You are on page 1of 9

Talanta 47 (1998) 143 151

Laser induced breakdown spectrometry of vanadium in titania supported silica catalysts

P. Lucena a, L.M. Cabal n a, E. Pardo b, F. Mart n b, L.J. Alemany b, J.J. Laserna a,*
a b

Department of Analytical Chemistry, Faculty of Sciences, Uni6ersity of Ma laga, E -29071, Ma laga, Spain Department of Chemical Engineering, Faculty of Sciences, Uni6ersity of Ma laga, E -29071, Ma laga, Spain Received 17 September 1997; received in revised form 4 February 1998; accepted 10 February 1998

Abstract The capability of laser induced breakdown spectrometry (LIBS) for vanadium determination in a x V 2TiO2 SiO2 catalyst is presented. The microplasma was generated onto the sample surface using a pulsed Nd:YAG laser operating in the second harmonic (532 nm). Laser produced plasmas were collected and detected using a charge-coupled device (CCD). In order to minimize the complex spectral interferences of emission lines and matrix effects a wide spectral range (210660 nm) was studied. The focusing of the laser beam on the surface was optimized to improve the signal-to-background ratio, and consequently the limit of detection. The analytical lines selected were used to evaluate the calibration curve. The detection limit for V was estimated to be 38 mg g 1 in 2TiO2 SiO2. The method precision expressed as relative standard deviation (RSD) was better than 6% in the concentration range 200 1000 mg g 1. 1998 Elsevier Science B.V. All rights reserved. Keywords: LIBS; Vanadium analysis; Catalysts; Quantitative determination

1. Introduction TiO2-supported vanadium oxides have been extensively studied and used due to their high catalytic activity and selectivity in many chemical reactions [1 7]. For instance, V/TiO2 is one of the most effective catalysts in the selective catalytic reduction (SCR) of NOx by NH3 [7,8]. Industrially, in these catalysts the anatase, a polymorph form of TiO2, is used as support for vanadium

* Corresponding author. Tel.: + 34 5 2131881; fax: + 34 5 2132000; e-mail:

oxide. Titania is the support more widely used for this purpose, although alumina, and to a lesser extent silica are also used. In spite of its wide use, titania suffers several drawbacks, including limited surface area, poor mechanical strength and a low sintering resistance. On the other hand, the interaction of vanadia with silica is weak and, consequently it results in a higher tendency to thermally induced aggregation with a poor dispersion of the active phase, while alumina-supported vanadia catalysts are susceptible to sulfation. The binary TiO2 SiO2 system, in principle, seems an ideal candidate to overcome the above disadvantages [9].

0039-9140/98/$19.00 1998 Elsevier Science B.V. All rights reserved. PII S0039-9140(98)00063-0


P. Lucena et al. / Talanta 47 (1998) 143151

It is envisaged that ternary V Ti Si systems display similar characteristics to those of binary VTi catalysts. In addition to the economic benets, the presence of silica grants much better mechanical properties to the system, which may allow its application in uidized bed reactors or extrusion into monolith reactors. The bulk, the surface physico-chemical properties and the reactivity of V Ti Si systems are strongly dependent on the method used for their preparation and the precursors used in the vanadia distribution. Vanadium loading determines the nature and distribution of vanadia species on the titania support. Therefore the structure and morphology of the TiO2 SiO2 substrate will determine to a large extent the dispersion of vanadia. In contrast to binary systems, ternary catalysts are still not fully understood. For VTi Si systems, it is very interesting to evaluate the dispersion and possible diffusion of vanadium into the support. Knowledge of the quantitative microstructure and composition are of great importance in understanding the possible correlations of distribution and composition with catalytic properties. Consequently, the development of analytical methods for these materials is of great practical interest. The capability of laser induced breakdown spectrometry (LIBS) for materials characterization has been widely demonstrated [10 14]. Recent analytical applications of LIBS include the determination of aluminum in zinc alloy [15], the simultaneous determination of aluminum, copper, iron, nickel, and zinc in alloys [16], the determination of copper in steel [17], and the detection of lead in concrete [18]. Surface analysis of photonicgrade silicon has been demonstrated [19,20]. However, no applications of LIBS for vanadium determination in catalysts have been previously reported. For this purpose, other surface analysis techniques (such as secondary ion mass spectrometry; X-ray photoelectron spectroscopy) can be used [21]. In comparison with those techniques, LIBS presents a number of advantages including the need for little or no sample preparation, the minute sample quantities needed, the possibility of work without controlled atmospheres and the rapid analysis time. In this paper, LIBS has been

evaluated for the quantitative determination of vanadium in x V2TiO2 SiO2 catalysts.

2. Experimental section

2.1. Apparatus
The LIBS system has been described in previous works [22,23]. Briey, it consisted of a pulsed Nd:YAG laser operating in the second harmonic (Continuum, model Surelite SLI-20, u = 532 nm, pulse width 5 ns), which was used to generate the microplasma. The laser energy at the sample was 4 mJ pulse 1. The laser beam was focused at normal incidence onto the sample surface using a planoconvex glass lens with a focal length of 100 mm and f -number of 4. The plasma image was collected by a planoconvex quartz lens with focal length of 100 mm and dispersed by an imaging spectrograph (Chromex, model 500 IS, tted with three indexable gratings of 300, 1200 and 2400 grooves mm 1). Two of the three gratings (300 and 2400 grooves mm 1) were employed in this study. The reciprocal linear dispersion was 20 and 2.5 nm mm 1, respectively. These values gave spectral coverages of 120 nm for the 300 grooves mm 1 grating and 15 nm for the 2400 grooves mm 1 grating with the detector used. The entrance slit width was 10 mm and the height was 10 mm. The spectrally resolved light was detected with a solid-state two-dimensional charge-coupled device (CCD) system (Stanford Computer Optics, model 4 Quik 05). The CCD consists of 752(h) 582(v) elements. The photoactive area is 6 4.5 mm2. The spectral resolution of the system was 0.16 and 0.02 nm pixel 1 using the 300 grooves mm 1 and the 2400 grooves mm 1 gratings, respectively. The CCD is equipped with an S 20 Q photocatode (spectral response from 180 to 820 nm) and an intensier system (microchannel plate, MCP). Operation of the detector was controlled by 4 Spec 1.20 software. Shutter and delay times can be selected in 50 ns steps. A fast photodiode was used as external trigger for exact synchronization of the incident laser pulse and opening of the camera shutter. The emission signal was corrected

P. Lucena et al. / Talanta 47 (1998) 143151


by subtraction of the dark current of the detector, which was separately measured for the same exposure time. Calibration of the detector system was conducted by using a mercury emission lamp and several lines emitted from a laser induced titanium plasma. Samples were placed on a manual XY Z translation stage to be moved with respect to the laser beam.

2.2. Samples
Catalysts were prepared using 2TiO2 SiO2 as a carrier for the ternary systems x V 2TiO2 SiO2. Silica microspheres (Aerosil 200 from Degussa) with size ca. 13 nm diameter and surface area of 200 m2 g 1 were used as starting material. The resulting x V 2TiO2 SiO2 material was obtained by co-deposition of titania and vanadia on the silica surface by incipient wetness impregnation, as described by Geuss et al. [24] and others [25 28]. The impregnation was made by adding a methanolic solution containing variable amounts of titanium tetraisopropoxide and vanadium acetil-acetonate to the support particles. The suspension was ultrasonically dispersed to ensure a good homogeneity. The supercial precipitation on silica occurs in a few minutes at room temperature. Powders were then washed and dried in air at 373 K overnight, and afterwards they were calcined at 773 K in air for 2 h. This procedure allows deposition of oxo-hydrated titanium vanadium, which yields dispersed TiO2, with vanadium incorporated into the TiO2 crystal lattice. Several samples with different vanadium loads were prepared, and they are labelled as x V2TiO2 SiO2, where x denotes the theoretical vanadium loading in mg g 1 and 2TiO2, two theoretical monolayers incorporated onto the silica surface. A monolayer was considered as the loading of titania for completed covering of silica surface by a 0.38 nm thick [26] lm of TiO2, which corresponds to the longest axis of the rutile unit cell. The titanium oxide incorporated on silica remains as small crystals of anatase covering the silica surface, as previously reported by Gala n-Fereres et al. [26]. A set of six sample pellets were grounded with a mortar and pelletized at a pressure of 7.5 Ton

cm 2 for 15 min. Approximately 0.15 g of mixture was pressed leading to samples of 13 mm in diameter and about 0.5 mm thick. Concentration of the calibration standards was in the range 20010000 mg g 1 vanadium in the 2TiO2 SiO2 support. In addition, for qualitative analysis other series of three pellets was prepared: one pellet with a 2TiO2 SiO2 support, the second one with vanadium in form of V2O5 and the third one with a 1:1 (w/w) V2O5 TiO2 mixture. As precision and accuracy of LIBS are highly dependent on sample composition, homogeneity, and surface condition, samples were carefully prepared according to the described methodology.

3. Results and discussion

3.1. Spectral analysis

In the analysis of solid samples by LIBS, the resulting plasma includes lines corresponding to the sample elements and matrix constituents. Thus, the goal in the quantitative determination of an element by LIBS is to nd a well-resolved line for the element of interest, free of matrix interferences. In the determination of V in x V 2TiO2 SiO2 samples, silica can be easily identied. A simple study of spectral range permits the choice of the optimal region for observing vanadium without spectral interferences of Si and O emission lines. The main problem in this kind of sample is the presence of TiO2. The large number of Ti lines along the UV-VIS region and their high intensities can complicate the qualitative and quantitative analysis. Five different spectral regions were studied for the most important emission lines of Ti and V in the range 210660 nm. The spectral window covering the range 404418 nm was chosen since it contains the most intense vanadium peak (411.18 nm) free from interference of neighbouring spectral lines. Fig. 1(ac) shows, respectively, the LIBS spectra corresponding to the 2TiO2 SiO2 matrix, to neat V2O5 and to a 1:1 (w/w) V2O5 TiO2 mixture. As shown in Fig. 1(c), the main peaks for Ti and V are clearly distinguished. In these gures, only the most intense emission lines were labelled [29]. Spectra were


P. Lucena et al. / Talanta 47 (1998) 143151

obtained using a single laser shot. Integration time and delay time were 1 ms and 300 ns, respectively.

3.2. Effect of laser beam focusing on the signal -to -background ratio
It is well-known that performance of LIBS for quantitative analysis is related among other factors to the signal-to-background ratio (S/B) [30]. Since the background in LIBS depends on the laser uence used, a study on the effect of laser focusing conditions on the S/B was performed. Fig. 2 shows the variation of the average signalto-background values and their precision (in terms of relative standard deviation, RSD) as a function of the relative lens-target distance. The values were calculated taking the background as the mean background signal along 30 pixels in an

interference free region and close to the peak of interest. In Fig. 2, the distance 0 indicates that the sample was placed at the lens focal length. Positive values of the relative focusing lens-target distance refer to the beam focused at a distance above the sample surface, while negative values refer to the beam focal position placed inside the material. From this gure, it is interesting to note that the S/B precision appeared approximately constant for relative lens-target distances from 2 to + 4 mm. However, the S/B reached a maximum when the focal point was placed 12 mm above the target surface. This value decreases drastically when the laser beam was defocused a few millimeters onto the sample surface because the laser uence decreases. Scanning electron micrographs of the craters produced by laser ablation in the 10000V2TiO2 SiO2 pellet surface at two differ-

Fig. 1. Single-shot LIB spectra corresponding to (a) 2TiO2 SiO2 support, (b) V2O5 and (c) 1:1 (w/w) V2O5 TiO2 mixture. The delay time was 300 ns. Acquisition time was of 1 ms. MCP gain: 700 V.

P. Lucena et al. / Talanta 47 (1998) 143151


Fig. 1. (Continued )


P. Lucena et al. / Talanta 47 (1998) 143151

Fig. 2. Signal-to-background ratio and its RSD (%) vs relative focusing lens-target distance. The V(I) line at 411.18 nm was used for the measurements and the results were obtained using the pellet 10000V 2TiO2 SiO2 support. Other conditions as in Fig. 1.

ent focusing distances are shown in Fig. 3. Micrographs A and B were taken respectively at the focal point and at a relative lens-target distance of + 2 mm, respectively. Two cumulative laser shots were used. Micrographs conrm that the 0 position presents the smallest irradiated surface. However, the efciency in ablating the sample, and consequently the S/B, can be increased when the laser beam is focused above the focal point ( + 2 mm) because the laser energy is still enough to allow ablation and the beam has a larger area of contact with the target surface. When further defocused, the laser uence becomes lower and it is less efcient in ablating the sample. Consequently, the optimum focal condition for both maximum ablation and higher signal-to-background ratio is at a focusing lens-target distance of + 2 mm. The irradiated areas and the laser uences for each focusing distance are summarized in Table 1. The ablated areas were calculated assuming an elliptical shape of the crater. As shown, a uence of 2.6 J cm 2 with ablated area of 18.6 10 2 mm2 were found for the optimum focusing conditions. It should be to noted that although the

plasma is formed at atmospheric pressure and above the focal point, the laser uence used is below the threshold uence for breakdown of air, and hence no lines corresponding to its several components are observed.

3.3. Quantitati6e analysis

It is well known that at early times following plasma formation, the LIBS spectrum is dominated by an intense radiation continuum and ionic emissions. Emission lines are broadened by the Stark effect. Temporal resolution was found to improve both linearity and signal reproducibility of the catalyst analysis. In this case, to compensate for the decreased signal at delayed integration, ten laser shots were accumulated. The optimal delay was estimated to be 1.3 ms after the laser shot. The use of internal standardization was necessary to compensate for the pulse-to-pulse variability and to minimize matrix effects. At delayed integration, continuum emission was reduced, but the intensity of the lines of the internal standard were also affected. For instance, the ionic Ti (II) line intensity at 416.37 nm decreased

P. Lucena et al. / Talanta 47 (1998) 143151


Fig. 3. Scanning electron micrographs of 10 000V2TiO2 SiO2 sample, showning the craters produced after two cummulative laser shots. (A) Sample placed at the focusing lens focal position. (B) Sample placed 2 mm below the focusing lens focal position.

signicantly, while the intensity of the neutral Ti line at 407.85 nm increased. For this reason, the atomic emission line of Ti was then chosen for internal standardization. In addition, this line satises the regular requirements of an internal standard, i.e. proximity to the analyte line and

freedom from interference with the neighbouring spectral lines. Several lines of V were evaluated to construct the calibration graph. The V(I) line at 411.18 nm was nally selected as it provides the largest sensitivity with the smallest standard deviation, thus


P. Lucena et al. / Talanta 47 (1998) 143151

Table 1 Irradiated areas and laser uences at different focusing lens-tosample distances Focusing lens-target distance (mm) 3 2 1 0 +1 +2 +3 Crater area (mm2, 102) 42.1 15.7 6.5 5.0 15.6 18.6 30.8 Fluence (J cm2)

1.0 2.6 6.2 8.0 2.7 2.6 1.3

presents good linearity, with correlation coefcient R 2 = 0.9966. Deviation from linearity was found at concentrations above 1000 mg g 1, probably due to self-absorption [31,32]. The detection limit (CL) was calculated from the formula: CL = 3s /S where s is the standard deviation of the V to Ti signal ratio at low concentration and S represents the method sensitivity calculated from the slope of the linear section of the calibration curve. The LOD from the time-resolved calibration graph using the s value at 200 mg g 1 V was 38 mg g 1 V. This value is well below the V level expected in x V2TiO2 SiO2 catalysts. The method precision was better than 6% RSD in the concentrate range 2001000 mg g 1. To check for the accuracy of the proposed method, a recovery experiment was performed. The results are summarized in Table 2. As shown, recovery values are satisfactory, better than 90%

The sample was 10 000 mg g1 V in a TiO2 SiO2 support.

leading to the best limit of detection. Fig. 4 shows the calibration graph for V in the 2TiO2 SiO2 support. This gure presents the ratio between the net line intensities of vanadium and internal standard as a function of V concentration. The net peak signal was obtained by subtraction of the background signal. As shown in Fig. 4, the curve

Fig. 4. Time-resolved calibration curve of V in a 2TiO2 SiO2 support. The intensity of the 411.18 nm V(I) line was ratioed to that of the 407.85 nm Ti(I) line. For construction of the calibration curve, mean values were calculated from six repeated measurements, each measurement representing an average of ten shots, from different target locations. The Inset shows the linear portion of calibration curve where each point represents the mean value.

P. Lucena et al. / Talanta 47 (1998) 143151 Table 2 Recovery of V in 2TiO2 SiO2 catalysts using laser-induced breakdown spectrometry Sample 1 2 3 4 Added V concentration (mg g1) 200 600 800 1000 Found V concentration (mg g1) 215.1 554.0 798.0 1030.3 Recovery (%) 107.5 92.3 99.75 103.3


Acknowledgements Research supported by Projects c PB94-1477 and PB93-1009 of the Direccio n General de Investigacio n Cient ca y Te cnica (Ministry of Education and Science, Madrid).

[1] M.A. Ban ares, L.J. Alemany, M.C. Jime nez, M.A. Larrubia, F. Delgado, M. Lo pez Granados, A. Mart nez Arias, J.M. Blasco, J.L.G. Fierro, J. Solid State Chem. 124 (1996) 69. [2] G.C. Bond, S. Flamerz, Appl. Catal. 71 (1991) 1. [3] S. Jansen, Y. Tu, M.J. Palmieri, M. Sanati, A. Andersson, J. Catal. 138 (1992) 79. [4] H. Kno zinger, E. Taglauer, Catalysis 10 (1993) 1. [5] J. Haber, T. Machej, E.M. Serwicka, I. E Wachs, Catal. Lett. 32 (1995) 101. [6] J.M. Gallardo Amores, V. Sa nchez Escribano, G. Busca, V. Lorentzelli, J. Mater. Chem. 4 (1994) 965. [7] J. M Jehng, I.E. Wachs, Catal. Lett. 13 (1992) 9. [8] Y. Cai, J. Ozkan, Appl. Catal. 78 (1991) 241. [9] L.J. Alemany, R. Mariscal, M. Galan-Fereres, M.A. Ban ares, J.A. Anderson, J.L.G. Fierro, Chem. Matet. 7 (1995) 1432. [10] L.J Radziemski, D.A. Cremers, in: L.J. Radziemski, D.A. Cremers (Eds.), Laser-induced Plasmas and Applications, chapter 7, Dekker, New York, 1989. [11] L. Moenke-Blankenburg, Laser Microanalysis, chapter 5, Wiley, New York, 1989. [12] L.J. Radziemski, D.A. Cremers, in: L.J. Radziemski, R.W. Solarz, J.A. Paisner (Eds.), Laser Spectroscopy and its Applications, chapter 5, Marcel Dekker, New York, 1987. [13] E.H. Piepmeier, Analytical Applications of Lasers, chapter 19, Wiley, New York, 1986. [14] S.A. Darke, J.F. Tyson, J. Anal. At. Spectrom. 8 (1993)

145. [15] D.E. Kim, K.J. Yoo, H.K. Park, D.W. Kim, Appl. Spectrosc. 51 (1997) 22. [16] T.L. Thiem, R.H. Salter, J.A. Gardner, Y.I. Lee, J. Sneddon, Appl. Spectrosc. 48 (1994) 58. [17] W.E. Ernst, D.F. Farson, D.J. Sames, Appl. Spectrosc. 50 (1996) 306. [18] A.V. Pakhomov, W. Nichols, J. Borysow, Appl. Spectrosc. 50 (1996) 880. [19] M. Hidalgo, F. Mart n, J.J. Laserna, Anal. Chem. 68 (1996) 1095. [20] J.M. Vadillo, S. Palanco, M.D. Romero, J.J. Laserna, Fresenius J. Anal. Chem. 355 (1996) 909. [21] G.C. Bond, S.F. Tahir, Appl. Catal. 71 (1991) 1 31. [22] M. Mila n, J.M. Vadillo, J.J. Laserna, J. Anal. At. Spectrom. 12 (1997) 441. [23] J.M. Vadillo, M. Mila n, J.J. Laserna, Fresenius J. Anal. Chem. 355 (1996) 10. [24] J.W. Geuss, in: G. Poncelet, P. Grange, P.A. Jacobs (Eds.), Preparation of Catalysts III, Elsevier, Amsterdam, 1983, p. 1. [25] N.E. Quaranta, V. Cortes Corbera n, J.L.G. Fierro, Stud. Surface Sci. Catal. 72 (1992) 147. [26] M. Gala n-Fereres, R. Mariscal, J.A. Anderson, L.J. Alemany, J.L.G. Fierro, J. Chem. Soc. Faraday Trans. 90 (1994) 3711. [27] M. Galan-Fereres, L.J. Alemany, R. Mariscal, M.A. Ban ares, J.A. Anderson, J.L.G. Fierro, Chem. Mater. 7 (1995) 1342. [28] L.J. Alemany, M.A. Ban ares, E. Pardo, F. Mart n, M. Gala n-Fereres, J.M. Blasco, Appl. Catal. 13 (1997) 289. [29] J. Reader, C.H. Corliss, W.L. Wiese, G.A. Martin, Wavelengths and Transition Probabilities for Atoms and Atomic Ions. NRDS-NBS 68, US Government Printing Ofce, Washington, DC, 1980. [30] T.R. Harville, R.K. Marcus, Anal. Chem. 65 (1993) 3636. [31] F. Leis, W. Sdorra, J.B. Ko, K. Niemax, Mikrochim. Acta 2 (1989) 185. [32] K.J. Grant, G.L. Paul, J. ONeill, Appl. Spectrosc. 45 (1991) 701.