Wet and dry etching

Etching techniques in combination with e-beam or photolithography are widely adopted in micro- and nanofabrication [72, 73, 74]. The process of etching can be defined as the transport of material from a solid into a mobile phase. Whether the mobile phase is liquid or gaseous, the etching technique is categorized as wet or dry etching, respectively.

The main properties of an etching process are the etch rate, material selectivity and direction dependence. In the etching of planar structures, as commonly done in semi-conductor processing, the etch rate is defined as the velocity of the reduction of layer thickness. When etching a non-planar structure, such as a NW, the situation is more complex, the etching can proceed inwards simultaneously on all sides of the NW. In this case the diameter reduction can be used to define the etch rate. Furthermore, if the etch-ing proceeds non-uniformly on the sides, the cross section becomes non-circular and the diameter is ill-defined.

Focusing on wet methods, the etching process can be broken down into 3 subprocesses:

1. Transport of reactants to the solid–liquid interface 2. Chemical reaction at the interface

3. Transport of the products to the liquid phase

The etch rate is determined by the slowest subprocess. In case the first step is the bottle-neck, the etching is diffusion-limited, if the second step, then it is reaction rate-limited. In the former case the process is isotropic, in the latter one anisotropy can arise if the reac-tion goes on with different rates along different crystallographic orientareac-tions. Considering dry methods, anisotropy can originate from the angular distribution of the incident bom-barding particles. In high vacuum a collinear beam can etch deep and narrow trenches, which we take advantage of in focused ion beam (FIB) milling.

Selectivity is the dependence of the etch rate on the chemical or structural composition of the etched material. Chemical selectivity is common in all chemical reaction-based etching methods, but is also observed in dry etching methods based on bombardment by inert particles, such as Ar sputtering. Sputtering has been introduced in the previous section as a thin film deposition technique. As the film grows on the substrate, material is removed from the source, which is essentially the etching counterpart of the deposition.

In sputter-etching the efficiency of momentum transfer depends on the relative weight of the bombarding particles and the atoms of the etched material, which leads to selectivity.

In the fabrication of InAs NW devices we used the following wet and dry etching methods:

ˆ Ammonium sulfide etching and passivation. InAs NWs have a native oxide layer on their surface. In order to form ohmic electrical contacts, this oxide layer must be removed. Ammonium sulfide ((NH₄)₂S) is a chemical commonly used in the oxide removal and passivation of III-V semiconductors [75]. It has been shown that the treatment in a highly dilute (∼0.1%) solution the process is practically self-terminating: only the oxide is removed, because the core InAs is etched with a much lower rate [76]. After the treatment the surface is covered with a passivating sulfide layer, which is very stable against reoxidation [77]. Using higher concentrations and a lithographic mask, the InAs NWs can be patterned. It has been observed that in this case the etching is anisotropic and leads to a double conical shape [78, 79].

ˆ Ar sputtering.As an alternative to the ammonium sulfide treatment, Ar sputtering can be used for the in-situ oxide layer removal in the vacuum chamber right before the metallic film deposition [80]. The two methods give comparable results for InAs NWs in terms of contact resistance.

ˆ Piranha etching. The piranha solution is a mixture of sulfuric acid (H₂SO₄) and hydrogen peroxide (H₂O₂), and can be diluted with water. It is an aggressive etchant and widely used for cleaning silicon wafers, because it quickly removes organic con-taminants. The piranha etching of semiconductors is a two-step process, first the hydrogen peroxide oxidizes the surface, then the oxide is dissolved by the acid. In Chapter 4 we show a technique we developed for the patterning of InAs NWs with a dilute piranha solution. The method relies on the observation that the sulfide layer protects the NW from the oxidizing agent and blocks the piranha etching.

ˆ Reactive ion etching (RIE). RIE is similar to sputter-etching, but instead of a noble gas, reactive gas is used as the atmosphere. After developing the mask in the EBL cycle, it is usual to treat the sample with O2RIE, which removes resist residues and other organic contaminants. Furthermore, to access the buried electrical gates in the bottom-gated Cooper pair splitters, we open windows in the silicon nitride layer using a CHF3/O2 RIE step [81]. The process parameters are listed in Table A.2 both for the O₂ and CHF₃/O₂ RIE processing.

ˆ Focused ion beam (FIB) milling.While the previous methods are used in com-bination with a mask, FIB milling is maskless, the pattern is directly etched. FIB milling is also based on the sputter-effect, the sample is bombarded with high energy (10-30 keV) ions focused in a beam. The beam scans the surface in the designed pattern, like in EBL. We used FIB milling with Ga ions in the fabrication of Cooper pair splitters to cut the NWs below the middle superconducting electrode between the QDs (see Figure 5.20(a)). A screenshot of the FIB milling software and a brief description of the process is shown in Figure A.7. The experimental results acquired in a CPSD processed with FIB milling are discussed in Section 5.5.

Galvanic effects in wet etching

In electrochemical etching an external current source is used to drive electrical current through a circuit consisting of the substrate to be etched, the etchant electrolyte and the counter electrode. The etch rate can be tuned with the current density.

However, when etching a semiconductor structure equipped with metallic contacts, a local galvanic element can be unintentionally formed and electrochemical effects can arise.

For example, in the processing of a GaAs field-effect transistor (FET), if the contact metal is exposed to the etchant, the etching proceeds much faster in the vicinity of the contact [73]. Surprisingly, a galvanic reaction can occur in deionized water as well, and lead to the rapid oxidation next to the contacts in GaAs FETs [82]. Here the oxygen dissolved in the water is the oxidizing agent, and an important message of the observation is to keep the rinsing time to minimum in wet etch processing.² We note that galvanic reactions can also occur in p/n junctions (without any metallic electrodes) [83, 84].

While galvanic element formation is often unfavorable, metal-assisted chemical etching (MacEtch) turns it into an advantage by artificially bestowing anisotropy to the etching process [85]. In MacEtch a noble metal, such as Au, Pt or Ag is deposited on the semicon-ductor surface which catalyzes the local oxidation. Oxidation is followed by the dissolution of the semiconductor oxide in an acid. The metal is not consumed in the etching process, and as the etching underneath the metallic layer proceeds, it descends into the semicon-ductor.

In Chapter 4 we demonstrate two novel wet etch methods, in which NW sections are thinned and electrical contacts are created in a self-aligned way. In the second method, where the contact formation precedes the etching step, we have found that galvanic effects emerge and enhance the etch rate. Although the effect is in principle localized to the vicinity of the contacts, we still rely on the lithographic mask to define the geometry.

2The dissolved oxygen can be removed by boiling the water or by bubbling N2.