Vol.:(0123456789) Journal of Thermal Analysis and Calorimetry (2025) 150:9039–9052 https://doi.org/10.1007/s10973-025-14338-x Thermal decomposition of sol–gel synthesized bismuth citrate Darren Fynn1,2  · Caren Billing1,2  · David Gordon Billing1,2 Received: 22 November 2024 / Accepted: 22 April 2025 / Published online: 24 May 2025 © The Author(s) 2025 Abstract This paper investigates the thermal decomposition of bismuth citrate sol–gel precursors using thermogravimetry, derivative thermogravimetry and differential thermal analysis, as well as Fourier transform infrared spectroscopy and powder X-ray diffraction with Rietveld refinement analysis. Citric acid effectively complexes Bi3+ to form bismuth citrate monohydrate after drying at 100 °C. Upon further heating, dehydration and decomposition of excess citric acid occurs, followed by the decomposition of bismuth citrate to form a mixture of α-Bi2O3 and bismuth oxide acetate at ~ 300 °C. An intermediate bis- muth subcarbonate phase was detected upon further heating before final decomposition to Bi2O3. The subcarbonate phase has not been reported in bismuth citrate decomposition studies before. A partial phase transition from α-Bi2O3 to γ-Bi2O3 was also observed after heating to 700 °C and 750 °C. Keywords Sol–gel · Thermal decomposition · Rietveld refinement · Bismuth oxide (Bi2O3) · Bismuth citrate Introduction Bismuth oxide (Bi2O3) is a pale-yellow ceramic material with a range of interesting properties that make it a suitable candidate for various applications, including photocatalysis [1, 2], gas sensing [3, 4], and solid oxide fuel cells [5–8]. These properties are due to the unique polymorphism of Bi2O3. α-Bi2O3 (space group 14, P21/c) is thermodynami- cally stable at room temperature [9, 10] and is an effec- tive photocatalyst for the degradation of pollutants such as dyes, pesticides and toxins [11, 12]. α-Bi2O3 undergoes a phase transition to δ-Bi2O3 (space group 225, Fm-3m) at 729–730 °C. δ-Bi2O3 is thermodynamically stable at high- temperature but only in a narrow temperature range as it melts at 824–825 °C [5–7]. δ-Bi2O3 has the highest ionic conductivity amongst the solid oxide ion conductors, hence it has been the subject of extensive research for applica- tion as a solid oxide fuel cell electrolyte [5–8]. δ-Bi2O3 is unstable below 730 °C and transforms to β-Bi2O3 (space group 114, P-421c) at 650 °C or to γ-Bi2O3 (space group 197, I23) at 639 °C, depending on the cooling conditions. β- and γ-Bi2O3 then transform to α-Bi2O3 between 650 and 500 °C, although γ-Bi2O3 may often persist to room tem- perature [9, 10, 13]. Both β- and γ-Bi2O3 are more effective as photocatalysts than α-Bi2O3 [14–17]. Metal oxide powders (including Bi2O3) can be synthe- sized from solution-state precursors using a sol–gel method. Citric acid (C6H8O7) is a popular small organic molecule in sol–gel chemistry due to its affordability, availability, and effectiveness as a chelating agent. When citric acid is used, the synthesis is specifically referred to as the citric acid or citrate sol–gel method [18]. The citrate sol–gel method has been used for the synthesis of Bi2O3 [19–23]. However, there is a lack of consistency regarding the thermal decomposition products of the bismuth citrate precursor, possibly due to deviations in experimental conditions. Astuti et al. [22] synthesized Bi2O3 nano-powders using the citrate sol–gel method by incorporating polyethylene glycol (PEG) after chelating the bismuth cations with cit- rate. They initially found a mixture of α and γ phases, at room temperature, after annealing at temperatures of 500 °C, 600 °C, and 700 °C. However, they later found only α-Bi2O3 after annealing at 700 °C but after annealing at 500 °C and 600 °C they found mixtures of α-Bi2O3, γ-Bi2O3 as well as bismuth(II) monoxide (BiO) [23]. In contrast, Anilkumar et al. [21] found only β-Bi2O3 at room tempera- ture after annealing at 400 °C, while Mansour [24] found * Caren Billing caren.billing@wits.ac.za 1 Molecular Science Institute, School of Chemistry, University of the Witwatersrand, Johannesburg, South Africa 2 DSI-NRF Centre of Excellence in Strong Materials, University of the Witwatersrand, Johannesburg, South Africa http://orcid.org/0000-0003-1296-5232 http://orcid.org/0000-0001-5462-1150 http://orcid.org/0000-0001-8995-924X http://crossmark.crossref.org/dialog/?doi=10.1007/s10973-025-14338-x&domain=pdf 9040 D. Fynn et al. a mixture of α- and β-Bi2O3 phases after annealing at 470 °C. Although γ-Bi2O3 is known to persist to room temper- ature, the reported formation temperatures of both γ- and β-Bi2O3 phases seem to suggest direct phase transitions from α-Bi2O3, and the persistence of β-Bi2O3 to room temperature also contradicts previous reports regarding Bi2O3 polymor- phism [9, 10, 13]. These discrepancies warrant further inves- tigation into the transformation mechanisms and stability of Bi2O3 phases synthesized using the citrate sol–gel method. Notably, the above research primarily focuses on the final oxide phases, overlooking key information about the initial bismuth citrate (BiC6H5O7) complex precursor [19, 20]. BiC6H5O7 is a commonly used drug in the treat- ment of peptic ulcers due to its activity against Helico- bacter pylori [25, 26]. Several studies have investigated the thermal decomposition process of BiC6H5O7 [19, 20, 24], but there is variability in the reported temperature ranges and decomposition products formed. Notably, all these studies used a heating rate of 10 °C min−1 in air. Radecki and Wesołowski [19] suggested that dehydra- tion of BiC6H5O7·2H2O (dihydrate) begins at 80 °C and ends by 200 °C. Srivastava et al. [20], who synthesised BiC6H5O7 by precipitation with sodium citrate, observed dehydration of BiC6H5O7·H2O (monohydrate) occurring in a wider range from 39 to 225 °C. They further suggest partial decomposition of BiC6H5O7 by 332 °C, giving a mixture of Bi2O3 and bismuth oxide acetate (BiOC2H3O2, BiOAc), with a 1:3 (Bi2O3:BiOAc) molar ratio. Mansour [24], who analysed purchased BiC6H5O7, proposed that partial decomposition of anhydrous BiC6H5O7 gives a mix- ture of Bi2O3, BiOAc, and BiO, with a 1:1:1 molar ratio, by 325 °C. BiOAc then decomposes further to form Bi2O3 by 400 °C (according to Mansour [24]) or by 410 °C (accord- ing to Srivastava et al. [20]). According to Mansour [24], BiO undergoes oxidation to form Bi2O3 up to 850 °C, but this is above the melting point of Bi2O3 [6–9]. In contrast, Radecki and Wesołowski [19] reported largely a single-step decomposition of BiC6H5O7 to Bi2O3 by 470 °C. The contrasting findings in literature regarding the for- mation of Bi2O3 through the thermal decomposition of BiC6H5O7 warrants further investigation to provide a more comprehensive understanding of the process. In the present work, we address this by investigating the thermal decom- position of BiC6H5O7 precursors prepared using the citric acid sol–gel method. The thermal decomposition pathway of the BiC6H5O7 precursors is probed using differential thermal analysis (DTA), thermogravimetry (TG) and derivative thermogravimetry (DTG). The decomposition products are identified using Fourier transform infrared (FTIR) spectroscopy and powder X-ray diffraction (XRD) followed by quantitative Rietveld refinement [27] of the XRD patterns. Experimental Sample preparation All samples were synthesised using a citric acid sol–gel method. Bismuth (III) nitrate pentahydrate (Bi(NO3)3·5H2O, 99.99% pure, Sigma-Aldrich) was dis- solved in glacial acetic acid (98% pure, MK Chemical) under mild heat and constant stirring. Citric acid mono- hydrate (C6H8O7·H2O, ACS reagent, Sigma-Aldrich) was then added such that the molar ratio of Bi3+:citric acid was 1:1 (BiCit1), 1:2 (BiCit2), and 1:3 (BiCit3), respectively. The glacial acetic acid (boiling point: 118 °C [28]) was subsequently evaporated and the nitrates decomposed to form nitrogen dioxide (NO2), observed by its distinctive reddish-brown colour [29], and oxygen. The white gel that remained was oven-dried overnight at 100 °C and then at 200 °C, with analysis of the residues after each step. BiCit3 samples were also annealed further in a calibrated Lenton chamber furnace between 300 and 700 °C in 100 °C intervals for two hours (at each temperature), and then finally at 750 °C. A heating rate of 10 °C min−1 was used, and the samples were left to cool to room temperature in the furnace. Sample characterisation All FTIR and XRD measurements were performed under ambient conditions. FTIR spectra were measured using a Bruker ALPHA II FTIR spectrometer equipped with a ZnSe crystal as part of the Eco-ATR sampling module. FTIR spectra were measured from 600 to 4000 cm−1 at a resolution 4 cm−1 with 24 scans per sample, and back- ground noise was subtracted. Powder XRD patterns were measured using a Bruker AXS D2 PHASER desktop diffractometer. The instrument was equipped with a sealed tube Co Kα1/Kα2 X-ray source (30 kV, 10 mA) with λ = 1.78900/1.79283 Å, a 0.6 mm fixed divergent slit, 2.5° primary and secondary beam Söller slits, an iron filter to attenuate Kβ radiation, and a Lynxeye linear position sensitive detector with an effective angular range of 5° 2θ. The instrument was calibrated using NIST SRM 660c (LaB6, lanthanum hexaboride) and the instrumental resolu- tion function was determined using the extended Thompson- Cox-Hastings pseudo-Voigt (TCHZ) approach as outlined by Dinnebier et al. [30]. XRD patterns were measured between 5 to 100° 2θ using a step size of 0.026° 2θ. For all the XRD patterns presented here, the crystalline phase identifica- tion and quantification was determined using the Rietveld method [27] as implemented in the Bruker AXS TOPAS software (Version 7). 9041Thermal decomposition of sol–gel synthesized bismuth citrate The thermal decomposition of the samples oven-dried at 100 °C (BiCitn-100, where n = 1–3) as well as the BiCit3 sample dried at 200 °C (BiCit3-200) was investi- gated through simultaneous TG and DTA using a Perki- nElmer STA600 with a heating rate of 10 °C min−1 in an atmosphere of technical grade air (African Oxygen Lim- ited) flowing at a rate of 20 mL min−1. Thermal analysis data acquisition as well as data processing and analysis were performed using PerkinElmer's Pyris software. For the residual mass analysis of the TG curves, the molecular masses used in the theoretical calculations are according to the International Union of Pure and Applied Chemistry (IUPAC) [31]. Results and discussion Characterisation of bismuth citrate precursors using FTIR spectroscopy and XRD The ambient XRD pattern of citric acid (black), which was treated the same as the other samples by dissolving in acetic acid followed by oven drying at 100 °C, is shown in Fig. 1a. According to Lafontaine et al. [32] this represents the anhydrous form of citric acid. XRD patterns of BiCit1 (blue), BiCit2 (grey), and BiCit3 (red) oven-dried at 100 °C (Fig. 1a) are significantly different compared to that of citric acid, which supports the effective complexation of Bi3+ by citrate during the sample synthesis. For BiCit2-100 and BiCit3-100, peaks at ~ 16°, 21°, 23°, 28° and 30° 2θ show a noticeable growth in the intensity, indicated by ∇ in Fig. 1a. These peaks are not distinct in the XRD pattern of BiCit1- 100. Since the relative intensity of these peaks increase from BiCit2-100 to BiCit3-100, it is proposed that they corre- spond to an increase in excess uncomplexed citric acid with increased amount of citric acid added, which is consistent with the formation of bismuth citrate complexes with an overall Bi3+:citrate ratio of 1:1. Given the established ability of Bi3+ to form complexes with high coordination numbers (ranging from three to ten), coupled with the variability in its coordination geometries [33], it was important to confirm a Bi3+:citrate ratio of 1:1. The dominance of the bismuth citrate diffraction pattern even in the presence of two times more citric acid (BiCit3-100) is due to the significantly higher scattering factor of Bi3+ ions [34]. The samples oven-dried at 100 °C overnight were ini- tially more white in colour (Fig. 1b), but after oven-drying at 200 °C, they were brown in colour (Fig. 1c). Despite the noticeably different visual appearance of the samples, there is a high degree of similarity between the XRD patterns (Fig. 1a) for BiCit3-100 (red) and BiCit3-200 (orange), which suggests that the BiC6H5O7 crystal structure is main- tained after oven-drying at 200 °C. However, in the BiCit3- 200 XRD pattern the peaks at 23° and 30° 2θ were no longer Excess citric acid BiCit3-300 BiCit3-200 BiCit3-100 BiCit2-100 BiCit1-100 Citric acid 5 10 15 20 25 30 35 40 (a) (b) (c)In te ns ity /a rb u ni ts λθ Fig. 1 a XRD patterns of the BiC6H5O7 sol–gel precursors and citric acid. b BiCit3 after oven-drying at 100 °C overnight, and c BiCit3 after oven-drying at 200 °C overnight 9042 D. Fynn et al. observed, while the peaks at 16°, 21° and 28° 2θ decreased in intensity (more similar to BiCit1-100), which suggests the thermal decomposition of excess citric acid during oven- drying at 200 °C. In contrast, the XRD pattern of BiCit3-300 (green) is significantly different compared to BiCit3-100 and BiCit3-200, indicating the complete thermal decomposition of the BiC6H5O7 complex during heating at 300 °C for two hours. These samples were also analysed using FTIR spectros- copy and the room temperature spectra are shown in Fig. 2a. The spectrum for citric acid (black) has a sharp absorption band at ~ 3500  cm−1, along with a broader absorption band centred at ~ 3280  cm−1, which are due to O–H stretching vibrations within the carboxylic acid groups (indicated by * in Fig. 2a, b) [35–37]. A less intense absorption band observed at ~ 3450  cm−1 can be assigned to the O–H stretch- ing vibration of the tertiary alcohol group (indicated by + in Fig. 2a, b) [35]. The weak absorption bands in the region between 2900 and 3000  cm−1 correspond to C–H stretching vibrations of the methylene (CH2) groups (indicated by ^ in Fig. 2a, b). Two prominent absorption bands in the range of 1700–1800  cm−1 correspond to carbonyl (C=O) stretch- ing vibrations (indicated by ∇ in Fig. 2a, b) [35–37]. The absorption bands observed between 1200 and 1300  cm−1 correspond to C–O stretching vibrations (marked with ↓ in Fig. 2a, b) [35–37]. These assignments are summarised in Table 1. The FTIR spectra of both BiCit2-100 (grey) and BiCit3- 100 (red) (Fig. 2a) display the characteristic citric acid absorption bands for the carboxyl O–H stretching (*), C=O stretching (∇) and C–O stretching (↓), but these bands are absent from the spectrum for BiCit1-100 (blue). The absence of the characteristic carboxyl O–H stretching bands (*) in BiCit1-100 suggests the deprotonation of the carboxylic acid (a) (b) (c) Fig. 2 a FTIR spectra of the BiC6H5O7 sol–gel precursors and citric acid. Skeletal formulae of b anhydrous citric acid, and c BiC6H5O7, high- lighting the functional groups responsible for the stretching vibrations observed in the spectra 9043Thermal decomposition of sol–gel synthesized bismuth citrate groups when forming the BiC6H5O7 complex. Additionally, the lack of the characteristic C=O stretching absorption bands (∇) is due to the delocalisation of electrons across the two C–O bonds in the carboxylate groups when deproto- nated (represented by orange lines in Fig. 2c). When compl- exation between Bi3+ and the carboxylate groups occur, the symmetric and asymmetric CO2 stretching bands (typically around 1450–1360 cm−1 and 1650–1540 cm−1, respectively [37]) are more pronounced. Since these vibrations involve the carbon and both oxygens, they are referred to as “CO2” stretching vibrations as indicated in Fig. 2a (⇓). The persis- tence of the hydroxyl O–H stretching band due to the tertiary alcohol group (+) in the BiCit1-100 spectrum shows that this functional group does not participate in the complexa- tion of Bi3+, as indicated in Fig. 2c. The assignments for the spectrum for BiCit1-100 are summarised in Table 2. These results suggest that excess citric acid is present in BiCit2- 100 and BiCit3-100, but not BiCit1-100, which again is con- sistent with the formation of a complex with a Bi3+:citrate ratio of 1:1. The spectrum for BiCit3-200 (orange) is more compara- ble to that for BiCit1-100, showing that decomposition of excess citric acid occurs during oven-drying at 200 °C. The FTIR spectrum of BiCit3-300 (green) is significantly differ- ent compared to the spectra of all other samples analysed. This suggests the complete thermal decomposition of the citrate complex (BiC6H5O7) by 300 °C. These results all align with the conclusions drawn from the XRD analysis. Qualitative analysis of the BiCit3-300 XRD pattern (Fig. 3a) indicates the clear presence of α-Bi2O3 (P21/c, ICSD card no. 2374) [9], together with an additional phase (most likely a minor phase due to the lower intensity peaks). Previous studies suggested the formation of BiOAc during bismuth citrate thermal decomposition [20, 24], but the for- mation of BiOAc was only inferred using thermogravimetry. Unfortunately, the crystal structure for BiOAc has not been reported in established crystallographic databases, but pre- vious literature suggests BiOAc is isostructural with the bismuth oxide halide crystal structures (BiOCl - ICSD card no. 24608, BiOBr - ICSD card no. 24609 and BiOI - ICSD card no. 24610). These have P4/nmm space group symme- try, as well as similar Bi3+, O2− and halide crystallographic positions [38, 39]. The BiOAc crystal structure is described as also having P4/nmm space group symmetry with simi- lar Bi3+ and O2− crystallographic positions which form a Bi2O2 2+ layer, but with acetate replacing the halides which form a double interlayer (see Fig. S2.1 and corresponding crystallographic data in Table S2.1–2.3.) [38]. Assuming a P4/nmm space group for BiOAc, the addi- tional minor phase reflections were indexed (shown as orange * in Fig. 3a and are indexed in Table S1) and include (101) at 32.5° 2θ, (002) at 37° 2θ, (110) at 38° 2θ, (200) at 55° 2θ, and (103) at 64° 2θ. The most intense (101) BiOAc reflection overlaps with the (012) reflection of α-Bi2O3, causing a noticeable asymmetry in the combined peak pro- file. Rietveld refinements were done with respect to the diffraction patterns in Fig. 3a, b using the BiOCl structure (P4nmm space group) as the starting model. The refined c-axis length of BiOAc (~ 5.65 Å) is significantly smaller than that for the bismuth oxide halides (7.347 Å for BiOCl, 8.092 Å for BiOBr, and 9.128 Å for BiOI), while the refined length of the a-axis of BiOAc (~ 3.87 Å) and the bismuth oxide halides (3.883–3.984 Å) are well aligned. This sug- gests a preferential orientation of the approximately trigonal planar acetate ions along the c-axis direction [39]. Although the absence of a BiOAc crystal structure pre- vents accurate quantification using Rietveld refinement, alternative strategies were explored to estimate of the BiOAc concentration, as this may provide important insights into the thermal decomposition of bismuth citrate. The bismuth Table 1 Assignment of the characteristic absorption bands in the FTIR spectrum of citric acid in Fig. 2a Wavenumber range/cm−1 Relative intensity Assignment Symbol 3500 and 3280 Medium Carboxyl O–H stretch * 3450 Weak Hydroxyl O–H stretch + 2900–3000 Very Weak CH2 stretch ^ 1700–1800 Strong C=O stretch ∇ 1200–1300 Strong C–O stretch ↓ Table 2 Assignment of the characteristic absorption bands in the FTIR spectrum of BiCit1-100 in Fig. 2a Wavenumber range/cm−1 Relative intensity Assignment Symbol 3450 Weak Hydroxyl O–H stretch + 2900–3000 Very Weak CH2 stretch ^ 1650–1560 Strong Asymmetric CO2 stretch ⇓ 1540–1360 Strong Symmetric CO2 stretch ⋄ 9044 D. Fynn et al. oxide halides were thus used as proxies for BiOAc in the Rietveld refinements due to their reported structural similar- ity, and since a large contribution of the diffracted intensity is due to the significantly heavier Bi3+ ions [34, 38]. While this is not a commonly applied procedure, it highlighted the significant intensity difference between the major α-Bi2O3 and minor BiOAc phases (Fig. 3a), and produced a good fit for the minor phase, suggesting a reasonable approximation of the P4/nmm space group symmetry and the Bi3+ crystal- lographic positions. It should be noted that the different elec- tron counts for the anions (32 for acetate, 18 for Cl−, 36 for Br−, and 53 for I−) and their molar masses (59.0 for acetate, 35.5 for Cl−, 79.9 for Br−, and 126.9 for I−), among others, introduce error in the quantification by affecting the scatter- ing factor and mass % calculation. Consequently, the true concentration of BiOAc in BiCit3-300 should fall within the range defined by the concentrations estimated using BiOI (18 mass% at 300 °C) and BiOCl (31 mass% at 300 °C) as structural proxies. All the results of the quantitative phase analysis from the Rietveld refinement of BiCit3 (annealed at different temperatures) are summarized in Table 3. It is important to note that the approximate concentra- tions estimated from the Rietveld refinements using the bis- muth oxide halides as structural proxies are sensitive to the scattering power of the crystal structure model used. Spe- cifically, the use of the iodide ion, characterized by a much higher electron count (54 electrons) compared to the acetate ion (32 electrons) and consequently stronger X-ray scatter- ing, may result in an underestimation of the true BiOAc con- centration compared to the hypothetical use of an accurate BiOAc crystal structure, despite the fact that the majority of the diffracted intensity is contributed by the significantly heavier and electron-rich Bi3+ ions. Conversely, using the chloride ion, with a much lower electron count (18 electrons) compared to acetate and consequently weaker X-ray scat- tering, could lead to an overestimation of the true BiOAc concentration relative to the hypothetical scenario where the BiOAc crystal structure is available [30, 34]. Consequently, the true concentration of BiOAc in BiCit3-300 may fall within the range defined by the approximate concentrations estimated using BiOI (18 mass%) and BiOCl (31 mass%) as structural proxies. The results of the quantitative phase analysis from the Rietveld refinement of BiCit3 (annealed at different temperatures) are summarized in Table 3. The crystal structure of the bismuth oxide halides is shown in Fig. S2.1, with the corresponding crystallographic data in Tables S2.1–2.3. The XRD analysis therefore suggests the complete decomposition of BiC6H5O7 after annealing at 300 °C for 2 h. In contrast to the decomposition products reported by Mansour [24] (which included BiO alongside α-Bi2O3 and (2 00 )( 02 0) 300 °C Calculated _ Measured Difference _ Rwp = 7.06 _ α-Bi2O3 (P21/c) BiOAc (P4/nmm)* _ _ BiOAc* orth-Bi2O2CO3 (Imm2)+_ α-Bi2O3 _ orth-Bi2O2CO3+_ α-Bi2O3 _ ^ _ tetrag- Bi2O2CO3 (I4/mmm) Calculated _ Measured Difference _ Rwp = 5.79 Calculated _ Measured Difference _ Rwp = 7.47 400 °C 500 °C * * * * * * ** * * * * + * +*+ + + * + * * * + + + + * + * + (1 01 )( 01 2) (0 02 ) (1 10 ) (2 00 ) (1 03 ) (0 02 ) (1 01 )( 01 1) (1 03 )( 01 3) (1 10 ) (2 13 )( 12 3) (0 02 ) (1 01 ) (1 03 ) + + + + ^ ^ + ^ + ^ + + ^ + ^ ^ + ^ + + (1 10 ) # # λθ (a) (b) (c) Fig. 3 Rietveld refinement of the XRD patterns of BiCit3 annealed at different temperatures: a BiCit3-300, b BiCit3-400, and c BiCit3- 500. The measured patterns (blue) were refined using a mixture of phases, including α-Bi2O3 (grey), BiOAc# (*, orange), orthorhombic Bi2O2CO3 (+, green), and tetragonal Bi2O2CO3 (∨, magenta). #The bismuth oxide halides (BiOCl, BiOBr or BiOI) were used as proxies for BiOAc due to their structural similarity [38] 9045Thermal decomposition of sol–gel synthesized bismuth citrate BiOAc), no reflections due to BiO were observed in the XRD patterns presented throughout this study. While BiO is known to exist in the gas phase [28], its existence in the solid state is uncommon. The 2+ oxidation state of bismuth is extremely rare, making solid BiO an unusual, unstable and not a frequently reported oxide of bismuth [28, 40, 41]. The absence of BiO reflections in the XRD patterns is there- fore not surprising especially when annealing occurs in air. Therefore, based on the XRD analysis of BiCit3-300, the thermal decomposition of BiC6H5O7 results in a two-phase mixture consisting of α-Bi2O3 and BiOAc. This agrees with the thermal decomposition step reported by Srivastava et al. [20]. Rietveld refinement of the BiCit3-400 XRD pat- tern (Fig. 3b) revealed a three-phase mixture consisting of α-Bi2O3, orthorhombic bismuth subcarbonate (orth- Bi2O2CO3; Imm2, ICSD no. 94740) [42] (indicated by green + in Fig. 3b) as well as the bismuth oxide halide-type BiOAc phase. As before, the bismuth oxide halide crystal structures were used as proxies for BiOAc to estimate the concentrations of BiOAc from the quantitative phase analy- sis, which showed a notable decrease in the BiOAc con- centration compared to BiCit3-300 (Table 3). This, together with the emergence of orth-Bi2O2CO3 suggests that the thermal decomposition of BiOAc leads to the formation of orth-Bi2O2CO3. In the BiCit3-400 XRD pattern, the most intense peak associated with orth-Bi2O2CO3 is at 35° 2θ and is due to (103) and (013) reflections. Additional prominent peaks of orth-Bi2O2CO3 are observed at 15° 2θ for the (002) reflection, and at 28° 2θ due to (101) and (011) reflections. A complete list of bismuth oxide halide-type BiOAc and orth-Bi2O2CO3 reflections identified in the BiCit3-400 XRD pattern are given in Table S1. The refined crystal structure of orth-Bi2O2CO3 is shown in Fig. S3.1, with the crystal- lographic data in Table S3.1. Rietveld refinement of the BiCit3-500 XRD pat- tern (Fig. 3c) reveals the absence of the bismuth oxide halide-type BiOAc phase. The measured pattern can be described by a three-phase mixture consisting of α-Bi2O3, orth-Bi2O2CO3, and a newly formed tetragonal Bi2O2CO3 phase (tetrag-Bi2O2CO3; I4/mmm, ICSD no. 36245) [43] (indicated by magenta ∨ in Fig. 3c). The quantitative phase analysis results (Table 3) show a significant increase in the estimated concentration of Bi2O2CO3 in BiCit3-500 com- pared to BiCit3-400. Some of the increase in orth-Bi2O2CO3 will likely be due to the decomposition of the remaining BiOAc, however, this cannot account for the entire increase in concentration and for the emergence of tetrag-Bi2O2CO3. An additional source of carbon is necessary to explain the increase in Bi2O2CO3. Although the actual source of the additional carbon is not known at this stage, a potential source is likely from the excess uncomplexed citric acid. Citric acid does not fully decompose below 500 °C (see Fig. S6) but rather partially decomposes and forms intermedi- ates like aconitic acid (C6H6O6) or methyl maleic anhydride (C5H4O3) [44]. The CO2 released from the thermal decom- position of these intermediates may provide the additional carbon necessary for Bi2O2CO3 formation. This was shown to be possible in a hydrothermal synthesis where the decom- position of urea provided the CO2 necessary for Bi2O2CO3 formation from Bi2WO6 [19]. Additionally, in the thermal analysis of bismuth citrate, an amorphous coked residue was shown to form between 300 and 350 °C [45]. At higher tem- peratures this could possibly be oxidised to provide the CO2 needed for Bi2O2CO3 formation. orth-Bi2O2CO3 is likely more thermodynamically stable at room temperature due to its lower symmetry compared to tetrag-Bi2O2CO3. tetrag-Bi2O2CO3 is probably metastable and persists to room temperature, similar to the behaviour observed for γ-Bi2O3 [9, 10, 13] or the metastable aragonite polymorph of calcium carbonate (CaCO3) [46]. The most prominent reflection of tetrag-Bi2O2CO3, in the BiCit3- 500 XRD pattern, is the (103) reflection at 34.5° 2θ and an additional notable reflection is the (002) reflection at 14° 2θ. A complete list of tetrag-Bi2O2CO3 and orth-Bi2O2CO3 reflections identified in the BiCit3-500 XRD pattern is pro- vided in Table S1. An interesting observation is the visual similarity of the XRD patterns for tetrag-Bi2O2CO3 and Table 3 Quantitative phase analysis determined from the Rietveld refinement of XRD patterns for BiCit3 after annealing for two hours at each temperature # Bismuth oxide halides (aBiOCl, bBiOBr or cBiOI) were used as proxies for BiOAc due to their structural similarity [38] Annealing Phase/mass% Temperature/°C α-Bi2O3 BiOAc# orth-Bi2O2CO3 tetrag- Bi2O2CO3 γ-Bi2O3 300 69a, 74b, 82c 31a, 26b, 18c – – – 400 68a, 69b, 71c 10a,b, 7c 22a,c, 21b – – 500 58 – 27 15 – 600 100 – – – – 700 94 – – – 6 750 95 – – – 5 9046 D. Fynn et al. orth-Bi2O2CO3. This is most apparent when comparing the (103) reflection and the (002) reflection, which have similar positions and relative intensities in the two phases, although they are shifted towards lower 2θ angles for tetrag- Bi2O2CO3. This suggests a high degree of structural similar- ity between the two phases, and this is further supported by visually comparing the refined crystal structures of orth- Bi2O2CO3 (Fig. S2) and tetrag-Bi2O2CO3 (Fig. S3.2). The refined crystallographic data for tetrag-Bi2O2CO3 is given in Table S3.2. The Bi2O2CO3 phases (both orth and tetrag) are no longer observed in the XRD patterns for BiCit3 when annealed at 600 °C and above (Fig. 4), indicating thermal decomposi- tion of Bi2O2CO3 occurs between 500 and 600 °C. Rietveld refinement of the XRD pattern for BiCit3-600 (Fig. 4a) shows all reflections can be solely attributed to α-Bi2O3. The refined crystal structure of the resulting α-Bi2O3 phase is shown in Fig. S4, with the crystallographic data given in Table S4. The Rietveld refinement of the XRD patterns for BiCit3- 700 and BiCit3-750 (Fig. 4b, c) reveals α-Bi2O3 as the major phase, however, several additional peaks are also present. The most prominent of these is located at 32.75° 2θ, situ- ated between the (120) and (012) reflections of α-Bi2O3. This reflection can be assigned to the (301) and (310) reflec- tions of γ-Bi2O3 (I23, ICSD card no. 2376) [9] (indicated by sky blue ∇ in Fig. 4b). Another significant peak at 36° 2θ corresponds to the (222) reflection of γ-Bi2O3. Distin- guishing between β-Bi2O3 and γ-Bi2O3 solely based on the peaks at 32.75° 2θ and 36° 2θ is challenging, as these posi- tions could also correspond to the (201) and (002) reflec- tions of β-Bi2O3, respectively. Nevertheless, additional peaks at 46.75° 2θ and 51.5° 2θ are present in the XRD pattern. These peaks are absent for both α-Bi2O3 and β-Bi2O3 [9] but match the (402) and (420) reflections at 46.75° 2θ, and the (422) reflection at 51.5° 2θ for γ-Bi2O3, therefore confirming its formation. The results of the quantitative phase analyses from the Rietveld refinements of BiCit3-700 and BiCit3-750 are summarized in Table 3. Although γ-Bi2O3 is known to be metastable and could persist to room temperature, its presence in BiCit3-700 is notable, as previous reports suggest that the formation of γ-Bi2O3 proceeds through a δ-to-γ transition upon cooling [9, 10, 13]. Since a temperature of 700 °C is not sufficient for the α-to-δ phase transition in Bi2O3 to occur, the identifica- tion of some γ-Bi2O3 together with α-Bi2O3 for BiCit3-700, suggests a potential direct phase transition from α-Bi2O3 to γ-Bi2O3 at this lower temperature. Astuti et al. [22, 23] also found a mixture of α and γ phases at room temperature after annealing nanoparticles at 500 °C and 600 °C. However, this 600 °C 700 °C 750 °C R R R α α γ α γ λθ (a) (b) (c) Fig. 4 Rietveld refinement of the XRD patters of BiCit3 annealed at different temperatures: a BiCit3-600, b BiCit3-700, and c BiCit3-750. The measured patterns (blue) were refined using α-Bi2O3 (light grey) for BiCit3-600, and a mixture of α-Bi2O3 and γ-Bi2O3 (∇, sky blue) for BiCit3-700 and BiCit3-750 9047Thermal decomposition of sol–gel synthesized bismuth citrate incidental finding can only be verified by a comprehensive in-situ variable temperature study of the material which is not the scope of this work. BiCit3-750 has a similar phase composition as BiCit3-700 (Table 3), but it is annealed above the α-to-δ phase transition temperature so γ-Bi2O3 is likely formed through the traditional γ-to-δ phase transition. The γ-Bi2O3 peaks identified in the XRD patterns are indi- cated by sky blue ∇ in Fig. 4b, c and listed in Table S1, with the refined crystal structure of γ-Bi2O3 shown in Fig. S5. Thermal analysis of bismuth citrate precursors using DTA, TG and DTG Although the ambient characterization study provides valuable insights, particularly regarding the formation and identity of the decomposition products, it limits the direct determination of the decomposition step temperatures. Addi- tionally, the isothermal annealing steps can potentially limit the usefulness of the quantitative phase analysis to describe the actual decomposition pathway. The annealing steps may drive unintended or unwanted side reactions, which are kinetically limited, during annealing or cooling, and this can alter the final phase composition. Unlike the charac- terisation study discussed above, thermal analysis provides real-time insights into the thermal behaviour of these mate- rials. Through DTA, TG, and DTG, the temperatures of the decomposition steps can be directly determined. Addition- ally, through residual mass analysis of TG curves, in con- junction with the complementary ambient characterization, a more comprehensive understanding of the decomposition products, and consequently the decomposition pathway, can be developed. The XRD and FTIR spectroscopic analysis of BiCit3- 100 and BiCit3-200 suggested complete thermal decompo- sition of excess citric acid had occurred by 200 °C. This is also observed in the thermal analysis data. The DTA curve (Fig. 5a) of citric acid (black) shows a sharp endothermic peak at ~ 150 °C (inset of Fig. 5a) followed by a broader endothermic feature spanning from ~ 165 to 240 °C. This observation aligns well with previous reports for citric acid decomposition [44, 47]. Endothermic features with a simi- lar profile to that observed for pure citric acid are seen in the DTA curves of BiCit2-100 (grey) and BiCit3-100 (red) (Fig. 5a). Radecki and Wesołowski [19] and Srivastava et al. [20] previously attributed the endothermic features to a dehydration process. This would clearly not be the case for citric acid which was shown to be in the anhydrous form. Additionally, these endothermic features are only present for BiCit2-100 and BiCit3-100 which contain excess cit- ric acid, and not for BiCit1-100 (blue) and BiCit3-200 (orange) which do not contain excess citric acid (Fig. 5a). This supports the thermal decomposition of excess citric acid occurring between ~ 130 and 240 °C. The very broad peak (from ~ 165 to 240 °C) may point to this being a slow process which is why annealing at 200 °C for two hours was sufficient to decompose the excess citric acid. Furthermore, the DTG curves display a peak at ~ 220 °C for citric acid (black) and BiCit3-100 (red), and at ~ 170 °C for BiCit2- 100 (grey), denoted by (1) in Fig. 5b. Such a peak is absent for BiCit1-100 (blue) and BiCit3-200 (orange). With the area under the peak representing the mass loss during this thermal event, the significant area under the peak for BiCit2- 100 (13.8 mass% loss) and BiCit3-100 (31.5 mass% loss) thus represents the decomposition of citric acid rather than dehydration. The residual mass after the entire thermal cycle also pro- vided some insight into the extent of hydration of BiC6H5O7 after drying at 100 °C (Fig. 5c). The theoretical residual mass for the thermal decomposition of BiC6H5O7·2H2O (dihydrate) to form Bi2O3 is 53.7 mass%, whereas it is 56.0 mass% for BiC6H5O7·H2O (monohydrate) and 58.5 mass% for anhydrous BiC6H5O7. The observed residual mass of 55.3 mass% for BiCit1-100 (where no excess citric acid is present) thus aligns well with the theoretical residual mass for the monohydrate. Thus, given the theoretical residual mass for BiC6H5O7·H2O after complete dehydration is 98.9 mass%, it was observed that this residual mass is attained at ~ 150 °C on the TG curve for BiCit1-100, which suggests that dehydration of BiCit1 is complete by this temperature. Notably, the dehydration process is complete just before the onset of mass loss step (1) seen in DTG curves (Fig. 5b) and the onset of the endothermic features seen in the DTA curves between ~ 130 and 240 °C for citric acid, BiCit2-100 and BiCit3-100 (Fig. 5a). This again supports that mass loss step (1) seen for BiCit2-100 and BiCit3-100 corresponds to the thermal decomposition of excess anhydrous citric acid, as dehydration would have occurred before this mass loss step. The residual mass after cooling back to room tempera- ture as determined from the TG curves (Fig. 5c) reveals the trend where BiCit3-100 has the lowest residual mass (29.0 mass%), followed by BiCit2-100 (38.4 mass%) and BiCit- 100 (55.3 mass%), showing a progressive increase in total mass loss with increasing excess citric acid content. Hence the evidence collectively suggests that the endo- thermic features between ~ 130 and 240 °C seen in the DTA curves (Fig. 5a), as well as mass loss step (1) in the DTG and TG curves (Fig. 5b, c), corresponds to the thermal decomposition of excess citric acid (after the dehydration of BiC6H5O7·H2O), which is in agreement with the results from the ambient characterisation discussed above and occurs according to Eq. (1). (1) BiC6H5O7(s) + xC6H8O7(s) + 4.5xO2(g) → BiC6H5O7(s) + 6xCO2(g) + 4xH2O(g) 9048 D. Fynn et al. (a) (b) (c) m m m m Citric acid M as s lo ss r at e/ % m in –1 ∆T /° C Exo Endo Fig. 5 Thermal analysis, including a DTA, b DTG and c TG curves, of citric acid and BiC6H5O7 precursors with varying citric acid content. Theoretical residual masses were determined relative to four (4) moles of BiC6H5O7⋅H2O and are indicated by * 9049Thermal decomposition of sol–gel synthesized bismuth citrate where x = 0 for BiCit1-100, x = 1 for BiCit2-100 and x = 2 for BiCit3-100. Despite the observed differences in the thermal behaviour related to excess citric acid decomposition, the overall simi- larities in the DTA and DTG curves of all the samples above 250 °C (Fig. 5a, b) suggest a shared pathway for the main decomposition of BiC6H5O7. The DTA curves for BiCit1, BiCit2, and BiCit3-100 (Fig. 5a) all have two distinct exo- thermic peaks—a prominent peak at ~ 310–330°C, followed by a smaller peak at ~ 380 °C. These peaks are more over- lapped for BiCit3-200. This suggests a multi-step decom- position pathway for BiC6H5O7 in these samples, with at least two distinct thermal events. This observation contrasts with the single-step BiC6H5O7 decomposition reported by Radecki and Wesołowski [19]. The DTG curves for BiCit1, BiCit2, and BiCit3-100 (Fig.  5c) also show a high degree of similarity above 250 °C, advocating a shared pathway for the decomposi- tion of BiC6H5O7, irrespective of the initial presence or absence of excess citric acid. A significant mass loss step is observed in the (D)TG curves (denoted as (2) in Fig. 5b) for BiCit1, BiCit2 and BiCit3-100, as well as BiCit3-200, at ~ 300 °C which aligns with the onset temperature of the more prominent exothermic peak in the DTA curve (Fig. 5a). Mass loss step (2) is closely followed by a much less significant mass loss process in the (D)TG curves at ~ 325 °C (denoted by (3) in the inset of Fig. 5b, c), although it is not as clearly evident for BiCit3-200. Mass loss step (3) aligns well with the end of the more promi- nent exothermic peak in the DTA curve (Fig. 5a). The final mass loss step occurs at ~ 375 °C and is most noticeable in the (D)TG curves of BiCit1-100 and BiCit3-100 (denoted (4) in Fig. 5b). Mass loss step (4) aligns well with the smaller, higher temperature exothermic peak in the DTA curve. The presence of these distinct mass loss steps on the (D)TG curves suggest a four-step thermal decomposition mechanism for BiC6H5O7, which has not been reported before. From the TG curve of BiCit1-100 (Fig. 5c), the resid- ual mass is approximately 62.1 mass% at ~ 320 °C after mass loss step (2). Mansour [24] proposed a decomposi- tion product mixture of Bi2O3, BiOAc, and BiO in a 1:1:1 molar ratio. Theoretically, this decomposition pathway would result in a residual mass of 58.6 mass%, which is notably lower than the value observed. The ambient XRD study of BiCit3-300 also did not indicate the presence of BiO in the decomposition products. Srivastava et al. [20] proposed an alternative decomposition pathway for BiC6H5O7, leading to a mixture of Bi2O3 and BiOAc in a 1:3 molar ratio. While the ambient XRD study of BiCit3- 300 confirms the presence of a two-phase mixture of Bi2O3 and BiOAc, the 1:3 molar ratio suggested by Srivastava et al. [20] would result in a theoretical residual mass of 79.2 mass%, which is significantly higher than the 62.1 mass% observed for BiCit1-100. In this work, we pro- pose the decomposition pathway leading to the formation of a mixture of Bi2O3 and BiOAc (BiOC2H3O2) in a 1:2 molar ratio which would give a theoretical residual mass of 62.1 mass%, assuming BiC6H5O7·H2O corresponds to 100 mass%. This decomposition pathway is described by Eq. (2): The XRD study found ~ 7:3 mass ratio of Bi2O3 and BiOAc which translates to ~ 1.5:1 molar ratio (not 1:2) for BiCit3-300. This discrepancy is not unexpected since the sample was held at 300 °C for two hours allowing for sig- nificant conversion to the oxide, whereas for the thermal analysis BiCit1-100 was simply heated at 10 °C min−1 until 500 °C. The distinct mass loss step (3) has not been previously reported in the literature for the thermal decomposition of BiC6H5O7. At 375 °C, after mass loss step (3), the residual mass is 58.5 mass%. Although the decomposition pathway suggested by Mansour [24] (Bi2O3, BiOAc, and BiO in a 1:1:1 molar ratio) gives a residual mass close to this value (58.6 mass%), it was not considered since there is no evidence for the formation of BiO. The XRD study did, however, reveal the presence of Bi2O2CO3 after annealing BiCit3 at 400 and 500 °C for 2 h respectively. This observation suggests that mass loss step (3) is associated with the complete decomposition of BiOAc resulting in the formation of Bi2O2CO3, as shown in Eq. (3): The theoretical residual mass after the formation of Bi2O2CO3 (from the decomposition of BiOAc) is 58.6 mass% (relative to four (4) moles of BiC6H5O7·H2O) which is in excellent agreement with the experimental observation from the TG analysis (58.5 mass%). The final mass loss step (4) is then due to the thermal decomposition of Bi2O2CO3 leading to the formation of Bi2O3 as shown in Eq. (4): The overall thermal decomposition from room temperature is therefore described by Eq. (5): (2) 4BiC6H5O7(s) + 14 O2(g) → Bi2O3(s) + 2BiOC2H3O2(s) + 20 CO2(g) + 7 H2O (g) (3) 2BiOC2H3O2(s) + 4 O2(g) → Bi2O2CO3(s) + 3 CO2(g) + 3 H2O (g) (4)Bi2O2CO3(s) → Bi2O3(s) + CO2(g) (5) 4BiC6H5O7 ⋅ H2O(s) + 18 O2(g) → 2Bi2O3(s) + 24 CO2(g) + 11 H2O (g) 9050 D. Fynn et al. Conclusions Literature presents contrasting findings regarding the thermal decomposition of BiC6H5O7 to form Bi2O3. This study inves- tigated the thermal decomposition of BiC6H5O7 sol–gel pre- cursors, synthesized using the citrate sol–gel method. Samples dried and annealed at different temperatures were analysed using ambient FTIR spectroscopy and XRD with Rietveld refinement, as well as the thermal analysis techniques TG, DTG and DTA. Our results confirm the complexation of Bi3+ by citric acid to form BiC6H5O7·H2O after drying at 100 °C. The initial step upon heating includes: 1. Dehydration (by ~ 150  °C) followed by the thermal decomposition of excess citric acid (~ 130–240 °C): BiC6H5O7 then decomposes to form Bi2O3 via the following steps (although O2, CO2 and H2O have been omitted here for simplicity, the full balanced equations are presented in Eqs. 1–5): 2. Thermal decomposition of BiC6H5O7 (~ 300 °C) accord- ing to the stoichiometric ratio: We also observed a potential direct BiOAc → Bi2O3 decomposition reaction occurring during isothermal annealing at 300 °C according to the XRD results. The bismuth oxide acetate (BiOAc) then decomposed to form Bi2O3 via an intermediate subcarbonate phase: 3. Thermal decomposition of BiOAc (~ 325 °C): 4. Thermal decomposition of Bi2O2CO3 (~ 375 °C): Steps (3) and (4), which involve the formation and decomposition of Bi2O2CO3, confirmed by XRD and ther- mal analysis results, have not been previously reported. XRD results show that initially orth-Bi2O2CO3 is formed, but tetrag-Bi2O2CO3 is also formed after annealing at higher temperatures. It is possible that the exact synthetic route may influence the decomposition pathway of BiC6H5O7 as it appears that residual carbon may be involved in Bi2O2CO3 formation. The identification of this additional intermedi- ate step in the phase transition to form Bi2O3 highlights the need for further investigation into the mechanism by which it occurs. BiC6H5O7 ⋅ H2O + xC6H5O7 → BiC6H5O7 + xC6H5O7 → BiC6H5O7 4 BiC6H5O7 → Bi2O3 + 2 BiOAc 2 BiOAc → Bi2O2CO3 Bi2O2CO3 → Bi2O3 Supplementary Information The online version contains supplemen- tary material available at https:// doi. org/ 10. 1007/ s10973- 025- 14338-x. Acknowledgements The authors hereby acknowledge the support of the National Research Foundation (NRF, South Africa, Grant No: 141966) and Department of Science and Innovation (DSI, South Africa, Grant UID: 41292) through the Centre of Excellence in Strong Materials (CoE-SM), as well as the University of the Witwatersrand towards this research. Opinions expressed and conclusions arrived at, are those of the authors and are not necessarily to be attributed to the NRF, DSI-NRF CoE-SM and/or the University of the Witwatersrand. Author contributions DF: Conceptualization, Formal analysis and investigation, Writing—original draft preparation. CB: Conceptual- ization, Writing—review and editing, Supervision, Resources, Funding acquisition. DB: Conceptualization, Supervision, Resources, Funding acquisition. Funding Open access funding provided by University of the Witwatersrand. Data availability The data that support the findings of this study are available from the corresponding author, Caren Billing, upon reason- able request. Declarations Conflict of interest All authors declare no conflict of interest, finan- cial or otherwise, related to the research and findings presented in this manuscript. Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. References 1. Reverberi AP, Varbanov PS, Vocciante M, Fabiano B. Bismuth oxide-related photocatalysts in green nanotechnology: a critical analysis. Front Chem Sci Eng. 2018;12(4):878–92. https:// doi. org/ 10. 1007/ s11705- 018- 1744-5. 2. Mane V, Dake D, Raskar N, Sonpir R, Stathatos E, Dole B. A review on Bi2O3 nanomaterial for photocatalytic and antibacterial applications. Chem Phys Impact. 2024;8: 100517. https:// doi. org/ 10. 1016/j. chphi. 2024. 100517. 3. Ghuge RS, Shinde MD, Rane SB. Bismuth-based gas sensors: a comprehensive review. J Electron Mater. 2021;50(11):6060–72. https:// doi. org/ 10. 1007/ s11664- 021- 09174-2. 4. Ge L, Mu X, Tian G, Huang Q, Ahmed J, Hu Z. Current applica- tions of gas sensor based on 2-D nanomaterial: a mini review. Front Chem. 2019;7:839. https:// doi. org/ 10. 3389/ fchem. 2019. 00839. https://doi.org/10.1007/s10973-025-14338-x http://creativecommons.org/licenses/by/4.0/ https://doi.org/10.1007/s11705-018-1744-5 https://doi.org/10.1007/s11705-018-1744-5 https://doi.org/10.1016/j.chphi.2024.100517 https://doi.org/10.1016/j.chphi.2024.100517 https://doi.org/10.1007/s11664-021-09174-2 https://doi.org/10.3389/fchem.2019.00839 https://doi.org/10.3389/fchem.2019.00839 9051Thermal decomposition of sol–gel synthesized bismuth citrate 5. Azad AM, Larose S, Akbar SA. Bismuth oxide-based solid elec- trolytes for fuel cells. J Mater Sci. 1994;29(16):4135–51. https:// doi. org/ 10. 1007/ BF004 14192. 6. Shuk P, Wiemhöfer HD, Guth U, Göpel W, Greenblatt M. Oxide ion conducting solid electrolytes based on Bi2O3. Solid State Ion. 1996;89:179–96. https:// doi. org/ 10. 1016/ 0167- 2738(96) 00348-7. 7. Sammes NM, Tompsett GA, Näfe H, Aldinger F. Bismuth based oxide electrolytes—structure and ionic conductivity. J Eur Ceram Soc. 1999;19(10):1801–26. https:// doi. org/ 10. 1016/ S0955- 2219(99) 00009-6. 8. Drache M, Roussel P, Wignacourt JP. Structures and oxide mobility in Bi-Ln-O materials: heritage of Bi2O3. Chem Rev. 2007;107(1):80–96. https:// doi. org/ 10. 1021/ cr050 977s. 9. Harwig HA. On the structure of bismuthsesquioxide: the α, β, γ, and δ-phase. Z Anorg Allg Chem. 1978;444:151–66. https:// doi. org/ 10. 1002/ zaac. 19784 440118. 10. Levin EM, Roth RS. Polymorphism of bismuth sesquiox- ide. I. Pure Bi2O3. J Res Natl Bur Stand Sect Phys Chem. 1964;68(2):189–95. https:// doi. org/ 10. 6028/ jres. 068A. 019. 11. Iyyapushpam S, Nishanthi ST, Pathinettam PD. Photocatalytic degradation of methyl orange using α-Bi2O3 prepared without surfactant. J Alloys Compd. 2013;563:104–7. https:// doi. org/ 10. 1016/j. jallc om. 2013. 02. 107. 12. Sood S, Umar A, Kumar Mehta S, Kumar KS. α-Bi2O3 nanorods: an efficient sunlight active photocatalyst for degradation of rhoda- mine B and 2,4,6-trichlorophenol. Ceram Int. 2015;41(3):3355– 64. https:// doi. org/ 10. 1016/j. ceram int. 2014. 10. 038. 13. Blower SK, Greaves C. The structure of β-Bi2O3 from powder neutron diffraction data. Acta Crystallogr C. 1988;44:587–9. https:// doi. org/ 10. 1107/ S0108 27018 70116 61. 14. Zahid AH, Han Q. A review on the preparation, microstruc- ture, and photocatalytic performance of Bi2O3 in polymorphs. Nanoscale. 2021;13(42):17687–724. https:// doi. org/ 10. 1039/ D1NR0 3187B. 15. Jalalah M, Faisal M, Bouzid H, Park J-G, Al-Sayari SA, Ismail AA. Comparative study on photocatalytic performances of crys- talline α- and β-Bi2O3 nanoparticles under visible light. J Ind Eng Chem. 2015;30:183–9. https:// doi. org/ 10. 1016/j. jiec. 2015. 05. 020. 16. Selvamani T, Anandan S, Granone L, Bahnemann DW, Ashok- kumar M. Phase-controlled synthesis of bismuth oxide poly- morphs for photocatalytic applications. Mater Chem Front. 2018;2(9):1664–73. https:// doi. org/ 10. 1039/ C8QM0 0221E. 17. Iyyapushpam S, Nishanthi ST, Pathinettam PD. Enhanced pho- tocatalytic degradation of methyl orange by gamma Bi2O3 and its kinetics. J Alloys Compd. 2014;601:85–7. https:// doi. org/ 10. 1016/j. jallc om. 2014. 02. 142. 18. Danks AE, Hall SR, Schnepp Z. The evolution of “sol–gel” chemistry as a technique for materials synthesis. Mater Horiz. 2016;3(2):91–112. https:// doi. org/ 10. 1039/ c5mh0 0260e. 19. Radecki A, Wesołowski M. The thermal decomposition of bismuth(III) compounds used in medicine. Thermochim Acta. 1976;17:217–29. https:// doi. org/ 10. 1016/ 0040- 6031(76) 85028-9. 20. Srivastava A, Gunjikar VG, Sinha APB. Thermoanalytical studies of zinc citrate, bismuth citrate and calcium citrate. Thermochim Acta. 1987;117:201–17. https:// doi. org/ 10. 1016/ 0040- 6031(87) 88115-7. 21. Anilkumar M, Pasricha R, Ravi V. Synthesis of bismuth oxide nanoparticles by citrate gel method. Ceram Int. 2005;31(6):889– 91. https:// doi. org/ 10. 1016/j. ceram int. 2004. 09. 002. 22. Astuti Y, Listyani BM, Suyati L, Darmawan A. Bismuth oxide prepared by sol–gel method: variation of physicochemical charac- teristics and photocatalytic activity due to difference in calcination temperature. Indones J Chem. 2021;21(1):108–17. https:// doi. org/ 10. 22146/ ijc. 53144. 23. Astuti Y, Nurhayati S, Arnelli. Effect of calcination temperature on the morphology and photocatalytic activity of bismuth oxide. AIP Conf Proc. 2022. https:// doi. org/ 10. 1063/5. 01041 24. 24. Mansour SAA. Thermal decomposition of anhydrous bismuth citrate. Thermochim Acta. 1994;233(2):257–68. https:// doi. org/ 10. 1016/ 0040- 6031(94) 85119-0. 25. Yang N, Sun H. Biocoordination chemistry of bismuth: recent advances. Coord Chem Rev. 2007;251(17–20):2354–66. https:// doi. org/ 10. 1016/j. ccr. 2007. 03. 003. 26. Ge R, Chen Z, Zhou Q. The actions of bismuth in the treat- ment of Helicobacter pylori infections: an update. Metallomics. 2012;4(3):239. https:// doi. org/ 10. 1039/ c2mt0 0180b. 27. Rietveld HM. A profile refinement method for nuclear and mag- netic structures. J Appl Crystallogr. 1969;2(2):65–71. https:// doi. org/ 10. 1107/ S0021 88986 90065 58. 28. Haynes WM, Lido DR, Bruno TJ. CRC handbook of chemistry and physics: a ready-reference book of chemical and physical data. 95th ed. Boca Raton: CRC Press; 2014. 29. Gill WE. Determination of NO2 and NO in air. Am Ind Hyg Assoc J. 1960;21(1):87–96. https:// doi. org/ 10. 1080/ 00028 89600 93433 13. 30. Dinnebier RE, Leineweber A, Evans JSO. Rietveld refinement: practical powder diffraction pattern analysis using TOPAS. Berlin: De Gruyter; 2019. 31. Wieser ME, Holden N, Coplen TB, Böhlke JK, Berglund M, Brand WA, De Bièvre P, Gröning M, Loss RD, Meija J, Hirata T, Prohaska T, Schoenberg R, O’Connor G, Walczyk T, Yoneda S, Zhu X-K. Atomic weights of the elements 2011 (IUPAC technical report). Pure Appl Chem. 2013;85(5):1047–78. https:// doi. org/ 10. 1351/ PAC- REP- 13- 03- 02. 32. Lafontaine A, Sanselme M, Cartigny Y, Cardinael P, Coquerel G. Characterization of the transition between the monohy- drate and the anhydrous citric acid. J Therm Anal Calorim. 2013;112(1):307–15. https:// doi. org/ 10. 1007/ s10973- 012- 2798-0. 33. Sadler PJ, Li H, Sun H. Coordination chemistry of metals in medicine: target sites for bismuth. Coord Chem Rev. 1999;185– 186:689–709. https:// doi. org/ 10. 1016/ S0010- 8545(99) 00018-1. 34. Brown PJ, Fox AG, Maslen EN, O’Keefe MA, Willis BTM. Inten- sity of diffracted intensities. In: Fuess H, Hahn T, Wondratschek H, Müller U, Shmueli U, Prince E, Authier A, Kopský V, Litvin DB, Rossmann MG, Arnold E, Hall S, McMahon B, editors. Inter- national tables for crystallography, vol. C. Chester: International Union of Crystallography; 2006. p. 554–95. https:// doi. org/ 10. 1107/ 97809 55360 20600 00600. 35. Poerwono H, Higashiyama K, Kubo H, Poernomo AT, Suhar- jono I, Sudiana IK, Indrayanto G, Brittain HG. Citric acid. Anal Profiles Drug Subst Excip. 2001;28:1–76. https:// doi. org/ 10. 1016/ S1075- 6280(01) 28002-1. 36. Stuart BH. Infrared spectroscopy: fundamentals and applications; analytical techniques in the sciences. Hoboken: Wiley; 2004. 37. Smith BC. Infrared spectral interpretation: a systematic approach. Boca Raton: CRC Press; 1999. 38. Aurivillius B, Gloor U. X-ray studies on bismuth oxide acetate CH3COO⋅OBi and related compounds. Acta Chem Scand. 1955;9:1213–8. 39. Sillen L. X-ray studies on BiOCl, BiOBr and BiOI. Sven Kem Tidsk. 1941;53:39–43. 40. Shannon RD, Prewitt CT. Effective ionic radii in oxides and fluo- rides. Acta Crystallogr B. 1969;25(5):925–46. https:// doi. org/ 10. 1107/ s0567 74086 90032 20. https://doi.org/10.1007/BF00414192 https://doi.org/10.1007/BF00414192 https://doi.org/10.1016/0167-2738(96)00348-7 https://doi.org/10.1016/S0955-2219(99)00009-6 https://doi.org/10.1016/S0955-2219(99)00009-6 https://doi.org/10.1021/cr050977s https://doi.org/10.1002/zaac.19784440118 https://doi.org/10.1002/zaac.19784440118 https://doi.org/10.6028/jres.068A.019 https://doi.org/10.1016/j.jallcom.2013.02.107 https://doi.org/10.1016/j.jallcom.2013.02.107 https://doi.org/10.1016/j.ceramint.2014.10.038 https://doi.org/10.1107/S0108270187011661 https://doi.org/10.1039/D1NR03187B https://doi.org/10.1039/D1NR03187B https://doi.org/10.1016/j.jiec.2015.05.020 https://doi.org/10.1016/j.jiec.2015.05.020 https://doi.org/10.1039/C8QM00221E https://doi.org/10.1016/j.jallcom.2014.02.142 https://doi.org/10.1016/j.jallcom.2014.02.142 https://doi.org/10.1039/c5mh00260e https://doi.org/10.1016/0040-6031(76)85028-9 https://doi.org/10.1016/0040-6031(76)85028-9 https://doi.org/10.1016/0040-6031(87)88115-7 https://doi.org/10.1016/0040-6031(87)88115-7 https://doi.org/10.1016/j.ceramint.2004.09.002 https://doi.org/10.22146/ijc.53144 https://doi.org/10.22146/ijc.53144 https://doi.org/10.1063/5.0104124 https://doi.org/10.1016/0040-6031(94)85119-0 https://doi.org/10.1016/0040-6031(94)85119-0 https://doi.org/10.1016/j.ccr.2007.03.003 https://doi.org/10.1016/j.ccr.2007.03.003 https://doi.org/10.1039/c2mt00180b https://doi.org/10.1107/S0021889869006558 https://doi.org/10.1107/S0021889869006558 https://doi.org/10.1080/00028896009343313 https://doi.org/10.1080/00028896009343313 https://doi.org/10.1351/PAC-REP-13-03-02 https://doi.org/10.1351/PAC-REP-13-03-02 https://doi.org/10.1007/s10973-012-2798-0 https://doi.org/10.1016/S0010-8545(99)00018-1 https://doi.org/10.1107/97809553602060000600 https://doi.org/10.1107/97809553602060000600 https://doi.org/10.1016/S1075-6280(01)28002-1 https://doi.org/10.1016/S1075-6280(01)28002-1 https://doi.org/10.1107/s0567740869003220 https://doi.org/10.1107/s0567740869003220 9052 D. Fynn et al. 41. Shannon RD, Prewitt CT. Revised values of effective ionic radii. Acta Crystallogr B. 1970;26(7):1046–8. https:// doi. org/ 10. 1107/ s0567 74087 00035 76. 42. Grice JD. A solution to the crystal structures of bismutite and bey- erite. Can Miner. 2002;40:693–8. https:// doi. org/ 10. 2113/ gscan min. 40.2. 693. 43. Lagercrantz A, Sillen LG. On the crystal structure of Bi2O2CO3 (bismutite) and CaBi2O2(CO3)2 (beyerite). Stockholm: Almqvist & Wiksell; 1948. 44. Barbooti MM, Al-Sammerrai DA. Thermal decomposition of cit- ric acid. Thermochim Acta. 1986;98:119–26. https:// doi. org/ 10. 1016/ 0040- 6031(86) 87081-2. 45. Li H, Jiang P, Zhang J, Luo X, Long Z. An one-pot approach to hierarchical flower-like Bi2O2CO3/Bi2WO6 heterojunctions: self-assembly, in situ carbonization, and enhanced visible light photocatalytic activities. Mater Express. 2021;11(4):533–9. https:// doi. org/ 10. 1166/ mex. 2021. 1934. 46. Carlson WD. The polymorphs of CaCO3 and the aragonite-calcite transformation. Rev Miner. 1983;11:191–225. 47. Wiecinska P. Thermal degradation of organic additives used in colloidal shaping of ceramics investigated by the coupled DTA/ TG/MS analysis. J Therm Anal Calorim. 2016;123:1419–30. https:// doi. org/ 10. 1007/ s10973- 015- 5075-1. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. https://doi.org/10.1107/s0567740870003576 https://doi.org/10.1107/s0567740870003576 https://doi.org/10.2113/gscanmin.40.2.693 https://doi.org/10.2113/gscanmin.40.2.693 https://doi.org/10.1016/0040-6031(86)87081-2 https://doi.org/10.1016/0040-6031(86)87081-2 https://doi.org/10.1166/mex.2021.1934 https://doi.org/10.1007/s10973-015-5075-1 Thermal decomposition of sol–gel synthesized bismuth citrate Abstract Introduction Experimental Sample preparation Sample characterisation Results and discussion Characterisation of bismuth citrate precursors using FTIR spectroscopy and XRD Thermal analysis of bismuth citrate precursors using DTA, TG and DTG Conclusions Acknowledgements References