The Sm-Nd Isotopic Method in the Geochronology Laboratory of the University of Brasília

Nd isotopes represent one of the best tools to investigate the processes involved in the evolution of the continental crust and mantle. This is due mainly to the similar geochemical behaviour of Sm and Nd, both light rare earth elements, which inhibits their fractionation during most varied geological processes. In order to carry out crustal evolution studies in central Brazil, the Sm-Nd isotopic method was implanted at the Geochronology Laboratory of the University of Brasília. The Sm-Nd separation methodology is basically that described in Richard et l. (1976), with the addition of some improvements. In this study we describe in detail the methodology used in Brasília. Precision and accuracy were checked with the international standards such as JB3, BCR-1, BHVO-1 and La Jolla, and the following results were found: JB-3 (Nd=15.74 ppm and Sm=4.28 ppm), BCR-1 ( 143Nd/144Nd=0.512647±8, Nd=28.73 ppm and Sm=6.66 ppm), BHVO-1 (Nd=24.83 ppm and Sm=6.2 ppm) and La Jolla ( 143Nd/144Nd=0.511835±14).

commonly used is the method that uses the anion exchange resin and acetic acid -nitric acid and methanol as eluent (Hooker et al. 1975, O'Nions et al. 1977).It is more efficient than the one that uses HDEHP, however it is necessary to work with the 142 Nd isotope which has 142 Ce, as an important isobaric interferent.A fast technique is the chromatography of high resolution Liquid Chromatography of High Resolution -HPLC (High Performance Liquid Chromatography), and Ionic Chromatography of High Resolution -HPIC (High Performance Ion Chromatography) which present high resolution and speed (Cassidy & Chauvel 1989).

Separation Technique Used in Brasília
In Brasília we decided for the method using the HDEHP, with the use of the commercial LN-spec resin.This also consists of teflon powder impregnated with HDEHP, industrially prepared with a very small grain size (270 to 150 mesh).It presented very good efficiency for the separation of Ce and this element is almost totally absent in the Nd fractions (isobaric with 142 Ce).Another form of eliminating the interference of Ce is to avoid the use of 142 Nd isotope for fractionation and isotope dilution calculations.
Significant amounts of Pr were found in the Nd fraction, however this element does not represent an interferent when Nd isotopic measurements are done in the metallic form (Nd + ) instead of the oxide form, when small interference with Pr oxides, with masses 158 and 159 happen (Richard et al. 1976).In the same way, samples with high Ba concentrations, can also present an ineffective separation of Ba in the secondary column, which was proven to the case in the experiments in this study.However this does not interfere either in the Nd measurement, because it is ionized before the Nd analysis start due to its lower ionisation temperature (Thirwall 1982).Most of the Ba can be removed during the process of chromatographic separation, eluting the solution with HNO 3 in a cationic resin, before the secondary elution for separation of Sm and Nd (Verma 1991), or still using the methodology presented by Stray & Dahlgren (1995), which combines classic chromatography of ionic change and HPIC for the separation and quantification of REEs in geological samples, based on the method of Le Roex & Watkins (1990).
The separation procedure used in the University of Brasília was efficient for the separation of Sm and Nd and, although time consuming, the procedure is simple and sufficiently effective.This is based, firstly, in the separation of the REE group, using a cation resin (primary column) followed by the extraction of Nd and Sm through a partition separation of phase-reverse (HDEHP), both using HCl as eluent.

Extraction of Sm and Nd of Rock Samples
Sample Digestion 50 to 100 mg of rock sample is mixed with the spike and the mixture is dissolved in teflon bombs covered with a steel jacket, using 1 mL of distilled conc.HNO 3 and 4 mL of concentrated and An.Acad. Bras. Ci., (2000) 72 (2) distilled HF.Dissolution is followed by evaporation using a mounted evaporation system with infrared lamps and teflon capsules.The residue is taken again in HF:HNO 3 (4:1) mixture and back to the oven at ca. 190 • C for 4 days.After complete dissolution, the sample was dried down and 2 mL of concentrated HNO 3 were added.The solution was dried down again, a new attack with 6 mL of distilled 6N HCl follows.The solution should be absolutely clear and homogeneous at this stage.Complete evaporation of the solution sample and addition of 2 mL of distilled 2.5N HCl follows.During the course of the study the evaporation procedure was modified.Samples are now evaporated on hot plates, inside clean air cabinets (class-100 air) placed in fume cupboards.This allows cleaner environment for the evaporation, and also allows the evaporation of a large number of samples at the same time.
As already observed in other studies (Getty et al. 1993, Rehkamper et al. 1996and Sato 1998), we had successful attacks using Savilex capsules.Some samples, however, such as garnet-rich rocks and some ultramaphic rocks were not entirely dissolved and needed to be transferred to bombs.

Separation of the REEs (Calibration of the Primary Column)
A quartz column (i.d.= 8 mm and height = 15 cm) was packed with ∼ 2.2g (it evaporates for 60 • C) or 12 cm of cation resin Bio-Rad AG 50W-X8 200-400 mesh in aqueous solution.The sample solution is eluted in the column using HCl (Fig. 1).The REE group is collected in the fraction between 1 and 15 mL of 6N HCl, after elution with 32 mL of 2.5N HCl.Together, with the REEs, Y and Ba are also collected (Richard et al. 1976).The column is regenerated with ca. 15 mL of 6N HCl and stored in diluted acid solution.
The 2.5N HCl was standardized by titrimetry, with NaCO 3 (anhydrous) as base and methyl orange as indicator.

Separation of Sm and Nd (Calibration of the Secondary Column)
The secondary columns are made of teflon (Savilex) (i.d.= 5 mm and height = 10 cm) and packed with LN-Spec resin (liquid resin HDEHP-270-150 mesh powdered teflon coated with di-ethylexil phosphoric acid).Height of resin bed is 6.5 cm.The REE fractions were totally evaporated and re-dissolved in 200 µL of 0.18N HCl.This solution was loaded into the LN-Spec column.The Nd fraction was collected in 4 mL of 0.3N HCl after the initial 10 mL of 0.18N HCl (Fig. 2).After the extraction of Nd, 2 mL of 0.3N HCl were discarded and the Sm fraction was collected in 4 mL of 0.4N HCl, with a flow speed of ∼ 1 mL/30 min.The regeneration of the resin was achieved with 6 mL of 6N HCl.The column was conditioned again with the purified 3 mL of H 2 O (Nanopure) followed by 2×3 mL of 0.18N HCl.

Chromatographic Columns for Minerals
Smaller columns, where set up for extracting Sm and Nd from low-REE materials.
As shown in Fig. 3a, the calibration showed the efficiency of the small columns for the separation of the REEs, using the Bio-Rad AG50W-X8 resin.The resolution of the secondary column (Fig. 3b) was also very good for separation of Sm-Nd, Ce-Nd, and Sm-Gd.

REE Separation (Primary Column)
The column was packed with cation resin to a height of 7 cm (column of i.d.= 5 mm and height of 10 cm).The flow speed was set at ca. 1 mL/10 min.250 µL of sample were added and washed with three times 250 µL of 2.5N HCl.7 mL of 2.5N HCl were discarded and the REE fraction comes out in 5 mL of 6N HCl.The column was regenerated with 15 mL of 6N HCl.Satisfactory results have also been obtained working with a larger aliquot of sample (500 µL).

Sm-Nd Separation
The column was packed with the LN-spec resin to a height of 7 cm (∼ 100 mg) and 0.5 cm of anionic resin Bio-Rad 200-400 mesh was added on top.The flow speed was set at 1 mL/50 min.100 µL of sample in 0.18N HCl was added and washed three times with the same amount of 0.18N HCl.The elution was carried out (Fig. 3b) with 0.18N HCl.The first 6 mL were discarded and Nd was extracted in 3 mL of 0.3N HCl.Elution of 2 mL of 0.3N HCl, followed and Sm is extracted in 3 mL of 0.4N HCl.The column is regenerated with 5 mL of 6N HCl.

Partition Coefficient and Separation Factor
The selectivity (Table I) of the liquid HDEHP, using a fine graned (270-150 mesh) powdered teflon as support, showed better efficiency than using a coarse powder (60-40 mesh), due to the increase in the ion exchange capacity of the resin.In Fig. 4a, the ion exchange reverse behaviour of HDEHP resin is observed, where the light REEs are first eluted.For the separation of Ba-La-Ce-Pr-Nd, the best concentration of HCl to be used proved to be 0.18N.This has been already observed empirically during the calibration and later on with the determination of the partition coefficient (Fig. 4b).For the separation of Sm-Eu-Gd the best concentration of HCl proved to be between 0.35 and 0.4N (Fig. 4c and d).In this concentration, a large separation factor (ratio between the coefficients of distribution of these elements) is observed.The separation of Nd-Sm happens in concentrations ≤0.3N (Fig. 4e and f).The use of stronger acids will result in the elution of some of the Sm into the Nd fraction.

Analytical Procedure (Mass Spectrometry)
The fraction collected in the secondary column is evaporated with 2 drops of 0.025N H 3 PO 4 .The residue is dissolved in 1 µL of 5% distilled HNO 3 loaded onto a Re filament of double An.Acad. Bras. Ci., (2000) 72 (2) Fig. 3a -Columns calibration for minerals analysis: REEs elution, where was necessary the concentration correction from blank value obtained in ICP/AES for Ba element.
2 filament assembly.The mass spectrometer used was a Finnigan MAT 262 with 7 collectors and the analyses have been accomplished in static mode.The 143 Nd/ 144 Nd ratio was normalised using 146 Nd/ 144 Nd = 0.7219 and the decay constant used was the value revised by Lugmair & Marti (1978)   different rock standards used varied between 0.0006-0.0016%and the analytical uncertainty for 147 Sm/ 144 Nd ratio was smaller than 0.19%.

Analytic Control
International rock standards with very well-known Sm and Nd concentrations and Nd isotopic composition were analyzed.The analyses of these standards were necessary to test the reproducibility of the data and mainly to evaluate the reliability of the methodology in rocks of unknown composition.The standards used (Table III) were: BCR-1 (Ballast-USGS), JB3 (Basalt-GSJ), BHVO-1 (Basalt Hawaiian-USGS), JG2 (Granite-GSJ) and La Jolla.The rock standard most frequently used for interlaboratorial comparison is the BCR-1 (Table IV).Therefore, some analyses of this standard were carried out during this study.They revealed an average value for the 143 Nd/ 144 Nd ratio of 0.512632±2 (1σ , n=5).Concentration of Nd and Sm, were 28.73 ppm and 6.66 ppm respectively being in close agreement with the data obtained in other laboratories (Table V).The BCR-1 standard, that has been used for so many years, is not available anymore.Therefore, we also used the BHVO-1 standard (Table IV).However, because BHVO-1 is only a recent rock standard, not much data from other laboratories are available in the literature.
Nevertheless, results for BCR-1, as well as for BHVO-1, presented good precision and accuracy, with the values being comparable to the recommended values.
The results for JB-3 and JG-2 were used in the preliminary evaluation of the method.The results obtained for standard JB-3 (a basalt) (Table IV) were satisfactory, presenting good precision, although the Sm and Nd concentrations were consistently 2% below the expected value.The concentrations results for JG-2 (Table IV) showed much greater scatter around the average, showing low precision for the final mean value.This value, however, is still statistically the same as the recommended value.One interpretation for the low precision of the data obtained for JG2 is that the standard is not as homogeneous (regarding the REE) as the other three rocks standards.This is only to be expect, because JG2 is a granite standard.
The data for the La Jolla standard in this study tended to be somewhat lower (  is, however, statistically the same as the recommended if 2σ uncertainties are considered. The average value obtained for the La Jolla standard is comparable with the values obtained in other laboratories (Table V).Our mean value of 0.511835±0.000014is statistically the same as the expected value of 0.511850 (average of all laboratories).
Nd procedure blanks (Table VI) decreased from 385 pg in the beginning of this study to ca. 74 pg towards the end.Nowadays these blanks are commonly around 30 pg.Blanks tend to be smaller when dissolution is carried out in Savillex capsules.the LN-Spec resin contribute little to the total blank indicating that the largest contribution for the blank is the memory effect in the dissolution vessels, as well as the reagents.VIII) are isobaric interferents with 152 Sm + .These interferences can be suppressed by the average discharge due to the low ionization potential of these species, while the emission of the 152 Gd + is inhibited by its high ionisation potential.In this study it was observed that the signal of BaO + is reduced to negligible proportions during the analysis of Sm and its main contribution, as 138 BaO + , will only interfere with masses 154 and 155 of Sm, not representing therefore any important interference.Gd was found in inexpressive amounts, as shown in Fig. 6.The amount of Gd is very small and 160 Gd + does not appear in the mass scan.To avoid any possible interference of Gd with 152 Sm + , used for the concentration calculations, we carried out an evaluation of the amount of Sm collected in the chromatographic column (Figs. 6 and 7).Maximum contribution of the mass 155 and detection of the mass 160 is observed when Sm is collected in 4 mL of 0.5N HCl, indicating presence of Gd and/or the presence of oxide species as the 144 NdO + or even 144 SmO + (Fig. 7).In the 3 mL Sm fraction an insignificant amount of the mass 155 is present, which is probably due only to the contribution of the La oxide ( 139 La + present in the scan).It is  observed that, even collecting the Sm fraction in of 4 mL of eluent, the contribution of the isotope 160 of Gd is insignificant, being smaller than 5 mV, what means a contribution smaller than 0.05 mV of 152 Gd + , being therefore, negligible.amounts, little instability is observed in the beam.This is well displayed in Fig. 8 which shows that presence of Ce in the Nd fraction is incapable of generating beam instability.Other isotope present during the analysis of Nd is 141 Pr + , but its presence does not represent isobaric interference in the metal form in spite of this.Its interference only happens when Nd is analyzed in the oxide form.

CONCLUSIONS
The improved chemical extraction used in this study following the preliminary determination of the partition coefficient of REEs in the HDEHP resin uses the best possible conditions for extraction of Sm and Nd.The efficiency of the chemical separation appears in the spectrometric analysis where interferences with other elements are kept to a minimum.
In the mass spectrometry the presence of Ce and mainly Pr in the Nd fraction was observed.There is absolutely no isobaric interference between Sm and Nd using this procedure.In the Sm fraction there are detectable amounts of Eu, but this does not represent a problem, because its isotope do not interfere with the determination of the Sm isotopic ratios.The main isobaric interference of Sm, Gd, is not detected.Sm and Nd concentrations obtained for the JB3 and BCR-1 rock standards show good accuracy and precision, and isotopic ratio of 143 Nd/ 144 Nd for BCR-1 and La Jolla are comparable to the values obtained in other laboratories, demonstrating the reliability of the methodology used at the University of Brasília.
2. Representation of the Calculation of the Atomic Mass: An. Acad.Bras.Ci., (2000) 72 (2) (2) 3. Calculation of the Concentrations: no. of moles = mass of the element/A mass of the element = C(µg/g) × W (mass in g) Using the equations ( 1) and ( 2) and simplifying the equation above and rearranging for Cn(µg/g):

Calculation of Nd Concentration
Using the equations mentioned, for a mixed spike of Sm and Nd enriched in 150 Nd (used in the laboratory of geochronology of UnB), we have: where: C S = concentration 150 Nd in the spike = 0.01691492 µmols/g P .A. = atomic mass of Nd (144.24mol/g) A = natural abundance of 150 Nd = 5.6251% (Russ et al. 1971 andWasserburg et al. 1981) where: R M = (146/150)m = measured in the spectrometer and corrected for mass fractionation.R S = (146/150) of the spike = 0.00464026 (calculated from Table II) R n = (146/150) natural = 3.05352835 (calculated from Table I -Wasserburg et al. 1981)

Calculation of Sm Concentration
For a mixed spike of Sm and Nd enriched in 149 Sm: where: R M = (147/149) measured in the spectrometer and corrected for mass fractionation R S = (147/149) in the spike = 0.0033163 (calculated from Table V) R n = (147/149) natural = 1.088396 (calculated from Table IV -Wasserburg et al. 1981) C S = concentration of the 149 Sm in the spike = 0.01334888 µmols/g P .A. = 150.35mol/g A = natural abundance of the 149 Sm = 13.8504%

Correction for Mass Fractionation
Thermal ionisation sources, involving the evaporation of the sample starting from a warm filament is subject to the effect of mass fractionation (Potts 1987).
The exact mathematical form of the fractionation law, that describes the instrumental mass fractionation by thermal ionisation is not very well known.The isotope chosen for the correction of the mass fractionation should have great mass difference and their ratio should be also relatively a close to the unit to minimize uncertainties.For Nd the best choice is 150 Nd/ 142 Nd and the second best is 146 Nd/ 142 Nd (Wasserburg et al. 1981).The factor of fractionation of the isotope of Nd for unit of atomic mass, it is defined as (Wasserburg et al. 1981): For cases where fractionation is small, these three different laws supply the same corrected values for the required accuracy.However, for cases in which the fractionation is important, each law will supply a different value for the corrected values depending on the choice of the isotope (u,v) used to calculate the correction.
For Nd and presumably also for Sm, the law of exponential fractionation seems to be better to correct the instrumental fractionation than the linear or power law.Therefore, the use of linear or power laws can introduce significant errors in the ratio 143 Nd/ 144 Nd, and this only happens when running a highly fractional analysis.This effect can be minimised by the choice of a different pair of isotope for the normalisation.When 146 Nd/ 142 Nd is used the effect in 143 Nd/ 144 Nd is insignificant; however, for this normalisation, significant errors can be introduced in 150 Nd/ 144 Nd.Also when we compare data from different laboratories that normalise in a different way, the use of the linear and power laws introduce small, but significant errors.If the data collected have a small fractionation (−0.001 < α < +0.001), the choice of the fractionation law is not necessary.
The fractionation law was suggested in Russell et al. (1978) for the analysis of Ca and that An.Acad. Bras. Ci., (2000) 72 (2) also used the Rayleigh fractionation law.
that corresponds to the equation of the Linear law.Solving for f (Boelrijk 1968): Nd isotope -Application in the UnB Laboratory Reference: 146 Primary: 150 Secondary: 144 Linear law where:  Linear law where: M 1 = 149/152 and 1 = −3 M 2 = 147/152 and 2 = −5 The ratio (147/149) corrected = T 2 /T 1 (used for calculation of the concentration)  where, P a = weight of the sample (mg) and P s = weight of the spike (mg).
Fig. 7 -Mass interference for Sm.A -increased scale, B -with 2.6 volts for 149 Sm.
any isotope j: reference isotope (Nd-146; Sm-152) u: primary isotope (Nd-150; Sm-149) v: secondary isotope (Nd-144; Sm-147) A: atomic mass iA: atomic mass of an isotope An: atomic mass of a normal element As: atomic mass of a spike W: mass in grams P: n • of atoms C: conc. in g/mL F: atomic fraction Condensed Notation: R S : ratio of the spike R N : ratio of the normal element R M : mix (spike + normal) : difference among atomic mass of = uA−jA; = vA−jA; = iA−jA

TABLE II Statistical treatment for the obtained data
* analytical error for the ratio 147 Sm/ 144 Nd.

TABLE VI Experimental blank data for Nd in the Geochronol- ogy Laboratory of the Brasília
The Nd total blanks observed in Geochronology laboratory of the UnB is within the range of values obtained in other laboratories of the world (TableVII and Fig. 5).As shown in TableVI An.Acad.Bras.Ci., (2000)72 (2)

TABLE VII Summary of obtained data for total blank in other geochronology laboratories
Isobaric Interference for Sm and Nd Samarium 136 CeO + , 136 BaO + (Table

TABLE VIII Mass interferences in Sm isotopic determinations
Table IX) of BaO + is trivial.Most of Ba is vaporized before the Nd analysis.Additionally, the natural abundance of the interfering masses ( 130 Ba + and 134 Ba + ) are low (0.11% and 2.42% respectively).Ce interferes (TableIX) with 142 Nd, however, due to the difficulty in separating Ce of Nd totally, it was preferred to use mass 146 as the reference isotope for the fractionation and isotopic dilution calculations.Previous studies have reported the presence of Ce as an important factor causing instability of the ion beam.In this study even when present in great An.Acad.Bras.Ci., (2000)72 (2)

TABLE II Isotopic Ratios for Nd in mixed Spike, calibrated against with the Blankwash-Tech and BCR-1 rock standards
Spike previously calibrated in the laboratory of Montreal.

TABLE IV Isotopic Ratio for natural Sm used in the Isotopic Dilution
Wasserburg et al. 1981.et al. 1981.

TABLE V Isotopic Ratio for natural Nd in mixed Spike, cali- brated using Blankwash-tech and BCR-1 rock stan- dards.
* Montreal.