I have a device with too many external contacts exposed to ESD, I search for a way to reduce cost of ESD protection.
According to datasheet BAT54 can stand 1 A non-repetitive peak forward current. So if I put a 10 kΩ resistor in front of it, it should stand ESD up to 10 kV. Will it work?
Is your resistor rated for 10kV?
This sort of thing works, within its limits, but you will usually want a physically larger then typical SMT part for the resistor, maybe a larger sort of MELF or a 1206 with a slit in the board under it to increase creepage.
Another useful trick if the input signals are slow is to slug it with a 100nF or so of cap, the usual ESD test sources are of the order of 100pF or so, which means that if you contrive for your input to look like 10k+100nF you inherently have a 1000:1 reduction in pulse height before you even consider the resistor and clamp diode.
If you can, the easiest way around it may well be to use the MCU's input pin protection. Some are rated quite well and just an external series resistor is enough to pass the ESD testing and have reasonable performance out in the field. It's not a free lunch, though: once you let transients deep into the board, various ways of coupling to other signals - be it electromagnetically or by conduction - become a concern, and it may be more trouble than it's worth. In a simple device where an MCU is a "dominant" device in the design and is directly connected to a lot of external I/O pins, it may be of some use.
According to datasheet BAT54 can stand 1 A non-repetitive peak forward current.
When using anything not explicitly specified for use in ESD protection you have to make test articles and test them anyway. And perhaps test the diodes from several vendors to qualify them all at once - there well may be differences that change their behavior in this application.
The maximum non-repetitive peak current rating may well be hard to translate into ESD suppression behavior - see what test circuit they test it in, and compare to standard waveforms used in ESD testing. With fast transients, the parasitic capacitances that shunt the resistor come into picture, so it's not exactly trivial to predict behavior without testing or lots of relevant experience and/or correctly used modeling tools (expensive ones at that).
It's also not unexpected to find devices that "look worse on paper" yet outperform those that look "better" in terms of specs that are somewhat related to ESD suppression applications but not specific to them. So it may well be that, say, a cheaper switching diode will work just as well. Or not. It's all tradeoff between how many things you can try. A well thought out ESD suppression evaluation board helps here - but you'd be designing it.
I have a device with too many external contacts exposed to ESD, I search for a way to reduce cost of ESD protection.
Many vendors are offering compact ESD protection chips for multiple lines, at low cost for mobile devices where ESD is a big concern. It may even turn out that discrete diodes and resistors cost more. There are hundreds of chips to choose from in that market I believe.
In regards to reducing costs: you're making thousands or tens of thousands of it then, right? Because if not, then it may be a tight squeeze between savings in component cost vs. saving engineering time with all the overheads. BOM cost optimization trades off engineering time. There is money on both sides of that balance scale. I'm sure you know it, it's just worth pointing out for anyone else who runs into this and may be forgetting :)
Arrays are a good way to save component count. Have you considered a TVS array such as Diodes D1213A-04S-7?
Resistor and capacitor arrays can also be used to good effect.
