The influence of true simultaneous internal standardization and background correction on repeatability for laser ablation and the slurry technique coupled to ICP emission spectrometry† Joachim No�lte,*a Franziska ScheZer,b Sabine Mann,‡c and Michael Pauld aBodenseewerk Perkin-Elmer GmbH, D-88647 U� berlingen, Germany bMartin-Luther-Universita�t Halle-Wittenberg, Institut fu�r Technische Chemie, D-06108 Halle, Germany cInstitut fu�r Umweltforschung, Universita�t Dortmund, D-44221 Dortmund, Germany dPerkin-Elmer Deutschland, D-88647 U� berlingen, Germany Received 4th November 1998, Accepted 5th February 1999 For the solid sampling techniques laser ablation and the slurry technique, the influence of simultaneous measurement vs.sequential measurement of the signal and background and the internal standard was investigated on an identical data set. The experiments were carried out with an array spectrometer, which generates spectra rather than single points.These spectra, typically measured around the analyte wavelength, are measured simultaneously. The ICP spectrometer used has a ‘profiling’ mode, whereby the entrance slit is moved fourfold and the spectra are recorded with a small spectral shift in a sequential manner. These images are then combined to form a spectrum which is composed of simultaneous and sequential measured data. Hence, by carefully selecting the data points in a measured spectrum, the diVerence between simultaneous and sequential measurements can be estimated from an identical data set.In the example studied, simultaneous background correction and simultaneous internal standardization (typical for array ICP-OES) gave a mean RSD of 3.6%, sequential background correction and simultaneous internal standardization (typical for simultaneous ICP-OES) produced an average RSD of 6.1% and sequential background correction and sequential internal standardization (typical for sequential ICP-OES) resulted in a mean RSD of 6.4%.The RSDs reflect the solid sampling procedure. Some applications of laser ablation and slurry sampling are presented. (ICP-MS) than for LA-ICP-OES. In a literature survey,4 not Introduction taking monographs and review articles into account, 101 In inductively coupled plasma optical emission spectrometry references were found for LA-ICP-MS compared with 35 for (ICP-OES), the analyte signal is always calculated as the net LA-ICP-OES. The use of an array-type spectrometer greatly signal, being the diVerence of the gross signal (analyte plus improves the quality of data for ICP-OES combined with background) minus the background signal (measured oV-line).LA.5,6 The slurry technique is usually associated with AA (228 The analyte signal and the background signal are influenced references) and there are considerably fewer publications on by the excitation temperature, which depends on the mass slurry ICP-OES (58 references).However, a large number of being introduced into the plasma. The eVect on the analyte papers on the slurry technique for ICP-OES have been pubsignal can be compensated with the use of an internal standard. lished for materials which are comparatively easy to digest or The height of the background signal in practice is only have weak chemical bonds (e.g., biological materials or available by measurement. If a steady-state signal can be coal ).5–8 For these materials, slurry sampling represents a generated, the fact that the signal and background are meas- practical technique.ured in a sequential manner with conventional ICP emission In this work, material which is diYcult to digest was used spectrometers does not present a problem. to investigate the potential and limitations of these methods Solid sampling techniques typically produce a far less uni- of solid sampling and how simultaneous measurements can form flow of sample to the plasma,1 which causes a fluctuating help to improve the reproducibility.signal. The only way to approach this situation is to measure all signals at the same time. However, in conventional ICP emission spectrometers, both signals are measured sequentially, Experimental even in most types of simultaneous spectrometers. The advent Instrumentation of array detectors for ICP emission spectrometry allowed the measurement of both signal contributions simultaneously.2,3 The ICP emission spectrometers used were Optima 3000 and Much more information is available for laser ablation (LA) 3300 DV (Perkin-Elmer, Norwalk, CT, USA).The Optima combined with inductively coupled plasma mass spectrometry series of spectrometers use two segmented charge coupled device (CCD) detectors (SCD), one for the UV section and the other for visible light. The detector and optics have been †Presented at the 8th Solid Sampling Spectrometry Colloquium, described in detail elsewhere.9,10 The Optima 3000 has a Budapest, Hungary, September 1–4, 1998.vertical torch with a radial view plasma and the Optima 3300 ‡Present address: Perkin-Elmer Deutschland, D-88647 U� berlingen, Germany. DV has a horizontal torch with a choice of axial11 or radial J. Anal. At. Spectrom., 1999, 14, 597–602 597Table 1 Operating conditions for the ICP emission spectrometers standard should be measured at the same time. The diVerence in repeatability when varying the degree of simultaneity was Power 1300 W studied in a data set generated with SRM BCR 142 ( light Outer gas flow rate 15 L min-1 sandy soil ).For this purpose, a ‘worst case scenario’, a soil Intermediate gas flow rate 0.5 L min-1 with a high degree of heterogeneity, was chosen in order to Carrier gas flow rate 0.7 L min-1 Viewing height (radial viewing) 15 mm have a more pronounced eVect. For this comparison of sequential and simultaneous measurement modes on the identical instrument and data set, view.12 The operating parameters for the plasma are summar- the profiling option of the ICP spectrometer was used.It ized in Table 1. should be emphasized that the described profiling option is The ICP mass spectrometer used was an ELAN 6000 not the recommended mode of recording data. Its only purpose (Perkin-Elmer SCIEX, Concord, ON, Canada). It has a dual in this context is to use a sequential measurement mode in the stage detector for pulse and analog reading.The quadrupole system. Profiling is a fast sequential measurement option while settling time is 200 ms for the pulse mode and 3 ms for the partially retaining simultaneity. When invoking profiling, the analog mode. The integration time was set to 1 s and the dwell entrance slit is moved one quarter of the slit width four times time to 10 ms. The plasma power was set at 1000 W with a consecutively. The spectral image falls on to the detector four outer gas flow rate of 15 L min-1 and a carrier gas flow rate times with the wavelength slightly shifted.By carefully selecting of 0.95 L min-1. the points that are used for peak maximum and background For laser ablation, a Laser Sampler 330 (Bodenseewerk correction, one can either choose a situation where all points Perkin-Elmer GmbH, U� berlingen, Germany) was used. It are gathered simultaneously (the points which are spaced four houses an excimer laser filled with KrF operating at the data points apart) or a situation where the points are gathered fundamental wavelength at 248 nm.The energy pulse was sequentially (all points not four data points apart). Fig. 1 greater than 20 mJ with a pulse duration of 10 ns. The unit illustrates the data generation process for the profiling option. was operated at a pulse frequency of 5 s-1. A Tygon tube The repeatability vs. processing mode was calculated for a (1 m×6 mm id) was used to carry the ablated material directly data set generated by laser ablation of SRM BCR 142 ( light to the injector of the plasma torch.Axial viewing was used sandy soil ). Twelve firings were made on a single spot on the throughout. sample. After a read delay time of 20 s,6 the spectra were For slurry sampling, a magnetic stirrer was used. The slurries recorded with an integration time of 0.1 s using the profiling were aspirated using standard 0.6 mm id tubing materie was set intentionally to such a short time in (PTFE) and pumped at 1 mL min-1 through a standard cross- order to have a time comparable to the laser pulse frequency flow nebulizer with GemTips, connected to a Scott-type spray of 5 s-1 or a time lag between two firings of 0.2 s.When using chamber. It was tested for horizontally and vertically this option, four images are generated, so the total measureoriented plasmas. ment time was 0.4 s. Eight element and wavelength combinations were recorded. The selection is based mainly on Sample pre-treatment freedom from spectral interference and on the amplitude of the signals and included Cu at 327 nm, Fe at 273 and 302 nm, Laser ablation.If block samples (such as metals or glass) Mg at 279 nm, Mn at 260 nm, Si at 221 nm, Ti at 368 nm and were to be analyzed, disks of 32 mm diameter or small pieces Zn at 206 nm; the internal standard was Al at 237 nm. glued to a mounting block were directly mounted in the sample The overall relative standard deviation (RSD) for eight cell of the laser sampler.Powdered samples (soils) were pressed wavelengths and corrected with one internal standard was to pellets after the addition of 20% Spectroblend (Chemplex calculated for these 12 firings. For further data processing, the Industries, Tuckahoe, NY, USA). The mixture was pressed RSDs were again averaged. If background correction and for 3 min at 690 MPa in an IR press.6 internal standard were applied simultaneously, an average RSD of 3.6% was calculated.The opposite, performing all Slurry technique. The slurry technique was used for zeoliths measurements sequentially, produced an RSD of 6.4%, while (faujasite LSX, mordenite, pentasil ZSM-5 and the layer sequential background correction but simultaneous internal silicate hectorite) and boron nitride (BN). Particle size recstandard gave 6.1%, which does not diVer significantly. One ommendations in the literature are generally 2 or 3 mm.13–17 can deduce from this that sequential and simultaneous ICP The zeolith samples were ground to this size with a ball mill emission spectrometers, which typically measure the back- made of stainless steel (Wissenschaftlicher Gera�tebau, ground in a sequential manner, produce a similar quality of Magdeburg, Germany).For zeoliths, it was found that after data, which contrast with those generated with an array type 15 min a median particle size of 3 mm resulted. After prolonged instrument.All subsequent experiments were conducted with- periods of grinding, the particles coagulated to larger particles, out the profiling option so as to ensure simultaneous back- so grinding was stopped after 15 min. ground correction. Also the integration time was set to be Amounts of 50–100 mg of sample were weighed and 100 mL identical for all elements in a given method, so all readings of 2% v/v HNO3 were added to the powder. For the prepwere taken simultaneously in order to allow real-time internal aration of boron nitride slurries, one drop of 0.1% Triton-X standardization. solution was added.The mixture was stirred vigorously with As mentioned above, the measuring time was deliberately a magnetic stirrer, after mixing, prior to sample uptake and set to be very short. For normal applications, longer measuring during the measurement.18 times are used (typically 1–20 s), resulting in a repeatability of around 0.5–2%.6,19 Results and discussion Measuring mode and repeatability Laser ablation applications In a laser ablation application, ICP-MS and ICP-OES were For laser ablation, a repeatability of typically around 10% is achieved.This can be improved to a few per cent by internal compared for a homogeneity test of the main components in glass. For homogeneity tests, the absolute concentration is not standardization.6 However, it is strongly influenced by the homogeneity of the sample and the elements to be determined.of importance but rather how precise a measurement can be made in order to distinguish between data points gathered for For a non-steady-state signal it is important that the internal 598 J. Anal. At. Spectrom., 1999, 14, 597–602A B C D Fig. 1 The profiling option of the Perkin-Elmer Optima 3000 series ICP emission spectrometer produces four simultaneous images of spectra which are generated sequentially by slightly moving the entrance slit four times ( left), so the spectra A, B, C and D are collected one after another. These four images are then overlaid to produce one composite image (right). All spectra are plotted as intensity vs.wavelength. the same material. The major components of this speciality butions of the background have a comparatively larger impact than for higher concentrations. Thus, good repeatability and glass were Ba (53%), Al (2.6%) and Si (16%). Single shots were fired at the glass. Fig. 2 shows the dot raster generated.a good linear calibration function are indications of proper correction of background fluctuations. A calibration with four Si was used as an internal standard. The RSDs of the resulting signals were then calculated. For ICP-MS, being a fast sequen- standards gave correlation coeYcients of 0.995 for all elements. Examples of calibration functions of two critical tial technique, the RSDs were 2.3% for Al and 2.5% for Ba. For ICP-OES, with all simultaneous measurements, the corre- elements, Pd and Pt, are shown in Fig. 3. sponding RSDs were 0.6 and 0.5%, respectively, allowing statements on the homogeneity of the sample. Slurry technique applications In another application, contaminants in silver alloys were For applications of the slurry technique, zeoliths and BN were determined. These included Au (at 208.209, 242.795 and used. To characterize the particle sizes of the materials, scan- 267.595 nm), Bi (at 190.178 and 223.061 nm), Cd (at 214.438, ning electron microscope (SEM) images were recorded as 226.502 and 228.802 nm), Cu (at 224.700, 324.754 and shown in Fig. 4 for zeolith. The chemical composition of the 327.396 nm), Ni (at 221.647, 231.604, 232.003 and zeoliths varied considerably (Table 2). The data were measured 341.476 nm), Pb (at 220.353 nm), Pd (at 324.270, 340.458 and after a microwave digestion with HF, HCl and HNO3. The 363.470 nm), Pt (at 214.423 and 265.945 nm), Sb (at 217.581 water content was determined after heating at 1000 °C for 2 h.and 231.147 nm) and Zn (at 202.548, 206.200 and 213.856 nm). The sum of the components is around 93% for the first five The repeatability was found to be around and below 1% using materials and 100% for the other two. the Ag line at 206 nm for internal standardization. The concen- On introducing particles into the sample introduction trations of these elements were relatively low (in the range up system, namely into the large volume Scott spray chamber, a to 500 mg kg-1).For low concentrations, the noise contriconcern is that there would be a prolonged stabilization time of the signal when the sample is introduced or removed. Fig. 5 and 6 show wash-in and wash-out for hectorite. After about Fig. 2 Microphotograph of glass with equally spaced laser craters 641 Pd 324.270 Pt 265.945 0 0.05 0 0.05 321 Intensity/counts s–1 Concentration (wt%) Correlation coefficient : Correlation coefficient : 0.9996 0.9992 Fig. 3 Calibration diagrams for Pd at 324.270 nm and Pt at 265.945 nm from an excimer laser, operating at 248 nm, across the surface for homogeneity studies with ICP-MS and ICP-OES.With kind per- for purity determinations using LA-ICP-OES of silver alloys with the Ag line at 206.117 nm for internal standardization. mission from R. Sievers and S. Rings, University of Bonn, Germany. J. Anal. At. Spectrom., 1999, 14, 597–602 599Fig. 4 SEM image of zeolith NaY using a magnification of 5000.One can clearly identify clusters of crystals. The individual crystals have a diameter of slightly above 1 mm. Fig. 6 Wash-out behavior for hectorite introduced as a slurry using the same conditions as described earlier. The remaining high signal for Na can be attributed to contamination of the water used for dilution. The stability of the signal of this element in solution is comparable to that of the elements meared as slurries in Fig. 5. Fig. 5 Time diagram to illustrate the stabilization period for introducing slurries.In this case, hectorite was aspirated. At 4 s, a new measurement is made. It takes about 1 min until the signal reaches steady-state conditions. After this period, the signal remains stable. The intensities for Ca and K were multiplied by a factor of 10 for better visibility. 1 min, the signal stabilizes, which corresponds to the stabilization time for liquid analysis. For the remaining time of the Fig. 7 Stability of slurries of boron nitride.Occasional spikes are quasi-steady-state signal, the precision was calculated for a observed for all signals. When calculated for all 50 replicates, the series of five consecutive points and was of the order of 1–2%. RSD is 3.2%. On introducing slurries of BN, occasional spikes were observed, as shown in Fig. 7, that degraded the repeatability to an extent not acceptable when using a set of five consecutive trations. The normal routine in ICP-OES is to use matrixmatching even for aqueous solutions.This becomes more measurements as one would do in a routine application. The RSD calculated for all 50 readings in the given example was important the higher is the concentration of the matrix. It would be expected that for slurry sampling, a high matrix 3.2%, still too high for stoichiometric calculations but acceptable for purity determination. Since all elements follow the eVect would result. In contrast, some workers have suggested calibrating against aqueous standards for ZrO2 powders.20 same trend, the precision can be improved significantly by taking one element as internal standard.When Fe at 259 nm Since no matrix-matched standards were available, we tested whether it would be possible to calibrate against aqueous is used as an internal standard, the RSD for 50 readings is then 0.5% (Fig. 8). Since the concentration ratio in the solid standards. The results are presented in Table 3. The zeolith of faujasite type, LSX, shows acceptable agree- is important for stoichiometric calculations, this procedure is acceptable.ment with the results for the digested material. The results for the slurry are generally higher, which is also reflected by the As a next step, we tried to quantify the element concen- Table 2 Chemical composition (% m/m) of zeoliths determined after digestion Faujasite Pentasil Layer silicate Component LSX Mordenite-I Mordenite-II ZSM-5 hectorite SiO2 32 66 63 81 50.8 Al2O3 25 9 10 2.3 — Na2O 0.15 5 5.9 1.9 2.6 MgO — — — — 24.6 Li2O 7.8 — — — 2.3 H2O 28 14.3 13.8 8.1 21 Sum 93 94 93 93 101 600 J.Anal. At. Spectrom., 1999, 14, 597–602120000 10000 80000 60000 40000 20000 0 1 7 13 19 25 31 37 43 49 55 61 67 73 79 85 91 97 4 10 16 22 28 34 40 46 Replicate Intensity/counts s–1 52 58 64 70 76 82 88 94 Fig. 9 Repeat experiment to verify Si signal spikes (arrows) on reinserting the sample tube after the aspiration of air for 4 s.go in order to compensate for the low readings found during the quasi-steady-state measurements. Fig. 8 By referencing the readings of Fig. 7 to an internal standard A repeat experiment to verify this behavior is depicted in (in this case Fe at 259.940 nm), the RSD improves to 0.48% (n=50). Fig. 9 for the Si signal in pentasil ZSM-5. At first, it was assumed that the introduction of air would raise the signal intensity. In order to check this hypothesis, nitrogen and fact that the sum is calculated to be 105% for the slurry technique in contrast to 93% for the digested sample.When oxygen (up to 10% of the carrier gas) were added with the aid of another external mass flow controller. However, the signal calculating the Si/Al ratio, which is an important figure for characterizing a zeolith, there is acceptable agreement between intensity increased by only about 10%. It appears that the first particles entering the plasma still find the plasma in a much slurry (151) and digest (1.1 : 1).For the other materials, the Si and Al values diVer considerably, as does also the Si/Al hotter state. The introduction of more particles then cools the plasma until an equilibrium is reached. During this initial ratio. Si and Al make up the structure with strong bonds via an oxygen atom. The other constituents are loosely bound period, the particles are in an environment hot enough to ‘crack’ them completely. This assumption is also backed by and therefore show much better agreement. It is assumed that the particles are ‘digested’ in the plasma the fact that the Si/Na ratio increases when the temperature in the analyte channel of the plasma, reflected by the ratio of starting from their surfaces.The elements on the surface of the particles or with loose bonds are therefore more easily Ca ionic-to-atomic transitions, is raised by changing the power of the rf generator or changing the carrier gas flow.removed and atomized. This assumption is further backed by the fact that in zeolith NaY, Al is not homogeneously distributed but found preferentially on the surface. For this material, Conclusions the Al concentrations measured in the digest and slurry are in good agreement. Also, mordenite-I is a finer material than Solid sampling for ICP-OES has the potential to present analytical information when all information is collected in real mordenite-II, which is otherwise very similar in chemical composition and identical in structure.As a consequence, the time. A key role is the use of an internal standard. It is important that this is matched as closely as possible to the agreement is much better for the material which has a greater portion of smaller particles. Hence it can be concluded that analytes with respect to excitation conditions. Calibration reacts sensitively to the nature of the calibration standard, so the general rule to matrix-match must also be followed here.This applies to the material type and size distribution. it is imperative that the chemical composition and physical properties resemble each other as much as possible. During the stability runs, it was observed by coincidence that spikes of Si and Al signals appeared immediately after If these basic rules are followed, more information can be extracted when using the lateral information provided by laser re-aspirating the sample during a run when the tube had slipped out of the slurry mixture, similar to those shown in ablation with excellent precision and good accuracy.The slurry technique is very simple to realize. The main Fig. 9. These spikes were not observed for the other constituents. The magnitudes of these spikes were so large that when attraction of this technique lies in the minimum sample pretreatment for materials which are diYcult to digest. The the maximum intensity was used for concentration calculation, the result was very close to the expected concentration.Hence particle size is very important. In our experiments, it was typically around 3 mm, which appears to be too large for this these spikes appear to indicate in which direction one should Table 3 Chemical composition (% m/m) of zeoliths determined using slurry technique Faujasite Pentasil Layer silicate Component LSX Mordenite-I Mordenite-II ZSM-5 hectorite SiO2 39 18 12 35 12 Al2O3 30 3.2 0.7 0.5 — Na2O 0.3 6 6.8 2.8 2.8 MgO — — — — 23 Li2O 8 — — — 1.2 H2O 28 14.3 13.8 8.1 21 Sum 105 42 33 46 60 J.Anal. At. Spectrom., 1999, 14, 597–602 6017 E. Poussel and J.-M. Mermet, Spectrochim. Acta, Part B, 1996, type of material. The wash-in and wash-out behavior is 51, 75. favorable, as are the short- and long-term reproducibilities, 8 J. No� lte, J. Scho�ppenthau, L. Dunemann, T. Schumann and which can be further improved with an internal standard L. Moenke-Blankenburg, J. Anal.At. Spectrom., 1995, 10, 655. measured simultaneously. The slurry technique thus far has 9 T. W. Barnard, M. I. Crockett, J. C. Ivaldi and P. L. Lundberg, yielded acceptable to good results for elements on the surface Anal. Chem., 1993, 65, 1225. 10 T. W. Barnard, M. I. Crockett, J. C. Ivaldi, P. L. Lundberg, D. A. and loosely boto the core material when calibrated against Yates, P. A. Levine and D. J. Sauer, Anal. Chem., 1993, 65, 1231. aqueous standards. In order to obtain good results for elements 11 J. C. Ivaldi and J. F. Tyson, Spectrochim. Acta, Part B, 1995, that have strong bonds in a material, which consists of larger 50, 1207. particles, at present a perfect matrix match seems to be 12 M. DuVy and R. Thomas, At. Spectrosc., 1996, 17, 128. essential. In order to calibrate against aqueous standards for 13 J. G. Williams, A. L. Gray, P. Norman and L. Ebdon, J. Anal. At. Spectrom., 1987, 2, 469. these types of samples, more research is needed. 14 L. Ebdon, M. E. Foulkes and S. J. Hill, J. Anal. At. Spectrom., 1990, 5, 67. 15 I. Halicz and I. B. Brenner, Spectrochim. Acta, Part B, 1987, 42, 207. References 16 D. A. Laird, R. H. Dowdy and R. C. Munter, J. Anal. At. 1 M. Thompson, C. D. Flint, S. Chenery and K. Knight, J. Anal. Spectrom., 1990, 5, 515. At. Spectrom., 1992, 7, 1099. 17 P. Goodall, M. E. Foulkes and L. Ebdon, Spectrochim. Acta, Part 2 J. C. Ivaldi and J. F. Tyson, Spectrochim. Acta, Part B, 1996, B, 1993, 48, 1563. 51, 1443. 18 L. Ebdon and A. R. Collier, J. Anal. At. Spectrom., 1988, 3, 557. 3 J.-M.Mermet and J. C. Ivaldi, J. Anal. At. Spectrom., 1993, 8, 795. 19 M. Ducreux-Zappa and J.-M.Mermet, Spectrochim. Acta, Part B, 1996, 51, 333. 4 M. Sperling, PELIDAS—Personal Literature Data System, 20 R. Lobinski, W. Van Borm, J. A. C. Broekaert, P. Tscho� pel and Bodenseewerk Perkin-Elmer GmbH, U� berlingen, 1998. G. To� lg, Fresenius’ J. Anal. Chem., 1992, 342, 563. 5 J.No� lte, L. Moenke-Blankenburg and T. Schumann, Fresenius’ J. Anal. Chem., 1994, 349, 131. 6 J.No� lte, Fresenius J. Anal. Chem., 1996, 355, 889. Paper 8/08591I 602 J. Anal. At. Spectrom., 1999, 14, 597