RELATIONSHIP AMONG DERIVATIVE SPECTROSCOPY, COLOR, CRYSTALLITE DIMENSIONS, AND A1 SUBSTITUTION OF SYNTHETIC GOETHITES AND HEMATITES 1

Transcription

1 Clays and Clay Minerals, Vol. 34, No. 6, , RELATONSHP AMONG DERVATVE SPECTROSCOPY, COLOR, CRYSTALLTE DMENSONS, AND A1 SUBSTTUTON OF SYNTHETC GOETHTES AND HEMATTES 1 C. S. KOSMAS, 2 D. P. FRANZMEER, AND D. G. SCHULZE Agronomy Department, Purdue University West Lafayette, ndiana Abstract--Nine hematites and 22 goethites were synthesized by a variety of methods to obtain monomineralic samples having a range of A1 substitutions and particle sizes. The second derivative ofabsorbance and Munsell color designations were calculated from visible reflectance spectra obtained from the dry powders. Unit-cell dimensions, A1 substitution, infrared band positions, mean crystallite dimensions (MCD) from X-ray powder diffraction, and particle size from fiber-optic Doppler anemometry (FODA) were determined. Previously reported correlations between A1 substitution, goethite unit-cell dimensions, and OH-stretching and -bending band positions were confirmed. For hematite, the position of the second derivative peak at ~600 nm was negatively correlated with A1 substitution (r = -.86). Munsell value and chroma were positively correlated with A1 substitution (r =.94 for both), but hue was not related to A1 substitution. Hue appeared to become redder, however, as particle size measured either by FODA or MCD increased. For goethite, the position of the second derivative minimum at ~ 485 nm was negatively correlated with A1 substitution (r = -.99). Munsell hue appeared to be related to both A1 substitution and MCD perpendicular to (110), MCD11o, with hues becoming redder with increasing A1 substitution and yellower with increasing MCD 11o. Correlations between Munsell value and chroma and parameters such as A1 substitution, particle size, and OH-stretching and -bending band positions were poor, but goethites synthesized by oxidation of Fe 2 solutions at room temperature had higher chromas than goethites synthesized hydrothermally from an Fe 3 system. Visually determined colors agreed well with calculated ones. Second-derivative spectra and color designations calculated from visible spectra appear to be potentially useful for quickly estimating other properties of goethite and hematite, such as A1 substitution and particle size. Key Words--Aluminum, Color, Fiber-optic Doppler anemometry, Goethite, Hematite, Visible spectroscopy. NTRODUCTON Soil scientists have long used color to judge the moisture and aeration condition of soils and to help explain how soils formed. The relected visible spectra of soil constituents, especially iron oxide minerals, determines what we perceive as color. Soil color is usually described by comparing it with standard color charts. The reflected visible spectrum, recorded with a spectrophotometer, can also be analyzed in detail to yield specific information about the crystallochemical properties of the minerals. From the results of such an analysis, certain mineralogical information can be predicted from color observations made in the field. Recently, Kosmas et al. (1984) demonstrated that the second derivative of the reflected visible spectrum can be used to identify some of the major iron oxide minerals. Their work suggested that additional information about some crystallochemical properties of the iron oxide minerals could be obtained from an analysis of the second derivative of the spectrum. Therefore, this work was undertaken: (1) to determine how various crystallochemical properties affect the position and amplitude of second derivative peaks of reflected visible spectra of synthetic goethites and hematites; (2) to determine how these properties relate to each other and to Munsell hue, value, and chroma calculated from spectral data; and (3) to compare Munsell colors read from color charts with those calculated from spectral data. MATERALS AND METHODS Hematites and goethites were synthesized under the various conditions described below to produce different crystal sizes and amounts of A1 substitutions. After the syntheses were complete, all samples were washed free of salts either by repeated centrifuge washings or by dialysis against deionized water. They were then dried at 60~ and gently crushed. 1 Journal Paper No. 10, Present address: Athens Faculty of Agriculture, Laboratory of Soils and Agricultural Chemistry, Botanicos , Athens, Greece. Copyright , The Clay Minerals Society 625 Hematite synthesis Sample H1-1. Hematite was produced by heating synthetic maghemite at 650~ for 3 hr (Sidhu et al., 1980). Sample H1-2. Hematite was produced by heating syn-

2 626 Kosmas, Franzmeier, and Schulze Clays and Clay Minerals Table 1. Quantities of Fe and A1 used for the synthesis of synthetic goethites. Coprecipitated Additional Fe(NO3)3 A(NO3)3 A(NO3)3 Sample~ (mmole) (mmole) (mmole) G G G G G G G See text for preparation procedures. thetic goethite at 320~ for 3 hr (Rend6n and Serna, 1981). Sample H2-1. Ferrihydrite was precipitated by adding 95 ml of 3.7 M NH4OH to 0.1 mole of Fe(NO3)3 in 800 ml of deionized water, after which the volume was brought to 1000 ml with water. The precipitate was centrifuged and washed three times with deionized water. The suspension containing the precipitate was made to 500-ml volume with deionized water, 0.43 g Of (NH4)2C204"H20 was added to make the solution 5 x 10-3 M, and the ph was adjusted to 7.0 with dilute HNO3. The suspension was then placed in an oven at 70~ for 15 days. t was shaken and the ph was readjusted to 7.0 once each day (Fischer and Schwertmann, 1975). Samples H3-1 to H3-5. Ferr/hydrite was precipitated by the addition of 11 ml of 3.7 M NH4OH to 15 rumple of Fe + A as Fe(NO3)3 and Al(NO3)3 in 1000 ml of water. A1/(A1 + Fe) molar ratios were: 0 (H3-1), 0.02 (H3-2), 0.06 (H3-3), 0.10 (H3-4), and 0.18 (H3-5). The precipitates were washed three times with deionized water, suspended in 1000 ml of water, and adjusted to ph 7.0 with NH4OH or HNO3. They were shaken and the ph was readjusted to 7.0 once each day. Samples to H4-3. Five hundred milliliters of 0.05 M Al(NO3)3 was slowly poured into 300 ml of 5 M KOH under constant stirring. Either 150 (H4-1), 180 (H4-2), or 210 ml (H4-3) of this solution was poured into a 2-liter stainless steel beaker and enough additional 5 M KOH added to give an OH- concentration of 0.3 M after precipitation of the A1 + Fe and dilution to 1000 ml final volume. Fifty milliliters of 1 M FefNO3)3 was then quickly poured into the beaker. The beaker was swirled to mix the contents thoroughly, and deionized water was added to make 1000 ml. The suspension was then placed in an autoclave at 120~ for 12 hr. Only hematite formed at these levels of A1 additions. Goethite synthesis Samples G-1 to G1-4. Five hundred milliliters of 0.05 M Al(NO3)3 was slowly poured into 300 ml of 5 M KOH under constant stirring. Either 0 (G 1-1), 15 (G 1-2), 40 (G1-3), or 60 ml (G1-4) of this solution was then poured into a 1000-ml polypropylene bottle and enough additional 5 M KOH added to yield an OH- concentration of 0.3 M after precipitation of the Fe + A1 and dilution to 1000 ml final volume. Fifty milliliters of 1 M Fe(NO3)3 was then quickly poured into the bottle, the bottle was swirled to mix the contents thoroughly, and deionized water was added to make 1000 ml. The bottle was capped, shaken, and placed in an oven at 70~ for 14 days and shaken once a day (Lewis and Schwertmann, 1979). Samples G2-1 to G2-4. Ferrihydrite was precipitated by adjusting 0.05 moles of Fe(NO3)3 in 900 ml ofdeionized water to ph 11.5 with 2.5 M NaOH. The resultant suspension was kept at 70~ for 1 week (G2-1), 45~ for 4 weeks (G2-2), 22~ for 6 weeks (G2-3), and 3~ for 22 weeks (G2-4). Samples G3-1 to G3-4. The following quantities of A13 and Fe 2+ as A1C13 and FeC12.4 H20 were dissolved in ~ 150 ml ofdeionized water: G3-1, 0 mmole A mmole Fe; G3-2, 1.5 mmole A mmole Fe; G3-3, 14.5 mmole A mmole Fe; G3-4, 21.8 mmole A mmole Fe. NaHCO 3 (0.05 M) was added to bring the ph to 7.0, and the final volume was brought to 1000 ml. The suspension was then oxidized by slowly bubbling air through it at the rate of ~4 bubbles/rain for 8 weeks at room temperature (Goodman and Lewis, 1981). The samples were shaken once each day. The ph changed from 7 to 8.5 during the reaction. Samples G4-1 to G4-7. The quantities of Fe(NO3)3 and Al(NO3)3 listed in Table 1 under "coprecipitated" were dissolved in 500 ml of deionized water. This solution was then poured into 2000 ml of 0.57 M NH4OH with vigorous stirring, and the resulting suspension was allowed to stand at room temperature for 24 hr. The precipitate was washed on a Buchner funnel with deionized water. The additional quantities of Al(NO3)3 listed in Table 1 were dissolved in 300 ml of 2 N KOH, and the washed precipitate was added to the solution. These suspensions were then placed into stainless steel beakers and autoclaved for 12 hr at 120~ (Thiel, 1963). Samples G5-1, G5-2, and G5-3. The procedure for these samples was the same as that for samples G4-2, G4-5, and G4-7, respectively, except that the suspensions were heated in a high-pressure reactor vessel for 12 hr at 155~ Chemical analysis Oxalate-extractable Fe (Feo) was determined by shaking the sample with 0.2 M ammonium oxalate at ph 3 for 4 hr in the dark (Schwertmann, 1964). Total Fe (FeO and A1 (Aid were determined using a Varian

3 Vol. 34, No. 6, 1986 Properties of synthetic goethite and hematite vs. A1 content 627 Table 2. A1 substitution, ratio of oxalate-extractable to total Fe, unit-cell dimensions, mean crystallite dimension perpendicular to (104) (MCD~o4), and particle size of the synthetic hematites. Unit-cell dimensions A Particle size substitution MCD~04 FODA 4 Sample (mole %) FeJFe? (~) (~) (A) (A) H-1 (PHFMH) H 1-2 (PHFG) H2-1 (PSH) H3-1 (HA18) H3-2 (HA14) H3-3 (HA15) H3-4 (HA16) H3-5 (HA17) H4-1 (HA26) H4-2 (HA27) H4-3 (HA28) See text for preparation procedures. z dentification of samples in Kosmas (1984). 3 Ratio at oxalate soluble Fe (Feo) to total Fe (FeO. Fiber optic Doppler anemometry. Techtron atomic absorption spectrophotometer after dissolving the sample in a mixture of aqua regia and HF (Bernas, 1968). Visible spectroscopy Visible spectra were measured using a Cary 17DX spectrophotometer equipped with a Model 1711 diffuse-reflectance accessory containing a ring collector. Powder mounts were prepared by passing the sample through a 50-urn sieve, backfilling the sample into a hole (27 mm in diameter) in an Al disk (3-mm thick), then gently pressing the powder against unglazed paper. Reflectance measurements were made relative to a BaSO4 standard from 770 to 380 nm in 1-nm steps. The BaSO4 standard was prepared by spraying Kodak white reflectance coating (Eastman Kodak Corporation, Rochester, New York) onto a sandblasted A1 plate to give a coating 1-2-mm thick. Munsell color designations were calculated from the reflectance curve as follows: The CE tristimulus values were first calculated using the color matching functions of the CE 1931 standard colorimetric system and CE standard illuminant C (Wyszecki and Stiles, 1982). A computer program was then used to convert the CE tristimulus values into MunseU notations? Hues are expressed in both the familiar numeral-letter format (e.g., 10YR) 3 The program to convert CE to MunseU notations was obtained from Fred Billmeyer, Department of Chemistry, Rensselaer Polytechnic nstitute, Troy, New York This conversion can also be made using charts as illustrated in Wyszecki and Stiles (1982). Enlarged versions of these charts are available from Munsell Color, 2441 North Calvert St., Baltimore, Maryland Accurate interpolations of hue and chroma from the charts, however, is difficult. The computer program makes this interpolation more accurately and quickly. and in a numeric format in which the number 5 = 5R, 10 = 10R, 15 = 5YR, 20 = 10YR, etc. The second derivative of the smoothed spectral data was calculated using a numerical method (Savitzky and Golay, 1964; Steinier et al, 1972), as described by Kosmas et al. (1984). Visual color determination The color of the dry powders was determined by comparison with the Munsell Soil Color Charts under sunlight. Determinations were made by two experienced observers. X-ray powder diffraction Random powder mounts for X-ray diffraction (XRD) were prepared by mixing nine parts of sample with one part (by weight) of 1-#m corundum (Buehler Micropolish Linde C 1.0 um o~-a1203, No ) as an internal standard (Bryant et al, 1983). XRD patterns were obtained using CuKa radiation and a Siemens Type F diffractometer equipped with a graphite difffracted-beam monochromator, a 1 ~ divergence slit, and a 0.25-mm receiving slit. The patterns were stepscanned at 0.02~ increments using a 10-s counting time. Unit-cell dimensions ofgoethite were calculated using the 110, 130, and 111 XRD reflections because these three lines were accurately measured on all patterns, even if the lines were broad. Unit-cell dimensions of hematite were calculated using the 012, 104, 110, 113, 116, 214, and 300 reflections. The widths at half-height (WHH) of selected XRD reflections were corrected for instrumental broadening by subtracting the WHH derived from coarse quartz. The mean erystallite dimensions were calculated from the corrected

4 628 Kosmas, Franzmeier, and Schulze Clays and Clay Minerals Sampie ~ Table 3. nfrared and visible spectral properties and eolorimetrie properties of synthetic hematites. nfrared Visible Calculated Munsell parameters 2nd deriv. Peak Peak Peak peak position Munsell color (cm ~) (cm-~) (cm ~) (nm) Hue number Hue Value Chroma from charts H R R 3/6 H YR YR 3/6 H YR YR 3/4 H YR YR 3/4 H YR YR 3/4 H YR YR 3/4 H YR YR 4/8 H YR YR 5/8 H R R 3/6 H R R 4/8 H R R 3/6 See text for preparation procedures. WHH using the Scherrer formula (Klug and Alexander, 1974). nfrared absorption nfrared absorption (R) spectra were obtained using a Perkin Elmer Model 180 grating R spectrophotometer and KBr disks (1 mg for goethite, 0.5 mg for hematite per 300 mg KBr) and a scan rate of 50 cm-v min. The band positions were measured graphically after expanding each absorption band of interest. Fiber-optic Doppler anemometry Particle size was measured using fiber-optic Doppler anemometry (FODA) (Ross et al, 1978; Wu, 1983) after the sample had been dispersed in water with ZONYL FSN fluorosurfactant as a dispersinng agent (Berkheiser and Monsees, 1982). The FODA technique measures the mean particle size of a sample in a relatively concentrated suspension (as high as ~ 1 g/liter for goethite and hematite). The instrument measures Brownian motion by measuring the Doppler shift of laser light scattered by the suspended particles. A computer program then relates the light scattering to particle size. Hematite RESULTS AND DSCUSSON The parameters measured for hematite are listed in Tables 2 and 3; the correlations between pairs of hematite parameters are summarized in Table 4. Samples H-1 and H1-2 are not included in this analysis because data from these two samples did not always agree with data from the other 9 samples. The data from these two samples are, however, plotted in Figure 3 for comparison. This omission is justified because samples H-1 and H-2 were synthesized by heating maghemite or goethite powders to convert them to hematite, whereas the other hematite samples were synthesized in aqueous systems. Reflectance spectra (Figure 1) of hematite samples H4-3 and H3-2 show almost complete absorption between 360 and 550 nm and an absorption band centered at about 660 nm. The greater overall reflectance of sample H4-3 vs. sample H3-2 is shown in the higher Munsell value and chroma (Figure 1, Table 3). Spectral curves can be enhanced by calculating the second derivative of the absorbance (Kosmas et al., 1984). n Figure 2, the upper graph shows the absor- Table 4. Correlation coefficients for the hematite samples (N = 9). Property Property Mole % A1 2. a dimension -.999** 3. c dimension -.75**.73* 4. MCD Particle size.71" -.69* -.65*.68* 6. R peak **.88**.90** R peak *.74*,38.63* R peak **.77**.47.58* nd der. pos. -.86**.85**.71" Hue number -.21, ** -.79** 11. Value.94** -.93** -.81"* Chroma.94** -.94** -.80**.06.66* * Significant at 95% level. ** Significant at 99% level..72*.79**.98**.83**.67*.69* ** -.78** -.78"* -.90** -.90** -.70* -.69* -.89** **

5 Vol. 34, No. 6, 1986 Properties of synthetic goethite and hematite vs. A1 content ' i 0.4-7_ t o 35O G3-1 8.SYR 6.1/9.4 46o 46o 56o 5;0 66o 0;0 76o 7;0 o6o Wavelength (nra) Y 6.3/7.3 H4-3 om 7.8R 3.3/7.2 ~ 0.3- G5-3 '~ L6YR 4.0/7.5,~m $ 2.5/4.6 Figure 1. Reflected visible spectra of synthetic hematite (H3-2 and H4-3) and goethite (G3-1, Gl-1, and G5-3) samples. Reflectance is a dimensionless quantity ~efl.. ~ --H4-t, 6.3Z A H3-5, t2.5~ A1 0 i i i i 6- x. bance and reflectance curves for two hematite samples, and the lower graph shows the second derivatives of the absorbance curves. The subtle maxima and minima in the reflectance and absorbance curves are represented by sharper second derivative peaks. For hematite, the second derivative maximum at about 600 nm (P in Figure 2) and the minimum at about 550 nm shifted to lower wavelengths with increasing A1 substitution. The relationship of the 600-nm peak and A1 substitution is plotted in Figure 3e; it has a correlation coefficient of (Table 4). The hematite properties examined can be divided into two groups: properties that varied with A1 substitution and properties that varied with crystallite or particle size. Properties that varied with A substitution at >99% confidence level include the a and c dimensions of the unit cell, the positions of the R bands at about 550 and 335 cm 1, Munsell value and chroma, and the position of the second-derivative visible spectroscopy peak at about 600 nm (Table 4). Particle size from FODA and the position of the R band at about 470 cm -1 correlated with A1 substitution at the 95% confidence level. Properties which correlated with A1 substitution also correlate with each other. The a dimension decreased linearly with increasing A1 substitution (Figure 3a). The relationship is not significantly different from that found by Schwertmann et al. (1979). The c dimension decreased from A at 0 mole % A1 substitution to about 13.73/~ at 3.8 mole % A1, but was essentially constant at about from 5 to 13 mole % A (Figure 3c). More scatter was observed in the c dimension values than in the a dimension values. The Munsell hue did not correlate with A1 substitution, but it did correlate with mean crystallite dimension perpendicular to the (104) plane (MCDo4) and to particle size (Table 4). Samples with MCD104 <1000 /k appeared to be less red than those with MCDl04 values > 1000 A (Figure 3b). Whether the relationship is linear could not be determined, because data points were not scattered evenly around a straight line, but were grouped into two clusters (Figure 3b). '2_ aso 4;0 4;0 56o 5;0 600 s; ;0 a6o Wavelength (nm) Figure 2. Top: Visible absorbance and reflectance of two synthetic hematite samples containing different amounts of A1 substitution (mole %). Bottom: Second derivatives of absorbance for those samples showing the position (P) of peak near 600 nm. Reflectance and absorbance are dimensionless quantities. Barron and Torrent (1984) found no relationship between hue and mean crystallite dimensions as estimated from the 110 reflection. They did not, however, report data for the 104 reflection. The width of the 104 reflection depends mainly on plate thickness and to a limited extent on plate width, whereas the width of the 110 reflection depends only on plate width (Schwertmann et al, 1979). Because the present data suggest a relationship between hue and MCD04, in contrast to Barron and Torrent (1984) who found no relationship between hue and MCD, lo, hue may be more a function of plate thickness than plate width. The hues of hematites H-1 and H1-2 formed by heating did not, however, appear to fall on a line defined by the other 9 hematites, suggesting that MCDlo4 was not the only factor influencing their hue. Munsell value increased linearly with A1 substitution (Table 4, Figure 3d), in agreement with data presented by Barron and Torrent (1984). The present data, however, were about 0.5 value units less than theirs, but the slopes of the regression lines were similar. Chroma increased linearly with increasing A1 substitution (r =.94; Table 4; Figure 3f). Barron and Torrent (1984) found no relationship between these two parameters. The large number of samples (68) that they studied possibly encompassed more variation than the smaller number (9) of samples examined in this study. Neither Munsell value nor chroma showed a systematic variation with MCD~04.

6 630 Kosmas, Franzmeier, and Schulze Clays and Clay Minerals Table 5. A1 substitution, ratio of oxalate-extractable to total Fe, unit-cell dimensions, crystallite size, and particle size of synthetic goethites. Unit-cell dimensions Crystallite size Particle A subst. MCD02o MCD,0 MCD02 ~ size FODA 4 Sample t (mole%) Feo/Fet 3 (2) (~) (~) (/k) (/~) (~) (~) G-1 (PSG) G1-2 (GA18) G1-3 (GA19) G1-4 (GA20) G2-1 (PG70D) G2-2 (PG45D) G2-3 (PG22D) G2-4 (PG3D) G3-1 (PGLPH) nd 3300 G3-2 (GA2) nd 3400 G3-3 (GA8) nd 3610 G3-4 (GA9) nd 3200 G4-1 (GA29) G4-2 (GA30) G4-3 (GA31) G4-4 (GA32) G4-5 (GA33) G4-6 (GA34) G4-7 (GA35) G5-1 (GA38) G5-2 (GA40) G5-3 (GA42) nd = not determined. See text for preparation procedures. 2 dentification of samples in Kosmas (1984). 3 Ratio of oxalate-extractable Fe (Feo) to total Fe (Fet). 4 Fiber-optic Doppler anemometry. Table 6. nfrared and visible spectral properties and colorimetric parameters of synthetic goethites. nfrared peak position Visible, 2nd derivative Calculated Munsell parameters Sam- e-oh ~i-oh 3'-OH Position Amplitude pie t (cm ~) (cm -t) (cm ~) (nm) (x 10 4) Hue number Hue Value Chroma G Y G Y G Y G nd nd nd nd G YR G YR G YR G YR G YR G YR G YR G YR G Y G YR G YR G YR G YR G YR G YR G YR G YR G YR Munsell color from chart 10YR 7/8 10YR 7/8 10YR 7/8 10YR 5/8 7.5YR 6/8 7.5YR 6/8 10YR 5/6 10YR 5/8 7.5YR 5/8 7.5YR 5/8 7.5YR 5/8 7.5YR 6/8 7.5YR 5/6 nd = not determined. See text for preparation procedures.

7 Vol. 34, No. 6, 1986 Properties of synthetic goethite and hematite vs. A1 content 631 5"051 a t ~ -5.o3 : t. E== m 5.02 ~ 7.5- t ].!.,,, t0 15 A (mole %) ~ C 9 0 ' ' MCD 104 (1) d , o U t3.74, i 9 o ' i ,,!,,,, A (mole %) 605- e!! A1 (mole ) 15 f ~ 600- E m A Q r ~- 6" ~ 595- "13 t- ru 590-.,.,., 4" A1 (mole ~) m, 0 ' ' ~ 0 5 t0 Al (mole %) 15 x H2-1; o H3-1 to H3-5; & H4-1 to H4-3; v Ht-t; o Ht-2 Figure 3. samples. Relationships of several parameters to A1 substitution or mean crystallite dimension (MCDo4) for the hematite Goethite Characterization data for the synthetic goethite sampies are reported in Tables 5 and 6; correlations between pairs of these parameters are listed in Table 7. Absorbance and the second derivative of absorbance are plotted in Figure 4. For goethite, the 447-nm maximum and the 423-nm minimum in the second derivative curves are characteristic of this mineral and do not change with A1 substitution (Kosmas et al, 1984). The position of the minimum at about 485 nm, however, indicated by arrows in Figure 4, shifted to smaller wavelengths with increasing A1 substitution (Figure 5d). The calculated regression line is: deriv, pos. = (mole % A1), where n = 22 and r = The amplitude of this curve (A in Figure 4), also increased with increasing A1 substitution (r =.91; Table 7). Because of the better correlation and because the position was influenced less by the spectra of other minerals than the amplitude, peak position is the preferred parameter. For goethite, the second derivative maximum centered at about 570

8 632 Kosmas, Franzmeier, and Schulze Clays and Clay Minerals "~ - G4-6, t8.9~ A1.,.\\1 ~ 0,, r ~q 8- A 9 L a 1- T = *t*t*:~** ;0 5oo' 589 6oo' 6;0 ' Wavelength (nm) 70O E 0 0 r: ** ~ ~ ~ ~ 4 ~ ~ ~ * N N Figure 4. Top: Absorbance of a pure goethite and one containing 18.9 mole % A1 substitution. Bottom: Second derivatives of absorbance for these samples showing the positions (arrows) and amplitude (A) of the peak at ~490 nm, Absorbance is a dimensionless quantity. nm shifted to higher wavelengths with increasing A1- substitution (Figure 4), the opposite direction of the shift for the hematite peak at about 600 nm. Many other properties were also highly correlated with A substitution, including: a, b, and c unit-cell dimensions; MCD02o; 6-OH and ~-OH R peaks; and Munsell hue and value. Previous work has shown that increasing ionic substitution of A1 for Fe in the goethite structure results in smaller unit-cell dimensions (Thiel, 1963; J6nfis and Solymfir, 1970; Schulze, 1984) and shifts in the OH-stretching and OH-bending band positions to higher wavenumbers (Schulze and Schwertmann, 1984). The present data are in agreement with these results. Hue decreased with increasing A1 substitution (Figure 5a), but A1 substitution was not the only factor that influenced hue. Multiple linear regression analysis showed that both A1 substitution and crystallite size (as measured by MCD1,0) were important factors. The final regression equation was:

9 V01. 34, No. 6, 1986 Properties of synthetic goethite and hematite vs. A1 content : A L b == 20 JE: 17.5" d] 6,t 3] 0 i v x x E o 8- " ::4 * ~ " 7- v v,a (>?vv, 8-9 ' 'o )15 A1 (mole x x C "~ 500- A a A m v O R 490- v ~ 480- V "6 0 v v ' ' ' ' A (mole ~) "0 r- cu 470- ' ' ' ' A (mole %) C ' ' ' ' 0 5 t A (mole %) Figure 5. x Gi o G2 4 G3 vg4 og5 Relationships of several parameters with A1 substitution for goetbite samples. HUE = (mole % A1) (MCD~]0) , where HUE is the hue number. All regression coefficients were significant at the 99.9% level. The negative coefficient for A1 substitution means that goethite hues became redder as A1 substitution increased, whereas the positive coefficient for MCDllo indicates that goethite hues became yellower as MCD 1o became larger. Similar relationships were found for MCDo2o. Munsell value decreased with increasing A1 substitution (Figure 5c), but the correlation was poor. Multiple linear regression indicated that A1 substitution was the main factor influencing value. For all goethite samples synthesized from Fe 3+ solutions (excluding the G3 samples synthesized from Fe2+), Munsell value increased with increasing crystallite size, as measured by MCDo:o or MCDo2~, with a high correlation coefficient (R 2 >.90). f series G3 samples were included in the statistical analysis, the correlation coefficient decreased considerably, because the series G3 samples had higher Munsell values than predicted from the relationship for the other samples (Figure 5c). MCDo21 could not be measured for series 3 samples. Representative visible spectra (Figure 1) show that the spectra for the goethites from the Fe 3+ and Fe 2+ systems were qualitatively different. The spectra for the goethites from the Fe 3+ system, illustrated by samples G - 1 and G5-3, showed absorption bands at about 500 and 660 nm and well-defined reflectance maxima at about 580 nm. n contrast, the 500- and 660-nm absorption bands of the goethites from the Fe 2+ system (sample G3-1) were broader, and the reflectance maximum at 580 nm was not as well defined. This difference is apparently reflected in the much higher chromas ( ) of the goethites from the Fe 2 system (G3 samples) compared with the chromas ( ) ofgoethites from the Fe 3+ system (other than G3 samples, Table 6, Figure 5b). The higher overall reflectance of samples G3-1 and G 1-1 produced the higher Munsell values, 6.1 and 6.3 (Table 6), as opposed to the lower Munsell value (4.0) for sample G5-3. Comparison of calculated and visually determined colors n general, the colors calculated from the spectral data agreed well with the colors determined visually from the Munsell soil color chart (Tables 3 and 6). For

10 634 Kosmas, Franzmeier, and Schulze Clays and Clay Minerals most samples, calculated hues corresponded to the nearest hues in the Munsell soil color chart. Values generally agreed within one unit. Chromas also appeared to agree well, considering that the Munsell soil color charts have a limited number of chips (only chroma 4, 6, and 8) available for comparing with mineral samples. SUMMARY AND CONCLUSONS The position and amplitude of the second derivative of the reflected visible spectrum peak at about 485 nm strongly correlated with A substitution in synthetic goethite; the position of the peak at about 600 nm correlated with A1 substitution in hematite. Moreover, several unit-cell dimensions and R peak positions correlated with A1 substitution, as previously reported. For hematite, hue was related to mean crystallite dimensions perpendicular to (104), but not to A1 substitution. Both value and chroma increased linearly with increasing A1 substitution. For goethite, hue was related to both A1 substitution and MCD~ 10- Munsell value tended to decrease with increasing A1 substitution. Thus, goethite samples became darker (decreasing value) but hematite samples became lighter with increasing A1 substitution. The chroma ofgoethite showed no trends with either A1 substitution or MCD~0, but goethites synthesized from an Fe z+ system had higher chromas than those from an Fe 3+ system. The relationship ofgoethite properties and the oxidation state of the environment in which they form needs more study. Colors calculated from spectral data agreed well with colors read from a Munsell chart. REFERENCES Barron, V. and Torrent, J. (1984) nfluence of aluminum substitution on the color of synthetic hematites: Clays & Clay Minerals 32, Berkheiser, V. E. and Monsees, M. B. (1982) Dispersion of clays on graphite supports for X-ray microprobe analysis: Soil Sci. Soc. Amer. J. 46, Bernas, B. (1968) A new method of decomposition and comprehensive analysis of silicate by atomic absorption spectroscopy: Anal Chem. 40, Bryant, R. B., Curl, N., Roth, C. B., and Franzmeier, D. P. (1983) Use of an internal standard with differential x-ray diffraction analysis for iron oxides: Soil ScL Soc. Amer. J. 47, Fischer, W. R. and Schwertmann, U. (1975) The formation of hematite from amorphous iron(ll) hydroxide: Clays & Clay Minerals 23, Goodman, B. A. and Lewis, D.G. (1981) Mfssbauer spectra of aluminous goethites (a-feooh): a t. Soil Sci. 32, J6n~s, tc and Solymfir, K. (1970) Preparation, x-ray deriva- tographic and infrared study of aluminum-substituted goethites: Acta Chim. Acad. Sci. Hung. 66, Klug, H. P. and Alexander, L. E. (1974) X-ray Diffraction Procedures for Polycrystalline and Amorphous Materials: 2nd ed., Wiley, New York, 996 pp. Kosmas, C. S. (1984) Visible spectra and color of synthetic Al-substituted goethites and hematites: Ph.D. thesis, Purdue Univ., West Lafayette, ndiana, 186 pp. (Diss. Abstr. nt. 4fi:1969-B). Kosmas, C. S., Cuff, N., Bryant, R. B., and Franzmeier, D. P. (1984) Characterization of iron oxide minerals by second derivative visible spectroscopy: Soil Sci. Soc. Amer. J. 48, Lewis, D. G. and Schwertmann, U. (1979) The influence of aluminum on the formation of iron oxides. V. The influence of A1, OH, and temperature: Clays & Clay Minerals 27, Rend6n, J. L. and Serna, C. J. (1981) R spectra of powder hematite: effects of particle size and shape: Clay Miner. 16, Ross, D. A., Dhadwal, H. S., and Dyott, R. B. (1978) The determination of the mean and standard deviation of the size distribution of a colloidal suspension of submicron particles using the fiber optic Doppler anemometer, FODA: J. Colloid nterface Sci. 64, Savitzky, A. and Golay, M. J. E. (1964) Smoothing and differentiation of data by simplified least squares procedures: Anal. Chem. 36, Schulze, D. G. (1984) The influence of aluminum on iron oxides. V. Unit-cell dimensions of Al-substituted goethites and estimation of A1 from them: Clays & Clay Minerals 32, Schulze, D. G. and Schwertmann, U. (1984) The influence of aluminium on iron oxides: X. Properties of Al-substituted goethites: Clay Miner. 19, Schwertmann, U. (1964) DifferenzierungderEisenoxidedes Bodens durch Extraktion mit Ammoniumoxalate-LiSsung: Z. Pflanzenerniihr., Diing., Bodenkunde 105, Schwertmann, U., Fitzpatrick, R. W., Taylor, R. M., and Lewis, D. G. (1979) The influence of aluminum on iron oxides. Part 11. Preparation and properties of Al-substituted hematites: Clays & Clay Minerals 27, Sidhu, P. S., Gikes, R. J., and Posner, A. M. (1980) The behavior of Co, Ni, Zn, Mn and Cr in magnetite during alteration to maghemite and hematite: Soil Sci. Soc. Amer. J. 44, Steinier, J.,Termonia, Y.,andDeltour, J. (1972) Comments on smoothing and differentiation of data by simplified least squares procedure: Anal. Chem. 44, Thiel, R. (1963) Zum System afeooh-aaooh: Z. Anorg. Allg. Chem. 326, Wu, Paul Pao-Lo (1983) Measurement of particle interactions in colloidal systems using a fiber optic Doppler anemometer (FODA) and viscometry: Ph.D. Thesis, Purdue Univ., West Lafayette, ndiana, 295 pp. (Diss. Abstr. nt. 44: ). Wyszecki, G. and Stiles, W. S. (1982) Color Science: Concepts and Methods, Quantitative Data and Formulae: 2nd ed., Wiley, New York, 950 pp. (Received 17 October 1985; accepted 16 May 1986; Ms. 1531)