TL;DR: Super duplex 2507 has roughly 2.7 times the yield strength of annealed 316 and a pitting resistance number (PREN) of about 41 to 42.5. It costs more, welds less forgivingly and has temperature limits, so use it when the environment is hard, not by default.
The situation: in my thread on 304 versus 316 near the sea I said that if the part is permanently wet and warm, I would stop arguing about 304 and 316 and look at duplex. Here is the case, with numbers.
Strength
The 2507 data in our archive (Alleima tube) gives a yield of 550 MPa at 20 C, tensile 800 to 1000 MPa, elongation 25 % and hardness 28 to 32 HRC. Annealed 316 datasheets list 205 MPa yield and 515 MPa tensile. That is about 2.7 times the yield, which can mean thinner walls for the same pressure, or a stronger part at the same thickness.
Corrosion
PREN is about 41 to 42.5 in the same sources. PREN is a ranking index, not a guarantee, but a number like that puts 2507 in a different league from 304 and 316 for chloride pitting and crevice attack. That is why duplex shows up in subsea, desalination and chemical plant.
Other properties
Thermal conductivity is low, 14 W/m.K at 20 C, similar to austenitic stainless. Thermal expansion is lower: 13.5 per million per kelvin between 30 and 100 C against 16.5 for 316 on the UpMet sheet, which helps when stainless is bolted to carbon steel. Stiffness is 200 GPa at 20 C and 180 GPa at 300 C.
The catches
- Temperature. Duplex stainless steels are usually kept below roughly 250 to 300 C in service because long exposure above that can embrittle them. Check the datasheet for your case.
- Welding and fabrication. Duplex is less forgiving than 316. Heat input and filler choice matter, and a poor weld can throw away the corrosion advantage. Not every shop is set up for it.
- Cost and supply. It costs more per kilogram, though thinner walls sometimes pay that back.
How I would decide
If the part is only splashed, the water is cool and the budget is tight, 316 with good design may be enough. If it is immersed, warm, under deposits, in a crevice you cannot design out, or costly to repair, start with duplex and turn the question around: what would it take to justify the cheaper grade?
For anyone who has specified or welded duplex: what was the most expensive surprise, fabrication, supply, or something about the environment you did not see coming?
TL;DR: A36 is a cheap, weldable, forgiving structural steel with a guaranteed minimum yield of 250 MPa. It is heavy for its strength and it rusts. In anything that moves or flies it loses; in a frame or a base plate it is hard to beat.
The situation: A36 is the default structural steel in much of the world, and its datasheet is refreshingly simple. In our archive: minimum yield 250 MPa (36 ksi), tensile 400 to 550 MPa (58 to 80 ksi), elongation 18 to 21 % (SSAB), density 7.8 g/cm3, stiffness 200 GPa. What should you read into that?
It is a minimum-yield steel, not a performance steel
The yield figure is a floor the mill has to meet, and the plate on your bench can come in above it. If your strength margin is thin, a floor is not a promise about your plate, so ask for the certificate.
For its strength it is heavy
Compare 6061-T6 aluminium in the same archive: 276 MPa yield at 2.7 g/cm3. A36's yield is a little lower than that, at nearly three times the density (7.8 against 2.7). In a bridge or a building frame weight is not the point, and A36 wins on cost, availability and weldability. In anything that moves or flies, it loses badly.
The elongation matters
18 to 21 % means the steel stretches visibly before it lets go, which is a big part of why structural codes are comfortable with it: an overloaded member usually gives you a warning in the form of a bend instead of a sudden break. A harder, higher-strength steel can trade some of that forgiveness away.
It rusts
A36 is plain carbon steel, so it needs paint, galvanizing or a protected environment. One detail I find useful: A36's 250 MPa yield is higher than annealed 304 and 316 stainless (205 MPa each in our tables). When you pay more for stainless you are paying for corrosion resistance and non-magnetic behaviour, not for strength.
Where I would use it: frames, base plates, brackets, fixtures and welded structures where mass and corrosion are handled some other way. Where I would not: weight-critical parts, anything wet and unpainted, and anything that needs guaranteed toughness at low temperature without checking the details first.
What is the most surprising place you have seen A36 used, or misused?
TL;DR: A melting point is a number for the periodic table, not a design rule. Tungsten is very dense, hard to make and shape, brittle at room temperature, and oxidises in air when hot. Where it works, the environment is controlled and the mass is either a feature or affordable.
The situation: tungsten has the highest melting point of any metal, 3422 C or 3695 K in the sources in our archive. So why is not every rocket nozzle, furnace and re-entry part made of it?
The numbers
Density is 19.25 g/cm3, more than four times Ti-6Al-4V (4.43 g/cm3) and about two and a half times steel. To make that concrete: a 10 mm cube of tungsten weighs about 19 g, the same cube of Ti-6Al-4V about 4.4 g. Elastic modulus is 411 GPa, about twice steel's. Thermal conductivity is 174 W/m.K, thermal expansion is very low at about 4.5 millionths per kelvin, and Vickers hardness of the pure metal is listed at 3.43 GPa.
Why not everywhere?
- Weight. In anything that flies, density is a tax on every kilogram of launch mass. A tungsten part has to justify being four times denser than titanium.
- Making and shaping it. With that melting point you cannot cast it the usual way. It is made by sintering powder and is hard to machine, so parts are expensive and shapes are limited. It is also known to be brittle at room temperature, so you design around it.
- Oxidation. Tungsten oxidises in air at high temperature (general knowledge), which is why it is so often used in vacuum, in inert gas, or under a coating. A material that melts at 3422 C is only useful at 2000 C if the atmosphere lets it survive.
Where it does win
Places where mass is a feature or the temperature is extreme and the environment is controlled: radiation shielding and ballast, electron-beam and plasma-facing parts, and nozzle throat inserts in some rocket designs. That last one is from general knowledge, not from our data.
My take: the real design question is "what does it cost me in weight, in manufacturing and in atmosphere to keep it at that temperature?" For most hot parts the answer is that a coated alloy, a carbon composite or a ceramic is the better compromise.
Which hot-section jobs do you think tungsten is genuinely the right answer for, and where is it more of a showpiece than a solution?
TL;DR: At room temperature annealed Ti-6Al-4V is not dramatically stronger per kilogram than 7075-T6. It wins on temperature, corrosion and strength in small sections, and it loses on heat conduction, stiffness and cost.
The situation: titanium Ti-6Al-4V shows up in airframes, engines, fasteners, spacecraft hardware and implants, and the usual explanation is "it is strong and light." True, but incomplete, so I ran the numbers from our archive.
The numbers (Boze datasheet)
Density 4.43 g/cm3. Yield strength 830 to 950 MPa annealed, 1000 to 1100 MPa solution treated and aged. Elastic modulus 113.8 GPa. Melting range 1604 to 1660 C. Thermal conductivity only 6.7 W/(m.K) at 20 C.
Strength per density
Divide yield by density: annealed Ti-6Al-4V is about 187 MPa per g/cm3, and the aged condition 226 or more. 7075-T6 aluminium (Kaiser data, 503 MPa over 2.8) comes out at about 180. A quenched and tempered 4140 steel (740 MPa over 7.85) is about 94, and 6061-T6 (276 MPa over 2.7) about 102. So titanium clearly beats steel and 6061, but at room temperature it is not far ahead of 7075-T6.
So why titanium?
- Temperature. Aluminium alloys lose strength quickly as they warm up, while titanium keeps working where aluminium is out of the game. That is general engineering knowledge, and it is why engine and hot-structure parts go titanium.
- Corrosion. It tolerates seawater and many chemicals, which is also why it is the implant material of choice.
- Strength in a small section. Aged titanium reaches 1000 MPa and more, which is hard to match with aluminium when space is tight.
The downsides people skip
- Heat. 6.7 W/(m.K) is far below aluminium (167 W/(m.K) for 6061), so heat stays where it is made. That hurts electronics enclosures and makes machining hard, because heat piles up at the cutting edge.
- Stiffness. 113.8 GPa is a bit more than half of steel's roughly 200 GPa and about 1.6 times aluminium's, so a titanium part can flex more than a steel part of the same size.
- Cost. Expensive to buy and to machine.
My rule: if the part runs hot, sits somewhere corrosive, or must carry high loads in a small section, titanium makes sense. If it is a room-temperature bracket where 7075 or even 6061 would do, titanium is often a status choice rather than an engineering one.
Where have you seen titanium chosen when aluminium would have done the job, or the other way round?
TL;DR: At the same strength class the 4140 and 4340 numbers in our datasheets are the same. The real differences are hardenability in thick sections and the strength range the grade is sold in. Choose the condition, not just the grade name.
The situation: someone says "use 4340, it is better than 4140", someone else says it is the same thing for half the price. I went to the datasheets in our archive to see who is right.
At the same class they match
Voestalpine's 4140 sheet lists two quenched and tempered classes: 665 MPa yield with 850 to 1000 MPa tensile, and 740 MPa yield with 930 to 1080 MPa tensile. Elongation is 12 %, impact energy 42 J, hardness 269 to 331 HB. The 4340 sheet uses lettered conditions, and T and U are the same two classes: 665 and 740 MPa yield, 850 to 1000 and 930 to 1080 MPa tensile, elongation 13 % and 12 %, impact energy 35 J and 42 J. Compare those lines and you cannot tell the steels apart.
But the 4340 sheet keeps going
Conditions V, W, X, Y and Z push yield through 835, 925, 1005, 1080 and 1125 MPa, with tensile up to 1550 MPa. And watch what happens to toughness: elongation falls from 13 % to 5 %, and impact energy falls from 42 J to 9 J. Same grade, completely different part. The sheet also marks the top conditions with an asterisk, noting such bars are typically ordered annealed for machining and then heat treated.
So where do they really differ?
4340 has nickel on top of chromium and molybdenum, and nickel lets steel harden deeper, so the core of a thick section can match the surface. That is general metallurgy, not a number from our tables. In a thin bar the two behave alike. In a big shaft or forging, 4340 is the safer way to get uniform properties through the section. 4140 is the cheaper, easier-to-find workhorse for axles, gears, tooling and general machinery.
How I would choose
1. Section size: thin or medium bars, 4140 is usually enough; thick sections with high strength demands, look at 4340.
2. Strength class: decide the yield and toughness you need, then find the condition that gives both. The highest condition is not the best one, as the impact numbers show.
3. Heat treatment: these values belong to defined quench and temper routes. If your heat treater cannot hold that route on your section size, the grade difference is academic.
4. Cost and supply: 4140 wins unless section size forces your hand.
Comparing two steels at different conditions is how people convince themselves one grade is stronger. Compare the same condition and the same section.
For people who machine or heat-treat these every day: at your sizes, do you actually see a difference, or is it mostly habit and whatever the supplier stocks?
TL;DR: 7075-T6 has about 80 % more yield strength than 6061-T6, but the same stiffness, less ductility, lower thermal conductivity, and it is usually not welded. Start with 6061 and only move up when strength is the real limit.
The situation: you are designing a bracket or a frame and the numbers make 7075 look like a free win. Our archive lists 7075-T6 at 503 MPa yield and 572 MPa tensile (Kaiser, tube and pipe) against 276 MPa and 310 MPa for 6061 in T6 and T651 (ASM), with a density of 2.8 versus 2.7 g/cm3. Why would anyone pick 6061?
Stiffness does not change
Elastic modulus is about 71 GPa for 7075 and 68.9 GPa for 6061. If a part fails because it flexes too much, the stronger alloy will not make it stiffer. Strength stops yielding and breaking; it does nothing for deflection. Plenty of designs are stiffness-limited, and for those the extra money for 7075 buys nothing.
Ductility goes the other way
Elongation is 11 % for 7075-T6 and 17 % for 6061-T6 in these tables. The stronger alloy forgives less before it breaks, which matters in fatigue-sensitive or impact-prone parts and anything you bend.
Heat flow
Thermal conductivity is 130 W/m.K for 7075-T6 and 167 W/(m.K) for 6061. If the part carries heat (a heatsink, a bracket under an electronics box, an enclosure wall), 6061 moves it noticeably better.
Welding and forming
6061 is the one people weld, and it machines and anodizes easily. 7075 is generally treated as a no-weld alloy because it tends to crack in the heat-affected zone, and it is more sensitive to stress corrosion in some tempers and grain directions. That is general industry experience, not a number from our tables, so check the temper and orientation for your part.
The T6 trap
These numbers are for T6. Strength depends on temper and thickness, and the Kaiser figures come from tube and pipe. Plate and thick sections can differ, so check the temper and form on the certificate.
What I would do
Start with 6061 and calculate the margin. Move to 7075 only when the shortfall is genuinely strength, not stiffness, heat or weld access. 7075 earns its price in highly loaded, weight-critical machined parts such as aerospace fittings, structural brackets and tooling plates.
Anyone swapped 6061 for 7075 and found it did not fix the problem? I would like to hear what the real limit turned out to be.
TL;DR: On paper the strength is the same (205 MPa yield for both, annealed). The extra money for 316 buys molybdenum, which helps against chloride pitting. It does not make 316 immune to seawater, and for permanent immersion I would look at duplex.
The situation: you are speccing stainless for something near the coast or on a boat, and every supplier says "use 316, it is marine grade." Is it really worth the premium over 304? Here is how I think about it.
Strength is not the difference
In the annealed condition the datasheets in our archive put them on top of each other. 304 plate: 205 MPa minimum yield, 515 MPa tensile, 40 % elongation, about 92 HRB (Outokumpu). 316: 205 MPa and 515 MPa as well (ATI, UpMet), 40 % elongation, 95 HRB on one sheet. Density is about 8 g/cm3 for both. So you are not buying a tougher steel. You are buying chemistry.
What the premium actually buys
316 contains about 2 to 3 % molybdenum and 304 contains none. Molybdenum helps stainless resist pitting and crevice corrosion in chlorides, and seawater is chloride soup. That is the whole reputation of 316 in marine use.
Where "marine grade" goes wrong
It is not a force field. In warm, stagnant seawater, under deposits, or in tight crevices under washers and gaskets, 316 can pit and crevice-corrode too. Brown staining on a 316 rail is common and usually cosmetic; a pit under a fastener is not. Design details, cleaning, and whether water can sit in a gap often matter as much as the grade.
My rough guide (opinion, not a code)
- Indoor, fresh water, food, inland architecture: 304 is usually fine. Do not pay the premium out of habit.
- Coastal air and splash: 316, with crevices designed out.
- Permanently immersed, warm, under deposits, or costly to repair: look at duplex. Our archive lists super duplex 2507 at 550 MPa yield at 20 C (Alleima tube), more than double annealed 316, with a PREN of about 42.5 on the same sheet. It is a different decision, not a free upgrade: more cost, tighter welding practice, temperature limits.
If you do use 316 near the sea
Avoid crevices where you can. Rinse salt off with fresh water now and then, because dried chloride keeps working. Clean and passivate welds, and keep carbon-steel grinding dust off the stainless, because embedded iron rusts and makes the steel look like it is failing.
Where I might be wrong: datasheets give mechanical properties, not corrosion rates. Temperature, chloride level, surface finish and welds change everything, so treat the grade as one input among several.
Has anyone watched 316 fail in seawater? Was it the grade or the detail around it?
TL;DR: In the annealed condition 316L is essentially non-magnetic. But a magnet can still stick to a perfectly good 316L part (cold work, welds), and a magnet cannot prove a part IS 316L. Use a certificate, not a fridge magnet.
The situation: you order 316L sheet or tube, a magnet clings to a bent edge or a weld, and the first thought is "I was sold the wrong steel." This comes up constantly, so here is what is actually going on.
Why 316L normally is not magnetic
316L is an austenitic stainless steel. Austenite does not respond to a magnet the way carbon steel does, which is why people choose it near sensors, instruments and vacuum hardware. As supplied it is soft and ductile: our archive lists a density of 8.03 g/cm3, 170 MPa minimum yield, 485 MPa tensile and 40 % elongation (ATI plate, sheet and strip).
Three reasons a magnet can still stick
- Cold work. Bending, drawing or heavy machining can turn some austenite into martensite, which is magnetic. A tight bend can attract a magnet while the flat plate does not.
- Welds. Weld metal often contains a little ferrite on purpose to resist cracking, and ferrite is magnetic. A magnet that grabs the bead but not the plate next to it is normal.
- It is not 316L. A magnet cannot tell 316L from 304, and ferritic or martensitic stainless steels are strongly magnetic. A magnet test proves very little either way.
What I would actually do
If magnetic response matters (sensors, vacuum chambers, sensitive instruments), skip the argument about the grade name and ask for the relative permeability of the part as delivered and after forming and welding. Put the limit on the drawing. If the grade matters, get the mill certificate or a composition check.
L or no L?
The L means low carbon, which helps after welding, and it costs a little strength: our datasheets list 316 at 205 MPa minimum yield against 170 MPa for 316L. For most non-magnetic jobs that gap is not what decides it.
Where I might be wrong: how magnetic a cold-worked part gets depends on composition and how hard you deform it, so I cannot give you a number for a given bend. Real measurements beat my generalisations.
Have you had a 316L part surprise you with a magnet? Was it cold work, a weld, or a mislabelled bar?
What Metalysis reports: the UK powder maker says it has produced C-103, a niobium-hafnium-titanium alloy powder, to commercial specification. It makes the powder with its FFC electrolysis process, which reduces metal oxides rather than melting metals together. Oxygen content is stated as below 500 ppm, and the powder can be spheroidised in-house for additive manufacturing. The intended uses are rocket nozzles and thrust and combustion chambers, plus hypersonic, satellite propulsion and turbine parts.
Why it matters for materials: C-103 is a refractory niobium alloy for high-stress rocket parts such as nozzles, and powder is the raw material for printing them. A different way of making powder can ease supply, since the company says its route is not constrained by hafnium metal, but it still needs hafnium oxide.
Our take: a commercial specification applies to the powder, not to printed or qualified hardware; the release gives only the oxygen limit and says the full specification is available on request. No customers are named, and the market figures are unsourced: the stated growth rate of about 32% is higher than the 28% a year that $46m to $200m gives over six years (23% over seven, 34% over five), and the release gives no base year, source or method. Our niobium alloys page puts C-103 at 8.85 g/cm3 with a melting point of 2350 °C plus or minus 50, and a minimum yield strength, for fully recrystallised material, of 276 MPa at room temperature (tested in air) that falls to 124 MPa at 1093 °C and 55 MPa at 1371 °C (tested in vacuum) (Smiths High Performance). These are datasheet minimums, not values for powder or printed parts.
What we would watch: the first printed C-103 parts with measured properties. What would you want in a powder spec?
What WSU reports: a team led by computer scientist Jana Doppa used an AI search to choose laser settings for printing GRCop-42, NASA's copper-chromium-niobium alloy for rocket combustion chambers, and says it printed the alloy at 500 W for the first time. The search started from 37 earlier unsuccessful configurations and used a budget of 40 experiments over three months, finding six working settings at different laser powers out of more than 100 million possible settings. The release links the paper in the Proceedings of the AAAI Conference on Artificial Intelligence (DOI 10.1609/aaai.v40i47.41428); the release itself names no process, but the paper's abstract names directed energy deposition.
Why it matters for materials: for an alloy that is hard to print, each trial is costly, so finding a working window with few trials matters. The release says one print can cost hundreds of dollars and checking a sample can take days.
Our take: it shows printability on a research setup, not qualified rocket-grade material. The release gives no strength, porosity or conductivity numbers for the parts. Democratizing and less expensive are not quantified beyond an unsourced claim that ninety percent of commercial printers cannot print this alloy, and the release gives no power figure for other printers and does not say 500 W is within their range, though it implies the new settings could let common commercial printers print the alloy. GRCop-42 has no profile of its own in our archive yet. The nearest copper chamber alloy we cover is CuCrZr, with a thermal conductivity of 320 W/m-K at 20 °C (Luvata) and yield strengths from 233 MPa for a heat-treated printed part (Desktop Metal) to at least 500 MPa for a wrought R540 temper (Aurubis); these are datasheet values, not rocket tests. Those are the kinds of numbers a GRCop-42 print will be judged against.
What we would watch: the first measured conductivity and porosity of AI-found prints. What would you check first?
What Stoke Space says: its first vehicle and the larger Nova Block 2 both use stainless steel skin-and-stringer construction, which it says needs almost no fixed tooling. The reusable upper stage re-enters behind an actively cooled metallic base heat shield with liquid hydrogen flowing through it, which Stoke's article contrasts with passive ceramic tiles. Block 2 is quoted at 15 t to low Earth orbit in a fully reusable configuration, and the page says Nova can deliver more than 4 t to GTO (it names no block and no reuse mode). Pathfinder flies seven Zenith engines; its first flight is planned for the first part of 2027 and Block 2's for 2029.
Why it matters for materials: a cooled metal shield replaces a ceramic insulator with plumbing, so the open question is how the metal and its cooling passages survive repeated re-entry heating, thermal stress and fatigue.
Our take: this is an architecture announcement from the company. No alloy grade, temperatures or reuse life are given, the heat-shield metal is not named, and the comparison with tiles is one-sided and unsourced. Almost no fixed tooling and all flight dates are company claims or targets. For context, our type 301 stainless page (Stoke names no stainless grade) lists a thermal conductivity of 16.3 W/m-K at 20 to 100 °C (annealed, ATI), while our CuCrZr page lists about 320 W/m-K for a copper alloy at 20 °C (Luvata). We are not saying which metal Stoke uses, because it does not say.
What we would watch: the named shield alloy, coolant flow and any flown data. What would worry you most in a cooled shield?
What TCT Magazine reports: Skyrora ran six hot-fire tests of its 3.5 kN low-Earth-orbit engine at its Midlothian site. The article describes a titanium-coated component and says Skyrora is said to have cut weight by about 44% by using titanium instead of a nickel-based alloy; it does not say who made the claim or what the weight refers to. It says the component was engineered with Skyrora's in-house Skyprint 2 hybrid directed-energy-deposition printer, under the EU-funded MADE-3D programme. No damage to the modified components was reported.
Why it matters for materials: swapping a nickel-based alloy for titanium usually trades high-temperature strength for much lower mass, so it tends to suit a part that sees moderate heat or is protected. That is our inference; the article says only that the part is designed for the engine's demanding thermal environment.
Our take: the saving is close to what the density gap alone would give. In our archive Ti-6Al-4V is 4.43 g/cm3 and Inconel 718 is 0.297 lb/in3, about 8.22 g/cm3, so titanium is roughly 46% lighter for the same volume, for those two grades. Which grades Skyrora used is not stated, and the page calls the part titanium-coated and also speaks of using titanium instead of a nickel-based alloy, so we cannot tell a surface layer from a bulk swap. The 44% has no baseline mass, and test durations, measured thrust, pressures and temperatures are not given. Six hot-fires with no reported damage do not show it is flight-qualified.
What we would watch: which part it is, how hot it runs, and whether it survives repeated firing. Would you trust titanium there?
What UT San Antonio reports: NASA is giving the university $1 million over two years, started in November 2026, through its M-STAR program, to study icy regolith, water ice mixed with lunar soil in and around permanently shadowed regions, and how it behaves when excavated and handled. The team will use a chamber that controls temperature, pressure, atmosphere, humidity and radiation to study how daily temperature cycling changes cohesion, flow, mechanical properties and water movement. Sub-teams will study low-temperature rheology and compaction (hardness and scoopability against ice amount) and the mechanical properties of individual grains under cryogenic conditions, and will develop sensors that survive the cold and measure the chemistry.
Why it matters for materials: simulants made from crushed Earth rock are a common stand-in for lunar soil, but ice may change how a soil sticks, flows and resists digging, which would matter for excavation and drilling hardware. A UT San Antonio researcher says the mix may be hard to dig up; the grant exists to find out.
Our take: this is a university announcement of a grant that has just started. It reports no results, no temperatures, ice fractions or strengths, and does not say whether real samples or a simulant will be used. As a baseline, our lunar simulant page lists JSC-1A with a bulk density of 1.5 to 1.7 g/cm3 (McKay et al., 1994) or 1.57 to 1.64 g/cm3 (Klosky et al., 2000) and a specific gravity of 2.9 (Alshibli and Hasan, 2009), all as cited on the ARES page, which does not say how compacted or moist the samples were. Ice-bearing equivalents, and the cohesion and hardness numbers the team says it will study, are what we would like to see.
What we would watch: whether real samples or a simulant are used, and the first strength-against-ice-content curve. What ice fraction would you test first?
What Xinhua reports: a fourth round of outside-the-station exposure experiments is under way, covering eight projects and 26 sample units, exposed for one to two years to vacuum, atomic oxygen, radiation, sharp temperature swings and micrometeoroids. A stainless magnesium alloy is described as the world's first in-orbit check of its space stability. Solid-liquid composite and gel lubricants reportedly reached near-zero wear, and simulated lunar-soil bricks stayed intact after a year of exposure. The quoted expert, Pang Zhihao, says the results may transfer to aviation, nuclear power and precision transmission parts.
Why it matters for materials: a rolling programme of one-to-two-year exposures is another source of low-Earth-orbit durability data, like MISSE, covering a magnesium alloy, space lubricants and lunar-construction bricks.
Our take: none of the claims comes with property values, test methods or a paper. World's first and near-zero wear are unquantified, the magnesium alloy's composition is not given, and the whole article rests on one expert interview. These are leads to chase, not numbers to cite. For context, our magnesium WE43 page lists a minimum 0.2 % proof stress of 172 MPa, a minimum tensile strength of 220 MPa and a 2 % minimum elongation (Luxfer's guaranteed minimums for a heat-treated WE43 casting). WE43 is a rare-earth casting alloy and Xinhua does not name its alloy, so this shows the type of figure to expect, not a like-for-like baseline.
What we would watch: whether a data release or paper follows, with the alloy named. What would you want reported before you trusted it?
A small satellite just 3D printed a beam in open space. Not inside the space station, where printing has been done before, but outside, in vacuum. The company is Orbital Matter and the satellite is called Replicator 2.
I find this one exciting. Today every solar panel and antenna has to be folded up tightly to fit inside a rocket, then unfolded in orbit. If a satellite can print its own structure once it is up there, that whole problem starts to go away. You launch a small box and it grows its own arms.
The way it works is simple. The printer pushes out a resin in a spiral and ultraviolet light hardens it on the spot. The company says the beam is 30 cm long and took about half an hour to print.
I would be a little careful with the headlines though. TechCrunch wrote that the beam is half a metre and that it already opened a solar array. The company's own page says 30 cm, and calls the solar array the next step. So the print looks real, but some of the bigger claims are ahead of what the company itself says.
The thing I keep wondering about is the plastic. Space is hard on plastics. Sunlight, oxygen atoms and the swing between hot and cold wear them down over time. Printing a beam is one thing. A beam that stays straight and strong for years is another, and nobody has shown numbers for that yet. I hope the company publishes them.
Still, for a satellite the size of a suitcase, this is a big step. The next mission is planned for 2028 with a much larger solar array.
What do you think: will satellites print their own parts in ten years, or will this stay a demo?
What Via Satellite reports: Rocket Lab has pushed the debut of its Neutron rocket to next year, 2027, after finding air pockets between the layers of carbon fibre in a propellant dome made on an automated fibre placement machine. This follows a tank that ruptured in testing in January; Rocket Lab said then that the rupture exposed a manufacturing defect in a tank hand-laid by a contractor. The article does not name the contractor or say the hand-laying caused the defect, and the account is Rocket Lab's own. No launch date beyond 2027 was given.
Why it matters for materials: voids between plies are a classic composite-manufacturing defect. They generally lower interlaminar strength and are hard to see without inspection, so they are caught by process control and non-destructive testing, not by the material datasheet. Moving from hand layup to automated placement changes which defects show up; it does not remove them.
Our take: we have only a trade-press account of Rocket Lab's statements, with no void size, fraction or location, and no independent inspection. What our data can add is how directional the material is: our CFRP page lists a unidirectional IM7/HexPly 8552 laminate at 2724 MPa tensile strength along the fibres but only 64 MPa across them (Hexcel data, 25 °C, dry). That is in-plane transverse tension, not interlaminar strength, which our archive does not list, but it shows how much weaker the resin side is, and voids between plies mainly hurt resin-dominated properties such as interlaminar strength.
What we would watch: whether Rocket Lab publishes inspection findings or a revised process, and how the dome is made after the change. How would you qualify a large composite tank?
What CompositesWorld reports: DLR's SpaceMast, a carbon-fibre-reinforced composite mast with a thin-walled double-omega cross-section that flattens into a band and winds onto a spool, completed what DLR researcher Martin Zander calls its first major operational test in orbit, after about five months in space. According to SpaceNews, it launched on 30 March 2026 on SpaceX Transporter-16, carried by D-Orbit's ION Satellite Carrier. The test ran more than 8.5 hours: the mast extended to 1.7 m, then fully retracted and stowed. The mast carries a tip camera meant to image the host vehicle, according to a 2025 DLR conference paper; the article reports no images. More cycles are planned, with extensions up to 2.5 m. DLR says its patented Floating-Core-Interface nearly doubles bending strength.
Why it matters for materials: a thin composite shell is elastically flattened and coiled, then deployed again. That tests whether the fibre and resin system survives repeated stowage strain, which is a different question from how it ages in vacuum, thermal cycling or atomic oxygen. The article gives no data on those.
Our take: this is trade press relaying a DLR researcher's LinkedIn post (he is co-founding a spin-off) and DLR material. Fibre, resin, layup, mass and orbit are not given, and nearly doubles has no numbers behind it. For a sense of how directional a laminate is, our CFRP page lists a unidirectional IM7/HexPly 8552 laminate at 2724 MPa tensile strength along the fibres and 64 MPa across them (Hexcel data, 25 °C, dry), so layup matters a great deal for any thin composite shell; the article does not give SpaceMast's layup.
What we would watch: the cycle count at which stiffness changes, and any thermal data. What failure would you look for first?
What NASA Glenn reports: a team led by research chemical engineer Allison Christy, with summer interns, mixed a biodegradable plastic with simulated Moon dust and simulated Mars dust. Bacteria could make the plastic from crew waste or CO2. Dust type and amount were varied to tune properties, and the page says the dust made the plastic stronger and easier to process, without giving numbers. Samples are in Glenn's Lunar Environment Structural Test Rig, and several are slated for the MISSE-23 exposure outside the space station. A NASA technical-reports record (ID 20260007758) for a talk by the same researcher on regolith composites names the polymer as poly(3-hydroxybutyrate), PHB.
Why it matters for materials: this is a particulate composite, a bio-made thermoplastic filled with mineral regolith. The tests that matter are cold embrittlement over temperature cycles and what vacuum, ultraviolet light and atomic oxygen do to it. The page says the rig tests how the material holds up in extreme temperatures and that MISSE-23 will expose samples to intense conditions outside the station; it does not say which, and reports no results yet.
Our take: the feature page does not name the polymer or give loadings, strengths or a temperature range, so on-demand tools and full recyclability are prospective, and the page does not say whether making the plastic from waste has been done in space. For context, our PHB page lists unfilled P(3HB) with a tensile strength of 34.2 MPa and an elongation at break of 3.8 %, a stiff, brittle baseline. Our page on an 80:20 LHS-1 simulant and PLA printed composite lists flexural strengths of 24 MPa and 16.6 MPa (0 and 90 degrees, independent laboratory).
What we would watch: the first numbers from the test rig and from MISSE-23. What filler loading would you try first?
What DLR reports: on Artemis I in 2022, two torso phantoms, each built from 38 tissue-equivalent plastic discs, flew with dosimeters. One wore the StemRad AstroRad vest and one did not. No significant solar storm occurred, so the team extrapolated the measured Van Allen belt data to past storms. The extrapolation indicates the vest would cut effective dose by about 60% for an August 1972-type event and just under 40% for an October 1989-type event, and in the model a full-body shield of the same mass gave roughly 30% less protection than the targeted vest. The flown vest weighed about 26 kg. The paper is in Science Advances.
Why it matters for materials: hydrogen-rich polymers such as polyethylene tend to stop protons well per kilogram, so the design question is where the mass goes, not only which polymer is used. DLR says only that the vest core is a polymer with a high hydrogen content and does not name it.
Our take: the storm figures are modelled, not measured in a storm, from two female torso phantoms on a single flight, and four of the paper's 21 listed authors have StemRad, the vest's developer, as their affiliation (Crossref record). DLR gives the vest mass (about 26 kg) and notes that mass and volume are generally limiting in spaceflight, but gives no target mass or launch-cost figure. As plain reference values for polyethylene, a common hydrogen-rich polymer, our HDPE page lists densities of 0.959 g/cm3 (SCG H5818J) and 0.955 g/cm3 (machining sheet), and our borated polyethylene page lists 1.04 g/cm3 for one 5 % boron grade (JCPE5) and 1.01 g/cc for another (MarShield).
What we would watch: whether a real solar particle event gives the same ratio, and what the vest weighs after redesign. Would you fly the mass?
What the paper reports: researchers at the Shanghai Institute of Ceramics, writing in the Journal of Advanced Ceramics, made a hafnium-zirconium solid solution with boron, carbon and nitrogen sharing the non-metal sites (Hf0.8Zr0.2B0.1C0.5N0.4) and combined it with silicon carbide. Under a 2600 °C plasma flame the composite showed negative ablation rates of -0.049 and -0.287 µm/s, which the authors say clearly beats a Hf0.8Zr0.2C-SiC baseline. They credit higher fracture toughness, graphite that forms during ablation and arrests cracks, and a staged oxidation that builds a dense HfZrO2-SiO2 barrier.
Why it matters for materials: the authors say existing high-entropy ultra-high-temperature ceramic work is largely confined to the metal sites. This one changes the non-metal side instead, aiming to keep a protective oxide scale and resist cracking in hot plasma.
Our take: we have read only the abstract, so the sintering route, test duration, heat flux and sample size are not known to us. A negative ablation rate usually means the oxide layer grew faster than material was lost, but the abstract does not explain its sign convention, so treat the number as a lead, not a verdict. As far as the abstract shows, it names one new composition and one comparator, Hf0.8Zr0.2C-SiC, in a 2600 °C plasma flame, and it does not say what the two rates refer to, which is a long way from flight performance. For reference, our hafnium carbide page lists melting points of 3890 °C (supplier data) and 4255 K plus or minus 30 K (Zhang and McMahon), about 90 K apart, which our page flags as unreconciled, and the zirconium diboride page lists 3245 °C and a thermal conductivity of 60 W/m-K (Sani et al. 2016).
What we would watch: test duration and heat flux once the full paper is read, and independent repeats. What would you ask the authors first?
What JHU APL reports: the lab developed FAST CAR2, a way to make carbon-carbon, a heat-resistant thermal-protection material, in one step. It introduces a carbon-bearing material into a carbon-fibre scaffold and consolidates it in minutes with field-assisted sintering, which the release calls essentially a rapid, high-current pressing method, instead of the repeated densification cycles used today. A 1-inch cylinder took a few days, under a tenth of the conventional time, and the team then made several 6-inch cylinders and one 8-inch cylinder. Pores were micrometres wide, against up to about 1 mm in industrial carbon-carbon, and the samples passed high-velocity oxygen-fuel heat tests that the release calls on par with industry products.
Why it matters for materials: the bottleneck in carbon-carbon is the slow, repeated step of filling pores with carbon. If one pressing can do it without hurting the microstructure at full part size, lead time and cost drop. That is a process claim, not a new material.
Our take: the release shows no strength, thermal-shock or oxidation numbers against conventionally made carbon-carbon, does not name the carbon-bearing precursor, does not say which samples were heat-tested at what temperature or for how long, and describes density only as comparable, if not superior. Shuttle-type re-entry carbon-carbon is not used bare: our archive page for reinforced carbon-carbon, the Space Shuttle material, describes a silicon carbide outer coating against oxidation, and lists a multi-mission design maximum of 1811 K, with coating recession negligible below 1967 K in arc-jet tests (NASA TM-104792). Any new route will be judged against data like that.
What we would watch: published mechanical and oxidation data, and whether 8-inch cylinders can become real parts. What numbers would convince you?
What NASASpaceflight reports: Ship 41 was rolled to Pad 2 carrying two heat-shield tiles recovered from Ship 40 after Flight 13, fitted at the aft end near the flaps. The article calls it the first time flown Starship tile hardware has been assigned to fly again. A Starship shield has about 20,000 tiles, and Ship 40 spent 24 days in the Indian Ocean after splashdown before it was returned to Starbase.
Why it matters for materials: reuse is the whole point of a ceramic tile shield, and it is a statistics problem. Tiles can fail through cracking, coating wear and lost attachments, and one or two samples cannot show how often that happens across a vehicle. The article notes that where a tile is missing, the area relies on an ablative backup that is not fully reusable, and that SpaceX is using curved tiles and different retention hardware in the areas most prone to tile loss.
Our take: this is a September article written before Flight 14 (targeted for no earlier than 28 September), mostly relaying SpaceX's own documentary episode, with no performance data and no named tile material or coating. We have not found a published inspection of those two tiles. We would call it a sensible first step and nothing more. For a yardstick from a vehicle that did fly many times, the Space Shuttle's LI-900 tile in our archive has a density of 144 kg/m3 (uncoated tile) and a through-thickness thermal conductivity of 0.0476 W/m-K at TPSX standard conditions (printed uncertainty about 18 %), both from NASA's TPSX database. Starship's tile material is not named in the source, so we are not comparing the two.
What we would watch: a published inspection of those two tiles, in particular the coating condition and whether any tiles were lost. What would you want measured on a reflown tile?
What NASA reports: after the Artemis II flight, engineers found 9 spots of char loss on the Orion heat shield, against 103 on Artemis I. The shield is 186 blocks of Avcoat, an ablative material, and the capsule splashed down on 10 April 2026. NASA credits a changed reentry path, with a shorter maximum downrange distance, which it says limits the time spent in the temperature range tied to the Artemis I damage.
Why it matters for materials: an ablator like Avcoat protects by charring and wearing away on purpose, so how the char forms, cracks and holds together over a temperature range is a large part of the story. A single property value will not tell you that. This page does not itself explain the cause of the Artemis I damage; a separate NASA article it links to says gases inside the Avcoat could not vent and dissipate as expected.
Our take: the headline compares counts of spots, not area or mass lost, across two flights with different trajectories. The post-flight laser scans are being compared with pre-flight measurements to work out how much material receded, and that recession will then be compared with analysis predictions; sample extraction at Marshall is continuing, with material tests and thermal analysis to follow. We read this as good evidence that the flight path is a design variable alongside the material, not yet proof that the material itself is fixed. For scale, the Avcoat page in our archive lists a room-temperature density of 530 kg/m3 and a thermal conductivity of 0.0502 W/m-K at -182 °F (154 K), with 0.109 W/m-K at 491 °F, all from the 1965 NASA report TN D-2991 on the Apollo-era grade Avcoat 5026-39-HC G, with the table cited.
What we would watch: the recession scans, and whether the Artemis III upgrade aimed at more uniform block permeability shows up in the data. If you have worked on char behaviour, what would you want published first?
