Some VLSI theory questions

With exams coming up, the pressure is one. If someone can answer thesequestion would be great.

Does the top metal layer generally has lower resistance per unit then all other layers? (I believe the answer is yes)

This question relates to self-heating:

In my course notes, there is a line saying "The top metal layer is far away from substrate and is isolated by field oxide that is very low thermal conductivity". The question is: Is that how we prevent self heating or is that a bad thing that top metal layer is far away from...

Guard rings. When we use double guard ring, we need to connect one guard ring to Vdd. It is obviously connected to different Vdd pad then the circuit. But these different Vdd pads can still be connected to the same power supply?
Signal routing. This one has to do with the return currents.As you know, the return currents are the highest in the nearest GND lines and decrease with the increase in the distance from the signal line.The question is: how come the current return in the closest signal line is negligible?

Thanks for all the answers.

I love it when people give you the question and answer and then ask you to justify their answer.

By "unit" I presume you mean "square." The answer is, usually top metal layer has a lower sheet resistance than lower metals, but there is no rule saying it must be for all processes. You should consult your design rules.

The top metal layer is far away from the substrate because there are a bunch of other metal layers below it between it and the substrate. There's nothing magic about the top layer, its just the furthest layer from the substrate. It's usually used for conveying power or ground. I don't really think "self-heating" is much of an issue with regards to metal. Power consumed in the power grid is negligible compared to other power consumers like transistors. The fact is that although the top metal layer is probably a good heat sink (and the oxide a good insulator, preventing you from using the metal as a heat sink), it sinks the heat into the package which itself is probably reasonably insulative, so I'm not sure its doing much to help or hurt. It may help spread the heat out a little if you were using it (at the cost of increased capacitance, which may be a bad thing). Reality is, the power grid is probably the best you are going to get as a heat spreader.

I've never heard of guard rings used for digital circuits. They are typically used to shield sensitive analog circuits or perhaps because of fear of latchup. As for whether or not you can connect it to the same power supply, the answer is it all depends on how much you care about the quality of the VDD. There is typically a large bypass cap on the outside of the chip, and inductance on the bond wires (not to mention all the other wires), so the VDD supply off-chip may be quite clean even when an on-chip supply is quite dirty. Its possible that with two different pads which are double bonded to the same pin one may be dirty and the other clean. Of course, its also possible they will both be dirty.

Okay, I've never heard of the term "return currents". The only thing I can imagine you are referring to is IR drop. As you move about on the chip, the voltage supply rails will change voltage because the metal has a finite sheet resistance and there is finite current, and so from any point on the supply rail looking into the "power supply" you'll see a different R. The further away you get, the bigger the R will be. Also, what you really care about is an integral of IR since the I will increase as you get closer to the supply (just like a tree gets thicker and thicker the closer you are to the ground) If you are right near a supply, R is probably so low its negligible.