22
\$\begingroup\$

I know that in virtually every case, PCBs start as a substrate with a layer of copper laminated on top, which is then etched away, leaving the traces. But why are the traces not added on an empty substrate instead?

I am well aware that the etching process is standard and very well understood by now, but isn't it a bit wasteful to etch that much copper away (or leave that much unused copper if you're maximizing it)?

Is the reason for this simply that additive manufacturing processes are just too costly, far outweighing the cost of etchant or copper?

\$\endgroup\$
3
  • 14
    \$\begingroup\$ In general, it's easier to remove material than it is to add. Think about most fabrication operations. They cut, drill, grind, and otherwise remove material. \$\endgroup\$
    – DKNguyen
    Commented Nov 7, 2021 at 17:51
  • 13
    \$\begingroup\$ How are you thinking of applying copper to the board in a way that has such good adhesion? Also note that many boards have large ground pours, etc., so very little copper is removed. You've also got to handle vias. \$\endgroup\$
    – Transistor
    Commented Nov 7, 2021 at 17:54
  • 4
    \$\begingroup\$ One word: cheaper. \$\endgroup\$
    – winny
    Commented Nov 7, 2021 at 18:09

7 Answers 7

35
\$\begingroup\$

I know that in virtually every case, PCBs start as a substrate with a layer of copper laminated on top, which is then etched away, leaving the traces.

This is not fully correct: in modern PCB manufacturing processes, copper is both added and removed.

Often the PCB substrate begins with 17 µm thick copper. First step is to drill the holes, and then an electroless process is used to form a thin layer of copper on the inside of the holes. Then the trace mask is applied and electrolytic copper plating starts. This process adds copper to all exposed areas, bringing the total copper thickness to 35 µm typical (or any other thickness needed). It is important that the thin copper everywhere under the mask connects the traces together so that electrolysis works (it needs electricity to conduct).

Finally, there remains only the original 17 µm of copper to remove in non-trace areas. So the full thickness of the copper does not have to be removed. Furthermore, the copper in the etchant can often be recycled into copper for the plating process, though that requires some chemistry.

This Altium article on PCB processing provides a great overview, including this diagram: Processing diagram

\$\endgroup\$
1
  • 4
    \$\begingroup\$ Excellent. This should be the accepted answer as it highlights that a modern professional process is in fact additive as well as subtractive. Another issue is the resolution and feature size. A household printer can create a photoresist mask to a resolution of 1200 dpi, whereas professional resolution is much greater (with the theoretical maximum resolution being the size of molecules and atoms). \$\endgroup\$
    – tim
    Commented Nov 8, 2021 at 11:29
18
\$\begingroup\$

Part of the PCB manufacturing process involves bonding copper foil onto the substrate. This is hard enough to do well when it's done across the entire panel.

But why are the traces not added on an empty substrate instead?

Because the traces are tiny, and need to be bonded onto the panel in such a way that they are laid out accurately with respect to one another and won't come off with soldering or mechanical stress. Short answer -- no one has figured out how to do it economically.

There are PCB manufacturing processes that involve printing a PCB with conductive ink. I don't think I've ever seen such a board, even in a cheap consumer-grade device. Simply based on the fact that it hasn't taken over the world, I think we're back to the notion that etching is easier.

isn't it a bit wasteful to etch that much copper away

Yes, it's just a little bit wasteful. But it's a lot less wasteful in commercial operations, because the copper can be reclaimed from etchant in a way that also regenerates the etchant. For the price of some electricity, the PCB manufacturer has some copper bars to sell back up the chain, and doesn't have to buy as much etchant.

Or leave that much unused copper if you're maximizing it?

Usually when you leave copper on a board it has nothing to do with concerns about using up copper or etchant -- it has to do with using that copper as a ground plane, or bus bars, or some other reason that you want to have copper on the finished board.

\$\endgroup\$
5
  • 2
    \$\begingroup\$ Just to add - if you have large "bare" areas on a PCB, the manufacturer will often want to add copper balancing, which essentially puts a grid of copper back in those areas. It's done do both minimize the amount of etchant "used up", and also to help ensure a uniform thickness when bonded. \$\endgroup\$
    – SiHa
    Commented Nov 8, 2021 at 8:02
  • 1
    \$\begingroup\$ I've also never handled a printed-trace PCB but from what I've seen/read they're not compatible with standard manufacturing workflows. On the first boards it was as fundamental as not being able to solder to them with normal solder; this may have changed. Instead they're more for integrating simple circuits into rapid prototypes. \$\endgroup\$
    – Chris H
    Commented Nov 8, 2021 at 9:50
  • 4
    \$\begingroup\$ @SiHa -- for multilayer boards, oh yes. I was on a project where we learned that at the same time as our board manufacturer, but only after some board failures. Because electrically it needed to be non-conductive, we ended up with fields of copper dots to establish the inter-layer spacing, and we were thankful that it worked. \$\endgroup\$
    – TimWescott
    Commented Nov 9, 2021 at 0:33
  • 1
    \$\begingroup\$ @TimWescott - yes, it was usually a grid of isolated squares, for us, if memory serves. \$\endgroup\$
    – SiHa
    Commented Nov 9, 2021 at 7:17
  • \$\begingroup\$ I've seen conductive ink used to make a digitizing tablet (a grid of traces that measures the position of an object) \$\endgroup\$
    – Rich
    Commented Nov 11, 2021 at 0:34
7
\$\begingroup\$

What would it take to selectively add Cu to a PCB? You would need a mask, and then a sputtering machine that can move in x-y directions. For fine resolution, neither would be cheap but let's say both these provisions are there.

At what temperature would this process be run? Cu melting pt = 1085 C and Boiling pt = 2562C. FR-4 material will melt near 130C.

Therefore, Cu Foils (produced separately) are attached to FR-4 panels using silver epoxy resins to provide reliable adhesion. Later on, after a masked exposure Cu that is not needed is etched away chemically. Commercially, Cu that etches away can be retrieved and reused.

I have heard of silver paste that can be used like a pen to do minor corrections and/or to route traces but have never used the technology myself. Luckily, it is always possible to use wires or 0ohm bridges to correct minor problems on pcbs.

\$\endgroup\$
4
  • \$\begingroup\$ Thanks, I find the first half of your answer to be a very good example of what problems might pop up with additive manufacturing! \$\endgroup\$
    – David Cian
    Commented Nov 7, 2021 at 18:42
  • 2
    \$\begingroup\$ Manufacturing time would be another consideration. Whereas, an exposure takes a few seconds and then the unwanted Cu can be etched away, an x-y machine will take ages compared to that process. Expensive machinery and time. Exposure lamp is cheap, CNC machinery will not be. \$\endgroup\$
    – Syed
    Commented Nov 7, 2021 at 18:48
  • 1
    \$\begingroup\$ "Temperature" really only exists in an equilibrium state. A tiny amount of molten copper won't carry enough heat energy to melt the FR-4 bulk. That's not the problem. But how would the solidified copper then adhere to the FR-4? \$\endgroup\$
    – MSalters
    Commented Nov 8, 2021 at 13:35
  • 1
    \$\begingroup\$ @MSalters, if sputtered Cu could adhere, the process would reduce to a Cu printer head sputtering tiny pieces of Cu, OR perhaps coat the whole surface with adhesive and sputter tiny solid Cu particles where needed. There would still be an annealing phase and Cu being a good conductor of heat would simply not allow localized heating solutions. Manufacturing time and complexity would still go up compared to exposure and etching. Since this mechanism/material is undiscovered, a foil is still the best first step. And, since Cu is recovered, there seems to be no urgent need to rock the boat. \$\endgroup\$
    – Syed
    Commented Nov 8, 2021 at 14:17
7
\$\begingroup\$

There are CIC - printed "conductive ink circuits" that use silver inks printed on flexible film (usually PET - the same material used to make soda and water bottles). PEN is used if higher temp or more dimensional stability is needed.

Silver ink is more expensive than copper, the screen printing process is not as high resolution as copper etching and silver ink cannot carry as much current as copper. Finally, it is not as durable as copper.

However, it can be very thin and flexible for thin stack ups like inside mobile devices or wearables. Light weight and form to a part. Some companies are also using these film circuits for Film insert molding and make the film part of the structural plastic part. In that case, the film is usually polycarbonate instead of PET.

\$\endgroup\$
4
\$\begingroup\$

One important aspect that other answers didn't touch:

Conductive traces and areas on the PCB need to be contiguous in order to serve their purpose.

Additive manufacturing processes are not very good at this. They all involve separate particles that are expected to come in contact in the process. They may as well fail and there indeed will be a great number of oportunities to fail.

And what is worse, to fail partially, passing the QA initially and bringing later a whole lot of the central EE problem - the bad contacts.

In contrast, the copper foil is pretty much contiguous before etching.

Etching may fail, too, with traces being over-etched (leading to no contact where it has to be) or gaps under-etched (leading to shorts or insufficient gap width). But: these failures can be tested for with unambiguous result and/or detected by visual inspection early in the process, generally before populating the PCB.

\$\endgroup\$
2
\$\begingroup\$

isn't it a bit wasteful to etch that much copper away (or leave that much unused copper if you're maximizing it)?

  1. Etched copper is not lost forever, copper cloride (or any other similar by-product) is electrolized back to metallic copper which is then recycled.

  2. Having more copper on the PCB than strictly necessary is good: it improves cooling, EMC, and reduces negative effects of ground loops.

\$\endgroup\$
1
\$\begingroup\$

The tracks and groundplane on a PCB are formed by removing copper from where it is not wanted, but the copper in the through-holes is applied additively- normally using electroplating. The result is called a plated through hole.

In the nineties there were attempts to build the tracks in the same way, starting with a conductive ink, but reliability was poor- bridging was one of the problems- so manufacturers decided to focus on copper recovery from the etchant to reduce the cost.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.