What is an electrostatic chuck (ESC)? How vacuum tools hold the wafer, and why helium flows behind it

What is an electrostatic chuck (ESC)?

An electrostatic chuck (ESC, sometimes called an e-chuck) is the stage a wafer sits on inside vacuum tools for etch, ion implantation and film deposition. Electrodes are buried inside it; when a voltage is applied, static electricity pulls the whole wafer flat and holds it there, through a very thin insulating layer on the surface. It also looks after temperature: helium flows between the back of the wafer and the chuck (helium backside pressure) to carry heat down.

Why a vacuum tool needs static electricity, and why the gas is helium, is on the glossary card. This article goes one layer further: how the static grip works, how the helium is watched, and how a wafer is gripped and released.

Why does the wafer need holding? How is a mechanical clamp different from an electrostatic chuck?

A wafer stays in the chamber for anything from tens of seconds to a few minutes, and the whole time it has to lie flat, stay put and keep a steady temperature. At ordinary air pressure, tools for coating photoresist, probing or dicing use suction chucks: air is pumped out from underneath and the atmosphere above presses the wafer down. Inside a chamber there is no atmosphere to do the pressing, so that trick stops working.

The other approach is a mechanical clamp: a ring presses down on the edge of the wafer from the front. It is simple, but it brings three problems:

• The clamped edge isn't processed: the strip under the ring can't become chips.
• It touches the front: as the ring presses down and lifts off, the contact and rubbing can shed particles.
• It only holds the edge: nothing holds the middle, so with gas flowing behind the wafer, the temperature across it is harder to keep even.

An electrostatic chuck touches only the back and grips the whole surface at once, so the wafer lies flat and the front stays clean. Many vacuum tools now use one.

How does static electricity hold the wafer?

Picture two metal plates with an insulating layer between them, connected to a power supply: the plates take opposite charges and attract each other. The electrode inside the chuck is one plate, the silicon wafer is the other, and the insulating layer on the chuck's surface sits between them.

This pull depends strongly on distance: the closer they are, the tighter the grip, and a little more distance makes it fall off sharply. That is why the insulating layer is very thin and the surface very flat. Where the wafer is warped, or a particle is stuck under it, that spot isn't held tightly.

Depending on whether the insulating layer conducts at all, there are two common types:

• Coulomb type: the insulating layer conducts almost nothing, and the charges stay on the electrode and on the wafer, pulling on each other. When the power goes off, the wafer is released quickly.
• Johnsen-Rahbek (J-R) type (named after two Danish engineers): the insulating layer is made to conduct just a little on purpose. A tiny current flows to the surface where the wafer and the chuck touch, and charge gathers there, so the same voltage gives a tighter grip. The cost: after the power goes off, that small current takes longer to die away, and the wafer is released later.

There are also chucks whose electrode is split in two halves, one positive and one negative, which can grip without waiting for the plasma to light.

Why does helium flow behind the wafer?

The plasma keeps pouring energy onto the wafer, so it heats up, and when its temperature is a little off, the shapes etched into it come out different (see "What is plasma?").

In a vacuum there is no air to carry heat away (see "Why do semiconductor tools need a vacuum?"). The wafer does lie on the chuck, but under magnification both surfaces are bumpy, and they really touch only at tiny points, like two sheets of sandpaper laid on each other. The gaps between those points are vacuum, and heat barely crosses them.

So helium is fed into that gap. The gas atoms bounce back and forth between the wafer and the chuck, carrying heat down to the metal base below, which has coolant flowing through it. Fine grooves on the chuck's surface spread the helium across the whole back of the wafer.

A little more helium pressure carries heat a little better, but that pressure is higher than the pressure in the chamber, so it pushes the wafer up. The chuck's grip has to be strong enough to hold it down.

What does a rising helium flow mean?

Helium slowly leaks out from under the edge of the wafer into the chamber, and a little leakage is normal. The tool keeps the backside pressure at its setpoint and tops up whatever leaks away, so the flow it feeds in is the amount leaking out: the higher the flow, the bigger the leak. There are three common causes:

• Something on the back: a single particle props up a small patch of the wafer. One university cleanroom's operating procedure says it plainly: if the helium leak is too high and the tool alarms, clean the back of the sample and try again.
• A warped wafer: it's hard to get the whole wafer to sit flat, so gaps stay open behind it.
• An old chuck: as its surface wears and picks up deposits, the grip and the fit slowly get worse.

With too much leakage, the temperature can't be controlled; what happens next is in "What is plasma?".

How is a wafer gripped and released? Step by step

1. Set it down: the robot arm carries the wafer into the chamber, lift pins rise through the chuck to take it, the arm pulls back, and the pins lower the wafer onto the chuck. The pins hold the wafer up so the arm can slide out; without them the arm would be stuck under the wafer.
2. Apply the voltage: the electrode is switched on and the wafer is pulled flat. It has to be flat before the back can seal in the gas.
3. Turn on the helium and check: does the backside pressure come up, and is the flow within range? This is the check that the wafer really is held. If it fails, the tool usually won't go on, and throws an alarm.
4. Run the process: the plasma is lit and the etch runs, with the helium flow watched the whole time. A steady temperature means the whole wafer etches the same; if the flow shoots up, the tool stops.
5. Stop the helium and pump it out: when the etch is done, the helium is switched off and what's left behind the wafer is pumped away.
6. Discharge (dechuck): the voltage is switched off, and a stretch of alternating positive and negative voltage, applied while the plasma is still on, can neutralize the leftover charge. Switching off doesn't always clear the charge at once, and the J-R type in particular takes longer.
7. Lift and hand off: the lift pins push the wafer up and the arm reaches in to take it. If the charge hasn't fully cleared, the wafer is still stuck to the chuck, and when the pins push, it may slide, jump, or even break.

What happens on the floor when the chuck won't grip or won't let go?

• It won't grip at the start: the helium pressure won't come up, the tool won't go on, and a message along the lines of "wafer not chucked" appears. This is an interlock, the tool stopping itself; pressing reset over and over doesn't help, and the wafer just gets pushed around once more (see "What are tool alarms and warnings?"). Check first: is the back of the wafer clean, is it warped, what do the chuck's power supply and leakage current show, and are the helium line fittings tight?
• The flow creeps up day by day: the chuck is wearing slowly, and each wafer is still within range; only the trend shows it (see "What is FDC?").
• The chuck has to be replaced: open the chamber, swap the part, pump down, leak-check, then do a qual run; only when that passes can product run (see "A tool has been fixed. Why does it still need a qual run?").

In fab24hr

Tap "Clock in", choose "A day in the fab", then on the "Your role today" card tap "Production fab", swipe to the equipment engineer and tap "Etch" under the name.

• Late-night shift, "Three Calls": at 03:40 Process calls: Chamber A of ETCH-03 threw a helium backside pressure alarm at 02:47. You're at home and can only see the night shift's photo of the back of the tool: helium flow in Chamber A is 14.2, Chamber B usually reads 8, but that's the result; what you're looking for is the ESC leakage-current lamp. Whether to get the vendor to bring an ESC in the morning right now is up to you.
• Morning shift, "Chamber open": the old ESC you take out shows surface wear, and an arc burn too. Wafer 7 may still be in the chamber, and you can back it out yourself in manual mode: check that the chuck has fully discharged, then raise the lift pins. The new ESC goes in with eight bolts in a star pattern, so the chuck sits flat; the leak check may find that the O-ring on the ESC helium fitting isn't seated. Skip the leak check, and the qual wafer's helium flow goes over the warning line, so the chamber has to be opened again.

Swipe to Manufacturing instead, and on the late-night shift, if you press reset once before making the call, the tool throws the same alarm with one more line underneath: WAFER NOT CHUCKED.

To watch a wafer go onto the chuck, open the plasma etch chamber, tap the electrostatic chuck under the wafer to read about it, then press "See it work" to see the lift pins set the wafer on the chuck and raise it again afterwards.

The companies, people, tool IDs and numbers in the game are fictional; how electrostatic chucks and backside helium work comes from public, general sources. For other jargon, see Chip fab jargon in plain English.