Researchers create laser light interconnects on silicon

Really ????

And what about the energy needed for the laser beam ?

This sounds like as much sense as the 'hydrogen economy' !!!!

Graham

On a sunny day (Mon, 18 Sep 2006 11:12:34 +0100) it happened Eeyore wrote in :

Think, just for a moment, the article mentions 500x speed. We are at say 4GHz, so 2THz. The parasitic C and L from a 1 mm chip connect, you can calculate that no? Also calculate the R, then apply 2THz in your spice, and see what comes out at the other end as peak to peak signal. Tiny laser dots, nanometer size optical wave guides, and tiny Si photo diodes are _the only way_ to transfer a signal at that speed, unless you can find a way to make such a small transmission line in silicon with proper impedance matching to the driver and receiver. And, termination = losses. An other advantage is that you can take the signal, serialise, go optical, and of chip with a fibre, all on the same silicon. So perhaps no longer 1000 pin chips, but just like we got rid of the flat cables in the IDE drive, go serial, an optical chip interconnect. All on silicon... see how the CMOS camera sensors are making inroads into _anything_ replacing more complicate CCDs, and going where these are not feasible. All on silicon is important.

This is the future, in my view.

Yeah, all we need to complement it is them holographic memories I've been hearing are just around the corner... for the last >30 years.

Wha?... bad as that? It's just another fraud?:-)

On a sunny day (Mon, 18 Sep 2006 08:49:23 -0400) it happened George Macdonald wrote in :

Hearing about? They will be at IBC 2006 in Amsterdam in a few days.

Know anything about Ikegami?

Ever used one of their cameras?

You are getting behind, read the wrong papers perhaps?

Juppy hupppy wuppy duppy

EE Times has a page that announces a similar breakthrough every week, but I haven't heard of any of these breakthroughs going commercial.

The laser makes heat, its driver makes a lot more, and the detector speed will be dominated by capacitance. True, the middle part, the optical waveguides, will be fast, but that's always been the easy part.

The NYT article is full of obvious errors.

And didn't Intel just pull the plug on this technology? Yeah, here it is:

I think they paid a few per cent of what Intel had invested; Intel just dumped it to get it off the balance sheet.

John

Don't you love that term "Off-Load"? Doesn't that mean "Dump" ?:-)

...Jim Thompson

On a sunny day (Mon, 18 Sep 2006 08:34:29 -0700) it happened John Larkin wrote in :

Not exactly this sort of thing.

It is now at an university, because the manufacturing is so simple, it is reasonable to expect some applications soon (say 5 years). It is not like fusion power that moves exponentially in the future, now at 100 years from now. It is a low $$ thing that any semiconductor company can play with.

yes

Nope, you must have heard of switching transistors.

That is not exactly true, yes, if you use a photo-diode on a high impedance input opamp, but even 30 years ago we used feedback to lower input impedance... in Vidicon tube camera preamps.

mm, I think _there_ are a few things that would be 'new territory'. At least for me.

Intel also killed their DLP chips, Texas is doing well with those I think. Intel is headed by a marketing druid, possibly assisted by some bean counters, and lawyers to sue everybody who uses the word 'inside' on their website. Marketing druid .. engineers fired (read the news) ...

It is a project at an university. Who will manufacture this is the one who buys the patent. Intel is irrelevant to this.

Well, seems our view are not the same, time will tell how this will pan out. But I feel OK about this one, other then 'quantum computer', 'nuclear fusion', bushism, what not.

To produce a THz modulated signal?

You don't seem to appreciate the trade-offs involved in designing switching transistor drives.

where the dector capacitance messes up the high frequency noise levels. Been there, been screwd by that.

-- Bill Sloman, Nijmegen

So can you explain to an EE newbie exactly what the break-through claims to be? Isn't this like a Laser Diode (laser on a chip) except on a silicon substrate (if I'm getting this right)? The idea is silicon is easy and cheap to manufacture right? What was the more common way? Sapphire? Is that more expensive because of the process or the material (sand versus rare gem)?

Very sorry for the rookie-ism, milk

I think it might be the bonded GaAs layer (can't make lasers on Si, at least not easily) and the idea of using trenches as fibers.

Yup, but you could already epoxy and wire-bond a compound-semiconductor laser on top of a silicon chip. I don't see a breakthrough here. Still no silicon laser, still no bulk deposition of non-silicon lasers on silicon.

Lots of people have developed SiO2 and polymer and PLZT optical waveguides on silicon, for 10 years at least, and a couple of them have even managed to stay in business. Nothing new here, either.

John

On a sunny day (18 Sep 2006 09:52:52 -0700) it happened snipped-for-privacy@ieee.org wrote in :

Well, let us say that we once will be at a point where we can switch at .1ps.... Sure 2THz is a limit. That 2THz would not be a modulation limit for _light_ I think?

I am the first one to admit I have never had anything switch that fast. But what basic physics argument exists against switching light in the Thz range? Maybe you only need to inject some electrons to make the laser work. Sort of a threshold... I can imagine some mechanism like that, but I think there will be no need to _analog_ modulate so the driver is always in a linear range. Also there is the issue of static and dynamic logic design, if you can send an data impulse and clock impulse, then you can energise very short periods only, to flip a bit elsewhere.

Bill, I dunno enough about that, I take your word for it, never worked at THz speeds, photo diode, photo transistor, sure there are limits. Limits are there to be overcome.

Research advances. First I would expect them to make some device with optical on chip interface to fiber... maybe for home high speed networks.... Not that I need more then DSL, but some will want HDTV via fibre. There is a market, there is a need, and so then there is funding for further research. I most certainly do _not_ believe in things like making HDMI interface wireless and messing up the whole radio spectrum like the Intel marketing druid seems to want?

Well, my best answer for now (10 in the evening), all I know. Time will tell.

Most of the problems related to get everyone a, say, 100Mbps data connection revolve more around the labor cost associated with installing copper, fiber, or whatever -- particularly the "last mile" thereof --, and not a limitation on current technology.

In other words, even if you could build a single 2THz optical switch today for pennies, I don't think it would really change how long (or how costly) it's going to be until you can watch The Simpsons, Family Guy, Robot Chicken, Futurama, and Sealab 2021 all simultaneously in HDTV.

What's Sealab 2021 like? Never heard of it

martin

Back to insulting people just for the sake of an insult? Still can't get how to conduct a discussion in a civilized way? You really nead to learn the basic manners before posting your childish insults.

NNN

Here, read the account in EDN

It was Intel and UCSB that called it a breakthrough.

I've been reading of these 'optical breakthroughs' for the past 25yrs. Don't hold your breath.

What do you not understand about >30 years.

I'm not impressed by techno-weenie talk!

Oh, it's been longer than that - holographic was all set to replace core memory back in uhh, ~1969.