GISAXS sample requirements
Commercial silicon wafers are an ideal substrate choice, as they are cheap, smooth, flat, and have well-defined surface chemistry. However even glass microscope slides will yield good data, as well as other common substrates (e.g. ITO). Samples are typically diced to be 1 cm by 1 cm, but samples can be bigger (up to a few inches in size) or smaller (a few mm in size).
Thin films From monolayers to microns can in principle be studied. Ideal film thickness is ~50 nm to ~300 nm; but ultimately it is the scientifically relevant thickness that should be targeted. Ultrathin layers will of course have lower total scattering intensity (and thus may require longer exposure times). Very thick (e.g. micron) layers can be studied with GISAXS, although the large roughness typical of such films will make some kinds of measurements impossible. In particular, the film will lack a well-defined critical angle, which makes the usual above/below critical measurements impossible (and of course x-ray waveguiding will also not be possible).
In principle any substrate can be used. However, GISAXS works best when the substrate is very smooth (on a nanometer and micron lengthscale), and also very flat (on a micron and macroscopic scale).
If the substrate is rough, substantial diffuse scattering will be observed (possibly overwhelming the desired signal); and the film itself is likely to be rougher (making critical angle measurements difficult; see above).
If the substrate is not flat:
- Sample alignment will be difficult, as there are in fact multiple sample planes.
- Relatedly, the incident angle may be ill-defined; measurements related to the critical angle will be similarly skewed (as there are multiple planes satisfying the critical angle).
- The measurements may be impacted, blurring/distorting the image.
- The specular x-ray beam itself may be focused or defocused (depending on the sign of the substrate curvature). In reflectivity measurements, this strongly skews the data (e.g. it can give rise to apparently greater than 100% reflection). In GISAXS, this effect may similarly distort intensities and peak widths.
Note that nominally flat substrates (e.g. silicon wafers) can become bent due to processing conditions. E.g. the stress of a spin-coater can kink a wafer.
In principle, any substrate material can be used. There are advantages to having a substrate material that has a larger critical angle than the film being studied:
- In reflectivity, the region between the two critical angles will generate pseudo-waveguide modes, and provides a sensitive measure of film properties (e.g. absorption).
- In GISAXS, one can similarly take advantage of waveguide-like modes and intensity enhancements due to the strong reflection from the substrate.
Substrate thickness does not have any impact on GISAXS measurements. Thicker substrates can be advantageous in terms of maintaining rigorous substrate flatness (especially relevant for reflectivity); but even regular (0.5 mm thick) Si wafers are typically fine.
The substrate may have coatings without impacting measurements. Of course, any roughness introduced by intervening coatings will yield diffuse scattering. Moreover, the substrate layers will have different critical angles, which must be taken into account in interpreting data. But overall, substrate coatings are not problematic.
Generally, the thin film of interest is the outermost layer for GISAXS measurements. It is, however, possible to probe buried layers. Note, however, that one must consider the critical angles, and absorption lengths of layers placed on top of the film of interest. If the top layers are too thick/absorptive, then no signal will be measured. And, of course, to probe a buried layer one must be above the critical angle of all the superposed layers.
Despite these guidelines, in reality a wide variety of substrates are suitable. In practice, commercial single-crystal silicon wafers are an ideal substrate choice, as they are cheap, smooth, flat, and have well-defined surface chemistry. However even glass microscope slides will yield reasonable data, as well as other common substrates (e.g. ITO).
A GISAXS beam is typically ~100 μm wide by ~50 μm tall. Because of the shallow grazing-incidence angle, the small beam height is nevertheless projected into a large stripe; usually 1-12 mm long. As such, a stretched rectangle (microns wide by mm long) of the sample surface will be probed. In principle, samples as small as 0.5 mm × 0.5 mm can be measured (with a corresponding decrease in total scattering). On the other hand, samples as large as many inches can typically be accommodated. The ideal sample size is ~10 mm × ~10 mm. This size captures most of the x-ray beam, and makes alignment relatively simple and robust.
Note that the beam projection mentioned above means that a long stripe of the sample is being measured, including the edge of the substrate. Some film preparation methods may yield a different structure on the edge versus the center of the sample. For instance, spin-coating often generates a 'lip' of thicker material at the edge. In GISAXS, this thick region may in fact dominated the observed scattering signal (e.g. giving rise to a seemingly isotropic signal, even though the thin film region was anisotropic/aligned).
These effects can be mitigated by cleaving substrates to avoid edge effects, or by removing material near the edge (dabbing with a solvent-soaked swap, or simply using a razor blade). Edge effects are especially important in GTSAXS, which specifically focuses on measuring the near-edge material.
Lastly, but most importantly, the sample must have some structure to be probed. A smooth and homogeneous film without any structure will not yield any scattering signal (other than an oscillation of specular intensity arising from the film thickness). A disordered film without any well-defined structure will yield diffuse scattering, but nothing else. GISAXS is thus targeted at films that have well-defined nanostructure. (GIWAXS is targeted at films with well-defined molecular-scale structure.)