Journal of Colloid and Interface Science 442 (2015) 89–96

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Preparation of small size palladium nanoparticles by picosecond laserablation and control of metal concentration in the colloid

http://dx.doi.org/10.1016/j.jcis.2014.11.0660021-9797/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author.E-mail address: emilia.giorgetti@fi.isc.cnr.it (E. Giorgetti).

E. Giorgetti a,⇑, P. Marsili a,b, S. Cicchi c, L. Lascialfari a,c, M. Albiani c, M. Severi c, S. Caporali c,e,M. Muniz-Miranda c, A. Pistone d, F. Giammanco b

a Istituto dei Sistemi Complessi, Consiglio Nazionale delle Ricerche, Via Madonna del Piano 10, 50019 Sesto Fiorentino (FI), Italyb Department of Physics ‘‘E. Fermi’’, University of Pisa, Largo Bruno Pontecorvo 3, 56127 Pisa, Italyc Department of Chemistry ‘‘Ugo Schiff’’, University of Firenze, Via della Lastruccia 3, 50019 Sesto Fiorentino (FI), Italyd Department of Electronic Engineering, Industrial Chemistry and Engineering, University of Messina, C.da di Dio, 98166 Messina, Italye Consorzio INSTM, Via Giusti 9, 50123 Firenze, Italy

a r t i c l e i n f o

Article history:Received 17 October 2014Accepted 27 November 2014Available online 4 December 2014

Keywords:Pd nanoparticlesLaser ablationMetal clustersPd oxides

a b s t r a c t

We assessed a method for the preparation of small, highly stable and unprotected Pd nanoparticles bypicosecond laser ablation in 2-propanol. The nanoparticles can be extracted from 2-propanol bycentrifugation and redispersed in water, where a strongly negative f-potential assures long term stability.The proposed procedure permits reduction of particle size down to 1.6 nm and optimization of thePd(0):Pd(II) ratio which, in the best cases, was of the order of 6:1. The increase of this ratio with ablationtimes has been correlated to the high temperature conversion of PdO to metallic Pd by a simpletheoretical model. A study of the relationship between colloid absorption at 400 nm and Pd concentrationpermitted the role of PdO in the determination of the UV–vis spectra to be clarified and the limits of theMie theory for the evaluation of colloid concentration to be established. The absorption at 400 nm can beused as a fast method to estimate the Pd content in the colloids, provided that a calibration of the ablationprocess is preliminarily performed.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

Pd is widely used, as metal or in the form of its salts, to catalyze anumber of synthetic processes [1,2]. In the form of nanoparticles(NPs), it is employed in sensors [3,4] and photoresponsive devices[5,6], as well as for the decoration of polymers [7] and carbon nano-tubes for the development of novel heterogeneous catalysts [8]. Thelarge surface to volume ratio exhibited by NPs is expected to greatlyimprove the catalytic activity [9,10]. NPs size also represents a keyparameter in the development of nanohybrids that are based onPdNPs-decorated Multi Walled Carbon Nanotubes (MWCNTs)[11,12]. Preliminary results on the decoration of functionalizedMWCNTs with AuNPs [13] suggested that NP size determines sur-face coverage of the nanotubes, as small NPs are the most efficient.

In addition to the size of the nanostructures, catalytic processesstrongly depend on the efficient interaction between catalyst andenvironment, so that surface cleanliness is expected to play a majorrole. In this regard, Pulsed Laser Ablation in Liquid (PLAL) seems tobe a promising procedure for the preparation of NPs because, unlike

current wet chemical methods, it permits the synthesis of stablemetal colloids also in pure solvents and, under proper fabricationconditions, with no need of stabilizing agents [14].

With respect to the most studied metals, such as Au and Ag[14], only a limited number of papers is available in the literaturewhich describe PLAL of Pd. Most of them use the fundamentalwavelength or the second harmonic of Q-switched Nd:YAG lasers,namely 1064 or 532 nm, and nanosecond (ns) pulses [15–22], anddifferent liquids, such as deionized water (DIW) [15,16,18,19,21–24], acetone [17,20,23–25] or other organic solvents [20,24]. Forsuch preparation conditions [15–25], which mostly involve highlaser fluences on target (namely from 1 up to 40 J/cm2), the PdNPsobtained by PLAL appear strongly polydisperse and relatively big,with average size ranging from 5 to 15–20 nm, which seem toincrease with increasing laser fluence [19,20]. For example, in thecase of DIW [15,18,19,22,23], the NPs size is always well above5 nm, with large statistical distribution and strong evidence of coa-lescence, although f-potential and long term stability tests of thecolloids were not reported. Examples of small PdNPs are describedin Ref. [24], where 12 picosecond (ps) ablation of Pd at 532 nm andhigh repetition rate (10–220 kHz) is reported in DIW, acetone andethanol, the smallest NPs being those obtained in ethanol.

90 E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96

A major problem when performing the ablation of Pd, particu-larly in the case of DIW, is surface oxidation, which can be difficultto control and depends on several parameters, among which pulseduration and laser wavelength, as reported in Ref. [18]. In particu-lar, XPS tests performed on PdNPs obtained in DIW with ns pulsesand 1064 nm, permitted observation of both Pd(II) and Pd(IV) [18].A considerable reduction of surface oxidation with respect to DIWis expected when performing the ablation in organic solvents, asobserved in Ref. [23], which compares results obtained with DIWand acetone. However, the use of organic solvents under PLAL con-ditions was observed to strongly promote solvent decompositionand subsequent formation of amorphous carbon, which can passiv-ate the surface of NPs. This phenomenon has already been detectedby means of Raman or XPS spectroscopy in the case of Au [26] andPd [20,23,24] NPs prepared by PLAL in acetone and ethanol.

The considerations discussed above suggest that it is importantto find a convenient trade-off between the requirements of smallsize, stability and surface cleanliness, in order to synthesize PdNPswith interesting potentialities for catalytic applications. Therefore,this paper is aimed at establishing a method for the preparation ofstable colloidal suspensions of small (mean diameter < 5 nm) andunprotected PdNPs by PLAL, in view of integration with the abovementioned applications and, particularly, with novel heterogeneouscatalysts. For this purpose, we analyzed PdNPs obtained in 2-propa-nol under ‘‘gentle’’ ablation regime with the second harmonic wave-length of a ps Nd:YAG laser. We call our regime ‘‘gentle’’ since in ourexperimental conditions (fluences below 1 J/cm2, ps pulses and lowrepetition rate) NP formation occurs in the absence of any phenom-ena related to shock wave formation, cavitation bubbles or brightplasma emission [27], generally, but only qualitatively claimed asresponsible for particle growth [18]. We carried out a characteriza-tion of the colloids by UV–vis, Raman, XRD and XPS spectroscopyand TEM and HRTEM imaging. Moreover, as the concentration isan essential parameter to evaluate the activity of a catalyst, we alsocompared UV–vis experimental data and their theoretical modelingby the Mie theory, with direct Inductively Coupled Plasma AtomicEmission Spectrometry (ICP-AES) concentration measurements, inorder to verify the existence of a simple relation between absor-bance at a fixed wavelength and Pd concentration.

2. Materials and methods

We prepared PdNPs by laser ablation of a Pd target with thefundamental wavelength or the second harmonic of a mode-lockedNd-YAG laser (EKSPLA PL2143A: rep. rate 10 Hz, pulse width 25 psat 1064 nm and 18 ps at 532 nm). The experimental set up isdescribed in detail in Ref. [26]. We performed the ablation in dou-bly DIW (18.2 MX cm @ 25 �C), acetone or 2-propanol. Acetoneand 2-propanol, of 99.9% purity (Chromasolv� for HPLC grade)were purchased from Aldrich and used as received. The Pd targetwas supplied by Aldrich (0.5 mm thick, 99.9% purity).

Off-line UV–vis spectra were recorded with a Varian Cary 4000using a 1 mm or 1 cm path-length quartz cell.

The f-potential of the colloids in water was measured with aZetasizer Nano ZS90, Malvern Instruments.

Samples of PdNPs for Transmission Electron Microscopy (TEM)inspection were obtained by dropping a small amount of colloidonto carbon-coated copper grids, followed by evaporation. Lowresolution images were recorded with a Philips CM12, 120 kVand High Resolution TEM (HRTEM) images were obtained with aJeol JEM 2010 HRTEM operating at 200 kV. Particle mean diameterand dispersivity were determined by fitting the measured statisti-cal distributions with a lognormal function.

X-ray Powder Diffraction (XRD) data were collected with aBruker New D8 Advance diffractometer using Cu Ka radiation.The Pd colloid was deposited on a silicon single crystal and

scanned over the 2h range from 26� to 60� using a 0.02� step for3 s on every step (total acquisition time 90 min). No filter wasemployed and the diffracted X-rays were collected using a Lynx-eye_Xe detector.

X-ray Photoelectron Spectroscopy (XPS) experiments werecarried out in an ultrahigh vacuum (UHV, 10�9 mbar) systemequipped with a VSW HAC 500 hemispherical electron-energy ana-lyzer using a non-monochromatic Mg Ka X-ray source (1253.6 eV;anode operating at 10 kV and 10 mA). The spectra were collected inthe constant-pass-energy mode (Epas = 44 eV). The overall energyresolution was 1.2 eV. The binding energies were calibrated onthe basis of the aliphatic C 1s peak at 284.8 eV. Samples for XPSmeasurements were prepared by drop-casting a small quantity ofthe colloid onto a glass slide and leaving it to dry at room temper-ature. Prior to the elemental scans, a survey scan was measured forall the samples in order to detect all the elements present. The spec-tra were deconvoluted by using XPS Peak 4.1 program with a Gauss-ian–Lorentzian mix function and Shirley background subtraction.

Raman spectra were measured using a Renishaw RM2000microRaman apparatus, equipped with an Ar+ laser emitting at514.5 nm. Sample irradiation was accomplished by using the 50�microscope objective of a Leica Microscope DMLM. The backscat-tered Raman signal was filtered by a double holographic Notch fil-ter system and detected by an air-cooled CCD. All spectra werecalibrated with respect to a silicon wafer at 520 cm�1.

The determination of the Pd concentration in the samples wasperformed in triplicate by a Varian 720-ES Inductively CoupledPlasma Atomic Emission Spectrometry (ICP-AES). 50 lL of eachsample were diluted to 5 mL with 0.1% suprapure nitric acidobtained by sub-boiling distillation, spiked with 0.5 ppm of Ge,used as an internal standard, and analyzed. Calibration standardswere prepared by gravimetric serial dilution from commercialstock standard solutions of Pd at 1000 mg/L. Wavelengths usedfor Pd determination were 324.270, 340.458, 342.122 and360.955 nm, whereas for Ge the line at 209.426 nm was used.The operating conditions were optimized to obtain maximumsignal intensity, and between each sample, a rinse solution consti-tuted by 2% v/v HNO3 was used.

3. Results and discussion

Initially, we prepared PdNPs in DIW, with 1064 nm and a fluenceon target of 0.5 J/cm2. As reported in the Supporting information(SI1 and SI2), HRTEM inspection, UV–vis and XPS spectroscopy con-firmed the expected presence of PdO, while the NP size was larger(7 nm) than our desired goal. Furthermore, these colloids exhibita strong tendency to coalesce and they collapse in a few days afterfabrication. According to recent literature, acetone permits the syn-thesis of very stable, small and quasi monodispersed AuNPs, whenthe ablation is carried out using a wavelength of 532 nm [26] while,in the case of PdNPs, it permits considerable reduction of surfaceoxidation with respect to water [23,24] and also the production ofsmall size NPs [24]. However, in spite of their excellent stability(two years shelf life with no detectable sign of spectral changes)and considerable reduction of size (5 nm) and oxide contributionwith respect to DIW, the NPs that we obtained in acetone areembedded in an abundant glassy matrix, which is likely composedof amorphous carbon, originating from solvent decompositionduring the ablation (as described in the Supporting information,SI3). Therefore, we maintained short laser wavelengths and thelow fluence, but changed to a different and scarcely studied solvent,namely 2-propanol [20], with the aim of finding the best compro-mise among NP size, colloid stability and surface contaminationfrom Pd oxides or carbon. Fig. 1 shows typical TEM micrographsof PdNPs obtained in 2-propanol using 532 nm radiation and

Fig. 1. TEM micrographs of PdNPs obtained in 2-propanol with 532 nm and 0.15 (a) and 0.5 (b) J/cm2 fluences. The insets show the statistical distribution of the particle size.

Fig. 2. Absorbance at 400 nm, registered shot by shot during the 532 nm ablation ofPd in 2-propanol with 0.15 (black line) and 0.5 J/cm2 (gray line). OPL = 1 cm.

E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96 91

different energies per pulse, corresponding to 0.15 and 0.5 J/cm2

fluence. The glassy matrix detected in the case of PdNPs in acetoneis not observed in Fig. 1, although we cannot exclude the presenceof amorphous carbon in small amounts. At low ablation fluence(Fig. 1a) the NP size (2.1 nm size, r+= 1.5 nm and r� = 0.7 nm 1/eright and left half widths of the Lognormal statistical distribution)is smaller than that obtained using acetone as solvent. However,some large NPs are present, whose size is of the order of 10 nm.In contrast, at higher fluence, the NPs appear well disaggregatedand smaller, with 1.5 nm average size, r+ = 0.7 nm and r� = 0.5 nmand no evidence of large NPs (Fig. 1b).

Contrary to PdNPs obtained in DIW, which collapse in a fewdays, PdNPs obtained in 2-propanol are stable for months. Further-more, if an aqueous environment is required for particular applica-tions, they can be centrifugated and subsequently redispersed inDIW. Redispersed PdNPs are much more stable than those obtaineddirectly by ablation in DIW, as suggested by their f-potential. Typ-ical values of f-potential of PdNPs obtained in DIW are only weaklynegative (�14 mV for the sample of SI1 and SI2), while the f-poten-tial of redispersed PdNPs is strongly negative (�47 mV for the sam-ple of Fig. 1b).

In order to assess the possibility of controlling NP properties andPd concentration in the colloids, we prepared several sets of sam-ples with two different fluence values, namely 0.15 and 0.5 J/cm2,and for different ablation times. As mentioned in the Introduction,this range of fluence ensures the absence of shockwave and cavita-tion effects [27]. In this regime, ablation occurs as the interplayamong different phenomena, namely vaporization, thermoionicand photon-assisted thermal ionization, as extensively describedin Ref. [28]. The ablated material, basically consisting of ions andatoms, is confined by the surrounding liquid to a thin layer of vapor-ized solvent (the so-called plume), where aggregation occurs untilplume pressure overcomes ambient pressure. In this framework,NP formation and stabilization occurs in a time of the order of hun-dreds of microseconds after the end of the laser pulse and no furtheraggregation occurs in between two pulses, as confirmed by thealmost linear growth of the UV–vis absorption versus the numberof pulses [29].

On the assumption that the concentration of the colloid is pro-portional to its absorption at a fixed wavelength, we stopped theablation when the absorbance A at 400 nm (A(400)) in a 1 cm pathlength cell reached predetermined values. We chose 400 nm as thereference wavelength, since this wavelength is sufficiently faraway from the expected contributions of Pd oxides to Mie scatter-ing [30–33].

Fig. 2 shows in situ measurements of A(400) versus number oflaser shots: A(400) grows monotonically and progressively tends

to a saturation value, which also grows with energy. It eventuallystops when the energy transmitted by the 2 cm-high liquid columnfalls below the ablation threshold which, in our case, is around0.07 J/cm2. This explains the quick saturation of the process at0.15 J/cm2.

In order to discriminate among the different ablation regimescorresponding to different slopes in the A(400) versus time curves,we sampled the colloids at different values of A, namely: (i) at earlyablation stages, (ii) in the linear regime, (iii) close to the onset ofsaturation and (iv) far beyond saturation. Table 1 summarizesthe list of the prepared samples. Samples A–C were prepared with0.15 J/cm2 in one day. We then prepared two series of sampleswith 0.5 J/cm2 in different days (samples E1 to I1 and D2 to I2).The samples belonging to these series and labeled with ‘‘2’’ wererepeated 3 times, so that Table 1 reports the average parametersof absorbance and concentration. In contrast, NP size (average,minimum and maximum values) and statistical distribution andXPS evaluation of the Pd(0):Pd(II) ratio were obtained from onerepresentative sample of each batch. Typical TEM images and sta-tistical distribution of samples B, C and and F2–I2 are reported inthe Supporting information as Figs. SI4–SI9, while samples A andE2 are described in Fig. 1.

We measured the Pd concentration in our colloids by ICP-AES.The results are also reported in Table 1. In our working conditions,Pd concentration is quite low at 0.15 J/cm2 ablation, with negligiblevariations among the three available samples. Here, due to quicksaturation of the process at a very low value of the absorption,we could span only a small interval of A(400). In contrast, in thehigh fluence regime, we could tune the concentration from 3 upto 25 mg/L upon proper increase of the ablation time.

Table 1List of samples prepared in 2-propanol with by ablation with 532 nm.

Sample Fluence (J/cm2) A(400) Size (r+/r�) (nm) Min/max size (nm) XPS (Pd(0)/Pd(II)) ICP-AES (mg/L) Mie (mg/L)

A 0.15 0.32 2.1 (+1.5/�0.7) 0.3/19 2.5 ± 0.1 10.0 17B 0.15 0.38 2.7 (+2/�0.9) 0.4/11 3.8 ± 0.1 9.9 20C 0.15 0.44 3.7 (+3/�1.2) 0.5/18 4.9 ± 0.5 10.5 24D2 0.5 0.17 3.6 9E1 0.5 0.30 7.5 16E2 0.5 0.28 1.5 (+0.7/�0.5) 0.2/6.7 2.2 ± 0.1 5.6 15F1 0.5 0.54 13.9 29F2 0.5 0.48 1.6 (+1/�0.6) 0.2/5.8 4.6 ± 0.5 12.0 26G1 0.5 0.88 20.0 47G2 0.5 0.76 1.4 (+0.8/�0.5) 0.4/6.7 4.9 ± 0.1 17.5 41H1 0.5 1.05 24.0 56H2 0.5 0.90 1.4 (+0.9/�0.5) 0.4/3.1 5.0 ± 0.1 19.0 48I1 0.5 1.19 25.0 64I2 0.5 1.20 1.4 (+0.7/�0.4) 0.3/3.1 5.8 ± 0.1 20.0 64

Fig. 4. MicroRaman spectrum of a dried drop of a Pd colloid obtained in 2-propanolby ablation with 532 nm and 0.5 J/cm2 (G type in Table 1). Exciting line: 514.5 nm.

92 E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96

In regard to the absorbance, the values reported in Table 1 canbe considerably different from those measured on-line. In particu-lar, the values close to or beyond saturation are higher than thosereported in Fig. 2, where the upper limits of A(400) are 0.35 and 0.8.Actually, since the ablation is performed without stirring, theshot-by-shot measure of A(400) reported in Fig. 2 is affected byfluctuations in the colloids, by solvent viscosity and diffusionphenomena. Therefore, Table 1 reports absorbance data obtainedoff-line and after thorough stirring.

The off-line UV–vis spectra of a series of samples obtainedwith 0.5 J/cm2 are shown in Fig. 3. In the UV region of the spec-trum, two well-resolved shoulders around 225 and 258 nm areclearly visible. These spectral features, that are superimposedover the normal interband transitions of the colloids, can becaused by quantum size effects that, due to the very small sizeof the PdNPs, lead to the appearance of a HOMO–LUMO gap atthe Fermi level and thus to a molecular-like absorption behavior[34], or by solvent degradation. In fact, the presence of someamorphous or graphitic carbon at the surface of our NPs, althoughnot directly proved by TEM inspection as in the case of acetone(SI3), is demonstrated by Raman spectroscopy. The characteristicD and G bands at 1360 cm�1 and 1590 cm�1 are clearly visible inFig. 4 [20].

Samples A–C and E2–I2 were characterized by XPS. Weevaluated the ratio between Pd(0) and Pd(II), which is reported inTable 1. It is noted that this was not possible for sample D2, dueto its high dilution. In all other cases, the XPS tests gave evidencefor the presence of Pd(II). In particular, with both ablation fluences,the amount of Pd(II) decreases with A(400). For 0.15 J/cm2 ablation,the Pd(0):Pd(II) ratio increases from 2.5:1 to 4.9:1. The same trend

Fig. 3. Off-line UV–vis absorption spectra of some of the samples of PdNPsbelonging to series ‘‘2’’, described in Table 1 and obtained in 2-propanol at 532 nmwith 0.5 J/cm2.

is observed for 0.5 J/cm2 ablation. In this case, the ratio grows from2.2:1 for the sample with the lowest absorbance, to 5.8:1 for thesample obtained in saturation regime. Fig. 5 shows a typical result,corresponding to a sample obtained with 0.5 J/cm2 and A(400) = 0.9.The doublet of Pd 3d core transition (Fig. 5a) is, in this case, charac-terized by a strong asymmetry toward high binding energy (BE).Therefore, the fitting requires the use of two sets of doublets forthe Pd 3d3/2 and the Pd 3d5/2 peaks. BE values for the two Pd 3d5/2

peaks were 335.3 and 337.3 eV, respectively. The more intensepeak, located at lower BE, is consistent with Pd(0) (thin gray linein Fig. 5a), whereas the weaker one, at higher BE value, indicatesthe presence of Pd(II), i.e. of PdO (thick gray line in Fig. 5a)[35,36]. In contrast with previous reports [18,21,24], no evidenceof Pd(IV) was found, neither in this case, nor in the other samplesstudied. In principle, the presence of some Pd carbide cannot beexcluded. Indeed, there are reports in the literature [37] showingthat PdCx (x = 0.13–0.15) can be observed in the XPS region charac-teristic for the carbon 1s core transition associated with a peak atabout 282 eV. Fig. 5b shows deconvolution of the XPS spectrum ofcarbon for the same sample of Fig. 5a. The spectrum is dominatedby aliphatic/elemental C, with a weak shoulder at higher BE values(287 eV) that can be assigned to carboxyl (ACOA), with no evidenceof PdCx. This suggests that, if present, its contribution to the spec-trum is very low and hidden by more abundant C forms.

HRTEM characterization showed the coexistence of Pd and PdONPs with a clear polycrystalline nature observed in high magnifica-tion images (Fig. 6) and their associated fast Fourier transform(FFT) patterns (inset in Fig. 6). Metallic palladium and palladiumoxide nanoparticles have been clearly recognized as confirmed bya lattice spacing of 2.25 Å, which corresponds to (111) planes ofthe face-centered cubic phase of Pd metal, or a lattice spacing of

Fig. 5. (a) Deconvolution of the XPS spectrum of Pd (3d3/2 and 3d5/2 transitions). Circles represent experimental points, whereas thick and thin gray lines represent thecontribution of Pd(II) and Pd(0), respectively. (b) Deconvolution of the XPS spectrum of carbon (1s core transition). Circles represent experimental points, whereas thick andthin gray lines represent the contribution of aliphatic and ‘‘oxidized’’ (carboxy) carbon, respectively.

Fig. 6. HRTEM micrograph of sample H2 with corresponding FFT analysis (inset).

Fig. 7. XRD data of a Pd colloid deposited on a Si single crystal. The inset shows amagnification of the spectrum between 39� and 46�.

E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96 93

2.14 Å, which corresponds to (110) planes of the tetragonal phaseof PdO. In contrast, HRTEM did not show the characteristic latticeplanes of graphite, suggesting that carbon, as detected by Ramanspectroscopy, is mostly present in its amorphous phase.

In spite of the high noise level, mainly due to the very smallsize of our PdNPs, the complex average crystalline compositionof our Pd colloids is demonstrated by the XRD spectrum reportedin Fig. 7. Apart from a very broad, dominant peak located at lowangles (8–20�) (data not shown) attributable to amorphous car-bon, the spectrum exhibits intense diffraction lines attributableto carbon in different crystalline forms (cubic and hexagonal).However, several crystalline phases of Pd are also detectable. Inparticular, the XRD peaks indexed as (101) PdO and (111)Pd(0) were identified at 33.6� and 40.1� respectively. It is notedthat two more XRD peaks, located at 39.4� and 45.8�, can be rea-sonably attributed to the (111) and (200) diffraction of a‘‘stressed’’ Pd(0) structure. Such an effect is a typical diffractionfeature indicating formation of interstitial alloys. In this case,two elements are available to form interstitial Pd alloys, i.e. Cand H. Due to the larger atomic size of the doping element, Pdcarbide is characterized by a larger lattice with respect to Pdhydride, so that its diffraction lines are located at lower angles,namely at about 38.8� for (111) PdCx versus 39.4� for PdHx [38].On the basis of these considerations, it seems reasonable to attri-bute the peaks at 39.3� and 45.8� to (111) and (200) planes ofPdHx, whereas PdCx, if present, is well below the detection limitof this analytical technique.

4. Discussion

The concentration of Pd in colloids can, in principle, be esti-mated indirectly by modeling the UV–vis spectra by the Mie theory[30,39]. In contrast to gold or silver, the UV–vis spectral features ofmetals such as Pd, which do not exhibit any plasmon resonance, donot depend on NP size and statistical distribution. Consequently, inthe absence of oxides, the absorption spectrum features of Pdcolloids are expected to be determined only by the interband tran-sitions of the metal and, hence, the absorption value at fixed wave-lengths by the concentration of metallic Pd [30], with no influencedue to particle size or shape. Therefore, a straightforward way topredict metal concentration would be the measure of the absor-bance A at any fixed wavelength, with no need of supportingTEM studies [40]. Although not tested by independent concentra-tion measurements, this method was previously proposed in Ref.[19]. In order to verify its feasibility, Table 1 also reports theconcentrations of our samples in 2-propanol, as evaluated by theMie theory, assuming metallic PdNPs. Inspection of Table 1immediately suggests that the case of Pd is more complex thanexpected. There is clear inconsistency between the ICP-AES resultsand theoretical predictions and the expected linear relationbetween absorption and concentration fails. For example, samplesA–C exhibit different values of A(400), but have the same experi-mental concentration, while samples A and E1 exhibit the sameA(400) but have considerably different concentrations. To bettervisualize this point, Fig. 8 reports the experimental values of Pd

Fig. 9. PdNP temperature versus laser fluence under 532 nm and ps pulsesirradiation.

94 E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96

concentration versus A(400) for all the samples prepared with 0.5 J/cm2 per pulse (black stars and black circles refer, respectively, toseries ‘‘1’’ and ‘‘2’’). In this ablation regime, the experimental Pdconcentration scales linearly with A(400) up to A � 1 and thensaturates. Fig. 8 also reports the theoretical values (gray starsand gray circles refer, respectively, to series ‘‘1’’ and ‘‘2’’) obtainedfor the same samples by the Mie theory (gray line). In this case,there is no saturation effect and, moreover, the calculated valuesof Pd concentration are much higher than the experimental ones.

The difference between the curves of Fig. 8 is too large to besimply ascribed to uncertainty in the dielectric constant of Pd,which has been investigated much less than the most studied casesof Ag or Au. The fact that an increase of the absorption is notreflected in a proportional increase of the concentration suggeststhe onset of different mechanisms occurring during the ablationprocess, namely the strong tendency of Pd to oxidize, the decom-position of the solvent with formation of amorphous carbon andthe very small size of the NPs. The spectral changes in the high fre-quency region of the UV–vis spectrum of the colloidal samples thatcan be related to these effects are not easily predictable, with con-sequent large errors in the estimation of Pd concentration. Forexample, SI10 shows Mie simulations of the UV–vis spectra oftwo Pd colloids, namely samples A and H2, assuming pure Pd orpure PdO NPs. For both assumptions, the calculated Pd concentra-tions are very different from the measured values. In particular, acomparison between theoretical spectra of Pd and PdO NPs showsthat the same value of A(400) corresponds to significantly differentmetal concentrations of Pd, the concentration of PdONPs beingmuch lower with respect to PdNPs.

Moreover, it is well known that Pd(0) oxidation to PdO occursabove 350 �C and that, in turn, PdO decomposes above 900 �C[41]. High temperature conversion of Pd(II) to Pd(0) [18] canexplain the saturation effect observed in Fig. 8, where A(400) growsfaster than Pd concentration. It is also in agreement with XPS data,which show that the samples corresponding to the saturationregions of Fig. 2 exhibit a higher Pd(0):Pd(II) ratio. This suggeststhat, in this irradiation regime, the process of conversion of PdONPsto PdNPs dominates over material extraction from the target.

To verify this hypothesis, we calculated the effect of laserabsorption on the temperature of PdNPs by using a model exten-sively described in Ref. [28]. The results are illustrated in Fig. 9,which reports, as a function of fluence, the temperature of a singlePdNP. With respect to the case treated in Ref. [28], where 532 nmpulses almost match the plasmon resonance of AuNPs, here theabsorption cross section of PdNPs at the same laser wavelengthis much lower. Consequently, for our irradiation conditions, thetemperature of PdNPs grows under laser absorption, but remainsalways below the melting point. Therefore, according to Fig. 9,

Fig. 8. Pd concentration versus absorbance at 400 nm calculated by Mie theory(gray line and symbols) or directly measured by ICP-AES (black symbols). Black lineis a guide to the eye; stars and circles refer to samples belonging series 1 and 2 ofTable 1, respectively. OPL = 1 cm.

we can expect a favorable formation of PdO above the ablationthreshold (0.07 J/cm2) and an efficient conversion of PdO to Pdabove 0.3 J/cm2. Although a more detailed theoretical modeling isbeyond the scope of this work, the trend of Fig. 9 seems to be in fairagreement with our experimental results. Studies on the oxidationof PdNPs produced in DIW have been discussed previously (Ref.[18]), where ablation was induced with infrared and UV radiation,by using ns pulses and high fluences. In that case the oxidation rateof PdNPs was higher by using UV-radiation than IR, which caused ahigher plasma temperature (by a factor of two with respect to thatreached with UV light), in agreement with our findings.

In summary, the presence of Pd oxides in the colloid stronglyaffects the absorption spectrum. In general, this is true not onlyin 2-propanol, but also in DIW or acetone. The poor knowledgeof Pd and PdO dielectric constants, the different crystalline phasesthat the oxide can exhibit and the possible formation of core–shellstructures, make any theoretical prediction of the UV–vis spectraof the colloids very difficult, impairing the use of Mie theory fora straightforward evaluation of the metal content. Nevertheless,in addition to presenting for the first time to our knowledge, anattempt to model UV–vis spectra of Pd-based colloids, we haveverified that, for the case of 2-propanol, a linear relationship existsbetween Pd concentration and colloid absorption at 400 nm, whenablation times are below the saturation regime (see Figs. 2 and 8).Therefore, when a knowledge of the Pd concentration in thecolloids is mandatory, this linearity range can be checked by cali-brating the process at a fixed fluence through UV–vis spectroscopyand subsequent ICP-AES measurements. Then, A(400) can be takenas a reliable indication of Pd content in the sample.

5. Conclusions

PdNPs obtained by PLAL are interesting candidates forapplication to heterogeneous catalysis and, in particular, for thedevelopment of MWCNT@Pd nanohybrids, as already reported inRefs. [12,17]. To achieve this goal, preparation of small (surfaceto volume ratio plays a key role in all catalytic applications), cleanand stable NPs is mandatory. In addition to the above mentionedcharacteristics, we stress that a viable use of PLAL-synthesized Pdcolloids, with their unique properties of purity, requires a reliablecontrol, and hence prediction, of Pd concentration. This is straight-forward when wet chemical methods are adopted for the synthe-sis. In contrast, it is difficult in the case of PLAL-prepared PdNPs.Unlike the case of Au colloids, which have been intensively studiedfor many years [39], the absence of a plasmon resonance in theUV–vis spectrum of Pd colloids, the strong tendency to oxidizeand the poor knowledge of the dielectric constants of the metaland its oxides, represent serious problems to determine the metalconcentration from the spectral properties of samples.

E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96 95

In general, the currently available studies on the preparation ofPLAL-synthesized Pd colloids have employed a wide range ofsolvents and laser parameters [15–25], but did not focus on thecontrol of particle size, concentration and surface properties. Toour knowledge, these issues are partially addressed only in Ref.[21], where the authors explore a different ablation method: themetal target is substituted by a suspension of Pd powders in waterand the ablation takes place at the liquid–air interface. Thismethod generated very small NPs with good yields and the pres-ence of Pd oxides could be reduced by flowing Ar.

In this scenario, the present paper proposes a method for thepreparation of small and stable NPs of Pd in colloidal suspensions,whose concentration and surface properties can be controlled byadjusting the solvent and the ablation parameters. In particular,we performed an investigation on the PdNPs obtained in 2-propa-nol by pulsed laser ablation under low fluence (0.5 J/cm2), under-taking a morphologic (TEM and HRTEM), spectroscopic (UV–vis,Raman, XPS, XRD) and analytic (ICP-AES) characterization. Further-more, the UV–vis absorption spectra has been simulated and amodel of the interaction between the laser pulse and the Pd targethas been proposed. According to our experimental results, theseNPs represent the best trade-off between stability and surfacecleanliness, although carbon and oxide on NPs surface cannot byeliminated completely.

The main results obtained from our study are the following:

� In comparison to other solvents, 2-propanol leads to theformation of quasi monodisperse and very small PdNPs,whose size can be reduced down to 1.6 nm. The colloidsare stable for months, and even more after re-dispersionin water, as suggested by the highly negative value of thef-potential.

� Surface oxidation of the PdNPs, as well as the presence ofamorphous carbon originating from solvent decompositionduring PLAL, are well-known effects [20]. This issue hasbeen addressed herein and it has been shown that surfaceoxidation in 2-propanol can be controlled by tuning fluenceand ablation time.

� The change of the Pd(0):Pd(II) ratio with ablation time wasmeasured by XPS and related to the high temperatureconversion of PdO to metallic Pd by a simple theoreticalmodel, which permits evaluation of the temperature of theablated material. This model, which represents an extensionto Pd of one previously proposed to describe the photofrag-mentation of AuNPs in water [28], can be considered as thefirst step towards a theoretical description of NPs formationunder ‘‘gentle’’ PLAL conditions.

� For the first time, colloid concentration can be controlledand predicted, for the case of PLAL-synthesized PdNPs, byfollowing the on-line UV–vis spectra during laser ablation,after calibration by ICP-AES.

In summary, stable Pd colloids consisting of very small NPs(1.6 nm size), having 20–30 mg/L concentration and dominatedby Pd(0) over Pd(II) can be easily produced in 2-propanol. Hence,we consider that the particular characteristics of these colloids willenable an efficient coverage of MCWNTs to be achieved and willlead to the development of stable and robust nanohybrids [13].

The presence of some surface contamination, related to forma-tion of amorphous C, cannot be avoided completely. However, it isnot expected to impair their catalytic activity, but, indeed, it can bean enhancing factor, especially for reduction reactions or reactionthat are mediated by the absorption of hydrogen on PdNPs surface.It was indeed, both theoretically and experimentally, proven thatthe presence of carbonaceous material on the surface of PdNPscan enhance the catalytic performances and the ‘‘catalytic life’’ of

the NPs [42,43]. In these sense, we can anticipate that nanohybridsconsisting of MWCNTs decorated with our PdNPs obtained in 2-propanol act as efficient and robust catalysts in reactions of reduc-tion and isomerisation of double C@C bonds and carbonyl groups.Such results are the object of a forthcoming paper.

Acknowledgments

The authors thank Regione Toscana POR CRO FSE 2007_2013Asse IV Project-Nanocube for financial support and Patrizia Cantonof the University of Venezia, Italy for useful discussions.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jcis.2014.11.066.

References

[1] Jie Jack Li, G.W. Gribble (Eds.), Palladium in Heterocyclic Chemistry – A Guidefor the Synthetic Chemist from Tetrahedron Organic Chemistry Series, vol. 26,Elsevier, 2007.

[2] J. Tsuji (Ed.), Palladium in Organic Synthesis – Topics in Organometallics, vol.14, Springer, 2005.

[3] H. Hosseini, S.J. Tabatabaei Rezaei, P. Rahmani, R. Sharifi, M. Reza, Nabid, A.Bagheri, Sens. Actuators B 195 (2014) 85.

[4] Q. Wang, X. Cui, J. Chen, X. Zheng, C. Liu, T. Xue, H. Wang, Z. Jin, L. Qiao, W.Zheng, RSC Adv. 2 (2012) 6245.

[5] Y. Shiraishi, N. Toshima, K. Maeda, H. Yoshikawa, J. Xu, S. Kobayashi, Appl.Phys. Lett. 81 (15) (2002) 2845.

[6] S. Thomas, P. Bazin, L. Lakiss, V. de Waele, S. Mintova, Langmuir 27 (2011)14689.

[7] A. Leyva-Pérez, J. Oliver-Meseguer, P. Rubio-Marqués, A. Corma, Angew. Chem.Int. Ed. 52 (2013) 11554.

[8] J.P. Tessonnier, L. Pesant, G. Ehret, M.J. Ledoux, C. Pham-Huu, Appl. Catal. A:Gen. 288 (2005) 203;A. Corma, H. Garcia, A. Leyva, J. Mol. Catal. A: Chem. 230 (2005) 97.

[9] M.A. Mahmoud, D. O’Neil, M.A. El-Sayed, Chem. Mater. 26 (2014) 44.[10] S. Guo, S. Zhang, S. Sun, Angew. Chem. – Int. Ed. 52 (2013) 8526.[11] A. Reyhani, S.Z. Mortazavi, S. Mirershadi, A.Z. Moshfegh, P. Parvin, A. Nozad

Golikand, J. Phys. Chem. C 115 (2011) 6994.[12] S.Z. Mortazavi, P. Parvin, A. Reyhani, R. Malekfar, S. Mirershadi, RSC Adv. 3 (5)

(2013) 1397.[13] L. Lascialfari, P. Marsili, S. Caporali, M. Muniz-Miranda, G. Margheri, A. Serafini,

A. Brandi, E. Giorgetti, S. Cicchi, Thin Solid Films 569 (2014) 93.[14] H. Zeng, X.W. Du, S.C. Singh, S.A. Kulinich, S. Yang, J. He, W. Cai, Adv. Funct.

Mater. 22 (2012) 1333.[15] Chang-Bing Hwang, Yaw-Shyan Fu, Yi-Ling Lu, Shu-Wen Jang, Pi-Tai Chou, C.R.

ChrisWang, Shuchun Joyce Yu, J. Catal. 195 (2000) 336.[16] T. Okada, J. Suehiro, Appl. Surf. Sci. 253 (2007) 7840.[17] M. Akbarzadeh Pasha, R. Poursalehi, M.A. Vesaghi, A. Shafiekhani, Physica B

405 (2010) 3468.[18] S.Z. Mortazavi, P. Parvin, A. Reyhani, A.N. Golikand, S. Mirershadi, J. Phys.

Chem. C 115 (2011) 5049.[19] G. Cristoforetti, E. Pitzalis, R. Spiniello, R. Ishak, M. Muniz-Miranda, J. Phys.

Chem. C 115 (2011) 5073.[20] G. Cristoforetti, E. Pitzalis, R. Spiniello, R. Ishakc, F. Giammanco, M. Muniz-

Miranda, S. Caporali, Appl. Surf. Sci. 258 (2012) 3289.[21] T. Nishi, N. Suzuki, N. Takahashi, K. Yano, J. Nanopart. Res. 15 (2013) 1569.[22] J. Kim, D.A. Reddy, R. Ma, T.K. Kim, Solid State Sci. 37 (2014) 96.[23] A. De Bonis, M. Sansone, A. Galasso, A. Santagata, R. Teghil, Appl. Phys. A 117

(2014) 211.[24] N.G. Semaltianos, P. Petkov, S. Scholz, L. Guetaz, J. Colloid Interface Sci. 402

(2013) 307.[25] A. De Bonis, M. Sansone, L. D’Alessio, A. Galasso, A. Santagata, R. Teghil, J. Phys.

D: Appl. Phys. 46 (2013) 445301.[26] E. Giorgetti, M. Muniz-Miranda, P. Marsili, D. Scarpellini, F. Giammanco, J.

Nanopart. Res. 14 (2012) 648.[27] M. Tiberi, A. Simonelli, G. Cristoforetti, P. Marsili, F. Giammanco, E. Giorgetti,

Appl. Phys. A 110 (4) (2013) 857.[28] F. Giammanco, E. Giorgetti, P. Marsili, A. Giusti, J. Phys. Chem. C 114 (2010)

3354.[29] E. Giorgetti, A. Giusti, S. Laza, P. Marsili, F. Giammanco, Phys. Status Solidi A

204 (6) (2007) 1693.[30] U. Kreibig, M. Vollmer, Optical Properties of Metal Clusters, Springer-Verlag,

Berlin, 1995.[31] J.A. Creighton, D.G. Eadon, J. Chem. Soc. Faraday Trans. 87 (24) (1991) 3881.[32] B.T. Sullivan, Appl. Opt. 1990 (1964) 29.[33] P.O. Nolsson, M.S. Shivaraman, J. Phys. C: Solid State Phys. 12 (1979) 1423.

96 E. Giorgetti et al. / Journal of Colloid and Interface Science 442 (2015) 89–96

[34] J. Calvo-Fuentes, J. Rivas, M.A. López-Quintela, in: Encyclopedia ofNanotechnology, second ed., Springer Science+Business Media B.V.,Dordrecht, The Netherlands, 2012.

[35] M. Brun, A. Berthet, J. Bertolini, J. Electron Spectrosc. Relat. Phenom. 104(1999) 55.

[36] K. McElney, C.M. Crudden, J.H. Horton, J. Phys. Chem. C 2009 (1901) 113.[37] Y.-F. Han, D. Kumar, C. Sivadinarayana, A. Clearfield, D.W. Goodman, Catal.

Lett. 94 (3–4) (2004) 131.[38] S. Ordez, F.V. Dìez, H. Sastre, Appl. Catal. B: Environ. 3 (2001) 113.

[39] W. Haiss, Nguyen T.K. Thanh, J. Aveyard, D.G. Fernig, Anal. Chem. 79 (2007)4215.

[40] F. Mafuné, J. Kohno, Y. Takeda, T. Kondow, J. Phys. Chem. B 105 (22) (2001)5114.

[41] N.N. Greenwood, Alan Earnshaw, Chemistry of the Elements, Pergamon Press,Oxford, 1984.

[42] M. Wilde, K. Fukutani, W. Ludwig, B. Brandt, J.-H. Fischer, S. Schauermann, H.-J.Freund, Angew. Chem. Int. Ed. 47 (2008) 9289.

[43] K.M. Neyman, S. Schauermann, Angew. Chem. Int. Ed. 49 (2010) 4743–4746.