Re:stamping process is not useful for mass product
on
Printing Chips
·
· Score: 2, Interesting
The idea of stamping an entire wafer at once, and getting the pattern correct is pretty heady stuff, considering the inherent difficulty in getting the overlay on a single field correct. The primary showstopper for stamping the entire wafer at once is obviously going to be the ability to get overlay to the underlying pattern *perfectly* over the entire area of the wafer. Let's say the quartz stamper is in perfect condition (no particles, and no residue from previous stampings, and no broken impression teeth). Raising the temperature of the stamper by even a miniscule amount will cause the stamper to expand, creating more than enough X and Y scaling error to prevent all but the centermost chips from even being *close* to matching the underlying pattern. Hence no continuity layer to layer and no usable chips except in the center of the wafer.
Also, let's say that the stamper can do 1 stamp per minute (aggressive, but ya gotta make some assumptions). How do you determine when the first little imprinter point snaps off, and every subsequent stamp creates a dead die. If it happens during the first several hundred wafers patterned, and the stamper isn't changed for thousands of stamps, there's going to be an awful lot of non-yielding die at end of line. That's a real bummer, because nobody buys the chips that fail sort/etest for anything more challenging than pretty keychains/ornaments.
Another problem is going to be that the surface of a wafer is *not* flat. Run a wafer through a diffusion furnace, and it warps like an album left in the sun (ok, so I'm dating myself... but the analogy is valid). Let's assume that the wafer bows up at the edges, relative to the center. If you try to press the entire wafer at once, you are going to get excessive pressures at the edges, while the center of the wafer isn't yet touched by the stamper. As a result, the center of the wafer isn't going to get any pattern, and so the center of the wafer won't yield usable die. Bummer.
Another issue is that *all* wafers end up with particles on the surface, be it aluminum, stainless steel, tantalum, or just plain old dust. What happens to the little imprinter fingers when you try to press them into a hunk of steel? I'll give you a hint, it'll be like holding your fingers out straight and punching a bowling ball... your fingers are gonna break. How well are your fingers going to be at pressing anything after that? On the wafer scale, any stamper which hits a die that has a surface defect will result in that die being defective on all subsequent pressings. More keychain ornaments, but less working chips and much less profitable.
In closing, let's consider one other little issue. In patterning, the goal is to have the sidewalls as nearly perpendicular to the surface as possible. A cross section view of a line should look like a skyscraper, with vertical sides, and not like a pyramid/trapezoid. In order to stamp and be able to extract the stamper from the imprinted surface without ripping off teeth, the impresser has to be tapered to minimize friction effects. Etchers and implanters really don't do well with tapered sidewalls on the pattern... you lose resolution of the resultant structures/implants.
stamping process is not useful for mass production
on
Printing Chips
·
· Score: 1
Ok, lets start with the basics of photolithography, and then compare how this new method works:
In the current manufacturing method, the entire wafer is coated with a photosensitive material (photoresist), and then the desired image is sequentially shined onto small areas of the wafer to pattern a few die at a time (usually a 2 or 4 die group known as a field, depending on the chip size). To complete the patterning of a single wafer can take several minutes, as the stepper/scanner machine has to custom align and expose each field individually.
This patterning is done using a reticle, which is a quartz plate with the desired pattern printed on it with a chrome layer. The reticle between the light source and the photosensitive wafer surface. The reticle patterned surface never physically touches any surface, so that no defects are created. Even a single spec of dust on the chrome side of the reticle will kill all die patterned using this reticle.
Once the sequential patterning of the entire wafer surface has been completed, the wafer is sent to a developer, where the exposed photoresist is stripped off the wafer using a chemical, leaving photoresist only in the areas which were not exposed to light (areas which were under the chrome parts of the reticle).
Now let's consider the direct printing methods:
One applies a polymer, similar to the photoresist but without the photosensitive chemical additives. The desired pattern is physically pressed into the polymer using a mask. This patterning stamping is repeated for each die, until the entire wafer has been processed, and then the wafer is sent on for processing (implant, etch, whatever). There is no develop process, as the image was stamped directly onto the wafer surface.
Another method uses a quartz contact surface and a laser to transfer the pattern to the wafer.
This is an important distinction, between optical and direct contact patterning... the reference pattern **directly contacts the surface to be patterned**. Let's assume there are 100 fields on the wafer which need to be patterned. Now let's be optimistic and say that the stamper can last for 500 stampings. That means that every 5 wafers, you'll need a new stamper. Replacing a stamper is not going to be a simple process... and time is definitely money in the semiconductor industry. Having a tool sit idle after every 5 wafers patterned is simply unacceptable. Also, the direct contact between the stamper and the surface will result in polymer adhering to the stamper, which will cause pattern to be blocked. Think of a cookie cutter with a closed top surface... how many cookies can you stamp out before it gets clogged with dough? All it takes is for one or two features on the stamper to be clogged, and then every die patterned after that will be dead. So, if the stamper get's corrupted after 50 stampings, the remaining 450 fields in the 5 wafer patterned set will be useless. That's 200 good chips, and about 20,000 bad chips. Now take into account that modern processors have 15-20 layers of patterning required for each chip, meaning that each chip must get perfectly stamped 20 times... what are the odds of getting even 1 good die? The answer is slim to none.
As for the chemical savings, the only chemical in the photo process which is not used in the stamping process is the developer solution. Developer solution for photolithography is usually a strong basic solution (tetramethyl ammonium hydroxide is a very common developer solution, as it is water soluble). So you don't use one of the chemicals. That's like saying that by getting rid of the power steering in your car, you are saving on all that power steering fluid getting into the environment. In the big scheme of things, this is a non-issue.
Another issue is the overlay to previously printed layers. The pattern must be *precisely* aligned to the layer underneath, otherwise the electrical connnections won't be correct. Using optical patterning, the corrections can be made by tilting/rotating the reticle, varying scan speeds, and the image can be optically expanded/shrunk for scaling corrections. With direct patterning, the stamped features cannot be corrected for other than the most gross alignment issues.
In short, the use of direct patterning is interesting for the laying down of a single layer of small holes a couple of times, but getting a yielding device from it? Get real. It ain't gonna happen. The best that it could be used for is the production of things like diffraction gratings or similar single layer non-critical patterning where small defects will not affect the overall image.
One of my favorite cue sticks is a Predator cue. One of the manufacturer's claims to fame is the use of a robot during development and testing http://www.predatorcues.com/oldenglish/whybuy.htm to perform repeatable shots over and over.
It would be interesting to combine the repeatability of an Iron Willie robotic arm, and the laptop for number crunching, with a feedback loop to allow for corrections (an iterative learning process). Now *that* might be able to shoot (basic and rudimentary) pool. However, without first learning the basics of forming a bridge, point of aim, stroke, etc... the use of a computer by a novice pool player will be of little to no help.
Let's look at those little bits from florida (the renowned "missing chads"), which completely altered the course of the United States government for the following 4 years. Now *that* is some seriously Most Significant Bits.
Slashdot Mirror
Also, let's say that the stamper can do 1 stamp per minute (aggressive, but ya gotta make some assumptions). How do you determine when the first little imprinter point snaps off, and every subsequent stamp creates a dead die. If it happens during the first several hundred wafers patterned, and the stamper isn't changed for thousands of stamps, there's going to be an awful lot of non-yielding die at end of line. That's a real bummer, because nobody buys the chips that fail sort/etest for anything more challenging than pretty keychains/ornaments.
Another problem is going to be that the surface of a wafer is *not* flat. Run a wafer through a diffusion furnace, and it warps like an album left in the sun (ok, so I'm dating myself... but the analogy is valid). Let's assume that the wafer bows up at the edges, relative to the center. If you try to press the entire wafer at once, you are going to get excessive pressures at the edges, while the center of the wafer isn't yet touched by the stamper. As a result, the center of the wafer isn't going to get any pattern, and so the center of the wafer won't yield usable die. Bummer.
Another issue is that *all* wafers end up with particles on the surface, be it aluminum, stainless steel, tantalum, or just plain old dust. What happens to the little imprinter fingers when you try to press them into a hunk of steel? I'll give you a hint, it'll be like holding your fingers out straight and punching a bowling ball... your fingers are gonna break. How well are your fingers going to be at pressing anything after that? On the wafer scale, any stamper which hits a die that has a surface defect will result in that die being defective on all subsequent pressings. More keychain ornaments, but less working chips and much less profitable.
In closing, let's consider one other little issue. In patterning, the goal is to have the sidewalls as nearly perpendicular to the surface as possible. A cross section view of a line should look like a skyscraper, with vertical sides, and not like a pyramid/trapezoid. In order to stamp and be able to extract the stamper from the imprinted surface without ripping off teeth, the impresser has to be tapered to minimize friction effects. Etchers and implanters really don't do well with tapered sidewalls on the pattern... you lose resolution of the resultant structures/implants.
In the current manufacturing method, the entire wafer is coated with a photosensitive material (photoresist), and then the desired image is sequentially shined onto small areas of the wafer to pattern a few die at a time (usually a 2 or 4 die group known as a field, depending on the chip size). To complete the patterning of a single wafer can take several minutes, as the stepper/scanner machine has to custom align and expose each field individually.
This patterning is done using a reticle, which is a quartz plate with the desired pattern printed on it with a chrome layer. The reticle between the light source and the photosensitive wafer surface. The reticle patterned surface never physically touches any surface, so that no defects are created. Even a single spec of dust on the chrome side of the reticle will kill all die patterned using this reticle.
Once the sequential patterning of the entire wafer surface has been completed, the wafer is sent to a developer, where the exposed photoresist is stripped off the wafer using a chemical, leaving photoresist only in the areas which were not exposed to light (areas which were under the chrome parts of the reticle).
Now let's consider the direct printing methods: One applies a polymer, similar to the photoresist but without the photosensitive chemical additives. The desired pattern is physically pressed into the polymer using a mask. This patterning stamping is repeated for each die, until the entire wafer has been processed, and then the wafer is sent on for processing (implant, etch, whatever). There is no develop process, as the image was stamped directly onto the wafer surface.
Another method uses a quartz contact surface and a laser to transfer the pattern to the wafer. This is an important distinction, between optical and direct contact patterning... the reference pattern **directly contacts the surface to be patterned**. Let's assume there are 100 fields on the wafer which need to be patterned. Now let's be optimistic and say that the stamper can last for 500 stampings. That means that every 5 wafers, you'll need a new stamper. Replacing a stamper is not going to be a simple process... and time is definitely money in the semiconductor industry. Having a tool sit idle after every 5 wafers patterned is simply unacceptable. Also, the direct contact between the stamper and the surface will result in polymer adhering to the stamper, which will cause pattern to be blocked. Think of a cookie cutter with a closed top surface... how many cookies can you stamp out before it gets clogged with dough? All it takes is for one or two features on the stamper to be clogged, and then every die patterned after that will be dead. So, if the stamper get's corrupted after 50 stampings, the remaining 450 fields in the 5 wafer patterned set will be useless. That's 200 good chips, and about 20,000 bad chips. Now take into account that modern processors have 15-20 layers of patterning required for each chip, meaning that each chip must get perfectly stamped 20 times... what are the odds of getting even 1 good die? The answer is slim to none.
As for the chemical savings, the only chemical in the photo process which is not used in the stamping process is the developer solution. Developer solution for photolithography is usually a strong basic solution (tetramethyl ammonium hydroxide is a very common developer solution, as it is water soluble). So you don't use one of the chemicals. That's like saying that by getting rid of the power steering in your car, you are saving on all that power steering fluid getting into the environment. In the big scheme of things, this is a non-issue.
Another issue is the overlay to previously printed layers. The pattern must be *precisely* aligned to the layer underneath, otherwise the electrical connnections won't be correct. Using optical patterning, the corrections can be made by tilting/rotating the reticle, varying scan speeds, and the image can be optically expanded/shrunk for scaling corrections. With direct patterning, the stamped features cannot be corrected for other than the most gross alignment issues.
In short, the use of direct patterning is interesting for the laying down of a single layer of small holes a couple of times, but getting a yielding device from it? Get real. It ain't gonna happen. The best that it could be used for is the production of things like diffraction gratings or similar single layer non-critical patterning where small defects will not affect the overall image.
It would be interesting to combine the repeatability of an Iron Willie robotic arm, and the laptop for number crunching, with a feedback loop to allow for corrections (an iterative learning process). Now *that* might be able to shoot (basic and rudimentary) pool. However, without first learning the basics of forming a bridge, point of aim, stroke, etc... the use of a computer by a novice pool player will be of little to no help.
Let's look at those little bits from florida (the renowned "missing chads"), which completely altered the course of the United States government for the following 4 years. Now *that* is some seriously Most Significant Bits.