A Terabyte of Data on a Laptop Hard Drive

KaosConMan writes: "TechnologyReview.com has an article describing a new technique being developed by General Electric and IBM to further decrease the size needed to magnetically store data. This new technique could produce 150 gigabits per square centimeter-- that's ~57,000 songs on an iPod or a terabyte on a laptop size hard drive!"

Comments

Re:Yawn

[Com2Kid]

AC thus spoke:

This is already obsolete. Terabytes of information on a creditcard sized medium have been announced years ago.

And it was replied:

Along with anti-gravity, ways to earn infinate money, and the secret of eternal youth.

The only difference is that this announcement comes from an actual lab with people who have actual degrees.

1TB iPod

[i1984]

I can't wait for my 1 terabyte iPod!

Lets see...a 10GB iPod costs $399 -- that's $39.90 per gigabyte. So extrapolating to 1000 gigabytes...yes...we'll have a $39,900 iPod!

I'll take two of them; just let me find my checkbook. Oh shoot, I must have left it in the McLaren...I just hope it isn't in the Bentley. Well, I'll just have my chauffeur bring it 'round in the helicopter. Do you have a pen I could borrow?

The real question is whether this technology will be better (and cheaper!) than any other high capacity memory when it's (maybe) released in 2008.

I have my hopes pegged on static random access technology that doesn't depend on disk technology. Instant power on and no difference between storage and application memory are likely to be killer technologies.

Security implications?

[CProgrammer98]

The article states that the new technology will only use one magnetic grain per bit as opposed to the hundreds currently used.

I wonder if this means that once a cluster is overwritten, there is no ghosting effect that could allow the previous data to be retrieved. Once the data's gone, it's gone. A single magnetic grain can only be set one way!

So NSA or whoever won't be able to retrieve those docs you wiped just before they busted into your home/office....

In the light of this, this tech. it is probably not in the security industry's best interest!

So as well as getting space for all your music and porn, you don't have to worry about the data persisting on your drive when you want to remove it... all in all a good thing!

Re:How much capacity is 57,000 songs?

[Jugalator]

Hm...

1 Tb is 10^12 bytes right? Ok, not exactly, but the correct magnitude? :)

1 * 10^12 / 57 * 10^3 = 17,543,859 bytes/song.

So it seems the author is using 17 Mb mp3's or something... Must be one of those "wooo i need l33t 640kbps mp3z cuz 256kbps dont r0x0rz".

Or it's just an approximation. :-)

Re:Circular?

[clark625]

You would think that a pyramid would be a neat advance, but it's probably not feasable. The costs associated currently are too high for consideration.

The problem with pyramids is the formation. Hard drive manufacturers like to use Si, polymers, metals, and the like. Doing a photolithography step on them isn't difficult--but finding an etchant that will prefer to etch in a pyramidal shape is rather tough. If you can find one, it usually will etch down such that you end up with pyramids facing into the substrate.

Not that it can't be done. I recall some work done creating pyramids on GaAs substrates. It may be extendable to other material systems as well. But GaAs is a zincblend crystal structure--not a diamond structure like Si is. The zincblend readily makes itself agreeable to off-axis etching (especially if you get the proper offcut wafer).

Maybe if the polymer could be self-assembling and would itself produce the desired pyramidal shapes, then everyone would be happy. Doing a metalization step over top of that would not be difficult at all. But I'm not terribly great at dealing with polymers--I don't know where the limits of self-assembly are. I'm sure someone else does, though--and I know of several journal papers you might consider.
Anyways--your idea is good, but it's an ideal. At this moment, there doesn't really exist a practical way to make metalized pyramids without steps that would either be prohibitively expensive (I'm talking processing time here), or steps that would require too large of feature size.