Facile photochemical synthesis and characterization of highly fluorescent silver nanoparticles

Article abstract of DOI:10.1021/ja900201k

Highly fluorescent silver nanoparticles (AgFNP) have been prepared by a facile photochemical method, yielding these materials in just a few minutes and with excellent long-term stability. The method makes use of photogenerated ketyl radicals that reduce A

Full text of DOI:10.1021/ja900201k

Published on Web 09/10/2009
Facile Photochemical Synthesis and Characterization of
Highly Fluorescent Silver Nanoparticles
Luca Maretti, Paul S. Billone, Yun Liu, and Juan C. Scaiano*
Centre for Catalysis Research and InnoVation and Department of Chemistry, UniVersity of
Ottawa, 10 Marie Curie, Ottawa K1N 6N5, Canada
Received January 10, 2009; E-mail: tito@photo.chem.uottawa.ca
Abstract: Highly fluorescent silver nanoparticles (AgFNP) have been prepared by a facile photochemical
method, yielding these materials in just a few minutes and with excellent long-term stability. The method
makes use of photogenerated ketyl radicals that reduce Ag+ from silver trifluoroacetate in the presence of
amines. While as functional materials these AgFNP can be described as of nanometer dimensions, we
believe that the luminescence arises from particle-supported small metal clusters (predominantly Ag2).
The materials have been characterized by electron microscopy, fluorescence and absorption spectroscopy,
fluorescence lifetime studies, and 19F NMR spectroscopy. Exploratory work shows that the fluorescence
from AgFNP can be efficiently quenched by paramagnetic quenchers, and these studies have been
combined with electron paramagnetic resonance work.
Introduction
other chromophores near the surface can undergo enhanced
transitions, sometimes resulting in increased luminescence.9,10
Research on fluorescent nanoparticles concentrates mainly on
semiconductor particles, usually referred to as quantum dots.1
Among these, CdSe particles, either core-only or core-shell,
are the most widely studied. In the case of CdSe core-shell
quantum dots, the shell (frequently ZnS)2 makes the particles
more strongly fluorescent and significantly more robust. The
fluorescence of CdSe quantum dots can be readily tuned across
the visible spectrum. In spite of their widespread use, quantum
dots frequently present problems related to the intrinsic blinking
of their luminescence, and to toxicity issues that limit their
applications in the health sciences.3 Further, while a shell makes
them more robust, it also limits their use in some sensing
applications, since the same shell that brings these benefits also
reduces their sensitivity to potential solution analytes.4,5
In contrast with metal nanoparticles in the nanometer scale,
very small clusters can show remarkably strong emission.11-29
In the case of silver, clusters such as Ag2, Ag3, and Ag4 have
(9) Lakowicz, J. R. Anal. Biochem. 2005, 337, 171–194.
(10) Tam, F.; Goodrich, G. P.; Johnson, B. R.; Halas, N. J. Nano Lett.
2007, 7, 496–501.
(11) Zheng, J.; Nicovich, P. R.; Dickson, R. M. Annu. ReV. Phys. Chem.
2007, 58, 409–431.
(12) Zheng, J. P. J. T.; Dickson, R. M. J. Am. Chem. Soc. 2003, 125, 7780–
7781.
(13) Vosch, T.; Antoku, Y.; Hsiang, J.-C.; Richards, C. I.; Gonzalez, J. I.;
Dickson, R. M. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 12616–12621.
(14) Petty, J. T.; Zheng, J.; Hud, N. V.; Dickson, R. M. J. Am. Chem. Soc.
2004, 126, 5207–5212.
(15) Yu, J.; Patel, S. A.; Dickson, R. M. Angew. Chem., Int. Ed. 2007, 46,
2028–2030.
Metal nanoparticles, particularly those from coin metals, are
expected to have lower toxicity and can be readily prepared
reproducibly and with excellent solution stability. Particles of
a few nanometers in diameter frequently display a plasmon
absorption that in the case of gold and silver is readily detectable
in the visible spectral region.6-8 This absorbance is due to the
collective oscillation of the electrons in the conduction band
and is not associated with any significant emission, although
(16) Zheng, J.; Dickson, R. M. J. Am. Chem. Soc. 2002, 124, 13982–13983.
(17) Rabin, I.; Schulze, W.; Ertl, G. J. Chem. Phys. 1998, 108, 5137–
5142.
(18) Zhang, J.; Xu, S.; Kumacheva, E. AdV. Mater. 2005, 17, 2336–2340.
(19) Alqudami, A.; Annapoorni, S. Plasmonics 2007, 2, 5–13.
(20) Shen, Z.; Duan, H.; Frey, H. AdV. Mater. 2007, 19, 349–352.
(21) Shang, L.; Dong, S. Chem. Commun. 2008, 1088–1090.
(22) Rabin, I.; Schulze, W.; Ertl, G. Chem. Phys. Lett. 1999, 312, 394–
398.
(23) Rabin, I.; Schulze, W.; Ertl, G.; Felix, C.; Sieber, C.; Harbich, W.;
Buttet, J. Chem. Phys. Lett. 2000, 320, 59–64.
(1) Alivisatos, A. P. Science 1996, 271, 933–937.
(2) Dabboussi, B. O.; Rodriguez-Viejo, J.; Mikulec, F. V.; Heine, J. R.;
Mattoussi, H.; Ober, R.; Jensen, K. F.; Bawendi, M. G. J. Phys. Chem.
B 1997, 101, 9463–9475.
(24) Felix, C.; Sieber, C.; Harbich, W.; Buttet, J.; Rabin, I.; Schulze, W.;
Ertl, G. Phys. ReV. Lett. 2001, 86, 2992–2995.
(25) Fedrigo, S.; Harbich, W.; Buttet, J. J. Chem. Phys. 1993, 99, 5712–
5717.
(3) Medintz, I.; Mattoussi, H.; Clapp, A. Int. J. Nanomed. 2008, 3, 151–
167.
(26) Treguer, M.; Rocco, F.; Lelong, G.; Le Nestour, A.; Cardinal, T.;
Maali, A.; Lounis, B. Solid State Sci. 2005, 7, 812–818.
(27) Makarava, N.; Parfenov, A.; Baskakov, I. V. Biophys. J. 2005, 89,
572–580.
(4) Heafey, E.; Laferrie`re, M.; Scaiano, J. C. Photochem. Photobiol. Sci.
2007, 6, 580–584.
(5) Rene-Boisneuf, L.; Scaiano, J. C. Chem. Mater. 2008, 20, 6638–6642.
(6) Burda, C.; Chen, X.; Narayanan, R.; El-Sayed, M. A. Chem. ReV. 2005,
105, 1025–1102.
(28) Maali, A.; Cardinal, T.; Treguer-Delapierre, M. Physica E 2003, 17,
559–560.
(29) Sengupta, B.; Ritchie, C. M.; Buckman, J. G.; Johnsen, K. R.;
Goodwin, P. M.; Petty, J. T. J. Phys. Chem. C 2008, 112, 18776–
18782.
(7) Henglein, A. Chem. ReV. 1989, 89, 1861–1873.
(8) Kamat, V. K. J. Phys. Chem. B 2002, 106, 7729–7744.
9
13972 J. AM. CHEM. SOC. 2009, 131, 13972–13980
10.1021/ja900201k CCC: $40.75  2009 American Chemical Society

Highly Fluorescent Silver Nanoparticles
A R T I C L E S
well characterized fluorescence.17,23,25,30 This emission has
molecular-type properties, relatively long lifetimes and a good
mirror-image relationship between absorption and emission.
However, these aggregates by themselves are effectively too
small and unstable for many practical applications, and it is
desirable to stabilize or encapsulate them with other materials
to make them suitable for applications in sensing, chemical
biology, or advanced materials.13
rescence measurements were performed on an Easylife LS (Photon
Technology International). The excitation was performed with a
440 nm pulsed LED, and the broadband emission was measured
using a 10 nm bandpass filter centered at 540 nm.
All absorption spectra were recorded on a Varian CARY-50
UV-vis spectrophotometer. All samples were placed in UV quality
quartz cuvettes, 1 cm path length unless otherwise noted. The
resolution of our measurements was set to 1 nm. Some experiments
required the use of cuvettes with a 3 mm path length, since
otherwise the absorbance exceeded 3.0 and was beyond the reliable
dynamic range of the spectrometer.
The vast majority of reductive processes leading to metal
nanoparticles rely on the use of thermal methods, such as citrate
or borohydride reduction.31,32 Some applications make use of
photochemical methods that offer temporal and spatial control
(as needed for imaging applications) and where the rate of
reduction can be readily controlled by adjusting the incident
light intensity.33,34 Remarkably, most reports concerned with
photoinitiated synthesis of metal nanoparticles do not give
particular consideration to the absorption properties of the
substrate, the detailed emission properties of the light source,
or the possibilities that quenching mechanisms will vastly reduce
the efficiency with which the synthesis takes place. In recent
publications we have addressed some of these issues in relation
tothephotochemicalsynthesisofsilverandgoldnanoparticles.33-35
In this contribution we report a new method for the facile
photochemical synthesis of highly fluorescent functional silver
nanoparticles. These particles are very stable and can be reliably
and reproducibly synthesized. While the observable silver
fluorescent nanoparticles (AgFNP) are of nanometer dimensions,
we believe that the emission arises from nanoparticle-supported
small metal clusters, a conclusion based on their spectroscopic
properties, and the fact that no “plasmon emission” should be
anticipated. Luminescence from small clusters on particle
surfaces is rare but not unprecedented.26
The samples were exposed to ultraviolet light in a Luzchem
photoreactor equipped with UVA lamps. Typically the unit was
operated with four lamps, corresponding to about 26 W/m2 with
approximately 4% spectral contamination, mostly visible and UVB
light.
19F were performed on a Bruker AVANCE 300 NMR spectrom-
eter. The solvent used in all cases was toluene-d8. 19F spectra were
recorded using a 2.59 s acquisition time with 0.3 Hz of line
broadening, proton decoupling, and 12626 Hz spectral width. An
external chemical shift standard of 0.05% trifluorotoluene (-63.72
ppm) was used to calibrate 19F chemical shift values.
EPR measurements were performed using a JEOL FA-100
X-Band EPR spectrometer equipped with a JEOL ES-UCX2
cylindrical cavity. Samples were deaerated with nitrogen prior to
measurements and held in clear-fused silica tubes (5 mm diameter)
purchased from Wilmad. All spectra were recorded over 2 min at
0.1 mW power, modulation width of 0.01 mT, time constant of
0.03 s, and a sweep width of 5 mT.
Fluorescence microscopy studies employed a Leica DMLS
optical microscope, with a 50 W mercury lamp which provided
the excitation lamp, and a Leica DFC 300 FX camera. Solutions
were spin-coated on 1 inch diameter fused silica discs at 2000 rpm
for 20 s, using a spin-coater from Specialty Coatings, Inc. The
solution was prepared by dissolving 200 mg of polystyrene in 5
mL of toluene and then adding appropriate amounts of silver
trifluoroacetate, I-2959, and hexadecylamine. Solutions were used
on the same day in which they were prepared. Films were exposed
Experimental Section
Unless otherwise indicated, solvents and reagents were purchased
from Aldrich and used as received. Irgacure-2959 (I-2959) (1-[4-
(2-hydroxyethoxy)phenyl]-2-hydroxy-2-methyl-1-propan-1-one) was
a generous gift from Ciba Specialty Chemicals. I-2959 is an efficient
source of ketyl radicals.35,36
The size of the nanoparticles was determined by transmission
electron microscopy (TEM), while the crystalline structures of the
nanoparticles were characterized by high-resolution transmission
electron microscope (HRTEM). Both TEM and HRTEM studies
were performed using a JEOL JEM-2100F field emission transmis-
sion electron microscope equipped with an ultra high resolution
pole-piece operating at 200 kV. An Oxford energy dispersive X-ray
spectrometer (EDS) attached to the JEM-2100F microscope was
used to determine the elemental composition of the nanoparticles.
TEM specimens were prepared by placing microdrops of nanopar-
ticle solution directly onto a copper grid coated with carbon film
(300 mesh, EMS).
to UVA light at approximately 13 W m-2
.
Results
Photochemical Synthesis and UV-Visible Spectroscopy.
Experiments were performed in toluene and in tetrahydrofuran
(THF), mostly under nitrogen and occasionally under air. The
samples were deaerated by bubbling with nitrogen for at least
30 min. Both solvents of choice are transparent in the UVA
region used for exposure during the AgFNP synthesis.
Ketyl radicals are known to be strong reducing agents,
suitable for metal ion reduction.35,37-39 In our case we chose
I-2959 as our source of ketyl radicals; the choice reflects the
excellent absorption properties of I-2959 in the UVA region,
the high quantum yield of radical generation, and the fact that
the short triplet lifetime of I-2959 reduces the likelihood of
excited state quenching by other solution components,36 notably
amines40 and silver ions.33 Scheme 1 shows the proposed
mechanism for the formation of silver nanoparticles.
All the steady-state emission spectra were recorded using a
Photon Technology International spectrofluorimeter. Lifetime fluo-
(30) Felix, C.; Sieber, C.; Harbich, W.; Buttet, J.; Rabin, I.; Schulze, W.;
Ertl, G. Chem. Phys. Lett. 1999, 313, 105–109.
The spectral changes observed upon irradiation (4 UVA
lamps) of a quartz cuvette containing 2 mM each of AgCF3COO,
I-2959, and cyclohexylamine are shown in Figure 1. Ap-
proximately 80% of the maximum achievable absorption is
(31) Turkevich, J.; Stevenson, P. C.; Hiller, J. Discuss. Faraday Soc. 1951,
11, 55–75.
(32) Brust, M.; Walker, M.; Bethell, D.; Schiffrin, D. J.; Whyman, R.
J. Chem. Soc., Chem. Commun. 1994, 801–802.
(33) Scaiano, J. C.; Aliaga, C.; Maguire, S.; Wang, D. J. Phys. Chem. B
2006, 110, 12856–12859.
(37) Hada, H.; Yonezawa, Y.; Akio, Y.; Kurakake, A. J. Phys. Chem. 1976,
(34) McGilvray, K. L.; Decan, M. R.; Wang, D.; Scaiano, J. C. J. Am.
Chem. Soc. 2006, 128, 15980–15981.
80, 2728–2731.
(38) Kometani, N. D. H.; Asami, K.; Yonezawa, Y. PCCP 2002, 4, 5142–
(35) Marin, M. L.; McGilvray, K. L.; Scaiano, J. C. J. Am. Chem. Soc.
2008, 130, 16572–16584.
5247.
(39) Henglein, A.; Tausch-Treml, R. J. Colloid Interface Sci. 1981, 80,
84–93.
(36) Jockusch, S.; Landis, M. S.; Freiermuth, B.; Turro, N. J. Macromol-
ecules 2001, 34, 1619–1626.
(40) Cohen, S. G.; Parola, A.; Parsons, G. Chem. ReV. 1973, 73, 141–161.
9
J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009 13973

A R T I C L E S
Maretti et al.
Scheme 1. Mechanism for the Formation of Stabilized AgFNP
Figure 2. Growth of the 450 nm absorption band upon irradiation (350
nm, four lamps) in toluene. Initial concentrations: 2 mM silver trifluoro-
acetate, 2 mM I-2959, 2 mM amine. Reaction performed in 0.7 × 0.3 cm
quartz cuvette.
reached within 3 min of exposure. The maximum is initially at
447 nm and gradually shifts to 449 nm; it is likely that this
shift simply reflects an underlying increase in light scattering
or the formation of plasmon absorbance due to nanoparticles
(vide infra). The particles produced in this manner were stable
for at least several months, even in the presence of air and
without protection from ambient light.
In a typical experiment, a nitrogen-saturated THF solution
of 5.0 mM AgCF3COO, 5.0 mM cyclohexylamine, and 5.0 mM
I-2959 was irradiated with UVA light. Upon irradiation, we
observe the growth of a sharp absorption band centered at 438
nm with a smaller peak at 390 nm (Figures 1 and below in
Figure 5). Control reactions were performed in the absence of
the photoinitiator at the same working concentrations, and no
absorption was observed. The concentration of starting reagents
was also varied from 2 to 30 mM for each component. When
the concentrations were kept equimolar and below 10 mM, an
absorption band centered at 438 nm was observed to grow upon
photolysis in THF as solvent. The most significant difference
between these experiments was that the value at which the
absorption band was found to plateau increased with increasing
concentrations. However, the decay of the absorption band (vide
infra) was also accelerated at higher concentrations; thus, 5 mM
was adopted as the optimal compromise between signal intensity
and stability. The spectra of reaction mixtures performed at
various concentrations are included in the Supporting Informa-
tion (Figure S2). At higher concentrations (30 mM of each), a
thermal reaction occurred over several hours and similar
absorption and emission peaks were observed. This thermal
reaction was extremely sensitive to the age and quality of both
reagents and solvent and was of limited use because of its
irreproducibility. The use of amines as reducing agents had
previously been reported as a route to metallic nanoparticles.41
The effect of oxygen on the synthesis of AgFNP was also
examined in THF. The only significant effect observed in aerated
solutions was an approximately 2 min induction period, which
we attribute to the sacrificial use of some of the initially
photogenerated radicals to consume the dissolved oxygen; we
note that since the samples are not stirred and the irradiation
times quite short, oxygen is unlikely to be significantly
replenished from the gas phase. Interestingly, the reaction of
ketyl radicals with oxygen yields superoxide radicals (or HO2•
in nonpolar solvents).42,43 It would seem that under our
experimental conditions, HO2• does not reduce Ag+ to Ag0. The
maximum under these conditions was at about 438 nm, slightly
blue-shifted with respect to that observed in toluene.
While cyclohexylamine proved to be effective in the forma-
tion and stabilization of these particles, other amines were also
found to work. While our search was not exhaustive, for
example, hexadecylamine also gave excellent results. Many
other amines, including triethylamine and butylamine, were
tested and did not yield fluorescent particles, resulting sometimes
in cloudy orange/brown suspension, suggestive of extensive
nanoparticle aggregation. Absorbance spectra of samples pre-
pared with triethylamine and in the absence of amine are
presented in the Supporting Information (Figure S3). With
triethylamine, a week absorption band is visible at 415-420
nm, with no associated fluorescence, and there is significant
optical scattering at longer irradiation times, indicative of the
formation of large aggregates of Ag nanoparticles or bulk silver.
The growth plot for hexadecylamine also shows a maximum
at 449 nm and is slightly lower than that for cyclohexylamine.
The two growth plots are compared in Figure 2. Interestingly,
with hexadecylamine at irradiation times longer than 300 s we
observe a small band at 552 nm, possibly due to larger
aggregates (vide infra).
While toluene proved an excellent solvent for AgFNP
synthesis, a few other solvents were tested. Among these, THF
proved to be a convenient one.
(41) Newman, J. D. S. B. G. J. Langmuir 2006, 22, 5882–5887.
(42) Maillard, B.; Ingold, K. U.; Scaiano, J. C. J. Am. Chem. Soc. 1983,
105, 5095–5099.
Figure 1. UV/vis absorption spectra following irradiation (350 nm, four
lamps) of a toluene solution containing 2 mM silver trifluoroacetate, 2 mM
I-2959, 2 mM cyclohexylamine. Reaction performed and monitored directly
in a 0.7 × 0.3 cm quartz cuvette.
(43) Ingold, K. U.; Paul, T.; Young, M. J.; Doiron, L. J. Am. Chem. Soc.
1997, 119, 12364–12365.
9
13974 J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009

Highly Fluorescent Silver Nanoparticles
A R T I C L E S
Figure 5. UV/vis absorption spectrum of Ag particles in toluene prepared
using the same procedure as in Figure 3 followed by transfer to toluene.
The inset shows the band we believe to be due to the residual plasmon
absorption.
Figure 3. UV/vis absorption spectra at various time intervals during
irradiation (350 nm, four lamps) directly in THF. Initial concentrations: 5
mM silver trifluoroacetate, 5 mM I-2959, 5 mM cyclohexylamine. Reaction
performed in 1 × 1 cm quartz cell.
Figure 6. Absorption (red), emission (green), and excitation (blue) spectra
of Ag particles after 4 min of irradiation in THF under the conditions of
Figure 3 and resuspension in toluene.
Figure 4. Growth and decay of the absorbance at 438 nm following 9
min. UVA exposure under the same conditions as in Figure 3.
precipitated from solution; when too much 2 was precipitated
due to extended storage at -5 °C, the particles decomposed
quickly upon warming to room temperature. In addition we
observed shifts in the absorption and emission bands to ∼450
and 530 nm, respectively (vide infra). This change may be due
to the difference in dielectric constants of the two solvents44 as
well as to the degree of surface coverage by the quaternary
ammonium salt.
It was interesting to note that in addition to the strong
absorption band (at 452 nm after toluene resuspension), we also
observed a weak but characteristic peak at approximately 390
nm, as shown in the inset in Figure 5. As we will argue in the
Discussion, we do not believe that the bands around 450 nm
are due to the plasmon band, but it is likely that the absorbance
at 390 nm (with no associated emission) is due to plasmon
transitions.
Fluorescence Spectroscopy. The nanoparticles described in
the preceding section showed strong luminescence in the visible
region. Figure 6 shows the emission, excitation, and absorbance
spectra from a sample irradiated for 4 min under the conditions
described above.
Detailed emission and excitation spectra are included in the
Supporting Information. Interestingly, in THF the emission was
quite red-shifted (λmax 580 nm) with respect to the maximum
at 528 nm observed in toluene. The sample yielding the
fluorescence spectra in THF solution contains not only silver
nanoparticles but also the photoproducts originating from the
In contrast to the case of toluene where particles are stable
for months, this is not the case in THF, where particles decay
over a few hours, as shown in Figure 4. The thermal decay of
the absorbance in THF follows a clean monoexponential decay
with a lifetime of 154 min.
We believe that the instability in THF is due to dissolution
of the ammonium trifluoroacetate salt which forms a major part
of the stabilizing layer on the particles, as indicated by NMR
results (vide infra). Given the lack of long-term stability shown
in Figure 4, one could assume that synthesis in THF solvent
was simply not practical. However, imaging data (vide infra)
suggests that use of THF for nanoparticle synthesis leads to
narrower polydispersity than in the case of toluene. This led us
to explore ways of purifying particles originally made in THF
to enhance their long-term stability. Transfer to toluene was an
obvious choice, given the recorded stability in this solvent. After
THF evaporation we added toluene to yield a bright yellow
solution and a white precipitate which was filtered and
characterized as 4-ethanoloxybenzoic acid (an oxidation product
of the benzoyl radical of Scheme 1). We found that upon
redispersion in toluene, the particles were stable for several
weeks when stored at -5 °C. Upon standing at -5 °C, a white
precipitate is formed and characterized as excess cyclohexy-
lammonium salt 2 present in the reaction mixture. Sometimes
after removal of the sample from the fridge and re-equilibration
to room temperature the particles decomposed quickly giving
a brown solution. We believe the stability following removal
to room temperature is an indication of how much salt 2 had
(44) Link, S.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103, 8410–8426.
9
J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009 13975

A R T I C L E S
Maretti et al.
Table 1. Fluorescence Quantum Yields in Toluene for Various
AgFNP Determined Using Fluorescein as a Reference Standard
and for Different Synthesis Solvents
a
solvent
synthesis amine
ΦF1 ((10%)
toluene
toluene
THF
hexadecylamine
cyclohexylamine
cyclohexylamine
0.11
0.15
0.11
a Fluorescein standard. Excitation wavelength 450 nm.
photodecomposition of I-2959 (these products emit weakly at
465 nm). Instead, the fluorescence spectra obtained in toluene
solutions of samples originally prepared in THF reflect AgFNP
that have been purified by the cooling/filtration process.
The determination of quantum yields was carried out using
fluorescein as an emission standard. It is important to select a
reference compound with an adequate spectral overlap with the
material under study.5 Relevant spectra are shown in the
Supporting Information. We note that all too frequently
Rhodamine 6G is used as a reference, even when it has little or
no overlap with the emission of the sample under study.5
Equation 1 shows the expression used for the case of toluene
and includes an appropriate refractive index correction.
Figure 7. Fluorescence lifetime of Ag particles in toluene, as prepared in
Figure 1 and diluted to an appropriate concentration, with excitation at 440
nm and detection at 550 nm (red). The instrument response function (black)
is also included.
Isample
Aref
η2
η2
sample
Φsample ) Φref
×
×
×
(1)
Iref
Asample
ref
where Φref is the known quantum yield of the reference
compound, Isample and Iref are the integrated fluorescence intensi-
ties of the sample and reference, respectively, Aref and Asample
are the absorbance of the reference and the sample at the
excitation wavelength, and ηsample and ηref are the refractive index
of the sample solvent and reference solvent, respectively.
The fluorescence quantum yields obtained are summarized
in Table 1 and are based on the integrated fluorescence between
470 and 700 nm. The absorbance of each sample and the
fluorescein reference at 450 nm (excitation wavelength) was
0.090 (Figure S4).
Figure 8. 19F NMR spectra of concentrated (A) and dilute (B)
hexadecylamine-Ag particles following the cooling/filtration procedure
described in the text. Spectrum C is the independently prepared hexade-
cylammonium trifluoroacetate salt. All samples were prepared in toluene-
d8, and additional details are included in the Experimental Section.
The fluorescence quantum yields obtained are quite high,
making their emission readily observable under weak irradiation
conditions, such as those available from laboratory TLC lamps.
An example of the observed emission is presented in the Table
of Contents graphic for this contribution.
Fluorescence Lifetimes. In our laboratory we are used to
examining fluorescence decay behavior in quantum dots, where
complex decays requiring lifetime distribution analysis are the
rule. We were thus surprised to discover that in the case of
AgFNP the decays (>98%) could be readily fitted with a
monoexponential expression. In the case of toluene the decay
was fit to a monoexponential function with a lifetime of 2.6 (
0.1 ns and a goodness of fit ꢀ2 ) 1.46. The corresponding decay
and instrument functions are shown in Figure 7. Similar lifetimes
have been recorded in DNA encapsulated systems,13 while in
argon matrixes the lifetime of Ag2 clusters has been reported
as 4.6 ns.45
presence of photochemically generated ketyl radicals but the
product is not fluorescent. 19F NMR is a sensitive technique
offering excellent diagnostics for the environment of 19F atoms.
The 19F NMR data for AgFNP prepared in toluene with
hexadecylamine are shown in Figure 8. Spectrum C shows a
reference spectrum from hexadecylammonium trifluoroacetate.
Spectrum A clearly shows that the 19F peak is considerably
shifted to -73.76 ppm (compared to -76.71 ppm for the
reference). As the sample is diluted (spectrum B), the relative
importance of the reference peak also increases. In both
spectrum A and spectrum B, a significant broadening of the
peak centered at -73.76 is observed.
A number of important conclusions can be drawn from the
NMR spectra of Figure 8. First, the trifluoroacetate anion is in
a different environment when the AgFNP are present, as would
be expected if they form part of a nanoparticle stabilizing layer.
Second, the increased peak width is consistent with restricted
mobility and an exchange time similar to or longer than the
time scale of the NMR response. Third, the fact that the
proportion of free anion increases upon dilution is a clear
indication of a dynamic equilibrium between associated and free
ions. Our experiments do not show if all, or only part, of the
stabilizing anions are exchangeable.
NMR Spectroscopy. Our synthesis experiments showed
unequivocally that the presence of trifluoroacetate anions was
critical for the preparation of highly fluorescent AgFNP. That
is, while the reduction of Ag+ by I-2959 would occur regardless
of the presence of this anion, the particles would not normally
fluoresce. For example, silver nitrate is readily reduced in the
NMR experiments were also performed when the stabilizing
salt was based on cyclohexylamine; the corresponding spec-
(45) Konig, L.; Rabin, I.; Schulze, W.; Ertl, G. Science 1996, 274, 1353–
1355.
9
13976 J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009

Highly Fluorescent Silver Nanoparticles
A R T I C L E S
Figure 10. Fluorescence microscopy image (100×) of a polystyrene (PS)
film containing hexadecylamine-AgFNP with (microscope) excitation at
436 nm and broadband detection at λ > 470 nm. The sample was prepared
by UVA photolysis of a spin-coated PS film containing 20 mM silver
trifluoroacetate, 20 mM I-2959, and hexadecylamine for 20 min. Spot size
is diffraction-limited (>300 nm) and shows correctly particle location but
not dimensions.
Figure 9. TEM images of AgFNPs prepared using different procedures:
(A) particles synthesized in THF with cyclohexylamine, redispersed in
toluene, and dropcast from toluene; (B) particles synthesized in toluene
with cyclohexylamine, purified as discussed in the text, and dropcast from
toluene; (C) particles synthesized in toluene with hexadecylamine, purified
as discussed in the text, and dropcast from toluene; (D) HR-TEM image of
a large particle in sample a. Scale bars are 10, 20, 20, and 2 nm for A, B,
C, and D, respectively.
Fluorescence Microscopy in Polymer Films. While this study
is at present at a very preliminary stage and its lithographic
applications will be the subject of a future report, fluorescence
microscopy images in polystyrene films (as close as possible
to the toluene solvent used here) were used to explore the nature
of the fluorescent material studied here. For this purpose,
solutions of similar composition as those described above, but
containing polystyrene, were spin-coated on fused silica disks
and exposed to UVA light. Figure 10 shows an image obtained
in this manner.
Figure 10 demonstrates unequivocally that the emission arises
from discrete particles and thus Ag emissive clusters must be
located at or near the AgNP surface. The fact that small
(predominantly Ag2) fluorescent clusters are associated with
larger nanoparticles is also consistent with the fluorescence
quenching studies reported in the next section. While our data
are consistent with the absorption, emission spectra, and
fluorescence lifetime for Ag2, we cannot rule out possible
involvement of species such as Ag3 and Ag4. A fluorescence
spectrum of a single fluorescing particle from Figure 10 has
been included in the Supporting Information.
Fluorescence Quenching Studies. Our recent studies of CdSe
quantum dots4,46-48 have led us to the conclusion that nitroxides
are excellent quenchers of nanoparticle fluorescence and that
these studies can yield information on the form of quenching,
the nature of binding interactions, the ligand environment at
the nanoparticle surface, and the dynamics of quencher-surface
interactions. Among the nitroxides tested, 4-amino-TEMPO,
4aT, proved very convenient, with excellent surface binding
properties.4,47 Amines are known to bind well to noble metal
particles.8
troscopic data are included in the Supporting Information.
Briefly, similar trends are observed in all systems examined,
but for cyclohexylamine the 19F peak is broader and shows some
shifts with concentration, suggesting that more than one site
may be available to the trifluoroacetate anion. The fact that the
behavior depends on the counterion (i.e., the protonated amine)
suggests that the ammonium ion must also be part of the
stabilizing layer, as suggested in Scheme 1.
1
Attempts to monitor H NMR for the amine were inconclu-
sive, although it is clear that the presence of AgFNP influenced
the spectrum, also consistent with the amine forming part of
the particle-stabilizing layer (Figure S6b). There are literature
reports indicating that base sites may contribute to AgFNP
stabilization.29
Nanoparticle TEM Imaging. TEM imaging experiments were
performed on samples dispersed in toluene, using both solutions
obtained directly in this solvent, as well as particles originally
made in THF and then resuspended in toluene, as described in
the AgFNP synthesis section. This purification procedure was
found to minimize the presence of organic crystals in the image.
Measurement of a range of particles prepared in THF leads
to an average size of (3.4 ( 0.7) nm for AgFNP obtained by
the same protocol as in Figure 9 (see histogram in Supporting
Information).
Analysis of the crystal lattice spacing clearly shows Ag(1,1,1)
with its characteristic 0.236 nm spacing (Figure 9d). We note
that while some single crystal particles are detected, the majority
proved polycrystalline. EDS characterization of the particles
showed that they are composed of metallic silver, as expected.
XPS measurements were attempted on the same samples used
for TEM microscopy but only confirmed the presence of Ag0
and Ag1+, the latter likely from residual unreacted silver
trifluoroacetate.
Stern-Volmer analysis is a common technique used to
acquire information regarding fluorescence quenching mech-
(46) Laferriere, M.; Galian, R. E.; Maurel, V.; Scaiano, J. C. Chem.
Commun. 2006, 257–259.
(47) Maurel, V.; Laferriere, M.; Billone, P.; Godin, R.; Scaiano, J. C. J.
Phys. Chem. B 2006, 110, 16353–16358.
(48) Scaiano, J. C.; Laferriere, M.; Galian, R. E.; Maurel, V.; Billone, P.
Phys. Status Solidi A 2006, 203, 1337–1343.
9
J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009 13977

A R T I C L E S
Maretti et al.
anisms49,50 and has proven useful in the interpretation of the
quenching experiments with CdSe quantum dots.47,48,51,52
Briefly, when the mechanism for deactivation of an excited state,
fluorescent in the present case, is a simple competition between
first-order processes that determine the excited state lifetime, τ
(fluorescence, intersystem crossing, thermal deactivation), and
a bimolecular quenching process by some quencher Q with rate
constant kq, the fluorescence intensity in the absence of quencher,
F0, and in the presence of quencher, F, can be related by eq 2,
where kqτ ) KSV.50
Figure 11. Stern-Volmer plot of fluorescence quenching of AgFNPs with
4aT in toluene. When analyzed using eq 2, the quenching rate constants
obtained at low 4aT concentrations are above the diffusion-controlled rates,
indicating a static quenching mechanism involving binding to the particle
surface. The efficiency of quenching is significantly higher in the absence
of cyclohexylamine (b) than in the presence of 40 µM cyclohexylamine
(9) which implies a competition for surface sites between 4aT and
cyclohexylamine.
F0
) 1 + kqτ[Q] ) 1 + KSV[Q]
(2)
F
Quenching experiments were performed in toluene with
AgFNP prepared directly in this solvent. The corresponding
Stern-Volmer quenching plot is shown in Figure 11 and also
includes a 4aT quenching plot recorded in the presence of 40
µM cyclohexylamine. The fact that cyclohexylamine attenuates
the quenching by 4aT is a clear indication that the amine groups
in 4aT and cyclohexylamine compete for binding sites at the
AgFNP surface. The data has been fit with a parabola for
convenience; the linear coefficient of the fit corresponds to the
initial slope (i.e., lim[4aT] f 0), yielding values of 44 200 M-1
and 30 100 M-1 in the absence and presence of 40 µM
cyclohexylamine, respectively.
It is important to note that both primary amine and free radical
functionalities are essential for quenching where the amine group
provides only a binding interaction, as cyclohexylamine addition
by itself has almost no effect on fluorescence. On the other hand,
TEMPO itself is a poor quencher, and 15 mM TEMPO leads
to a quenching ratio F0/F of 2.6; the same value can be achieved
with ∼30 µM 4aT (see Figure 11). Thus, 4aT is approximately
400 times more efficient than TEMPO as a quencher. Quenching
data for TEMPO have been included in the Supporting Informa-
tion.
Figure 12. (A) EPR spectra of 18.0 µM 4aT in deaerated toluene; (B)
deaerated toluene solution of concentrated AgFNPs (A448 ) 1.3) and 18.0
µM 4aT. A decrease in the relative intensity of the high-field feature of
4aT in the presence of AgFNPs is observed. All EPR parameters for both
spectra are included in the Experimental Section.
Given Ksv ) kqτ, where kq is the bimolecular rate constant
for excited state quenching and τ the excited state lifetime, then
it is possible to extract kq, given the knowledge of τ from
fluorescence studies (τ ∼ 2.6 ns in toluene). This analysis yields
a value of kq ) 1.7 × 1013 M-1 s-1 (from the initial slope in
Figure 11). This number is 3 orders of magnitude larger than
that for a diffusion-controlled process in toluene.49 This type
of behavior is normally referred to as “static quenching”, that
is, diffusion is not required for quenching to occur, with the
quencher already bound to the fluorophore before excitation
takes place. In our system this requires 4aT to bind strongly to
the AgFNP, in a similar fashion that 4aT binds to quantum
dots.4,46 Figure S6 shows fluorescence lifetime measurements
in the absence of 4aT and in the presence of 17 µM 4aT,
corresponding to the F0/F value of 2.1 in Figure 11. The fact
that the fluorescence lifetime remains unchanged, while the
maximum intensity decreases, is also indicative of a static
quenching mechanism. The reasons for the upward curvature
at moderately high 4aT concentrations are analyzed in the
Discussion.
EPR Studies. EPR experiments were performed to observe
if binding to the surface of the Ag particles was showing the
same type of anisotropic line broadening observed in the case
of 4aT bound to CdSe quantum dots.47 As seen in Figure 12,
there is no broadening of the low and midfield nitroxide peaks
upon addition of concentrated Ag particle solution. However,
there is a relative decrease in the peak-to-peak height of the
high-field line, which was reproducible with many different
batches of particles. Control experiments were performed by
addition of synthetically prepared cyclohexylammonium trif-
luoroacetate, and no effects on the isotropic 4aT spectrum were
observed. Therefore, we conclude that the small degree of
anisotropy observed in Figure 12B is a result of 4aT binding to
(49) Turro, N. J.; Ramamurthy, V.; Scaiano, J. C. Principles of Molecular
Photochemistry: An Introduction; University Science Publishers: New
York, 2008.
(50) Turro, N. J. Modern Molecular Photochemistry; Benjamin/Cummings
Publishing Co.: Menlo Park, 1978.
(51) Billone, P. S.; Maretti, L.; Maurel, V.; Scaiano, J. C. J. Am. Chem.
Soc. 2007, 129, 14150–14151.
(52) Jin, W. J.; Fernandez-Arguelles, M. T.; Costa-Fernandez, J. M.; Pereiro,
R.; Sanz-Medel, A. Chem. Commun. 2005, 883–885.
9
13978 J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009

Highly Fluorescent Silver Nanoparticles
A R T I C L E S
the surface of the Ag particles.53,54 The absence of broadening
on the two low-field features suggests that the local environment
of the bound nitroxide is close to that of free nitroxide where
its axial degrees of rotational freedom are not hindered by a
closely packed ligand environment.55
with those synthesized in THF and then transferred into toluene.
As a functional material, the description as 3.4 nm fluorescent
silver nanoparticles is accurate but somewhat misleading. Metal
nanoparticles of this size are not expected to fluoresce, and
plasmon absorptions do not have a corresponding emission band.
We believe that the emissions are due to small silver clusters,
predominantly Ag2 supported by the readily detectable nano-
particles and stabilized by the presence of protonated amines
and trifluoroacetate anions. Further support for the presence of
small fluorescent clusters comes from the remarkably clean
monoexponential decay of the fluorescence, resembling that of
simple molecules and a characteristic that is not usually observed
with fluorescent nanoparticles, such as quantum dots, where
fluorescence decay follows complex decay patterns. Other
researchers have detected fluorescence in the 2-3 ns range and
attributed it to Ag2 clusters.13,17,59 We have verified that any
Ag+ complexes that may form with either the starting reagents
or any of the photoproducts are nonfluorescent, precluding the
possibility of silver complexes contributing to the observed
emission. We have also observed that the organic photoproducts
themselves, which have been isolated, are not fluorescent either.
Small clusters are known to have a tendency for agglomera-
tion. These aggregates of clusters still retain the key emissive
properties of the individual clusters, although not surprisingly
the solution behaviors lack the sharp excitation and emission
profiles17 that characterize small clusters in matrixes at cryogenic
temperatures. The fact that these clusters, most likely Ag2,
survive on the nanoparticle surface may reflect the intrinsic bond
strength of 160 kJ/mol in diatomic silver and further stabilization
due to the amine and trifluoroacetate moieties.
A few studies have also noted that an excess of positive
charges tends to facilitate particle or cluster stabilization.18,39,60
In our case NMR provides unequivocal evidence that trifluo-
racetate ions are part of the surface coverage and perhaps a few
ions are needed to neutralize excess Ag+ charges. However,
ammonium ions derived from the amine are clearly also on the
surface. Most likely it is these hydrophobic groups (e.g., from
hexadecylamine) that ultimately provide the compatibility with
a very nonpolar solvent such as toluene. Small amines appear
to fail to provide equivalent stabilization.
Once the particles have been purified and the excess salt (such
as cyclohexylammonium trifluoroacetate) separated, the material
retains some affinity for free amines. In the case of nitroxide
free radicals, for instance, we see much higher quenching
efficiency with 4aT than with TEMPO, which lacks the primary
amine functionality. The quenching by 4aT is remarkably
efficient, even when compared with the quenching of quantum
dot emission by the same quencher.4,47 These results can only
be explained by invoking static quenching in which 4aT is
strongly bound to the AgFNP surfaces, which appears to provide
excellent access to the fluorescent clusters these particles
support. The upward curvature (Figure 11), while small, is rather
intriguing. In the case of quantum dots, quenching slots by 4aT
leads to negative (downward) curvature, attributed to the
replacement of TOPO (trioctylphosphine oxide) at high con-
centration, while in the low concentration regime, 4aT binds to
available sites.4,47,51 Using a similar rationale for AgFNP, one
would conclude that the more difficult sites to access (i.e., at
higher concentrations) are also the ones in closest proximity to
Discussion
Numerous methods are available for the reliable synthesis
of metal nanoparticles, including silver.6,56 Most of these
methods employ reductants such as borohydride and stabilize
the particles in solution with a variety of coatings to avoid or
minimize precipitation.32,57 Some of these protected layers can
have biological functions incorporated. Typically silver nano-
particles show a distinct plasmon band which in our past work
has been between 390 and 420 nm.8,33 We were initially
surprised when the absorption band obtained was closer to 450
nm (see Figure 1). We now believe that this absorbance is not
due to the silver plasmon band but rather to the presence of
small silver clusters. Previous studies have identified absorbance
bands at 385, 410, and 442 nm, depending on their local
geometries, as belonging to Ag dimers in Ar matrixes,22 the
latter which corresponds well to the absorption we observe in
the present case at around 450 nm. As evident from the TEM
images (Figure 9) and the observed weak absorption band at
390 nm in certain samples, large (3.4 nm) particles are formed
in the photoreaction. We have been unable to determine the
chemical yield of the small silver clusters versus the yield of
larger nanoparticles. While the spectroscopic data for the
particles are well documented58 (and gives about 0.5 nM
particles), that for the clusters is not; further, it is unclear if
surface enhancement effects may influence the cluster optical
properties. Gel permeation chromatography (GPC) was used in
an attempt to verify that the fluorescence of the samples was
associated with species attached to the larger nanoparticles but
the fluorescence was unstable under these conditions, possibly
due to on-column interactions.
In our case we have selected a photochemical reaction in order
to generate the reducing species required, as shown in Scheme
1. While photochemical methods have been used before, the
choice of chromophore and wavelength are frequently not
optimized; the rationale for making suitable choices was
discussed in a recent contribution.35 Our approach has been to
generate well-established reductive species, such as ketyl
radicals, under conditions where the precursor (I-2959 in our
case)34,35 has strong absorptions, and where radicals are
produced in fast processes that minimize or preclude quenching
events. For example, although Ag+ is an excellent excited-state
quencher,33,38 it is unable to quench significantly the triplet state
of I-2959, which has a lifetime of just a few nanoseconds.36
Our transmission electron microscopy measurements reveal
particles of about 3.4 mm in diameter, somewhat more poly-
disperse when they are prepared directly in toluene, compared
(53) Wertz, J. E.; Bolton, J. R. Electron Spin Resonance. Elementary Theory
and Practical Applications; McGraw-Hill: Toronto, 1972.
(54) Weil, J. A.; Bolton, J. R. Electron Paramagnetic Resonance:
Elementary Theory and Practical Applications, 2nd ed.; Wiley-
Interscience: Hoboken, NJ, 2007.
(55) Chechik, V.; Wellsted, H. J.; Korte, A.; Gilbert, B. C.; Caldararu, H.;
Ionita, P.; Caragheorgheopol, A. Faraday Disc. 2004, 125, 279–291.
(56) Eustis, S.; El-Sayed, M. A. Chem. Soc. ReV. 2006, 35, 209–217.
(57) Mirkhalaf, F.; Paprotny, J.; Schiffrin, D. J. J. Am. Chem. Soc. 2006,
128, 7400–7401.
(59) Fedrigo, S.; Harbich, W.; Duttet, J. J. Chem. Phys. 1991, 99a, 5712.
(60) Messere, A. G., A.; Garella, I.; Temussi, F.; Di Blasio, B.; Fiorentino,
A. Synth. Commun. 2004, 34, 3317–3324.
(58) Link, S.; Wang, Z. L.; El-Sayed, M. A. J. Phys. Chem. B 1999, 103,
3529–3533.
9
J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009 13979

A R T I C L E S
Maretti et al.
the emissive clusters supported by the nanoparticles, thus leading
to more efficient quenching (upward curvature). In any event
the effect is small and quenching remarkably effective, even
more so when the short excited state lifetime (∼2.6 ns) is taken
into account.
nonpolar organic media may have applications in polymeric
functional materials; further, photochemical synthesis provides
a route for imaging and the efficient fluorescence a mechanism
for mapping material distribution. These nanoparticle-supported
clusters may also have interesting catalytic properties.
One may ask if small silver clusters, specially Ag2, are free
in solution or associated with AgNP such as imaged by TEM
and shown in Figure 9; as already indicated, we favor the latter
explanation. The data in Figures 10 and 11 provide support for
this interpretation. Quenching data support that 4aT binds
strongly to the luminescent species (i.e., quenches efficiently),
and earlier work shows strong binding to nanoparticles.4,8,47,48,61
Thus, when we add 40 µM cyclohexylamine, as it binds to the
nanoparticle surface we would expect more 4aT to become
available in the solution and less at the particle surface. Thus,
if the emissive Ag2 was located in the solution, addition of
cyclohexylamine would lead to enhanced quenching by 4aT,
since displacement of 4aT from the nanoparticle surface would
make more available for quenching in the solution. In contrast,
we observe reduced quenching, consistent with the fact that
cyclohexylamine displacement of 4aT from the surface also
reduces abundance of quenchers in the proximity of the
fluorescent clusters. Further, it would be hard to see how Ag2
free in solution could lead to curved Stern-Volmer plots, since
Ag2 should show molecular-type quenching behavior.
Fluorescence microscopy images in polystyrene films (a
polymer structurally similar to the toluene solvent used here)
demonstrate unequivocally that the emission arises from discrete
particles, thus also supporting the conclusion that the emissive
clusters are located at or near the AgNP surface. One such image
is shown in Figure 10.
In conclusion, we recently pointed out that ketones are good
sensitizers for nanoparticle synthesis not because of the energy
they can deliver but rather because of the free radicals they can
generate.35 The synthesis and characterization of AgFNP shows
the generality of these concepts and the fact that ketyl radicals
are excellent tools for nanoparticle synthesis. In the example
reported here, nanoparticles synthesized are around 3 nm in
diameter but their fluorescent properties are derived from small
silver clusters, predominantly Ag2, that are believed to be located
at the nanoparticle surface or stabilization layer. It is possible
that interaction with the surface contribute to the high emission
quantum yield observed (0.11-0.15). The new AgNFP are
interesting functional materials with potential applications in
imaging and nanophotonics. Their excellent long-term stability
and excellent solubility in organic solvents may prove key to
future applications.
Acknowledgment. We acknowledge financial support from the
Natural Sciences and Engineering Research Council of Canada
(NSERC), CFI, and the Government of Ontario. Thanks are due to
Professor Gonzalo Cosa and Mr. Pierre Karam (McGill University)
for their help in acquiring the spectrum in Figure S10.
Supporting Information Available: Absorption spectra, com-
parison of excitation and emission spectra in THF and in toluene,
spectral data used to obtain fluorescence quantum yields, the
fluorescence lifetime in the presence of 4aT, 19F and H NMR
data, the histogram of particle size distribution and Stern-Volmer
plot of fluorescence quenching by TEMPO. This material is
available free of charge via the Internet at http://pubs.acs.org.
In the past few years a number of contributions have reported
on the bright emission from small silver clusters;11 their
properties are such that aqueous systems are interesting from a
biological perspective and have already been examined in DNA
and in cells.13,27,29,62,63 On the other hand, AgFNP soluble in
JA900201K
(61) Thomas, K. G.; Kamat, P. V. J. Am. Chem. Soc. 2000, 122, 2655–
2656.
(63) Richards, C. I.; Choi, S.; Hsiang, J. C.; Antoku, Y.; Vosch, T.;
Bongiorno, A.; Tzeng, Y. L.; Dickson, R. M. J. Am. Chem. Soc. 2008,
130, 5038.
(62) Ritchie, C. M.; Johnsen, K. R.; Kiser, J. R.; Antoku, Y.; Dickson,
R. M.; Petty, J. T. J. Phys. Chem. C 2007, 111, 175–181.
9
13980 J. AM. CHEM. SOC. VOL. 131, NO. 39, 2009