You'll need to use a PCB fab capable of going down to 0.1mm trace width/spacing and 0.2mm holes, so I'd suggest JLCPCB. These are BGA chips and will require an electropolished solder stencil, ENIG finish, and sobriety from caffeine. But, these chips are a dollar, might as well not spend $40 on a board for them- just go with HASL and some flux. YOLO. Did you know you can use a 3D printer as a pick and place machine? Yeah that project is coming up soon.
These cute little BGA/CSP's were really tiny! Luckily the KiCAD default library had ATTiny20 footprints with silkscreen right at the component edges. So, there's only precise alignment and nothing else.
I used Kester 186 flux because it's tacky but doesn't char at the 235C melt point and dissolves readily in isopropanol. Had this been a BGA256, Kester 186 may not have been a good choice due to the cleaning issues of getting the isopropyl to dissolve it out and I could have used Kester 951, but honestly, I don't love that stuff since it doesn't hold the BGA in place with its thicc-ness.
As a side-note, if you could apply the Kester 186 with a stencil and create a serpantine or grid pattern, you would have a high enough surface area to be able to clean it out in an isopropyl bath in a reasonable amount of time. One way to do this is to cool it to a point where it doesn't wick away from the stencil. That's assuming you order a BGA stencil for a part that should not have solder paste... If you're designing a board with QFN's/0402/0603/0805 components and BGA/CSP's, when exporting Gerber's you need to export, then delete the chips before exporting the paste layer to avoid those components showing up in the stencil. They are flux only. Don't make the same mistake I did and burn $50 of chips because of my sheer hubris.
I used my new Miniware MHP30 heating plate which I can't recommend enough. It is truly a pleasure to have in the lab. Even the USB cables were selected thoughtfully, being silicone and flexy enough to avoid knocking the heating pad over. Best $100 I used all year!
Definitely use Kester 186 for smaller CSP parts. Also, be sure to get your flux from a reliable source since they're basically organic chemistry tubes and I know enough about org-chem to say that after sitting long enough at room temperature "shit happens". The stuff goes bad. The reducing ability drops and the viscosity stops being perfect. For a BGA system where you need the surface tension to perfectly isolate the balls and wick to pads, be sure your flux is less than 6-12 months old. Just order a new tube with your CSP's from DigiKey.
Inspection of the balls during reflow is a good practice to correct issues in the hot state and maximize your chances to touch the hot plate on accident. I recommend getting one of those cheap, $15 USB microscopes found all over eBay. They really work, and if they stop after a year, just buy another one.
This yielded a pretty nice shot of the CSP-12 ATTiny.
Since I see distinct balls, that indicates a lack of bridging and as a result, it's safe to assume the rest are OK, too.
The chemicals in the flux are best removed since who knows what will happen to them as they age. Better safe than sorry, so I tossed the completed board in some warm isopropyl alcohol. 90% is back in stock at pharmacies due to no more pandemic shortages, so I could finally clean again!