American Mineralogist, Volume 85, pages 473–479, 2000

Raman spectroscopic characteristics of Mg-Fe-Ca pyroxenes

E. HUANG,* C.H. CHEN, T. HUANG, E.H. LIN, AND JI-AN XU

Institute of Earth sciences, Academia Sinica, Taipei 115, Taiwan

ABSTRACT
Raman spectra of several compositions of (Mg, Fe, Ca)SiO3 pyroxenes were collected at ambient
conditions. More than 10 Raman vibrational modes were observed for these pyroxenes in the
wavenumber range between 200 and 1200 cm–1. In general, these pyroxenes are characterized by (1)
the Si-O stretching modes above 800 cm–1; (2) the Si-O bending modes between 500 and 760 cm–1;
(3) SiO4 rotation and metal-oxygen translation modes below 500 cm–1. For a constant Ca content,
frequencies of the Raman modes in the enstatite-ferrosilite (opx) and diopside-hedenbergite (cpx)
series generally decrease with an increase in Fe content. This phenomenon is attributed to an in-
crease in both the bonding lengths and the reduced mass as Fe2+ is substituted for Mg. However, two
modes at ~900 cm–1 in the enstatite-ferrosilite series increase in frequencies as Fe content increases.
A possible explanation is to the shortening in the Si-O-Si bridging bonding bonds when the M2 sites
are preferentially occupied by the iron cation. The effect of Fe substituting for Mg on the frequency
shift in cpx is less profound than opx because the larger M2 was occupied by calcium and the
substitution of iron and magnesium in the M1 site results in a less significant change in the bond
length. The major-element composition of the (Mg, Fe, Ca)-pyroxenes, especially the orthopyroxene
series, can be semi-quantitatively determined on the basis of the peak positions of their characteris-
tic Raman modes.

INTRODUCTION EXPERIMENTAL METHODS

Spectroscopic methods have become a vital investigative Chemical compositions of all natural single-crystals of py-
tool in determinative mineralogy (Calas and Hawthorne 1988). roxene (Table 1) were analyzed by the electron microprobe. In
The infrared (IR) absorption and Raman scattering methods each sample, several different spots were analyzed to check
probe the lattice vibrations of a material. Applications of these the homogeneity of the sample. We also examined series of
techniques to the study of rock-forming minerals, which en- synthetic orthopyroxene with compositions (En97.5, En80, En75,
ables us to understand the lattice dynamics of minerals (e.g., En70, En60, En50, En40, En35, En30, En25, En17, and En10) that were
McMillan and Hofmeister 1985) and for the calculation of the produced by H. Yang. See Yang and Ghose (1994) for details.
thermodynamic parameters of the minerals (e.g., Kieffer 1985; The chemical composition of all the specimens in this study is
Xu et al. 1995). plotted in Ca-Mg-Fe pyroxene diagram (Fig. 1). Strictly speak-
Minerals of the pyroxene group (MSiO3, where M are cat- ing, wollastonite is not a pyroxene. However, it is also included
ions such as Mg, Ca, Fe2+) are one of the most abundant rock- in this study because it is one of the end-members in the
forming minerals on Earth (Deer et al. 1966). Raman MgSiO3-FeSiO3-CaSiO3 composition diagram.
spectroscopic studies on pyroxenes are relatively few Raman spectra were excited using the 514 nm line of an
(McMillan and Hofmeister 1988) and most are limited to end- Argon-Ion laser. We used a Raman spectrometer of the
member compositions (e.g., White 1975; Mao et al. 1987; Ghose Renishaw Company, which contains a micro-objective that fo-
et al. 1994; Hugh Jones et al. 1998; Wang et al. 1994; Chopelas cuses the size of the laser beam to about 5 µm at the surface of
1999). Here, we report a systematic survey on the Raman spectra the sample. A charge-coupled device is used as a detector to
of (Mg, Fe, Ca)-bearing pyroxenes over a wide range of com- collect the signal in an 180° geometry. The position of the
position. The results are used to infer and interpret the possible Raman peaks was determined by the PeakFit program. Typical
bonding characters in these minerals. Moreover, the feasibility acquisition time was about 10 minutes. Normally, the intensity
of using Raman spectroscopy as an indicator of chemical com- of the modes varies with the orientation of the single crystal
position of pyroxenes is discussed. under investigation whereas the wavenumber of the modes re-
mains the same. Therefore, in each sample, several orienta-
tions are attempted and an average in the peak position was
taken for each vibrational mode. The spectral resolution for
*E-mail: [email protected] each mode is on the order of ±1 cm–1.
0003-004X/00/0304–473$05.00 473
474 HUANG ET AL.: RAMAN OF PYROXENES

TABLE 1. Chemical composition of natural pyroxene specimens

No. Mineral Source SiO2 FeO MgO CaO Al2O3 Na2O MnO TiO2 Cr2O3 Total
1 Enstatite Bamble 57.84 6.66 34.84 0.25 0.39 0.01 0.12 —— —— 100.11
2 Enstatite Bamble 56.73 8.76 33.12 0.26 0.09 0.05 0.10 0.05 0.00 99.15
3 Enstatite St. Ludger-de-Milo 54.06 13.35 29.35 0.24 3.78 0.01 0.27 0.06 0.10 101.22
4 Enstatite Sh. John Lake 52.60 14.24 28.08 0.20 3.68 0.00 0.19 0.05 0.12 99.16
5 Enstatite Tsune Cho 54.22 16.82 26.23 1.31 1.49 0.03 0.43 0.30 0.04 100.87
7 Enstatite Ekersund 50.56 26.97 18.01 2.02 1.17 0.05 0.42 0.19 0.04 99.43
8 Augite Hualalai 49.37 5.89 16.24 21.80 5.34 0.43 —— 0.65 0.61 100.33
10 Augite Ney Mineral Co. 54.47 15.46 15.39 12.01 1.64 0.39 0.08 0.04 0.03 99.51
11 Augite Ney Mineral Co. 53.99 16.83 14.38 12.50 1.80 0.52 0.14 0.06 0.00 100.22
13 Diopside Natural Bridge 54.87 0.24 18.30 25.63 0.11 0.34 0.04 —— —— 99.53
14 Diopside Birds Creek 55.11 0.82 17.89 25.06 0.09 0.06 0.05 0.21 0.00 99.29
15 Diopside Templeton 53.69 2.83 15.94 24.72 2.11 0.70 0.14 0.10 0.00 100.23
16 Diopside Pierrepout 54.08 5.31 14.48 23.97 1.18 1.10 0.21 0.05 0.03 100.40
17 Diopside Otter Lake District 52.34 10.01 11.65 23.64 0.86 0.82 0.27 0.01 0.01 99.61
18 Hedenbergite Obira mite 50.20 22.56 2.98 23.42 0.24 0.05 1.02 0.02 0.01 100.50
19 Wollastonite Willsboro 51.44 0.00 0.04 48.83 0.02 0.03 0.01 0.03 0.01 100.41
* Chen et al. (1992).
† Jarosewich et al. (1980).

TABLE 2. The Raman active vibrational modes of the pyroxene in this study
Composition ν1 ν2 ν2' ν3 ν3' ν4 ν5 ν6 ν7 ν8 ν9 ν10 ν11 ν12
En.975Fs.025 239 – 304 343 – 402 421 446 523 541 552 583 663 686
En.80Fs.20 233 – – 335 – 396 412 – 518 538 – 573 658 678
En.75Fs.30 230 – – 333 – 392 408 – 516 537 – 570 656 676
En.70Fs.40 229 – – 331 – 390 406 – 515 536 – 566 655 675
En.60Fs.50 – – – 327 – 383 – – 511 533 – 562 652 671
En.50Fs.60 – – – 325 – 380 – 416 511 531 – 558 649 669
En.40Fs.65 – – – 319 – 370 – 406 508 528 – 554 645 665
En.35Fs.70 – – – 320 – 372 – 406 508 528 – 553 645 665
En.30Fs.75 – – – 315 – 365 – 405 506 526 – 550 642 663
En.25Fs.83 – – – 313 – 360 – 400 504 524 – 548 640 661
En.17Fs.83 – – – 312 – 356 – 395 502 523 – 547 637 660
En.10Fs.90 – – – 305 – 349 – 392 502 519 – 537 633 657
#1
En.90Fs.10 235 – 297 341 – 400 – 441 521 537 549 – 662 683
#2
En.87Fs.13 234 – 297 339 – 398 – 439 518 537 548 – 661 682
#3
En.80Fs.20 231 – 295 338 – 397 – 437 515 534 543 – 657 680
#4
En.78Fs.22 231 – 295 337 – 395 – 437 513 537 544 – 656 680
#5
En.72Fs.26Wo.03 230 – 291 335 – 394 – 436 515 533 545 – 657 670
#7
En.52Fs.44Wo.04 224 – 292 330 – 386 – 420 513 530 539 – 653 672
#10
En.47Fs.27Wo.26 218 – 290 – 357 382 427 – – 527 558 – 665 –
#11
En.44Fs.29Wo.27 218 – 289 – 359 383 424 – – 525 553 – 667 –
#13
En.50Wo.50 230 255 – 325 359 391 – – – – 558 – 665 –
#14
En.49Fs.01Wo.50 230 255 – 325 360 392 – – 508 528 559 – 665 –
#15
En.45Fs.04Wo.50 230 255 – 325 359 392 – – 509 529 558 – 665 –
#16
En.42Fs.08Wo.50 229 253 – 325 359 392 – – 508 528 559 – 664 –
#17
En.34Fs.16Wo.50 232 251 – 321 354 386 – – 505 529 557 – 663 –
#18
En.09Fs.39Wo.52 233 – – 307 338 375 404 – 495 522 549 – 660 –
#8
En.46Fs.09Wo.44 – – – 323 352 387 – 461* 508 531 556 – 662 –
#19
Wo 226*,237 257* 303 321*,337 – 400 412 – 485* – – 581 636 688

RESULTS Augite and pigeonite show little variation in chemistry and

More than ten Raman vibrational modes were observed in therefore are difficult to find a systematic change in the Raman
these pyroxenes in the wavenumber range between 200 and modes with respect to the iron content. For a constant Ca con-
1200 cm–1. (This number is far less than the expected number tent, most Raman frequencies in the enstatite-ferrosilite and
of modes, see below.) The pyroxenes analyzed are classified diopside-hedenbergite series decrease with increase iron con-
as five groups as (1) synthetic orthopyroxene (opx, enstatite- tent. We observed a linear relationship between the Raman fre-
ferrosilite series); (2) natural orthopyroxene; (3) clinopyroxene quencies and Fe content for all observed modes in these two
(cpx, diopside-hedenbergite series); (4) subcalcic cpx (augite); series of pyroxenes. In general, the slopes (dν/dx, where x is
and (5) wollastonite (Wo). Table 2 lists the observed frequen- the Fe2+ content in mol%) of Raman modes in the opx is steeper
cies and representative Raman spectra are shown in Figure 2. than those in the cpx (Table 3). All modes in opx series show
The variations of Raman frequency with Fe content (in negative slope except for ν14 and ν15, which show a positive
mol%) for the opx series are in Figures 3 and 4, respectively. correlation with Fe2+ content. All modes in cpx series show
 HUANG ET AL.: RAMAN OF PYROXENES 475

TABLE 1.—Extended
Mg*=Mg/(Mg+Fe) En Fs Wo
0.90 0.90 0.10 0.00
0.87 0.87 0.13 0.00
0.80 0.79 0.20 0.00
0.78 0.78 0.22 0.00
0.74 0.72 0.26 0.03
0.54 0.52 0.44 0.04
0.83 0.46 0.09 0.44
0.64 0.47 0.27 0.26
0.60 0.44 0.29 0.27
0.99 0.50 0.00 0.50
0.97 0.49 0.01 0.50
0.91 0.45 0.04 0.50
0.93 0.42 0.09 0.50
0.67 0.34 0.16 0.50
0.19 0.09 0.39 0.52
1.00 0.00 0.00 1.00

TABLE 2. —Extended
ν13 ν14 ν15 ν16 ν17 ν17' ν17
'' FIGURE 1. The chemical composition of the pyroxene specimens
750 854 933 1013 1033 – – determined by the electron microprobe in this study. Solid circles =
748 862 935 1006 1025 – – natural crystals; open circles = synthetic polycrystals.
748 866 935 1007 1022 – –
748 868 936 1004 1013 – –
745 876 940 1003 – – –
747 880 942 998 – – –
740 884 942 994 – – –
– 884 946 995 – – –
– 888 945 993 – – –
– 886 946 992 – – –
– 884 946 991 – – –
– 890 949 988 – – –

746 857 938 1010 1025 – –

748 858 938 1009 1025 – –
746 860 939 1003 1019 – –
– 860 941 1004 1019 – –
– 864 941 1003 1014 – –
– 880 948 997 1008 – –

738 – 927 – 1023 1048 1060

736 – 927 – 1023 1049 1061

– 852 – 1011 – 1048 –

– 852 – 1012 – 1048 –
– 854 – 1011 – 1047 –
– 856 – 1011 – 1044 –
– 853 – 1012 – 1043 –
750 853 907 1012 – 1031 –
769* 862 – 1007 – 1038 –

– 883 970* 997 1020 1044 –

negative slope except for ν1, ν14, and ν15, which show nearly
flat to positive correlation with Fe2+ content. Accordingly, the
Raman frequencies for the end-member compositions in the FIGURE 2. Representative Raman spectroscopic patterns of the five
opx and cpx series can be obtained by the extrapolation of the types of (Mg, Fe, Ca)SiO3 pyroxenes.
linear relationships shown in Figures 3 and 4. Raman frequen-
cies for the end-member Mg-Fe-Ca pyroxenes from the ex-
DISCUSSION
trapolated lines in Figures 3 and 4 are listed in Table 4. The
extrapolated mode frequencies of the end-members are in ex- Characteristic Raman-active vibrational modes of
cellent agreement with those reported by Chopelas (1999) for pyroxene
enstatite (see Fig. 3) and in agreement with those of Zhang and The Raman-active modes detected in natural and synthetic
Chopelas (unpublished data) for hedenbergite (see Fig. 4), re- orthopyroxene specimens listed in Table 2 are designated ν1 to
spectively. ν17 with increasing wavenumber. Note that difference in the sym-
476 HUANG ET AL.: RAMAN OF PYROXENES

metry between orthopyroxene and clinopyroxene would yield cient to identify the modes with nearly the same wavenumber.
spectra that are not exactly analogous. Some modes observed in If a photomultiplier tube was used in stead of CCD as a detec-
the opx series will not appear in the cpx series or wollastonite. tor, the background can be minimized and some of the weaker
In the enstatite-ferrosilite series of the natural specimens, three modes can be detected (Chopelas 1999, personal communica-
moderate to weak modes (ν1, ν2, and ν3) below 360 cm–1, two tion). A further reason may be that the disorder created by cat-
moderate modes (ν11 and ν12) in the range from 600 to 700 cm–1, ion substitution broadens the peaks, making them indistinct.
and three intense ones from 900 to 1050 cm –1 are the most promi- Because the symmetry differs for opx and cpx, a direct com-
nent Raman vibrational modes (Fig. 2). Depending on orienta- parison of modes may be misleading. However, it is conve-
tion, the relative intensity of these modes may vary from run to nient to list all modes with a similar frequency together.
run and may also vary with composition in this series (Fig. 13 In general, these pyroxenes are characterized by (1) the Si-
in Lin 1995). The Raman spectra of the synthetic opx are in O nonbridging and bridging modes at 1020 ± 50 cm–1 and 900
general similar to those of the natural samples. The synthetic ± 50 cm–1, respectively; (2) the Si-O bending modes at 600 ±
samples lack ν2 and ν9 (except for En97.5) but have additional 90 cm–1 and (3) metal-oxygen bending and stretching modes
ν10 when compared with the natural samples. Moreover, some below 600 cm–1. Peaks in the 800–1100 cm–1 region are gener-
modes such as ν5 and ν17 only present in Mg-rich opx synthetic ally assigned to Si-O stretching vibrations, normally related to
samples. The diopside-hedenbergite series differs from the opx the non-bridging Si-O bonds (McMillan and Hofmeister 1988;
by the appearance of four Raman modes below 360 cm–1, only Lazarev 1972; Farmer 1974; McMillan 1984). The wavenumber
one mode from 600 to 700 cm–1 and two modes from 900 to of the Si-O stretching modes of the bridging O atoms (Si-Obr)
1050 cm–1. The mode appears at ~250 cm–1 in the cpx series is is assigned to be in the range between 650–750 cm–1. The bend-
termed ν2' rather than ν2 because it has a much higher frequency ing modes are those from the O-Si-O bends with a wavenumber
than ν2 in opx. The Raman spectra of augite studied have the between 500–590 cm–1. Therefore, in the opx series, there are
characteristics of both opx and cpx. However, it differs from four modes from the stretching bridging O atoms, three modes
opx by the lack of ν10 and ν12 and differs from cpx by the lack from the stretching nonbridging O atoms, four modes from the
of ν2' and the presence of ν15. However, due to the limited num- bending O-Si-O. In addition, two modes may result from the
ber of the specimens examined, the result is not conclusive and Mg-O octahedron (375–490 cm–1) and two modes from the Fe-
may not be taken as a general character Raman modes for O octahedron (225–325 cm–1). The assignments are in Table 4.
subcalcic cpx. The pyroxenoid wollastonite shows six modes Raman measurements should be carried out by inciting the
below 360 cm–1 and four modes in the range between 900 to polarized laser beam along some oriented crystals. The signals
1050 cm–1, some which are drastically different from the py- should be taken along some polarized directions. Mao et al. (1987)
roxene minerals, as indicated in Table 2. The characteristic found a little variation in the wavenumber with the orientation
Raman modes for the five groups of specimens in pyroxene of the crystal. In our measurements, we did not observe such
(and pyroxenoid) are compared in Table 5. phenomenon. The modes observed represent the average vibra-
Qualitative identification of the opx and cpx series can be tional modes of the pyroxene in random orientation.
achieved by checking the presence or absence of the character-
istic Raman modes (Table 5). However, one should keep in Variation of the Raman peaks with Fe2+ content
mind that the absence of the Raman modes mentioned above The vibrational frequency of a simple harmonic oscillator
in each series does not positively rule out the presence of the can be expressed as ν = 1/(2π)*(k/µ)1/2 where µ is the reduced
mineral in that series, only the presence of the characteristic mass, k is the force constant which is assumed to be an inverse
modes in a series indicates the presence of the minerals in that function of interatomic distance, r. The reduced mass increases
series. Quantitatively, the opx and cpx can be distinguished by with the Fe content in the pyroxene solid solutions. The nega-
the frequency of ν14. The frequency of ν14 is always greater than tive correlation of the Raman frequency with Fe content is at-
854 cm–1 in opx while that of the cpx is less than 856 cm–1. tributed to the increase in both reduced mass and bond length
as Fe substitutes for Mg (see below).
Mode assignment The structure of pyroxene minerals consists of chain SiO4
Orthopyroxene has the Pbca space group (D2h) with Z = 8. tetrahedra parallel to the c axis, that are linked laterally by lay-
The 240 total vibrational modes are 30A1g (R) + 30B1g (R) + ers containing six to eightfold-coordinated M cations. The M
30B2g (R) + 30B3g (R) + 30A1u + 30B1u (IR) + 30B2u (IR) + 30B3u cations can reside in two crystallographic distinct sites, M1
(IR) for orthoenstatite, where R and IR denote Raman- and and M2. M1 is nearly a regular octahedron, but the larger M2
infrared-active modes, respectively (Chopelas 1999). There are site coordination is irregular and varies according to the atom
fewer the optical modes for the C2/c end-member cpx, Ca(Mg, present, sixfold for Mg, eightfold for Ca and Na. In
Fe)Si2O6, series because Z is 4. The Raman-active modes in orthorpyroxenes, long-range order occurs between divalent Mg
cpx are 14 Ag + 16 Blg (Rustein and White 1971). We observed and Fe atoms in the two M sites, with Fe preferring the larger
less than 20 Raman-active modes in each solid solution. From M2 sites (Ghose 1965). In the diopside-hedenbergite series,
intensity considerations and compared to Chopelas (1999; Table the larger cation, Ca2+ predominately occupies the greater M2
1) most modes are assigned as Ag. The reasons for the low num- sites whereas the smaller Mg2+ and Fe2+ cations are randomly
ber compared to theory might be that some of the modes are distributed among the smaller M1 sites.
degenerate, some modes might be too weak to be observed, or The average O-M1 interatomic distance of opx increases
the resolution of the present Raman spectroscopy is insuffi- with Fe content (Domenegheti et al. 1985). A similar trend also
 HUANG ET AL.: RAMAN OF PYROXENES 477

FIGURE 4. Raman frequencies vs. Fe content for the diopside-

hedenbergite series. Solid circles: composition with Wo = 50 mol%.
Open triangles = composition with Wo = 44 mol% (no. 8, Table 2).
Open squares = unpublished data from Zhang and Chopelas.

FIGURE 3. Raman frequencies vs. Fe content for the enstatite-

ferrosilite series. Solid circles = natural crystals; open circles = synthetic
polycrystals. Data of enstatite reported by Chopelas (1999) are shown
as open squares. exists for M2. Therefore, most of the vibrational modes in opx
decrease in wavenumber as Fe content increases. However, for
the cpx series substitution of Fe for Mg does not cause a seri-
TABLE 3. The slope of Raman frequency versus composition ous distortion in the octahedron and, therefore, does not cause
(dν/dx) for various modes in the opx and cpx series a significant variation in the Raman frequency as compared
Enstatite–ferrosilite Diopside–hedenbergite with opx series. The relatively rigid Si-O bands can be affected
series series by the cation substitution. The frequency of the ν1 mode of cpx
ν1 –0.3196 +0.0423 shows a positive correlation with the Fe content and cannot be
ν2' – –0.1339
ν2 –0.1684 –
accounted for by the mechanism proposed above. However, if
ν3 –0.4018 –0.2296 225 cm–1 (from Zhang and Chopelas 1994) is taken as the ν1
ν3' – –0.273 mode frequency of the end-member hedenbergite, then a nega-
ν4 –0.6097 –0.216
ν5 –0.5542 – tive slope is established. However, in both of the opx and cpx
ν6 –0.6223 – series, some other modes show a positive dν/dx slope which
ν7 –0.2388 –0.179; –0.828 needs to be accounted for.
ν8 –0.2457 –
ν9 –0.2777 –0.1189 In pyroxene, the interatomic distance of Si-Obr decreases as
ν10 –0.461 – the mean ionic radius of the cation increases while the inter-
ν11 –0.3308 –0.0639
ν12 –0.3114 –
atomic distance of Si-Onbr increases as the mean ionic radius of
ν13 –0.143 – cation increases (Cameron and Papike 1981; Domeneghetti et
ν14 +0.4067 +0.0016 al. 1985). The mean ionic radius of the M-cation increases with
ν15 +0.1916 +0.0097
ν16 –0.2792 – Fe2+ substitution for magnesium in pyroxenes. Therefore, the
ν17 –0.5234 – modes of which the frequency shows positive correlation with
ν18 – –0.2098 the Fe content may be related to the stretching between the bridg-
478 HUANG ET AL.: RAMAN OF PYROXENES

TABLE 4. Raman modes of the end-member Mg-Fe-Ca pyroxenes

Enstatite Fesilite Diopside Hedenbergite Wollastonite Mode type
ν1 238 206 230 234 226 M-O stretch
ν2' – – 255 242 237, 337 Ca-O stretch
ν2 298 281 – – – Mg-O stretch
ν3 344 304 327 304 303 M-O stretch
ν3' – – 360 333 321, 337 Ca-O stretch
ν4 407 346 393 372 – Mg-O stretch
– – – – 400, 412 Ca-O stretch
ν5 422 367 – – – Mg-O stretch
ν6 447 685 – – – Mg-O stretch
ν7 522 498 510, 530 492, 521 485 O-Si-O bend
ν8 543 518 – – -— O-Si-O bend
ν9 551 523 559 547 – O-Si-O bend
ν10 582 536 – – 581 O-Si-O bend
ν11 665 632 665 659 636 Si-O-Si bend
ν12 685 653 – – 688 Si-O-Si bend
ν13 751 737 – – – Si-O-Si bend
ν14 856 897 853 855 883 Si-O(br) stretch
ν15 931 950 1010 1012 970 Si-O(br) stretch
ν16 1013 985 – – 997 Si-O(br) stretch
ν17 1035 983 – – 1020 Si-O(br) stretch
ν17' – 1047 1027 1044 Si-O(br) stretch

TABLE 5. Characteristic Raman modes of the five groups of specimens in pyroxene

200–360 cm–1 900–1050 cm–1 ν2' ν2 ν6 ν9 ν10 ν12 ν15 ν14
synthetic opx 2–3 modes 2–3 modes – + ± ± + + + >854
natural opx 3 modes 3 modes – + + + – + + >854
cpx 4 modes 2 modes + – – + – – – <856
augite 2 modes 4 modes – + – + – – + –
Wo 6 modes 4 modes + + + + + – + 883

ing oxygen and silicon because the Si-Obr distance decreases with in the opx and cpx series, respectively.
an increase in the Fe content. According to this interpretation, The cpx series plotted on Figure 4 showing linear relation-
the Si-Obr stretching mode frequencies are less than those for the ship of mode frequencies vs. Fe content have Wo contents near
Si-Onbr modes in the opx series, which is contrary to Williams’ 50 mol%. Raman mode frequencies of cpx with a large amount
(1995) scheme. Moreover, the Si-Onbr mode frequencies gener- of Ca content would yield a recognizable difference from the
ally decrease with the Fe content in Mg-Fe silicate solid solution linear trends shown in Figure 4. For instance, cpx with the com-
such as the forsterite to fayalite series where there are no bridg- position of En0.46Fs0.09WO0.44 has it Raman mode frequencies
ing bonds (e.g., Guyot et al. 1986). deviate from the linear trends and can be resolved by the Raman
spectroscopy.
RAMAN SPECTROSCOPY AS A MICROPROBE FOR
PYROXENES ACKNOWLEDGMENTS
Based on Raman peak positions, the major-element com- We thank H.K. Mao for providing synthetic samples of orthopyroxene and
helpful comments during the preparation of this manuscript. We acknowledge
position of the (Mg, Fe, Ca)-pyroxenes can be estimated. The the help from V. Cheng for the preparation of the diagrams. This is a publication
benefit of using Raman as a tool for chemical identification of of the Institute of Earth Sciences, Academia Sinica.
minerals is its nondestructive nature, and the sample size un-
der investigation can be as small as a few micrometers. REFERENCES CITED
The Raman modes showing the greatest variation in fre- Calas, G. and Hawthorne, F.C. (1988) Introduction to spectroscopic methods. In
quency with the Fe2+ content are potentially capable of being Mineralogical Society of America Reviews in Mineralogy, 18, 1–9.
Cameron, M. and Papike, J.J. (1981) Structural and chemical variations in pyroxenes.
used as an index for determining the Fe content in the opx and American Mineralogist, 66, 1–50.
cpx series. The best Raman modes for this purpose in the opx Chen, C.H., Presnell, D.C., and Stern, R.J. (1992) Petrogenesis of ultramafic xeno-
series are ν6 (447–385 cm–1) and ν14 (856–897 cm–1 ) pairs that liths from the 1800 Kaupulehu flow, Hualalai Volcano, Hawaii. Journal of Pe-
trology, 33, 163–202.
show about 60 cm–1 across the series. The ν15 (931–950 cm–1) Chopelas, A. (1999) Estimates of mantle relevant Clapeyron slopes in the MgSiO3
mode can also serve as a supplementary mode for the determi- system from high-pressure spectroscopic data. American Mineralogist, 84, 233–
244.
nation of the composition. These modes are characteristic and Deer, W.A., Howie, R.A., and Zussman, J. (1966) Introduction to the Rock Forming
fairly intense. However, in the cpx series, because only the ν3 Minerals. Wiley, New York.
(327–304 cm–1 ) and ν3' (360–333 cm–1 ) pairs show a variation Domeneghetti, M.C., Molin, G.M., and Tazzoli, V. (1985) Crystal-chemical impli-
cations of the Mg2+-Fe2+ distribution in orthopyroxenes. American Mineralo-
more than 20 cm–1. These provide a weaker basis for the com- gist, 69, 987–995.
position indices. The uncertainties in the determination of the Farmer, V.C. (1974) Orthosilicate, pyrosilicates, and other finite-chain silicates. In
Fe content with the Raman modes are of the order of 3 and 6% V.C. Farmer, Ed., The Infrared Spectra of Minerals, p. 285–303. Mineralogical
Society of London.
 HUANG ET AL.: RAMAN OF PYROXENES 479

Ghose, S. (1965) Fe 2+-Mg 2+ order in an orthopyroxene. Zeitschrift für spectroscopy to analysis and identification of minerals in thin section. Scan-
Kristallographie, 122, 81–99. ning Microscopy, 1, 495–501.
Ghose, S., Choudhury, N., Chaplot, S.L., Pal Chowdhury, C., and Sharma, S.K. Rutstein, M.S. and White, W.B. (1971) Vibrational spectra of high-calcium pyroxenes
(1994) Lattice dynamics and Raman spectroscopy of protoenstatite Mg2Si2O6. and pyroxenoids. American Mineralogist, 56, 877–887.
Physics of Chemistry and Minerals, 20, 469–477. Wang, A., Han, J., Guo, L., Yu, J., and Zeng, P. (1994) Database of standard Raman
Guyot, F., Boyer, H., Madon, M., Velde, B., and Poirier, J.P. (1986) Comparison of spectra of minerals and related inorganic crystals. Applied Spectroscopy, 48,
the Raman microprobe spectra of (Mg,Fe)2SiO4 and Mg2GeO4 with olivine and 959–968.
spinel structures. Physics of Chemistry and Minerals, 13, 91–95. White, W.B. (1975) Structural interpretation of lunar and terrestrial minerals by
Jarosewich, E., Nelsen, J.A., and Norberg, J.A. (1980) Reference samples for elec- Raman spectroscopy. In C. Karr, Jr., Ed., Infrared and Raman Spectroscopy of
tron microprobe analysis, Geostand. Newsletter, 4, 43–48. Lunar and Terrestrial Minerals, p. 325–358. Academic Press, New York.
Kieffer, S.W. (1985) Heat capacity and entropy: systematic relations to the lattice Williams, Q. (1995) Infrared, Raman and optical spectroscopy of Earth materials.
vibrations. In Mineralogical Society of America Reviews in Mineralogy, 14, In T.J. Ahrens, Ed., Mineral Physics and Crystallography: A Handbook of Physi-
65–126. cal Constants, p. 291–302. American Geophysical Union, Washington, D.C.
Lazarev, A.N. (1972) Vibrational Spectra and Structure of Silicates. Translated by Xu, J., Huang, E., Lin, J.F., and Xu, L.Y. (1995) Raman study at high pressure and
G.D. Archard, trans. editor V.C. Farmer, Consultants Bureau, New York, 302 p. the thermodynamic properties of corundum: Application of Kieffer’s model.
(in Russian) American Mineralogist, 80, 1157–1165.
Lin, I.H. (1995) Raman Spectroscopy study of (Ca, Mg, Fe)-Pyroxene. Master The- Yang, H. and S. Ghose (1994) Thermal expansion, Debye temperature and Gruneisen
sis, National Taiwan University, 74 p. parameter of synthetic (Fe,Mg)SiO3 orthopyroxenes. Physics of Chemistry and
McMillan, P. (1984) Structural studies of silicate glasses and melts: applications Minerals, 20, 575–586.
and limitations of Raman spectroscopy. American Mineralogist, 69, 633–644.
McMillan, P.F. and Hofmelster, A.M. (1988) Infrared and Raman spectroscopy. In MANUSCRIPT RECEIVED SEPTEMBER 29, 1998
Mineralogical Society of America Reviews in Mineralogy, 18, 99–159. MANUSCRIPT ACCEPTED NOVEMBER 16, 1999
Mao, H.K., Hemley, R.J., and Chao, E.C.T. (1987) The application of micro-Raman PAPER HANDLED BY ANASTASIA CHOPELAS