write coil on the head of a hard disk drive (updated with higher resolution images)

[rjlittlefield]

I recently had occasion to take apart a hard disk drive. My original motivation was mainly to harvest one of the platters to use a first-surface mirror, but of course I stuck various pieces under a dissecting scope to see what they looked like. I was confused by some of what I saw, or more accurately by some of what I did NOT see, so I dug in to take a closer look.

Here's a visual sequence, zooming in to focus on details of the write coil and its exquisitely small pole pieces that form the individual recorded bits on the spinning platters.

5X objective, cropped here to about 3.2 mm field width on subject.




Shot with a 10X objective, the structure of the write coils can be seen clearly inside the white circle that I've added. The view may be a little confusing because we are looking through both the side and the top of a transparent slab. The pole pieces extend from the center of the coil toward the surface of the head that flies next to the platter, culminating in that bright bar and associated small triangle, which in use are separated from the platter by only a few nanometers.



Closer, at 20X. The view of the write coils is starting to break up here, mainly because this large NA objective does not like to look through the clear slab at an angle. But the pole pieces are exposed on the surface, and some more detail of those is appearing here.




The following is as close as I can get, 50X NA 0.55, cropped to actual pixels. I have taken care to sharpen this with a simple USM (unsharp mask), because Topaz AI was starting to make some changes that I did not trust to accurately represent the structure.



I think that the active region of the pole pieces is somewhere around the cross of that T shape in the center of the circle that I've drawn. But I'm not sure that I have read the literature properly about that.

In any case, the active region is certainly too small to be seen clearly in this picture, because even at this magnification it's only about the size of 1 pixel.

This drive is far from current technology -- it's only 250 GB, made in 2005 -- but even so just running the math based on platter size, number of surfaces, and number of bits, says that the available area per bit is only about 0.1 micron diameter. The number would be correspondingly smaller for say the 4 TB drives that I use for removeable backup. Note that the area per bit is a product of track spacing and linear span of each bit, so even the tracks have to be awfully close together.

I'll add more information in a later posting. Right now I really need to hit "Submit" and move on to some other work.

As always, I hope you find this interesting!

--Rik

Edited title, 2025/04/07

[WojTek]

Hi Rik,
Yes, very interesting story, thanks for sharing!
Best, ADi

[donpirhana]

Hello Rik,

Very interesting read.

I have a couple of drives that are ready for the rubbish bin (20+ year old 4GB pata drives!) and your post has me reaching for the screwdriver. Unlike my last post re mangled read/write heads, I hope I can extract the fiddly bits intact and without damage.

[rjlittlefield]

Thanks for the feedback and other info.

Don, I look forward to seeing what you find in that old 4 GB drive. My understanding is that "in the old days", reading and writing were both done using a magnetic coil, but these days only the writing is done with a coil while reading is done using one or another magnetoresistive effect. I do not know when the transition occurred, or how it might affect the visual appearance of a head.

Copying here some information that I posted a couple of weeks ago in another thread:

Here is one good article: https://hddscan.com/doc/HDD_from_inside.html

About 2/3 of the way down, that article explains that the black lump with the embossed pattern on it is called a "slider", although its real function most of the time is not to slide but rather to fly, holding the read/write parts a very small distance from the surface of the disk platter. The actual read/write elements are too small to be seen in this photo and are not shown in the article, although the article does draw a circle around where they are located.

Other articles explain that, depending on disk density, the magnetically active area has critical dimensions that are probably less than 0.1 micron. That size would be not resolvable by any optics in visible light. Scanning electron images can be found by searching Google images for "SEM hard disk head". One article I found that way is at https://www.researchgate.net/... 1-nm_Clearance . It dates from 2010, and is evaluating a slider design that flies the head only 1 nanometer away from the disk surface. I'm not sure what a typical height is for today's consumer-class drives, but I get the feeling that it's around 5 nanometers.

--Rik

[rjlittlefield]

One technical snippet worth mentioning... For the 5X shot, I was annoyed by large halos caused by the interplay between large aperture and deep subject and the way the light was reflecting off the metal. So I decided to stop way down, using an iris mounted behind the Mitutoyo objective, to a point where live view showed just enough resolution that I thought would look OK at forum size, as the start of a zooming-in sequence. The working aperture turned out to be something like effective f/55 (=208 mm focal length divided by 3.8 mm aperture diameter) instead of the normal effective f/18.6 . In addition to shrinking the halos, that let me increase the step size to effectively 200 microns (shot at 100 microns, processed every other frame). It did, however, cost a lot of resolution. Here is a comparison of corresponding frames, each focused on the write coil:



--Rik

[fdupre]

Hello everyone, hello Rik,
Excellent ! Photographing a hard drive head is a very good idea in my opinion. Your photos are very interesting.
Kind regards,
FRanck

[rjlittlefield]

Thanks again!

And, the how-it-was-shot story is now posted at viewtopic.php?f=8&t=46590 .

--Rik

[rjlittlefield]

Update, 1.5 years later. This is shot with epi illumination through a beamsplitter behind the objective, using a Nikon M PlanApo 40X NA 0.80 objective plus a 1.4X teleconverter, giving approximately 56X optical magnification. Focus-stacked with 0.5 micron steps.

Crossed eye stereo, +-8 degrees so 16 degrees total separation. Yes, the copper-colored elements sit behind the flat face where the white stripes are. The separation is about 6 focus steps, so visually about 3 microns. But that's looking through dense medium, so it's probably more like 4-5 microns physical depth.



Below is a closer crop. This is actual pixels on Canon R7 camera, so about 0.057 microns on subject per pixel in the image.



With a modern hard drive, say 4 TB, that means each bit is about the size of 1 pixel in this image. The active area of these read/write heads is a few 10's of nanometers across, and the heads are kept aligned with the disks to that precision, while the disk spins underneath them at typically 120 rotations per second (7200 rpm). As I understand it, only one head is used at a time and the data is transferred bit-serially, so 250 megabytes/sec transfer speed means 2 gigabits per second streamed through the active head while alignment is dynamically maintained within a few nanometers. The last drive I bought was 8 TB, quantity one in retail packaging, for less than $200. I have literally no idea how electromechanical devices can be manufactured with this precision for that price. SSD's seem trivial in comparison.

The drive shown here is not modern. It's only 250 GB. But the heads are still too small to image well with visible microscopy.

--Rik

[BugEZ]

I am a little late to the party but I wanted to toss in my 2 cents worth.

First of all amazing photographs Rik! Amazing what you see when you look. Even more amazing what you see when you look carefully.

In the 5x photo the large rectangular grey match book cover thingy immediately caught my attention. That is because it appears to have an etched micro pad. In the aircraft gearboxes I worked on prior to retirement we had oil face seals that incorporated etched micro pads to help the carbon seal float a few microns away from the flat steel runner. The non rotating seal was carbon as it helped avoid wear during startup. The etched pads in the rotating runner took advantage of viscous drag to pull air into the pad. The air became slightly compressed and pushed the carbon seal away a few microns. The air also pushed oil mist back into the gearbox reducing oil loss. This greatly reduces seal heating and prolonged seal life.

I believe that on these disc drives the head is not lowered onto the platter till they have spun up to speed. I am not sure what material they use for the match book cover thingy…

Keith

[fdupre]

Hello everyone,
This is incredible! This precision.
Thanks for sharing, Rik.