Overview

Metal Adhesion

Cleanliness

The most common reason for metal adhesion issues is poor cleanliness, particularly from grease/non-polar compounds on the sample surface. These require a non-polar solvent (Toluene) to be removed.

Metal adhesive layer

Most metals adhere poorly to silicon.  In general for metal deposition on silicon substrates, an adhesion layer of 3-5 nm of titanium or chromium should be used. Anecdotal reports suggest that Cr may potentially offer better adhesion to oxides with Ti offering better adhesion to bare (native oxide removed) silicon. However, a thermodynamic look at silicide and oxide formation suggests that Ti is superior in both cases (Pretorius 1978), while a direct measurement of peel strength shows similar results between the two (Buchwalter 1995). Thus adhesion on silicon (or SiC) may be maximized with a BOE dip to remove the native oxide layer, with Ti used as an adhesion layer.

Use of these materials is to provide an interface between two bulk materials (i.e. silicon and the metal to be deposited). Adhesion is promoted via silicide or oxide formation at the interface. Therefore, more is NOT necessarily better. Beyond ~2 nm, a film of Ti (for example) on a polished silicon surface will be uniform and ideal for adhesion promotion (Vogt 1994). To add for some margin of nonuniformity, a film thickness of 3-5 nm may be used, potentially up to 10 nm (Mattox 1973). Beyond this point, additional thickness will begin to increase resistivity and may lead to generally poorer adhesion than a thinner film (e.g. Boh 2009). Films below 50 nm should absolutely be used, as thicker films will have high stress that significantly degrade adhesion. The exception to this rule would be the case of a porous/rougher substrate than single crystal silicon, which may require a thicker film to fill in pores. However, even glass slides only need ~10 nm of Ti to provide a continuous coating (as seen in this commercial use).  It should be noted that neither Ti or Cr is an effective diffusion barrier, and other materials will be required if a diffusion barrier is needed. Other considerations such as optical/plasmonic effects may also necessitate a modification of the adhesion layer thickness to enhance device performance (Zheng 2007, Aouaini 2009, Madsen 2017).

Resist Profile

Commonly used positive photoresists do not offer an ideal profile for lift off. To obtain an undercut profile, a sample may be soaked for ~5 minutes in chlorobenzene or toluene (Examples: Toluene via UCSB and Chlorobenzene in Collins 1982.) The wafer is then developed and metallized.

An interesting potential alternative approach for relatively larger features may be to use an optical diffuser above the photomask during exposure (Lee 2004 and Lee 2005), however this has not (to my knowledge-JCW) been attempted at BNC and the practical issues would need to be discussed with BNC staff before attempting.

Heated Solvent Liftoff

Many liftoff recipes will recommend a heated bath of NMP, typically at 80 C.

Intrinsically safe/explosion proof hotplates are required to heat solvents

Heating of solvents is intrinsically dangerous, as normal laboratory hotplates contain components that can and will act as ignition sources. For example, the manual from the Thermo Scientific Super-Nuova plates plainly state:

• “Do not use in the presence of flammable or combustible materials — fire or explosion may result. This device contains components which may ignite such materials. Not rated for use in hazardous atmospheres.”
• “Use caution when heating volatile materials; top surface and element can reach the “Flash point Temperature” of many chemicals. These plates are not explosion proof. Fire or explosion may result. Unit contains components which may ignite such materials.”
• “Do not use in highly corrosive atmospheres; corrosive fumes and spills may damage top and internal components, creating shock hazard.”

Normal hotplates may not be used for heated liftoffs, or used in any of the solvent hoods at BNC. The only safe hot plates to use in a solvent hood are known as "explosion proof" or "intrinsically safe". These plates do not contain components that may cause ignition.

As BNC does not have hoods with built in intrinsically safe hotplates, only intrinsically safe plates that have been placed in the hoods may be used for heating solvents. The only explosion proof/intrinsically safe hotplate that currently is available for purchase is the Thermo Scientific Explosion-Proof SAFE-T HP6 Hotplate. It is specifically rated for use in Class 1 (“locations in which flammable vapors and gases may be present”), Group D (“Propane, Acetone, Ammonia, Benzene, Butane, Ethanol, Gasoline, Methanol, Natural Gas”) environments (defined here) such as a fume hood with solvent vapors present. The cheapest source for purchase (As of 10/17/2018) is Laboratory-Equipment.com.

Never use any other hotplate in a solvent hood, and alert BNC staff if any other plate is being used in a solvent hood.

References