Analysis of the phase-dependent fluorescence decay time of treated wood using FD-FLIM

1 Introduction

Compared to other renewable raw materials, wood is particularly environmentally friendly due to the possibility of cascade usage. To exploit the full potential of cascade utilization of wood in the future, material recycling of post-consumer wood should be given preference over energetic usage [1]. But, wood is often processed on the surface before utilization. For instance, glazes, paints, oils, or plastic coatings are added for a long service life or optical reasons. Thus, post-consumer wood can contain various pollutants or contaminants, which are components that are non-wood related and can affect the technical properties of wood. A distinction can be made between mechanically and non-mechanically removable contaminants. Mechanically separable contaminants include metal parts, glass, or asphalt, which can be removed through a sieve or a magnetic separator after the wood is shredded. Binding agents, wood preservatives, acrylic varnish, or adhesives cannot be separated mechanically and are thus non-removable. Varnishes and glazes have different effects on the wood. Glazes are absorbed by the wood and are usually translucent on the surface of the wood. Varnishes build a solid layer of paint on the wood surface. Both treatments are harmful to the environment if not disposed off and further processed properly. To ensure proper disposal and excluding pollutants from the wood cycle, a post-consumer wood directive provides instructions on the recycling process of post-consumer wood in Germany since 2002. According to regulations [2], the accumulating post-consumer wood is divided into 4 categories: Category A1 being untreated post-consumer wood; A2 being wood treated neither with halogen-organic substances nor with wood preservatives; A3 being wood contaminated with halogenated organic components; A4 being pollutants contaminated post-consumer wood. Currently, only A1 wood can be used for material utilization. However, a lot of recycling companies have a mixed wood pile containing A1 to A3 post-consumer wood. Automated sorting of class A1 from the A1 to A3 pile would result in a percentage increase of A1 to be used and thus significantly increase material recycling. The volume of additionally recyclable post-consumer wood of class A1 is estimated to be 8 million tons per year [3, 4].

Currently, there is no automated sorting technology integrated into the recycling process to extract the wood of class A1 from the A1–A3 pile due to the lack of classification methods. Research shows that the identification of wood via the near-infrared (NIR) spectrum is possible in theory [5]. The disadvantage of the NIR spectroscopy is that the absorption band of water caused by variable moisture contents superimposes the NIR spectrum of wood desired for classification [4]. The method of X-ray fluorescence is also unsuitable for wood sorting, as only pollutants such as heavy metals with a high atomic number can be detected [6]. Additionally in [7] a classification of wood types by the fluorescence spectra in the visible spectral region that claims to be unaffected by thermal noise and water absorption is under research. As a result, the post-consumer wood is sorted manually by professional staff or not at all, contrary to the principle of sustainability.

Lignin in wood fluoresces when excited in the ultraviolet or visible spectrum [8]. Due to this, in the publication [9], the wood species beech, maple, spruce, pine, larch, and oak have been distinguished by the FD-FLIM (frequency domain fluorescence lifetime imaging microscopy) method, which is primarily used in the context of biochemistry or biomedicine [10]. To identify wood species, the potential impact of wood treatment on the fluorescence characteristics of wood has to be known. Thus, it was investigated whether the FD-FLIM method can also differentiate treated wood with non-mechanically removable contaminants from untreated wood. Therefore, samples were prepared using two wood types with varnish and glaze as treatment. FD-FLIM measurements were carried out, resulting in an intensity and phase-dependent fluorescence decay time image. The measured phase-dependent fluorescence decay time was evaluated to show the impact of treatment on the fluorescence characteristics of wood.

2 Theory of fluorescence

Materials that exhibit autofluorescence, including wood, respond to excitation with a characteristic fluorescence signal [11]. As further outlined by [9], to measure the fluorescence decay time of a material, a continuous and modulated excitation is needed, when measuring the fluorescence decay time using the FD-FLIM measurement method. The excitation can be done by a Fourier-transformed test function like a sinusoidal or rectangular oscillation at a defined modulation frequency ω. The fluorescence signal follows the sinusoidal/rectangular excitation with a phase shift ϕ. Additionally, the fluorescence signal is attenuated in its amplitude and shifted in its average value. In Figure 1, the amplitude of excitation B, the amplitude of the fluorescence emission b, the average value of excitation A, and the average value of the fluorescence emission a as well as the phase shift ϕ are shown schematically.

Figure 1:

Schematic representation of the phase shift and the amplitude attenuation.

Using the defined modulation frequency ω and the phase shift ϕ, the phase-dependent fluorescence decay time τ _ϕ can be calculated according to Eq. (1) [11]:

The usual approach to calculate the intensity-averaged fluorescence decay time τ _f is shown in Eq. (2), as provided by [11], considering the fractional intensity f with ∑f _i = 1 [11, 12].

3 Experimental setup

3.1 FD-FLIM

To analyse the fluorescence decay time characteristics of treated and nontreated wood, an experimental setup similar to [9] is used, containing a PCO FD-FLIM camera and an OMICRON laser assembled to a PSM 1000 microscope (see Figure 2). The selected laser is controlled by the camera and provides a sinusoidally/rectangularly intensity modulated excitation signal. As modulated excitation source, three lasers having wavelengths of 405 nm, 445 nm or 488 nm can be used. The light signal is directed to the wood sample via the beam input of the microscope. The time-shifted fluorescence response signal is captured by a 10× lens from the manufacturer MOTIC and subsequently detected by the 1004 × 1008 photodiodes of the FD-FLIM camera, which is attached to the top of the microscope. Additionally, two optical filters are used, an optical band-pass filter to narrow the bandwidth of the excitation light and an optical long-pass filter to block reflections and stray light in the emission path. The measurement setup is calibrated over a reference slide with a defined decay time of 3.75 ns. A reference measurement on the slide quantifies the uncertainties of the system with ±0.01 ns. The measured data is transferred from the camera to the computer via a USB interface. NIKON’s NIS-Elements software is used for image acquisition and to process the data and calculate the fluorescence intensity and the phase-dependent fluorescence decay time images of the wood samples.

Figure 2:

Prepared wood samples are divided into three sections, whereby section (1) contains the painted wood using acrylic clear coat, (2) contains the lacquered wood using wood preservative glaze, and (0) contains the untreated wood (left), acrylic clear coat and the wood preservative glaze in a petri dish (right).

4 Data evaluation

Each individual wood and coat surface is measured 16 times on every sector using the FD-FLIM method. For fluorescence decay times the uncertainties have a Gaussian distribution [12, 13]. Therefore a Gaussian analysis is performed to evaluate the phase-dependent fluorescence decay time after combining the 16 FD-FLIM measurements. The measurement data of the phase-dependent fluorescence decay time is imported into the MATLAB workspace in the form of TIF (Tagged Image Information) files. A histogram of the phase-dependent fluorescence decay time image is created and the measured data is interpolated and smoothed. Afterwards, the maximum and the corresponding fluorescence decay time, which describes the expectation value of the normal distribution, are determined. Also, the corresponding standard deviation assuming a gaussian normal distribution and the coefficient of determination R² is calculated. The gaussian curve is then plotted to graphically evaluate the measured data. The generated graphs are stored in PNG format and the values of the maxima, the standard deviations, and the coefficient of determination are saved in a CSV file format.

To analyse the impact of the coating and the excitation wavelength on the phase-dependent fluorescence decay time, the percentage shift δ of the fluorescence decay time from coated wood τ _cw concerning the initial fluorescence decay time of the uncoated wood τ _w is calculated. A percentage shift δ of 100 % is equivalent to the fluorescence decay time of plain coating τ _c .

Furthermore, the previously mentioned phase-dependent intensity-averaged fluorescence decay time is determined to compare the measured decay time with the theoretical values of coated wood. The fractional intensities f _c , f _w are calculated as the mean value of the normalized intensity images of the FD-FLIM measurements individually for the wood I _w and the coating I _c divided by the sum of the normalized intensity I _w + I _c , with f _c + f _w resulting as 1.

5 Results

The results of the data evaluation is visualized in histograms. The frequency of the phase-dependent fluorescence decay time is normalized to the respective maximum. As an example the graphical results from the phase-dependent fluorescence decay time image of larch are shown in Figures 3 and 4:

Figure 3:

Gaussian curve of the phase decay time of non-treated larch (blue), with ACC-treated larch (red) and plain ACC (green) at an excitation of 405 nm.

Figure 4:

Gaussian curve of the phase decay time of non-treated larch (blue), with WPG-treated larch (pink) and plain WPG (green) at an excitation of 405 nm.

Table 1 shows the phase-dependent fluorescence decay time and standard deviations for the samples. If the resulting values of the phase-dependent fluorescence decay time are compared in Table 1, it can be derived that the phase-dependent fluorescence decay time of the treatments ACC and WPG are longer compared to larch and oak. The values of the phase-dependent fluorescence decay time of the treated wood show a convoluted fluorescence decay time of wood and treatment respectively. Hereby, the fluorescence decay time of wood is shifted in the direction of the fluorescence decay time of the treatment. This shift in fluorescence decay time can also be obtained from Figures 3 and 4. Additionally, if the surfaces of oak and larch are treated using WPG, a longer fluorescence decay time is obtained in comparison to the ACC-treated woods. Thus, a differentiation of the surfaces can be made using their fluorescence decay time for a constant excitation wavelength. As expected, the absolute fluorescence decay time increases at higher excitation wavelengths using the described experimental setup.

The determined δ-value, the fractional intensity of the wood, and the theoretically calculated intensity-averaged fluorescence decay time of the treated wood samples are entered in Table 2.

The percentage shift δ of a treated oak is up to 71 % and thus higher compared to equally treated larch (55 %). Furthermore, if the wood is treated using WPG, the percent shift is higher compared to the ACC treatment. But the WPG also shows a higher variance if the percentage shifts of the three excitation wavelengths are compared. Hence, it can be concluded that the ACC coating has a smaller contribution than wood to the mixed fluorescence decay time of ACC with wood. WPG-treated wood behaves in a contrary way since hereby it is evident from the calculated expectation values that WPG has a greater impact than wood on the mixed fluorescence decay time. A possible reason is the deeper absorption of the WPG in wood compared to the ACC. A deeper absorption causes a deeper fluorescence interaction volume, whereas the ACC remains on the surface and is less detrimental to the wood structure. Overall, the percentage shift decreases when the sample is excited at a higher excitation wavelength as a result of the correlation between the excitation-dependent fluorescence behaviour of the wood lignin and the coating additives. Thus, the phase-dependent fluorescence decay time measured using the current experimental setup and an excitation wavelength of 405 nm is more suitable to differentiate between untreated and treated wood surfaces and also between different treatments of the wooden surface.

Table 2 also displays the theoretically determined intensity-averaged fluorescence decay times τ _f after Eq. (2). The theoretically calculated values of τ_f are consistently higher at all excitation wavelengths than the measured fluorescence decay times except for the decay time for oak and larch treated with ACC at 488 nm excitation. These two values with the ACC treatment are identical to the untreated woods within one standard deviation. At an excitation wavelength of 405 nm, the fluorescence decay times are 0.09–0.23 ns shorter than the calculated expectation values. At 445 nm excitation, the disparity increases in the range of 0.37 ns–0.56 ns. At an excitation wavelength of 488 nm, the theoretical decay time is shortened by up to 0.21 ns for ACC treatment or lengthened by 0.71 ns for the WPG treatment. Because of the significant deviation from the measured values, Eq. (2) does not apply to the discussed problem here. The coating and glaze do not homogenously mix with wood but change the molecular structure of wood. Thus, the treatment has an excessive impact on the measured mixed fluorescence decay time which allows differentiation of the treated from the untreated wood.

6 Conclusions

The analysis shows that the WPG has a greater effect on the mixed fluorescence decay time from wood and treatment than ACC. Additionally, it can be concluded that the percentage shift of the decay time due to the coating or glaze is stronger at a shorter excitation wavelength. Hence, to differentiate treated and nontreated wood and to distinguish the treatment method, the measurement of the phase-dependent fluorescence decay times at a low excitation wavelength is preferable. In addition, the theoretical value of the intensity-averaged fluorescence decay time was calculated, which differs significantly from the determined values of the phase-dependent fluorescence decay time in particular at longer excitation wavelengths. The high difference between the theoretical and measured fluorescence decay times of the mixture indicates that the fractional intensity of the individual fluorescence signals is only one contribution to the total mixed fluorescence decay time. Molecular effects as well as the reabsorption of the fluorescent signals of the woods by the ACC on the surface have to be taken into account. To build a model, the investigations have to be continued using other wood species as well as wood that is lacquered with different coatings and defined thicknesses.