Geochemistry, Geophysics, Geosystems

Absolute calibration of low- and high-field magnetic susceptibilities using rare earth oxides

Authors


Abstract

Low- or high-field magnetic susceptibility measurements have been routinely utilized in or even beyond paleomagnetism and environmental magnetism. Susceptibility meters, such as Bartington or Kappabridge, are now commercially available and have enabled us to measure a variety of samples readily. However, interlaboratory or interinstrument calibration is not yet faithfully performed on absolute or even relative scales. The reason is that no standard material is generally accepted for susceptibility calibration. Gadolinium or other rare earth oxides are considered as ideal standard materials because of the high chemical stability. We defined reference susceptibility values for all available paramagnetic rare earth oxides after carefully examining the Landolt-Börnstein database and selected four compounds Nd2O3, Gd2O3, Dy2O3 and Er2O3 as standard materials. We carried out calibration measurements using Bartington and Kappabridge susceptibility meters, a Magnetic Property Measurement System (Quantum Design) and a Vibrating Sample Magnetometer (Princeton Measurements Corporation). Linear relationships between measured and reference values were observed for all the instruments. In some cases, however, measured values deviated significantly from expected values, suggesting that calibration is indeed necessary. If any instrument is calibrated with these rare earth oxides, we can reliably compare absolute susceptibility values measured with any kind of instrument in different laboratories.

1. Introduction

Low-field magnetic susceptibility is the most commonly measured parameter in paleomagnetism and environmental magnetism. Nevertheless, absolute or even relative calibrations of susceptibility have not drawn serious attention to the rock magnetism and paleomagnetism community. The reason is that susceptibility measurements were primarily used for obtaining the temporal or spatial variation patterns, for example, to correlate multiple sediment cores or to acquire dominant frequencies in time series analyses [e.g., Blum, 1997]. In these cases, measured susceptibility values were not necessarily expressed in calibrated values with appropriate units.

Relative interlaboratory calibration of susceptibility using the same type of instrument (e.g., Bartington or Kappabridge susceptibility meter) was already exercised several times [e.g., Snowball et al., 1994]. In these calibration exercises one or a few sorts of reference samples (e.g., sediments containing biogenic magnetites) were prepared by a single laboratory and distributed to participating laboratories. Although these exercises were worthwhile in the light of cross checking susceptibility values with other laboratories, such reference samples are not generally available and cannot be applied for other type of instruments. It should be pointed out that the absolute value of susceptibility remained unconfirmed.

Absolute calibration of susceptibility is by far more important than relative interlaboratory calibration. If a reference material carries a well-defined susceptibility value and is generally available, absolute calibration can be performed without any distributor laboratory. In addition, such a reference material enables us to compare measured susceptibility values to those measured with different instruments. In order to prepare standard samples with a variety of size and shape, reference materials are preferably provided in powder form.

Some of paramagnetic compounds containing 3d transition metals (Mn, Fe, Cu etc.) have been traditionally used as standard materials for susceptibility calibration. This is because paramagnetic materials have neither field intensity nor frequency dependence of susceptibility. In addition, grain size or hysteresis effects are not necessary to be considered for paramagnetic materials contrary to ferromagnetic or ferrimagnetic substance. These materials have well defined mass-specific susceptibility values but are chemically unstable due to their nonstoichiometry. In contrast, rare earth oxides are chemically stable and most of them are paramagnetic. Therefore one rare earth oxide, Gd2O3, was chosen as a reference material for both interlaboratory and interinstrument calibration [Jackson, 2000; Jackson et al., 2002; Sagnotti et al., 2003]. Gd2O3 has been used also in the inorganic chemistry community as a susceptibility reference material [e.g., Mele et al., 2009].

Sagnotti et al. [2003] reported that highly uniform mass-specific susceptibility values were gathered from twelve laboratories using Bartington and Kappabridge susceptibility meters. Although their relative interlaboratory calibration exercise seemed to work well, the two susceptibility meters provided lower values by 9% and 6% respectively than the Gd2O3 reference value of 1.845 × 10−6 m3/kg calculated from the Landolt-Börnstein database [Holtzberg et al., 1970]. In the later version of this database [Köbler and Sauer, 1982], it was shown that an opposite sign was erroneously assigned to one of the paramagnetic Curie temperatures, which led to an incorrect reference susceptibility value for Gd2O3.

Paramagnetic rare earth oxides have systematically varying susceptibility values corresponding to the atomic number of the rare earth element. If we use several sorts of rare earth oxides, we can avoid being hindered by mistakenly assigned reference values and can confirm a linear relationship between measured and reference values. In this study, we examine reference susceptibility values of rare earth oxides, then measured their low- and high-field susceptibility values using various instruments.

2. Reference Susceptibility Values of Rare Earth Oxides

The Curie-Weiss law describes temperature dependence of magnetic susceptibility for paramagnetic materials

equation image

where χ is susceptibility, T is the absolute temperature, C is the Curie constant, and θ is the paramagnetic Curie temperature [Foex, 1957]. The Curie constant C is expressed as

equation image

where N is the number of magnetic atoms, μB is the Bohr magneton, peff is the effective Bohr magneton number and k is the Boltzmann constant. Susceptibility values can be given by specifying the two parameters: θ and peff, and hence we can obtain a reference susceptibility value at any temperature according to the equations (1) and (2). We adopted 293 K as a standard temperature, and corrected measured susceptibility values to those at 293 K based on the Curie-Weiss law. Mass-, volume-, or mole-specific susceptibilities are given by assigning corresponding N.

In the Landolt-Börnstein database, several sets of θ and peff for a single rare earth oxide are denoted as cited from original literatures. In handbooks [e.g., Lide, 1998], a susceptibility value at a fixed temperature (e.g., 293 K) is given by averaging all the sets of θ and peff or selecting some sets in the database. For Gd2O3 two θ values are negative [Foex, 1957; Hacker et al., 1964] and the other is positive [Arajs and Colvin, 1962] in the table of the Landolt-Börnstein database III/4a [Holtzberg et al., 1970]. Based on these θ and peff, Gd2O3 reference susceptibility values at 293 K denoted in several handbooks vary in a range from 1.70 × 10−6 to 1.84 × 10−6 m3/kg [e.g., Lide, 1998]. The value of 1.845 × 10−6 m3/kg adopted by Sagnotti et al. [2003] is close to the highest value that was obtained by simply averaging all the sets of θ and peff. In a later version of the Landolt-Börnstein database III/12c, however, the positive θ was corrected to be negative [Köbler and Sauer, 1982]. The susceptibility versus temperature measurements of Sagnotti et al. [2003] confirmed indeed the negative value for θ of Gd2O3. They determined an extrapolated Curie temperature of −18 K and we also did so. Such a negative paramagnetic Curie temperature implies that there are small anitiferromagnetic interactions between some clustered Gd3+ ions [e.g., Mele et al., 2009; Rada et al., 2009]. A same correction of sign was made also for dysprosium oxide in the database III/12c, suggesting that such errors are inevitable in this kind of database. Therefore the reference susceptibility value of 1.845 × 10−6 m3/kg for Gd2O3 [Sagnotti et al., 2003] should be regarded as just a wrong value.

If we use several sorts of rare earth oxides as reference materials, we can achieve more reliable calibration instead of relying on a single compound as Gd2O3. Even for a single compound a variety of θ and peff, which were obtained based on different methods or instruments of susceptibility measurements, are denoted in the Landolt-Börnstein database. Simple averaging does not always provide a reliable susceptibility. Not just depending on the measured values cited in the database, we need to examine the cited θ and peff in light of theoretical aspects.

Fortunately peff, which strongly influences susceptibility values, can be well theoretically predicted for rare earth oxides to be g √(J(J + 1)), where J expresses the magnitude of resultant total angular momentum number for rare earth elements and g is the Landé's g factor [Chikazumi, 1978]. J comes from the vectorial summation of orbital and spin angular momenta. Gadolinium, which has the largest magnitude of spin angular momentum within rare earth elements, does not exhibit a largest peff in its oxide form. This is because the orbital angular momentum is not quenched in rare earth elements contrary to 3d transition metals such as iron. In some rare earth elements J take to be zero as for La, Eu and Lu, and J becomes largest for Ho.

In Figure 1 we plotted all the data sets of susceptibility at 293 K calculated from θ and peff cited in Landolt-Börnstein database III/4a and III/12c for rare earth oxides. There is a significant variation in the susceptibility values even for a single rare earth oxide. Instead of simply averaging all the cited data sets, we chose a data set with closest peff to the predicted. Taking the magnitude of susceptibility values and their commercial availability into consideration, we selected Nd2O3, Gd2O3, Dy2O3 and Er2O3 as reference materials for susceptibility measurements (Table 1). These mass-specific susceptibility values span almost one order of magnitude from 3.83 × 10−7 to 3.07 × 10−6 m3/kg at 293 K.

Figure 1.

Variation of reference mass susceptibility values of rare earth oxides against the number of 4f electrons. Small dots denote all the susceptibility values cited in the Landolt-Börnstein database III/4a [Holtzberg et al., 1970] and III/12c [Köbler and Sauer, 1982], and open circles indicate the selected values for which the measured effective Bohr magneton number is closest to the theoretically predicted value.

Table 1. Reference Mass Susceptibility Values for Selected Rare Earth Oxidesa
Rare Earth OxidePredicted peffTabulated peffTabulated θ (K)χ at 293 K (m3/kg)
  • a

    Abbreviations are as follows: peff, effective Bohr magneton number; θ, paramagnetic Curie temperature; χ, mass susceptibility. Predicted peff is given by g√(J(J+1)), where g is Landé's g factor and J is the magnitude of resultant total angular momentum number. Tabulated peff and θ were taken from the Landolt-Börnstein database III/4a [Holtzberg et al., 1970] and III/12c [Köbler and Sauer, 1982].

Nd2O33.623.65−32.03.83 × 10−7
Gd2O37.947.90−18.41.74 × 10−6
Dy2O310.710.6−17.03.07 × 10−6
Er2O39.589.56−9.602.48 × 10−6

3. Calibration Measurements

3.1. Materials and Instruments

We purchased Nd2O3, Gd2O3, Dy2O3 and Er2O3 powder that the manufacturer (Kojundo Chemical Lab. Co. Ltd.) ascertains the purity of 99.9%. The powder materials were packed into 1, 7 or 10 cm3 plastic cubes. The 1 and 7 cm3 cube are provided by Natsuhara Giken and widely used by Japanese paleomagnetists for susceptibility, remanence and hysteresis measurements. The 10 cm3 cube is not presently used for routine measurements, but is utilized to check volume effects in this calibration exercise. After taring a particular set of cube and cap, rare earth oxide powder was pushed into a cube as tightly as possible by a wooden stick and then capped. Mass of the cube filled with powder was measured and then we sealed them by a Scotch tape to avoid escaping powder. We prepared totally twelve cubes in order to measure susceptibilities by Bartington MS2 and Kappabridge KLY3 meters. For measurements using a Magnetic Property Measurement System XL (MPMS, Quantum Design) we sealed a few hundreds of mg powder by Saran Wrap and inserted it into a plastic straw. Aluminum foil, which has negligibly small susceptibility, was used to wrap a few tens of mg oxides for a Vibrating Sample Magnetometer (VSM, Princeton Measurements Corporation's MicroMag™ 3900) measurements.

Before measuring our own calibration samples, we measured the manufacturer-provided calibration samples for Bartington and Kappabridge meters. The Bartington susceptibility meter was connected to a dual frequency sensor MS2B operated at 0.47 kHz (low frequency (LF)) with the range of 1.0 or 0.1 SI. To measure 1 cm3 cubes, we inserted a 1 cm3 cube into an adapter made by polystyrene so as to position a 1 cm3 cube precisely at the center of coil, and measured it with an empty 1 cm3 cube to correct diamagnetic effects of the cube and adapter. For 7 and 10 cm3 cubes, diamagnetic susceptibilities of empty cubes were measured and subtracted from bulk susceptibility values.

We performed calibration measurements on cloudy or rainy days to minimize temperature variation during measurement. Rare earth oxide samples were stored in a desiccator to avoid hydration after packing powder into cubes. In order to make the samples equilibrium to room temperature near 20°C, samples were left next to the instruments more than one hour before starting measurement. In addition we recorded the atmospheric temperature in a course of measurement and corrected the measured susceptibility values to those at 293 K. We repeated susceptibility measurements seven times for each cube and calculated averages and standard deviations from resultant five values after excluding maximum and minimum values.

We used MPMS that is capable of generating up to 5 T with an alternating current (AC) susceptibility option. Measurement was performed for both low- and high-field susceptibilities at a fixed temperature of 293 K. For low-field measurements, an alternating magnetic field was generated with an amplitude from 0.05 to 0.42 mT and at a frequency of 500 Hz, which is close to that of the Bartington's low-frequency mode. The low-field susceptibilities were determined by fitting the linear relationship between field and magnetic moment. For high-field susceptibility measurements, we applied a maximum field of 1 T and then gave a series of back field down to 0.7 T. Again, we confirmed the linear relationship of magnetic moment against applied field between 0.7 and 1.0 T. It should be noted that the high-field susceptibility value of MPMS was previously checked by a manufacturer-provided palladium pellet with a similar size as the rare earth oxides and confirmed the error was only 0.3%.

For VSM measurements it is necessary to adjust the sample position relative to the pickup coils each time before measurement. We carefully performed positioning to minimize errors less than 0.5%. The magnetic moment was calibrated using a NIST's yttrium iron garnet (SRM2853). The magnetic field generated by the electromagnet was calibrated using two Hall probes of a same type: one was placed at the sample position and another was at about 1 cm away. The latter Hall probe exhibited lower field values by 2.7%. When actually measuring magnetic moment of a sample in an applied field, the field was monitored by the 1 cm away probe and then corrected to the field value at the sample position. The high-field susceptibilities were determined in a range of 0.7 to 1 T by VSM as well as by MPMS.

3.2. Calibration Results

We measured mass specific low-field susceptibility values for rare earth oxides packed in three kinds of cubes of 1, 7 and 10 cm3 by using Kappabridge and Bartington meters (Table 2 and Figure 2). When using a Kappabridge meter, measured susceptibility values exactly matched the reference values irrespective of cube's volume. Also we can see extremely high linearity in a wide range from 3.83 × 10−7 to 3.07 × 10−6 m3/kg. Although our Bartington susceptibility meter gave good linearity as well for each cube size, the measured values are always lower than the expected. The 10 cm3 cubes yielded 10.8% lower susceptibilities on average and 7 and 1 cm3 cubes exhibited lower values by 14.5% and 12.9% respectively. This may indicate size effect of cubes because Bartington MS2B coil is much smaller than that of Kappabridge. Sagnotti et al. [2003] reported that Bartington gave 4.5% lower values than Kappabridge on average from twelve laboratories by circulating ten Gd2O3 cubes (ASC Scientific, 2 × 2 × 2 cm).

Figure 2.

Comparison of reference and measured low-field susceptibility values by using Bartington (Bart), Kappabridge (KLY) susceptibility meters and Magnetic Property Measurement System (MPMS). Rare earth oxides (Nd2O3, Gd2O3, Dy2O3 and Er2O3) were packed in 1, 7, and 10 cm3 cubes for Bart and KLY, and sealed with plastic wrap for MPMS. Measured susceptibility values were corrected to 293 K based on the Curie-Weiss law.

Table 2. Measured Susceptibility Values for 1, 7, and 10 cm3 Cubes Using Bartington and Kappabridge Susceptibility Metersa
Cube Volume (cm3)Rare Earth OxidePowder Mass (g)Reading (×10−5)Average ± Standard DeviationMeasurement Temperature (°C)χ at 293 K (×10−7 m3/kg)
1234567
  • a

    Readings for the Bartington meter were measured in the low-frequency mode (0.47 kHz) with the range of 0.1 SI. Averages and standard deviations of readings were taken from five readings after excluding maximum and minimum readings, and converted to mass susceptibility χ following Dearing [1999]. Temperature corrections were applied to mass susceptibilities according to the Curie-Weiss law (equation (1) in the text).

Bartington
1empty cube + adapter −0.8−0.7−0.7−0.7−0.7−0.7−0.8−0.7 ± 0.1  
 Nd2O32.2246.66.76.66.66.66.66.66.6 ± 0.021.53.3 ± 0.0
 Gd2O32.69240.640.740.640.640.540.640.640.6 ± 0.021.515.4 ± 0.0
 Dy2O33.28286.886.686.586.686.486.486.486.5 ± 0.121.526.7 ± 0.0
 Er2O33.58876.176.076.075.975.976.075.876.0 ± 0.121.521.5 ± 0.0
7empty cube −0.2−0.3−0.3−0.3−0.3−0.3−0.2−0.3 ± 0.0  
 Nd2O313.1141.541.741.541.641.541.641.541.5 ± 0.121.53.26 ± 0.00
 Gd2O317.02254.3253.9254.0253.9253.9253.9253.8253.9 ± 0.021.515.05 ± 0.00
 Dy2O321.28554.1553.9554.7553.6553.9554.0553.8553.9 ± 0.121.526.22 ± 0.00
 Er2O322.10460.2460.1460.4460.4460.0460.1460.3460.2 ± 0.121.520.99 ± 0.06
10empty cube −0.3−0.2−0.2−0.3−0.3−0.2−0.3−0.3 ± 0.1  
 Nd2O319.9467.067.067.067.066.967.066.967.2 ± 0.021.53.39 ± 0.00
 Gd2O324.31381.3380.9379.6380.8379.7379.4380.2380.5 ± 0.621.515.73 ± 0.02
 Dy2O329.08789.5789.0788.8790.3788.5789.9791.0789.8 ± 0.621.527.31 ± 0.02
 Er2O333.90735.8730.2732.5731.7734.7733.1736.0733.8 ± 1.725.522.05 ± 0.05
 
Kappabridge
1empty cube + adapter −0.7901−0.7965−0.7889−0.7892−0.7913−0.7983−0.7895−0.7913 ± 0.0030  
 Nd2O32.2247.6127.6197.6117.6087.6057.6117.6087.610 ± 0.00221.53.796 ± 0.001
 Gd2O32.69246.5946.5846.6446.5846.6146.6046.6546.60 ± 0.0221.517.70 ± 0.01
 Dy2O33.28299.0899.0599.1098.9899.0698.9499.0799.05 ± 0.0421.530.59 ± 0.01
 Er2O33.58887.1687.1187.1287.0687.1187.0987.0987.10 ± 0.0121.524.63 ± 0.00
7empty cube −0.3118−0.3792−0.3256−0.3298−0.3341−0.3370−0.3298−0.3313 ± 0.044  
 Nd2O313.1149.2749.1849.2949.3249.2249.2049.2649.25 ± 0.0421.53.802 ± 0.003
 Gd2O317.02300.1300.1300.5300.3300.4300.0300.1300.2 ± 0.121.517.75 ± 0.01
 Dy2O321.28650.3650.6650.0650.6649.9650.3649.8650.2 ± 0.321.530.74 ± 0.01
 Er2O322.10543.8543.9543.9543.7543.7543.9543.8543.8 ± 0.121.524.76 ± 0.00
10empty cube −0.2092−0.2115−0.1924−0.1819−0.1818−0.1944−0.1976−0.1951 ± 0.098  
 Nd2O319.9474.6474.7274.7074.6874.7174.7474.7574.71 ± 0.0221.53.775 ± 0.001
 Gd2O324.31427.1427.5427.4427.4427.6427.4427.6427.5 ± 0.121.517.68 ± 0.00
 Dy2O329.08881.9882.4882.4882.2882.7882.9882.8882.5 ± 0.221.530.52 ± 0.01
 Er2O333.90822.9822.4822.8822.8822.8822.6822.9822.8 ± 0.125.524.73 ± 0.00

Low-field susceptibilities measured with MPMS also showed high linearity in the same range (Figure 2). However, the measured values are lower than the expected by 4.0–6.2%. We calculated susceptibilities from the linear relation between amplitude of alternating field and magnetic moment that MPMS provided. For the AC mode of MPMS we could calibrate neither amplitude of alternating field nor magnetic moment independently. Therefore it is impossible to infer what causes the lower susceptibility values. If we obtain these calibration data for standard materials, nevertheless, it is possible to calibrate MPMS susceptibility values, which can be directly compared to those measured by Bartington or Kappabridge meters.

Figure 3 shows high-field susceptibilities for the four rare earth oxides (Nd2O3, Gd2O3, Dy2O3 and Er2O3). Both MPMS and VSM measurements gave values remarkably close to those expected, although only Dy2O3 exhibited slightly lower susceptibilities when measured with MPMS. The applied magnetic field and the induced magnetic moment were simultaneously measured for high-field measurements and then high-field susceptibility was calculated from the gradient of the moment versus field. Such excellent calibration results suggest that both field and moment are accurately determined for our small sized rare earth oxide samples. The manufacturers provided magnetic moment calibration specimens that are palladium for MPMS and yttrium iron garnet for VSM. They are proved to be faithful calibration materials and have similar size and shape with those of our rare earth oxides. Raw high-field susceptibility values measured by MPMS and VSM can be regarded reliable, if once applied field is calibrated accurately to the actual value at the sample position.

Figure 3.

Comparison of reference and measured high-field susceptibility values by using Magnetic Property Measurement System (MPMS) and Vibrating Sample Magnetometer (VSM). Measured susceptibility values were corrected to 293 K based on the Curie-Weiss law.

We would like to show how actual calibration is performed using a Bartington susceptibility meter with a MS2B sensor for 1 and 7 cm3 cubes containing the four rare earth oxides. First of all, it is necessary to choose a cube or another type of holder with which you want to measure natural samples and the diamagnetic background susceptibility of the holder needs to be measured. A reference bulk susceptibility value of a calibration sample can be calculated by multiplying the mass-specific reference susceptibility value (Table 1) with the mass of powder packed in the cube. If temperature is deviating from the standard temperature of 293 K, susceptibility should be corrected according to the Curie-Weiss law. Taking zero, put the calibration cube in the coil and push the measurement button as you usually measure natural samples, then record a raw reading shown on the meter's panel. A linear calibration function is obtained between readings and bulk reference susceptibilities from the four rare earth oxides (Figure 4). Then you just need to measure and record readings on natural samples routinely. Based on the calibration function you can convert raw readings to bulk susceptibilities that can be normalized by mass or volume of natural samples. Certainly these obtained mass- or volume-specific susceptibilities are absolutely calibrated values, even if the instrument does not give expected susceptibilities when following the standard measurement protocol that the manufacturer provided.

Figure 4.

Calibration exercise for a Bartington susceptibility meter with a sensor MS2B using 1 and 7 cm3 cubes containing Nd2O3, Gd2O3, Dy2O3 and Er2O3. Measurements were performed in the low-frequency mode (0.47 kHz noted as “LF”) with the range of 0.1 SI for 1 cm3 cubes and 1.0 SI for 7 cm3 cubes. A relationship between reference bulk susceptibility x (m3) and reading y is approximated by a linear function y = a + b*x. a = 0.334 ± 0.563 and b = 8.61 × 109 ± 7.89 × 107 (m−3) for the 1 cm3 cubes, a = 1.75 ± 3.52 and b = 8.48 × 109 ± 7.78 × 107 (m−3) for 7 cm3 cubes.

4. Chemical Stability of Transition Metal Bearing Compounds

We examined aging effects on low-field susceptibility using seven sorts of paramagnetic hydrates, oxides and carbonates that contain transition metals such as manganese (Mn), iron (Fe) and copper (Cu). Unlike rare earth oxides, these compounds are chemically unstable even at room temperature. The powder or grains were packed in 7 cm3 cubes whose mass and low-field susceptibilities were measured nine years ago. These cubes were sealed in plastic bags and had been kept at room temperature for nine years, and then we measured again the mass and susceptibilities. Low-field susceptibilities were measured by two different Bartington meters with MS2B sensors spanning the nine years. They were intercalibrated in relative scales using several newly packed standard samples that contain 3d transition metal compounds.

Ratios for mass and low-field susceptibility spanning nine years were plotted (Figure 5). Four materials, MnSO4(NH4)2SO4•6H2O, CuSO4•5H2O, Mn2O3, and FeSO4(NH4)2SO4•6H2O, showed no significant differences in mass but the susceptibilities have changed by several percents. Two iron-bearing hydrates, FeSO4(NH4)2SO4•12H2O and FeSO4•7H2O, exhibited significant reduction in mass and enhancement in mass-specific susceptibility. Such a susceptibility enhancement should be an apparent effect due to efflorescence. Near edges of the cubes some tints of stain were observed for both compounds. Manganese carbonate (MnCO3), whose color was completely changed from beige to black in nine years, showed 20% reduction in mass and 17% decrease in susceptibility. At room temperature manganese carbonate slowly dissolves into manganese dioxide and carbon dioxide gas. This slow oxidation should have caused reduction of mass and susceptibility for manganese carbonate.

Figure 5.

Nine year aging effects on mass and mass-specific low-field susceptibility for transition metal bearing compounds: dot 1, FeSO4(NH4)2SO4•12H2O; dot 2, FeSO4•7H2O; dot 3, MnSO4(NH4)2SO4•6H2O; dot 4, CuSO4•5H2O; dot 5, Mn2O3; dot 6, FeSO4(NH4)2SO4•6H2O; dot 7, MnCO3. Efflorescence of FeSO4•7H2O and FeSO4(NH4)2SO4•12H2O resulted in apparently enhanced mass susceptibilities, whereas oxidation of MnCO3 led to reduction in mass and susceptibility.

5. Discussion

Rare earth oxides fulfill necessary requirements for reference materials of susceptibility: paramagnetic behavior over a broad temperature range, powder which can be packed in any size and shape of containers, low electrical conductivity, and high chemical stability. Paramagnetic compounds bearing transition metal have been traditionally used as susceptibility standard materials in paleomagnetics laboratories [e.g., Thomas et al., 2003] because they are readily available at relatively low cost. However, one of the serious drawbacks is that these compounds are chemically unstable, which is associated with several possible valence states that transition metals can take. Our 9 year storage test for transition metal bearing paramagnetic compounds demonstrated that the mass of some compounds was reduced by more than 10% and susceptibilities were increased or decreased due to hydration, efflorescence or oxidation (Figure 5). Even if no significant change in mass was observed, susceptibilities have changed by several percents. The outermost 3d electrons in transition elements are associated with both chemical bonding and magnetism. In contrast, the 5d and 6s electrons in rare earth elements play only a role in chemical bonding resulting in stable trivalent oxides. The unpaired spins of 4f orbital, which are shielded inside the electron cloud, are responsible for the paramagnetism of rare earth oxides but not affected by chemical bonding.

In the interlaboratory calibration experiment using Gd2O3 by Sagnotti et al. [2003], ten Kappabridge susceptibility meters were tested prove to be cross calibrated within 1% but the average low-field susceptibility value was about 6% lower than the reference value of Gd2O3. They suggested that Kappabridge values should be multiplied by 1.06 to achieve absolute calibration. However, our Kappabridge KLY3 meters did yield the measured values within 1% to the reference values for four rare earth oxides (Nd2O3, Gd2O3, Dy2O3 and Er2O3) using three types of cubes (1, 7 and 10 cm3) (Table 2 and Figure 2). We assure that Kappabridge meters provide extremely accurate absolute values and no correction is needed when calibrated by the manufacturer-provided standard sample and temperature effect is considered based on the Curie-Weiss law. The 6% difference of Sagnotti et al. [2003] is due to their incorrectly high Gd2O3 value of 1.845 × 10−6 m3/kg that originated from the wrongly assigned sign for the paramagnetic Curie temperature θ in the Landolt-Börnstein database III/4a [Holtzberg et al., 1970]. In the later version III/12c [Köbler and Sauer, 1982] the sign of the paramagnetic Curie temperature for Gd2O3 was corrected, but we need to caution that susceptibility values of Gd2O3 still remain uncorrected in some handbooks.

Calibration relying on a single sort of material is prone to be affected by an error in the reference value. As in our absolute calibration exercise, several sorts of reference material should be combined to achieve reliable calibration in order to avoid errors inevitable in databases. Another advantage using several sorts of material is that linearity check between measured and reference values is possible. A reference susceptibility value at room temperature (e.g., 293 K) can be calculated from a set of paramagnetic Curie temperatures θ and effective Bohr magneton numbers peff. These two parameters are deduced from measured temperature dependences of susceptibility that are based on a variety of measurement methods and conditions as shown in original literatures [e.g., Arajs and Colvin, 1962]. As the number of available sets for a rare earth oxide never exceeds five, averaging all the available sets may results in an inaccurate reference value. For each rare earth element the effective Bohr magneton number is theoretically given from the orbital and spin angular momentum number [Chikazumi, 1978]. Instead of simply calculating the average, we did choose a data set having the closest peff to the theoretically expected.

Presently, a variety of susceptibility meters are commercially available and many kinds of specimens are prepared for measurements using these instruments. No standard susceptibility meter have been established yet by the paleomagnetics community. Although attention is drawn to interlaboratory calibration involving a particular measuring equipment, the calibration between different instruments measuring the same parameters needs to be improved, even in a single laboratory. In addition, low- and high-field susceptibilities are compared to separate paramagnetic and ferrimagnetic signals or measured susceptibility values are sometimes combined with other magnetic parameters (e.g., anhysteretic remanent magnetization (ARM)). For interinstrument calibration a combination of instrument's pickup coil and sample's size and shape significantly affect measured susceptibility values. Size effect is pronounced when using a small pickup coil as in a Bartington susceptibility meter (Figure 2). Although manufacturers provide calibration samples with particular shape and size, measured values still do not match reference values. It is advisable to prepare rare earth oxide samples that are packed in a same type of sample holder as natural samples you measure. Then you need to do calibration measurement for each instrument, and to record ambient temperature that is necessary to correct based on the Curie-Weiss law.

We do not intend to favor a particular instrument for susceptibility measurement over other instruments, rather we would like to stress that any instrument can be used and the measured susceptibility values can be directly compared if calibration is performed on an absolute scale. At present paleomagnetists should hesitate to compare published susceptibility data to their own data even if expressed in mass- or volume-normalized values with appropriate units. This is because absolute susceptibility calibration has never established and even published data must be regarded as relative values inherent to the instrument. If we secure the absolute values of susceptibility by calibrating using several kinds of rare earth oxides, we can literally exchange and confidently utilize susceptibility data that other paleomagnetists and rock magnetists provided using any instrument and any size or shape of specimen.

6. Conclusions

We could define reference magnetic susceptibility values for paramagnetic rare earth oxides that have ideal characteristics as susceptibility standard materials. We adopted reference values associated with expected effective Bohr magneton number, instead of averaging all existing data. Absolute susceptibility calibration is now recommended by using four kinds of rare earth oxides: Nd2O3, Gd2O3, Dy2O3 and Er2O3. These rare earth oxides are chemically stable, commercially available to any laboratory, and have a rather wide range of susceptibility values (3.83 × 10−7 to 3.07 × 10−6 m3/kg at 293 K). With some commonly used instruments for low- and high-field susceptibility measurements, linear relationships between the measured and reference values were always confirmed. However, measured susceptibility values somehow deviate from the reference values in many cases, indicating that calibration is needed for any specific instrument by preparing calibration samples with same size and shape as natural samples. If calibration is once accomplished using these rare earth oxides, both low- and high-field susceptibilities can be directly compared in absolute values even when obtained by any instrument at any laboratory.

Acknowledgments

We would like to thank Hirokuni Oda, Naoto Ishikawa, Toshitsugu Yamazaki, and Hitoshi Fukusawa for participating our previous interlaboratory calibration using transition metal bearing compounds. Yuhji Yamamoto and Joe Stoner, who are the members of Science and Technology Panel of Integrated Ocean Drilling Program (IODP), are acknowledged for encouraging publication. We are indebted to Simo Spassov and two anonymous reviewers for constructive comments that significantly improved the manuscript.

Ancillary