The carbides of the transition metals in Groups IV - VI have extremely high melting points (Table 8.1) and are therefore referred to collectively as the “refractory carbides.” In addition to their stability at high temperatures, these compounds are extremely hard (Table 8.2), finding industrial use in cutting tools and wear-resistant parts. Their hardness is retained to very high temperatures, and they have low chemical reactivity – they are attacked only by concentrated acid or base in the presence of oxidizing agents at room temperature, and retain good corrosion resistance to high temperatures. The refractory carbides are strong, with Young’s modulus values – a measure of elastic deformation resistance – rivaling those of SiC at room temperature. In addition, they have good thermal shock resistance and good thermal conductivity, permitting heat to be drawn away from the working surface of the tool. This gives them a benefit over other refractory materials, which do not conduct heat so well. (Table 8.3)

Tungsten carbide, WC, is the most commonly used for fabrication as “cemented carbide” tools for cutting steel, in which the carbide is bonded in a metal matrix, usually cobalt. Cobalt is used because it wets the carbide particles and therefore behaves as a good binder without having significant ability to dissolve the carbide, so that the carbide is left pure in the bound form. However, pure WC-Co cemented carbides tend to weld locally with the steel being cut. TiC, TaC, and NbC are often used in conjunction with WC because TiC locally forms a layer of TiO² or TiO₂³ which protects the tool from wear, and TaC and NbC raise the melting temperature and oxidation resistance of the tool.

It must be noted that there is some variation in the literature with respect to the reports of assorted mechanical and thermodynamic properties of the refractory carbides. The transition metal carbides show a range of nonstoichiometries and possibilities for vacancy ordering, so the precise phases being tested for a given property are often unclear. Furthermore, small concentrations of oxygen present as metal oxide are famously difficult to remove – or even detect – and can be expected to affect the properties of the material.

Mixed-metal carbides have been examined for their melting point and hot-hardness behavior as well. The hardness arc-cast or zone-melted samples of (Ta_0.8Hf_0.2)C_1+x was tested (by indentation using diamond or B₄C tips) and compared with that of the similarly-prepared pure-metal carbides TaC_1-x and HfC_1+x over the temperature range 800 - 2000 ^oC.⁴ In all cases the hot hardness decreased with increasing temperature. For temperatures up to 1400 ^oC the hardness order increased as HfC_1+x < (Ta_0.8Hf_0.2)C_1+x < TaC_1-x, but over 1400 ^oC the mixed carbide began to outperform the tantalum carbide. Hardnesses varied from 600 kg/mm² to 150 kg/mm² over the temperature range tested.

The melting points of the mixed-metal carbides outperform those of the pure-metal carbides, as well. Samples prepared by vacuum reduction of the mixed oxide powders at 2000 ^oC, followed by sintering at 2200 ^oC and 2500 ^oC in good vacuum were tested for their melting behavior.⁵ 8TaC^.ZrC and 4TaC^.HfC had melting points 3890 ^oC and 3990 ^oC, respectively, somewhat higher than those of the pure-metal carbides (measured at 3470, 3750, and 3840 ^oC for Zr, Hf, and Ta carbides, respectively). The higher melting points in the mixed-metal carbides was attributed to composition changes due to the selective evaporation of carbon during melting.

In addition to the technological uses of transition metal carbides which exploit their high hardness and stability at high temperatures, certain carbides have been examined for their catalytic properties in a number of reactions. This is in addition to their being potential supports for more traditional catalytic materials (Ni, Pt, Rh, etc.) due to their high heat stability.

The observation has been made that tungsten was active as a catalyst and showed good selectivity toward xylene formation during the isomerization and hydrogenolysis of 1,1,3-trimethylcyclopentane, but only after an induction period, as is also characteristic of platinum and paladium.⁷ Base transition metals do not behave this way. The behavior was explained by invoking the formation of tungsten carbide on the surface of the metal; WC is similar to Pt in selectivity for neopentane isomerization as well.⁸ Mo₂C has been found to behave similarly to Ru in CO-H₂ reactions.⁹ These are merely two examples: in fact, carbides of the Group IV-VI metals have been studied for their activity in oxidation, hydrogenation/dehydrogenation, isomerization, hydrogenolysis, and CO-H₂ reactions, and in many cases have been found to rival the performance of the less economic Group VIII metals. While the refractory carbides do not show high activity for oxidation reactions (for example, the rate of H₂ oxidation follows the order metal >> carbide > oxide for Group V and VI metals¹⁰, and the rate of NH₃ oxidation over refractory carbides is lower than that over Group VIII metals¹¹), they are as active as the transition metals themselves for hydrogenation and dehydrogenation reactions. In isomerization reactions, WC, Pt, and Ir are unique in their high activity and selectivity.¹² However, it is not necessary for the refractory carbides to be more or even equally active in catalyzing given reactions compared with the noble metals, because the lower cost of the carbides will in many cases offset the losses in catalytic activity.

The Group IV-VI metal carbides are, however, difficult to prepare as high purity (e.g. free of surface oxide or graphitic and amorphous carbon contamination) and high surface area powders. There have been a number of approaches to synthesizing the carbides in a way that produces a high specific surface area, including carburization of a spray of the oxide powder using a CH₄/H₂ gas mixture,¹³ reaction of the metal oxide vapors with activated carbon,¹⁴ ultrasound irradiation of the metal carbonyl,¹⁵ carburization of a precursor deposited onto a support such as alumina,^16,17 and similar approaches.¹⁸ The problem of surface contamination has been addressed as well, with most solutions centering on the idea of activation of the surface of the carbide via a thermal treatment prior to catalytic use. The thermal treatment includes heating in vacuum, which has been found to activate TaC, TiC, and WC for hydrogenation of ethylene,^19,20 or reduction in flowing hydrogen gas.^14,21 These methods, of course, are for removing surface oxygen; surface carbon (amorphous or graphitic) is difficult to remove.

The catalytic uses of the transition metal carbides are mentioned in the context of this project because the low-temperature precursor methods reported in Chapter 9 have the potential for yielding high-surface area carbides possibly useful for catalytic applications. However, the structural and refractory characteristics of the carbides were the focus of the project reported in Part II of this thesis, and no exploration of the catalytic potential, even characterization as basic as finding the surface area via BET analysis, was considered.

Most of the transition metal monocarbides form in the B1 (NaCl) structure, fcc metal with carbon occupying the octahedral interstitial sites. The shortest M-M distance is about 30% greater in the B1 carbide than in the pure metal for the Group IV and V carbides, but drops to less than 10% greater for the Group VI or VII carbides.¹⁸ At 100% site occupancy, the stoichiometry of the carbide is MC_1.0, though this situation is rarely realized. The concentration and ordering, if any, of the vacancies that result from a nonstoichiometric M-C ratio have a great effect on the thermodynamic, mechanical, electronic, and magnetic properties of the metal carbides; however, the details of these effects are a matter of some debate in the literature, due to the difficulties inherent in synthesizing pure compounds and in measuring the exact details of the crystal structure of a given sample. The metal carbides share many characteristics with the metals themselves, having a plastic deformation like the fcc metals which, while lowering the high-temperature hardness, protects parts fabricated from the carbides from catastrophic failure in response to stresses.

Most of the Group IV-V metal carbides conform to the Hägg rules, which were developed empirically to “predict” the structures of the transition metal borides, carbides, halides, and nitrides.²² The structure adopted by the metal carbide is determined by the ratio of the radius of the nonmetal atom (r_x) to that of the metal atom (r_m). For r = r_x / r_m < 0.59, the structures adopted are the simple A1, A2, A3, and hexagonal lattices. For r > 0.59, more complex structures form to prevent the expansion of the lattice – which would have been required for the simple structure to accommodate the large nonmetal atom – from taking the metal atoms beyond the distance for favorable metal-metal interactions. The monocarbides take on an fcc metal lattice with carbon atoms on the octahedral interstitial sites, while random occupation of half of the O_h sites in M₂C or M₃C leads to the L3’ (anti-NiAs) structure, and carbon occupation of the trigonal prismatic sites in the hcp lattice formed by the tungsten atoms leads to the CdI₂ structure.² Non-Hägg structures are known among the carbides as well, a major example being Cr₂C₃. Specific phases available to each metal will be discussed in Section 8.4 below.

One much-noted feature of the structure of the transition metal carbides is that the lattice adopted by the metal in the carbide is never that of the parent metal. That is to say, if the metal has an hcp lattice, its carbide has the metal on an fcc lattice; the fcc parent metal occupies a non-cubic lattice in the daughter carbide; and bcc parent metals have both fcc and hcp lattices available in the daughter carbides. This has been explained using Engel-Brewer theory of metals,^23,24 in which the structure adopted by a metal or alloy depends on the s-p electron count.¹⁸ With increasing s-p electron count the metal structure progresses from bcc to hcp to fcc across the transition series. Likewise, the Group IV and V metal carbides MC form in the B1 structure rather than a hexagonal form because the incompletely filled bands of the host metals can accommodate a high ratio of sp-electron-rich carbon to metal. In Group VI the stoichiometry M₂C occurs often, while Groups VII and VIII, when they form carbides at all, take on metal-rich stoichiometries M₃C and M₄C, consistent with an attempt to avoid filling antibonding levels in the metal bands.²⁵

The nature of the bonding in the monocarbides is a matter of some debate, although all agree that the simple ionic model (M⁺C^- or M^-C⁺) is not consistent with the properties of the carbides. Ionic materials will not typically slip on the {111} planes due to the strong repulsive (coulombic) interactions across the shear plane in the half-glide position; instead, they will slip on the {110} or {100} planes.² The catalytic behavior of WC is similar to that of Pt, as noted in Section 8.1.2, and this is explained according to the addition of the carbon valence electrons to those of tungsten to give a count equal to that of platinum. However, the exact direction of electron transfer is the subject of some controversy. X-ray photelectron spectroscopy (XPS) and electronegativity considerations suggest simple M --> C electron donation, but the XPS data is questionable due to the possibility for backdonation or screening effects.²⁵ Furthermore, simple M ‡ C donation would result in ionic compounds such as the alkali and alkaline earth metal carbides. However, these materials are resistors with low optical transparency or reflectivity, and readily hydrolyze to the metal oxide and hydrocarbon. In contrast, the transition metal carbides are conductors with a shiny metallic and colored appearance and are hydrolytically stable. Band occupation suggests C --> M electron transfer. Carbon appears to combine its sp electrons with the metal spd bands, leading to the similarity between the crystal structures, reactivities, and catalytic activities of the refractory carbides and the Group VIII metals. Such a donation scheme may be used to explain, if crudely, the trend in melting point maxima, which occurs in Group VI for the metals, Group V for the carbides, and Group IV for the analogous mononitrides: the maxima may be associated with the half-filled d shell.²⁵

The Group IV-V carbides are able to form continuous solid solutions with each other over a wide range of compositions, but are only partially miscible with the carbides of the Group VI - VIII metals.¹⁸ Ceramic materials of mixed C, N, and O composition are also common, with oxycarbides of definite stoichiometry having been reported for a number of the early transition metals. Even “pure” metal carbides tend to contain small amounts of dissolved oxygen.

The distinction between “traditional” methods and “precursor” methods made in this section is admittedly somewhat arbitrary, since any starting material besides the elements themselves could be considered to be “precursors.” However, the separation has been made so that simple starting materials are included with “traditional” methods, and starting materials requiring some high degree of design and synthetic effort have been designated “precursor” methods.

The reaction temperatures for synthesis of the refractory carbides according to Equations 8.3a, b are listed in Table 8.5. However, most carbide systems must be heated for several hours at temperatures over 2000 ^oC to ensure compositional homogeneity. With good vacuum, such treatment also will remove oxygen contamination from many of the metal carbides. Reports regarding the range of temperatures required to form the carbides according to other reaction schemes give variable numbers.

Many of the reported precursor methods for making high-temperature carbides focus on the synthesis of alpha- or beta-SiC rather than on the carbides of the early transition metals. In addition the solid-state reaction between silica and carbon (Acheson process),⁶ beta-SiC has been made by pyrolysis of high-Si rice hulls (a carbothermal reduction of SiO₂ collected by the rice plant and locallized in the hulls). This process is commercially practiced. It has also been synthesized, more notably, by the pyrolysis of polymers such as poly(dimethylsilane) or poly(phenylmethylsilane).²⁶ These polymers are synthesized via the Wurtz-type coupling of appropriate chlorosilanes, e.g. coupling in the presence of sodium or potassium metal (M) with loss of MCl driving the polymerization. Linear polycarbosilanes produce nearly zero ceramic yield: branching on the silicon backbone is crucial. The presence of Si-H and Si-vinyl functionalities lead to the potential for hydrosilation cross-linking and a still-higher ceramic yield.

Similar work has been done to give organometallic precursors for the transition metal carbides. This work is dogged by the relatively few compounds of the Group IV - V metals which are stable, have a low C/M ratio (so that excess free carbon is avoided), and which do not contain oxygen. Those which do fulfill these criteria tend to be sublimable, leading to a low ceramic yield, since the opportunities for polymerization or cross-linking of these materials are few. Seyferth and Tracy attempted to couple dimethyltitanocene with methyldichlorosilane in the presence of sodium metal to produce a copolymer precursor for (Si, Ti)C, but little of the product was soluble in toluene and the ceramic yield was low.³ Dichlorobis(2,4-pentanedionato)titanium(IV) was reacted similarly with methyldichlorosilane in the presence of Na, but the products were soluble only in THF and contained chlorine. Finally, dimethylmetalloocene, which decomposes in ultraviolet light to Cp₂M and 2 ^.CH₃ (M=Ti, Zr, Hf), was used with the polysilane made by the Würtz coupling of HMeSiCl₂ to produce air-sensitive metal-containing polysilanes with ceramic yields ranging from 44% (Hf) to approximately 80% (Ti). The polymers were at least sparingly soluble in aromatic or ethereal solvents, sometimes more soluble, depending on the specific reaction conditions.

Unpublished research performed in the Seyferth labs polymerized zirconium and hafnium metallocene methacrylates for use as precursors for ZrC and HfC. The precursors were solids or viscous polymers. Liquid precursors were obtained when substituted metallocene dichlorides were polymerized with n-butyllithium and a diyne. These gave ceramic yields near 60%, but the process suffered from the inability to scale it up.

Most other research involving organometallic precursors for the refractory carbides has centered on the small-molecule, monomeric metallocene compounds for use with various forms of chemical vapor deposition of films of these the carbide materials.

Precursors for the refractory carbides have been made from the metal alkoxides transesterified with polyhydroxyl alcohols and mixed with phenolic resins or furfuryl alcohol to provide excess carbon. In most of these cases the precursor was a solid, though some soluble precursor polymers resulted. Past work along this line will be discussed in more detail in Chapter 9.

Complete characterization of the refractory carbides is difficult. Since many of the important mechanical and catalytic properties are sensitive to a number of factors which tend to vary widely among samples, there is a variation in literature reports regarding measurement of these properties. These factors include (1) the crystal structure and lattice parameters, including the presence of vacancy ordering; (2) the chemical composition, including not only the overall carbon-to-metal ratio present in the bulk sample, but the amount of free carbon versus combined (lattice) carbon; (3) the impurity concentration, particularly that of oxygen; (4) the overall defect structure, including grain size, dislocations, and porosity; and (5) the sample homogeneity.

The crystal structure can be found and the lattice parameters measured by X-ray and neutron diffraction experiments. Sample homogeneity may be deduced from X-ray diffraction (XRD) data by noting the sharpness of the splitting the K_alpha-1 and K_alpha-2 lines, and crystalline impurity phases present in concentrations over approximately 5% may often be detected using XRD. Even large oxygen impurity concentrations, however, may escape notice due to the formation of a solid solution with the carbide. The effects of medium-to-large oxygen impurity concentrations on lattice parameters are not well-described. Very careful XRD analysis may also detect ordering of carbon atoms or vacancies in the structures of the carbides of the lighter metals, when such ordering is accompanied by a distortion of the lattice. The specific details of the ordering would not be easy to determine even using modelling techniques such as Rietveld analysis;²⁷ ordering without an accompanying symmetry or lattice parameter change would only be detectable using a complex Fourier analysis of the XRD peak intensities.²⁸ Neutron or electron diffraction experiments, on the other hand, would give unambiguous information regarding the ordering of carbon atoms on the interstitial sites, however, electron diffraction sample preparation is made difficult by the need to make a thin section of the brittle and possibly porous carbide material, and neutron diffraction is not immediately available to most researchers.

Chemical analysis of the bulk material primarily yields information about the quantity of carbon in the sample. The total carbon content may be found by heating the carbide in a stream of oxygen, completely converting it to the metal oxide. The CO₂ generated is captured and measured by weight or by the conductivity of a combusted CO₂-O₂ mixture in a commercial Leco unit. This process is complicated by the high temperatures tolerated by the refractory carbides, since it relies on all of the carbon’s being driven off as CO₂. Diffusion of carbon and/or oxygen through the metal carbide and/or oxide is slow even at very high temperatures. Free carbon (graphite) may be determined by dissolving the carbide in a mixture of hydrofluoric and nitric acids (warning: the dissolution reaction is violent and exothermic). The unbound carbon does not dissolve; it is collected, washed, and analyzed as CO₂.

Vanadium carbide has been formed by heating V₂O₅ or V₂O₃ with carbon for two hours at 1800 ^oC in 1-10 torr of CO. V₂O₅ begins reacting with carbon at 435 ^oC, and the oxygen concentration is relatively easily reduced to below detectable limits by higher-temperature treatments or by the reaction of vanadium metal or hydride with carbon. Loss of vanadium at high temperatures and low carbon content presents a difficulty, but nearly-saturated VC can be heated to 2000 ^oC without loss of vanadium. The carbide has been made by treatment of VCl₄ in an atmosphere of hydrogen and methane at 1500 - 2000 ^oC. The two main phases are cubic VC, available over the range VC_0.78-VC_1.0, and the hexagonal beta-V₂C, available for C/V atom ratios of 0.4-0.5 between approximately 1500 - 2000 ^oC, but presenting a very narrow range of stable compositions at room temperature. Reports of V₅C and V₄C₃ have been discredited. VC will react with dry HCl gas at 750 ^oC to produce VCl₂, methane, and hydrogen, and has a high rate of oxidation in air, with powdered VC and V₂C being attacked slowly by air even at room temperature. V₂C is soluble in hot 50% HCl solution, leaving a carbon residue, but VC is inert under these conditions; both of the vanadium carbides are attacked by concentrated nitric, sulfuric, and perchloric acids.

Niobium (V) oxide begins to react with carbon at 675 ^oC, forming NbO₂ and NbC_x below 1200 ^oC and forming an NbC_xO_y solid solution between 1450 and 1500 ^oC. Pure NbC is accessible by heating the metal and carbon powders directly, but high temperatures and heating times are required to complete the reaction and drive off oxygen and nitrogen contaminants. The conditions are made less stringent by the presence of an H₂ atmosphere. NbCl₅ heated in the presence of hydrogen and methane forms the pure carbide at 900 -1000 ^oC. The Nb-C system has the cubic phase NbC and two crystal forms (alpha, room temperature, with a very narrow composition range and beta, existing between 2300-3000 ^oC and over the C/Nb ratio range 0.4-0.5) of Nb₂C. A third, zeta phase of Nb₂C (C/Nb range 0.5-0.7) has been suggested to exist on the basis of a single weak powder pattern line, but has not been verified. The lattice parameter increases as the C/Nb atom ratio approaches unity, and is increased as well by the presence of oxygen and nitrogen. NbC is inert even to boiling aqua regia but will dissolve in HNO₃/HF mixtures; it is severely corroded in air at temperatures above 1100 ^oC. Its color ranges from grey (NbC_0.9) to lavender (NbC_0.99).

The tantalum carbide system is relatively easy to free of oxygen impurities, but due to the slow rate of carbon diffusion it tends to have inhomogeneities in its bulk composition. Evaporation of carbon at temperatures above 2400 ^oC renders the use of high temperatures to establish a uniform composition problematic. Reaction from the elements in vacuum begins at approximately 1000 ^oC but is slow to reach completion. Use of hydrogen or methane atmospheres increases the reaction rate but requires a post-synthesis vacuum annealling step to remove dissolved hydrogen. TaC cannot be made from TaCl₅ in a hydrocarbon/hydrogen atmosphere due to the formation of metallic tantalum, but has been made with varying success by heating Ta wire in methane. Arc-melting tends to produce C-deficient, inhomogenous carbides. The Ta-C system has the cubic phase TaC and a hexagonal compound Ta₂C (actually the C6 anti-CdI₂ structure type due to ordering of the carbon atoms) with an α-β transition near 2000 ^oC. A ζ phase has been claimed at 2000 - 3000 ^oC over the C/Ta atom ratio 0.70-0.75 but its existence remains a point of debate. The cubic phase persists over a wide temperature and composition range, possibly as low as TaC_0.58 and verified down to at least TaC_0.74. The composition TaC_0.89 is the highest-melting substance known. Very small amounts of carbon (Ta₆₄C) result in a tetragonal distortion to the normally bcc Ta parent lattice. The TaC lattice varies linearly with composition, with the equation C/Ta = -25.641 + 5.9757a. It is grey and metallic in appearance up to about TaC_0.85, then becomes increasingly brown with rising carbon content until the golden TaC_0.99 is reached. It is the most acid-stable of the refractory carbides, dissolving in a nitric/hydrofluoric acid mix, and reacts with pure oxygen above 800 ^oC. Loss of carbon results from lower-temperature reactions with oxygen in air.

The specific goals of Part II of this thesis were the synthesis of a industrially-useful liquid precursor for tantalum and hafnium carbides, TaC and HfC. As indicated in Section 8.3.2 above, most precursor methods have resulted in solid or at best soluble polymeric materials. Liquid precursors were preferred in this project so that removal of solvent during fabrication of HfC or TaC monoliths could be avoided. The liquid precursors were to be used in conjunction with proprietary technology owned by Foster-Miller, Inc., of Waltham, Massachusetts, to fabricate high-performance parts of solid TaC or HfC. Factors of concern included (1) ease of synthesis, particularly of scale-up; (2) ease of handling (e.g. room-temperature stability, reasonable air-stability); (3) clean decomposition to the target carbide without undue contamination by metal oxide or by excess graphitic carbon; (4) low volatility – e.g. the material should decompose in situ during pyrolysis, rather than distill; (5) low gas production on decomposition, so that bubbles in the solid would not become points of structural weakness; (6) low viscosity over a temperature range near or just above room temperature (e.g. the precursor was to be a liquid or low-melting solid). Cost was an addtional concern but was left to be addressed in later phases of the project.

Early work on the project focused on the use of organometalic precursors for TaC and for HfC. The main idea behind this approach was that the precursor was best formed by putting the M-C bonds in place during the synthesis, eliminating any need to remove heteroatom bonds (particularly oxygen) from the metal center via carbothermal reduction during the pyrolysis step. Large, waxy groups or smaller, symmetry-destroying groups, would enforce the liquid form of the precursor. Efforts first focused on using CpTaCl₄ or Cp₂TaCl₂ as starting materials, and using organolithium or Grignard reagents to add alkyne groups based on 1-hexyne, trimethylsilylacetylene, and triethylsilylacetylene by a salt metathesis; however, the products consistently decomposed, possibly due to reduction of the tantalum by the organolithium or -magnesium reagent. A similar approach using hafnocene and substituted hafnocenes (e.g. bis(methylcyclopentadienyl)hafnium dichloride, bis(vinyldimethylsilylcyclopentadienyl) hafnium dichloride) gave liquid precursors which pyrolyzed to HfC, and which were obtained in pure form. Stirring pentakis(dimethylamido)tantalum with alkynes, with loss of ≥ 2 dimethylamine groups, gave liquid precursors which pyrolyzed to TaC but which defied purification and characterization attempts. Thus this work showed some success, but was dogged by the sensitivity of the organometallic compounds made. The organometallic approach has not been discussed in detail in this Thesis.

Practical concerns of scalability were a major factor in the abandonment of this approach in favor of a sol-gel-type route in which the alkoxide in alcohol solution would be hydrolyzed to form the precursor, which would be reduced carbothermally during pyrolysis. This route was superior in ease of synthesis, scale-up, handling, and cost, but was found to give a too-low mass-of-carbide-per-gram-of-precursor. It also formed a glassy, bubble-ridden solid which was high in excess carbon. The work is discussed in Chapter 9, and has been continued by researchers at Foster-Miller, Inc.

It should be noted by the reader that the work described in Chapter 9 involved a practical challenge to the academic chemist. Traditional methods of purifying chemical materials for characterization include primarily crystallization and distillation or sublimation. However, the materials synthesized in these chapters were designed not to crystallize, and were expected to decompose on heating rather than to distill. Therefore, purification for the sake of getting NMR, MS, or elemental analytic evidence of the structure claimed was rarely possible. Production of a formulation with reproducible properties and pyrolysis behavior was given priority over isolation and characterization of an exact precursor molecule.