Oftentimes, a high resolution electron microscope and imaging of a film cross-section can be adequate to retrieve a thickness value as demonstrated for 150 nm thick cellulose films (Puspasari et al., 2015). However, only a few examples are available for cross-sectional analysis of thinner biopolymer films for mere film thickness determination due to laborious sample preparation (Weißl et al., 2019b), low contrast of most of organic matter in electron microscopy, and its tendency to decompose in an electron beam. Indeed, the most used tool for a thin film thickness measurement is a stylus profilometer that tracks the surface conformation by mechanically moving a stylus across the film substrate and monitoring the force applied (Niegelhell et al., 2016b; Niegelhell et al., 2017; Weißl et al., 2018). For recording film thickness, scratching of the film on the substrate allows creation of a contour of the substrate, and the film and the height difference between the substrate and the film surface is monitored (Siderov et al., 2013). The force and material applied to scratch the film is important, as it is crucial not to indent the substrate. Another known challenge is that when the contour is analyzed by recording the morphology, contact is required with the surface in the measurement and it can destruct a soft film and alter the recorded value of thickness.

Similar to the profilometry, atomic force microscopy (AFM) is used for thickness determination by analysis of height profiles as shown for dry and for wet cellulose films (Fält et al., 2004; Ganner et al., 2015; Reid et al., 2016). Disadvantage of the scratch method here as well is the potential damaging of the substrate and hence affecting the thickness value. Indeed, when using razor blade for scratching, a 1–2 nm penetration also to the substrate when determining thickness of cellulose films, has been found (Fält et al., 2004). To circumvent this, polymer film thickness can be determined by controlling the loading force of the AFM tip on film surfaces, scratching the film with the tip and recording the height profile (Ton-That et al., 2000). Prior to extraction of the height profile from AFM analysis, it is vital to apply corrections to the image to account for the plane tilt and scanner movements (Canale et al., 2011).

Sarfus is a setup where an optical microscope in cross-polarization mode is used in conjunction with specific substrates, so called surfs. The surfs consist of a silicon substrate with a top coating that does not polarize light on reflection and therefore leads to a signal enhancement of the film attached to that layer (Chiang and Yang, 2011). In combination with the use of a calibration standard, layer thicknesses down to 0.3 nm can be determined. The method can also be used in combination with scratching the surface, to further reduce potential experimental errors. An example where the method has been used to visualize biopolymer samples are chitosan—silane hybrid thin films (Figure 1) (Spirk et al., 2013). Layer thicknesses in the range between 8 and 100 nm were analyzed with this fast technique with high throughput.

QCM-D is a gravimetric method to determine deposited mass on a substrate. The detection is based on a change in the resonance frequency of the sensor when its mass changes by deposition of a thin film. Furthermore, the adsorption and desorption of molecules on such films can be studied (Reviakine et al., 2011). The Sauerbrey equation (1959) (Equation 1) is used for mass (Δm) determination for rigid films and requires input of change in frequency Δf, overtone number n, crystal sensor specific constant C (ng cm⁻² Hz⁻¹). Δ m = − C Δ f n (1)

QCM-D has been utilized to determine thickness of cellulose thin films by calculating the mass of the film via Eq. 1 using the changes in frequency response before and after deposition of a cellulose layer on the QCM-D sensor (Peresin et al., 2012; Tammelin et al., 2015; Palasingh et al., 2021). This mass together with an input value of (assumed) density, ρ, is then used to extract film thickness (d) according to Eq. 2: d = Δ m ρ (2)

Mohan et al. (2012) used the QCM-D thickness determination method for trimethylsilyl cellulose and cellulose films accompanied by density values that were determined by X-ray reflectivity (XRR) to 1.0 and 1.5 g cm⁻³, respectively.

The challenge of the thickness determination using the resonance frequency before and after the film deposition is that it is sensitive to contamination that may take place when the sensor is removed, coated, and inserted again into the surface sensitive device. Another advantage of the QCM-D method is that also changes of layer thickness by swelling either via gas phase (humidity) or liquid water can be tracked (Niinivaara et al., 2016).

However, soft systems with dissipation >0 are sensitive to viscoelastic properties and the Sauerbrey equation is inadequate to represent the film mass. The Voigt model is a viscoelastic model applied for analysis of viscoelastic systems in QCM-D (Reviakine et al., 2011). The model accounts for viscosity and elasticity of the film through complex shear modulus (Voinova et al., 1999), which has been used in analysis of biopolymer films, for instance, xyloglucan and gums (Eronen et al., 2011), carboxymethyl cellulose (Eronen et al., 2011; Orelma et al., 2011) and chitosan (Orelma et al., 2011) on cellulose films. Further models and guideline for choosing and interpreting the data can be found in Reviakine et al. (2011) However, analysis of soft and hydrated biopolymer films using Voigt model has been shown to deviate from values from surface plasmon resonance (SPR) (Mohan et al., 2014) that is considered to be a direct analysis to gain value for film mass. Hence, the latter is often the method of choice over QCM-D when it comes to determination of thickness.

SPR is an optical technique that relies on the interaction of light with surface plasmons that are present on metal surfaces. In a typical setup, the incident light beam’s angle is varied, and the intensity of the reflected light is recorded. At a specific angle, the intensity of the reflected light has a minimum as the plasmons resonate. Under resonance conditions, the surface plasmons react sensitive towards changes of their environment, i.e., permittivity of the surrounding media (e.g., solvent, film). This also applies when additional layers or molecules are added via adsorption measurements in situ as they trigger changes in the permittivity of the entire thin film assembly (Jung et al., 1998). However, as it is a surface sensitive phenomenon, typically films with layer thicknesses larger than one micron cannot be analyzed. The phenomenon can be explained by classic physics as shown in Eq. 3 where w is the angular frequency, c is the speed of light in vacuum, ε _o is the permittivity of refractive element (prism), θ_c is the incident angle and ε ₁ and ε ₂ are the permittivity of the ambient medium and the metal surface. ω c ε 0 s i n ( θ c ) = ω c ε 1 ε 2 ε 1 + ε 2 (3)

The main information that can be extracted from such SPR curves is the refractive index and the layer thickness, with resolution limits in the submonolayer regime (Niegelhell et al., 2016a) by employing multi-layer fitting on the basis of the Fresnel equations. These fitting procedures yield a set of continuous layer thickness-refractive index pairs, whose unique solution requires either the use of a multi wavelength setup (multi-color method) or to investigate the film in different media [two media approach (Peterlinz and Georgiadis, 1996)]. In adsorption conditions, also the kinetics of layer growth/molecules adsorption can be observed in situ using SPR, with the advantage that the layer thickness determination is not influenced by the softness of the films. If the refractive index increment dn/dc is known, the de Fejter equation Eq. 4 can be directly used to determine the amount of adsorbed materials (Γ) from the shift in the surface plasmon resonance angle (Δθ) without applying multi-layer fitting procedure (k and d_p are instrument constants) (De Feijter et al., 1978). Γ = Δ Θ × k × d p d n / d c     (4)

Via the density of the adsorbed compounds, the layer thickness can then be easily evaluated. In principle also the density can be determined at the interfaces. Sampl et al. (2019) showed that the outermost layer of cellulose thin films (3 nm) has a different refractive index and density than the bulk film. Combining SPR with QCM-D, swelling studies can be performed that allow the monitoring of water uptake inside different biopolymer thin films (Kontturi et al., 2013).

Ellipsometry is used to determine optical constants and thickness of a material (Azzam et al., 1978). It measures change in polarization of reflected or transmitted light. Upon reflection at a plane surface, linearly polarized radiation generally becomes elliptically polarized. Its polarization state can be described by two ellipsometry parameters: amplitude ratio (Ψ) and phase shift difference (Δ) of two mutually orthogonal polarized components of the reflected waves. The change in polarization is the ellipsometry measurement, commonly written as: ρ = tan ( ψ ) e i Δ (5)

The measured Ψ and Δ cannot be converted directly into the optical constants of the material but requires modeling. Description of some models for thin biological films can be found in Arwin (2000). When a suitable model is chosen, the ellipsometry data are fitted to the model. The optical models give then the complex refractive indices ñ (Eq. 5) that is expressed as the sum of n and extinction coefficient (k), and finally film thickness. n ˜ = n + i k (6)

Typically, for enabling the measurement the surfaces must be light reflecting, and the film has no (or low) light absorption or the complex refractive index of the film is known (Tengvall et al., 1998). When the refractive index of the material is known, the determination of thickness is simple (Tengvall et al., 1998; Höök et al., 2002). In some cases, when refractive index is unknown, measuring at multiple wavelengths (spectroscopic ellipsometry) can provide both refractive index and thickness. The thickness determination of thin films requires building a multilayer model that includes the substrate, e.g., silicon, silicon dioxide, biopolymer layer and air and is often performed using commercial software with embedded models and for example, using refractive indices of the known layer materials and iterative fitting [e.g., (Eriksson et al., 2007)].

Reflectance using X-rays (XXR) (Foster et al., 1990; Kontturi and Lankinen, 2010), neutrons (Thomas and Penfold, 1996) and light (Buron et al., 2006; Cranston and Gray, 2008) have been used for thin film characterization, among them thickness or change in thickness. The analysis of thin films is based on an incident beam that penetrates to the sample reaching to regions that have different refractive indices, e.g., at a film-substrate interface. The beam is reflected from the interfaces and interference produces a Fresnel reflectance pattern. The ratios of heights of the maxima in the pattern relate to film thickness and interface width and hence lead to possibility to resolve film thicknesses. (Foster et al., 1990). The periodicity of the oscillation is inversely proportional to the film thickness. Lu et al. (2020) have analyzed cellulose thin films (8–88 nm) using XRR. X-ray reflectivity can be also assessed to observe a response of cellulose thin films upon humidity variations. In such experiments, the films are placed in humidity chambers that have windows for the X-ray beams. Figure 2 shows such results of cellulose thin films whose thickness increased from 45 to 52 nm when humidity increased from 0 to 70% (Figure 2A). Figure 2B shows that XRR yields also different densities for the bulk layer and the layer that is exposed to the air interface.

An estimation of film thickness is valuable for many purposes but an utmost necessity for some analyses. Hydration and swelling of hydrophilic biopolymers direct many of their properties and already in ambient conditions they interact with humidity. Craig and Plunkett (2003) pioneered a solvent exchange method to determine water in polyelectrolyte thin films. Kittle et al. (2011) applied the method to determine amount of water (Γ_water) in cellulose thin films (Eq. 7) by exchange of water to heavy water (D₂O) and utilizing the density difference for determining the amount of water in a layer, according to Eq. 8. Γ water = − C ( Δ f n ) water (7) ( Δ f n ) water = ( Δ f n ) film − ( Δ f n ) bare ( ρ D 2 O ρ H 2 O ) − 1 (8) Where ρ D 2 O and ρ H 2 O are the densities of D₂O and H₂O, respectively. The (Δf/n) water is then be used to determine the water concentration in the film. The method has recently been used for determination of water content in polyelectrolyte multilayers (Kittle et al., 2021). The film thickness affects the amount of bound water and needs to be known to enable direct quantitative comparison. Furthermore, the film thickness has been shown to affect, e.g., adsorption of xyloglucan on cellulose films (Kittle et al., 2018). Also SPR has been used for swelling determination by utilizing the inverse proportionality of volume and optical intensity (Erdoğan et al., 2008). However, also in this case, the film thickness was determined independently using ellipsometry. The power of combined methods.

The mostly used methods for determining thin film thickness are tracking conformation (profilometry, AFM), XRR, SPR, QCM-D and ellipsometry (Figure 3). Each of the method have their benefit but also a challenge. AFM and profilometry thickness determination requires scratching of the film that may lead to an artifact of also changing the substrate. The measurement itself may also be affected if the film is deformed by the stylus or AFM tip. Thickness analysis of films with QCM-D is non-destructive, however a challenge is that often the film is formed in a spin coating or other device that requires measurement of the sensor before deposition and after and is a source for an artifact for this gravimetric method. Another challenge arises from that a density value is needed for calculating thickness and in the case only an estimate is available, decreases accuracy. SPR, ellipsometry and XRR are non-destructive methods which allow for extremely precise film thickness determination (±0.1 nm). They also provide additional information on film structure (XRR: roughness, density) and optical constants (SPR, ellipsometry: refractive indices). Their drawback is that the required modelling procedures are more laborious than, e.g., measuring a step height of a scratched surface using AFM.

The coupling of methods may give rise to additional information, particularly if, e.g., QCM-D and SPR are coupled, when exploring swollen films. This allows to determine the water content of the films. Another example was described by Gesang et al. (1995) who compared AFM and ellipsometry analysis of synthetic organic films. They pointed out that the areas investigated by ellipsometry and AFM are very different, the ellipsometry reaching about 2–3 mm² (the exact value depending on the angle and focus of incident light), and the AFM thickness determination being local with the step height assigned to smaller areas, such as in hundreds of nanometers up to microns in frame dimensions. Discrepancy was detected above film thicknesses of 10 nm between ellipsometry and AFM analysis (smaller values with AFM), likely to be due to the restriction of AFM to measure large steps in height. The study used a scratched square to reveal the substrate and identified two sources of potential error: the square created by AFM tip can be that the tip cannot penetrate the film fully during scanning square formation. Other artifact can be that the tip cannot reach the bottom of the scanning square during imaging (Gesang et al., 1995). However, the AFM was rewarded to be potentially very accurate technique that does not rely on models and hence valuable for further development of models for ellipsometry. Also Mykhaylyk et al. (2007) suggest AFM techniques valuable for thickness determination, for cross-checking an ellipsometric thickness obtained from an assumed value of the thin-film refractive index, or for determination of unknown thin-film optical parameters when combined with ellipsometry measurements. They investigated polystyrene films on silicon substrates and found good agreement between AFM and ellipsometry in the 80–130 nm thickness range. The complementarity of ellipsometry and AFM has also been pointed out in that AFM provides the topographical details that are challenging to access with ellipsometry (Siqueira et al., 1995; Kosaka et al., 2005). Cranston and Gray (2008) compared cellulose nanocrystal film thicknesses using scratch-height analysis with AFM, ellipsometry, wavelength-dependent optical reflectometry and angle-dependent optical reflectometry and recorded 191 ± 4, 184 ± 2, 195, and 174 ± 11 nm, respectively for 10 bilayers. The wavelength-dependent optical reflectometry values for film thickness were consistently larger than the other methods and was denoted to be due to the homogeneous film model that may not represent the real film configuration.

It seems that the often present structure, configuration, and roughness that lead to heterogeneous films are generally sources of uncertainty in the film thickness analysis. Some techniques are facile to identify surface roughness and thickness, e.g., AFM. While in other techniques, the film is modeled as a homogeneous structure and lead to discrepancy with the reality. The ability to swell and film porosity as well require attention when embarking in biopolymer thin film thickness determination. Error from swelling and change in free volume is challenging to account for. For example, on the case of the methods requiring a density input, this becomes a variable. Similarly, with changing volume fraction, the refractive index is modulated. Humidity is a challenge in layer thickness determination of biopolymer thin films in general. Care needs to be taken so that values are given either at a specific humidity or over a wide range of humidity levels to provide reproducible thin film data.