Promoting Recycling Efficiency Through the Use of Sub-Terahertz Waves for Proper Wood Identification

Abstract

Past studies have reported that carbon dioxide emissions during combustion vary depending on the tree species used as fuel. It has also been reported that the moisture content of wood affects combustion efficiency. From this perspective, identifying the tree species and moisture content is crucial for utilizing waste wood as a resource. Therefore, this study verified the effectiveness of non-destructive diagnosis using terahertz waves. Samples with adjusted moisture content were prepared for eight types of wood. Each wood sample was irradiated with multiple broadband terahertz electromagnetic waves, and their transmission characteristics were compared. Experimental results revealed a strong negative correlation (Pearson Correlation coefficient: −0.98~−0.71 square meter/gram) between the sample’s specific gravity and transmittance when irradiated with 65 GHz and 90 GHz sub-terahertz waves. This trend was particularly pronounced during 90 GHz sub-terahertz irradiation. Furthermore, it was found that the trend in transmittance variation differed depending on the wood’s moisture content. These results indicate that terahertz waves are effective as a wood identification method capable of distinguishing between coniferous and broadleaf trees. Furthermore, they are considered effective for predicting wood moisture content. This research is expected to contribute to promoting wood recycling and the sustainable use of wood resources.

1. Introduction

As global warming and climate change worsen worldwide, many countries are accelerating decarbonization policies and setting carbon-neutrality goals [1]. In this context, bio-based value chains are increasingly recognized not only as carbon sinks but also as material and energy systems that can support a circular economy when biomass streams are allocated to their most appropriate use [2]. Wood valorization technologies can substantially enhance sustainable resource management, economic growth, and environmental protection by converting wood waste into valuable products [3]. Turning to Japan’s domestic forestry sector, both private and national forests suffer from inadequate conservation and management, with insufficient selective thinning and ineffective utilization of forest resources [4]. Consequently, wildfires have become frequent in recent years, yet fundamental countermeasures remain unestablished. Meanwhile, biomass power plants are being constructed nationwide as a form of renewable energy. However, since imported wood chips are used as part of the fuel, revitalizing domestic forestry and achieving self-sufficiency through effective utilization of forest resources are urgent priorities. For example, among the species native to the Mediterranean region—the Turkish stone pine (Pinus brutia Ten.), black pine (Pinus nigra L.), and black locust (Robinia pseudoacacia L.)—the black pine’s trunk and branches are known to have a high calorific value due to their low moisture and ash content and high lignin content [5].

Forest resources include materials for construction and furniture, as well as fuel. Maximizing their value as resources likely requires accurately sorting the diverse types of wood for their specific uses. One of the most critical factors in this sorting process is the difference in tree species. This is because the most effective way to utilize wood varies depending on the species. Additionally, differences in wood moisture content are also a crucial factor in sorting operations. If wood can be sorted based on moisture content differences, it becomes possible to accurately determine the processing, transportation, storage, and utilization methods for the wood. This is expected to significantly contribute to improving the recycling efficiency of forest resources. However, in practice, both harvesting and post-consumer streams are highly heterogeneous. They may contain mixed species, variable particle sizes and moisture content, and in the case of waste wood, coatings, preservatives, or heavy metal contamination [6].

Recent studies show that optical sensing can support wood identification and sorting. Near-infrared (NIR) spectroscopy has revealed that waste-wood streams are highly variable, meaning this variability must be quantified to build robust models; and results depend strongly on model choice and validation strategy under realistic sorting conditions [7,8]. Camera-based methods such as convolutional neural networks (CNNs) can identify wood species from standardized wood-core images [9], while NIR- hyperspectral imaging has been shown to discriminate visually similar high-value timbers [10]. However, these approaches face practical limitations in real recycling and biomass processing lines. They capture primarily surface information and are sensitive to surface conditions (such as dust, coatings, stains), contamination, and variations in lighting or imaging setup. Their limited penetration depth in bulk chips means performance degrades when samples are heterogeneous, mixed, or contaminated, which are precisely the conditions encountered in most waste-wood streams.

This research proposes a method using terahertz waves to accurately sort harvested timber according to species or moisture content for specific applications.

Terahertz sensing has increasingly been considered for industrial non-destructive evaluation because it can penetrate many dry, non-polar materials while remaining highly sensitive to water and dielectric-property differences, enabling both material discrimination and moisture-related measurements [11]. Wood-derived materials have been assessed as terahertz waves optical elements, supporting the feasibility of transmission-based sensing in wood and cellulose systems [12]. In wood science, terahertz waves time-domain spectroscopy has been applied directly to wood species identification [13] and nondestructive species recognition using terahertz waves spectral features [14]. Moreover, terahertz waves properties (e.g., absorption and refractive response) of common wood species have been characterized, providing a physical basis for species-dependent attenuation and potential density-related trends [15]. Terahertz waves have also been used to simultaneously estimate density, moisture content, and fiber direction [16], to monitor moisture diffusion in wood [17], and to predict moisture content from terahertz waves time-domain signals [18]. Beyond species and moisture characterization, terahertz signals can capture thermal process related changes, enabling assessment of water content and delamination states in wood materials under pyrolysis temperature conditions [19]. Utilizing this identification model is expected not only to improve recycling efficiency through effective timber resource utilization but also to propose a resource circulation model for sustainable forestry. This is anticipated to greatly contribute to decarbonization and the formation of a circular economy society.

1.1. Biomass Raw Materials

In December 1997, the Kyoto Protocol was adopted at the Kyoto Conference on Global Warming Prevention (COP3) held in Kyoto to reduce greenhouse gas emissions in developed countries. Furthermore, in December 2015, the Paris Agreement was adopted at the 21st Conference of the Parties (COP21) to the United Nations Framework Convention on Climate Change held in Paris. This Paris Agreement establishes a shared long-term global goal to limit global warming caused by greenhouse gases to well below 2 degrees Celsius above pre-industrial levels and to pursue efforts to limit the temperature increase to 1.5 degrees Celsius [20]. Following the adoption of the Paris Agreement, various social science analyses have been conducted in Japan and around the world [21].

In recent years, global attention has been turning to biomass feedstocks to curb car- bon dioxide, a greenhouse gas contributing to global warming. Biomass feedstocks refer to renewable organic resources derived from plants and animals. Among these, plant-derived biomass feedstocks absorb carbon dioxide from the air and release oxygen through photosynthesis during plant growth. Therefore, by offsetting the carbon dioxide emissions generated when processing or burning the cultivated plants, it is possible to reduce total carbon dioxide emissions [22].

Other advantages of biomass feedstocks include the absence of depletion concerns and ease of processing. Particularly due to this processing advantage, biomass resources are utilized in numerous products such as food fertilizers, livestock feed, biofuels, and bioplastics. Among these, wood containing high-strength pulp fibers is used not only for construction materials and daily necessities but also for various applications including bioplastic feedstocks and reinforced fibers. In Japan, the total demand for wood products in 2022, converted to logs, was 85.09 million cubic meters, representing a 3.6% increase from the previous year [23]. Globally, although industrial log consumption in 2022 decreased by 2% from the previous year, it still amounted to a significant 2.026 billion cubic meters [23]. Furthermore, FAO statistics published in GLOBAL FOREST PRODUCTS 2023 state that the global log harvest for 2023 was 1925 million cubic meters [24].

The recycling of these wood materials is being actively promoted. For example, waste wood generated during wood processing, along with tree branches and stumps, is finely shredded and processed into wood chips [25]. These wood chips are utilized as pulp for paper manufacturing and as fuel chips for biomass power generation. Indeed, Japan’s firewood consumption in 2022 reached 17.39 million cubic meters, an increase of 18.0% compared to the previous year [23]. The Forestry Agency of Japan attributes this to the growing prevalence of power plants utilizing woody biomass. On the other hand, imported wood chips and chips derived from imported logs account for 430,000 tons of the 11.06 million tons of domestic wood chip consumption [23]. For the management of national forests, forest density adjustment—one of the measures to prevent forest fires—is being implemented, and there is potential for increased domestic timber production going forward. To promote biomass recycling utilizing Japan’s abundant timber resources, it is necessary to actively utilize domestically produced wood.

In the recycling of such plant-derived biomass feedstocks, the moisture content of wood is a critical factor. In the construction field, wood moisture content generally serves as a benchmark for diagnosing material degradation based on cracking and deformation and is one factor in determining the reusability of building materials. From a recycling perspective, wood moisture content alters the energy required for water evaporation and impacts combustion efficiency [26]. Also, research data has been reported indicating that the number of pollutants generated by the combustion of wood pellets increases or decreases depending on the moisture content of the wood [27]. Furthermore, sorting by raw material type is also a crucial factor in wood recycling. For example, trees are divided into conifers and broadleaf trees, each possessing distinct characteristics. In “Comparative study for hardwood and softwood forest biomass: Chemical characterization, combustion phases and gas and particulate matter emissions”, Amaral et al. found that conifers contain more lignin than broadleaf trees, and this difference in lignin content results in differing CO2 emissions during combustion [28]. Furthermore, in “Evaluation of hardwood or softwood bark biomass as feed materials for aqueous-phase reforming gasification process”, Meryemoglu et al. found that differences in lignin content led to variations in the amount of hydrogen produced during bark hydrolysis [29]. Thus, the species of wood significantly influences both its consumption as fuel and the synthesis of hydrogen fuel derived from it.

1.2. Terahertz Waves

In recent years, development has progressed on the creation of a gun diode utilizing microwaves generated by applying an electric current to gallium arsenide crystals. Advances in the development of this high-frequency device have drawn attention to terahertz waves, a type of electromagnetic radiation. Terahertz waves are electromagnetic waves with frequencies between 0.1 and 10 THz, existing in the frequency band between radio waves and light waves. This electromagnetic wave possesses both the straight-line propagation characteristics of light waves and the penetrating power of radio waves, while also being non-invasive to the human body. Therefore, unlike X-rays, it does not cause cumulative damage to cells, enabling prolonged use. It also possesses greater penetration depth than infrared radiation. Consequently, research is actively progressing in areas such as non-destructive testing and applications in plastics recycling [30,31].

Research on using terahertz waves for wood involves non-destructive testing to assess internal conditions, such as evaluating wood quality including growth rings [32], and predicting wood moisture contents [33]. All these studies focus on non-destructive testing for assessing internal conditions. On the other hand, few studies have compared the response of terahertz waves to different wood species. Furthermore, since wood is a non-polar material like plastic and concrete, it is conceivable that terahertz waves—which penetrate non-polar materials and are absorbed by substances like water—could be utilized for sorting by exploiting differences in response.

1.3. Research Objectives

This study compared the terahertz wave transmittance of wood, a cellulose-derived biomass material. By comparing wood properties using terahertz waves, it is possible to identify different materials, offering potential applications as sensing technology in recycling. For example, identifying wood species allows waste wood to be sorted by tree type.

Furthermore, differences in the moisture content of the wood itself may cause variations in its response to terahertz waves. Differences in moisture content led to variations in the probability of incomplete combustion and the amount of heat released during combustion. Particularly in the construction field, the moisture state of wood can cause deformation or cracking, making moisture content prediction crucial. Therefore, this study also discusses the identification of wood species and the differences in terahertz wave responses based on moisture content.

1.4. Prospect of Sub-Terahertz Applications

By sorting wood based on species and moisture content, it becomes possible to select wood suitable for specific conditions. Hardwoods with low moisture content and lignin content achieve high combustion efficiency and low carbon dioxide emissions. It is believed that efficiently utilizing sorted wood as biofuel can suppress total carbon dioxide emissions while utilizing waste wood as fuel.

2. Materials and Methods

2.1. Samples for Measurement

Multiple pieces of wood of different types and sizes were prepared for the samples. The wood samples are classified into eight types as shown in

Table 1. Details of wood samples.

Type	Coniferous /Broadleaf	Size	Pieces
Japanese Ceder	Coniferous	40 mm × 40 mm × (10 or 14 or 18) mm	3 per size (9 total)
Hiba
Western red ceder
Redwood
Radiata Pine
Chestnut	Broadleaf
Zelkova serrata
Ipe

(Japanese Cedar: Cryptomeria japonica, Zelkova: Zelkova serrata, Hiba: Thujopsis dolabrata, Western red cedar: Thuja plicata, Redwood: Sequoia sempervirens, Ipe: Tabebuia spp., Chestnut: Castanea crenata, Radiata pine: Pinus radiata). All these wood samples were provided by house builders as materials used for residential construction. Among these, Cedar, Hiba, Western red cedar, Redwood, and Radiata pine are classified as coniferous woods, while Zelkova, Ipe, and Chestnut are classified as Broadleaf woods.

As shown in Table 1, the wood samples were pre-cut by the housing manufacturers into three sizes: 40 mm in length, 40 mm in width, and thicknesses of 10 mm, 14 mm, and 18 mm. To avoid the influence of knots and other factors on the measurements, three specimens were prepared from each of three sizes for each tree species (total of 9 specimens).

2.2. Humidity Conditioning of Wood Samples

For wood samples, the saturated salt method was used to prepare the water content. A beaker containing distilled water and saturated aqueous solutions of the reagents listed in

Table 2. Reagents for and humidity and saturation that can be conditioned(Humidity: Refer to JIS standards/Solubility: Refer to material safety data sheet from Kanto Kagaku Co.; Tokyo, Japan) [34].

Reagent	Humidity at 20 Degrees Celsius (%)	Solubility at 20 Degrees Celsius (g/100 mL)
LiCl	11.1~12.6	83.2
MgCl2	33.1 ± 0.2	54.3
NaBr	59.1 ± 0.5	73.3
K2SO4	97.6 ± 0.6	11.1

was prepared in advance. Referring to Table 2, the reagents were mixed while being added using a spoon and a magnetic stirrer. The reagent was added until the added reagent had a dissolved residue.

The moisture content inside the sample was prepared by sealing the bat containing the sample and humidity conditioning reagent in a desiccator and maintaining a constant humidity inside the desiccator. The moisture content was gradually increased by changing the type of humidity conditioning reagent. To clarify the progression of hydration and the change in the transmission of the samples to electromagnetic waves, the moisture content was prepared in two types: one in a dry state and the other in which the hydration progressed to a maximum level.

The moisture content was calculated based on the following equation.

U = (W₁ − W₀)/W₀ × 100

(1)

U is the moisture content (%), W₀ is the mass of the wood in its dry state (g), and W₁ is the mass of the wood with the moisture content prepared (g).

The dried wood samples were stored for two months in a desiccator containing saturated LiCl solution (around 12% humidity at 20 degrees Celsius, listed in Table 2). After storage, the wood mass was measured at three-day intervals. If the moisture content fluctuated within the range of −0.1% to 0.1%, the wood was judged to be in a stable dry state.

The conditioned wood samples were stored in a desiccator (AS ONE, Tokyo, Japan) alongside the dry wood samples with saturated MgCl₂ solution (7 days with humidity of 33.1% at 20 degrees Celsius, listed in Table 2), saturated NaBr solution (4 days with humidity of 59.1% at 20 degrees Celsius, listed in Table 2), and saturated K₂SO₄ solution (20 days with humidity of 97.6% at 20 degrees Celsius, listed in Table 2). After storage, the wood samples were deemed conditioned based on mass change at the three-day intervals, as same as the dry wood samples.

The moisture content of the conditioned wood is as follows: Japanese cede; 7.9~10.8%, Zelkova; 8.8~11.4%, Hiba; 8.1~11.7%, Western red cedar; 10.0~13.6%, Redwood; 8.1~11.1%, Ipe; 7.1~10.1%, Chestnut; 10.0~13.6%, Radiata pine; 9.1~16.4%.

2.3. Terahertz Measurement System

As an experimental method, we adopted transmission measurements using terahertz waves.

For terahertz wave transmission measurements of wood, we used the apparatus shown in Figure 1. Based on this device utilizing terahertz waves, development of equipment suitable for identifying plastic materials and research into related analytical techniques are already underway [35].

The apparatus consists of a light source, a detector, a lens, and a stage for fixing the sample. The light source employed a Gunn diode, while the detector utilized a Schottky barrier diode. Both were equipped with horn antennas. The light source emits sub-terahertz waves, with separate transmitters prepared to irradiate electromagnetic waves at 30 GHz (Vega tech., Tokyo, Japan; 200 mW), 65 GHz (Amtechs corp., Tokyo, Japan; 25 mW), and 90 GHz (SPACEK Labs., Santa Barbara, CA, USA; 15 mW). Each detector employs detectors manufactured by ERAVANT Company (Torrance, CA, USA) with the following performance characteristics: 1300 mV/mW (30 GHz), 1000 mV/mW (65 GHz), 800 mV/mW (90 GHz). Horn antennas were used according to the following specifications: WR-28 (26.5~40.0 GHz), WR-15 (50.5~75 GHz), WR-12 (75~110 GHz). This utilizes sub-terahertz waves, which are low-frequency electromagnetic waves within the terahertz range known for their particularly high transmissibility. The detector receives sub-terahertz waves transmitted through the sample, while the lens focuses sub-terahertz waves emitted into the air.

2.4. Measurement Method

The wood sample was fixed to the stage portion of the apparatus for transmission measurements. We measured a wood sample using the following parameters: electromagnetic wave frequency, sample type, sample size, and moisture content. Terahertz waves emitted from the light source passed through the samples and were received by the detector. Based on the intensity of the received electromagnetic waves, the transmittance was obtained by comparing the intensity of the electromagnetic waves in air with that after passing through the sample.

We compared the transmittance values obtained from the measurements and analyzed the relationship between sample differences and changes in transmittance.

3. Results and Discussion

3.1. Regarding Wood Species and Transmittance

Figure 2 shows the relationship between the transmittance of 65 GHz terahertz wave and the specific gravity of the sample in the dry state and after humidity conditioning, respectively. Figure 3 shows the relationship between the transmittance at 90 GHz and the sample specific gravity for the dry and humidified samples, respectively. Figure 4 shows the relationship between the transmittance of the dry sample at 30 GHz and the specific gravity of the sample. For wood, specific gravity (g/m²) is generally used as the standard, so a graph was created with specific gravity as the variable [5]. For each of the three thickness of wood, three pieces were measured in sets to determine their degree of transparency. Additionally, each figure includes a graph showing the relationship between the average transmittance and specific gravity of wood of various thicknesses, along with its correlation coefficient.

Comparing the same type of wood, we can see differences in transmittance among the nine samples. We suspect this difference in transmittance among samples of the same type of wood species may be due to the amplitude of the incident terahertz waves or interference from the terahertz waves reflected from the samples.

3.1.1. Sample Reaction Trends

Figure 2a and Figure 3a show that a negative correlation coefficient ranging from −0.98 to −0.71 square meter/grams exists between the specific gravity (kinds) of dry wood samples and the transmittance of the wood. These graphs show that as the specific gravity of the dry wood sample increases, the transmittance of the sample at 65 GHz and 90 GHz is inversely proportional to the specific gravity of the sample. On the other hand, as shown in Figure 4, this relationship was not observed at 30 GHz, where the correlation coefficient range is −0.20~0.02 square meter/grams. This indicates that there is a negative correlation between the specific gravity and the transmittance of the samples when the wood is irradiated with electromagnetic waves at 65 GHz and 90 GHz.

Comparing Figure 2a,b, the negative correlation between the specific gravity and transmittance of the samples is more clearly seen in Figure 2a, which shows the samples in a dry state, than in Figure 2b, which shows the samples in a humidified state. This same trend can be observed in Figure 3a,b.

Furthermore, when irradiating with 90 GHz sub-terahertz waves instead of 65 GHz waves, a clear negative correlation was observed between wood density and its permeability. As density decreases, the density of wood fibers decreases, making it easier for gaps to form between fibers. Water entering these gaps increases the amount of water fixed within the wood. Furthermore, as is particularly understood in relation to Drude, terahertz waves located in the low-frequency band are easily absorbed by polar molecules in water and are strongly affected by moisture. Therefore, high frequency 90 GHz sub-terahertz waves are less affected by moisture content and are expected to exhibit high permeability in wood with low specific gravity.

The correlation coefficient is similar constant with variations in wood thickness. Therefore, the influence of sample thickness on transmittance is small.

3.1.2. Sorting of Wood Species

Coniferous trees have longer fibers than broadleaf trees. Also, the graph shows that coniferous wood tends to have a higher specific gravity than broadleaf wood. Therefore, this study examined the transmittance values that could be extracted for approximately half of the 72 wood samples, along with the proportion of coniferous wood. Figure 2, Figure 3 and Figure 4 show the reference transmittance values as dotted lines in the graphs.

For example, in the 65 GHz transmittance measurements shown in Figure 2a, when collecting wood samples in a dry state with a transmittance of 0.5 or higher, 39 samples were recovered, representing nearly half of the 72 wood samples. Furthermore, of the 39 selected samples, coniferous wood accounted for approximately 79% (31 samples). Similarly, from Figure 2b, in the 65 GHz transmittance measurements, we can collect 39 samples with a transmittance of 0.17 or higher from the humidity conditioned wood samples, and coniferous wood accounted for approximately 85% of these. From Figure 3a, in the 90 GHz transmittance measurements, we can collect 36 samples with a transmittance of 0.35 or higher for the dried wood samples, and coniferous wood accounted for approximately 83% of these. From Figure 3b, in the 90 GHz transmittance measurements, we can collect 36 samples with a transmittance of 0.04 or higher from the humidity-conditioned wood samples, and coniferous wood accounted for approximately 83% of these.

Thus, it is possible to extract many coniferous trees from multiple wood samples based on transmittance at a specific terahertz frequency.

3.2. Moisture Content of Wood Sample

Figure 5 shows a graph of transmittance when 65 GHz and 90 GHz electromagnetic waves were irradiated onto dried and conditioned samples of radiata pine. All nine samples were measured after drying or conditioning the same sample. Figure 6 shows a graph of the transmittance of Hiba samples divided into three thicknesses when irradiated with 65 GHz and 90 GHz electromagnetic waves. Here too, measurements were taken after drying or conditioning three identical-sized wood samples.

Figure 5 shows that differences in moisture content among the wood samples cause variations in transmittance, with higher moisture content generally resulting in lower transmittance. The experimental results clearly indicate that the difference in moisture content distinctly separates the transmittance values, particularly when irradiated with 90 GHz electromagnetic waves. Figure 6 further shows that, for samples of the same size, those in the humidified state with higher moisture content tend to have lower transmittance than those in the dried state.

Based on the above, it has been confirmed that the response to terahertz waves changes with variations in wood moisture content. Based on these results, it is possible to measure wood moisture content by utilizing terahertz waves. In the paper by Yu et al., measurements were performed using broadband electromagnetic waves with a THz-TDS (Time-domain Spectroscopy) system [33]. In contrast, this study adopted measurements using a compact diode device capable of irradiating electromagnetic waves at a single frequency. Utilizing this miniaturized device enables implementation in the field, suggesting enhanced potential for the societal deployment of terahertz wave-based devices.

This paper performed irradiation on wood of different thicknesses. It discusses the effect of wood thickness on transmittance and provides a more comprehensive examination of terahertz wave response in wood.

4. Conclusions

In this study, we conducted experiments to compare the response characteristics of wood, a cellulose-derived biomass material, to terahertz waves at specific frequencies based on its terahertz spectrum. In the experiments, we irradiated wood samples from different tree species with terahertz waves, calculated the transmittance from the intensity of the transmitted terahertz waves, and compared the transmittance values. The experimental results revealed the following:

• Regarding wood species, wood exhibited a negative correlation with both specific gravity and transmittance at specific terahertz frequencies of 65 GHz and 90 GHz. This trend was particularly pronounced when irradiated with 90 GHz sub-terahertz waves.
• Regarding wood moisture content, the negative correlation between specific gravity and transmittance was more clearly evident in dry wood than in wet wood, due to differences in the inherent moisture content of the wood.

By utilizing differences in wood’s transmittance, it is possible to identify both tree species and moisture content based on transmittance when irradiated with terahertz waves of specific frequencies. Mechanizing wood identification enables the identification of wood that should be removed during the recycling process. Machine-based wood sorting can reduce the environmental impact when using wood as fuel. Furthermore, predicting moisture content enables determining the most suitable processing and storage methods for the wood.

Therefore, sorting technology utilizing terahertz waves is considered to contribute to promoting wood recycling.