss spectroscopy
functionals. In accordance with literature results,47 we included also (EELS). Their composition can be expressed as Co0.78Fe2.22O4
a U term for Fe atoms, amounting to 4 eV, to reproduce at best the and Mn0.77Fe2.23O4. Octahedral MnO NPs are similar in size to
DOS of bulk Fe3O4. All atoms except hydrogen were assigned a split- Fe3O4 NPs but have larger size dispersion. a-MnS NPs are much
valence basis set of double zeta quality for valence electrons, while larger and have spheroidal shape. Electron diffraction (Fig. 2)
norm-conserving Troullier–Martins pseudopotentials were adopted allowed us to identify the nanocrystals as cubic ferrites (spinel
to describe the core electrons. The reciprocal space was sampled structure) Fe3O4, CoFe2O4, MnFe2O4 and as rock-salt structure
with a 8 ꢁ 8 k-points grid. Convergence of the wavefunction was MnO and a-MnS. It should however be noted that the poor
facilitated by imposing an electronic temperature of 300 K within a resolution of the diffraction pattern prevented us to distinguish
Fermi–Dirac occupation statistics. All geometries were fully between magnetite (Fe3O4) and maghemite (g-Fe2O3), so that
optimized. Fig. 7 compares energies belonging to different partial oxidation of magnetite NPs cannot be excluded.
computations, i.e. having different vacuum energy levels, and
The FT-IR spectra of the NPs confirmed that they are coated
hence the DOS of 1c and energy levels of 2 were aligned so that with a layer of fatty acid in the carboxylate form. The spectra of
the distance between the lowest eigenvalues shown in the figure Fe3O4 and a-MnS NPs are shown in Fig. 3 as examples of NPs
reproduces the value obtained in a gas phase computation coated with oleic and stearic acid, respectively. Both spectra
including both molecules in a non-interacting configuration.
display the characteristic peaks of fatty acid anions: the C–H
Molecular DFT calculations related to the cycloaddition of 2 stretching vibration of CH2 and CH3 groups (2956–2851 cmꢀ1
)
and 1c, 1c-Fe(OH)3 and 1c-Fe2(OH)4 were carried using the and the symmetric and asymmetric stretches of the COOꢀ
OPBE functional, which was shown to perform well with Fe(III) group (ca. 1550 and 1430 cmꢀ1). The weak peak at 3004 cmꢀ1
Table 1 Main features of as-synthesized metal oxide and sulfide nanoparticles
a
b
NP
CFe (g lꢀ1
)
CM (g lꢀ1
)
Dispersion solvent
Coating
hdi
sd
sd/hdi (%)
Fe3O4
CoFe2O4
MnFe2O4
MnO
a-MnS
3.22
2.31
2.16
—
—
Toluene
Toluene
Toluene
n-Hexane
n-Hexane
Oleic acid
Oleic acid
Oleic acid
Stearic acid
Stearic acid
11.9
10
8.6
11.6
44
0.6
2.0
1.2
1.6
5
20
14
14
27
c
0.85 (Co)
0.73 (Mn)
1.21 (Mn)
1.05 (Mn)
—
12
a
b
c
Median equivalent diameter (nm). Diameter standard deviation (nm). CoFe2O4 NPs have elongated shape with major axis = (12 ꢂ 3) nm
(24%), minor axis = (9 ꢂ 2) nm (20%), aspect ratio = (1.3 ꢂ 0.2) (15%).
18052 | New J. Chem., 2019, 43, 18049--18061 This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019

View Article Online
NJC
Paper
As can be seen from Table 3, the ‘‘metal free’’ reaction
required long reaction times and yielded poor reactant conversion
at room temperature (entry 1). By limiting the reaction time to 24 h
at 45 1C, the conversion of the reactants was also low (entry 2). At
higher temperature (entry 3) the reaction proceeded smoothly as
expected for a typical thermal cycloaddition. Similar cycloadditions
display similarly high yield when carried out in water.56 Both
reactant conversion and reaction times were satisfactory in the
presence of Fe3O4 NPs at 45 1C (entry 5), while at room tempera-
ture some amount of unreacted 1a was recovered (entry 4).
At higher temperatures, good results were obtained (entries 6
and 7) although in these cases any catalytic activity of the Fe3O4
NPs can be hardly revealed. Not unexpectedly, the hetero-
geneous reaction mixture in the presence of bulk Fe2O3 did not
produce appreciable results. On the other hand, a homogeneous
solution of iron(III) oleate were also ineffective, probably due to
the crowded, hexacoordinated nature of the iron atom in this
complex. In the presence of cobalt and manganese ferrite NPs
(entries 8 and 9) at 45 1C, results similar to that observed with
Fe3O4 NPs were achieved. As far as manganese oxide and sulfide
nanoparticles are concerned (entries 10–13), their ability to act as
catalysts was poorer in comparison to the ferrite NPs. The ligand
density is low enough to allow the reactants and products to
diffuse to and from the metal oxide surface resulting in a
generally good catalytic activity. The better performance of ferrite
NPs, in particular Fe3O4, must be ascribed to the nature of the
surface catalytic in view of the similar ligand density among the
NPs. In all cases listed in Table 3 the observed regioselectivity,
favouring the 4-cycloadduct in all cases, was higher than of the
uncatalysed reaction although the results are not impressive
since the regioselectivity is partial and not much different from
that of the uncatalysed reaction (vide infra).
Fig. 1 TEM images of as-synthesized (a–e) and used (f) nanoparticles.
(a) Fe3O4; (b) CoFe2O4; (c) MnFe2O4; (d) MnO; (e) a-MnS; (f) Fe3O4 after
5 runs.
(C–H stretching of sp2 carbon) and the strong peak at 590 cmꢀ1
(Fe–O stretching) are only present in the spectrum of oleic acid
coated Fe3O4 NPs. The density of the fatty acid ligands on the
NP surface affects both the number of catalytic sites and the
ease by which the reactants and products can diffuse to and
from the NP. The ligand density collected in Table 2 were
calculated by dividing the amount of fatty acid measured by
elemental analysis by the surface area of the inorganic core
calculated from TEM data. (The relevant equation is developed
in the ESI†). The ligand density is approximately constant
among the NPs (1.44–1.52 molecules per nm2), except for the
higher value of MnFe2O4 NPs (1.93 molecules per nm2). These
values are at the low side of the ligand density range (1.2–
6.6 molecules per nm2) for long-chain ligands, calculated from
the data collected in ref. 52.
A related methodology has been published,57 in which
naked maghemite (g-Fe2O3) NPs supported on hydroxyapatite
were shown to catalyse the reaction of an organic halide with
an alkyne in the presence of NaN3 in water at 100 1C. This
methodology is different from our one as to the nature of the
NPs (naked, supported vs. coated, colloidal), the different
reactants and solvent, and the reaction temperature. Further-
more, we extended the scope of the reaction and compared the
activity of four different oxides and a sulfide.
In order to further investigate the behaviour of the metal
oxide and sulfide NPs listed in Table 1, we considered the
cycloaddition between benzylazide 1b, phenylazide 1c and
methyl propiolate 2 (Scheme 1 and Table 4). While the reaction
times were dependent upon the azide, the reactant conversion
was satisfactory for all the NPs. Again, the best conversion and
3.2 Azide–alkyne cycloadditions
Because of the well-known relevance of the azide–alkyne cyclo- regioselectivity were achieved in the presence of the Fe3O4 NPs.
additions,53,54 we decided to investigate the reaction between Because of the close similarity of the 1H-NMR spectra of the
azides 1 and methyl propiolate 2 or dimethylacetylene dicarb- isomeric cycloadducts 4 and 5, the independent synthesis of the
oxylate 3 (Fig. 4) in the presence of catalytic amounts of the NPs 4-substituted 1,2,3-triazoles 4 were performed in the presence of
listed in Table 1. The appropriate reaction conditions were copper(I) oxide in toluene (see Experimental section).
established by studying the reaction between the azide 1a and
By submitting azide 1a to the reaction with dimethylacetylene-
methyl propiolate 2 in anhydrous toluene (Scheme 1 and dicarboxylate 3 (DMAD), the cycloadduct 6 was obtained quanti-
Table 3), this azide was chosen since further synthetic trans- tatively in the presence of both Fe3O4 and MnFe2O4 at 20 1C
formations of cycloaddition products can be envisaged.55
(Scheme 2 and Table 5). The uncatalysed cycloaddition also
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18053

View Article Online
Paper
NJC
Fig. 2 Electron diffraction patterns of as-synthesized (a–e) and used (f) nanoparticles. (a) Fe3O4; (b) CoFe2O4; (c) MnFe2O4; (d) MnO; (e) a-MnS; (f) Fe3O4
after 5 runs. Patterns in (a), (b), (c) and (f) are consistent with the spinel structure; patterns in (d) and (e) are consistent with the rock-salt structure.
Table 2 Ligand density of the metal oxide and sulfide nanoparticles
Ligand density
NP
Surface ligand C (w/w%) H (w/w%) (molecules per nm2)
Fe3O4
CoFe2O4 Oleic acid
MnFe2O4 Oleic acid
MnO
a-MnS
Oleic acid
3.75
10.77
14.41
6.12
0.59
9.48
13.28
6.68
1.44
1.52
1.93
1.45
1.51
Stearic acid
Stearic acid
1.84
2.03
respect to Sharpless’ ‘‘click’’ azide–alkyne cycloaddition since it
is well-known that it does not occur onto 1,2-disubstituted
acetylenes. However, a few examples of Cu(I) complexes have
been shown to catalyse the cycloaddition to internal alkynes.59
The recycling of the NPs was studied for the cycloaddition of 1c
to 2 catalysed by Fe3O4 NPs. We performed two series of experi-
ments. The first one follows a conventional protocol: the NPs were
magnetically recovered, washed, and immediately dispersed in fresh
solvent. In the second series, the magnetically recovered NPs were
Fig. 3 Selected FT-IR spectra of NP catalysts. Blue: fresh Fe3O4 NP
catalyst. Red: Fe3O4 NP catalyst after 5 runs of the 1c + 2 cycloaddition.
Black: a-MnS catalyst.
occurred in the same conditions requiring longer reaction washed, dried, and weighted before dispersion in fresh solvent in
times. Notwithstanding it is known that Huisgen cycloadditions order to measure the loss of NPs due to manipulation. Fig. 5 shows
usually occur at high temperature,58 this latter result should be the cycloaddition yield for 5 cycles. The isolation yield, as measured
related to the high dipolarophilic aptitude of DMAD. The in the conventional recycling experiments, decreases with recycling
obtainment of the 1,2,3-triazole 6 represents a dissimilarity with and falls below 50% of the initial value at the fifth cycle.
18054 | New J. Chem., 2019, 43, 18049--18061 This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019

View Article Online
NJC
Paper
the fifth cycle. The recycling ability of these NPs is not striking
but one should recall that they are cheap and do not present
serious waste management problems being composed of bio-
compatible iron oxide and fatty acids so extended recycling is
not as important as for precious or toxic catalysts.
To better understand the fate of the catalytic NPs, we
analysed Fe3O4 NPs after the fifth conventional run by TEM
and FT-IR. The FT-IR spectra of fresh and used catalyst are
similar (Fig. 3), showing that NPs maintain the oleic acid
coating. Note that no peaks attributable to 1c or 2 can be seen.
The iron oxide NP core maintained the magnetite structure
after the five cycloaddition runs as shown by the ED patterns in
Fig. 2. TEM images show that the NPs formed large agglomerates
with size 4100 nm, in which the Fe3O4/oleic acid NPs maintain
their individuality (Fig. 1). The size of the NPs is unaffected by the
use as a catalyst. Thus, in addition to NP loss by manipulation,
another cause of decreased activity of the NPs is the agglomeration
of the otherwise unchanged NPs that slows down the diffusion of
reactants and products to and from the NP surface and lowers the
number of catalytic sites available for reaction. These results suggest
that the recycling ability of the NPs could be increased by
(i) improving the magnetic separation by optimized magnetic
field gradients and (ii) sonicating the dispersed NPs.
Fig. 4 Organic reactants for the NP-catalysed azide–alkyne cycloaddition.
The comparison of the present results with those previously
obtained through an uncatalysed cycloaddition must be limited
to the reaction between phenylazide 1c and methyl propiolate
2. In fact, cycloadducts 4a and 5a are novel products, while the
1,2,3-triazoles 4b and 5b arising from the reaction of benzylazide
Scheme 1 NP-catalysed cycloaddition between azides 1a–c and methyl
propiolate 2.
1b and methyl propiolate 2 were obtained as unique regioisomers
From the second series of experiments, we learned that the in the presence of copper(I)60 or Cp*RuCl(PPh3)2 catalysts,
NP mass decreases after each cycle due to manipulation (see respectively. As described by Huisgen,62 phenylazide 1c reacted
ESI,† Fig. S3). However, it was not possible to envisage the with methyl propiolate 2 in the absence of solvent for 12 days at
presence of iron in the mother solution i.e. in the cycloadducts r.t. plus 34 h at 60 1C yielding a mixture of 1-phenyl-4-
mixtures. It is likely that the tiny amount of Fe3O4 NPs (5.5 mgFe methoxycarbonyl-1,2,3-triazole 4c and 5-methoxycarbonyl isomer
per cycle on average, compared to B750 mg of cycloadducts) 5c in 88 : 12 ratio (83% combined yield). In a previous report,63 we
was lost by washing of the mother solution with water followed investigated the reactive behaviour of phenylazides and methyl
by crystallisation of the crude triazoles. When the isolation propiolate in boiling tetrachloromethane. After 36 h, the 1c + 2
yield is scaled by the remaining amount of NPs, it remains cycloaddition gave 4c :5c = 75:25 (496% overall yield). Comparison
constant for four cycles and falls to 80% of the initial value at of these results with Table 4 shows that NP catalysis provides better
61
Table 3 Cycloaddition between azide 1a and methyl propiolate 2a
b
b
Entry
NPs
wFe/w1a (%)
wM/w1a (%)
Time (h)
T (1C)
4a + 5a (%)
4a : 5a
1
2
3
4
5
6
7
8
—
—
—
—
—
—
21
21
21
21
10
6.4
—
—
—
—
—
—
—
—
—
—
—
3.7 (Co)
2.2 (Mn)
3.6 (Mn)
3.6 (Mn)
3.1 (Mn)
3.1 (Mn)
240
24
8
24
7
3
2
8
6
7
40
8
20
45
95
20
45
75
95
45
45
45
45
45
45
45c
57d
89
57d
95
88
80
93
88
66
83
62
78
51 : 49
57 : 43
68 : 32
65 : 35e
83 : 17e
67 : 33e
68 : 32e
75 : 25e
71 : 29e
65 : 35e
65 : 35e
70 : 30e
69 : 31e
Fe3O4
Fe3O4
Fe3O4
Fe3O4
CoFe2O4
MnFe2O4
MnO
MnO
a-MnS
a-MnS
9
10
11
12
13
40
a
b
c
d
In the presence of iron(III) oleate or bulk Fe2O3: (4a + 5a) o 10%, 4a : 5a undetermined. Weight ratio. Unreacted 1a, 55%. Unreacted 1a, 40%.
e
As determined by 1H NMR analysis.
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18055

View Article Online
Paper
NJC
Table 4 Cycloaddition between azides 1b, c and methyl propiolate 2
Entry
NPs
R1
wFe/w1b,c (%)
wM/w1b,c (%)
Time (h)
T (1C)
4 + 5 (%)
4 : 5
1
2
3
4
5
6
7
8
—
CH2Ph
CH2Ph
CH2Ph
CH2Ph
CH2Ph
CH2Ph
Ph
Ph
Ph
Ph
Ph
—
48
23
15
—
—
—
54
26
16
—
—
—
—
8.5 (Co)
5.1 (Mn)
8.2 (Mn)
7.1 (Mn)
—
22
24
24
20
21
22
20
8
8
8
8
8
75
45
45
45
45
45
90
45
45
45
45
45
85
96
90
88
80
84
99
99
95
95
92
95
73 : 27
Fe3O4
CoFe2O4
MnFe2O4
MnO
a-MnS
—
Fe3O4
CoFe2O4
MnFe2O4
MnO
82 : 18a
76 : 24a
78 : 22a
81 : 19a
78 : 22a
75 : 25
—
82 : 18a
82 : 18a
78 : 22a
74 : 26a
77 : 23a
9
9.6 (Co)
5.5 (Mn)
9.1 (Mn)
7.9 (Mn)
10
11
12
a-MnS
Ph
As determined by 1H NMR analysis.
a
was able to furnish selectively both the 4-methoxycarbonyl-1,2,3-
triazoles 4b60 and 4c.20 In both cases, yields 490% were achieved in
a few hours at RT.
Compared to the above mentioned Huisgen-type cyclo-
additions,58,62 our present results show that the catalytic effect
of the studied NPs is such to increase the reaction rate without
significantly enhancing the regioselectivity in favour of the
4-substituted triazoles 4. This behaviour may be ascribed to
the reversible formation of a labile azide-NP intermediate that
reacts with the alkyne faster than the free azide. The inter-
mediate arises from the interaction between the azide moiety
and the uncoordinated metal ion at the NP surface, as shown in
Scheme 3. This hypothesis is based on the following considerations.
First, we have shown that the catalytic effect is due to the NP
surface, most probably to under-coordinated metal ions, since
both bulk iron oxide and iron(III) oleate do not increase the
cycloaddition rate. The lack of azide degradation, the absence
of by-products, and the occurrence of the cycloaddition to a
1,2-disubstitued acetylene are consistent with the formation of
an azide-NP intermediate. It seems likely that organic azides
ligate to surface iron ions since both complexes of Fe(III) with
Scheme 2 1,2,3-Triazole 6 obtained from the cycloaddition between
azide 1a and DMAD 3 in the presence of NPS (see also Table 5).
Table 5 Cycloaddition between azide 1a and DMAD 3
Entry NPs
wFe/w1a (%) wMn/w1a (%) Time (h) T (1C) 6 (%)
1
2
3
4
5
—
—
5
6
—
5
—
—
2
—
—
96
22
20
4
20
20
20
95
95
72
99
91
95
99
Fe3O4
MnFe2O4
—
Fe3O4
1
the N3 anion64 and transition metal complexes with aromatic
ꢀ
azide ligands65 are known, despite that Fe(III) complexes with
organic azide ligands have not been reported (at the best of our
knowledge). Ligation of the alkyne ester to surface iron ions is
unlikely as Fe(III) complexes with alkynes or esters seem to be of
very minor importance.
However, further investigations would be desirable to unravel
the details of the mechanistic features concerned to the catalytic
activity of the mentioned nanoparticles. In order to get further
support to the proposed mechanism of the NP-catalysed azide–
alkyne cycloaddition, we carried out DFT calculations that are
reported in the next subsection.
Fig. 5 Catalyst recycling for the 1c + 2 cycloaddition catalysed by Fe3O4
NPs. Blue: isolation yield. Orange: isolation yield scaled to the NP loss. All
yields are normalized to the value at cycle 1. Error bars represent 2
standard deviations.
3.3 Computational results
DFT calculations were carried out to assess the plausibility of our
working hypothesis that the increased reaction rate is related to
the adsorption of the azide on the NP surface via the interaction
between the azido group and under-coordinated surface metal
yield and similar regioselectivity in a much shorter time and at ions. Theoretical computations focused on the cycloaddition of
lower temperature. Furthermore, the use of a copper(I) nanocatalyst phenylazide 1c to methyl propiolate 2 catalysed by iron oxide NPs.
18056 | New J. Chem., 2019, 43, 18049--18061 This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019

View Article Online
NJC
Paper
Scheme 3 Proposed catalytic cycle for the NP-catalysed cycloaddition between azide 1 and alkyne 2 in the presence of magnetite NPs. The key step is
the adsorption of the azide onto an under-coordinated Fe site at the NP surface. For the sake of clarity, we pictured a single transition state structure but
it is understood that both regioisomeric transition states occur. A similar cycle is thought to be effective for mixed ferrites, MnO, and MnS.
The problem was attacked in two ways. First, we employed a
periodic model where 1c is ligated to an under-coordinated iron(III)
ion at the (100) surface of a magnetite (Fe3O4) slab. Periodic
calculations with full geometric relaxation provides a description
of the electronic structure of the 1c-magnetite system allowing us
to discuss the NP catalytic effect using a frontier molecular orbital
(FMO) approach. Since periodic calculations do not allow one to
locate transition states (TSs), we secondly used simple molecular
models (neutral Fe(OH)340 or Fe2(OH)4 fragments ligated to 1c) to
calculate the regioisomeric TSs of the 1c + 2 reaction.
Fig. 6 shows the unit cell of the magnetite(100) surface
adopted in the periodic computations and the 1c azide molecule
adsorbed in its minimum energy conformation. We are confident
that the results discussed below do not significantly depend on
the surface orientation.
According to the FMO approach, reactivity is enhanced when
the energy gap between the highest occupied (HOMO) and the
lowest unoccupied molecular orbital (LUMO) of the reactants
decreases. In molecular systems, the available electronic states
correspond to a set of discrete orbital energies. In periodic
systems, available electronic states can be found in extended
energy intervals (bands) and are collectively pictured by a
continuous curve representing the density of available states
(DOS). The larger the DOS in a given energy range, the larger
the number of states accessible to electrons. We note here that,
since magnetite is a magnetic material with a different number
Fig. 6 Minimum-energy structure of the 1c-magnetite adsorption
of spin-up and spin-down electrons, ab initio computations
complex. Colour code is as follows. White: hydrogen; black: carbon; blue:
provide two distinct sets of electronic states and hence two nitrogen; red: oxygen; pink: iron.
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18057

View Article Online
Paper
NJC
Adsorption of 1c onto the magnetite surface produces marked
effects on its electronic structure, which affect the HOMO–
LUMO gaps (Fig. 7). The overlap of azide and iron peaks in the
DOS shows that the molecular states of 1c hybridize with states
located on the magnetite iron atoms (see the states pictured in
Fig. S5 in the ESI†).
As concerns the spin-down electronic states (left panel), we
note that the peak corresponding to the HOMO(1c) state slightly
mixes with Fe orbitals, and its energy slightly lowers with respect to
the gas-phase (ꢀ0.1 eV). A little hybridization among Fe and N
states occurs also for the LUMO(1c) peak, leading to a much larger
energy shift of about ꢀ1.3 eV. Consistently, upon adsorption of 1c
onto magnetite, the HOMO(1c)–LUMO(2) gap changes slightly
with respect to the isolated molecules (see Table 6), while the
HOMO(2)–LUMO(1c) drops from 4.7 to 3.4 eV.
The hybridization of 1c frontier orbitals with spin-up magnetite
states is stronger and more intriguing. The LUMO(1c) orbital
hybridizes with states at the surface of magnetite, giving rise to a
peak at about ꢀ2.8 eV. The HOMO(1c) strongly mixes with magnetite
states giving rise to a broad band of HOMO-like azide–iron states
ranging from ꢀ5.5 eV to about ꢀ4.5 eV. The spin-up HOMO–LUMO
energy gaps between 1c-magnetite and 2 drop to 2.5 [HOMO(1c)–
LUMO(2)] and 3.5 eV [HOMO(2)–LUMO(1c)], values much smaller
than those of the 1c + 2 reaction. These spin-up 1c-magnetite hybrid
electronic states, which contain large contributions from the –N3
moiety, provide a faster HOMO-dipole controlled reactivity channel.
The computational results on the periodic model thus suggest
Fig. 7 Electronic states of isolated 1c and the 1c-magnetite complex
compared with the states of 2. The states of isolated 1c (middle) and 2 (left
and right ends) are represented as thick lines. The states of isolated 1c
having large contribution from the –N3 moiety are blue and the states of
having large contribution from the –CRC– moiety are black. The density
of states (DOS) of spin down (left panel) and spin up (right panel) electrons
of the magnetite-1c complex are portrayed as continuous curves.
Contributions from different moieties are color coded as follows. Blue: azide
nitrogens, grey: phenyl carbons, red: iron atom interacting with 1c. The Fermi
level is displayed as a dotted horizontal line (electronic states below the
Fermi level are occupied, states above it are empty). The energy shift of the
1c HOMO and LUMO upon adsorption are indicated by blue arrows.
Relevant HOMO–LUMO interactions are highlighted with dashed lines.
different DOS, one for each spin component. To get clear that the adsorption of azide 1c on the magnetite surface is likely
evidence of the energy range where the orbitals of a given atom to increase the cycloaddition rate because of lower energy gaps
group contribute significantly to global electronic states, we for both spin-down (azide-like orbitals) and spin-up (hybrid
projected the DOS of 1c-magnetite onto the orbitals of azide orbitals) electrons. It is less clear what effect adsorption might
nitrogens, phenyl carbons and the iron atom interacting with have on the regioselectivity since HOMO-dipole control is weaker
1c. The analysis conducted below safely assumes that the for magnetite-adsorbed 1c (dDE = 1.1 and 1.0 eV) than for isolated
reaction between 1c-magnetite and 2 involves electronic states 1c (dDE = 1.5 eV).
with significant orbital contribution coming from the –N3 (1c)
and the –CRC– (2) moieties.
To gain more insight, we then turned to the investigation of
the relevant TSs in the above delineated simple molecular
The DOS of 1c adsorbed onto the (100) ferrite surface is model. The geometry of the regioisomeric TSs of 1c-Fe(OH)3 +
shown in Fig. 7 along with the orbital energies of the isolated 2 2 and 1c-Fe2(OH)4 + 2 optimized at the OPBE/LANL2/LANL2DZ
and 1c molecules, while relevant HOMO–LUMO gaps have been level are shown in Fig. 8. (The 1c + 2 TSs and the 1c-Fe(OH)3 and
collected in Table 6. For the isolated molecules, we have highlighted 1c-Fe2(OH)4 adducts can be found in the ESI,† Fig. S6).
the MOs with a large contribution from the reactive moieties –N3
The presence of the Fe–O clusters does not affect much the
(blue, 1c) and –CRC– (black, 2). For the 1c-magnetite system, the TS structures. With respect to the 1c + 2 case, the TS leading to
contributions to the DOS from azide nitrogens (blue), phenyl 4 is more symmetrical than that leading to 5 and the length of
carbons (grey) and the iron atom interacting with 1c (red) show the forming bonds changes by less than 0.1 Å, except for one
where the electronic states relevant to the reaction are located. case. The activation energy E‡ = E(TS) ꢀ E(1c) ꢀ E(2) for the
Pictures of relevant electronic states of 2, 1c, and 1c-magnetite can cycloadditions between 2 and both ligated and non-ligated 1c
be found in the ESI† (Fig. S4).
leading to the 4 and 5 cycloadduct are collected in Table 7. The
As can be seen in Table 6, the HOMO–LUMO gaps between ligation of 1c to Fe–O clusters decreases the TS energy of both
1c and 2 are 3.2 and 4.7 eV, indicating HOMO-dipole control. regioisomeric pathways leading to a significant increase of the
Table 6 Computed HOMO–LUMO energy gaps (eV) for selected systems
System
Spin component
DE [HOMO(1c)–LUMO(2)]
DE [HOMO(2)–LUMO(1c)]
1c + 2 (gas phase)
1c-magnetite + 2
1c-magnetite + 2
Both
Down (minority)
Up (majority)
3.2
3.3
2.5
4.7
3.4
3.5
18058 | New J. Chem., 2019, 43, 18049--18061 This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019

View Article Online
NJC
Paper
transition metal oxides and sulfide nanocatalysts. These solvent-
dispersible, inexpensive magnetic nanoparticles allowed milder
reaction conditions and better product yields compared to the
corresponding examples of the classic Huisgen reaction. The regio-
selectivity of the catalysed cycloaddition towards the 4-cycloadduct
was to some extent improved with respect to the thermal process,
and the reaction to 1,2-disubstituted acetylenes was proved to be
feasible. Experimental and computational results support the
hypothesis that the reaction rate increase is due to enhanced
reactivity of the adduct between the azide and an under-
coordinated iron ion at the surface of the NPs with respect to
the free azide.
Conflicts of interest
There are no conflicts of interest to declare.
Acknowledgements
G. M. was financially supported by the Department of Chemistry of
`
the Universita degli Studi di Milano (PSR2015-1716FDEMA_09).
References
1 D. Wang and D. Astruc, Chem. Soc. Rev., 2017, 46, 816–854.
2 S. Shylesh, V. Schu¨nemann and W. R. Thiel, Angew. Chem.,
Int. Ed., 2010, 49, 3428–3459.
3 A. Padwa, in Comprehensive Organic Synthesis, ed. B. M.
Trost, I. Fleming and M. F. Semmelhack, Pergamon Press,
Oxford, 1991, vol. 4, pp. 1069–1109.
Fig. 8 Structure of the regioisomeric transition states of the 1c-Fe(OH)3 + 2
and 1c-Fe2(OH)4 + 2 reactions. Relevant distances are given in angstrom.
Colour code is as follows. White: hydrogen; grey: carbon; blue: nitrogen; red:
oxygen; pink: iron.
Table 7 DFT (OPBE/LANL2/LANL2DZ) activation energy E‡ (kcal molꢀ1
)
4 A. Michael, J. Prakt. Chem., 1893, 48, 94–95.
5 R. Huisgen, Proc. Chem. Soc., 1961, 357–396.
6 R. Huisgen, Angew. Chem., Int. Ed. Engl., 1963, 2, 565–598.
7 R. Huisgen, Angew. Chem., Int. Ed. Engl., 1963, 2, 633–696.
8 C. W. Tornøe, C. Christensen and M. Meldal, J. Org. Chem.,
2002, 67, 3057–3064.
for the 4- and 5-regioisomers of the cycloaddition between 2 and both
ligated and unligated 1c
4
5
1c
17.2
15.0
10.1
17.1
13.1
16.1
1c-Fe(OH)3
1c-Fe2(OH)4
9 V. V. Rostovtsev, L. G. Green, V. V. Fokin and K. B.
Sharpless, Angew. Chem., Int. Ed., 2002, 41, 2596–2599.
10 B. C. Boren, S. Narayan, L. K. Rasmussen, L. Zhang, H. Zhao,
Z. Lin, G. Jia and V. V. Fokin, J. Am. Chem. Soc., 2008, 130,
8923–8930.
reaction rate, in agreement with experiments. However, the
data are inconsistent with respect to regioselectivity. One model
favours the 4 cycloadduct whereas the other one favours the 5
cycloadduct. This is not unexpected due to (i) the simplistic
11 L. K. Rasmussen, B. C. Boren and V. V. Fokin, Org. Lett.,
2007, 9, 5337–5339.
models employed and (ii) the high accuracy (o1 kcal molꢀ1
)
needed to predict the regioselectivity. The computational
results based on the molecular model systems support the
FMO analysis of the periodic model calculations. We can thus
conclude that DFT calculations support the hypothesis that the
observed catalytic effect is due to the coordination of 1c to an
under-coordinated iron ion at the NP surface.
12 J. R. Johansson, T. Beke-Somfai, A. Said Stålsmeden and
N. Kann, Chem. Rev., 2016, 116, 14726–14768.
13 W. Song and N. Zheng, Org. Lett., 2017, 19, 6200–6203.
14 W. G. Kim, M. E. Kang, J. B. Lee, M. H. Jeon, S. Lee, J. Lee,
B. Choi, P. M. S. D. Cal, S. Kang, J.-M. Kee, G. J. L. Bernardes,
J.-U. Rohde, W. Choe and S. Y. Hong, J. Am. Chem. Soc.,
2017, 139, 12121–12124.
¨
15 S. Brase, C. Gil, K. Knepper and V. Zimmermann, Angew.
4. Conclusions
Chem., Int. Ed., 2005, 44, 5188–5240.
¨
A novel protocol for the catalysed azide–alkyne cycloaddition has 16 Organic Azides: Syntheses and Applications, ed. S. Brase,
been developed, which relies upon Earth-abundant, first-row K. Banert, John Wiley & Sons, Ltd, Chichester, 2010, pp. 389–490.
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18059

View Article Online
Paper
NJC
17 S. Pålhagen, R. Canger, O. Henriksen, J. A. van Parys, M.-E. 43 P. Calcagnile, D. Fragouli, I. S. Bayer, G. C. Anyfantis,
`
Riviere and M. A. Karolchyk, Epilepsy Res., 2001, 43, 115–124.
L. Martiradonna, P. D. Cozzoli, R. Cingolani and
´
18 J. Diot, M. I. Garcıa-Moreno, S. G. Gouin, C. Ortiz Mellet,
A. Athanassiou, ACS Nano, 2012, 6, 5413–5419.
K. Haupt and J. Kovensky, Org. Biomol. Chem., 2009, 7, 357–363. 44 H. Zhu, S. Zhang, Y.-X. Huang, L. Wu and S. Sun, Nano Lett.,
19 J. M. Aizpurua, R. M. Fratila, Z. Monasterio, N. Perez-Esnaola, 2013, 13, 2947–2951.
E. Andreieff, A. Irastorza and M. Sagartzazu-Aizpurua, New 45 E. Capetti, A. M. Ferretti, V. Dal Santo and A. Ponti, Beilstein
J. Chem., 2014, 38, 474–480. J. Nanotechnol., 2015, 6, 2319–2329.
20 G. Molteni, C. L. Bianchi, G. Marinoni, N. Santo and 46 J. M. Soler, E. Artacho, J. D. Gale, A. Garcıa, J. Junquera,
`
`
`
A. Ponti, New J. Chem., 2006, 30, 1137–1139.
21 T. Jin, M. Yan and Y. Yamamoto, ChemCatChem, 2012, 4,
1217–1229.
22 F. Alonso, Y. Moglie and G. Radivoy, Acc. Chem. Res., 2015,
48, 2516–2528.
P. Ordejon and D. Sanchez-Portal, J. Phys.: Condens. Matter,
2002, 14, 2745–2779.
47 H. Liu and C. Di Valentin, J. Phys. Chem. C, 2017, 121,
25736–25742.
48 M. Swart, J. Chem. Theory Comput., 2008, 4, 2057–2066.
23 M. B. Gawande, A. Goswami, F.-X. Felpin, T. Asefa, 49 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,
X. Huang, R. Silva, X. Zou, R. Zboril and R. S. Varma, Chem.
Rev., 2016, 116, 3722–3811.
24 M. Gholinejad and N. Jeddi, ACS Sustainable Chem. Eng.,
2014, 2, 2658–2665.
25 S. Kaur, V. Bhalla and M. Kumar, Chem. Commun., 2015, 51,
526–529.
26 R. Hudson, C.-J. Li and A. Moores, Green Chem., 2012, 14,
622–624.
M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone,
B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato,
X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng,
J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda,
J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao,
H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta,
F. Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin,
V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari,
A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi,
N. Rega, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross,
V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli,
J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski,
G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D.
27 Z. Zhang, C. Dong, C. Yang, D. Hu, J. Long, L. Wang, H. Li,
Y. Chen and D. Kong, Adv. Synth. Catal., 2010, 352, 1600–1604.
´
28 F. Chahdoura, C. Pradel and M. Gomez, ChemCatChem,
2014, 6, 2929–2936.
29 H. Woo, H. Kang, A. Kim, S. Jang, J. C. Park, S. Park, B.-S. Kim,
H. Song and K. H. Park, Molecules, 2012, 17, 13235–13252.
30 S. Ishikawa, R. Hudson, A. Moores and C.-J. Li, Heterocycles,
2012, 86, 1023–1030.
31 B. Kumar, K. H. V. Reddy, B. Madhav, K. Ramesh and
Y. V. D. Nageswar, Tetrahedron Lett., 2012, 53, 4595–4599.
32 M. R. Decan, S. Impellizzeri, M. L. Marin and J. C. Scaiano,
Nat. Commun., 2014, 5, 4612.
¨
Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and
D. J. Fox, Gaussian 09, revision C.01, Gaussian, Inc., Wallingford,
CT, 2010.
50 S. Mondini, M. Leonzino, C. Drago, A. M. Ferretti, S. Usseglio,
D. Maggioni, P. Tornese, B. Chini and A. Ponti, Langmuir, 2015,
31, 7381–7390.
33 J. E. Hein and V. V. Fokin, Chem. Soc. Rev., 2010, 39, 1302–1315. 51 S. Mondini, A. M. Ferretti, A. Puglisi and A. Ponti, Nanoscale,
34 A. M. Ferretti, A. Ponti and G. Molteni, Tetrahedron Lett.,
2015, 56, 5727–5730.
2012, 4, 5356–5372. Pebbles is freely available from the
authors, http://pebbles.istm.cnr.it.
35 R. Hudson, Y. Feng, R. S. Varma and A. Moores, Green Chem., 52 A. M. Smith, K. A. Johnston, S. E. Crawford, L. E. Marbella
2014, 16, 4493–4505. and J. E. Millstone, Analyst, 2017, 142, 11–29.
36 C. Della Pina, M. Rossi, A. M. Ferretti, A. Ponti, M. Lo Faro 53 W. Lwowski, in 1,3-Dipolar Cycloaddition Chemistry,
and E. Falletta, Synth. Met., 2012, 162, 2250–2258.
37 A. Gervasini, C. Messi, A. Ponti, S. Cenedese and N. Ravasio,
J. Phys. Chem. C, 2008, 112, 4635–4642.
38 A. Gervasini, C. Messi, P. Carniti, A. Ponti, N. Ravasio and
F. Zaccheria, J. Catal., 2009, 262, 224–234.
ed. A. Padwa, Wiley-Interscience, New York, 1984, vol. 1,
pp. 559–652.
54 C.-K. Sha and A. K. Mohanakrishnan, in Synthetic Applications
of 1,3-Dipolar Cycloaddition Chemistry Toward Heterocycles and
Natural Products, ed. A. Padwa and W. H. Pearson, John Wiley
& Sons, Inc., New York, vol. 59, 2002, pp. 623–679.
39 M. Tejeda-Serrano, J. R. Cabrero-Antonino, V. Mainar-Ruiz,
´
´
M. Lopez-Haro, J. C. Hernandez-Garrido, J. J. Calvino, 55 G. Molteni, Heterocycles, 2013, 87, 1765–1773.
´
A. Leyva-Perez and A. Corma, ACS Catal., 2017, 7, 3721–3729. 56 (a) Z. Li, T. S. Seoa and J. Ju, Tetrahedron Lett., 2004, 45,
40 G. Molteni, A. M. Ferretti, S. Mondini and A. Ponti,
3143–3146; (b) Z.-X. Wang and H.-L. Qin, Chem. Commun.,
J. Nanopart. Res., 2018, 20, 79.
2003, 2450–2451.
41 (a) L. Menini, M. C. Pereira, L. A. Parreira, J. D. Fabris and 57 S. R. Kale, S. S. Kahandal, M. B. Gawande and R. V. Jayaram,
E. V. Gusevskaya, J. Catal., 2008, 254, 355–364; (b) B. Sahoo, RSC Adv., 2013, 3, 8184–8192.
S. Kumar Sahu, S. Nayak, D. Dharaa and P. Pramanik, Catal. 58 R. Huisgen, R. Grashey, M. Seifeld, W. Wallbillich,
Sci. Technol., 2012, 2, 1367–1374. H. Knupfer and R. Schmidt, Ann. Chem., 1962, 653, 105–113.
42 M. Kazemi, M. Ghobadi and A. Mirzaie, Nanotechnol. Rev., 59 (a) S. Dıez-Gonzalez, A. Correa, L. Cavallo and S. P. Nolan,
2017, 7, 43–68. Chem. – Eur. J., 2006, 12, 7558–7564; (b) N. Candelon,
´
´
18060 | New J. Chem., 2019, 43, 18049--18061 This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019

View Article Online
NJC
Paper
´
`
D. Lastecoueres, A. K. Diallo, J. Ruiz Aranzaes, D. Astruc and 62 R. Huisgen, R. Knorr, L. Moebius and G. Szeimies, Chem.
J.-M. Vincent, Chem. Commun., 2008, 741–743; (c) S. C. Sau, Ber., 1965, 98, 4014–4021.
S. R. Roy, T. K. Sen, D. Mullangi and S. K. Mandal, Adv. 63 G. Molteni and A. Ponti, Chem. – Eur. J., 2003, 9, 2770–2774.
Synth. Catal., 2013, 355, 2982–2991.
60 H. A. Orgueira, D. Fokas, Y. Isome, P. C.-M. Chan and
C. M. Baldino, Tetrahedron Lett., 2005, 46, 2911–2914.
64 N. Wiberg, E. Wiberg and A. F. Holleman, Anorganische
Chemie, De Gruyter, Berlin, 103rd edn, 2016, Band 2, Teil
C, Kapitel XXIX, p. 1953.
61 T. Farooq, L. K. Sydnes, K. W. Tornroos and B. E. Haug, 65 S. Cenini, E. Gallo, A. Caselli, F. Ragaini, S. Fantauzzi and
Synthesis, 2012, 2070–2078. C. Piangiolino, Coord. Chem. Rev., 2006, 250, 1234–1253.
This journal is ©The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2019 New J. Chem., 2019, 43, 18049--18061 | 18061