Silicon Carbide

Silicon carbide is a compound derived from silicon and carbon at a 1:1 atomic ratio, with diamondoid sp3-hybridized bonding between those atoms.  Bulk silicon carbide is a hard, dark gray, exceptionally stable ceramic – highly resistant to chemical attack, mechanical abrasion, exposure to temperatures of nearly 3000°C, and the ultra-high pressures which characterize the most extreme ballistic impacts.  In recent years, silicon carbide has become the workhorse ceramic armor material in high-end military body armor systems, and is growing increasingly important in power electronics and other industries.

Silicon carbide has about 250 polytypes, some of which exhibit distinct mechanical, optical, and physical properties.  The mechanical properties of the most common 4H/6H mixed polytype (sintered, polycrystalline,) can be summarized:

Density: 
3.21 gm/cc

Hardness (Vickers):  

2300-2600 HV1, sintered polycrystalline alpha silicon carbide
Fracture toughness:  3 – 4 MPa*m1/2
Compressive strength: 3500-4900 MPa
Flexural strength:  360-550 MPa (three point bend, average), sintered polycrystalline alpha silicon carbide.
Melting point: Degrades at 2830°C ± 30°C
Sonic velocity:  13.700 km/s (L) 


Silicon Carbide History 

Although Swedish chemist Jöns Jacob Berzelius is best recognized for his discovery of metallic silicon, he was also the first to create silicon carbide. In a report he wrote in 1824, he noted that one of the samples he had created probably contained a chemical link between silicon and carbon.

Nothing further came of Berzelius’ discovery, which amounted to little more than a footnote, and silicon carbide was accidentally rediscovered in 1891 by Edward Goodrich Acheson, formerly an assistant to Thomas Edison, during experiments on the synthesis of abrasives in electric arc furnaces.  Acheson himself has left us with a concise and interesting account of this discovery:  

“[I] did quite a great deal of experimenting during the Winter on various lines. I think it was in February, 1891, I was working on the making of rubber synthetically. I succeeded in producing a small piece, when at this critical moment Mr. John S. Huyler came from New York to see our plant. He was not pleased with the prospects. … With this, he left me to my own resources. His remarks discouraged me regarding rubber; I dropped the subject and resolved to endeavor to produce an artificial abrasive. . .  

The value of a good abrasive was brought to my attention by a remark incidentally made in 1880 by Dr. George F. Kunz of Tiffany & Company, New York.  I also remembered the observation of clay impregnated with carbon I made at Gosford, and I decided to make experiments on impregnating clay with carbon under the influence of electric heat. An iron bowl, such as plumbers use for holding their melted solder, was attached to one lead from a dynamo and filled with a mixture of clay and powdered coke, the end of an arc light carbon attached to the other lead was inserted into the mixture. The percentage of coke was high enough to carry a current, and a good strong one was passed through the mixture between the lamp carbon and bowl until the clay in the center was melted and heated to a very high temperature. . . 

“When cold, the mass was examined. It did not fill my expectations, but I, by sheer chance, happened to notice a few bright specks on the end of the arc carbon that had been in the mixture. I placed one on the end of a lead pencil and drew it across a pane of glass. It cut the glass like a diamond.”

Acheson had assumed that the hard, blue, crystalline substance was a reaction product between the carbon he was using in his diamond experiments and the alumina (or corundum) contained in the clay, which is why the SiC he discovered was given the trade name “carborundum.”  It quickly became apparent that carborundum contains neither corundum, nor aluminum, nor oxygen, but the name stuck.

Several years after his serendipitous discovery, Acheson completed the development of an industrial process for the production of SiC powders in ton quantities.  This “Acheson Process” is still in use today.  It involves heating silica sand and coke in an electric furnace.  The initial application of heat ignites a self-sustaining endothermic reaction, which results in reactor temperatures which often approach 2600°C.  Relatively pure SiC is produced via this process — often in very large crystals that need to be milled down to a more manageable size.

This process is usually simplified SiO2 + 3C → SiC + 2CO.  But it contains two discrete steps that proceed in sequence, so a more accurate description would be:  

SiO2 + C → SiO + CO

followed by

SiO + 2C → SiC + CO

What the above makes clear is that most of the carbon utilized in the process is liberated in the form of CO gas.  We’ll come back to this later. 

Methods for purifying synthesized SiC also date back to the 19th century.  Along with the coke and silica, the Acheson process occasionally also uses salt (NaCl) and sawdust, albeit in much smaller quantities. The salt produces free chlorine, which combines with metal impurities to render them volatile and gaseous, and the sawdust forms channels that allow vaporous impurities and reaction products, like CO gas, to leave the growing SiC ingot.  These materials – and similar ones that are used to the same ends – are still in use today.

In yet another serendipitous discovery on Acheson’s part, it was quickly realized that the Acheson process for the production of silicon carbide can, with a very minor modification, result in an industrial-scale method for the manufacture of pure graphite.  For if the process’s reaction is forced to proceed at a higher temperature — over roughly 4150°C — the silicon itself is vaporized and high-purity graphitic carbon crystals are left behind.     

The silicon carbide and graphite produced via these processes were both immensely valuable to the burgeoning industries of the late 19th and early 20th centuries, though in opposite roles:  Silicon carbide proved indispensable as a hard abrasive powder in grinding and machining media, whereas graphite found many dozens of uses as a solid lubricant. 

Yet although the Acheson Carborundum factory was producing 500 tons of SiC per year by the late 1890s, and although silicon carbide powders were a cheap and common industrial commodity in the decades that followed, dense silicon carbide ceramic parts were extremely rare until the mid 1970s.  

The conventional manner of manufacturing ceramic parts was then, and remains today, pressureless sintering.  This involves pressing ceramic powders to shape, with or without a binder, and then heating those porous parts in a furnace, which bonds the ceramic particles to each other and burns off any porosity along with any binder which might have been utilized.  During this process, as they’re heated in the furnace, the porous ceramic preforms shrink– though, fortunately, in a controllable and predictable way.  If you’ve ever made clay pottery, you should have a generally correct understanding of the sintering process.

Unfortunately, pure silicon carbide has extremely poor sinterability, and was long considered “unsinterable.”  Applied to SiC, conventional pressureless sintering processes invariably result in parts with poor density and high porosity. Porous ceramic parts of this sort are generally extremely weak and extremely fragile — they are never useful towards armor applications and are rarely useful in any other role.  

There are very few other ways to manufacture ceramic parts.  Though hot-pressing can produce dense silicon carbide ceramics from SiC powders with the addition of 2% Al2O3, hot-pressing was, and remains, a slow, expensive, intrinsically limited, and low-throughput procedure — and, for just these reasons, it was and remains enduringly unpopular.  Moreover, growing very large SiC crystals in a controlled way was a severe challenge in those days, so that for all practical intents and purposes it was effectively impossible.

So, virtually without exception, early SiC was sold in powder form, or as powder embedded in a binder — for e.g., in sandpaper and various “bonded abrasive” synthetic whetstones.  The binder is frequently an organic resin.

Sintering experiments were not uncommon, and by the mid 1960s there was some limited success in sintering eutectic B4C-SiC blends at 2400°C, but only densities of up to 80-90% were attainable via this method, and this degree of 10-20% porosity is generally below any reasonable utility threshold.  Aluminum and aluminum compounds were also common experimental sintering aids at this time, and although they facilitated the production of hot-pressed SiC, they were unsuccessful as additives in pressureless sintering experiments. 

Things changed rapidly in the mid 1970s.  In 1974, General Electric scientist Svante Prochazka developed a low-cost sintering technique for silicon carbide utilizing boron and carbon powders as sintering aids.  Both are required; if only carbon is used, or if only boron is used, no sintering occurs.  And yet very small amounts of both are required — the optimal amounts being just 0.5% C and 0.34% B.  The temperatures required for sintering were lower than in previous experiments, at roughly 2100°C.

Pressureless sintered SiC was rapidly adopted following this discovery.  By 1976, it was already in military use.  By the early 1980s, sintered SiC had become a mundane material with dozens of applications in industrial and consumer goods, e.g. in oven heating elements and fishing rod guides. 

Various refinements were subsequently made to the General Electric B+C SiC sintering method.  In the 1980s, it was shown that the addition of aluminum in addition to boron and carbon promoted more complete densification and a finer grain structure, resulting in “ABC-SiC” which is today among the most common commercial grades, e.g. in Saint Gobain’s “Hexoloy SA.” At around the same time, Y2O3 was also shown to be a highly effective sintering aid, and SiC with yttria and alumina additions is commercially popular.  To this day, these variants remain the most popular commercial grades of silicon carbide.

Going back to production, recall that the synthesis of SiC powders generates large quantities of CO gas.  In fact, one liberates two molecules of carbon monoxide for every molecule of SiC one produces.  This sort of “dirty” industrial process is, in a political and regulatory sense, deeply unpopular in the West.  Such things have been frowned-upon for decades; deindustrialization and the offshoring of heavily polluting industries have become standard practice.  So although the USA was once the world’s primary producer of silicon carbide powders, today it accounts for less than 4% of global production.  I don’t suppose it will come as a surprise to learn that China has become the global center of SiC production — indeed, as of this writing, producing more than the rest of the world combined.  So a footnote must be added to the history of silicon carbide: Though developed and pioneered by American businesses, the production of SiC has been offshored.  This is a situation we are working to rectify.

As a footnote, silicon carbide is a rare, almost singular, example of a mineral that was found in nature after it was discovered synthetically.  In 1893, just two years after Acheson’s synthesis of the artificial version, French scientist Henri Moissan found hard, gem-like, and transparent stones in a meteorite impact crater.  He at first mistakenly assumed that these were impact diamonds, and only years later discovered that they were crystals of silicon carbide.  Natural silicon carbide was named Moissanite after its discoverer.  But “natural” silicon carbide exists only in very minute quantities on Earth, as non-oxide minerals tend to be unstable and their formation is disfavored.  Silicon carbide is far more common in space, where it’s formed by supernovae, found in comets and meteorites, exists in the crusts and mantles of carbon-rich and oxygen-poor planets, etc.

_________________________

Silicon Carbide Structure and Basic Mechanical Properties

Silicon carbide is made up of two light elements that prefer covalent, sp3 bonding.  Each Si atom is tetrahedrally bonded to four carbon atoms, and each carbon atom is similarly bonded to four silicon atoms, so it can rightly be said that SiC is made up of alternating C4Si and Si4C units.  The distance between carbon and silicon atoms is 0.189 nm, and the distance between adjacent carbon atoms is 0.308 nm

This layered, alternating structure can exist in two crystalline forms: Beta-SiC, which is a cubic zinc-blende (sphalerite) structure, and alpha-SiC, which is a catchall term for all of the hundreds of known hexagonal or rhombohedral polytypes.  These types are characterized by how their layers are stacked — which largely amounts to a sphere packing problem — with three possible layer/stacking variants, denoted A, B, and C.  Hexagonal forms are further indicated by an H, rhombohedral forms by an R, and the cubic form by a C.  

Beta-SiC is 3C-ABCABC – 3 repeating layers, cubic, and those repeating layers are in a simple ABC arrangement.

The simplest form of hexagonal alpha-SiC is 2H-SiC, which is a simple ABAB… zig-zag structure, also known as the wurtzite structure.

This can get much more complicated pretty quickly.  6H-SiC, which is one of the most common forms of alpha-SiC, is ABCACBABCACB… — in other words, it consists of repeating [ABCACB] layers.  But this is only one variant of two; there’s a different way to put 6H-SiC together.  And 6H-SiC is relatively simple; six, after all, is a very small number.  There are 2435 possible structures in the 18H polytype, and many thousands of possible structures in the common 21R polytype — the very vast majority of which have never been explored or recorded.  21R itself is by no means the limit; there are known 400H and 1200R polytypes.  The full range of SiC structures is in fact theoretically unbounded, and only a small fraction of SiC’s structural space has been explored.  A lot of it, >400H in particular, looks pseudo-random.

The cubic (3C) beta form of silicon carbide transforms to alpha-SiC at temperatures above 1700°C – usually to 4H-SiC and 6H-SiC, but frequently also to 10H, 15R, and 21R – and, as a rule, sintering temperatures for SiC ceramics are >1700°C, so, to a fair approximation, all of the SiC ceramics used in armor are some form of alpha-SiC.  If very high sintering temperatures are employed – over 2200°C – beta-SiC grains can transform into elongated, rod-like grains of hexagonal alpha-SiC, which is a strategy that can potentially improve fracture toughness.

Different impurity concentrations seem to favor the formation of different polytypes.  For instance, if aluminum impurities are found at less than 0.05% by weight, the 6H polytype appears to be favored; if aluminum is in the 0.05-0.06% range, the formation of 15R is much more likely; at >0.1% aluminum, 4H is most commonly observed.  (Aluminum impurities in bulk silicon carbide ceramics like those intended for armor are usually present at well over 0.1%, so the most common hexagonal polytype is 4H.)  

Different polytypes can, in pure form, exhibit different colors and opacities – though, in practice, all grades of polycrystalline SiC are usually dark gray or black due to the presence of iron impurities.  This is true even in the case of cubic beta-SiC:  Sintered beta-SiC is almost always black and opaque, though in pure form it is as colorless and as transparent as diamond.  (And indeed moissanite, SiC, gemstones are frequently colorless and transparent.)  Further, polycrystalline alpha-SiC made up of mixed polytypes is usually black regardless of impurity content.  

It’s possible to have mixed polytypes within a single grain of SiC.  For instance, a grain of 4H-SiC could contain a couple layers of 2H-SiC, and then return to its 4H structure.  This phenomenon is called a “stacking fault,” and is common to virtually all crystalline materials, including metals such as aluminum alloys and steel.  

Stacking faults can form upon ballistic impact, as can “twins” which are, practically speaking, large stacking faults with a characteristic mirrored structure that only form in response to pressure or shear stress.  

Stacking fault formation and twinning can be important mechanisms for energy absorption – which makes sense on an intuitive level, as it means that the crystal structure has slipped or compressed in a dynamic fashion under load – so it has been surmised that hexagonal SiC polytypes that form stacking faults and twins at a lower energy might exhibit better performance.  Better performance via an enhanced capability for kinetic energy absorption, potentially some measurable degree of high-rate ductility, or some combination of the two.  This remains an untested theory, but it follows from well-established metallurgical practice, where the same phenomena are very common and have been characterized in detail. 

Whether alpha-SiC exhibits better ballistic performance than beta-SiC is also something of an open question.  Certainly twinning and slip are more common in alpha-SiC, but it seems that beta-SiC may have slightly superior static mechanical properties, such as hardness and fracture toughness. Beta-SiC also exhibits a phase transformation at pressures of about 105 GPa, as it compresses to a rocksalt structure with a 20.3% reduced volume and correspondingly increased density.  This high-pressure rocksalt phase is metastable in the 35-105 GPa pressure range, and is reversed when pressures are withdrawn to levels below 35 GPa.  Phase transformations are generally associated with improved ballistic performance, for a wide range of reasons, but 105 GPa is well above most ballistic impact pressures, especially from small arms.  The boron carbide amorphization problem occurs at ~20-25 GPa.

All of the above aside, there is, as yet, no known relation between any silicon carbide structural type and silicon carbide’s ballistic performance, but that this topic is fertile ground for various theories and conjectures.

Overall, in a general sense, sintered silicon carbide will have the following mechanical properties:

Density: 3.16 gm/cc
Hardness (Vickers):  2300-2600 HV1
Fracture toughness:  2-4 MPa*m^1/2
Compressive strength:  3500-4900 MPa

Commercial grades that are sintered with an excess of sintering aids, or excess carbon, can exhibit slightly reduced mechanical properties.  

In the armor space, SiC is usually in competition with alumina on the low-end and boron carbide on the high-end.  So, to put SiC’s numbers into perspective, alumina’s mechanical properties are generally:

Density: 3.98g/cm3
Hardness (Vickers):  1200-1600 HV1
Fracture toughness:  3-5 MPa*m1/2
Compressive strength:  >2000MPa

And boron carbide’s are: 

Density: 2.51 gm/cc
Hardness (Vickers):  2400-3100 HV1
Fracture toughness:  2 – 4 MPa*m^1/2
Compressive strength:  ~3000 MPa

In comparison with alumina, SiC has superior hardness and toughness at a lower density.  Its mechanical properties are broadly comparable with boron carbide’s, though boron carbide is roughly 15% less dense.  Boron carbide’s achilles heel, which was alluded to earlier, is that its molecular structure begins to break down at pressures exceeding roughly 20 GPa.  This renders boron carbide absolutely unsuitable for heavy vehicular armor, and even for body armor designed to withstand potent threats such as those with cemented carbide or depleted uranium cores.  SiC, in contrast, actually becomes stronger and harder at high pressures.  So, all things considered, SiC is today’s ideal “conventional” ceramic armor material.  It is substantially lighter and stronger than alumina, and, though slightly heavier than boron carbide, it has none of boron carbide’s vulnerabilities.  There are unconventional research-grade ceramics that outperform SiC, but they’re not commercially available, and arguably not viable, as of this writing. 

Properties of Silicon Carbide

Silicon carbide has an extraordinary combination of useful properties, which are summarized below. 

Hardness

The bulk hardness of polycrystalline silicon carbide samples is generally in the 2300-2600 HV1 range, placing it high among the hardest commercially-available ceramic materials.  

Rather interestingly, many other carbide ceramics are in the 2000-2700 HV hardness range.  This includes WC, ZrC, NbC, HfC, and Be2C.  Mo2C and TaC are only slightly softer at roughly 1800 HV.  Titanium carbide, TiC, is somewhat harder than other metal carbides at 2800-3200 HV.  Boron carbide is slightly harder still, peaking at around 3400 HV.   Ultimately, of binary carbide ceramics, boron carbide has the greatest potential for hardness, followed closely by titanium carbide, which in turn is followed closely by silicon carbide in the third spot.

It should be noted that although high-quality grades of pure or single crystal boron carbide can exhibit superlative hardness values, most commercial grades of boron carbide are comparable in hardness to commercial grades of silicon carbide – they’re not significantly harder and in some cases are softer.  This is due to the sintering aids and process parameters that were required for the production of these grades.

Carbides are very hard materials in general – far harder, as a class, than the oxide and the nitride ceramics.  In fact, there’s no oxide that exceeds 1800HV.  

All carbides are substantially softer than cubic boron nitride (cBN) and diamond, which respectively have hardness values of approximately 5000 and 7500 HV.  However, cBN and diamond are metastable and totally unsinterable at any pressure from ambient through 1,000 MPa.  Indeed, they require 2-6 GPa of pressure for densification.  Industrial hot-presses for ceramic production don’t produce nearly as much pressure; they tend to peak at vastly lower pressures of around 50±10 MPa.  That such extreme high pressures are required for the densification of bulk diamond and cBN severely constrains the industrial utility and commercial potential of ceramic-like parts from those materials.

Alongside boron carbide and titanium carbide, SiC is the hardest material that can be densified without the use of extreme high-pressure apparatus.  This fact helps to explain the popularity of all three materials in industrial and military applications.

In single crystal experiments, on the (100) plane, 3C, 4H, and 6H SiC polytypes were compared, and it was found that 3C at 2500-3000 HV is slightly harder than the others. 4H measured in at 2600, and 6H at 2000-2500.  Other mechanical properties varied little across types, though the 4H and 6H polytypes had a measurably higher Young’s Modulus at 347 GPa vs. 315 GPa for 3C.

Impurity Content and Type

As SiC is a synthetic material, commercial grades of silicon carbide powder are usually highly pure, and even low grade material is typically in the 98-99% purity range.  The most common impurities, which together frequently exceed 1% by weight, are iron and aluminum.  

This is in contrast to most natural oxide ceramics. Low grade alumina, for instance, is often just 85% pure, with the remainder being various other species of oxide, typically SiO2. 

Mixed-polytype alpha-SiC is usually gray or black.  But, quite interestingly, beta-SiC is very different, and you can determine its impurity load from its color.  Absolutely pure 3C-SiC is colorless and transparent.  As it picks up small amounts of iron and aluminum impurities, it turns yellow and then green, and grows increasingly opaque.  As it picks up more metallic impurities, it turns an opaque and iridescent dark blue, and then gray/black.  The first crystals that Acheson saw were blue, and were likely an iron-rich and aluminum-rich form of beta-SiC, particularly as he was running his experiment with carbon particles suspended in iron-rich and especially aluminum-rich clay.  The SiC powders utilized in armor ceramic production are almost invariably black, because impurities are rarely very tightly controlled, and because mixed-polytype alpha-SiC powders are generally the norm.  Green beta-SiC powders of higher quality are not uncommon, but are rarely intended for use in armor ceramic production.

Thermal Stability

Silicon carbide is highly resistant to extreme temperatures.  In vacuum or air, it decomposes at 2400°C, though may begin to exhibit microscopic signs of degradation at 2000°C.  In a nitrogen atmosphere, SiC will begin to absorb nitrogen at 1400°C and begin to degrade at 1450°C; for, in a nitrogen atmosphere, Si3N4 is considerably more stable than SiC, and will form per the following equation:

3SiC + 2N2 = 3C + Si3N4 

Which equation is complicated only slightly if an SiO2 surface layer is present. 

SiC’s high thermal stability have made it very useful in metallurgical applications, for instance as a crucible material for molten alloys.  It was also very popular as a heating element – in fact, one of the first niche uses for SiC ceramics was in high-performance heating elements – though in this category it has largely been replaced by the intermetallic silicide MoSi2.  

Thermal Conductivity

SiC generally exhibits outstanding thermal conductivity, and it has every trait that one would look for in a high-thermal conductivity material:  Light constituent atoms with a low atomic weight differential, a very high elastic modulus, and a high sonic velocity.  

Thermal conductivity in ceramics is very sensitive to porosity, impurity content, crystal grain boundaries, and crystal defects.  Further, it varies by polytype.  With cubic 3C-SiC at  λ = 360 W/(m∙К), hexagonal 4H-SiC at λ = 370 W/(m∙К), and hexagonal 6H-SiC at λ = 490 W/(m∙К). 

Ultimately, the thermal conductivity of SiC at room temperature can range from roughly λ = 30-40 W/(m∙К) to λ = 490 W/(m∙К).  The former derived from polycrystalline sintered SiC ceramics of fairly low quality, whereas the latter value comes from pure single-crystal hexagonal SiC.  Typical polycrystalline material will come in at around λ = 100 W/(m∙К).

On the whole, however, SiC’s thermal conductivity is good.  Commercial grades of polycrystalline SiC far outpace competing ceramic and metallic materials, e.g. in this chart from Kyocera: