Mr. DUNLAP. Thank you, Mr. Chairman and members of the committee for an opportunity to talk to you this morning. And I ask that my written testimony and a number of articles on the protection act be introduced. As you indicated, I represent Intel Corp. I also represent the semiconductor industry, which is a group of 57 semiconductor manufacturers and users including the large manufacturers in California, Texas, and large computer companies.

I am going to try to explain the basic process of how to manufacture a semiconductor and, in doing so, explain what a copy is under the new technology, the semiconductor technology, and explain what we consider a copy, which is a little bit different from what our Founding Fathers had in mind when they wrote the Constitution.

And I would like you to keep in mind that there is a minority of intellectual property bar which believes that the chip topography can be covered under the current copyright law. I indicated that is a minority. There have been a couple of suits under the current law, but none of the plaintiffs have thought it worthwhile to go far enough in suits so that no court has actually ruled on this subject. I would like to use a projector [slide presentation]. First of all, I would like to show you what is a chip.

Mr. KASTENMEIER. May I inquire, Mr. Dunlap, how this may be reflected in the record?

Mr. DUNLAP. That which is shown visually on the screen is also reduced to writing and will be made available to the reporter. (See app. 2 at p. 365.)

Mr. KASTENMEIER. I think we will dim the lights. People in the room who can position themselves to see the screen are at liberty to do so.

Mr. DUNLAP. First of all, I want to show a chip. This is really what the user is going to see. In other words, these are the different pins where you get your electrical impluses, and this is the form that would go into your automobile, your robot, your calculator or whatever. And I also have taken the lid off it, and you can see it is the actual silicon.

OK; so now a chip is a collection of transistors on a single structure, and they all work together to perform a particular function. And so they can perform the measurements of fuel, measurement of emissions. They can perform movements of the robot and all these different things that I have listed here.

Now, the basic building block of the chip is the transistor. This is what is actually going to perform the functions. And this transistor is typically fabricated on a semiconductor material known as silicon. That is how we get Semiconductor Chip Protection Act.

The chip itself then is made by starting with a thin substrate here which we call a wafer, which is typically silicon. So it would be shiny like this side when it is unprocessed silicon. We are going to put it through a number of chemical and photographic and heat processes in order to make chips on it.

So on this side is what a finished wafer would look like. So you start out with this basic bare wafer. And now you're going to go through some processes where you're going to grow a layer of oxide. Then, you're going to put down a photographic material which we call resist. Resist is sensitive to being exposed to light.

So what we'll do now is we will basically take a picture on this resist. We will shine light on the wafer. Certain areas will be exposed to light; other areas will not be exposed. And then what happens is the areas that were exposed become very hard. The areas that were not exposed are still soft.

And so what you do now is subject the wafer to a chemical process and etch away the areas that were not exposed to the light. And so you make these patterns [indicating]. This is a very simple one-step example. You go through eight steps, and you make maybe hundreds of thousands of these on a single chip. But the basic idea of imprinting this pattern is what we're trying to protect.

Mr. SAWYER. Does the chip have to be silicon or can it be something else?

Mr. DUNLAP. In the early days it was germanium. It can also be saffire.

Mr. KASTENMEIER. Why is it a semiconductor chip as opposed to a conductor chip?

Mr. DUNLAP. Because what will happen is it has the ability to either conduct or not conduct.

If I go on with the process, the silicon is exposed here, and you will put in impurities. For example, you put in either boron or arsenic in here. And then you will be able to put electric potential up here. And in the normal state this, what we call will be a source and a drain here of impurities, but they can't communicate, so it does not conduct. But once you put some potential, some electrons on here, now it can conduct. And so they form a channel here, and so that's how they get semiconductor. Sometimes it conducts and sometimes it doesn't conduct.

Now, how do you design a chip? The first thing you have to decide is what it is going to do. You have to do a market study and understand what your customer wants for his computer. And somehow your marketing people determine that you want the chip to do a specific function. So you get a circuit designer who is going to develop an electronic circuit and represented in what we call schematic function. So he is going to represent it like this.

So he understands what this particular device will do, and this type of circuit diagram could be protected by patent. That is, the function that it is going to perform could be protected. This is not the subject of the copyright.

Now, in actual schematic-I mean this is just a very simple one-but an actual schematic of the chip that I showed you is much more complicated here, as you can see. It is many more lines, many more gates. And this is 1 page out of 20 pages. So you've got 20 pages of this to come up with a chip that has 100,000 transistors

or so.

But at this point you have got nothing but paper. So you're going to have to take this paper and put it into a form which can eventually be made into a pattern which will be printed on the chip.

And so now we have a layout designer who is going to take this circuit, and he is going to draw it out into a group of patterns which can be imprinted on this wafer. And this is what we are going to call mask work.

And I have here a drawing of a mask work. This is 20 times the size of the original chip. They draw it out like this. And it has different color for each of the layers. So there's eight different layers-well, I think it has six different colors, because some layers are not shown.

And this is really what we are trying to protect, not the chip. But we're trying to protect this picture. OK. So you take that layout [indicating], which you're going to have to put on the silicon. Sooner or later you take that layout, you embody it in a magnetic tape because that's easy to manipulate. You don't want to keep redrawing the thing.

And from there we're going to make a mask. And this mask will be either glass or metal plate-in this case it's metal-for each pattern, for each layer. And you then are going to be able to take a picture of the chip.

So you get this printer, we call it. You put this in the printer, shine light through this mask, and this is the actual size now. You see it will match up exactly with the wafer. And then light comes through certain areas and does not come through the other areas exactly like that does.

OK; now, if I could ask you to use your imagination for a moment, we're going to pretend this is the silicon. This is the wafer. OK. And this is the printer that I talked about. And these are the exact same thing that I showed you there, the same pattern that a mask would create. So what happens is we're going to take these pictures. So the first thing you do is you put down this layer. And again it's projected on the silicon. And then the areas here which you see in red would not be exposed. They'd be protected from the light. In the clear areas, the light would go through, and therefore you could etch away the material underneath and put the second layer down and then the next layer.

And then you just keep doing this process until finally you'll come out with the same pattern that we had before. This is exactly the same pattern that's on the chip. And that's exactly how you make a semiconductor chip.

Now, I want to stop here for a minute and indicate that the Copyright Office has in the past accepted these mylar prints as engineering documents. They have accepted them for copyright. They have also accepted the masks.

But the problem is that they have not accepted the chip. And my concern and the reason that we want the copyright law is to make clear that this pattern is protected. This form or any of those forms, are not available to the public. We treat these very carefully. These are important trade secrets. The only way you could get at the pattern is from the chip.

So we want to extend the idea of copying two ways. One way is that when I project light off of here onto the screen or onto the silicon, that's copy No. 1. No. 2 is the reverse of that, which is when a potential pirate who doesn't get access to this but gets access to the chip. When he takes that pattern and just measures it and draws it back on that drawing, that is a copy, because now what he has done is reverse the process. He's going to take it from here, from the silicon, put it back into this form, and then get it

into the mask. And then he'll compete with us, and he hasn't had to design it.

OK; so how to copy a chip is the next important issue. In section. 41 of the bill, embodying the pattern in a mask work is defined. That's what you're not going to be allowed to do. The pirate will not be allowed to take the mask work-that is, the patterns which are on the chip-and put them in a mask and then copy it. The chip I just showed you, is available publicly; it has to be. It's easy to pop the lid off. Now he takes a photograph of the chip and he very carefully measures this top layer. Then he goes through a chemical process and etches it away.

Now he's got another layer exposed. He measures this layer, etches it away. Then he just keeps doing that until he has completely reversed the process. And that's much easier than designing it in the first place.

Mr. SAWYER. Is that original drawing done by hand, the one that you started copying with?

Mr. DUNLAP. OK. The original drawing in the early days was done by hand, but now we have computer machines that do it, so that the layout person would draw it on a machine, just draw it on a computer just like you would type into a word processor today. And then it would be converted by the machine into a printed form.

Mr. KASTENMEIER. Obviously, it is easier to remove these layers, these etched layers, than it is to design it. But it is still not easy for a pirate to do an entire chip, is it?

Mr. DUNLAP. It's relatively straightforward. A chip-let's take that chip there, which maybe would cost you $4 million to design, the pirate can do it for $100,000.

Mr. KASTENMEIER. How much time would it take?

Mr. DUNLAP. It would take maybe 3 years, 31⁄2 years to do originally, and 1 year, maybe 11⁄2 years for the pirate to do it.

Mr. KASTENMEIER. It takes them 1 year, 11⁄2 years to do it? Mr. DUNLAP. I am sorry. I guess I was thinking of reverse engineering. I am thinking of the actual case of this chip. It was not copied. To do reverse engineering would take 11⁄2 years. To do just straightforward copying like this would take 3 to 5 months.

Mr. FRANK. The chips you said are designed for a specific user so that the pirate would then go and undersell the original manufacturer to that particular user? What does he do with the chips once he pirates it?

Mr. DUNLAP. Once he pirates it-

Mr. FRANK. I thought you said the first step is to see what the customer wants.

Mr. DUNLAP. Right. The right customer.

Mr. FRANK. Because that is the application.

Mr. DUNLAP. Sure. I would like to at this point go to two of the questions that were asked of earlier witnesses. I think you would have a better understanding of what we're talking about if we answer them now.

The first one is the case referred to, the Midway case. That really did not talk about this entire pattern here. What that case was primarily about was a rom-code. What it really involves is taking software, which is copyrighted, and putting software into sil

icon. So in the case of this chip, if someone copied this chip, we would be protected since we have software in this chip. And as you can see, this area here is relatively random. It's just a bunch of functions, but they all look different. The design is all different. This is regular [indicating].

And the reason is this is what we call memory. All it can do is remember 1-0; it either has charge on it or no charge on it. There are a number of bits in there. So the Midway case refers to when you put the program in here, when you fill this up with a bunch of 1's and O's, is that protected? And I think the majority of the courts say yes, it is. The software in silicon is generally protected. So this particular area today is probably protected. The rest of this is what's not protected. The rest of this is the random portion that has no software, and it's designed by painstakingly having the designer lay out this electronic function into this pattern.

The other question from Mr. Kindness was with respect to a book. OK. You could take the contents of a book and convert it into 1's and 0's, electronic impulses. So, for example, today you don't write a book in handwriting, you don't write it on a typewriter; you probably write it on a word processor. Well, with a word processor, what you do is you take the words and convert them into binary signals, into 1's and 0's, and you store it on a diskette of some type, magnetic material. What could conceivably happen, instead of storing the 1's and O's on a magnetic material, you could store the 1's and O's in silicon.

But again it's the expression of the book which the author wants to protect. It doesn't matter whether you store it in words or you store it on paper. You store the words on a magnetic media or you store the words in silicon. It's all the same expression of the book. And I think the Chip Protection Act would not affect that at all. Mr. KASTENMEIER. If you did store it in the chip you would have, you could also erase it and remove the bits of information that it has in there?

Mr. DUNLAP. Correct. Actually, in some chips you can and some chips you can't. It's the kind of thing-well, let's take a calculator. A calculator always has to be able to add and subtract. How do you add and subtract? You put it in there with what we call read-only memory, which means you can always add and subtract.

But on the other hand, you don't know what numbers you're going to add and subtract. So when you put the numbers into the calculator, you have to be able to change them. So that memory is called random access, which you can change. So some you can change and some you can't change. Some you can change if you expose it to light, which is what you call an eprom.

Mr. KASTENMEIER. Most of these chips have a a certain storage capacity. If you try to store in a chip more than the capacity, it just will not take the information in bits. The chip may not be able to easily eliminate as you put on this tape that which was previously stored.

Mr. DUNLAP. That's right. But this chip, when you're talking about capacity, you're talking about a memory. So it means in this particular chip you would only be talking about this area. There is a certain capacity. All this is just random logic. It doesn't have any capacity.