Hydrothermal synthesis of dioctahedral smectites

1 downloads 0 Views 848KB Size Report
The crystal-chemistry of a series of synthetic Al–Fe3+ smectites was studied in detail using near and mid infrared spectroscopy. Chemical and NIR data ...
Applied Clay Science 104 (2015) 96–105

Contents lists available at ScienceDirect

Applied Clay Science journal homepage: www.elsevier.com/locate/clay

Research paper

Hydrothermal synthesis of dioctahedral smectites: The Al–Fe3 + chemical series. Part II: Crystal-chemistry S. Petit a,⁎, A. Decarreau a, W. Gates b, P. Andrieux a, O. Grauby c a b c

Université de Poitiers, CNRS UMR 7285 IC2MP, HydrASA, 6 rue Michel Brunet, F-86073 Poitiers Cedex 9, France Monash University, Department of Civil Engineering, Wellington Road, Clayton, Victoria 3800, Australia Université Aix Marseille, CNRS UMR 7325 CiNaM, Campus Luminy, F-13288 Marseille Cedex 09, France

a r t i c l e

i n f o

Article history: Received 17 March 2014 Received in revised form 4 November 2014 Accepted 5 November 2014 Available online 28 November 2014 Keywords: Nontronite Infrared spectroscopy Near infrared Fe-beidellite Synthesis (b) parameter

a b s t r a c t The crystal-chemistry of a series of synthetic Al–Fe3+ smectites was studied in detail using near and mid infrared spectroscopy. Chemical and NIR data indicated a quite complete range of octahedral Al for Fe3+ substitution, and therefore, the solid-solution between beidellite and nontronite end-members was continuous and complete. The wavenumbers of several infrared absorption bands were correlated with the chemistry of the synthetic smectites, providing a useful tool to constrain their structural formulae and also for assisting in assignments of similar bands in natural smectites. The Al and Fe3+ cations were shown to be randomly distributed in the octahedral sheet of synthesized smectites. Despite the high availability of iron during synthesis, generally only a small amount of tetrahedral Fe3+ was observed. © 2014 Elsevier B.V. All rights reserved.

1. Introduction In 2:1 type clay minerals, natural smectites exhibit large chemical variability in both tetrahedral and octahedral sheets. For dioctahedral smectites, the most common octahedral cations are Al, Fe and Mg, while Al, and sometimes Fe, can substitute for Si in tetrahedra. For natural smectites, the complete octahedral Al–Fe3 + chemical series has already been described, but in the presence of Mg (Brigatti, 1983; Gaudin et al., 2004). While many papers deal with the synthesis of aluminous smectites (Kloprogge et al. (1999) and Lantenois et al. (2008) for a review), very few data exists dealing with the synthesis of Al–Fe3 +-smectites in the absence of Mg (Petit et al., 1995, and Andrieux and Petit, 2010 for a review of Fe and Al–Fe smectites synthesis). The synthesis of Mg free smectites is not straightforward and requires careful adjustment of pH and temperature, which depend specifically on the amount and relative proportions of Al and Fe (Andrieux and Petit, 2010). These smectites are particularly interesting as reference materials and may be suitable for providing spectroscopic fingerprints because their permanent charge resides only in the tetrahedral sites due to Al (or Fe3+) for Si substitutions. To better refine the structural chemistry of these smectites, Al and Fe3+ cations should be accurately assigned to octahedral and tetrahedral

⁎ Corresponding author at: Université de Poitiers IC2MP UMR 7285 CNRS B 27 - 4 rue Michel Brunet TSA 51106 F- 86073 Poitiers cedex 9.

http://dx.doi.org/10.1016/j.clay.2014.11.014 0169-1317/© 2014 Elsevier B.V. All rights reserved.

sites within the 2:1 layer structure; and they should be correctly distributed within the crystallographic sites. The aim of this paper therefore was to use Fourier transform infrared spectroscopy (FTIR) to study the crystal-chemistry of these Al–Fe3 + smectites. OH vibrational modes are the most often used as a local probe of the octahedral sheet composition, but lattice bands can also give structural information (Madejová et al., 1994; Petit et al., 1995; Gates, 2005; Petit and Madejová, 2013). For well-crystallized 2:1 type clay minerals, and especially trioctahedral talc (Petit et al., 2004a, 2004b; Petit, 2005), IR spectra are relatively well-resolved because bandwidths are narrow and spectral features can be accurately assigned. For dioctahedral smectites, local disorder and the inclination of the OH vibration cause broadening of absorption bands, resulting in difficulties with assignments. The IR study of the synthetic Al–Fe3 + smectitic series provides an opportunity to clarify the origin of some of these spectral features. 2. Experimental 2.1. Materials This study is based on 26 synthetic smectites samples (Table 1) most of which were previously described in Andrieux and Petit (2010). They are pure smectites, without any other crystalline phases present. Andrieux and Petit (2010) underlined the difficulty to synthesize Al–Fe smectites near the Al end-member. In order to have a more

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

twenty analyses of different particles were performed. Data given in Table 2 are the mean values of the measured chemistry of particles. XRD patterns of the (06–33) reflections were recorded using a Panalytical X'Pert Pro MPD diffractometer equipped with an X'Celerator detector operating with an angular aperture of 2.127°. Scanning parameters were 0.033°2θ as step size and 200 s as counting time per step over the 57–65° 2θ Cu Kα angular range. The divergence slit, the antiscatter slit and the two Soller slits were 0.25°, 0.5°, 2.3°and 2.3° respectively. The diffractometer was calibrated using a powder of silicon (00-0271402 ICDD reference XRD patterns). Reflections (111), (311) and (511) of Si were used. The differences between the theoretical and observed positions for the three Si reflections were less than 0.05° (2θ). The uncertainty on the position of the (06–33) reflections of smectites is then ± 0.002 Å. The (06–33) reflections were fit by Gaussian components using the Fityk software (free software Fityk 0.8.2 (http://unipress.waw.pl/fityk)). Mid infrared (MIR) spectra were obtained using KBr pressed pellets, 13 mm in diameter. The pellets were prepared by mixing 1 mg of sample with 150 mg of well ground KBr. Then, the mixture was pressed for 5 min at 5 kbars and 4 min at 12 kbars. The pellets thus prepared were kept at 110 °C overnight in an oven before analyses. The spectra were recorded at 4 cm−1 resolution, from 100 co-added scans, between 4000 and 400 cm−1 using a KBr beam-splitter and a DTGS/KBr detector on a Nicolet 760 FTIR spectrometer. The NIR spectra were recorded from 100 co-added scans at a resolution of 4 cm−1 between 4000 and 11,000 cm− 1 using a Nicolet 6700 FTIR spectrometer using a CaF2 beam-splitter and a SMART NIR Integrating Sphere using an InGaAs internal detector. NIR spectra were decomposed by minimising the difference in the fit of the sum of Gaussian components to the observed spectrum using the Solver AddIn in Microsoft Excel. Spectra were first truncated to an extended region of interest (EROI) from 4700 to 4200 cm−1. Two background Gaussian functions were fit to the tails of the EROI on either side of the ROI to be fit (Fit-ROI), which itself was constrained between 4270 and 4620 cm− 1. In general, these background functions were centered near 4700–4750 cm− 1 and 3840– 3860 cm−1. The center, width and intensity of the background functions were allowed to float unconstrained until an acceptable reproduction of the spectral wings on either side of the Fit-ROI was obtained. These background functions were then subtracted from the EROI, with the minimum value being forced to zero. Care was taken to ensure that this method of background subtraction did not result in spectral tilting or distortion. One to five Gaussian functions were then introduced at appropriate centers by initially adjusting centers (C), widths (W) and intensities (I) manually. The Solver AddIn was then used to iteratively decompose the selected components so that the least-squares difference between the sum of all components (i.e. the sum of squares of the fitted line) was minimised compared to the observed spectrum. For most spectra, all parameters (i.e., C, W, I) of selected components, or a given parameter for all components (e.g. W), were temporarily fixed to achieve a reasonable fit without straying to a false minimum,

Table 1 Experimental conditions of synthesis. pHi: initial pH of synthesis; pHf: final pH of synthesis; T°C: synthesis temperature. Stoichiometry of starting gel (Si/Al/Fe)

Synthesis conditions pHi

pHf

T (°C)

3.25/0/2.75 3.94/0.31/1.75

12.5 10.5 10.5 10.5 10.2 10.3 8.7 8.7 9.8 8.7 10 8.4 6.1 6.9 9.2 9.2 6.3 8.4 7.5 6.5 6.4 7.1 6.5 6.5 7.4 7.4

12.8 10.8 10.8 10.7 10.5 10.7 9.7 9.3 10.8 9.3 10.6 10.4 10.4 10.9 8.7 8.7 6.7 7.7 7.8 8 7.1 10.2 9.8 9.4 9.3 9.5

150 150 170 200 150 200 200 220 200 220 220 200 200 200 200 220 200 200 200 220 220 220 220 220 220 240

3.58/0.67/1.76 3.78/0.67/1.55 3.61/0.72/1.67 3.60//1.01/1.39 4.33/0.76/0.91 3.96/0.97/1.07

4.03/1.11/0.86 4.12/1.09/0.79

4.54/1.03/0.43 4.14/1.35/0.51 4.12/1.55/0.33 4.05/1.72/0.23 3.99/1.84/0.17

97

Sample

0 1 2 3 39 14 41 42 43 44 47 19 20 21 22 50 23 25 26 51 48 52 53 54 31 33

continuous evolution of the Fe3 +/(Fe3 ++ Al) chemical ratio, other smectites were added (samples 47, 52, 53, 54, Table 1). These smectites were synthesized for 1 month at 200 °C, following the same procedure as for the other samples (Andrieux and Petit, 2010) and Ca-saturated. A refined study (Decarreau et al., 2008) demonstrated that the sample 0 is pure nontronite with the following structural formula: (Si3.25 Fe3+0.75) Fe3+2 O10 (OH)2 Na+0.75. Notably, iron is strictly ferric in this synthetic series (Decarreau et al., 2008; Andrieux and Petit, 2010).

2.2. Methods Chemical analyses of the starting gels were performed with a JEOL JSM-5600LV scanning electron microscope (SEM) equipped with an EDX system (Bruker AXS Microanalysis). The analyses were made from pressed pellets prepared with 50 mg of gel, which were then carbon coated. Chemical analyses of smectite particles were performed using a Phillips FEI Tecnai G2 Transmission Electronic Microscope (TEM) equipped with an EDX system. Samples were sedimented onto collodion-carbon coated-copper grids. For each sample from ten to

Table 2 Octahedral Fe and octahedral Al obtained from NIR (see text) and structural formulae [Si4 − x (Al, Fe3+)x] (Al, Fe3+)2 O10 (OH)2 R+x] of some synthetic smectites from TEM-EDX chemical analyses of particles for Fe total, Al total, and Si. The layer charge was obtained from Fe total, Al total and Si according to the above structural formula. Tetrahedral Fe was obtained combining TEM-EDX and NIR data. Sample

0

1

Fe total Al total Si Fetet Altet Feoct Aloct Layer charge Feoct NIR Aloct NIR

2.75 0 3.25 0.75 0 2.00 0.00 0.75

2.13 0.34 3.52 0.13 0.35 2.00 0.00 0.48

2

3

39

14

2.04 0.48 3.47 0.07 0.45 1.97 0.03 0.52

1.80 0.56 3.64 0.00 0.36 1.80 0.20 0.36

47

20

44

21

41 1.93 0.68 3.51 0.13 0.36 1.80 0.20 0.49

42

19

43

1.43 0.96 3.61 0.00 0.39 1.43 0.57 0.39

1.58 0.97 3.50 0.00 0.50 1.58 0.42 0.50

52

53 0.58 1.87 3.60 0.00 0.40 0.58 1.42 0.40

50

54

22

48

51

26

23

33

31

25

0.42 1.88 3.70 0.00 0.30 0.42 1.58 0.30

2.00 2.00 2.00 2.00 1.97 1.89 1.85 1.84 1.83 1.80 1.80 1.79 1.74 1.64 1.26 0.80 0.60 0.57 0.54 0.49 0.45 0.38 0.38 0.36 0.34 0.34 0.00 0.00 0.00 0.00 0.03 0.11 0.15 0.16 0.17 0.20 0.20 0.21 0.26 0.36 0.74 1.20 1.40 1.43 1.46 1.51 1.55 1.62 1.62 1.64 1.66 1.66

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

but were allowed to float freely for the final iterations. Special care was taken to minimise the differences on each component W so that within a given spectrum, these did not vary significantly. Nonetheless, because of the strongly overlapping nature of the components in the NIR, the W may have little physical meaning. 3. Results 3.1. Chemistry of smectite particles TEM-EDX was used to measure Si, Al and Fe contents of particles for about half of the synthesized samples. However, some of the data had to be omitted due to an unexplained contamination (anomalous occurrence of K, Mg, Ca, and sometimes Si in excess). Only EDX data showing no K, Mg and Si anomalies were preserved. Unfortunately, Ca was always in excess in these samples and then could not be used to evaluate the layer charge of smectites. Al (total), Fe (total), and Si were measured (Table 2). Unfortunately too, for the samples richest in Al (31, 52, 53, 54, Table 1), the structural formula could not be established because of an excess of Si probably due to an intimate mixing with residual unreacted gel. This feature is related to the difficulty to synthesize Al rich smectites without Mg (Andrieux and Petit, 2010). The synthetic smectites are of medium charge (from 0.3 to 0.5 per O10 (OH)2) with a 0.42 mean value. The Fe/(Fe + Al) (total) ratio measured for smectite particles by TEM-EDX and for the gels are the same (Fig. 1). On the basis of these results we can safely assume that all synthesized smectites are of medium charge and that their Fe/(Fe + Al) (total) ratio are those of starting gels. The distribution of Al and Fe within tetrahedral and octahedral sheets will be discussed later. 3.2. FTIR data 3.2.1. NIR (ν + δ) OH combination bands As for many natural ferruginous smectites, the (ν + δ) OH combination bands are broad and poorly resolved (Fig. 2). All along the chemical series, the (ν + δ) OH combination bands shift progressively from the combination bands observed for natural nontronites (4370 cm−1) (e.g. Madejová et al., 1994; Bishop et al., 1999; Gates, 2005) to the combination bands observed for natural montmorillonites–beidellites (4550 cm−1) (e.g. Post and Noble, 1993; Post and Borer, 2002). Up to five Gaussian bands were needed to fit all spectra (Fig. 3, Table 3). The band near 4375 cm− 1 is the major band of sample (0) which contains only Fe3+ as octahedral cation and was then attributed to Fe3+2–OH combination (Bishop et al., 1999; Petit et al., 2002; Gates, 2005; Decarreau et al., 2008; Andrieux and Petit, 2010;

Fig. 1. Correlation between Fe/(Fe + Al) (total) atomic ratio in starting gels (SEM-EDX data) and in synthetic smectite particles (TEM-EDX data). Full line slope = 1.

Sample Fe 1 2 39 14 41 42 47 19

Absorbance

98

20 21 52 53 22 23

Al

25 26 31 33 4600

4500

4400

Wavenumbers

4300

4200

(cm-1)

Fig. 2. NIR combination bands of synthetic smectites. From the top to the bottom: decreasing Fe/(Fe + Al).

Neumann et al., 2011). Indeed, this band results clearly from the combination of νFe3 +2–OH (stretching — 3565 cm− 1) and δ Fe3 +2–OH (bending — 813 cm−1) observed in MIR (see below). This band occurs in all spectra except those of samples (31) and (33) which have the lowest amounts of Fe. For samples (0), (1), (2) and (3) that are the richest in iron, a small isolated band occurs at 4510 cm−1. As sample (0) does not contain any Al at all, this band cannot be attributed to Al2–OH or Al–Fe3+–OH combinations and is probably due to another type of combination. The integrated intensity of this band is about 10% of the Fe3+2–OH band. The next block of samples in Table 3 (39 to 22) can be adequately fitted with three bands located at about 4379, 4471 and at 4530 cm−1. The band at 4379 cm− 1 is due to Fe3+2–OH as seen above. The band at 4530 cm−1 can be attributed to Al2–OH combination (according to νAl2–OH (3620 cm− 1) and δAl2–OH (912 cm−1) observed in MIR, see below). The 4471 cm−1 band is attributed to Al–Fe3+–OH combination (according to νAl–Fe3 +–OH (3598 cm− 1) and δAl–Fe3 +–OH (875 cm−1) observed in MIR — see below and Petit et al. (1995)). Similar assignments were made by Bishop et al. (1999) and Gates (2005). The intensities of the Al–Fe3+–OH and Al2–OH combination bands tend to increase, while the Fe3+2–OH band intensity simultaneously decreases, with the increase of the Al content of samples. To fit the next block of samples in Table 3 (22 to 33) five bands with mean positions at 4379, 4419, 4471, 4530, and 4573 cm−1 were necessary. Note that Table 3 reports the fitting of spectrum (22) by 3 or 5 components, the latter being significantly better than the former, the sum of least squares difference being respectively 6.5 × 10−5 and 3.3 × 10− 5, the sum of squares of the fitted line being 0.0384 (see above Section 2.2). The 4379, 4471, and 4530 cm−1 bands are assigned to Fe3+2–OH, Al–Fe3+–OH, and Al2–OH combination bands respectively, as seen above. The 4573 cm−1 band reaches its maximum intensity for the most Alenriched samples (31 and 33). This band was then attributed to a second Al2–OH combination. By analogy, the 4419 cm−1 band was tentatively attributed to a second Al–Fe3+–OH combination.

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

99

Fig. 3. Decomposition of the NIR combination band of samples 47 and 53. Dotted line: experimental; solid line: fit; band 1: 2Fe–(OH); band 2: Al(2)–Fe–(OH); band 3: Al(1)–Fe–(OH); band 4: 2Al(1)–(OH); band 4′: 2Al(1)–(OH) + combination band (see text); band 5: 2Al(2)–(OH) (refer to Table 3).

While the two Al2–OH bands (4576 and 4536 cm−1) have similar intensities, this was not the case for the two Al–Fe3+–OH bands: the 4419 cm−1 band always has a lower contribution than the 4471 cm−1 band. Similar bands than those described here for the synthetic smectites were already observed for natural smectites (Gates, 2005), with the exception of the 4419 cm− 1 band. Note, however, that while Gates (2005) did not separate Al–Fe3 +–OH into two distinct bands, both second Al2–OH and Al–Fe3+–OH bands assigned here match the

positions he observed for natural Al-rich smectites with sufficient octahedral Fe content. Both the low intensity of the second Al–Fe3+–OH as assigned here, and a possible overlap with another band such as Al– Mg–OH, may account for the fact that this second Al–Fe3 +–OH has never been distinctly observed previously in natural Fe–Al smectites. Assuming that the absorption coefficients for all X–Y–OH bands are the same (Madejova et al., 1994; Bishop et al., 1999; Vantelon et al., 2001; Gates et al., 2002; Gionis et al., 2007; Chryssikos et al., 2009),

Table 3 Decomposition of NIR combination bands. Center: position of the component (cm−1); FWHM: Full Width at Half Maximum (cm−1); S%: relative integrated intensity of the component; a : S% of this component was reduced by the removal of 10% of the S (2Fe–(OH)) (see text). Sample

Fe3+2–OH

Al(2)–Fe3+–OH

Center

FWHM

S%

0 1 2 3

4375 4377 4379 4379

84 60 59 60

87 92 93 91

39 14 41 42 47 43 44 19 20 21 52 22 22 50 23 25 26 51 48 53 54 31 33 Mean

4375 4376 4376 4376 4373 4377 4376 4375 4372 4373 4380 4397 4395 4377 4383 4383 4374 4380 4379 4386 4386

53 52 52 51 63 54 53 40 54 55 66 57 49 67 46 59 74 53 61 65 61

98 93 83 81 79 72 78 71 79 76 41 5 4 6 4 3 8 7 9 15 4

4379

Center

4420 4411 4420 4420 4418 4420 4418 4412 4424 4429 4421 4419

FWHM

61 85 58 65 74 57 61 65 61 65 54

Al(1)–Fe3+–OH S%

4 22 8 8 8 8 8 9 10 9 5

Center

4460 4464 4464 4465 4469 4462 4462 4470 4473 4470 4470 4492 4486 4473 4476 4469 4460 4468 4464 4475 4485 4487 4482 4471

FWHM

33 45 52 52 61 52 52 43 50 48 65 65 63 65 55 65 74 63 65 65 65 65 72

Al(2)2–OH

Combination band S%

2 4 15 16 13 19 16 19 11 14 40 43 41 26 22 20 14 24 23 40 38 25 30

Center

FWHM

4510 103 4526 90 4529 83 4529 87 2Al(1)–(OH) 4539 36 4540 56 4535 55 4525 50 4540 63 4526 72 4524 73 4542 39 4541 59 4537 61 4543 67 4560 64 4534 61 4516 65 4520 69 4514 65 4509 76 4517 68 4514 65 4523 65 4540 65 4527 65 4533 61 4530

S%

Center

FWHM

S%

4576 4568 4577 4574 4573 4575 4574 4561 4579 4578 4581 4573

44 66 59 65 62 67 66 65 61 65 65

19 23 29 35 28 29 28 21 23 46 39

13 8 7 9 0a 3a 2a 3a 0a 0a 0a 3a 2a 2a 19 52 32 23 37 34 42 32 32 15 25 20 26

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

it is possible to calculate the octahedral chemistry (on the basis of O10 (OH) 2) of samples using the relation:Fe3+oct = (2 ∗ S (Fe3+2–OH) + S (Al(1)–Fe3+–OH) + S (Al(2)–Fe3+–OH)) / 100, where S is the relative integrated intensity of bands given in Table 3, and Al(1), Al(2) refer to the two Al–Fe3+–OH bands (Table 3), and Aloct = 2 − Fe3+oct. Results are given in Table 2, and were used to calculate the structural formulae of samples for which TEM-EDX data are available (Table 2). Using the octahedral chemistry and the integrated intensities of X– Y–OH combination bands, it is possible to compare the actual distribution of Al and Fe3+ in the octahedral sheet of the synthetic smectites with a theoretical random distribution of these cations. If P (x,y) is the occurrence of the pair (x,y) in the octahedral sheet, - in the case of a theoretical random distribution: 2

P ð Fe– FeÞr ¼ ð Feoct =2Þ



P ðAl–FeÞr ¼ Fe

oct =2

6975

Sample Al 33 31 26 25 23 22 21 20 19 44 42

2

P ðAl–AlÞr ¼ ðAloct =2Þ 

7081

Absorbance

100

41



 ðAloct =2Þ  2

where Fe3+oct and Aloct are as defined above; - while for the actual distribution measured from NIR:    3þ P ð Fe– FeÞ exp ¼ S Fe 2 –OH =100      P ðAl–AlÞ exp ¼ S Alð1Þ 2 –OH þ S Alð2Þ 2 –OH =100 P ðAl–Fe  Þ exp    3þ 3þ ¼ S Alð1Þ – Fe –OH þ S Alð2Þ – Fe –OH =100:

Results revealed that Al and Fe3+ atoms are randomly distributed within the octahedral sheet of synthetic smectites, all along the series (Fig. 4). 3.2.2. NIR 2 ν–OH first overtone bands NIR 2 ν-OH first overtone bands are broad, with notably a large band centered at about 6850 cm−1 due to water partially overlapping. Maxima are at about 6975 cm− 1 for iron-rich smectites, corresponding mainly to the first overtone of the Fe3 +2–OH vibrations, or at about

14 39 3 2 Fe 1

7600

7200

6800

6400

Wavenumber (cm-1) Fig. 5. NIR overtone bands of synthetic smectites. From the top to the bottom: increasing Fe/(Fe + Al).

7081 cm−1 for Al-rich smectites, corresponding mainly to the first overtone of the Al2–OH vibrations (Fig. 5). Following Petit et al. (2004a, 2004b), the relation between the ν OH wavenumber in MIR and the 2ν OH wavenumbers in NIR is: Wν OH = (0.5 × W2ν OH) + 85.6. Consequently, the MIR Fe3 +2–OH vibrations should be expected at about 3573 cm−1 and those of Al2–OH 3626 cm−1, which agrees well with literature (e.g. Farmer, 1974; Petit et al., 1995). 3.2.3. MIR bands MIR spectra of samples 22, 23, 50, 51 are dominated by features of residual gel well evidenced by characteristic Si–O band at 1040 and 790 cm−1 (Table 4), and were therefore not taken into account in the following discussion. 3.2.3.1. OH stretching bands. These bands are broad and are largely overlapped by H2O O–H stretching bands (Fig. 2 in Andrieux and Petit, 2010). The bands are from 3620 for Al-rich smectites to 3562 cm−1 for Fe3+-rich smectites and correspond to νAl2–OH and νFe3+2–OH, respectively (Table 4). For octahedral Fe3+ (Fe3+oct) contents b 1.4 the band position remains constant at about 3620 cm−1.

Fig. 4. Occurrence of octahedral cations pairs (Al–Fe) in synthetic smectites: theoretical for a random distribution of octahedral cations versus experimental calculated from NIR data (Table 2).

3.2.3.2. OH bending bands. Three bands located at 920, 877 and 814 cm−1 (Table 4) were observed in the synthetic series and were attributed to δAl2–OH, δAl–OH–Fe3 + and δFe3 +2–OH (Fig. 6) (e.g. Farmer, 1974; Petit et al., 1995). The δFe3 +2–OH band is relatively strong and well observed for 1.2 b Fe3+oct b 2. The δAl2–OH band appears for octahedral Al (Aloct) greater than 1.4 per O11 and is observed as a shoulder only. This last feature is probably linked to the poor crystallization of the Al-rich synthesized smectites (Andrieux and Petit, 2010). The δAl–Fe3+–OH band is observed between 0.36 b Fe3+oct b 1.9 per O11. The Al–Fe3 +–OH band position ranges from 870 to 877 cm−1. These wavenumbers are consistent with those reported by Gates (2005) for natural smectites, whose δAl–Fe3+–OH band ranged from 868 to 870 cm−1 for nontronites and from 872 to 875 cm− 1 for beidellites. However, contrary to Craciun (1984) and Vantelon et al. (2001), no linear correlation could be observed here between the Al–

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

101

Table 4 Position and assignment (see text for details) of MIR bands for the synthetic smectites. a: MIR spectra dominated by residual gel features; w: weak band; sh: shoulder. Ech

Fe3+oct (NIR)

νSi-O

0 1 2 3 39 14 41 42 47 44 43 19 20 52 53 54 31 33

2 2 2 2 1.97 1.89 1.80 1.79 1.85 1.71

993 1010 1009 1007 1018 1020 1020 1024 1020 1022 1020 1028 1028 1030 1031 1034 1036 1036

1.74 1.84 1.26 0.80 0.57 0.34 0.36

δAl2–OH

δAl–Fe3+–OH

877

920 sh 917 sh 917 sh

R1

δFe3+2–OH

851 851 w 847 w 847 w 856 w 856

812 814 813 814 815 816 815 816 814 816 814 814 814 814

872 872 864 870 870 875 875 874 877 877 877 877

R2

R3

M 3+0ct-Oap

710 708 w 708 w 708 w

669 672 669 675 675 679 678 679 677 681 679 683 683 690 698

771 w 770 w 770 w 765 w 777 w 777 w 775 w 775 w 777 w 780 w 752 w 764 w 764 w

611 w 609 w 611 w 608 w 615 w 606 w 615 w 612 w 616 w 617 w

0.54

1040

914

875

790

702 w

50a 23a 51a

0.65 0.38 0.45

1040 1040 1040

912 912 912

875 875 875

791 791 791

702 w 702 w 702 w

3.2.3.3. Lattice bands. A strong Si–O stretching band from 1036 to 973 cm−1 is observed. Higher wavenumbers indicated significant

1036

534

490 451

669 812

Sample Fe 0 1 Absorbance

2 3 39 14

41 44 20 52 53 54 33

Al 1100

900

700

500

Wavenumber (cm-1) Fig. 6. MIR spectra of synthetic smectites in the 1200–400 cm−1 region. From the top to the bottom: decreasing Fe/(Fe + Al).

Si-O

Si–O (M3+oct)

Si–O (M3+oct)

R4

νOH

598 584 w 584 w 582 w 588 w 587 w 588 w 586 w 582 w 584 w 586 w 588 588 592

490 490 490 490 495 493 495 497 496 497 497 495 495 503 515 526 534 534

449 451 451 451 453 455 453 457 453 455 455 455 455 459 463 466 474 475

420 435 424 432 430 430 430 436–420 426 424 420 434 435 428–410 sh 412 sh

3562 3564 3564 3562 3568 3562 3562 3569 3562 3566 3564 3562 3562 3568 3597 3618 3628 3628

608 w 615 w 617 617

713 w 711 w 711 w

22a

OH–Fe3+ band wavenumber and Fe3+oct (Table 4). Marchel and Stanjek (2012) found a correlation between the δAl–Fe3+–OH band position and the tetrahedral charge of smectites. Following these authors, the tetrahedral charge of the synthetic smectites studied here ranges from 0.35 to 0.52. While these values of tetrahedral charges are consistent with those expected using chemical data (Table 2), the dependence of the δAl–OH–Fe3 + band wavenumber on tetrahedral charge is only roughly observed: for synthetic nontronites with tetrahedral charges from 0.3 to 0.5, the band position remains at 877 cm−1. OH bending bands were not sufficiently resolved to evidence possible splitting of δAl2–OH and δAl–Fe3+–OH bands as it was observed for combination bands.

Si-O

665 665 665

414 sh 414 sh

516

460

3602

519 519 519

463 463 463

3611 3602 3602

amounts of unreacted gel, as said above. The wavenumber of the strong νSi–O band decreased with total Fe3 + content of the synthesized smectites. A linear relationship exists between the wavenumber of this band and the octahedral Fe3+ following this regression (Fig. 7): ν (Si–O) = −9.5 ∗ Fe3+oct + 1040 cm−1. This regression indicates that the wavenumber of the ν (Si–O) band for a pure aluminous ((Si3.6 Al0.4) Al2 O10 (OH)2 Na0.4)-beidellite would be at 1040 cm−1, and at 1021 cm−1 for a ((Si3.6 Al0.4) Fe3+2 O10 (OH)2 Na0.4)-nontronite. Most natural nontronites having Feoct ≈ 1.8 were observed to have a ν Si–O band near 1020 cm−1 (Gates, 2005). For Fe3 + contents of the synthetic smectites N 2, the decrease of the wavenumber is amplified and the following linear regression was obtained (Fig. 7): 3þ

νðSi–OÞ ¼ −32  Fe

tet

−1

þ 1019 cm

:

This regression line was obtained with only four data points and as such the equation is only indicative. Nonetheless, comparing the two relations for ν (Si–O), tetrahedral Fe has 3 times more influence than octahedral Fe in lowering the ν (Si–O) band wavenumber. From the latter regression, the wavenumber of the ν (Si–O) band would be at 1006 cm−1 for a pure (Si3.6 Fe3+0.4) Fe3+2 O10 (OH2 Na0.4) nontronite. These results are also in agreement with the position of this band observed for essentially end-member natural nontronites (e.g. samples N11 and N12 in Gates, 2005). Three other bands have their position clearly linked to Fe3+oct content: the M3+oct–Oap out of plane deformation band (669–698 cm−1) and two Si–O–M3 +oct bands (490–535 cm− 1 and 449–475 cm−1 respectively) (Table 4, Fig. 8). The slopes of the regression lines are different for the three bands. As above, positions of these bands for theoretical end-members are extrapolated (Table 6). Lastly, two small bands corresponding to M3+oct–Oap–Si are observed at 611 cm−1 for Al and 588– 598 cm−1 for Fe (Table 4). A low intensity band at 851–856 cm−1 is observed for samples with the highest Fe3+oct values (R1 band, Table 4) as previously observed by Decarreau et al. (2008). This band is similar to the 841–850 cm−1 band observed in IR spectra of natural nontronites (e.g. Gates, 2005) and whose attribution is still debated (Gates, 2008; Decarreau et al., 2008). As the band is observed for smectites which contain no or very few octahedral Al, this band cannot be attributed to a δAl–OH–Fe3+ vibration

102

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

region clearly indicates the presence of only one band corresponding to δFe3+2–OH vibrations. Thus, assignment of this 841–850 cm−1 band in synthetic nontronite to a δFe3+2–OH bending mode as previously hypothesized by some authors for natural nontronites (Goodman et al., 1976; Keeling et al., 2000; Gates, 2005) is therefore inadvisable. Three other small lattice bands (R2, R3, R4, Table 4) were observed near 775, 710 and 414–430 cm−1, but are unassigned here due to insufficient information as to their origin. 3.3. XRD data

Fig. 7. Correlation between the Si–O band position and octahedral and tetrahedral Fe content of smectites. Diamond: smectite having less than 2 octahedral Fe; disk: smectite having 2 or near 2 octahedral Fe and tetrahedral Fe.

mode as previously hypothesized by Komadel et al. (1995). Gates (2005, 2008) observed that the band at 842–850 cm−1 was only present for natural nontronites with Feoct ≥ 1.48 and assigned this as a second δFe3+2–OH bending band. Gates (2008) observed a strong correlation of the ratio of the integrated intensities of this band and the δFe2–OH band near 818 cm− 1 with tetrahedral layer charge. However, as has been demonstrated here for synthetic nontronites, the NIR combination

The shape and the position of the (06–33) reflection were carefully studied. Simulation of this reflection was adjusted by using only one Gaussian curve (Fig. 9), showing that the XRD pattern of each sample arises from only one smectite lattice. Accordingly, the (b) parameter of synthesized smectites was measured from the maximum of the (06–33) reflection with b = 6 ∗ d (06–33). The (b) parameter of synthesized smectites correlates linearly with the total amount of Fe (Table 5, Fig. 10). On the basis of an O10 (OH)2 structural formula, the relation is: ∘  3þ b A ¼ 0:111  Fe tot þ 8:944:

The synthesized smectites having a layer charge near 0.4, 8.944 Å is the (b) parameter of a theoretical pure Al-beidellite with the [Si3.6 Al0.4] Al2 O10 (OH)2 Na0.4 structural formula. The (b) parameter correlates also with the Fe3+oct content (Table 5, Fig. 10): ∘  3þ b A ¼ 0:114  Fe oct þ 8:940:

Si-O-M3+oct

Using the b values of samples (0), (1), (14) and (39), for which Fe3 +oct is 2 or very close to 2, it is possible to correlate qualitatively (b) with tetrahedral Fe (Fig. 10): ∘  3þ b A ¼ 0:103  Fe tet þ 9:164:

y = -24.7x + 539 R2 = 0.97

Taking into account the scattering of experimental points around the regression lines, the slopes of the three regression lines can be considered as similar. It is then not possible to differentiate the role of tetrahedral and octahedral Fe on the (b) parameter. Opposite correlations were found between (b) both total Al and Aloct (Fig. 10):

Si-O-M3+oct

∘  ∘  b A ¼ −0:122  Altot þ 9:232; b A ¼ −0:116  Aloct þ 9:170:

This suggests that tetrahedral and octahedral Al have a similar impact on the (b) parameter. Thus, while the presence of Fe has a positive y = -12.1x + 476 R2 = 0.94

M3+oct-O

y = -18.3x + 713 R2 = 0.91

Fig. 8. Correlation between lattice band wavenumbers and octahedral Fe.

Fig. 9. Experimental (06–33) XRD reflections and their decompositions using one Gaussian curve.

S. Petit et al. / Applied Clay Science 104 (2015) 96–105 Table 5 Experimental (b) parameter (6 ∗ d (06–33) from XRD) of synthetic smectites. Fe3+oct: octahedral Fe3+ (Table 2); Fe3+tet: tetrahedral Fe3+ and Altotal: Al total (Table 2). All for O10 (OH)2. Sample

Fe3+oct

Fe3+tet

Altot

(b) (Å)

0 1 14 19 23 31 39 41 43 52 53 54

2 2 1.89 1.74 0.38 0.34 1.97 1.80 1.64 1.26 0.80 0.57

0.75 0.13 0

0 0.34 0.56 0.96

0.07

0.48 0.68 0.97

9.240 9.186 9.168 9.132 8.982 8.970 9.156 9.126 9.138 9.096 9.048 9.002

1.87 1.88

effect on b, Al has a negative effect, but for both cations, it is not possible to distinguish the effect of tetrahedral substitution from octahedral substitution.

4. Discussion 4.1. Chemistry of the synthesized smectites Dioctahedral smectites were synthesized for 2 b Fe3+oct b 0.34 from NIR data on the basis of the following structural formula [Si4 − x (Alt, Fe3+)x] (Al, Fe3+)2 O10 (OH)2 Na+x. A quite complete Al–Fe3+ octahedral solid-solution having compositions ranging between the endmembers beidellite and nontronite was then obtained. The (Fe/(Fe + Al))(total) chemical contents of the synthesized smectites were the same as their corresponding starting gel compositions (Fig. 1), probably because of the low solubility of Al and Fe. A part of Si from the starting gel was lost to solution during the synthesis process originating in a tetrahedral charge for the synthetic smectites. Smectites were not synthesized for Fe3+oct values between 1.26 and 0.8 (Table 2). This result is linked to the difficulty in adjusting the pHf value to an acceptable value according to the (Fe/(Fe + Al))(total) ratio (Andrieux and Petit, 2010). A good agreement was observed between the octahedral composition of synthetic Fe rich smectites obtained by FTIR and TEM-EDX data (Table 2). For the Al rich smectites discrepancies appeared between the two sets of data, due mainly to the difficulty to synthesize well crystallized Al rich smectites that are not mixed with residual starting gel. As such these TEM-EDX data are probably less valid than the FTIR data. y = 0.103x + 9.164 R² = 0.92

y = 0.111x + 8.944 R² = 0.98

103

The layer charge (mean value 0.42) of the synthetic Al–Fe smectites is common for natural and synthetic smectites: Al-beidellites synthesized at higher temperature (280–350 °C) by Kloprogge et al. (1990) and Lantenois et al. (2008), and most natural nontronite (Gates et al., 2002) exhibited a similar layer charge. In the Al–Fe3 + smectites synthesized here, the major part of the layer charge is due to Al for Si tetrahedral substitution (Table 2). Tetrahedral Fe3+, which was obtained indirectly from chemical and NIR data, is low for all samples and detectable amounts are observed only for high (Fe/(Fe + Al))(total). Since Fe3+tet as high as 0.75 (per half unit cell) was observed for synthetic ferric-nontronites (Decarreau et al., 2008), the low level of Fe3+ for Si substitution observed here for the Al–Fe3+ series has thus probably not a steric origin (Decarreau and Petit, 2014). 4.2. Crystallography of the synthesized smectites From (06, 33) XRD data, only one crystallographic lattice was necessary to describe all smectite particles for each synthetic sample, which therefore means that Al and Fe atoms are not located in different smectite layers, but instead randomly occupy the same layers. The (b) lattice parameter is thus closely correlated with the chemistry of smectites. Several correlations between (b) parameter of smectites and total Fe content have been published using the position of the maximum of the (06–33) reflection (Eggleton, 1977; Brigatti, 1983; Köster et al., 1999). Recently Heuser et al. (2013), using natural smectites and three samples of this paper, gave a new correlation from a Rietvelt approach of XRD patterns: ∘  3þ b A ¼ 0:117  Fe tot þ 8:998:

The correlation obtained from this study is very similar: ∘  3þ b A ¼ 0:111  Fe tot þ 8:944:

The main result is the validation of all these correlations between Fe and (b) parameter, where Fe3+ can either be in octahedral and tetrahedral sites. For this Al–Fe synthetic smectite series, Fe3+tet and Fe3+oct have a similar (and inseparable) effect on the (b) parameter value. Data also suggest (Fig. 10) that Al in octahedral and tetrahedral sites have a similar, but opposite influence on the (b) parameter. The correlation between (b) and Fe3 +tot given here is valid for dioctahedral Al–Fe smectites, and cannot be necessarily applied to other clays containing other major elements or having other structures. As an example, the (b) value calculated using the above correlation for a ferripyrophyllite with Fetot = 1.96 (Badaut et al, 1992) would be 9.16 Å instead of the 9.06 Å actually measured. From synthetic Al–Fe3+ kaolinites, Petit and Decarreau (1990) and Iriarte et al. (2005) obtained this correlation:

b parameter (Å)

∘  3þ b A ¼ 0:074 Fe oct þ 8:946:

Fe oct Fe tet Total Fe Al oct Total Al

y = -0.122x + 9.232 R² = 0.94 y = 0.114x + 8.940 R² = 0.98

y = -0.116x + 9.17 R² = 0.93

Number of Fe or Al atoms per half unit cell Fig. 10. Correlation between the crystallographic (b) parameter and Al and Fe contents of synthetic smectites.

The Al/Fe3+ substitution in the synthetic kaolinite structure had a lower effect than in synthetic smectite, but a similar effect than in pyrophyllite. The less rigid structure of smectite probably allows more possibilities of cell distortion. 4.3. Crystal chemistry The random distribution of Al and Fe3+ in the octahedral sheet occurs despite the discrepancy between the ionic radii of the two cations: 0.535 Å for Aloct and 0.645 Å for Fe3+oct (Shannon, 1976). This random octahedral distribution was observed here for a synthetic dioctahedral series with homovalent cations. For di-trioctahedral smectites and those with octahedral heterovalent cations, random distributions of octahedral cations are generally not observed (Grauby et al., 1993, 1994).

104

S. Petit et al. / Applied Clay Science 104 (2015) 96–105

Segregations of Al or Fe and Mg in clusters were indeed observed simultaneously with the occurrence of Al–Mg or Fe–Mg pairs in adjacent octahedra, for synthetic smectitic series obtained under similar synthesis conditions as for this study. From these (Grauby et al., 1993, 1994) earlier studies, it is clear that M3+–Mg substitution in the octahedral sheet (giving a montmorillonitic character to the smectite) results in a non-random distribution of octahedral cations. The great variability of octahedral cation occupancy observed in natural smectite structures (Brigatti, 1983; Manceau et al., 1998; Vantelon et al, 2001, 2003; Gates, 2005, 2008; Cashion et al., 2008; Marchel and Stanjek, 2012) is then also observed for synthetic samples. 4.4. Applications to other smectites Using the various correlations obtained between the chemical composition of the synthesized smectites and the XRD and IR data, it is possible to give crystallographic and spectroscopic parameters of three theoretical end members: (Si3.6 Al0.4) Al2 O10 (OH)2 Na0.4, (Si3.6 Al0.4) Fe3 +2 O10 (OH)2 Na0.4, and (Si3.6 Fe3 +0.4) Fe3 +2 O10 (OH)2 Na0.4 (Table 6). These data can be very useful to study the crystal-chemistry of other smectites. For example Kloprogge et al. (1990, 2000), Kloprogge (2006) synthesized aluminous beidellites having no octahedral charge at 350 °C and from 0.5 to 1 kb. While the synthesis conditions of beidellite are very different (higher pressure/temperatures) to those of Al–Fe smectites studied here, it is possible to use the present results to predict (b) parameter and IR band positions of the Al-beidellite synthesized by these authors (Table 6). Goodman et al. (1976) studied seven nontronites samples by Mössbauer and MIR spectroscopy. For the Washington and Garfield samples the Si–O band was located at 1021 and 1020 cm−1. According to data from Table 6, these two samples contain essentially no tetrahedral Fe, which is consistent with the structural formula of the Washington nontronite. According to Goodman et al. (1976) the Garfield nontronite contains 0.05 tetrahedral Fe (for O11 equivalence) and Gates et al. (2002), using multiple techniques, concluded that the amount of tetrahedral Fe in Garfield nontronite is between 0 and 0.05. Recently Gates (2005) showed that the Si–O band for 12 natural nontronites ranged from 1016 (for the Cheney Washington sample) to 1002 cm−1 (for the nontronite from Clausthal Zellerfeld, Germany) in agreement with the results reported here for the synthetic nontronite, where tetrahedral Fe content caused the red-shift of this band. The position of 1002 cm−1 was observed for 0.46 Fe3+tet (for O11 equivalence). This band shifted to 1007 cm−1 for NAu-2 having 0.145 Fe3+tet and to 1016 cm−1 for NAu-1 with 0.03 Fe3+tet. Deviations in the relationship between synthetic and natural samples are probably related to larger differences in total layer charge and to the presence of octahedral Mg for the natural nontronites. The MIR approach presented here yields results and interpretations in general agreement with Gates et al. (2002) conclusions. For all nontronites studied by Goodman et al. (1976), except the Washington sample, the Si–O–M3 + band was located from 487 to 490 cm−1. The equation given in Fig. 8 confirms the lack of octahedral

Table 6 Data for theoretical end-members of synthetic smectites. In italic, expected value. XRD

MIR (cm−1)

b (Å)

Si–O

M3+0ct–Oap

Si–O–M3+

Si–O–M3+

End member (Si3.6 Al0.4) Al2(Si3.6 Al0.4) Fe2(Si3.6 Fe0.4) Fe2-

8.944 9.165 9.211

1040 1019 1006

712 675 675

539 490 490

476 452 452

Synthetic beidellite (Si3.65 Al0.35) Al2

8.96a

1047a

709a

535b

473b

a b

Kloprogge et al. (1990). Kloprogge (2006).

Al in these nontronite samples. For the Washington sample this band is located at 497 cm− 1. Using the equation given in Fig. 8 for this band, the octahedral Fe/(Fe + Al) ratio is 0.84, in reasonable concordance with the value given by Goodman et al. (1976): 0.73, and by Gates et al. (2002), Gates (2005): 0.75. 5. Conclusion A series of Al–Fe smectites, free from the influence of octahedral Mg, was synthesized. The smectites display a random distribution of octahedral Al and Fe occupancies ranging from 0 to 1.66 Al and 0.34 to 2 Fe. Relations between chemical composition and position of the X-ray 06,33 reflection indicate that total Fe and Al control the b lattice dimension in opposite fashion, with increasing Fe content increasing b and increasing Al content decreasing b. Tetrahedral substitutions were dominated by Al except when total Fe contents exceeded 1.8 per f.u., and thus exhibited a partitioning in Al–Fe distribution for this synthetic series. The synthetic Al–Fe smectite series enabled unambiguous identification of the origins of several IR bands observed in both the NIR and MIR regions for smectites. A strong correlation between the Si–O stretching frequency and total Fe content was observed for the synthetic series, and because the positions compared favorably with those of natural smectites, it enables accurate calculations of this band for samples from other studies. References Andrieux, P., Petit, S., 2010. Hydrothermal synthesis of dioctahedral smectites: the Al–Fe chemical series. Part I: influence of experimental conditions. Appl. Clay Sci. 48, 5–17. Badaut, D., Decarreau, A., Besson, G., 1992. Ferripyrophyllite and related Fe3+-rich 2:1 clays in recent deposits of Atlantis II Deep, Read Sea. Clay Miner. 22, 227–244. Bishop, J.L., Murad, E., Madejová, J., Komadel, P., Wagner, U., Scheinost, A., 1999. Visble, Mössbauer and infrared spectroscopy of dioctahedral smectites: structural analyses of the Fe-bearing smectites Sampor, SWy-1 and SWa-1. 11th International Clay Conference, June, 1997, pp. 413–419. Brigatti, M., 1983. Relationships between composition and structure in Fe-rich smectites. Clay Miner. 18, 177–186. Cashion, J.D., Gates, W.P., Thomson, A., 2008. Mössbauer and IR analysis of iron sites in four ferruginous smectites. Clay Miner. 43, 1–11. Chryssikos, G.D., Gionis, V., Kacandes, G.H., Stathopoulou, E.T., Suarez, M., Garcia-Romero, E., Sanchez Del Rio, M., 2009. Octahedral cation distribution in palygorskyte. Am. Mineral. 94, 200–203. Craciun, C., 1984. Influence of the Fe3+ for Al3+ octahedral substitutions on the IR spectra of montmorillonite minerals. Spectrosc. Lett. 17, 579–590. Decarreau, A., Petit, S., 2014. Al–Fe3+ partitioning between tetrahedral and octahedral sites in dioctahedral smectites. Clay Miner. 49, 657–665. Decarreau, A., Petit, S., Martin, F., Farges, F., Vieillard, P., Joussein, E., 2008. Hydrothermal synthesis, between 75 and 150 °C, of high charge, ferric nontronites. Clays Clay Minerals 56, 322–337. Eggleton, R.A., 1977. Nontronite: chemistry and X-ray diffraction. Clay Miner. 12, 181–194. Farmer, V.C., 1974. The Layer Silicates. The I.R. Spectra of Minerals: Monograph 4, Mineralogical Soc. London. , pp. 331–363. Gates, W.P., 2005. Infrared spectroscopy and the chemistry of dioctahedral smectites. In: Kloprogge, J.T. (Ed.), The Application of Vibrational Spectroscopy to Clay Minerals and Layered Double Hydroxides: CMS Workshop Lectures, Aurora, Colorado, pp. 125–168. Gates, W.P., 2008. Cation mass–valence sum (CM-VS) approach to assigning OH bending bands in dioctahedral smectites. Clays Clay Minerals 51, 10–22. Gates, W.P., Slade, P.G., Manceau, A., Lanson, B., 2002. Site occupancies by iron in nontronites. Clays Clay Minerals 50, 223–239. Gaudin, A., Petit, S., Rose, J., Martin, F., Decarreau, A., Noack, Y., Borscheneck, D., 2004. The accurate crystal chemistry of ferric smectites from the lateritic nickel ore of Murrin Murrin (Western Australia). II. Spectroscopic (IR and EXAFS) approaches. Clay Miner. 39, 453–467. Gionis, V., Kacandes, G.H., Kastritis, I.D., Chryssicos, G.D., 2007. Combined near-infrared and X-ray diffraction investigation of the octahedral sheet of palygorskyte. Clays Clay Minerals 55, 543–553. Goodman, B.A., Russell, J.D., Fraser, A.D., Woodhams, F.W.D., 1976. A Mössbauer and IR spectroscopic study of the structure of nontronite. Clays Clay Minerals 24, 53–59. Grauby, O., Petit, S., Decarreau, A., Baronnet, A., 1993. The beidellite-saponite series: an experimental approach. Eur. J. Mineral. 5, 623–635. Grauby, O., Petit, S., Decarreau, A., Baronnet, A., 1994. The nontronite-saponite series: an experimental approach. Eur. J. Mineral. 6, 99–112. Heuser, M., Andrieux, P., Petit, S., Stanjek, H., 2013. Iron-bearing smectites: a revised relationship between structural Fe, cell edge length b and refractive indices. Clay Miner. 48, 97–103. Iriarte, I., Petit, S., Huertas, F.J., Fiore, S., Grauby, O., Decarreau, A., Linares, J., 2005. Synthesis of kaolinite with high level of Fe3+ for Al substitution. Clays Clay Minerals 53, 1–10.

S. Petit et al. / Applied Clay Science 104 (2015) 96–105 Keeling, J.L., Raven, M.D., Gates, W.P., 2000. Geology and preliminary characterization of two hydrothermal nontronites from weathered metamorphic rocks at the Uley graphite mine, South Australia. Clays Clay Minerals 48, 537–548. Kloprogge, J.T., 2006. Spectroscopic studies of synthetic and natural beidellites: a review. Appl. Clay Sci. 31, 165–179. Kloprogge, J.T., Jansen, J., Ben, H., Gens, J.W., 1990. Characterization of synthetic Nabeidellite. Clays Clay Minerals 38, 409–414. Kloprogge, J.T., Komarneni, S., Amonette, J.E., 1999. Synthesis of clay minerals: a critical review. Clays Clay Minerals 47, 529–554. Kloprogge, J.T., Hickey, L., Frost, R.L., 2000. The effect of increasing layer charge on the infrared absorption spectra of synthetic beidellites. J. Mater. Sci. Lett. 19, 1131–1134. Komadel, P., Madejovà, J., Stucki, J.W., 1995. Reduction and reoxidation of nontronite: question of reversibility. Clays Clay Minerals 39, 467–477. Köster, H.M., Ehrlicher, U., Gilg, H.A., Jordan, R., Murad, E., Onnick, K., 1999. Mineralogical and chemical characteristics of five nontronites and Fe-rich smectites. Clay Miner. 34, 579–599. Lantenois, S., Muller, F., Bény, J.-M., Mahiaoui, J., Champallier, R., 2008. Hydrothermal synthesis of beidellites: characterization and study of the cis- and trans-vacant character. Clays Clay Minerals 56, 39–48. Madejová, J., Komadel, P., Cicel, B., 1994. Infrared study of octahedral site population in smectites. Clay Miner. 29, 319–326. Manceau, A., Chateigner, D., Gates, W.P., 1998. Polarized EXAFS, distance-valence leastsquares modeling (DLVS), and quantitative texture analysis approaches to the structural refinement of Garfield nontronite. Phys. Chem. Miner. 25, 347–365. Marchel, C., Stanjek, H., 2012. Cation ordering in cis- and trans-vacant dioctahedral smectites and its implications for growth mechanisms. Clay Miner. 47, 105–115. Neumann, A., Petit, S., Hofstetter, Th., 2011. Evaluation of redox-active iron sites in smectites using middle and near infrared spectroscopy. Geochim. Cosmochim. Acta 75, 2336–2355. Petit, S., 2005. Crystal-chemistry of talcs: a NIR and MIR spectroscopic approach. In: Theo Kloprogge, J. (Ed.), The Application of Vibrational Spectroscopy to Clay Minerals and Layered Double Hydroxides, CMS Workshop Lectures vol. 13. The Clay Mineral Society, Aurora, CO, pp. 41–64 (ISBN: 978-1-881208-14-1).

105

Petit, S., Decarreau, A., 1990. Hydrothermal synthesis and crystal chemistry of iron-rich kaolinites. Clay Miner. 25, 181–196. Petit, S., Madejová, J., 2013. Fourier transform infrared spectroscopy, chapter 2.7. In: Bergaya, F., Lagaly, G. (Eds.), Handbook of Clay Science second edition. Techniques and Applications vol. 5B. ISBN: 978-0-08-098258-8, pp. 213–231. Petit, S., Robert, J.L., Decarreau, A., Besson, G., Grauby, O., Martin, F., 1995. Contribution of spectroscopic methods to 2:1 clay characterization. Bull. Centres Rech. Explor. Prod. Elf-Aquitaine 19 (1), 119–147. Petit, S., Caillaud, J., Righi, D., Madejová, J., Elsass, F., Köster, H.M., 2002. Characterization and crystal chemistry of a Fe-rich montmorillonite from Ölberg, Germany. Clay Miner. 37, 283–297. Petit, S., Martin, F., Wiewiora, A., De Parseval, P., Decarreau, A., 2004a. Crystal-chemistry of talc: a near infrared (NIR) spectroscopy study. Am. Mineral. 89, 319–326. Petit, S., Decarreau, A., Martin, F., Buchet, R., 2004b. Refined relationship between wavenumbers of ν OH and their overtones for clay minerals. Phys. Chem. Miner. 31, 585–592. Post, J.L., Borer, L., 2002. Physical properties of selected illites beidellites and mixed-layer illite-beidellites from southwestern Idaho, and their infrared spectra. Appl. Clay Sci. 22, 77–91. Post, J.L., Noble, P.N., 1993. The near-infrared combination band frequencies of dioctahedral smectites, micas and illites. Clays Clay Minerals 41, 639–644. Shannon, R.D., 1976. Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallogr. A32, 751–767. Vantelon, D., Pelletier, M., Michot, L.J., Barres, O., Thomas, F., 2001. Fe, Mg and Al distribution in the octahedral sheet of montmorillonites. An infrared study in the OHbending region. Clay Miner. 36, 369–379. Vantelon, D., Montarges-Pelletier, E., Michot, L.J., Briois, V., Pelletier, M., Thomas, F., 2003. Iron distribution in the octahedral sheet of dioctahedral smectites. An Fe K-edge Xray absorption spectroscopy study. Phys. Chem. Miner. 30, 44–53.