1. Introduction

The ability to measure ultrathin films in different environmental conditions as they form, allows detailed studies to be conducted over a vast range of systems of biological and technological importance. Existing examples of ultrathin films include microelectronics, protective coatings, biocompatible coatings, pharmaceuticals and optical/magnetic devices [1–6].

Instruments that can detect real time changes in ultrathin films have been extensively developed as biosensors, for the detection of minute quantities of biomolecules. Surface plasmon resonance (SPR), optical waveguide lightmode spectroscopy (OWLS), quartz crystal microbalance with dissipation (QCM-D) and dual polarization interferometry (DPI) all provide label free dynamic responses to changes in ultrathin films. Real time responses allow events to be directly attributed to a physiochemical process, and the rates of interactions or reactions to be determined.

Measurements of mass and kinetics of molecular binding events have been one of the main purposes for biosensors. SPR, a widely used biosensor detects the change in a mass-related parameter (resonance angle shift) and QCM-D measures the wet mass (including trapped solvent) and viscoelastic properties [7,8]. The need exists to bridge the gap between the real time, bench-top accessibility associated with biosensors and the highly detailed, relatively static views provided by neutron and x-ray reflectometry.

Through the acquisition of layer thickness and refractive index (RI), structural information is gained that provides a link between a molecule’s structure and its function. For molecules at a surface, DPI provides information on their molecular dimensions (layer thickness), packing density (proportional to the RI of the adsorbed species), surface loading and stoichiometry (mass) [7].

The dual polarization interferometer supports light of two single mode polarizations, the transverse magnetic (TM₀) and transverse electric (TE₀) modes. The two polarizations allow the unambiguous determination of both layer thickness and refractive index (RI) of isotropic films. For measurements of biological films that are naturally heterogeneous the assumption of a uniform film was investigated by Mann et al. It was found that a uniform thin film was an excellent approximation for particle dimensions up to 50nm [9,10].

DPI is a solid state device and has no mechanical scanning to reduce its sensitivity or time resolution, while OWLS, another evanescent (decaying) wave technique, requires the movement of a goniometer. This reduces the sensitivity of OWLS and hinders detailed kinetic measurements. With low noise instrumentation, evanescent wave techniques, such as DPI, have the capability to resolve layer changes at the sub-angstrom level and RI increments of 10⁻⁷ [7].

The multiple path length dual polarization interferometer (MPL-DPI) builds on the design of the dual polarization interferometer (DPI). DPI has established itself as a method to study surfactants, biological films and polymer films [11, 12]. However, DPI can only calculate layers that are formed while a measurement is in progress and is dependent on uninterrupted acquisition. The MPL-DPI has a new ability that enables the characterization of surfaces coated outside the instrument and the interruption of in situ experiments for further ex-situ modifications. The ability to coat ultrathin films ex-situ increases the number of condensed matter systems that can be studied, since it allows much more flexible approaches for sample preparation.

This work presents the principles behind MPL-DPI and demonstrates some of its capabilities. Ex-situ coated polymer films were measured by MPL-DPI and verified using complementary techniques: spectral ellipsometry, atomic force microscopy and profilometry. A real time study of an ex-situ coated polymer film is demonstrated and the ability to measure a protein film adsorbing to the ex-situ coated polymer film is shown.

2. Dual slab waveguide interferometers

Both the new MPL-DPI and the previous DPI technique use stacks of five silicon oxy-nitride thin films, of alternating low and high refractive indices on a silicon wafer designed to provide two closely spaced single mode slab optical waveguides [Figs. 1(a) - 1(b)]. The fifth uppermost film forms the last cladding layer that is etched to expose the sensing waveguide.

 

Fig. 1 (a) Cross section of the DPI in x-z plane. (b) Cross section of the MPL-DPI in x-z plane. (c) Experimental arrangement for the DPI. (d) In situ experimental arrangement for the MPL-DPI.

Download Full Size | PDF

The waveguide modes are very sensitive to the interfacial conditions [13]. Adsorption and desorption of molecules on the sensing waveguide surface, or other RI variations in the waveguide cover medium change the detectable phases in light propagating along the waveguide that are recorded by a CMOS camera.

To monitor the phases, polarized light is uniformly illuminated over the end face of the device which excites all the modes equally [Figs. 1(c) - 1(d)]. The light in the waveguides can propagate only as distinct normal modes. This only permits a transverse magnetic (TM₀) and a transverse electric (TE₀) single mode to propagate in the waveguides. As the light exits, the mode emanating from the reference waveguide and the mode from the sensing waveguide diffract into the far-field where they form Young’s fringes. A numerical Fourier transform of the Young’s fringes is used to acquire the phases. Whether the interferometer is a DPI or a MPL-DPI the basic arrangement shown in Figs. 1(c) and 1(d) to take a measurement is the same, but the data analysis of the Young’s fringes is different.

The DPI is an in situ only device. Thus the phases for both TM and TE polarizations on one arm must be tracked from the initial conditions of the interferometer. If the measurement is disrupted and the phases are not tracked, the deposited layer thickness and RI cannot be determined. Additional layers can cause phase changes of a few 2π radians in each polarization. The phase that is measured completes a cycle every 2π radians and unless this is accumulated, the number of 2π cycles is lost, leaving only the initial and final phase positions known. A mechanism to determine the number of 2π radians while the phase is unmonitored would thus overcome the in situ only nature of the DPI.

The final clad layer of the MPL-DPI is etched to create two arms, each with a different length. A section between the two arms initially equals the length of the longest arm and decreases until it has a length equal to the shorter arm [Fig. 2(a) ]. The geometry of the final clad layer causes an interference pattern that can be used to determine the number of 2π cycles for each polarization when the measurement has been disrupted. To fit an ex-situ coated layer, the initial and final phase position along each arm and the initial and final ‘phase difference between each arm’ in both polarizations must be determined. The phase measured relates to the vertical position of the interference pattern on the camera and the phase difference between each arm can be found from the analysis of the interference pattern.

 

Fig. 2 (a) Top view of the MPL-DPI in the y-z plane, L1 = 10000μm, L2 = 7500μm, W1 = W2 = 600μm, W3 = 1000μm. (b) Real CMOS image of the Young’s Interference patterns. A phase difference between armsΦ of $5.8\times 2\pi$radians can be seen. (c) Illustrates the regions of the MPL-DPI that cause the different interference patterns on the CMOS camera.

Download Full Size | PDF

2.1 Analysis of the interference pattern

The Young’s interference pattern created on the CMOS camera shown in Figs. 2(b) and 2(c) can be separated into regions, each performing a role. The middle section of the interference pattern is caused by the cladded section on the MPL-DPI that connects the two path lengths of arms 1 and 2. This connecting section on the MPL-DPI causes a sloping Young’s fringe pattern that allows the phase difference (Φ), between arms one and two to be determined for each supported mode. The phase on each arm (ϕ) is acquired from the fringe patterns either side of the middle section. The two outer sections are reference fringes used to check alignment and anomalous effects not caused by the coating of the sensing waveguide.

2.2 Calculation of an average layer

DPI calculates the average RI and layer thickness on a single arm. The total phase change $\Delta \varphi$ in TM and TE ($\Delta {\varphi }_{TM}$,$\Delta {\varphi }_{TE}$) polarizations are used to determine the average properties. This is done by solving Maxwell’s equations for a uniform multiple stack dielectric system [14]. There will be one unique solution that satisfies both the TM and TE polarizations simultaneously. The total phase change $\Delta \varphi$ in TM and TE polarizations are given by Eq. (1) and Eq. (2):

For an ex-situ modified surface the initial phases (${\varphi }_i$) and final phases (${\varphi }_f$) for each arm remain known, but the number of 2π cycles p and q are not. To determine the number of 2π cycles an initial assumption of a layer with the same properties on each arm is made. To find the average layer properties for an ex-situ modified layer the phase change between the two arms$\Delta \Phi$is used in both polarizations, given by Eq. (3).

The temporary assumption of a uniform layer reduces the number of unknown variables found within Eq. (3). By setting$x_1=x_2=x$ and $n_1=n_2=n$ Eq. (3) becomes an equation with two unknown variables x and n. Solving the two unknowns x and n using the two known changes in phase difference between arms $\Delta {\Phi }_{TE}$and$\Delta {\Phi }_{TM}$ gives a unique solution. The solution is a prediction for the layer thickness, refractive index (RI), and a prediction of the number of complete phase cycles p and q for both arms. To find the average layer properties on each arm, the predicted number of complete phase cycles is used and Eq. (1) and Eq. (2) are solved.

A further check can be made of the layers calculated. The average layer fitted on each arm should cause the same change in$\Delta \Phi$, as the layer fitted that assumed the same properties on each arm. If this is not true then at least one of the two average layers obtained is incorrect. The condition that must be satisfied is given by Eq. (4).

Differences in the average layer properties between arms result in a small over- or under-estimation in the predictions of p and q. The true value can be obtained with ease if the difference between the actual phase and the predicted phase is smaller than π for each polarization, given by Eq. (5).

The condition for satisfying Eq. (5) is dependent on the uniformity of the layer. The uniformity depends not only on the homogeneity of the film thickness but also the uniformity of the RI. When a layer is coated with the same average properties on both arms, the predicted phase change equals the actual phase change.

For a less uniform film, the ability to satisfy Eq. (5) improves as the path length of the sensing area is reduced. Sensitivity is governed in part by the path length that the light traverses. Reducing the length, would require an improvement in another aspect of the instrument to maintain the same level of sensitivity. To retain the sensitivity the refractive index of the two waveguides within the MPL-DPI could be increased or phase measurements could be made to a higher precision.

2.3 Description of calibration

To obtain accurate measurements for layers deposited upon the sensing waveguide, it is necessary to calibrate the interferometer. The purpose of calibration is to determine the RI and the thickness of the sections of the interferometer that would otherwise generate errors.

Only the exposed sensing waveguide of a DPI requires calibration. To calibrate the sensing waveguide two calibration solutions are flowed sequentially over the exposed sensing waveguide surface. The calibration solutions that pass over the sensing waveguide act as layers of infinite thickness with known refractive indices. The transition from one RI of a solution to another causes a phase change in both polarizations. The phase changes are compared with the expected phase responses and any difference is accounted for by adjustment of the thickness and RI of the exposed sensing waveguide layer.

MPL-DPI can be calibrated in a similar manner to DPI using two solutions of known refractive indices. This method of calibration allows the properties of the exposed sensing waveguide, along each arm to be calibrated. However, unlike DPI the MPL-DPI also requires the fully clad section of the sensing waveguide to be calibrated.

To calibrate the fully clad section of the sensing waveguide a single calibration solution is passed over the exposed sensing waveguide. The phase difference between the two arms in both polarizations are measured and compared with the expected phase difference between the arms. Any difference is accounted for by the adjustment of the thickness and RI of the cladded region of the sensing waveguide layer.

A simpler calibration technique is possible at a decreased accuracy. This calibration is less involved and can be done quickly, relying on an isotropic sensing waveguide and an accurately known RI of the cladding that surrounds it. The simple calibration uses only one known RI hence it can be done in air. This is useful when accuracy is not pivotal but quick measurements of a range of samples are required.

3. Experimental details

3.1 Preparation of the surfaces

Silica <111> wafers (Compact Technology Ltd, UK), were diced into 22.0x6.0mm slides, (to approximate the dimensions of the MPL-DPI) for spectral ellipsometry(SE) and AFM studies. The MPL-DPI and silica wafers were carefully cleaned using acid peroxide solution (30 ml of 99% H₂SO₄ in 10 ml of 30% H₂O₂) for 90 seconds at 120°C, followed by 5.0% (v/v) Decon 90 (Decon Laboratories Ltd, East Sussex). The samples were then thoroughly rinsed with UHQ water to remove any remaining detergent, and dried, giving good quality hydrophilic surfaces. The oxide layer thickness on the silica wafers was 12Å as determined by SE.

To assess the accuracy of the MPL-DPI measurements poly(methylmethacrylate) (PMMA) was chosen. PMMA has been used for many ellipsometry studies, it is a common e-beam resist, offers high-resolution, high contrast and it is very stable [15]. These qualities made PMMA an excellent model polymer to pattern, probe and measure using different techniques.

For each MPL-DPI measurement, a complementary spectral ellipsometry (SE) measurement was made of PMMA on a silicon wafer. The silicon wafers were cut to match the shape of the MPL-DPI, to minimize the variation between films.

Although a direct SE measurement of PMMA coated on the MPL-DPI would have been ideal, the ellipsometry models could not fit the data adequately. The poor fit was due to the very thick layers of the amorphous silicon oxy-nitride (SiON) that formed the MPL-DPI.

Comparisons between different substrates (MPL-DPI and Silicon) were deemed acceptable because silicon oxide surfaces have similar chemical properties to silicon oxy-nitride and PMMA coatings were reproducible. Coatings made within the same batch had thickness variations of ~6%.

PMMA layers were produced by spin coating solutions of PMMA over the cleaned surfaces at a speed of 8000 rpm, followed by annealing at 170°C for 6 hours in a vacuum oven. Solutions of PMMA of molecular weight 950,000g.mol⁻¹ in anisol (MicroChem) at a concentration of 2% and PMMA of molecular weight 96,700g.mol⁻¹ (Fluka, 370037) in toluene at a concentration of 0.4% were used. All PMMA layers used in each of the different measurement techniques were coated within the same batch under identical conditions.

3.2 Materials

Only water with a minimum resistivity of 18 MΩ.cm from a ultrapure (UHQ) grade water system was used for the experiments (Purelab UHQ, Vivendi Water Systems Ltd.). Human serum albumin (HSA) was free of fatty acid and was used as supplied (Sigma Lot No A-3782). The molecular weight of HSA is 66 478Da calculated from the known amino acid sequence, and its isoelectric point is 4.7−4.8. The pHs of the buffer and protein solutions were fixed at pH 5 using phosphate buffer, at an ionic strength of 2mM. The HSA solution used in this work had a concentration of 0.5mg/ml.

3.3 Multiple path length dual polarization interferometer (MPL-DPI)

The MPL-DPI measurements were performed using an Analight® Bio200 (Farfield Group Ltd., Crewe, UK). The instrument uses a helium-neon laser (λ = 632.8nm) as the source of polarized light, contains a CMOS camera and houses a fluidic and temperature control system necessary to regulate the environment above the sensing surface. When the MPL-DPI chip was inserted into the device with a Kalrez gasket it formed a single channel that allowed solutions to be passed over the exposed sensing waveguide. The Analight® Bio200 fluidic system was cleaned using Decon 90, ethanol, water and dried using nitrogen prior to every experiment.

Before the PMMA layer was coated the MPL-DPI chip went through a calibration procedure. To calibrate the MPL-DPI the phases on each arm and the phase difference between the arms are tracked using the Analight® DAQ data acquisition software, as 80% ethanol solution was replaced by 100% water. After calibration, the MPL-DPI chip was removed and dried using an infra red lamp whilst the Analight® Bio200 was nitrogen dried. The MPL-DPI was then coated with PMMA as described previously.

To measure the PMMA layer in air the MPL-DPI chip was reinserted into the Analight® Bio200. The phases on each arm(ϕ) and the phase difference between arms(Φ) were recorded to calculate the initial layer properties. Any subsequent changes in the layer were recorded by monitoring the changes in phase of each arm independently.

The Analight® data acquisition software cannot always track the sudden RI change between air and water. When a measurement of PMMA in water was made the phases on each arm(ϕ) and the phase difference between each arm (Φ) were re-measured. The new measurement allowed a recalculation of the PMMA film. Any subsequent changes in the layer were then recorded by monitoring the changes in phase of each arm independently.

For the protein adsorption experiments, once the measurement in water was completed, the water was replaced with phosphate buffer at pH5. The experiment was started by the injection of a protein solution. The solution was then incubated on the chip surface for 10 min, followed by rinsing with the phosphate buffer.

Analysis software was written based on the principles set out above and was used to calculate the layers and their evolution with time.

3.4 Spectroscopic ellipsometry (SE)

A Jobin-Yvon UVISEL spectroscopic ellipsometer (SE) was used to characterize PMMA films. The SE measurements were performed over a wavelength range from 300 to 600 nm. The thickness (x) and refractive index (n) were calculated using the DeltaPsi II software, developed by Jobin-Yvon. The model for PMMA used a Cauchy dispersion relationship $n\left(\lambda \right)=A+B/{\lambda }^2+C/{\lambda }^4$with coefficients of A = 1.488, B = 2.898x10⁻³ and C = 1.579 x10⁻⁴. In circumstances where it was possible to calculate RI, the Cauchy coefficients A, B and the thickness were fitted.

For each SE measurement, the thickness of the silicon oxide layer was determined prior to being coated with PMMA. Once coated, the PMMA layer was measured.

3.5 Electron-beam lithography

Troughs to allow AFM and profilometry measurements of the PMMA layer thickness were made by high-resolution electron-beam lithography. After coating and annealing the PMMA as described, the PMMA film was patterned by direct e-beam lithography at 10 keV using a LEO 1530 Gemini field-emission gun (FEG) scanning electron microscopy (SEM) and Raith Elphy Plus lithography system with a laser interferometer stage. The PMMA was developed in 1:3 MIBK/IPA for 30 seconds and rinsed with IPA for 30 seconds. Well defined troughs of dimensions 72μmx72μm were patterned from the air/PMMA interface to the PMMA/silicon interface were patterned at different locations across the wafer.

3.6 Atomic force microscope (AFM)

The topography and thickness of PMMA films were investigated using AFM (Dimension 3100 AFM with a Nanoscope® IIIa controller, Veeco). Operated in tapping mode, standard non-contact “Golden” Silicon cantilevers (NSG 10) were used. The cantilever tip height was 10-20μm with typical tip curvature radius below 10nm and a cone angle less than 22°. The height of surface features were measured in the fast scan direction (at a typical rate of 0.2-0.9 Hz), averaging over 5μm in the slow scan direction parallel to the step being measured. The step height was calculated using the Extract Profile function in the Gwyddion 2.13 analysis software.

3.7 Profilometer

A profilometer (Dektak 150, Veeco) measured the PMMA film thickness. A stylus with a tip radius of 12.5μm was used. A 10.0mg force was applied to the samples at a scan rate of 1.6μm/s.

To calculate the film thickness, 10μm positions before and after the step were sampled. The changes in step height were averaged and the error calculated.

4. Results and discussion

4.1 Comparisons of single layer measurements

SE was the primary technique used to validate MPL-DPI. To confirm SE measurements, a silicon wafer coated with PMMA had areas of the film removed by direct e-beam lithography. These holes extended down to the silicon oxide surface. Seven holes were positioned in sample areas of the substrate that would contribute to a MPL-DPI measurement [Fig. 3(a) ].

 

Fig. 3 (a) Positions of removed PMMA with respect to the silicon wafer and locations of the sampled step heights. (b) AFM height data from corner of trench 4. 2D interpretation of the height and a 3D interpretation. (c) Averaged step height from one of the six sample regions within trench 4. (d) Single profilometry line scan of one of the seven trenches.

Download Full Size | PDF

The holes were measured by AFM and profilometry to determine the thickness by measuring the change in vertical displacement [Figs. 3(c) and 3(d)]. AFM gave an average of 42.3 ± 2.7nm and profilometry gave an average of 42.7 ± 2.3nm. Scans by SE over the region gave an average layer of 42.0 ± 1.5nm.

The PMMA layer on the MPL-DPI was measured as 41.7nm on arm 1 and 41.9nm on arm 2. The silicon oxy-nitride surface of the MPL-DPI had an rms roughness of 5.6nm. The roughness made an accurate determination by AFM challenging. AFM measured the layer coated on the MPL-DPI to be 42 ± 4nm. The small differences between each technique can be attributed to systematic errors in each instrument and surface roughness where each technique has a different sensitivity.

The similarity of the results found using SE, profilometry and AFM confirms the accuracy of the ellipsometry model used to compare MPL-DPI and SE measurements. The largest uncertainties in comparing SE and MPL-DPI come from the 6% variation in the ability to reproduce the PMMA films on different substrates

A small error in MPL-DPI measurements comes from the chips removal from the instrument. Alignment is lost and accidental coating of the end faces (where light couples and exits the MPL-DPI) is possible. With the use of reference arms built within the MPL-DPI these errors can be corrected. In this study a correction was not attempted, but the sizes of the errors were calculated. For a 52nm layer the largest error seen was 0.046nm and 0.0006 for the RI. For an 8nm layer the largest error seen was 0.025nm and 0.003 in the RI. In terms of percentage error, the thinner layer was more affected by the ex-situ modification.

For a coated PMMA layer roughly 50nm thick MPL-DPI measured the layer properties as x = 51.54 ± 0.05nm, RI = 1.508 ± 0.001 and x = 51.39 ± 0.03nm, RI = 1.505 ± 0.001 for arms 1 and 2 respectively. The complementary SE measurement fitted a layer with a thickness of 49.4 ± 1.7nm and a RI of 1.499 ± 0.004, (where all refractive indices were measured at a wavelength of 632.8nm).

For a coated PMMA layer roughly 8nm thick, MPL-DPI resolved both the RI and thickness where SE could not resolve these 2 parameters independently. The RI in the ellipsometry model had to be fixed, leaving thickness as the only free parameter. The MPL-DPI measured the layer properties as x = 8.38 ± 0.03nm, RI = 1.503 ± 0.005 and x = 8.57 ± 0.02nm, RI = 1.499 ± 0.004 for arms one and two respectively. The complementary SE measurement fitted a layer of 8.10 ± 0.47nm and 8.07 ± 0.47nm when the RI was fixed at either 1.499 or 1.503.

SE and MPL-DPI were thus in close agreement for all the films measured. The differences between measurements were within the uncertainty introduced by the use of two different substrates.

4.2 Real time measurements

After the properties of the ex-situ PMMA layer were calculated for one point in time, the evolution of the PMMA film or the growth of additional layers could be detected. PMMA was shown to swell slowly when immersed in water, as shown in Fig. 4(a) . The thickness increased, its density (RI) decreased while its mass per unit area was constant. Any technique that measured only the mass of the polymer could not have detected the structural changes of the PMMA film.

 

Fig. 4 Real time measurements. (a) PMMA swelling in water at 20°C. (b) HSA at pH5 adsorbing to the PMMA film shown in (a).

Download Full Size | PDF

For measurements of protein films that are naturally heterogeneous the assumption of a uniform film was used. As indicated previously, a uniform film is an excellent approximation for particles dimensions up to 50nm thick [10].

To demonstrate the MPL-DPIs sensitivity to additional layers, a measurement of HSA absorbing to a fixed 52.8nm PMMA layer was made. The high affinity adsorption of HSA shown in Fig. 4(b) was carried out at pH 5 at a protein concentration of 0.5mg/ml. The measured RI (density) of the layer had an initial spike, as the proteins arranged themselves on the surface. After 300 seconds of adsorption, a HSA layer with a thickness of 35Å and a RI of 1.445 was achieved. Mass per unit area was calculated, through the use of the de feijter formula [16] and the known parameters (x and RI), to be 2.1ng/mm². A relatively large error or change in the PMMA layer would lead to minor errors in the measurement of the protein layer. In principle, if the PMMA layer were to change by 2nm, with a RI change of 0.02, the measurement of the protein layer would have a small error of 0.5Å and a 0.004 error in RI.

While we have discussed only the simplest systems in this proof of principle study of MPL-DPI, this work has provided a foundation for the study of increasingly complex surfaces and systems. We will study much more complicated ex-situ control surfaces with distinct bionanotechnological applications in the near future.

5. Conclusion

We have presented a new interferometer (MPL-DPI) that allows the thickness and refractive index of in situ and ex-situ coated ultrathin films to be measured. The technique was demonstrated using films of PMMA and adsorption of human serum albumin. Good agreement was found with AFM, spectral ellipsometry and profilometry measurements, showing the high structural sensitivity of MPL-DPI.