Investigating structural aspects to understand the putative/claimed non-toxicity of the Hg-based Ayurvedic drug Rasasindura using XAFS

XANES- and EXAFS-based analysis of the Ayurvedic Hg-based nano-drug Rasasindura has been performed to seek evidence of its non-toxicity. Rasasindura is determined to be composed of single-phase α-HgS nanoparticles (size ∼24 nm), free of Hg⁰ or organic molecules; its structure is determined to be robust (<3% defects). The non-existence of Hg⁰ implies the absence of Hg-based toxicity and establishes that chemical form, rather than content of heavy metals, is the correct parameter for evaluating the toxicity in these drugs. The stable α-HgS form (strong Hg—S covalent bond and robust particle character) ensures the integrity of the drug during delivery and prevention of its reduction to Hg⁰ within the human body. Further, these comparative studies establish that structural parameters (size dispersion, coordination configuration) are better controlled in Rasasindura. This places the Ayurvedic synthesis method on par with contemporary techniques of nanoparticle synthesis.

1. Introduction

The discipline of Indian Ayurvedic medicine is distinctive in that it recognizes the therapeutic potency of heavy metals or their compounds when (i) mixed with organic molecules (herbs) and (ii) processed in a prescribed route employing natural products (Raha, 2013; Acharya et al., 2014¹). The advantages of Ayurveda over allopathic treatment include (i) cost-effectiveness (due to natural resources), (ii) longer shelf-life (due to the presence of metals) and (iii) minimized adverse side effects (Kumar & Gupta, 2011). In India, Ayurvedic medicine assumes crucial importance in bridging the gap between heavy patient load (rural/poor) and medical accessibility (Nandha & Singh, 2013). The scope of Ayurveda in India is aided by the availability of abundant medicinal plants (Raut et al., 2013; Panghal et al., 2010). Recognizing these facts, the Government of India (GOI) has taken keen interest in regularizing Ayurveda as a widespread alternative healthcare route. To practically realise this, the GOI has actively invested in Ayurvedic research at several institutes² with the objective of finding scientific evidence of the non-toxicity, standardization and effectiveness of Ayurvedic medicines (Valiathan, 2006; Bose & Saroch, 2012).

An objection raised against metal-derived drugs of Ayurveda (Patwabardhan, 2011; Baghel, 2013; Saper et al., 2004; Leonti & Casu, 2013; Nishteswar, 2013) is the lack of scientific evidence for its claimed non-toxicity (Sarkar et al., 2010; Kamath et al., 2012; Panda & Hazra, 2012). Some reports suggest that toxicity is neutralized by compound formation (Tabakova et al., 2006). However, their synthesis methods do not guarantee (nor experimentally prove) total elimination of the metallic phases or other toxic forms. In the absence of scientific validation, the present criterion of toxicity (or consequent ban) is defined by the metallic content (and not the chemical or structural forms) (Singh et al., 2011).

The general Ayurvedic synthesis method (`Bhasmikaran') (Kumar et al., 2006; Wadekar et al., 2005) includes:

(i) Ingredients (metal + organic material). The advantages of metal (or its compounds) are manifold: longer shelf-life, small attainable sizes, quick drug release, lower required dosage, and size-controlled tunability of surface plasmon resonance for targeted drug delivery (Liao et al., 2006).

(ii) Heating. The objective of heating is the formation of metal salts etc. The toxic reaction of elemental metals (M ⁰) within the body (C, H, O) proceeds by oxidation . This is curtailed by pre-forming stable [M ⁺B ^-] compounds before being administered into the body (Dubey et al., 2009; Chaudhary, 2011; Mukherjee et al., 2010³). Accomplishing 100% oxidation (zero M⁰ remaining) is the most crucial step for non-toxicity of the drug.

(iii) Repeated purification. Impurities (e.g. toxic organic forms)/residual metals are removed by controlled heating for a prolonged period of time (Krishnamachary et al., 2012).

(iv) Grinding. Reducing the particle size such that the end-product (`Bhasma') has a very fine texture (Bhasma) and no metallic shine (Patel, 1986; Shastry, 1979; Svoboda, 1998).

In modern perspectives, this process could be equivalent to the formation of metal–oxide (or compound) nanoparticles (Chakrapany & Singh, 2014; Sarkar & Chaudhury, 2010; Kulkarni et al., 2013; Rastogi, 2010; Adhikari & Thapa, 2014; Paur et al., 2011), that could act as drug-carriers in targeted drug delivery (Yonezawa et al., 2000; De Jong & Barm, 2008).

As mentioned earlier, complete oxidation/removal of the toxic organic form and quality control of the `Bhasma' particles have to be experimentally validated for Ayurvedic bhasmas to be credible and acceptable for use as drugs. Previous research includes reports on elemental analysis (Singh et al., 2009), toxicology studies (Vardhini et al., 2010), heavy metal bioaccessibility tests (Koch et al., 2013) and X-ray diffraction (XRD) (Kamath et al., 2012; Singh et al., 2009). However, none of them could unambiguously establish complete oxidation status or account for non-toxicity. In this work, we propose to accomplish this by X-ray absorption fine-structure (XAFS) investigation of the structure and chemical form of α-HgS-based Rasasindura. The latter is widely prescribed (<125 mg per day) for treatment of certain diseases (Varier, 1999; Gokarn et al., 2012; Patgiri & Gokarn, 2014; Mahdihassan, 1987; Ingole, 2013; The Ayurvedic Formulatory of India, 2003). Recent in vivo experiments (Anita et al., 2013; Dwivedi et al., 2012, 2013, 2014; Kanojia et al., 2013) have established the therapeutic effect of Rasasindura and ruled out toxicity. We clarify that our XAFS understanding/validation of non-toxicity will cover only the virgin medicine, i.e. before being administered into body. How it reacts within the body or remains non-toxic within the body is beyond the scope of this work. Systematic biochemical experiments are required to answer these questions.

XAFS is an oscillatory feature above the absorption edge of the constituent atom, resulting from the interference between the outgoing photoelectron wave and the backscattered wave from the surrounding atoms. By analyzing the period and amplitude of these oscillations, one can obtain detailed structural information around the excited atom. One can selectively excite different atomic species in the material (by tuning the incident X-ray energy around their binding energies) and obtain site-resolved structural information (Koningsberger & Prins, 1988). This element-specificity makes XAFS more sensitive to small amounts (≥3%) of defects/hidden phases that are not detected by XRD. Further, the region close to the edge of the XAFS spectrum is called XANES (X-ray absorption near-edge structure) and contains information about the oxidation state (single or multi-valence) (Koningsberger & Prins, 1988).

As mentioned above, elemental specificity of XAFS yields better-resolved results than XRD. In this case: (i) in the backdrop of the primary α-HgS phase there could be minute quantities of segregated unstable and toxic chemical phases (pure Hg, β-HgS, organic Hg compound) that are undetected by XRD; (ii) identification of possible core-shell structure of the nanoparticles; (iii) surface segregation (if any) for the nano-crystals; and (iv) identification of local defects (pores etc.)/disorder etc. within the nanoparticle. Each of these parameters, as explained later, is correlated with the toxicity or efficacy of drug delivery/action.

1.1. HgS-based Rasasindura

Due to high mobility, water-solubility and relative ease of oxidation⁴, metallic mercury (Hg⁰) is one of the most toxic elements known to humans; it interacts with the human body to form toxic methyl-mercury [(CH₃)Hg]⁵. While there is worldwide concern against Hg contamination in food/water/soil (Azimi & Moghaddam, 2013; Wang et al., 2004; Zhang & Wong, 2007; Mercury Study Report to Congress, 1997⁶), it is strange that Hg-based medicines are recommended in Ayurveda, often in concentrations larger than the WHO-permissible limit (>1 p.p.m.)⁷. From reported XRD results, the crystal structure of Rasasindura is known to be α-HgS. Preliminarily, the non-toxicity of Rasasindura can be correlated with the advantageous properties of bulk α-HgS (Frawlay, 2000):

(i) Hg has a special affinity for S (Wiberg & Wiberg, 2001), resulting in the formation of a strong Hg—S bond. This is supported by recent experiments on Hg-contaminated soil samples (Mishra et al., 2011), where it was observed that binding of Hg²⁺ with sulfur-containing groups (rather than oxygen-containing group) significantly arrests its reduction (Hg²⁺ → Hg⁰).

(ii) Low solubility and bioavailability (Koch et al., 2013) (0.001 g L⁻¹) of α-HgS results in a low accumulation in the human body (0.2% is absorbed in the gastrointestinal tract and only 0.02% reaches the kidneys) (Sin et al., 1989; Yeoh et al., 1986).

(iii) Toxicity[HgS] ≃ 10⁻⁴ × Toxicity[(CH₃)Hg] (Chuu et al., 2007; Huang et al., 2012; Liu et al., 2008).

(iv) The possibility of HgS → (CH₃)Hg conversion by human intestinal bacteria has been ruled out (Zhou et al., 2011).

However, these advantages of α-HgS may not hold good in nanoparticle form (Bhasma), i.e. for α-HgS nanocrystals: (i) surface energetics, increased strain and vacancy defects (Deneen & Carter, 2006; Perrey et al., 2005) may lead to the formation of local Hg⁰ sites, i.e. initiate the reverse reaction [Hg²⁺ → Hg⁰]. (ii) Increased disorder/strain can disintegrate the nanoparticle into toxic forms; therefore, robustness of the nanoparticle is of additional concern. (iii) Further, XRD cannot unambiguously rule out the presence of <5% Hg⁰ or organic Hg due to its limited resolution. To validate the reported non-toxicity of Rasasindura, we need to establish reliably that Rasasindura has (i) complete absence of the unstable and toxic chemical forms, viz. Hg⁰, β-HgS and organic Hg, and (ii) a robust nano-structure with minimal defects.

1.2. Synthesis method

Rasasindura, like any other licensed and standardized formulation, is subjected to standard operating procedures, including starting/in-process/finishing quality-control checks. Rasasindura was prepared by Arya Vaidya Sala (India) following three distinct steps. (i) Pre-treatment of Hg and S with herbal and milk products: Hg was ground with slaked lime on a mortar for three days and filtered through a fine cloth. The filtrate was ground with garlic and rock salt until it turned black in color and was then washed in water (Sharma, 1979a). S was heated with ghee and allowed to drop through a cloth into milk. The resultant granules were collected and washed with water (Sharma, 1979b). (ii) Mixing of Hg and S [Hg:S = 1:1] along with other herbal ingredients (Aloe vera juice) and ground for five days using an electric grinder, resulting in the formation of black-HgS (Kajjali) (Sharma, 1979c). (iii) Thermal treatment at 600°C of dried Kajjali, in porcelain pots, with porcelain lid and totally covered with seven layers of clay-smeared cloth. The whole pot is smeared with clay for total sealing. The pots are placed into an open hearth furnace (electrically operated) for 24 h with the temperature being raised from room temperature to 600°C. After 24 h, heating is cut off and the pots are allowed to cool down naturally for the next 24 h. The cooled porcelain pots are opened by cutting open the clay seal. The final product, Rasasindura, in the form of fine dust of brick red colour, will be found deposited on the inside roof of the porcelain lid, and is then scraped off (Hariprapannaji, 2004). It is triturated in a mortar and pestle and then passed through a fresh nylon cloth of 200 mesh.

Schematically, the entire synthesis process can be thus summarized:

For HgS formation, S + Hg → HgS, = −46 kJ mol⁻¹. The negative free energy change shows the feasibility of formation of some amount of HgS even before heat treatment of Kajjali (Svensson et al., 2006). Heat treatment of Kajjali is essential to decrease the proportion of unreacted S, and the herbal ingredients provide the acidic medium (catalyst) required for this reaction besides aiding in solidification of Hg (Singh et al., 2009).

To understand the relative stability and non-toxicity of Kajjali and Rasasindura, we studied their inorganic counterparts (viz. black- and red-HgS).

1.3. Inorganic black-HgS and red-HgS

In bulk form, black- and red-HgS generally represent the two structural forms of (inorganic) HgS.

(a) Symmetry. In bulk form, black- and red-HgS represent structural polymorphs, viz. cubic β-HgS (tetrahedral) and trigonal α-HgS (octahedral), respectively.

(b) Stability. Black-HgS is unstable below 600 K (Rickard, 2012), decomposes into red (α)-HgS, and is prone to oxidation. The instability of black-HgS may be due to the poor packing efficiency of the tetrahedral configuration (Karkare & Bahuguna, 2007).⁸ On the other hand, red (α)-HgS is reported to be stable (Bhuse, 2011) because of its significantly enhanced packing efficiency. The smaller misfit factor for the octahedral configuration in red (α)-HgS can be accommodated with a slight distortion of the octahedron and phase-stabilized. In summary, α-phase → stability → non-toxicity.

(c) Toxicity. Red (α)-HgS, the stable structural form, has not been explicitly reported as toxic in the literature (De, 2009⁹).

(Chuu et al., 2007; Huang et al., 2012; Liu et al., 2008). On the other hand, there are several reports on black-HgS being toxic (Morris, 1992¹⁰).

1.4. Pre-XAFS characterization of Rasasindura

Although colors are strong indicators of structural forms, they could be elusive in this case since Bhasma samples are formed of nano-sized particles where color could be size-dependent. To obtain a comprehensive overview of the Rasasindura structure (with reference to laboratory-based α-HgS nanoparticles, which we refer to as `red-HgS' hereon), we have employed complementary techniques: XRD/X-ray fluorescence (XRF)/Fourier transform (FT) Raman and IR/surface enhanced Raman scattering (SERS). XRD and XRF experiments were performed at the Indus-2 synchrotron source (India), where high-resolution information could be extracted due to the very high photon flux (∼10¹¹ photons s⁻¹). SERS, FT-Raman and FT-IR spectra were recorded at Bhabha Atomic Research Centre (India). The conclusions from these techniques, listed in Table 1, unanimously establish that Rasasindura has the same structure as red (α)-HgS and, additionally, is better ordered.

Table 1
Characterization of Rasasindura and red-HgS

Technique		Rasasindura	Red-HgS (lab)
XRD (Fig. S1)	Form	α-HgS; no Hg^0/organo-HgS phase detected	α-HgS; no Hg^0/organo-HgS phase detected
	Shape	Spherical	Spherical
	Particle size	24 nm	18 nm
XRF (Fig. S2)	Elements	Hg, S and Zn (0.6%)	Hg, S and Zn
		(Note that Zn is reportedly found as impurity in Hg)
	Hg:S ratio	1:1	1:1
FT-Raman (Fig. S3)	Form	α-HgS	α-HgS
	Bonding	Strong Hg—S bond	Strong Hg—S bond
	Order	Better ordered
SERS (Fig. S4)	Surface organic groups	Absent	Absent
FT-IR	Organic groups	Absent	Absent

2. XAFS experimental details

For XAFS experiments, Rasasindura was ground to a fine powder using mortar and pestle. The ground powder was mixed with AR-grade isopropyl alcohol in a beaker and the solution was allowed to stand undisturbed for 40 min. During this procedure, the larger particles settled down while the smallest particles remained in solution. The solution was decanted into a petri dish and left to evaporate overnight. The alcohol evaporated, leaving behind fine (∼5 µm-diameter) particles (Lahiri, 2016). 20 mg of these particles was mixed with 80 mg of cellulose and pressed into a 15 mm-diameter pellet, which was used for recording XAFS data.

Transmission-mode XAFS was recorded on Rasasindura and red (α)-HgS at the Hg L₃-edge (12.284 keV) at the Scanning EXAFS beamline (BL-09), Indus-2 Synchrotron Radiation Source, RRCAT (India). A pair of Si(111) crystals in parallel geometry was used as a double-crystal monochromator (DCM). A 1.5 m-long horizontal pre-mirror with meridionial cylindrical curvature was used to obtain a collimated beam on the first crystal of the DCM and reject higher harmonics from the XAFS spectrum. Incident and transmitted intensities were measured using N₂/Ar-filled ionization chambers.

XAFS data were processed using ATHENA software. The extracted XAFS oscillations, χ(k), were Fourier-transformed into real space χ(R) for fitting. {k = where m is the electron mass and E₀ is the edge energy of the relevant absorption edge.} XAFS oscillations in k- (inset) and R-space are shown in Fig. 1. A structural model was constructed using FEFF6.1 (Newville et al., 1995). The model parameters were allowed to vary while fitting (using FEFFIT) to yield the best-fit values for bond lengths (R), coordination numbers (N) and Debye–Waller factors (DWF or σ²). The R-factor was considered as an estimate of the quality of fit (Newville et al., 1995; Newville, 2001).

Figure 1 Comparison of the Hg L3-edge XAFS data on Rasasindura and red (α)-HgS with simulated XAFS data for α- and β-HgS. It is clear that the data for Rasasindura resemble α-HgS (not β-HgS). The inset shows the χ(k) data for Rasasindura.

3. Results and discussion

XAFS data for Rasasindura and laboratory-produced red (α)-HgS were fitted for k = 2.8–8.6 Å⁻¹; R = 1.1–3.4 Å. The limited k-range is due to (i) smearing of the Hg XAFS signal by the high core-hole lifetime of Hg, (ii) large disorder of higher shells (as shown later) and (iii) fast decay of oscillations from low-Z (sulfur) backscattering neighbours. EXAFS modelling was essentially limited to Hg—S fitting for which k = 2.8–8.6 Å⁻¹ is more than sufficient. In order to reduce error bars and have a sufficient number of points for fitting, simultaneous fitting of the data was carried out for different k-weights of the transform, viz. ; w = 0, 1. Some of the ripples observed in Fig. 1 are due to truncation effects from the limited data range and do not represent real data. Apparently, features around 2.5–3 Å (Fig. 1) look less sharp for Rasasindura. However, it should not be mis-interpreted as higher disorder; as we note from the analysis results (described below), Hg—S bond lengths in Rasasindura are displaced relative to each other such that their scattering contributions phase cancels. In fact, Rasasindura is found to be better ordered [than red (α)-HgS] from our analysis.

3.1. Phase segregation

Since changes in XANES (inset of Fig. 2a) are very subtle between α-HgS (Hg⁺²), Hg⁰ and Rasasindura, we have plotted derivatives of their XANES spectra (Patty et al., 2009; Colombo et al., 2013) in Fig. 2(a). XANES derivative features for Rasasindura, viz. pre-edge (A) and edge position (B), resemble α-HgS (Hg⁺²) (Fig. 2a) and are markedly different from Hg⁰ (C). A comparison of XANES derivative spectra of HgS compounds (Rasasindura, α-HgS) and Hg⁰¹¹ clearly reveals (i) a shift in the edge position (B) towards higher energy (E_shift = 5 eV) for Rasasindura, α-HgS, due to higher oxidation state (+2), and (ii) the presence of a pre-edge feature (A) in HgS due to S (2p)–Hg (6d) orbital hybridization, which is absent in Hg⁰. The main absorption edge (B) corresponds to an intra-site Hg (2p) → Hg (6d) transition, consistent with the dipole selection rule = ±1). While direct inter-site Hg (2p) → S (2p) transition is prohibited by the dipole rule, the pre-edge feature (A) is an indirect signature of the transition to S (2p) through Hg (6d)–S (2p) hybridization. Strong pre-edge (hybridization) features confirm strong Hg—S covalent bonds.

Figure 2 Comparison between (a) derivatives of XANES data on α-HgS, Hg^0, Rasasindura and red (α)-HgS (lab). Rasasindura closely resembles α-HgS and red (α)-HgS (lab). Clearly in Rasasindura, (i) the edge position is shifted with respect to Hg^0 (Eshift = 5 eV) and (ii) pre-edge features are present which are absent in Hg^0. The inset shows a comparison between XANES spectra for α-HgS (Hg^+2), Hg^0 and Rasasindura. Since the observed changes are subtle, we compared the derivatives of the respective XANES spectra. (b) Simulated derivative spectra for different fractions (xHg^0) of Hg^0.

Our next objective is to determine whether Hg⁰ is completely absent or partially present. Since our XRD spectrum does not show a signature of the Hg⁰ phase, one can presume that the Hg⁰ content (x) would be <5%. On the other hand, XANES derivative spectra show that the amplitudes of the (A, B) peaks and their ratio (X) are significantly reduced,

in Rasasindura (X = 1.2) with respect to α-HgS (X = 1.5). Since peak A (α-HgS) is negatively correlated with peak C (Hg⁰), a mixture of these phases (Hg⁰, α-HgS) could reduce the net amplitude around peak A and subsequently reduce X (as observed for Rasasindura). This implies that the reduced features of the Rasasindura spectra could be consistent with the co-existence of (Hg⁰, α-HgS) phases. To clarify this ambiguity, we simulated the derivative spectra for different fractions (x_Hg⁰) of Hg⁰ (Fig. 2b). For x_Hg⁰ = 0 → 30%, the amplitudes of the (A, B) peaks are progressively dampened while their ratio (X = 1.5) remains constant. For x_Hg⁰ ≥ 30%, the signature of Hg⁰ becomes dramatically conspicuous as the spectra evolve into peaks A′, C′ following peaks A, C; peak B is completely dampened. With increasing Hg⁰ content, the positions of A′, C′ move away from and towards A, C, respectively, such that the split becomes wider. The spectrum of Rasasindura is not consistent with any of these mixed-phase features; instead, X = 1.3 resembles XANES for Hg—S bonds in soil, water, etc. (Patty et al., 2009; Myneni et al., 2009; Colombo et al., 2014; Huggins et al., 2009; Rajan et al., 2008¹²). Further, XANES for Rasasindura is inconsistent with organo-Hg (Colombo et al., 2013). Reconciling XRD and XANES results, we unambiguously conclude that Hg⁰ and organo-Hg are absent in Rasasindura.

Compiling XANES and SERS results, we confirm that the chemical form of Rasasindura is single-phase α-HgS; the absence of unstable and toxic chemical forms (Hg⁰, β-HgS, organic Hg) is the most significant evidence towards confirming the non-toxicity of Rasasindura.

3.2. Crystalline structure details and degree of disorder

In Fig. 1, we compared Rasasindura/red (α)-HgS data with simulations for α-HgS and β-HgS crystal structures. Except for larger disorder, near-neighbor features (R < 2.5 Å) of both Rasasindura and red (α)-HgS clearly resemble α-HgS (in terms of peak position). Theoretical bond lengths for α-HgS are shown in Table 2. XAFS fit results (Fig. 3a) confirm the α-HgS configuration for S neighbors. Our bond length and coordination number results for Hg—S bonds are consistent with the reported literature (Charnock et al., 2003; Manceau & Nagi, 2008). [Other models (a) cubic = 6S, (b) β-HgS = 4S (R = 2.53 Å) yielded poor quality fits.] From Fig. 3(a) we observe that the Hg—S near-neighbor configuration is better ordered (lower σ²) in Rasasindura compared with laboratory-synthesized inorganic red (α)-HgS. An example of the quality of fit is shown in Fig. 3(b).

Table 2
Comparison of bond lengths (R)

	Hg—S1	Hg—S2	Hg—S3	Hg—Hg1	Hg—Hg2	Hg—Hg3
α-HgS (theory)	2.36 Å	3.1 Å	3.3 Å	3.79 Å	4.09 Å	4.14 Å
Red (α)-HgS (lab) (XAFS)	2.36 Å	3.10 Å	3.30 Å	Absent
Rasasindura (XAFS)	2.34 (2) Å	3.02 (4) Å	3.31 (5) Å	Absent

Figure 3 (a) XAFS fit results for Rasasindura and red (α)-HgS (lab). Rasasindura is more ordered than red (α)-HgS (lab). (b) Comparison of data with fit for Rasasindura.

In contrast, the Hg—Hg bond features (R > 2.5 Å) of α-HgS are conspicuously absent in the experimental XAFS spectra (Charnock et al., 2003; Kim et al., 2004) [although the XRD-generated radial distribution function (using RAD software) (Petkov, 1989) clearly reproduces the Hg—Hg peak]. XAFS and XRD experimental results can be reconciled by considering the possible role of the Hg—Hg disorder (). By simulating XAFS for different , we determined the critical disorder value for the Hg—Hg peak suppression to be ≥ 0.025 Ų (see Fig. S5a of the supporting information). Independently, we simulated XRD patterns with different values of . The XRD pattern for Rasasindura is insensitive to variation up to = 0.035 Ų (Fig. S5b). From XAFS and XRD simulations, we thus conclude and clarify that the apparent `absence' of Hg—Hg correlation features in the experimental XAFS spectrum is actually a reflection of the large disorder in Hg—Hg bonds. Despite Hg being the heavier atom, >> due to a strong Hg—S covalent bond and relatively weak van der Waal interaction between Hg—Hg. The HgS structure can be depicted as spiral –S—Hg—S—Hg—S– chains (Fig. S6a), in which Hg—S and Hg—Hg bonds are intra- and inter-chain, respectively. Any defect in the spiral (e.g. compression/elongation) affects the inter-chain Hg—Hg bond substantially while the strong Hg—S bond remains largely unaffected (Fig. S6b). This leads to > .

3.3. Analysis of the nanocrystal units

For D_Rasa = 24 nm, the surface–volume ratio of atoms is ∼2% (i.e. x_Hg^Surface = 2%). The XAFS coordination result N_Hg–S = 6 (±3%) is the site-averaged contribution from core (x_Hg^Core = 98%) and surface (x_Hg^Surface = 2%) sites. Considering coordination loss due to truncation at a bare surface, the net (theoretical) coordination for a particular bond length (R = Hg—S in this case) (Calvin et al., 2003):

i.e. 2%, which is less than the XAFS resolution ( = ±3%). Thus, our XAFS coordination result is apparently consistent with the theoretical size-dependent coordination loss for a chemically homogeneous defect-free nanocrystal. Any defect, if present, has to be accommodated within a 3% error bar of the XAFS coordination result. We provide defect estimates that are consistent with this constraint.

(a) Vacancy. Vacancy-induced extra coordination loss is not observed; one can rule out the presence of a (≥3%) vacancy in the nanoparticles.

(b) Surface segregation of Hg. Due to an insignificant surface–volume ratio (2%), the surface contribution to the XAFS coordination result is insignificant; hence, XAFS results do not reflect the surface coordination unambiguously. [We have calculated the critical core size (D_core ≤ 4 nm) for resolving surface coordination.] To estimate the possibility (and fraction) of preferential surface site occupancy by Hg, we considered the chemical non-uniformity across the nanocrystal and allowed x_Hg^Surface:x_Hg^Core to vary (instead of constraining x_Hg^Surface = 2%). By re-calculating the site-averaged Hg coordination as a function of x_Hg^Surface, we determined that x_Hg^Surface < 6% is consistent with the XAFS result.

(c) Organic molecules at the surface. Due to an insignificant surface–volume ratio and weak backscattering factor of organic elements, XAFS was unable to unambiguously detect the presence of surface organic molecules. We obtained the answer by employing SERS. SERS results convincingly ruled out the presence of organic molecules at the surface. This is remarkable since the Ayurvedic synthesis method involves organic molecules. We believe that organic molecules were removed during purification steps or heating.

(d) Porosity. To explore the presence of nano-pores (cylindrical) in the particle, we theoretically estimated the average coordination for pore parameters D_p, H_p, N_p; D_p = pore diameter, H_p = pore-height, N_p = number of pores. By varying one of these parameters, H_p = 1–10 nm (i.e. H_p^max = radius of the nanocrystal), we determined the range of (D_p,N_p)_Hp (i.e. for each H_p) that is consistent with N_expt 6 (±3%). Compiling all these calculation results, we obtained D_p^max = 10 nm, N_p^max = 20, V_p^max = 7%. [Note that (D_P,N_p)_Hp are negatively correlated.] V_P^max = 7% implies that pores occupy a small volume fraction of the nanocrystal, which is insufficient to generate large disorder and helps preserve the robust character of the nanoparticle. Minimal defects for these nanocrystals could be correlated with prolonged annealing and large particle size (D_Rasa 24 nm). Interfacial defects and broadening generally increase significantly for D ≤ 8 nm (Shibata et al., 2002; Chaudhuri & Paria, 2012; Alayoglu et al., 2009).

[Note that the above conclusions for Rasasindura also hold good for red (α)-HgS due to similar nanoparticle size and coordination results.] These pores could provide the pathway for drug binding.

A summary of the nanocrystal coordination analysis is the presence of a robust and chemically homogeneous (α-HgS) nanocrystal (Fig. 4). Including the error bar ( = ±3%), we could accommodate possible defects in the particle (viz. vacancy, surface segregation, nano-pore) and determine the upper limit of their content: <3–7%.

Figure 4 Model of a Rasasindura nanocrystal with 24 nm HgS core.

3.4. Implications of our structural results for toxicity and synthesis route

3.4.1. Toxicity

For successful non-toxic functioning of heavy-metal-based Ayurvedic medicine, the most crucial factors (with respect to toxicity) are: (i) no organic chemical form of metal (Hg) and complete oxidation of metals (no remnant of Hg⁰ metallic state) before entering the human body; (ii) integrated structure of the medicine so that the compound form is retained inside the body (no reduction to metallic form).

From our work, we observe that both criteria are met for Rasasindura. Our XANES/SERS/EXAFS results unanim­ously establish that Rasasindura is in the single-phase α-HgS stable form; unstable and toxic forms of Hg, viz. metallic Hg⁰ [i.e. Hg is completely oxidized], β-HgS and organic complex of Hg, are absent. Our structural results unanimously establish that Rasasindura is composed of robust (minimal defects) single-phase α-HgS nanoparticle units (D_Rasa 24 nm).

As already mentioned above, α-HgS is non-toxic which implies the same for Rasasindura. The stable α-phase form and robust character of the nanoparticle would help to maintain its integrity during the entire drug delivery process (Yoshitomi et al., 2009). Cinnabar, being one of the most stable and non-reactive classes of compounds addressing toxicity issues related to engineered nano-material is not particularly relevant in this case.

3.4.2. Synthesis

We compared the structures of Rasasindura and red-HgS, i.e. end products of organic and in­organic synthesis methods. Interestingly, we found that the particle size distribution is better controlled in Rasasindura.¹³ The distribution is Gaussian, with size = 3.5 nm and dispersion = 18%. On the other hand, the size-distribution of red (α)-HgS is non-Gaussian (D_peak ≃ 20 nm) and heavily skewed towards higher particle sizes; the total size distribution spreads over = 50 nm. The better size control in Rasasindura could be due to herbal coating, similar to the modern-day surfactant-mediated nanoparticle synthesis and sublimation process at the final stage. The coordination configuration for both show large disorder for Hg—Hg bonds but well defined Hg—S bonds. The Hg—S coordination distribution is better ordered in Rasasindura (Fig. 3) which could be due to prolonged annealing.

The efficacy of drug delivery and action is directly affected by particle size, size distribution and order (Tiwari & Tiwari, 2013). Size (and size distribution) can determine the in vivo distribution, biological fate, toxicity and targeting ability of these delivery systems. Although the implications of size (D_Rasa ≃ 24 nm) are not clear at the moment, we would like to cite the correlation between nano-drug and target cell sizes, as in targeted drug delivery (optimal particle size requirement is target specific) (Walkey et al., 2009; Smith et al., 2013; Niikura et al., 2013). Why Ayurvedic α-HgS (Rasasindura) rather than inorganic α-HgS nanoparticle (red HgS) is therapeutic could be related to their difference in size distribution; larger red-HgS particles (30–70 nm) may be failing to penetrate cells and thus be less effective. Further, the better ordered structure (and consequently predictable electronic structure) of Rasasindura can help maintain better equilibrium within the body.

Thus, our work not only helps to understand the non-toxicity of Rasasindura but also establishes the Ayurvedic synthesis method for a well controlled end-product.

4. Conclusion

We have employed XAFS (coupled with supporting techniques) to investigate the structure of Rasasindura. The main results are that Rasasindura has the same structure as non-toxic α-HgS, and toxic chemical forms, viz. elemental Hg⁰, organo-Hg, are completely absent. Our results also demonstrate that the nanocrystal (D_Rasa 24 nm) units of Rasasindura are robust, defect-free and free of organic molecules. The absence of these toxic chemical forms helps in the understanding of non-toxicity, and the robust character implies the nanoparticle integrity during drug release. Further, we observed that Ayurvedic synthesis yielded a better controlled end-product than laboratory-based red (α)-HgS: lower size dispersion and better ordered coordination configuration. With all these advantages, Rasasindura can be considered a potential therapeutic agent for target cell sizes (∼24 nm).

Footnotes

¹See also https://www.ccras.nic.in/PharmacopoeialWork/Links/Compfom/AyurvedicFarmocopia.pdf.

²https://www.bhu.ac.in/ims/ayurveda/ayurveda_about.htm; https://www.ayurvedamanuscripts.com/; https://www.rria.nic.in/; https://www.nia.nic.in/?ref=12&id=6.

³See also https://www.slideshare.net/technoayurveda/bhasma-nano.

⁴https://web.anl.gov/PCS/acsfuel/maroto-valer.pdf; https://copublications.greenfacts.org/en/dental-amalgam/l-3/4-health-effects-amalgams.html.

⁵https://www.cfspages.com/mmercmicro.html.

⁶See also https://www.epa.gov/hg/eco.htm; https://www.usgs.gov/themes/factsheet/146-00/;https://www.cseindia.org/node/439.

⁷https://pib.nic.in/newsite/erelease.aspx?relid=42213; https://www.udel.edu/chem/C465/senior/fall00/Mercury/health.htm.

⁸Packing fraction PF_β-HgS = 32%; PF_α-HgS = 58%.

⁹See also https://www.mercurynetwork.org/uk/wp-content/uploads/2009/11/Orthiel.pdf; https://www.hallegeologicalservices.ca/?p=692.

¹⁰See also https://www.lookchem.com/Mercury-II-sulfide/; https://www.answers.com/topic/mercuricsulfide; https://www.instruments-support/labs-facilities/chemistry-laboratories/list-of-chemicals/dangerous-chemicals.ill.eu/.

¹¹https://ftp.esrf.eu/pub/UserReports/44815_B.pdf.

¹²See also https://doesbr.org/PImeetings/2012/pdf/Tues/NagySBR2012.pdf.

¹³Transmission electron microscope result provided by Professor Sujit Roy, IIT-Bhubaneswar.