While one could successfully argue the ‘read’ heads are evolutionary, the same is not the case with the write heads. These new write heads are revolutionary. Much like the ‘read’ side of the equation a bit of history is required in order to understand just how much has changed in one generation.
In the last gen, 24TB drives, the write head tech was based upon a “traditional” inductive coil. One that “writes” data by magnetically changing the polarity of (relatively larger) grains of iron embedded into the lattice of the platters. Put in layman’s terms, the writer used is/was a very small and precisely aligned electromagnet. Sorta-kinda like the ones used at a junk yard for picking up cars. But instead of picking up cars, it was flipping tiny grains of iron and magnetizing (or demagnetizing) them by quickly powering up the magnet… and turning it off. Tried. True. Tech that Seagate has literally decades worth of experience at working around basically all the known limitations of this type of ‘writer’.
Sadly, as we mentioned previously, there is only so small, and so tightly you can pack the iron ‘bits’… as that electromagnet could only be made so small and/or so powerful. Basically, at ~2.5TB per platter mark, the coercivity of the magnetic media required to minimize bleed over (aka splash damage, aka it silently writes 1 or 0s to multiple bits at the same time!) is too high as to make an electromagnet work (or at least fit) inside such a constrained footprint. Quite honestly, few thought Seagate would ever hit the 2TB per platter mark before hitting this wall. The very fact that they hit 2.4TB is a testament to their ingenuity… but great engineers or not, they cannot cheat the laws of physics forever and were at a hard limit with no more room left to maneuver.
(Image curtesy of https://blog.westerndigital.com)
Thankfully, this was a well-known looming issue. So well known that decades ago the idea of Heat Assisted Magnetic Recording was born…as the coercivity (aka the intensity of the external magnetic field required to reduce the magnetization of a ferromagnetic material to zero after it has been fully magnetized or saturated) of magnetic material changes at given temperature points. Put in layman’s terms, “coercivity” measures how much power you need to use to hammer a magnetic ‘bit’ into changing from a 0 to a 1… and said power requirement wildly varies depending on how ‘hot’ or ‘cold’ the material is. This change can be so large that at higher temperatures, “hard” materials once deemed unusable suddenly become optimal choices. In fact, they become highly sought after as the chances of a random bit flip (aka “bit rot”) at normal operating temperatures (room to say 100F) are so low as to be non-issues. This is an actual issue that previously ‘soft’ materials could (but rarely did) suffer from.
(Image curtesy of https://optics.ansys.com)
The problem is we are talking about 800degrees (and above) Fahrenheit. Focused on a teeny tiny spot. If that is not tough enough of an engineering problem, HAMR requires it to be done as close to instantons as possible… as this heating is the limiting factor on HAMR’s write speed! Then, for bonus points, said material has to cool down (enough) before the next rotation completes so as not to bake the sensors in the ‘read’ head that are located mere millimeters in front of the write assembly. Oh, and just to make it spicy, none of this heating/cooling cycle can impact the alignment of the floating heads. Which are floating less than a hair’s breadth above the platter.
This is why a “write head” in a HAMR drive is different from that in older PMR or even SMR drives. Where once there was basically one main component, there are now three. The laser. The focusing array. The inductive coil. All in a package dubbed the ‘Plasmonic Writer’. A package that cannot be much bigger nor weigh much more than it did in previous non-HAMR designs.
So what do each of these components do? The near-infrared Laser Diode, or in Seagate speak the “nanophotonic laser”, fires a burst of coherent(ish) light in the general direction of the super lattice. While that is bit hyperbolic in how loosey-goosey a 810nm laser’s beams width can naturally be… when dealing with targets sized in the single digit nanometer range the fact remains that on its own this integrated ‘Laser’ would either heat too big a zone per firing to be useful or take too long to be useful (without blowing out the power budget that is).
To overcome this, Seagate uses a focusing array. Okay, technically, Seagate does not call it a focusing array nor even the more technically accurate optical waveguide. Instead, they break it down into two key sub-components… neither of which has array or guide in their name.
The first part is the photonic funnel, and as the name suggests, it “funnels” all the coherent(ish) light into an even tighter waveform. Allowing the vast, vast, vast (Seagate’s marketing claims ‘all’ but we highly doubt that) majority of the Laser emissions to accurately enter the center of the “quantum antenna”.
(Image curtesy of https://blog.westerndigital.com)
This “quantum antenna” in turn converts the LASER into quasiparticles called “plasmons”… or quantum-level oscillations of electrons induced by light. In layman’s terms, and yes, that too is a theme of this review, it turns the coherent light beam into an energy beam. A 1950s “heat gun” or “Martian Death Ray (of Doooooom)” if will. A highly accurate heat ray that is concentrated enough to darn near instantly heat a tiny ~30 to 35nm spot on the superlattice to above 800°F. At this point, the old school (with new school dimensions) electromagnet (“inductive coil” if one wants to be pedantic) crosses over this hot spot, changing the central ~5nm zone’s magnetic polarization into whatever has been called for. Then the arm continues it sweep past the quickly cooling spot. Rinse and repeat, and you have how HAMR writes to the superlattice.
