Now that we’re using a metal extrusion for most of the main body, we also need to use some glue to stick plastic bits to it (such as the window you see the display through), since there just isn’t enough room in the tiny body for strong snap-fits. This makes recycling harder, for two reasons: first, because now the device is harder to disassemble into its component materials; and second, because now the component materials will have some gunk on them (adhesive residue) that cause problems in recycling the metal or plastic.

Don’t Be Too Strong

Fixing the first problem is fairly simple: you just use weak enough glue (or a small enough amount of strong glue) that whoever disassembles the device can just pop the parts out by hand, overwhelming the strength of the glue. This way, disassembly doesn’t take significantly longer than it would with snap-fit parts. If the glue is too strong, you have to pry things out with a tool, or (if the glue is stronger than the plastic itself) you have to release the glue somehow. Most glues can be released by dissolving them in nasty solvents like acetone, or burning them off in a furnace, but some glues dissolve in water, and other glues melt at low enough temperatures that your plastic parts won’t be affected. The advantage of dissolving or melting your adhesives is that then they can be removed from your parts, avoiding the problem of getting gunk in the recycling furnaces.

Recycling The Glue

Aluminum parts with glue-gunk on them are still recyclable, and only have a slightly lower value than clean aluminum, because the gunk will get burned off during the re-melting process; this requires the recycler to have special emission controls on their smelter in order to avoid air pollution, and can introduce some impurities in the metal. Many smelters in the US and Asia have these emission controls, but even those that do will pay a higher price for clean scrap. I talked with five different scrap brokers (companies who take your scrap and sell it to smelters; you can’t really go directly to the smelter yourself unless you’re a big company with huge amounts of scrap), and they said contradictory things. One said that having even a tiny amount of glue reduced the scrap value from 40¢ per pound to 15¢ per pound; another said it reduced the value from 70¢/lb. to 60¢/lb; two other brokers said that small amounts of glue didn’t really matter, and didn’t reduce the value at all.

Plastic parts with glue-gunk on them are a much worse proposition. All four places I talked to said that even a little glue, even if it’s easily removable, cuts your value in half (from 40¢/lb. to 20¢/lb., say), and bad glue or a lot of glue cuts your scrap value to almost zero (maybe 5¢/lb.) This isn’t because of emissions concerns like aluminum, it’s because the melted glue mixes in with the plastic and causes the plastic’s mechanical and molding properties to get whacked (glue molecules get in the way of polymer-chain bonding.). This is in the worst-case scenario. The best case scenario is that the recycler will clean off the glue (with water or heat or both) and then that scrap will be as good as other clean scrap; this takes time and processing, though, which is why your resale value gets cut in half. So how do you find a good water-soluble glue that can be easily removed? Well, turns out that’s easy.

Water Soluble Glue

Cyanoacrylate (a.k.a. “superglue”, “krazy glue”, and various other brand names) is water-soluble. A Loctite rep told me “If you superglue your eye shut, a warm compress will unglue it in a day or so”, so presumably a tub of boiling water would get it off in a few hours. (A plastic part, that is, not your eye. Ahem.) By the way, Loctite is a great company environmentally (for a glue/chemical company), they’re good to buy from. Cyanoacrylate is probably your best bet, because it’s strong, so you don’t have to use very much, and it sets rapidly, which is good in manufacturing.

Common Elmer’s glue (”white glue”) and some wood glues are water-soluble. White glue is not feasible for much manufacturing because it takes minutes or hours to set instead of seconds; it usually is not a very strong adhesive, and can also shrink when setting, causing parts to move or bend. But there are also higher-performance kinds that have stronger adhesion and faster setting times; many wood glues are stronger than the wood they glue (though they’re still not very useful for plastic or metal).

Regitex, a Japanese company, specifically formulates their glues to be eco-sensitive. They have a whole line of water-soluble latex glues for different applications (envelopes, food containers, medical tape, shoes, and general use; one kind is even cured by UV light instead of drying, which is very fast.)

Hide glue (yes, glue made from animal hide) is another water-soluble glue which can bond metal, plastic, glass, and cloth, although it’s mostly used for wood. Sometimes it requires a bit of vinegar in the water to help dissolve it, but vinegar is a mild chemical. Hide glue is archival (three-hundred-year-old violins are made with it), and it requires just a thin coating, which is nearly transparent, so it should not interfere with aesthetics. However, like Elmer’s glue, it may take an hour to set, and it shrinks when setting. You can get it in liquid form or do-it-yourself granular form.

There are various other specialty glues that are water soluble: AquaBond makes temporary high performance adhesives for semiconductor processing; Mosaic Mercantile makes it for mosaic tiling; Dongguan Jelly makes it for fabric; 3M even uses water soluble adhesive in some tape for masking printed circuit boards during soldering.

Heat-Reversible Glue

Heat-reversible glues, from what I found, would be fine for metal parts, but they don’t help with recycling plastic. They don’t disappear under heat, they just lose their stick; so on plastic they melt into the mix and gunk up the polymer. There may be ones that vaporize or drip off entirely before the plastic melts, but I didn’t find one on the market.

The trusty hot glue that you use for making models and crafts is heat-reversible. It softens at 75°C, where you can sort of pull things apart, and melts at 130°C, where it doesn’t hold anything in place anymore; this is a lot cooler than most plastics melt, so on a disassembly line, you might be able to place the devices in an oven that melted the glue until it dripped off or got sprayed off. But this would be much more hassle and energy use than dealing with water-soluble glues. And hot glue’s melting point is so low it would melt on a car dashboard in an Arizona summer.

Apparently the archery industry has a couple heat-reversible glues, as well: Ferr-L-Tite and Bohning PowerBond.

There are also higher-performance heat-reversible glues being investigated. Active Disassembly Research ltd. in the UK doesn’t sell glue retail, but they develop heat-reversible glues, latches, hooks, screws, springs, and other fasteners in an impressive variety. Check out their video gallery (no direct link); they’re one of the main companies making devices which actively disassemble themselves. In the lab, researchers at Sandia National Labs have made heat-reversible tape, and scientists at Germany’s Fraunhofer Institute have made a heat-reversible glue with nano-scale ferrite particles that, when exposed to an oscillating magnetic field, jitter around to heat up the glue without heating up much else. Neither of these are commercial products yet, but they hopefully will be within the next several years.

Conclusion

Generally speaking, it’s best to avoid adhesives in a product in order to ensure easy recyclability, but that doesn’t have to be the end of the story. There are no doubt situations where using a reversible glue actually makes disassembly faster and easier than snap-apart fastening would, without too much cleanup required for recycling. In our case it’s an obstacle, but we can work around it.

This week’s episode of green design for Exbiblio is about metal. They recently decided to change plans about how the first release of the oPen will be made–instead of the whole body being recycled injection-molded plastic, most of the body will be an aluminum extrusion with holes machined into it for the screen and buttons, and there’ll just be plastic bits on the ends, much like an iPod Nano. The reasons for this had to do with schedule and design flexibility–we have a very tight schedule to make, and need to get to production as soon as possible, but still have not nailed down all of the design considerations. Using an extrusion with machined holes gives us a great deal of flexibility, as machining can be reprogrammed at any time to cut different holes, and extrusions are fast and easy to get into production–easier than injection-molding.

Using aluminum instead of plastic does increase the device’s environmental impact, in three ways: first, aluminum is more energy-intensive to produce than plastic; second, it’s more energy-intensive to manufacture with and requires harder tooling; third, having the case be made out of multiple different materials makes it harder to recycle because it needs to be more carefully disassembled and sorted. With a device this small, we need disassembly time to be extremely short, otherwise it won’t be worth anyone’s while to recycle it, because the amount of plastic and metal you get for the amount of time spent is small. I’ll talk more about design for disassembly in a later post.

Almost all consumer electronic devices which use extrusions are made out of aluminum. The ecological advantages of aluminum are that it’s a very common material (the most common metal in the Earth’s crust), it’s not toxic, and it’s very recyclable–according to the US Geological Survey, 44% of aluminum production in the US is recycled material (the industry calls it “secondary” production as opposed to “primary” production.) The disadvantage of aluminum is that extracting and refining it is enormously energy-intensive–so much so that it’s often called “solidified electricity”. The US EPA estimates that 2-3% of all electricity use in the US is for making aluminum, and New American Dream estimates that 16% of all electricity used in Oregon & Washington goes to aluminum smelting. (Also from that site: “every three months, Americans alone throw away enough aluminum [cans] to rebuild the entire US commercial air fleet.” So remember to recycle your cans.)

Exactly how energy-intensive is aluminum? An MIT study found “the production of 1 kg of aluminum requires on the order of 12 kg of input materials and 290 MJ of energy”. However, recycled aluminum only requires 5% as much energy to produce. This is why so much aluminum is recycled–it’s advantageous purely from a money point of view, even if you don’t care about the environment. Estimates vary on how much of the US’s total aluminum production is recycled, and it depends on the grade of aluminum you buy, but on average it’s at least 20%. Smelting tends to be a very large-scale operation, so everything gets mixed together in large batches–that unfortunately means it’s nearly impossible to source 100% recycled aluminum. (The upside is that everyone who buys aluminum of most grades gets some recycled content whether they care or not.)

Steel, on the other hand, is also extremely common (iron is the fourth most common material in the ground), and doesn’t require aluminum’s enormous energy budget–it uses less than 1/8 (almost 1/9) as much energy as aluminum to produce. And stainless steel has a great aesthetic, very classy without being pretentious. Steel is also recycled even more than aluminum: the USGS says that 71% of steel in America is recycled. (Surprisingly, they also say that the auto industry’s steel recycling rate is 102%, meaning that they’re recycling more steel from old cars than the steel they’re putting in new cars.) Unfortunately, however, steel cannot be extruded in the tiny dimensions of our device–I talked to about twenty companies throughout the country, and the thinnest wall they could extrude was usually the thickness of our entire device! Some places can do what’s called cold-rolling for thinner walls, but even then I couldn’t find a single company who felt they could make what we needed.

We also considered titanium briefly, but it has none of the advantages of aluminum and far worse disadvantages. It’s a rare metal; it’s not toxic itself, but its refinement requires toxic chemicals like chlorine and hydrochloric acid; and it uses over three times the amount of energy to produce as aluminum (twenty-six times that of steel). (See that same MIT study for details.) Titanium is also expensive, due to the difficulty and energy-intensity of processing it; it’s also very hard to work with: a European Commission study says that “80% of material from forged parts for the aerospace industry becom[e] turnings or other scrap. Titanium machines at between 10-20 times slower than aluminum and can account for 70-80% of the cost of the component.” So, no titanium for us. Best to leave it for high-performance engineering applications that require its impressive physical properties, or to the medical industry that likes its non-reactivity in the body.

Coatings

Bare aluminum gets smudgy and ugly when handled, and can get your hands a little smudgy too, so for aesthetics you should coat it with something. Paint is a poor choice, because all but a few special paints cause a great deal of pollution, as they require toxic solvents to set, and offgas volatile organic compounds while they’re new. (I hate to break it to you, but that “new car smell” isn’t very good for you.) Lacquering, plating, and buffing are also not so great. The two good choices are powder-coating and anodization.

Powder-coating is like paint, except there are no solvents. Instead, the pigment is a fine, usually non-toxic, dust that gets sprayed onto a product with a compressed-air sprayer (no CFC’s required). The powder sticks to the product is by static cling: the powder shooting out of the sprayer gets positively charged, and the product is hanging on a rack that’s negatively charged (meaning that the product has to be electrically conductive to powder coat it.) Then the powdered product is put in an oven to bake the powder into a hard coating; this does use some energy, but is still a much smaller impact than painting. The only problem with powder coating is that, like paint or lacquer, when you recycle the aluminum it’s coating, it burns off into toxic fumes, so a recycling plant needs to have special emissions-control equipment to do it cleanly. This means that powder-coated aluminum is less recyclable than bare aluminum (and painted or powder coated aluminum scrap is worth less money than bare aluminum.)

Anodization is a microns-thin layer of oxidation on the surface of the aluminum. It involves some nasty chemicals (depending on the color, these can include sulphuric acid, nitric acid, phosphoric acid, nickel acetate, and others; some use hexavalent chromium, but we will definitely avoid that), but small amounts of them (because the oxidation layer is so thin). The coating is non-toxic to the user, the concern is the waste and worker safety in manufacturing; any decent modern plant has emissions/effluent controls, but it would be better not to use toxins in the first place. The main advantage of anodization is that it does not hurt recyclability of the aluminum–it is such a thin coating (and even that coating is mostly aluminum itself) that anodized parts can be thrown right in with bare parts in recycling plants, and their value as scrap is just as high as bare aluminum.

In the end we decided to go for anodization. It was unclear which of the two choices was best (they both have advantages and disadvantages), and anodization looks much nicer.

The hope is that when mark II is ready towards the end of the year, it won’t have these integration problems, as most of the software kinks will have been ironed out.

Exbiblio, once again, is to be commended on its openness – but on this occasion, Synapse deserves credit too. As it happens, I was planning to write an account of my visit to Synapse.

Synapse is based in a former tram depot in a suburb of Seattle. Immediately I walk in, I get a different feel from Exbiblio – it’s open-plan, chatty, but business-like. The depot makes me think of the Google Garage. It has some of the make-shift romance of a start-up, though Synapse was founded in 2001 and employs around 40 people. Exbiblio, as I have mentioned before, is in the center of town, and has a bookish, intellectual feel, with the workers dispersed in offices along corridors, where they beaver feverishly away at their projects.

On its website, Synapse says it does technology- intensive product development. Since June, it has been working with Exbiblio to develop the oPen. Its other clients come from all over the world and include Microsoft, Samsung, Philips, Intel, Logitech, General Electric – well you name it.

The five-strong engineering team working on the oPen is managed by Dave Zucker (holding the notepad in the picture). You will often see Dave and other Synapse people around Exbiblio. Exbiblio’s Ian MacDuff (in the middle of the picture) is often at Synapse. It all seems very well integrated. They seem like a happy team, though I hope Ian won’t mind me saying that he looks a little stressed now that the project is behind its ambitious schedule.

I’ve heard Dave say a couple of times that the Synapse way is to “fail early, and fail often” – in other words to go all out for rapid development of prototypes, see what works, what doesn’t, and then quickly do another one, and then, if necessary, another one.

That is what has NOT happened with the oPen. Dave told me:

“We had said at the beginning that this was a two month project.”

I interject, “Was that possible?”

“It was possible. But Martin and I had discussions along the way where we made conscious decisions to push out the schedule in return for something – either we learned more, or got a design that we liked better. When you shoot for a prototype in two months, you definitely are going to make some compromises. You certainly couldn’t design something for production that was exactly for final version in two months.”

And so the deadline slipped, partly as a trade-off for more features and knowledge, and partly just because it slipped a bit.

The first prototype - videoed last week – is about one and a half times bigger than the desired size of the final version (this was planned). The idea of this prototype is to learn about memory, processor, and battery-life requirements. The engineers are also looking for ways to cut the size down.

Dave told me about one of the problems – er, challenges – already uncovered. When the first circuit boards came back from the factory they noticed some funny behaviour. Some diagnostics showed that two pins on the layout had been connected by a mistake in the design. It sounded as if this could be fixed quite easily.

I don’t really understand all the technicalities of Brian’s email but it doesn’t sound like the end of the world to me. I’m sure these glitches can be put right. But as we used to say back in my school days, Tempus Fugit.

It would have been nice to have a perfect prototype first time, but I think we all live in the real world and realise that “right first time” would require a lot of good fortune. It’s not really what Synapse promise (”Fail early and fail often”).

Tactless as ever, I asked Dave who pays for any delays. The answer is Exbiblio.

Well…The build of 20 units has not gone as cleanly as I had hoped.

There has been pretty low yield on the two PCBs. Less than 50% on the Main board and ~75% on the Button board.

For the Main PCB there are a lot of boards that have processor bus data lines shorted to each other.

For example one of the boards looks to have D8 and D10 shorted together.

If you look at the layout of the board, you’ll see that D8 and D10 are adjacent to each other at the board to board connector and under the SDRAM BGA. See the image below.

If you look at the left side, you’ll see two pins on the connector that are colored yellow. These are D8 and D10.

I suspect that they are shorted either under the J2 connector or under the U7 BGA.

I’m taking 10 boards over to PCA to morrow to have them x-rayed to see if we can locate the shorts, then fix them.

The fallout on the Button board seems to be a bit more random. On one board, the Left Illumination LED doesn’t work, on another the SM Bus data line is shorted to ground.

That being said, I’ve give 7 working board sets to Dave to get assembled into units. We’ll work on getting more working and assembled ASAP…

One of the things that is facinating about Apple is that they can ignore popular opinion, chart their own course, and end up shaping popular culture. Something to think about as we begin to share early versions of our own products with the public. When should Exbiblio listen to feedback, when should we ignore it, and how do we draw that line?

All the standard disclaimers apply here, there is much optimization to be done and several more months of hardware and software improvements before we are ready for our public beta, but it is actually coming together!

By the end of the week, we hope to have the new prototype integrated into our Exbiblio software application prototypes, so stay tuned for more news soon.

Check out the video:

Hugh is planning to post details from the interview this week, but I thought I’d get some images from our visit up on the blog today. Including, if you look closely, the first image of the scanning device (model) that we have ever shown to the public! Stay tuned for more details.