You’re going to have to talk to the metal bed about behaving if you want that to work.
Jokes aside… Metal conducts heat, very efficiently, if you just try to heat one section your heat will dissipate in all directions and you’ll just…
A.) End up driving that small section into oblivion
or more likely
B.) Never hit your temperature set point.
You’d also need a sensor at each location.
The only FEASIBLE (take that with a huge grain of salt) way you’d get around this is by constructing a bed made of individual “tiles” with a thermal insulator between them and individual heating elements and sensors on each one.
But then you’d have an entire laundry list of new issues to deal with.
I think putting 9x equally spaced TMP102 (+/- 5$) (single I2C line) on an aluminium backed board, should be good enough for a temperature bed mesh, a good balance of price and temperature accuracy
But the bed is not a rod, it is a thin 2.5mm aluminium sheet.
Especially in a chilled chamber, unheated exposed portions will not emit heat. While the printed on section will have plastic insulation on top.
And if lateral heat loss is a problem, it would make sense to then use a less conductive material like, steel, stainless or titanium.
It also is easy to bump of the number of resistors and grid rows as needed for finer control of where the heat is applied.
Here is a few thermal images of heaters behind metal sheets, you can see the portion behind the heater than where no heat is emitted
The heat gradient depends on many factor,
the power of the source, more power = sharper heat gradient
ambient temperature and air movement = colder air and more air movement means sharper heat gradient
heat conductivity of the material = less conductivity means sharper heat gradient
insulation = region with insulation to less insulation will have a sharper heat gradient
thickness = thinner sheet means sharper heat gradient