Etching: Digging nanoscale trenches

Beginner's introduction to etching.

To lay underground cables, pipes, building foundations, and other such structures, we need to dig the ground up and then lay them below the surface. A loose connection during laying can mean either digging up the place again, or in the case of building foundation, keep reinforcing the structure to keep it from falling. A well dug-in pipe or a cable can last for a very long time. A strong foundation makes for a sturdy building. Etching aids in the formation of lasting layers of a semiconductor device.

Figure 1: Location of etching in process flow.

Etching is the process of removing unwanted material from the substrate. There are some portions that need insulation in a conductive layer and some places which may require conductive material deposited into an insulating layer. Removing that part and subsequently depositing the required material is where it finds its use.

For all of our life on this planet, you can only exist in two states: wet or dry. The same goes for etching. Wet etching uses reactive compounds in liquid form to etch out layers and dry etching uses gas, or more specifically, ionized gas. Gas in this ionized state is called plasma. Ionization means the gaseous atoms are energized enough to either gain or lose an electron. The one that gains an electron is an anion (negatively charged), or if it loses an electron it becomes a cation (positively charged).

Wet etching involves immersing the wafer in a liquid that etches the top layer off. This etchant (compound that etches) should not react with photoresist, so that only the portions exposed are removed. The liquid, which is more often than not an acid, is very dangerous to touch. These acids have the capability of burning a hole straight through your body. There is no cure for an injury from this. All the compounds used in wet etching are enough to send an untrained person straight to the hospital or to their end.

Every element or compound used in CMOS fabrication has its own etching chemistry. Silicon has an alkaline (basic pH > 7.0) and an acidic chemistry. The acidic chemistry uses hydrofluoric acid (HF) as a component and alkaline uses potassium hydroxide (KOH) in the same way. Both of these compounds are insanely corrosive. Hydrofluoric acid has the infamy of being one of the deadliest acids on Earth, oddly enough it is a weak acid when it comes to chemistry standards. While the most corrosive acid title goes to Fluoroantimonic acid (HSbF₆), HF is combined with SbF₅ to create it. Potassium hydroxide (KOH) may not possess the same systemic biological dismantling property of HF, yet it can easily burn your skin. It is amazing how such deadly compounds can be used in our everyday lives. KOH is used to manufacture soap and works as a food thickening agent in yogurt, cheeses, cakes, and the like. It is quite similar to the pure sodium and chlorine problem we were told in school. Sodium is highly reactive and chlorine gas (Bertholite: Cl₂) was used to decimate troops during World War 1; and yet here we are using a compound which utilizes both elements in food to make it salty.

For example, etching silicon dioxide with hydrofluoric acid.

SiO₂​ + 6HF → H₂​SiF₆​ + 2H₂​O

Figure 2: Wet Etching in Chemical Bath (made using Inkscape).

As you have noticed in the figure (1) shown, wet etching is highly selective, it did not etch the layer below. Underneath the photoresist is a different story. There is an undercut which was made by wet etching. We have low controllability over what happens after the liquid enters through the photoresist and proceeds to etch out more than required. This is fine for final processing of wafer when we start working in microscale; but in nanoscale, it is undesirable. Wet etching occurs in all directions equally, this is called an isotropic process. You have noticed that when water enters a hole, it takes its shape. The same phenomenon occurs here but instead of just staying there, the etchant eats away at the hole.

To fix the problem of undercuts, we need a way that is anisotropic in nature; that is, it can be directionally controlled, allowing the area under photoresist to remain untouched. What ways do we have that can remove atoms from a material without targeting any area underneath photoresist? Dry etching provides one solution with sputtering. This method is usually discussed as a deposition method where a target material is bombarded with ions to throw out atoms which are then deposited on the substrate. We just shift the positions of the target and substrate, so now we are bombarding the substrate with ions.

Figure 3: Sputtering of deposited material (made using Inkscape).

Figure (2) tells the opposite story of wet etching. Instead of an undercut, we now have incomplete etching. This happens because the bombarding atoms sputter the material in random directions. Some of the etched material may redeposit back after hitting the walls and lose the energy required to eject completely. This phenomenon is used in bias sputtering to make uniform thin films but, unfortunately not the correct time for this to occur.

The problem becomes, how do we etch out atoms so that we gain accuracy and no redeposition. Well, the answer is quite simple; change the etched atom/molecule. Make the material into something that won't re-deposit on the surface. Behold, reactive ion etching (RIE); combining the best parts of both wet etching (selectivity), and sputter etching (directionality), we have a method that finally gives us straight walls. Instead of Argon atoms being ionized, we ionize gases with a reactive component like Fluorine (F), or Chlorine (Cl). These ions bombard the material to be etched on the substrate and remove them. The free radicals (neutral) Fluorine and Chlorine, react with the ejected material. No re-deposition will occur because the atom/molecule is no more, its properties have been changed.

For example: etching silicon dioxide using fluorine plasma.

Dissociation of gas in plasma into ions and free radicals.

SF₆​ + e⁻ → SF₅​+ + F + e⁻

CF₄​ + e⁻ → CF₃​+ + F + e⁻

^{Reaction with ejected silicon dioxide molecule.}

SiO₂​ + 4F → SiF₄​(gas) + O_2​

Figure 4: Reactive ion etching using sulphur hexafluoride (made using Inkscape).

Reactive ion etching has made sputter etching somewhat inferior. The core plasma/vacuum infrastructure is similar, but electrode design, plasma control, and chamber engineering are significantly different because the physics of etching is different. If the correct chemistry exists then it can be etched out using RIE. New and novel materials have use of sputter etching because figuring out its chemistry will be a long process. RIE is slow, expensive due to vacuum machinery, and can damage the surface if used for a very drawn out time. Wet etching is the black to RIE's white in these aspects. For non-critical layers (above the nanoscale layers), wet etching is the best technique.

The simple task of removing material turns into such a lengthy affair when working in small scales. So many parameters to think about and so many precautions. Those protocols are in effect to protect the operator of the machinery. A person well trained in etching will know it is not a trivial task and will do their utmost in protecting the device and themself. I just wish that India's Central Public Works Department could learn a thing or two about real engineering.

Go back to main page