Tech noteTN-05

The chip

TopicGold film, film quality, handling

Reading time7 min

IN SHORT

A sharp dip starts here: the dip can only be as sharp as the gold film, about 50 nm thick, smooth and clean.
Software can’t fix the film: a rough or patchy film scatters light and blurs the dip, and no processing recovers it (TN-04: Where the precision comes from).
The same gold carries the coating: it holds the layer that captures your molecules, such as our low-background AffiCoat (TN-06: The binding surface).

Overview

The SPR signal is generated in a gold film roughly 50 nanometres thick. Everything about that film (its thickness, smoothness, purity and how it is held to the glass) shows up later as the depth and width of the resonance, the noise on your baseline and how well your surface chemistry forms.

This note explains what makes a good film, how films are made, what degrades them, and what to look for in a sensor chip.

1 · THE FILM

Why gold, and why about 50 nm

Gold supports a strong plasmon in the visible and near-infrared, does not oxidize in air or buffer, and takes thiol chemistry directly (TN-06: The binding surface). Silver gives a sharper resonance but corrodes.

Thickness matters. Around 50 nm of gold gives the deepest dip, with a narrow width. Thinner films lose too much light into the plasmon, and the dip gets shallow and broad. Thicker films give a narrower but much shallower dip, because too little light reaches the surface.

Figure 1 — layer stack and gold thickness — Fig. 1. Left: the layers of a sensor chip, drawn to scale in nanometres (the glass is not). The plasmon field reaches through the surface chemistry and the bound proteins into the sample. Right: computed resonance for 40, 50 and 70 nm of gold with water on the surface. Around 50 nm gives the deepest dip; a thinner film gives a shallow, broad dip; a thicker one a narrower but much shallower dip. Fresnel calculation, wavelength readout.

2 · FILM QUALITY

What makes a good film

Five properties decide how a film performs, and each one shows up in your data when it is off.

The resonance can only be as sharp as the film allows. No software recovers a film that scatters, or a surface that won’t take chemistry evenly. That is why the film is where a lensless instrument cannot cut corners (TN-03: Lensless optics and wavelength readout).

Property	When it’s right	When it’s wrong, what you see
Thickness	Uniform, close to target	Shallow or broad dip; differences between channels and between chips
Smoothness	Low roughness, fine continuous grains	Light scatters, the dip broadens and shifts; a broad dip is tracked less precisely, so baseline noise goes up (TN-04: Where the precision comes from)
Purity	Clean gold, nothing trapped in or on it	Offsets and drift; monolayers form patchily, so coupling yields vary and background binding increases (TN-06: The binding surface)
Continuity	No pinholes, islands or voids	Weak or uneven signal; chemistry reaches the glass through the gaps
Adhesion	Firmly bonded, with a minimal adhesion layer	Film lifts or blisters during regeneration; a thick adhesion layer damps the plasmon (section 4)

Figure 2 — smooth vs rough film — Fig. 2. The film sets the shape of the dip, and the dip sets how precisely it can be tracked. Left: a smooth, pure, continuous film next to a rough or contaminated one, with bumps, gaps and residue. Right: the same 50 nm film computed smooth, and with extra optical loss standing in for roughness and contamination. The second dip is broader and slightly shifted. Illustrative.

3 · HOW FILMS ARE MADE

Physical vapour deposition

Sensor-grade gold films are made by physical vapour deposition (PVD): gold is turned into vapour in a high vacuum and condenses on the glass, a few atoms at a time.

The method is only part of the story. What decides the final film is how well the process is run: how clean the substrate is before coating, the vacuum quality, the deposition rate, thickness monitoring, and batch-to-batch consistency. Those are the skills of a specialized coating house, not of a general lab evaporator.

Method (all PVD)	How it works	Strengths	Watch-outs
Thermal evaporation	Gold heated in a crucible until it evaporates	Simple, pure gold, gentle on the substrate	Rate and thickness control depend on the setup; grain size depends on rate and substrate temperature
Electron-beam evaporation	An electron beam melts a small spot of the gold source	High purity, precise rate and thickness control	Needs good vacuum and process control to stay smooth
Sputtering	Ions knock gold atoms off a target	Dense, well-adhered films, uniform over large areas	Higher-energy process; can trap gas or roughen the film if not tuned
Template stripping (reference)	Gold deposited on an ultra-flat template, then peeled off	Near-atomically flat surfaces	A research technique; not practical for routine sensor chips

4 · THE ADHESION LAYER

Holding the gold to the glass

Gold does not stick well to glass on its own, so a few nanometres of an adhesion metal, usually chromium or titanium, go down first. That layer has a cost.

It damps the plasmon: adhesion metals absorb light, and weaken and broaden the resonance. The thicker the layer, the bigger the effect. And it can move: chromium and titanium can diffuse through the gold, along the boundaries between grains, and oxidize at the surface. Heat speeds this up. Over time that changes the film’s optical behaviour and the surface your chemistry sees.

So the adhesion layer should be just thick enough to hold the gold, deposited in the same vacuum run, and well controlled from batch to batch.

Figure 3 — adhesion layer — Fig. 3. Left: schematic of a gold film on its adhesion layer. Over time and with heat, adhesion metal can diffuse up the grain boundaries and oxidize at the surface. Right: computed resonance with no adhesion layer and with 2 and 5 nm of adhesion metal (chromium optical constants). The thicker the layer, the broader the dip and the less light is reflected around it, so the dip stands out less.

5 · CLEANING

Cleaning, and what it does to the gold

Cleaning removes organic contamination and old monolayers, but every method also acts on the gold. Plasma cleaning (oxygen or argon) removes organic layers and thiol monolayers efficiently. Oxygen plasma also leaves a thin gold oxide, and a monolayer built on it can trap that oxygen. Aggressive or repeated treatment can roughen the film.

UV-ozone is gentler on the film, but also oxidizes the surface. Strong oxidizing acids clean thoroughly but are hard on thin films and adhesion layers, and are best avoided for routine reuse.

For most users, a fresh, clean chip from a controlled batch is more reproducible than a heavily cleaned one. If you reuse chips, use the gentlest method that works and check the baseline before trusting the surface.

6 · HANDLING

Both faces matter

The chip couples optically to the instrument through its back side, so the back side matters as much as the gold. A fingerprint or dust on either face shows up as noise and an offset baseline; a scratch on the gold shows up in every run.

Hold chips by the short edges, and store them clean and dry.

Figure 4 — chip in its holder (photo) — Fig. 4. A sensor chip in its holder, held by the edges with gloves. Both faces of the chip matter: the gold side carries your surface, and the glass side is where the light comes in.

7 · OUR CHIPS

How we source our chips

Our sensor chips are coated by a world-class coating service, to our specification, on Dove-prism substrates made for our instruments. Every chip, and every channel on it, starts from the same optical point, which the lensless design relies on (TN-03: Lensless optics and wavelength readout).

The coating report for a batch is available on request.