Mechanisms of Thermal Color Change in Brown Elbaite–Fluorelbaite Tourmaline: Insights from Trace Elements and Spectral Signatures

Abstract

This study investigates the mechanism behind the heat-induced color change (brown to yellowish green) in Mn- and Fe-rich elbaite tourmaline under reducing atmosphere at 500 °C. A combination of analytical techniques including gemological characterization, electron microprobe analysis (EMPA), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), Fourier-transform infrared spectroscopy (FTIR), Raman spectroscopy, and ultraviolet–visible (UV-Vis) spectroscopy was employed. Chemical analysis confirmed the samples as intermediate members of the elbaite–fluorelbaite series, with an average formula of ^X(Na₀.₆₆ □₀.₂₆ Ca₀.₀₈) _Σ1.₀₀^Y(Li₁.₂₉Al₁.₁₀Mn₀.₃₁ Fe²⁺₀.₁₅Ti₀.₀₁Zn₀.₀₁) _Σ2.₈₇ ^ZAl₆^T[Si₆O₁₈] (BO₃)₃^V(OH)₃.₀₀^W(OH₀.₅₁F₀.₄₉) _Σ1.₀₀, enriched in Mn (17,346–20,669 μg/g) and Fe (8396–10,750 μg/g). Heat treatment enhanced transparency and induced strong pleochroism (yellowish green parallel c-axis, brown perpendicular c-axis). UV-Vis spectroscopy identified the brown color origin in the parallel c-axis direction: absorption bands at 730 nm (Fe²⁺ d–d transition, ⁵T_2g → ⁵E_g), 540 nm (Fe²⁺→Fe³⁺ intervalence charge transfer, IVCT), and 415 nm (Fe²⁺→Ti⁴⁺ IVCT + possible Mn²⁺ contribution). Post-treatment, the 540 nm band vanished, creating a green transmission window and causing the color shift parallel the c-axis. The spectra perpendicular to the c-axis remained largely unchanged. The disappearance of the 540 nm band, attributed to the reduction of Fe³⁺ to Fe²⁺ eliminating the Fe²⁺–Fe³⁺ pair interaction required for IVCT, is the primary color change mechanism. The parallel c-axis section of the samples shows brown and yellow-green dichroism after heat treatment. A decrease in the IR intensity at 4170 cm⁻¹ indicates a reduced Fe³⁺ concentration. The weakening or disappearance of the 4721 cm⁻¹ absorption band of the infrared spectrum and the near-infrared 976 nm absorption band of the ultraviolet–visible spectrum provides diagnostic indicators for identifying heat treatment in similar brown elbaite–fluorelbaite.

1. Introduction

Tourmaline is a borosilicate mineral with the following general formula: XY₃Z₆[T₆O₁₈](BO₃)₃V₃W. The X-site contains large cations such as Na⁺, Ca²⁺, K⁺, □(=vacancy); the octahedral Y-site hosts Mg²⁺, Fe²⁺, Fe³⁺, Mn²⁺, Al³⁺, Mn³⁺, Cr³⁺, Li⁺, Ti⁴⁺, etc.; the octahedral Z-site accommodates Al³⁺, Mg²⁺, Fe²⁺, Fe³⁺, Cr³⁺, and V³⁺; the T-site contains Si⁴⁺, Al³⁺, and B³⁺; the V-site includes O²⁻ and OH⁻; and the W-site includes F⁻, OH⁻, and O²⁻ [1,2,3,4,5,6]. The varying charges and sizes of these cations enable their accommodation in the X and Y sites, contributing to tourmaline’s structural complexity.

Based on the occupancies of the X and Y sites, tourmalines are classified into end-members, including schorl (Y = Fe²⁺), dravite (Y = Mg), tsilaisite (Y = Mn), olenite (Y = Al), and elbaite (Y = Li, Al) [1,2]. Complete solid solutions commonly exist between Mg-Fe and Li-Fe end-members but not between Mg-Li end-members. The Z-site is generally occupied by Al in all tourmaline species, though it may occasionally host other trivalent cations. Tourmaline’s diverse colors arise from transition metal ions (Fe²⁺, Fe³⁺, Mn²⁺, Mn³⁺, Cu²⁺, Cr³⁺, V³⁺, Ti³⁺ and Ti⁴⁺, etc.) and their concentrations. Hydroxyl groups occupy two distinct structural sites: (OH1) at the W position and (OH3) at the V position. These OH groups (OH1, OH3) are coordinated by neighboring cations, so their stretching frequencies reflect local cation environments, making vibrational spectroscopy a powerful probe for compositional complexity near hydroxyl sites [7].

Heat treatment is a common gem enhancement technique to improve clarity, color, and thus value. Examples include beryl heat-treated to form aquamarine [8], amethyst heat-treated to form citrine [9], yellow titanite transformed to reddish-brown via heat treatment [10], green apatite converted to sky blue [11], and so on. Heat treatment is widely used due to its harmlessness, ability to produce stable colors, and low cost—and tourmaline is no exception. Previous studies have explored tourmaline heat treatment [12,13,14,15,16], such as most “Paraiba tourmaline” being heat-treated from purple Cu-bearing tourmaline [17,18]. Gem color changes during heat treatment typically result from valence state transitions of chromophoric elements (e.g., Fe³⁺⇆Fe²⁺, Mn³⁺⇆Mn²⁺) under oxidizing or reducing atmospheres.

This study presents the characteristics of brown elbaite before and after heat treatment, using electron microprobe analyses (EMPAs), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS), infrared spectra (IR), Laser Raman, and ultraviolet–visible spectrophotometry (UV-Vis) analyses to discuss its properties and explore the color change mechanism.

2. Materials and Methods

2.1. Samples and Standard Gemological Methods

Samples were obtained from the jewelry market and numbered T2, T5, T23, and T29 (Figure 1), and the exact geological locality was not provided by the vendor. Surfaces were cut and polished for conventional gemological testing. Standard gemological characterization was performed at the Jewelry Testing Center of Guangzhou Polytechnic University to determine refractive index (RI), relative density, and fluorescence under long-wave (365 nm) and short-wave (254 nm) UV light. Optical properties and pleochroism were observed using a polariscope and dichroscope, respectively. Physical appearance was documented under fiber-optic lighting (ambient and transmitted) using a 60 W FABLE LED daylight bulb (5000–6000 K). Microscopic features were examined using a FABLE 4×–40× gemological microscope (Shenzhen, China).

2.2. Sample Preparation and Heat Treatment

Samples had table facets parallel to the c-axis and triangular cross-sections perpendicular to the c-axis, with an emerald cut. They were cut perpendicular to the c-axis, with one portion used for heat treatment and the other as a control (Figure 1a,b). A reducing atmosphere was created by adding graphite powder to the crucible and sealing the lid with clay. Heat treatment conditions are listed in

Table 1. Experimental protocol for heat treatment.

Sample No.	Temperature (°C)	Heating Rate	Holding Time (h)	Atmosphere
		°C/min
T2	500	1	0.5	reduction
T5	500	3	0.5	reduction
T23	500	4	1	reduction
T29	500	2	0.5	reduction

.

Unheated and heat-treated samples were analyzed via IR, Raman, and UV-Vis spectroscopy. Unheated samples were additionally analyzed via EMPA and LA-ICP-MS for compositional analysis. IR, Raman, and UV-Vis analyses were conducted at the Jewelry and Advanced Materials Research Center of Guangzhou Polytechnic University; EMPA and LA-ICP-MS were performed at Guangzhou Tuoyan Analytical Technology Co., Ltd., Guangzhou, China.

2.3. EMPA and LA-ICP-MS

Major elements were analyzed with a JEOL JXA-iSP-100 Electron Probe Microanalyzer (JEOL Ltd., Tokyo, Japan) equipped with five wavelength-dispersive spectrometers (15 kV voltage, 20 nA current, 10 μm beam spot). Data were corrected using the ZAF method, with standards including British MAC mineral/metal standards and Chinese national standard GSB. Trace elements of samples were determined using a New Wave Research 193 nm ArF Excimer laser-ablation system (Elemental Scientific Inc., Omaha, NE, USA) coupled to an iCAP RQ (ICPMS, Thermo Fisher Scientific, Waltham, MA, USA). The ICPMS was tuned using NIST 610 standard glass to yield low oxide production rates. 0.7 L/min He carrier gas was fed into the cup, and the aerosol was subsequently mixed with 0.89 L/min Ar make-up gas. The laser fluence was 3.5 J/cm², with a repetition rate of 6 Hz, a 30 μm spot size and analysis time of 45 s, followed by a 40 s background measurement.

2.4. Fourier-Transform Infrared (FTIR) Spectroscopy

FTIR analysis used a Bruker Tensor 27 Fourier-transform infrared spectrometer (Bremen, Germany, 32 scans, 4 cm⁻¹ resolution). Reflection spectra (1600–400 cm⁻¹; MIR: mid-Infrared) and transmission spectra (9000–4000 cm⁻¹; NIR: near-Infrared) were collected. Reflection spectra (parallel and perpendicular to the c-axis; Figure 1c) and transmission spectra (perpendicular to the c-axis) of unheated and heat-treated samples used MIR and NIR lamp fittings, respectively. The crystal’s c-axis was parallel to the optical axis (OA) of the indicatrix.

2.5. Laser Raman (LR) Spectroscopy

Raman spectroscopy was performed using a Renishaw InVia Reflex confocal microscope (Renishaw plc, Gloucestershire, UK) with a high-sensitivity Centrus CCD detector (1040 × 256 pixels). Spectra were acquired using a 532 nm laser focused through a 50× long-working-distance (LWD) objective, with excitation power set to 100% instrument output (sample power uncalibrated). The system used 1800 lines/mm grating and 65 μm entrance slit for high spectral resolution. Scattered light was collected over 100–4000 cm⁻¹ with 10 accumulations (2 s exposure each) to optimize signal-to-noise while minimizing laser-induced damage.

2.6. Ultraviolet–Visible (UV-Vis) Spectroscopy

UV-Vis analysis used a Guangdong BiaoQi Gem3000 spectrophotometer (Guangzhou, China, detection range: 1000–200 nm; signal-to-noise ratio: 450:1; 220 V, 250 W).

3. Results

3.1. Gemological Characteristics

Gemological properties (appearance, color, refractive index, density, UV fluorescence, pleochroism) are summarized in

Table 2. The gemological characteristics of the brown elbaite.

Sample No.	Weight (g)	Transparency	Refractive Index		Birefringence	UV Fluorescence		Color		Dichroism
			No	Ne		Long	Short	Unheated	Heat-Treated	Unheated	Heat-Treated
T2	0.511	transparent	1.640	1.620	0.020	inert	inert	yellowish brown	yellowish green	reddish brown + yellowish brown	yellowish brown + green
T5	0.470	transparent	1.640	1.620	0.020	inert	inert	reddish brown	yellowish green	Brown + light green	yellowish brown + green
T23	0.442	transparent	1.640	1.620	0.020	inert	inert	reddish brown	yellowish green	purple + reddish brown	yellowish brown + green
T29	0.407	transparent	1.640	1.620	0.020	inert	inert	yellowish brown	yellowish green	pink green + brown	yellowish brown + green

. The gemological properties of the samples are consistent with those of tourmaline. In particular, it is clear that the dichroism of the unheated samples is masked by their strong absorption, and the green and brown dichroism can be clearly seen in the parallel c-axis direction after heat treatment.

3.2. Chemical Composition and Species

The results of major and trace elements in the samples are listed in

Table 3. EMPA results for major elements and molecular formula calculation (wt.%).

Sample No.	T2		T5		T23		T29
	1	2	1	2	1	2	1	2
SiO2	38.94	38.79	38.76	38.82	38.59	38.67	38.79	38.63
TiO2	0.09	0.07	0.06	0.06	0.02	0.04	0.05	0.06
Al2O3	38.57	38.72	38.79	38.85	38.60	38.58	38.47	38.69
FeO	1.41	1.46	1.15	1.13	1.03	1.04	1.10	1.08
CaO	0.40	0.42	0.46	0.43	0.53	0.50	0.50	0.50
MnO	2.19	2.14	2.18	2.29	2.62	2.52	2.43	2.43
ZnO	0.11	0.07	0.07	0.07	0.00	0.04	0.04	0.03
Na2O	2.32	2.26	2.11	2.18	2.10	2.18	2.12	2.05
F	1.04	0.96	1.02	0.94	0.10	1.03	0.10	0.94
H2O *	3.36	3.40	3.37	3.41	3.37	3.36	3.37	3.40
B2O3 *	11.69	11.71	11.74	11.74	11.77	11.77	11.68	11.68
Li2O *	2.04	2.04	2.07	2.07	2.07	2.07	2.07	2.07
Total	102.18	102.05	101.79	102.01	101.72	101.80	101.63	101.57
O=F	0.44	0.41	0.43	0.40	0.42	0.43	0.42	0.39
Total *	101.74	101.64	100.37	101.61	101.30	101.37	101.21	101.17
T: Si	6.06	6.04	6.04	6.04	6.02	6.03	6.06	6.03
B	3.00	3.00	3.00	3.00	3.00	3.00	3.00	3.00
Z: Al	6.00	6.00	6.00	6.00	6.00	6.00	6.00	6.00
Y: Al	1.07	1.10	1.12	1.12	1.10	1.09	1.08	1.12
Ti	0.01	0.01	0.01	0.01	0.00	0.00	0.01	0.01
Mn	0.29	0.28	0.29	0.30	0.35	0.33	0.32	0.32
Fe2+	0.18	0.19	0.15	0.15	0.13	0.14	0.14	0.14
Zn	0.01	0.01	0.01	0.01	0.00	0.00	0.00	0.00
Li	1.28	1.28	1.30	1.29	1.30	1.30	1.30	1.30
X: Ca	0.07	0.07	0.08	0.07	0.09	0.08	0.08	0.08
Na	0.70	0.68	0.64	0.66	0.64	0.66	0.64	0.62
V: OH	3.00	3.00	3.00	3.00	3.00	3.00	3.00	3.00
W: OH	0.49	0.53	0.50	0.54	0.51	0.49	0.51	0.54
F	0.51	0.47	0.50	0.46	0.49	0.51	0.49	0.46

Note: The contents of B₂O₃ * and Li₂O * were calculated by LA-ICP-MS results, the H₂O * and Total * were calculated value; FeO = total Fe.

and

Table 4. LA-ICP-MS results for trace elements (μg/g).

Samples No.	Li	B	Na	Mg	K	Ca	Ti	V	Cr	Mn	Fe	Co	Ni	Cu	Zn
T2	9530.01	36,297.61	18,037.89	5.21	165.33	3187.23	1111.77	—	—	17,346.71	10,750.68	—	—	10.87	736.06
T5	9571.51	36,381.28	17,977.81	4.47	162.35	3480.91	942.31	0.40	—	17,871.53	9001.68	—	—	10.02	569.89
T23	9572.69	36,533.50	17,886.37	3.70	167.76	3948.25	582.43	0.36	—	20,669.52	8396.46	—	—	8.27	276.87
T29	9636.87	36,172.70	17,710.79	3.81	162.99	4100.36	748.28	—	—	19,608.02	8612.76	—	—	10.25	331.12

Note: “—” below detection limits.

, respectively (two EMPA points per sample). The main components included SiO₂ (38.59–38.94 wt.%), Al₂O₃ (38.47–38.85 wt.%), Na₂O (2.05–2.32 wt.%), and a small amount of MnO (2.14–2.62 wt.%), FeO (1.03–1.46 wt.%), and TiO₂ (0.02–0.09 wt.%). B₂O₃ and Li₂O were calculated by the LA-ICP-MS results. H₂O was calculated, and cation counts were determined assuming 31 anions (O, OH, F). In the structural formula calculation, all iron was assigned as Fe²⁺ to the Y-site, as the Fe³⁺/ΣFe ratio could not be determined by EMPA. It should be noted that some Fe³⁺ may occupy the Z-site. The provided formula is therefore an estimate based on this conventional approach. The calculation was performed on averaged compositional data, justified by the homogeneous nature of the samples without visible zoning. Aluminum was first fully allocated to the Z-site, with excess assigned to the Y-site. The average molecular formula was ^X(Na₀.₆₆ □₀.₂₆ Ca₀.₀₈) _Σ1.₀₀^Y(Li₁.₂₉Al₁.₁₀Mn₀.₃₁ Fe²⁺₀.₁₅Ti₀.₀₁Zn₀.₀₁) _Σ2.₈₇ ^ZAl₆^T[Si₆O₁₈] (BO₃)₃^V(OH)₃.₀₀^W(OH₀.₅₁F₀.₄₉) _Σ1.₀₀. The lack of Y-site cations also means the presence of Fe³⁺. LA-ICP-MS analysis indicated negligible V, with Cr, Co, and Ni below detection limits. Mg, K, and Cu contents were very low; detected transition metals included Fe, Ti, and Mn (Table 4).

It should be noted that the fluorine content (F apfu = 0.49–0.51, atoms per formula unit) places these samples close to the boundary between elbaite and fluorelbaite as defined by the IMA (International Mineralogical Association) nomenclature [19]. Given that F exceeds o.5 apfu in some individual spot analyses (e.g., T2-1, T5-1, T23-2), the samples are best described as intermediate members of the elbaite–fluorelbaite solid solution series.

3.3. Heat-Treated Samples’ Features

After heat treatment, the color changed from brown to yellowish green, with a lighter hue, improved transparency, and occasional cracks. Heat-treated samples exhibited strong pleochroism: brown in sections perpendicular to the c-axis (ordinary ray) and yellowish green in sections parallel to the c-axis (extraordinary ray) (Figure 1).

3.4. Infrared Spectroscopy

Spectra of unheated and heat-treated samples were compared (Figure 2, Figure 3 and Figure 4), with band assignments summarized in

Table 5. Assignments of IR bands (1600–400 cm⁻¹) in various tourmalines.

This Study		Elbaite S5 [20]	Elbaite Tg-1 [21]	Schorl [22]	Dravite [22]	Assignment
‖c-Axis Section	⊥c-Axis Section
1354(m)	1353(w)	1355	1357	1353	1345	vs BO3 or vas BO3
1299(s)	1298(s)	1309	1297	1255	1312
					1165	v Si–O–Al
1112(m)	1110(s)	1106	1112	1085	1112	vas Si–O–Si
	1053(w)		1056			vas Si–O–Si
1031(s)	1031(s)	1031	1031	1039	1065	vas O–Si–O
988(s)	992(s)	996	988	985	1040	vas O–Si–O
850(w)						vs O–Si–O
810(w)				800	802	vs O–Si–O
	791(s)	794	790	780	787	vs Si–O–Si
	753(w)	755	752	758	727	vs Si–O–Si
721(w)	716(s)	717	716	713	720	vs Si–O–Si
				651	650	v M–O (M = Fe, Mg, Al)
627(s)	631(s)	629	636	631		vs Si–O–Si
595(s)	596(m)	593	595	608		δ Si–O
565(s)			535	565	590	δ Si–O
510(s)	508(s)	515	503	510	520	δ BO3
				483	475	δ Si–O
452(m)	453(w)	458	449	449		v M–O
426(m)	429(w)	431	428	422	430	v M–O

Note: m—middle; s—strong; w—weak. v—stretching vibrations; v_s—symmetric stretching vibrations; v_as—asymmetric stretching vibrations; δ—deformation vibrations; “‖” means parallel; “⊥” means perpendicular.

and Table 6.

3.4.1. Reflective MIR Spectra (1600–400 cm⁻¹)

Unheated samples showed two distinct spectral regions: 1400–800 cm⁻¹ and 700–400 cm⁻¹. In sections parallel to the c-axis (parallel to the optical axis), bands appeared at 1354, 1299, 1112, 1031, 988, 810, and 721 cm⁻¹ and 627, 595, 565, 510, 452, and 426 cm⁻¹ (Figure 2a–d). In sections perpendicular to the c-axis, bands included 1353, 1298, 1110, 1053, 1031, 992, 791, 753, and 716 cm⁻¹, and 631, 596, 508, 453, and 429 cm⁻¹ (Figure 3a–d). While band positions were similar, spectral morphologies differed: sections parallel to the c-axis had unique 810 and 565 cm⁻¹ bands, while perpendicular sections had unique 791 and 753 cm⁻¹ bands. The sharp, weak bands are observed in each spectrum, and these bands’ displacement show no change after being heat-treated. Sharp, weak bands showed no displacement post-heating, but reflectivity increased in both directions (Figure 2).

The mid-infrared spectra of tourmaline are dominated by fundamental vibrations of infrared-active structural units: YO₆ octahedra, TO₄ tetrahedra, BO₃ triangles, hydroxyl groups, and water molecules.

The double bands at ~1353 cm⁻¹ and 1255 cm⁻¹ in schorl are attributed to antisymmetric BO₃ stretching [22]. Some scholars classify bands near 1300 cm⁻¹ as ν_as BO₃ [20] or ν_s BO₃ [21]. Thus, the 1354 and 1299 cm⁻¹ bands here are assigned to BO₃-related vibrations. The 510 cm⁻¹ (parallel to c-axis) and 508 cm⁻¹ (perpendicular to c-axis) bands are also BO₃-related, attributed to deformation vibrations [21].

Infrared vibrational modes of [SiO₄] tetrahedra ([TO₄]) include ν_s Si–O–Si (symmetric stretching), ν_as Si–O–Si (asymmetric stretching), ν_s O–Si–O, ν_as O–Si–O, and δ Si–O (bending) [20,21]. Observed bands (Figure 3) are assigned as follows: 1112 cm⁻¹ corresponds to the ν_as Si–O–Si mode; bands at 1031 cm⁻¹ and 988 cm⁻¹ are attributed to ν_as O–Si–O modes; features at 850 cm⁻¹ and 810 cm⁻¹ arise from ν_s O–Si–O modes; bands at 721 cm⁻¹ and 627 cm⁻¹ are assigned to the ν_s Si–O–Si mode; and absorptions at 595 cm⁻¹ and 565 cm⁻¹ originate from δ Si–O bending vibrations.

Absorptions at ~400 cm⁻¹ relate to metal cations in YO₆ octahedra (e.g., MgO₆ at 470 cm⁻¹, FeO₆ at 400 cm⁻¹). Bands at 449 cm⁻¹ and 428 cm⁻¹ in Fe-bearing elbaite are assigned to [MgO₆] and [FeO₆] [21], while 457 cm⁻¹ and 431 cm⁻¹ in elbaites correspond to ν Mg–O and ν Fe–O [20]. This study’s samples had weak bands at 452 cm⁻¹ and 426 cm⁻¹ (parallel to c-axis) and 453 cm⁻¹ and 429 cm⁻¹ (perpendicular to c-axis; Figure 2a), with minor band shifts across samples (Figure 2 and Figure 3). Given the samples’ Fe content (and low Mg), these bands may relate to Fe or Mn.

3.4.2. Transmission NIR Spectra (9000–4000 cm⁻¹)

Bands in 8000–4000 cm⁻¹ were divided into three groups: 8000–6000 cm^–1, 5500–4700 cm^–1, and 4600–4000 cm^–1 (Figure 4). Unheated sample T2 showed third-group bands at 4597, 4536, 4440, 4344, 4201, and 4170 cm⁻¹; first-group bands at 7138 and 6997 cm⁻¹; and second-group bands at 5184, 4871, and 4721 cm⁻¹ (Figure 4a). T5, T23, and T29 showed similar bands (Figure 4b–d).

Band positions were unchanged post-heating, but third-group absorption intensity increased in heat-treated samples (except T29; Figure 4). First- and second-group absorbance decreased in T2 and T29 but increased in T5 and T23. Notably, the 4721 cm⁻¹ band weakened significantly (tending to disappear; Figure 4) and vanished entirely at >600 °C [23]. First-group bands (8012, 7138 and 6997 cm⁻¹) enhanced post-treatment, with the 6997 cm⁻¹ band showing the greatest increase. These phenomena may relate to compositional changes along the c-axis or varying thickness. Alternatively, intensity changes in 4600–4300 cm⁻¹ post-heating/irradiation may reflect distances between metal ions and hydroxyl groups [13], as heat treatment alters cation valence states and ionic radii.

Bands in 8000–6000 cm^–1 are attributed to the first overtones of hydroxyl stretching modes (3600–3460 cm^–1) [4,24,25,26,27]. Specifically, 7000 and 6500 cm^–1 correspond to the first overtones of OH3 groups at 3600 and 3460 cm^–1 [4]. Others attribute 7138 and 6995 cm^–1 to the first overtones of OH vibrations in complex tripartite rings [20,27], with strong 6995 cm⁻¹ bands from pure hydroxyls and weak 7132 and 7185 cm⁻¹ bands from water-bound hydroxyls [20]. Thus, the bands at 7138 and 6997 cm^–1 are assigned to OH overtones, and 8012 cm^–1 to Fe^{2+ 5}T_2g–⁵E_g transitions [4].

Second-group bands (5500–4700 cm⁻¹) included 5182, 4871, and 4721 cm⁻¹ (Figure 4). The 4871 cm⁻¹ band is the third overtone of ν_as Si–O–Si [20,27]. Bands in 5400–4900 cm⁻¹ reflect combinations of water bending and stretching vibrations [27], with 5182 cm⁻¹ indicating water content. The disappearance of 4721 cm⁻¹ post-heating suggests that it relates to transition metal ions (e.g., Fe^2+/3+) whose valence states changed during treatment.

Bands in 4800–4000 cm⁻¹ arise from combinations of hydroxyl stretching and bending modes in the ^YM–OH unit (M = Al, Mg, Fe, Mn, etc.) [4]. Despite minor species-specific differences (due to Y-site cation variation), characteristic 4700–4100 cm⁻¹ bands in tourmalines (4600, 4540, 4440, 4340, 4170 cm⁻¹) are assigned to cationic hydroxyl unit vibrations (M–OH; M = Al, Mg, Fe, Mn, Li) [2,4,13,27]. ^YM–OH1 bands are influenced by Y-site cations; some attribute them to ^YAl–OH1 [24] or Al–OH combinations (~4600 cm⁻¹ [4]). Bands in 4500–4300 cm⁻¹ correspond to Fe-OH and Mg-OH, with 4540 cm⁻¹ corresponding to the ^YFe-OH, 4440 cm⁻¹ to the ^Y(Fe, Mg)-OH, and 4344 cm⁻¹ to the ^YMg-OH. Two bands in the range of 4300–4100 cm⁻¹ are assigned to the combination of stretching and bending modes of ^YM–OH3 (M = Al, Li, Fe, Mn) [24] or (Fe²⁺, Mn)-OH [20]. Given Y-site occupancy by Mn, Fe, Al, and Li, the bands at 4204 and 4171 cm⁻¹ are assigned to ^Y (Mn, Fe, Al, Li)–OH3.

Table 6. Assignments of NIR bands (9000–4000 cm⁻¹) in various tourmalines.

This Study		Green Elbaite [20]	Green Elbaite [4]	Pink Tourmaline [24]	Green Tourmaline [24]	Brownish-Green [24]	Assignment
Unheated	Heat-Treated
8012	8012 ↑		8000				First overtones of hydroxyl stretching modes
		7185
7138	7138 ↑	7132	7140
6997	6997 ↑	6995	7000
		6783	6795
5183	5183	5180					The bending vibration of the water and the stretching vibration
4871	4871	4873					νas Si–O–Si
4721	4721 ↓
4597	4597	4598	4597	4604	4604	4596	Combination of stretching and bending modes of M-OH units	YM-OH1
4536	4536	4544	4538	4541	4545	4541		YM-OH1
4440	4440	4444	4433	4444	4448	4444		YM-OH1
4344	4344	4344	4347	4347	4347	4344		ZM-OH3
4201	4201	4206			4211	4214		YM-OH3
4170	4170↓	4164	4171	4146	4177	4175		YM-OH3

Note: “↑”—increase in intensity; “↓”—decrease in intensity; v_as—asymmetric stretching_;^YM and ^ZM —the corner markers Y and Z represent Y and Z sites, respectively, and M represents the metal cations.

Table 6. Assignments of NIR bands (9000–4000 cm⁻¹) in various tourmalines.

This Study		Green Elbaite [20]	Green Elbaite [4]	Pink Tourmaline [24]	Green Tourmaline [24]	Brownish-Green [24]	Assignment
Unheated	Heat-Treated
8012	8012 ↑		8000				First overtones of hydroxyl stretching modes
		7185
7138	7138 ↑	7132	7140
6997	6997 ↑	6995	7000
		6783	6795
5183	5183	5180					The bending vibration of the water and the stretching vibration
4871	4871	4873					νas Si–O–Si
4721	4721 ↓
4597	4597	4598	4597	4604	4604	4596	Combination of stretching and bending modes of M-OH units	YM-OH1
4536	4536	4544	4538	4541	4545	4541		YM-OH1
4440	4440	4444	4433	4444	4448	4444		YM-OH1
4344	4344	4344	4347	4347	4347	4344		ZM-OH3
4201	4201	4206			4211	4214		YM-OH3
4170	4170↓	4164	4171	4146	4177	4175		YM-OH3

Note: “↑”—increase in intensity; “↓”—decrease in intensity; v_as—asymmetric stretching_;^YM and ^ZM —the corner markers Y and Z represent Y and Z sites, respectively, and M represents the metal cations.

3.5. Raman Spectroscopy

Laser Raman spectra were consistent across samples (Figure 5a–d). For T2, sections perpendicular to the c-axis showed no significant pre- vs. post-heating differences, with bands at 223, 242, 376, 408, 730, 756, 837, 1063, 1395, 3498, 3564, 3593, and 3653 cm⁻¹(Figure 5a). Four main Raman band groups (centered at 220, 370, 730, and 1060 cm⁻¹ in 1000–1200 cm⁻¹) are assigned in

Table 7. Assignments of Raman bands in various tourmalines.

This Study	Elbaite [4,7]	Liddicoatite [7]	Uvite [7]	Assignment
⊥c-Axis Section
223(s)	222			YO6 vibrations
242(w)	244			O–YAl–O bond bending
376(s)	373	383		ZAl–O bond stretching
408(w)	407			Onon–Si–Onon bond bending
512(w)	508			Oxygen vibrations in Si–O rings
	637			Si–Obr bond rocking
	727			“Breathing” of Obr in SiO rings
730(vs)	731	734	728	B–O bond stretching and O–B–O bond bending
756(sh)	760			Si–O bond stretching and Si–O–Si bond bending
837(w)	840	839		Si–Onon bond stretching
1063(s)	1059	1068	1070	Si–Onon bond stretching
			1131	δMgOH
1395(w)	1400	1400	1400	B–O stretching mode
		3460		ZAlZAlYAl-OH3
3498(m)	3490	3500		Two Z-site cations and one Y-site cation link OH3: ZAlZAlYAl-OH3 or ZAlZAlYLi-OH3 in liddicoatite; ZAlZAlYMg-OH3 in uvite
3560(sh)	3560		3570
3593(m)	3585	3597
3653(w)	3655	3640	3728	Three Y-site cations link OH1: YAlYAlYLi-OH1 orYAlYLiYLi-OH1 in liddicoatite; YMgYMgYMg-OH1 in uvite

Note: m—middle; s—strong; vs—very strong; w—weak; sh—shoulder; O_br, bridging oxygens in Si–O rings, which link two SiO₄ tetrahedra; O_non, non-bridging oxygens in Si–O rings; ^YAl and ^ZAl—the corner markers Y and Z represent Y and Z sites, respectively; “⊥” means perpendicular.

. Band intensity and shifts were consistent, aligning with chemical composition and optical parameters.

Prior research links specific Raman shifts to tourmaline structural vibrations, with diagnostic bands for classification. The band at 223 cm⁻¹ is assigned to δ(Si-O-Si) bending in SiO₆ octahedra [7,26], a universal feature in all tourmalines. The band at 376 cm⁻¹ corresponds to ν(Z-O) stretching in Al³⁺-dominated ZO₆ octahedra and the sharp band with marked intensity at 376 cm⁻¹ in both the spectra of the samples implies strong bonding of Al–O [4]. The band at 408 cm⁻¹ arises from ν(O-Si-O) bending in SiO₄ tetrahedra, characteristic of Fe-rich tourmaline and potentially linked to Fe³⁺ substitution at Y sites [28]. Bands at 636, 708, and 730 cm⁻¹ are attributed to ν_s(B-O) stretching in BO₃ triangles, indicative of Li-bearing tourmaline, while 756 cm⁻¹ originates from ν(O-B-O) asymmetric stretching in BO₃ groups. The region 837–1087 cm⁻¹ results from ν(Si-O) stretching in SiO₄ tetrahedra, with band shifts attributed to variations in Si-O bond lengths [26].

The bands ranging from 3400 to 3700 cm⁻¹ are related to OH vibrations (Figure 5). In liddicoatite (calcic-lithium tourmaline), 3640 cm⁻¹ corresponds to Y-site Al/Li occupancy; in Fe-substituted elbaite, this shifts to 3655 cm⁻¹ [7]. In the paper, the observed band at 3652–3653 cm⁻¹ aligns with elbaite (3655 cm⁻¹), suggesting ^YAl^YAl^Y(Li, Mn, Fe)-OH1 or ^YAl^Y(Li, Mn, Fe)^Y(Li, Mn, Fe)-OH1configurations. For liddicoatite, 3597 and 3490 cm⁻¹ reflect Y-site differences between Al and Li, assigned to ^ZAl^ZAl^YAl-OH3 or ^ZAl^ZAl^YLi-OH3; elbaite shows 3585 and 3500 cm⁻¹ bands [7]. This study’s 3594 cm⁻¹ band is intermediate between liddicoatite (3597 cm⁻¹) and lithium tourmaline (3585 cm⁻¹). The 3489 cm⁻¹ band (parallel to c-axis) aligns with liddicoatite (3490 cm⁻¹), while 3498 cm⁻¹ (perpendicular to c-axis) matches elbaite (3500 cm⁻¹), reflecting heterogeneous Y-site cation occupancy (Al, Fe, Li, Mn) in ^ZAl^ZAl^Y(Li, Al, Mn, Fe)-OH3. The 3560 cm⁻¹ shoulder (diagnostic of X-site Na⁺; [7]) confirms classification as sodium-dominant elbaite.

3.6. UV-Vis Spectroscopy

UV-Vis absorption intensity showed polarization dependence: bands parallel to the c-axis were stronger and more distinct than those perpendicular to the c-axis (Figure 6). Unheated samples exhibited three visible-region bands (415, 540, and 730 nm), causing the brown color, plus two ultraviolet bands (379 and 395 nm) with no color impact. Heat-treated samples showed reduced overall absorption, with the 540 nm band absent. A green transmission window formed in the visible region, shifting color from brown to green (Figure 1).

Unheated brown samples (parallel to c-axis) had 730, 540, and 415 nm bands. Post-treatment, the 540 nm band vanished, while 730 and 415 nm bands persisted. Spectra perpendicular to the c-axis remained nearly unchanged, retaining the brown color. Thus, heat-treated samples showed strong pleochroism parallel to the c-axis.

4. Discussion

4.1. Tourmaline Species of Samples

Mostly brown tourmalines are primarily dravite and some are uvite, ferromanganese elbaite [12,29,30]. The composition of tourmaline measured in the paper is ferromanganese elbaite and enriched Mn- (MnO: 2.14–2.62 wt.%), Fe- (FeO: 1.03–1.46 wt.%), and Ti-bearing (TiO₂: 0.02–0.09 wt.%), which lacks V, Cr, Cu, and Ni, and is likely colored by Fe, Mn, Ti, etc. [14]. The molecular formula calculation shows that the sum of Y-site cations is less than 3, and the number of silicon atoms is greater than 6, which means the existence of some trivalent iron. Fe³⁺ occupies the Z-site, resulting in the overall small Y-site, which also confirms the existence of Fe³⁺. Raman spectra show 3653 and 3585 cm⁻¹ bands (elbaite markers) and a Li-related 730 cm⁻¹ band [4,7]. The absence of the 1131 cm⁻¹ band (exclusive to Fe-uvite and dravite; [7]) confirms the classification as Fe- and Mn-bearing elbaite–fluorelbaite, consistent with compositional analyses.

4.2. Spectral Response to 500 °C Heating and Structural Stability

Tourmaline’s complex composition leads to varied vibrational band displacements. IR spectra show subtle pre- vs. post-heating differences (Figure 2 and Figure 3), making distinction challenging. However, post-treatment, the 4721 cm⁻¹ band vanished (Figure 4) and the near-infrared absorption at 976 nm weakened (Figure 6). Raman spectra perpendicular to the c-axis were unchanged post-heating, confirming structural integrity. Invariant OH-stretching bands (3498, 3593, and 3653 cm⁻¹) indicate unaltered cation occupancy at crystallographic sites.

4.3. Color Origin and Color Change Mechanism Induced by Heat Treatment

Brown arises from absorption of blue-green and some red light, with yellow-orange light transmitted. These unheated samples have 415, 540, and 730 nm absorption bands, which are the main reasons for the brown color of the samples (Figure 6). After heat treatment, the 540 nm absorption band of the parallel c-axis section of the samples disappears, forming a transmission window near 540 nm, thus showing yellow-green. However, the absorption spectrum characteristics in the perpendicular c-axis section are basically unchanged, and the color is still brown. The sample shows yellow-green and brown dichroism (Figure 1b).

Based on previous studies and the characteristics of the samples in this paper, it is believed that the absorption bands at 730, 540, and 415 nm are caused by Fe²⁺ d–d transition (⁵T_2g→⁵E_g) [21,31,32,33,34,35], Fe²⁺→Fe³⁺ intervalence charge transfer (IVCT) [32,36,37], and Fe²⁺→Ti⁴⁺ (IVCT) [32,35,36,37,38], respectively. And for the Mn²⁺ spin-forbidden transition ⁶A_1g→⁴A_1g, ⁴T_Eg generated the 412/414 nm absorption band [38], superimposed onto the band 415 nm. Furthermore, the comprehensive review by Vigier et al. (2025) [39] emphasizes that Fe²⁺→Ti⁴⁺ (IVCT) is associated with a strong, broad absorption band in the visible range and is a common cause of brown coloration in minerals and glasses. This strongly supports our assignment of the 415 nm band to Fe²⁺→Ti⁴⁺ (IVCT), which is a significant contributor to the brown color of the untreated samples in this study, alongside the Fe²⁺→Fe³⁺ (IVCT) band at 540 nm.

Combined with previous studies, the blue tourmaline has 540 nm and 730 nm absorption bands [13,32,35,40,41]. The researchers found that the blue tourmaline generally does not contain Ti (<0.01% or below the detection limit). This means that the 540 nm absorption band has nothing to do with Ti. When the researchers heat-treated the blue tourmaline in the air, the absorption band at 540 nm was significantly enhanced (No.11 [40]; TB1-1, Faye et al., 1974 [32]), and the Fe²⁺/Fe³⁺ ratio changed from 24:1 to 2:1 (TB1-1). A reasonable explanation is that part of the Fe²⁺ was oxidized to the Fe³⁺ by heat treatment process, and the 540 nm absorption band was strengthened with the increase in Fe³⁺. Similarly, pink and yellow tourmaline are iron-free, or rather, below the detection limit [4,38,42]. The band of 520 nm in the pink tourmaline is related to the Mn, and the tail absorption 320 nm trending to ultraviolet region in the yellow tourmaline is due to Mn²⁺→Ti⁴⁺ intervalence charge transfer (IVCT) [43]. Therefore, the 520 and 320 nm bands are unrelated to Fe. Moreover, the iron-related absorption bands are missing in pink and yellow tourmalines, including the 540 nm band. In other words, the band at 540 nm is related to Fe.

The green tourmaline heat-treated in the air produced the 540 nm absorption band and the samples turned brown (No.3 [40]; TGr-2 [32]). The iron of the green tourmaline is dominated by Fe²⁺ [32,35]. This means that the Fe²⁺ cannot generate the absorption band alone. In general, Fe²⁺ can be oxidized to Fe³⁺ when the tourmaline is heat-treated in air (oxidizing atmosphere). In other words, the appearance of the 540 nm band is coupled with the appearance of Fe³⁺. Similarly, the green tourmaline produces a band that is centered at 550 nm after irradiation, which is caused by the Fe²⁺→Fe³⁺ electronic transition due to the oxidation of Fe²⁺ to Fe³⁺ by irradiation [13]. Additionally, the intensity of the infrared band at 4170 cm⁻¹, attributed to Fe³⁺, significantly decreases after heat treatment (Figure 6), indicating a reduction in its concentration. Therefore, we believe that the 540 nm absorption band is caused by Fe²⁺→Fe³⁺ (IVCT).

The brown samples in this study turned green after heat treatment in the reducing atmosphere, and the absorption band at 540 nm disappeared. This is the reverse process of the above-mentioned irradiation or heat treatment in oxidizing atmosphere, that is, the reduction of Fe³⁺ to Fe²⁺ leads to the disappearance of the Fe²⁺→Fe³⁺ electronic transition that can produce the 540 nm band. In addition, the samples in the paper have high Mn contents (17,346.71–20,669.52 μg/g), and the absorption near 415 nm may have a 412/414 nm superposition caused by the Mn²⁺ d–d (⁶A_1g→⁴A_1g, ⁴T_Eg) spin-forbidden transition [38]. Heat treatment reduces part of Fe³⁺ to Fe²⁺ and Ti⁴⁺ to Ti³⁺, causing Fe²⁺→Ti⁴⁺ to decrease and the absorption at 415 nm to weaken. However, some of Mn³⁺ was reduced to Mn²⁺, and the Mn²⁺ that leads to an increase in the absorption band at 412/414 nm increases, so the absorption band intensity around 415 nm has changes insignificantly. The 520 nm absorption related to Mn³⁺ [13,25] also disappeared at the same time. The existence of the 520 nm absorption band may also be the reason for the asymmetrical absorption band of the 540 nm absorption band in the unheated samples.

5. Conclusions

The brown tourmaline belongs to Fe- and Mn-bearing elbaite–fluorelbaite, at 500 °C under neutral-reducing conditions, where its brown hue fades and transparency improves. The weakening or disappearance of IR bands at 4721 cm^–1 serve as diagnostic indicators of heat treatment.

The UV-Vis spectra of the natural brown samples (parallel to c-axis) show 730, 540, and 415 nm bands, attributed to Fe²⁺ d–d transitions (⁵T_2g→⁵E_g), Fe²⁺→Fe³⁺ IVCT, and Fe²⁺→Ti⁴⁺ IVCT, respectively. The 415 nm band may also include Mn²⁺ d–d transitions (⁶A_1g→⁴A_1g, ⁴T_Eg) at 412/414 nm.

The reduction heat treatment leads to the transformation of Fe³⁺ to Fe²⁺, resulting in the disappearance of the 540 nm absorption band caused by Fe²⁺→Fe³⁺ IVCT in the parallel c-axis direction, forming a green light transmission window, which is the reason for the green color. The parallel c-axis section of the samples shows brown and yellow-green dichroism after heat treatment.