Colloids and Surfaces B: Biointerfaces 109 (2013) 25– 31
Contents lists available at SciVerse ScienceDirect
Colloids and Surfaces B: Biointerfaces
jou rn al hom epage: www.elsev ier .com/ locate /co lsur fb
comparative study of economical separation and aggregationroperties of biologically capped and thiol functionalized goldanoparticles: Selecting the eco-friendly trojan horses foriological applications
.Shankaran Nair Research Institute for Nanotechnology and Bionanotechnology, Ambernath, MS, India
a r t i c l e i n f o
rticle history:eceived 25 October 2012eceived in revised form 3 February 2013ccepted 5 March 2013vailable online 28 March 2013
a b s t r a c t
We are presenting facile bio-fabrication of extremely stable gold nanoparticles (GNPs) using medicinalplant Azadirachta indica (commonly called Neem) and its comparison with most commonly used glu-tathione (GSH) protected GNPs in terms of stability under physiological conditions, seperation usingdensity gradient centrifugation and aggregation properties in the solution. There was dual peak at 536and 662 nm indicating the presence of non-spherical GNPs including triangles, rods and hexagons in case
eywords:old nanoparticlesucrose density gradient centrifugationlocculation parameterzadirachta indicalutathione
of A. indica mediated GNPs unlike citrate stabilized GNPs which exhibited single sharp peak. SphericalGNPs were separated from the consortium of uniquely shaped nanoparticles bio-fabricated using A. indicaleaf extract using sucrose density gradient centrifugation (SDGC).To comprehend the anti-agglomerationpotentials of A. indica leaf mediated GNPs and GSH-GNPs under physiological conditions, flocculationparameters (FP) were calculated and found to be least for A. indica leaf mediated GNPs, indicating theirexceptional stability.
. Introduction
Metal nanoparticles lack some of the cardinal attributes, whichake them incompatible for biological applications, such as fer-
ying drugs to the target and sensing inimical biomolecules. Mostommon fundamental problems associated with metal nanoparti-les are their reduced solubility in water or serum [1], toxicity [2],ifficulty in tuning the desired size and shape for many hemody-amic constraints [3]. Most of the chemical protocols optimizedo synthesize plethora of metal nanoparticles such as gold and sil-er, involve noxious materials as precursors, making them hostileor biological systems. The use of materials like plants [4], bacteria5] and fungi [6] offer numerous benefits because of their inherentroperties to fabricate metal nanoparticles without involving toxicrecursors.
Biologically synthesized GNPs exhibit exceptional stability as
ell as biocompatibility due to charismatic orchestrations of their
urfaces by biological peptides. Moreover, biological peptides can
also be exploited as natural linkers for anchoring drugs and othertherapeutic moieties [7].
Biological peptides can be a potential alternative to earlierattempts, in order to stabilize the GNPs in the solution for variousbiochemical and electroanalytical applications [8–10]. There aremany promising protocols for protecting surfaces of GNPs, includ-ing use of thiolated ligands such as poly (ethylene glycol) [8,11],silica encapsulation of metal clusters [12], polymeric cappingagents such as amino- or mercaptodextran [13] and Bovine SerumAlbumin [14]. However, all these methods suffer lack of genericprotocols for the functionalization of particles with biomolecules.
Amongst variety of biological systems used as molecular ves-sels to make GNPs, medicinal plants can be considered as themost reliable sources due to their exceptional biocompatibilityand lucid protocols involved in the synthesis of GNPs and othernanoparticles. During the synthesis of GNPs using plant extractssuch as Azadirachta indica leaf in our case, a consortium of uniquelyshaped GNPs such as triangles, cubes, hexagons and roughly spher-ical structures are predominant. For crucial applications like drugdelivery spherical nanoparticles of suitable size and shape must be
separated from the mixture [15]. Some of the purification and sep-aration techniques involve capillary electrophoresis [16], columnchromatography [17] and size selective precipitation [18]. Sedi-mentation coefficient differences between nanoparticles have been
xploited for separation of nanoparticles using centrifugation [19].ensity gradient centrifugation has proven to be successful in sep-ration of anisotropic nanoparticles from spherical ones or sortinganoparticles of different aggregation states [19,20].
In this article, we report the comparative analysis of biogenicnd chemogenic GNPs as ideal molecular armadas for carrying ther-peutic payloads to the cells and other green applications.
Following steps have been performed to prove the efficacy ofoth types of nanoparticles as an ideal nano-vessel to carry drugsoward solid tumors:
Designing an efficient protocol for the fabrication of nanoparticles:or synthesis of GNPs, we used A. indica leaf extract and citrateeduction method.
Surface protection of synthesized GNPs: It is a well known facthat the GNPs synthesized using plants are capped with proteinsnd/or enzymes to confer exceptional stability in the aqueousnvironment. Citrate stabilized GNPs were functionalized usinglutathione (GSH) at pH 7.2.
Separation of the nanoparticles using Sucrose density gradient cen-rifugation: In order to separate nanoparticles based on their sizend shape, SDGC was used as efficient method.
Flocculation studies of GNPs: These studies were performed toomprehend the efficiency of the linker or surface proteins to resisthe flocculation of the nanoparticles under physiological pH (7.2).
. Materials and methods
.1. Materials
Gold aurochlorate, tri-sodium citrate and sodium chloride wereurchased from Sigma–Aldrich, USA. All the experiments were car-ied out in ultrapure water (18 M�). In order to remove the tracesf metal contaminants glasswares were washed with aqua regia.
.2. Preparation of the plant extract
A. indica leaves were subjected to mild ultrasonication beforeaking the extract to remove contaminants which may interfereith results. For preparation of extract, 5 g of leaves were crushed
n 20 ml of distilled water and ultra filtered using 0.22 � filters to getlear extract and also to remove cellular debris. For all the experi-ental considerations, extract was diluted 100 times using double
istilled water. This extract was stored at 4 ◦C till further use.
.3. Synthesis of GNPs
A solution of 50,000 ppm gold aurochlorate was prepared inltrapure water. This stock solution was diluted to 100 ppm andsed to synthesize GNPs. For synthesis of GNPs, 20 �l of 50,000 ppmold salt solution was added in 10 ml boiling solution of reactionessel containing diluted plant extract (pH 6.14). The extract wasoiled till the appearance of wine red color. Chemogenic GNPs wereynthesized using citrate reduction method [21].
.4. Characterization
UV–vis Spectroscopy (Lambda 25 PerkinElmer, USA) was car-ied out using plant extract as the reference. Spectra were recordedsing a clean quartz cuvette having a path length of 1 cm. Exam-
nation of the morphology of GNP was done using field emissioncanning electron microscopy (FE-SEM) on a Carl Zeiss Microimag-ng, GmbH, Germany. 2–3 drops of the colloidal gold solution were
ispensed onto a silicon wafer and dried under ambient conditionefore examination. Transmission electron microscopy (TEM) on aarl Zeiss Micro imaging, GmbH, Germany was done by dispens-
ng few drops of colloidal gold solution onto a formwar-coated
: Biointerfaces 109 (2013) 25– 31
200-mesh copper grid. Involvement of diverse functional groupsand molecular interactions as well as molecular orientation of thecomplexes was verified using Fourier transformed infra red spec-troscopy (FTIR) on a MAGNA-550, Nicolet instruments, USA. Thesample preparation for this was done by loading 0.1 ml of each ofthe biogenic and thiol functionalized chemogenic GNPs in aqueousform onto the source.
2.5. Functionalization of GNPs using glutathione
Aqueous solution of GSH of 3000 ppm was prepared by dissolv-ing 0.03 g of GSH in 10 ml of nano-pure water. 300 �l of GNPs wereadded to 500 �l of GSH solution and total volume was made 3 mlto maintain ratio of GNPs to GSH as 1:50.
2.6. Stability test of biogenic GNPs and thiol functionalized GNPsusing flocculation parameter
To comprehend the changes in optical properties of both (bio-logically as well as chemically treated) GNPs in response to varyingconcentrations of NaCl, 10 �l of 5 M NaCl solution was added incuvette containing respective GNP solutions and UV–vis spectrawas recorded. This procedure was repeated till there is consider-able red shift in the peaks as compared to initial spectrum (beforeadding salt).
Aggregation property of the GNPs in solution was studied usingan empirical term called “Flocculation Parameter” which is mea-surement of integrated absorbance between longer wavelengths(500–600 nm in case of GNPs) [22]. Following equation was usedto compute the integrated absorbance:
P =800∫
600
IAbs(�)d� (1)
2.7. Density gradient centrifugation (DGC) of biogenic GNPs forpurification of spherical nanoparticles
2 ml of 10–60% sucrose solution (w/v) was layered successivelyin a 20 ml centrifuge tube starting from highest to lowest concen-tration. The gradient was top layered with 2 ml of respective GNPsolutions and centrifuged at 4000 rpm for 40 min and character-ized spectrophotometrically. Size and shapes of separated fractionswere studied after selecting a representative field during electronmicroscopic studies.
3. Results and discussions
3.1. Synthesis of GNPs
GNPs synthesized using ultra-filtered extract of A. indica leafexhibited a sharp peak at 536 nm and 662 m (Fig. 1a) with aprominent wine red color with bluish tinge on transmission, asa consequence of surface plasmon resonance (SPR). Dual peakscorrespond to transverse surface plasmon resonance (536 nm)and longitudinal surface plasmon resonance (662 nm) respectively.These peaks are indicative of presence of anisotropic GNPs and/ortheir agglomeration in the solution [23]. There can be many possibleexplanations for the anisotropy exhibited by the GNPs fabricatedusing A. indica leaf extract which are as follows:
1. Initiation of GNP synthesis in solution happens by formation of
the tiny clusters which behave as nuclei to guide further growth.The process of nucleation becomes ultra-fast due to depleted sol-ubility of the gold in aqueous extract of A. indica [24]. During thenucleation process of GNPs, surface energies of the nanoparticles
S. Pandey et al. / Colloids and Surfaces B: Biointerfaces 109 (2013) 25– 31 27
Fig. 1. (a) UV–vis spectroscopy of GNPs synthesized using A. indica leaf, the dual peak at 536 and 662 nm indicating presence of anisotropic GNPs in the solution (Inset displaysF e of anp t 523
s
2
3
a
Fp
E-SEM image of GNPs fabricated using A. indica leaf extract displaying the presencroteins) (b) UV–vis spectra of chemically synthesized GNPs showing sharp peak apherical nanoparticles).
are the prime factors which navigate the directional ripening ofmetal nanocrystals.
. Minimization of the surface energies of the nanoparticlesdepletes the surface area and lead to the fabrication of sphericalnanoparticles.
. Presence of the ions as well as capping agents such as shortpolypeptides in the extract of A. indica alter this surface energyand guide the growth of the crystal facets to be non-sphericalincluding nano-prisms and hexagons. The pattern of formationof nanoparticles may be similar to template assisted synthesis of
anisotropic metal nanoparticles [25,26].
On the other hand, citrate stabilized GNPs exhibited a sharp peakt 523 nm with very small area under the peak (Fig. 1b). Less area
ig. 2. UV–vis spectra of the fractions collected after sucrose density gradient centrifugaer Fig. 3.
isotropic nanoparticles. White arrow indicates the capping of the nanoparticles bynm (Inset shows HRTEM image of citrate stabilized GNPs showing the presence of
under the peak explains the homogeneous nature of the nanopar-ticles. Origin of such sharp peak along with the color changefrom pale yellow to wine red is due to the quantum mechanicalphenomenon called SPR as explained earlier. This unique opticalphenomenon is observed in GNPs due to entrapment of electronsin the nano-boxes resulting in quantum confinement effect. Unlikeprevious case, least anisotropy can be seen in GNPs synthesizedusing citrate.
FE-SEM image of GNPs synthesized using leaf extracts of A.indica (Inset of Fig. 1a) displays presence of gold nano-plates,
nano-triangles, nano-hexagons and nano-rods ranging from 10 to60 nm. Closer observation reveals a protective covering aroundthe nanoparticles (indicated by white arrow). These protectivecoverings by certain peptides confer exceptional stability to the
tion in case of biogenic and chemogenic nanoparticles. The fractions are named as
28 S. Pandey et al. / Colloids and Surfaces B: Biointerfaces 109 (2013) 25– 31
Fig. 3. Sucrose density gradient centrifugation of biogenic and GSH functionalized GNPs (i) Zones after the separation of the biogenic GNPs clearly showing the separationof the particles based on their size (ii) a, b and c, SEM image of biogenic GNPs separated from fractions a, b and c respectively (Scale bar 100 nm) (iii) Histogram displayingthe % distribution of biogenic GNPs of different shapes present in fractions a, b and c respectively (iv) Zones after separation of GSH functionalized GNPs (v) d and e TEMi e (Scaa
nis
3c
nwco(sp
astos∼tc
r(n
mage of GSH functionalized GNPs separated from fractions d (Scale bar 20 nm) and representative field shown in the image.
anoparticles in the solution. Inset of Fig. 1b displays the HRTEMmages of GNPs synthesized using citrate reduction method. Theize of the nanoparticles were found to be from 5 to 40 nm.
.2. Seperation of GNPs using sucrose density grediententrifugation
The choice of DGC for the separation of faceted GNPs such asano-triangles is due to high buoyant density of such structureshich disqualify separation by conventional equilibrium isopycnic
entrifugal strategies [27,28]. DGC displays noteworthy separationf the spherical GNPs based on their size and viscoelastic natureFig. 3i). Three separate bands (a, b and c) can be seen in the den-ity gradient made by overlaying different concentrations of highlyurified sucrose solution.
In fraction a, light pink colored GNPs displayed a sharp peakt 543 nm which correspond to the presence of nanoparticles ofize 40–60 nm (Fig. 2a). As seen in the SEM image (Fig. 3iia), clus-ered nanoparticles can be seen including triangles, spheres andther non-spherical nanoparticles. The percentage of the triangles,pheres and other non-spherical nanoparticles was calculated to be6, 9, and 5% respectively. A closer scrutiny of the SEM image shows
he presence of capping agent around the nanoparticles which mayonfer stability in the solution.
In fraction b, color of the nanoparticles was found to be deeped with an intense peak at 639 nm and a mild hump at 559 nmFig. 2b). Splitting of the peaks into two explains the presence ofon-spherical nanoparticles in the solution which can be seen in
le bar 30 nm) respectively (vi) Histogram showing the sizes of GNPs recorded from
SEM image (Fig. 3b). The highest concentration of gold nano-rods(56%) was found in this fraction.
In fraction c, the color of the nanoparticles was royal blue indi-cating the presence of larger size GNPs. There were multiple peaksobserved in UV–vis spectra at 575, 725, 910 and a mild hump at1125 nm (Fig. 2c). The peak at 575 nm is due to the presence ofspherical nanoparticles (80%) and intensity of the peak indicates thehigh concentration of the spherical GNPs. Another peak at 725 nmis assigned to dipole resonance of gold nanotriangles (∼9%). Fur-ther red shifts in the peak (910 and 1125 nm) can be due to thepresence of hexagonal, cubical GNPs and/or due to the ripening ofthese anisotropic nanostructures [29].
SEM image (Fig. 3c) shows the presence of distinct nanotrian-gles, cubes (∼2%) and other uniquely shaped structures (∼6%).
More interesting results were observed in case of citrate stabi-lized GNPs. There were two distinct zones exhibiting pink and royalblue color by virtue of SPR. In both the fractions, spherical nanopar-ticles of different sizes can be seen (Fig. 3iv). In fraction d, GNPsexhibited sharp peak at 532 nm owing to quantum confinement ofthe electrons in nanoparticles (Fig. 2d). Spherical nanoparticles canbe seen raging from 10 to 20 nm in the TEM image (Fig. 3vd). Sizeof the nanoparticles ranging from 10 to 20 nm can be seen in thehistogram (Fig. 3vid).
In fraction e, deep blue colored GNPs exhibited broad peak at
538 nm (Fig. 2e). Area under the peak is more than the previouscase which indicates the heterodispersity of the nanoparticles. TEMimage displays the presence of nanoparticles from 19 to 40 nm(Fig. 3ve). The particles size is also displayed in Fig. 3vie.
S. Pandey et al. / Colloids and Surfaces B: Biointerfaces 109 (2013) 25– 31 29
Fig. 4. FTIR Spectra of (a) A. indica extract (b) GNPs synthesized using A. indica extract indicating the surface protection of GNPs (c) Chemogenic GNPs (d) GSH functionalizedc
nefca1
3
daatbttusomohc
hemogenic GNPs.
The idea behind the separation was to get the ideal size of theanoparticles for plethora of biological nanoparticles. Optimumfficiency of delivery into the pulmonary system can be achievedor nanoparticles of diameter <40 nm [15]. Enhanced uptake effi-iency has also been shown for gastro-intestinal absorption [30]nd transcutaneous permeation [31], with particles size of around00 nm and 50 nm respectively.
.3. FTIR study of the nature of capping proteins
FTIR spectra (Fig. 4a) of the ultra-filtered extract of A. indicaisplay peak at 1630.12 cm−1 which is assigned to NH bendingrising due to presence of proteins or short peptides. Another peakt 2924 cm−1 corresponds to bending and stretching of CH func-ional groups associated with alkanes. These functional groups mayelong to the thiolated proteins or short polypeptides. Moreover,he high content of surfactants present in A. indica may also con-ribute to these peaks. In comparative spectra of GNPs synthesizedsing same plant (Fig. 4b), an intense peak at 1629 cm−1 corre-ponding to NH bends and CO stretch of carboxylic acid wasbserved. This may be due to interaction between SH group of
oieties present in A. indica extract and surface of GNPs. The affinity
f thiol groups toward gold is more as compared to NH2 [32]. Mildumps at 1436 cm−1 and 1484 cm−1 corresponds to CH bends ofarboxylic acid.
Attachment of glutathione on GNPs was confirmed on com-paring the FTIR spectra of GNP and GNP–glutathione displayed inFig. 4c and d. Reappearance of dual asymmetric vibrations withpeaks at 2922.94 cm−1 and 2856 cm−1 corresponds to CH stretchof aldehydes, SH stretch is s in Fig. 4c. The shift in the peak wasof 66 nm. This may be due to interaction between the SH groupof glutathione and COOH group of chemically synthesized GNPsas the affinity of gold toward the thiol is extremely high whichresults in to formation of a thiol–ester interaction. This leads tostabilization of chemically synthesized GNPs [32].
3.4. Stability of GNPs synthesized using A. indica vs glutathionecapped GNPs
Fig. 5 a and b shows impact of NaCl on the spectral propertyof biologically capped GNPs and glutathione functionalized GNPsrespectively. Qualitatively, the sign of flocculation was reflecteddue to change in the color of GNPs from pink to blue after additionof the salt. There was slight change in the color in case of biogenicGNPs even after addition of heavy amount of salt (approx. 6 M). Ini-tial peak of the biogenic GNPs was exhibited at 542 nm which was
red shifted to 558 nm after addition of the 6.6 M NaCl in increasingconcentrations. A red shift of 16 nm indicates the spatial intimacyof the nanoparticles in the solution due to dipole coupling betweenthe plasmon of adjacent particles of the GNPs under the influence
30 S. Pandey et al. / Colloids and Surfaces B: Biointerfaces 109 (2013) 25– 31
Fig. 5. (a) UV–vis spectra of biogenic GNPs displaying the exceptionally high stability due to the influence of capping proteins. (b) UV–vis spectra of GSH functionalizedchemogenic GNPs showing a red shift of 107 nm indicating low stability as compared to biologically capped GNPs (c) Flocculation parameters of the GNPs after adding NaClat pH 7.2. (d) TEM image of biogenic GNPs (e) TEM image of chemogenic GNPs showing higher rate of flocculation as compared to biogenic GNPs.
Fig. 6. A hypothetical mechanism of metallo–micellar formation in the aqueous extract of A. indica leading to exceptional stability and biocompatibility to GNPs synthesizedusing the same. Figure at the top explains the formation of micelle due to oil water interface. At the bottom, a possible mechanism of internalization GNP–micelle complexvia steps a, b and c is outlined.
faces B
oa
rGtlwr(a
ccba7tt
saTctcts
cmoG
Acbipmcafbw
4
se
[
[
[
[[
[
[
[[[
[[[
[[
[[[
[[
[[
S. Pandey et al. / Colloids and Sur
f salt [33]. The peak was found to be stable after the addition ofny further amount of salt.
In case of GSH-functionalized GNPs, sudden change in color fromed to violet took place after addition of the salt. GSH functionalizedNPs exhibited a red shift of 107 nm (540–647 nm) after the similar
reatment. The physical relation between the red shift and floccu-ation was given by Quinten and Kreibig [34]. According to them,
hen the distance between flocculating spheres is smaller than theadius of the spheres, the resonance occurs at longer wavelengthlonger than the resonance of isolated spheres). These resonancesre concentrated above 600 nm in visible spectrum.
The flocculation measurement was done adding different con-entrations of NaCl in GNPs as mentioned earlier (pH 7.2) andalculated using Eq. (1). It can be inferred from Fig. 5c, the FP ofiogenic GNPs increases with increase in salt concentration exceptfter adding 6.6 M NaCl where the FP was less than that after adding
M NaCl. This unusual finding may be due to presence of surfac-ants in the A. indica which may deplete the dielectric constant ofhe medium leading to agglomeration at this pH.
In case of GSH capped GNPs, the FP was found to follow theame trend. But it was higher than the biogenic nanoparticles afterddition of the respective amount of salt concentration (Fig. 5c).his indicates the enhanced stability of the biogenic GNPs overhemogenic GSH-GNPs. The cardinal role of capping proteins inerms of its efficiency to prevent flocculation of the GNPs can beonsidered as directly proportional to FP at a particular pH (7.2 inhe present case). The less FP in this case indicates the exceptionaltability of the GNPs [35].
The impact of flocculation after adding the final amount of NaClan be seen in the TEM images. Due to the potential impact of ther-odynamically stable capping around the biogenic GNPs, less signs
f flocculation (Fig. 5d) can be seen as compared to GSH protectedNPs (Fig. 5e).
High content of the oils or hydrophobic materials present in the. indica leaf extract may also contribute to exceptionally high bio-ompatibility, thus making it an ideal drug delivery vehicle to ferryiological payloads. A speculative scheme for the above explanation
s presented in the Fig. 6. Due to the hydrophilic end of the cappingeptides facing to the water, there might be the chances of for-ation of tiny metallo–micellar complexes. These nano-micelles
an easily traverse the cells via thermodynamically favorable inter-ctions through lipid bilayer of cell membranes. These uniqueeatures of the A. indica mediated GNPs make them robust nano-ullets to fetch therapeutic moieties across the cell membranesithout the need of active surface functionalization.
. Conclusion
A. indica leaf mediated syntheses of GNPs were found to beuperior to glutathione functionalized GNPs as an ideal drug deliv-ry vehicle to ferry plethora of payloads in the biological systems.
[
[[[
: Biointerfaces 109 (2013) 25– 31 31
A comparative stability of the flocculation parameter proved thatthe biogenic GNPs were found to be highly stable under theinfluence of very high salt concentration due to presence of bio-logical capping proteins exposing hydrophilic functional groups tothe water and hiding the hydrophobic to interiors. GSH-cappednanoparticles were feeble in stability under the same salt con-centration. Flocculation parameter was found to be inverselyproportional to the stability of the GNPs.
[2] R. Hardman, Environ. Health Perspect. 114 (2006) 165–172.[3] Y. Yin, A.P. Alivisatos, Nature 437 (2005) 664–670.[4] S. Shiv Shankar, A. Ahmad, R. Pasricha, M. Sastry, J. Mater. Chem. 13 (2003)
1822.[5] Y. Feng, X. Lin, Y. Wang, Y. Wang, J. Hua, Mater. Lett. 62 (2008) 4299.[6] M. Sastry, A. Ahmad, M. Khan, R. Kumar, Curr. Sci. 85 (2003) 162.[7] S. Pandey, G. Oza, A. Mewada, R. Shah, M. Thakur, M. Sharon, J. Mater. Chem. B
3893–3946.25] C. Burda, X. Chen, R. Narayan, M.A. El-Sayed, Chem. Rev. 105 (2005) 1025–1102.26] M.C. Daniel, D. Astruc, Chem. Rev. 104 (2004) 293–346.27] M.S. Arnold, A. Green, J.F. Hulvat, S.I. Stupp, M.C. Hersam, Nat. Nanotechnol. 1
(2006) 60–65.28] A. Green, M.C. Hersam, Nano Lett. 9 (2009) 4031–4036.29] J.E. Millstone, S. Park, K. Shuford, L. Qin, G. Schatz, C.A. Mirkin, J. Am. Chem. Soc.
127 (2005) 5312–5313.30] N. Hussain, V. Jaitley, A.T. Florence, Adv. Drug Deliv. Rev. 50 (2001) 107.31] A.K. Kohli, H.O. Alpar, Int. J. Pharm. (2004) 275.
32] J. Zhang, Q. Chi, J.U. Nielsen, E.P. Friis, J.T. Andersen, J. Ulstrup, Langmuir 16
(2000) 7229–7237.33] U. Kreibig, L. Genzel, Surf. Sci. 156 (1985) 678–700.34] M. Quinten, U. Kreibig, Surf. Sci. 172 (1986) 557–577.35] K.S. Mayya, V. Patil, M. Sastry, Langmuir 13 (1997) 3944–3947.
