Microwave detection and quantification of water hidden in and on building materials: implications for healthy buildings and microbiome studies

Visible signs of dampness and mold in buildings are epidemiologically associated with adverse health outcomes for occupants [1‐3]. However, there is a quantitatively large variance in studies of such health outcomes [4, 5], and there is a need for deeper understanding of these associations. In addition to factors such as genetic and experiential differences among occupants, a key confounding factor is the limited availability of data on the location and amount of water present. Moreover, information on the presence of water is critical for implementing any remedial action. The most added-value from improved water detection would be expected from methods that are noninvasive, quantifiable, spatially resolved, and able to detect hidden water. The capability to passively monitor over long timescales is also valuable, in order to monitor for water that may only appear sporadically.

Current non-destructive technologies for the measurement of water in buildings have significant limitations, especially in the detection of hidden water [6‐8]. Unaided visual inspection is widely used to assess outer surfaces for signs of water damage or mold, with improved sensitivity sometimes offered by infrared imaging of surface temperature [9]. Infrared imaging relies on detecting the temperature differentials that can form between wet areas, which are often relatively cool, and the surrounding dry areas [10]. To the best of our knowledge, the reviewed literature has not yet critically examined infrared detection of hidden water with regard to sensitivity, quantitative reliability, confounding factors, and how deeply into building materials infrared-based detection can penetrate.

The ability of microwaves to penetrate through walls make them an attractive solution for the detection of hidden water. Microwave aquametry [11] is already used to measure moisture during the preparation of building materials such as wood [12, 13] and concrete [14, 15], as well as in a range of other materials such as soils [16], seeds [17], cheese [18] and textiles [19]. Moisture monitoring within building walls has been performed with qualified success by measuring microwave transmission between probes drilled into the wall [20], however this partially-destructive technique has not been broadly adopted.

The present paper makes no claim to have developed a microwave technology that works in a practical way to detect hidden water in buildings. However, we provide a simple demonstration of the detection of small volumes of water in and around common building materials. Interdisciplinary collaboration and engineering efforts will be required to turn this demonstration into a practical device or application. Practical development will be further considered in the discussion.

We used a simple setup, consisting of two microwave horns (A-info, LB-OH-159-15-C-SF) connected to a vector network analyzer (Agilent, PNA N5222A), as shown in Fig. 1(a). This allowed us to measure the microwave reflection and transmission through test samples placed between the horns, as a function of microwave frequency.

The reflection and transmission are measured as S-parameters. As indicated in Fig. 1(a), S11 (S22) measures reflection of a signal sent from horn 1 (2), and S21 measures transmission from horn 1, through the test sample, to horn 2. Starting with a dry test sample, we used a pipette add water in 1 mL steps and monitored the resulting change in S-parameters, making measurements within a few seconds of each step. For an S-parameter S_γ (γ = 11, 22, 21), we define the change in reflection or transmission due to the added water as ΔS_γ = S_γ - S_γ0, where S_γ0 is the S-parameter measured without any water present. The 4–8 GHz bandwidth of our measurements was chosen to match the bandwidth of the available microwave horns, and the network analyser output power was 0 dBm (1 mW).

To demonstrate the suitability of microwaves for detecting water in inaccessible spaces, such as inside walls, we used a hollow concrete brick, shown in Fig. 1(b), with 7 cm of concrete above and below the central hole. The brick was dried in air for 1 week before the measurement. Figure 2 shows the changes in transmitted and reflected microwave signals as we added water with a pipette, creating a free-standing water layer in the hollow centre of the brick. We detect water volumes as small as 1 mL, and see a strong increase in absorption with increasing water volume. There is little change in the reflected signal with water volume, however we do see oscillations in reflectivity (and to a lesser extent absorption) as a function of microwave frequency. We attribute this to interference between reflections from the water-brick and water-air surfaces, which depends on the ratio of microwave wavelength to water layer thickness (see discussion below). We did not see oscillations as a function of water volume in this experiment. We interpret this lack of change with water volume as follows: the area covered by the water layer in the brick increased with volume, however the thickness (roughly 1–2 mm) remained constant. Absorption of water into the brick occurred over tens of minutes, and was negligible over the 7 min measuring time. Water loss due to evaporation, which is strongly dependent on airflow velocity [21] can also be assumed to be negligible within the confines of both the hollow brick and our lab.

Metallic objects in a building, such as pipes, will block microwave transmission. We show that water on a metallic surface can be detected through its influence on the reflected microwave signal. We used a 5 mm thick aluminium sheet as test sample, and created a free-standing water layer directly on top. As transmission through the aluminium was essentially zero, Fig. 3(a + b) show minimal variation in transmitted signal with water volume. However, Fig. 3(a + c) do show a strong decrease in reflection (S11) with water volume, and we again detect volumes down to 1 mL. This change in reflection signal, which was not seen in Fig. 2, is due to the water blocking the signal from the aluminium surface. We again see oscillations in reflection as a function of frequency but not water volume, due to interference between the water-aluminium and water-air interfaces, and the fact that increasing water volume did not change the water layer thickness.

To demonstrate the effect of water layer thickness, we used a Pyrex container as test sample, which ensured that the water layer thickness increased approximately linearly with water volume. Figure 4 shows the changes in microwave absorption and reflection, where we can see S-parameter oscillations as a function of both frequency and water thickness. Figure 4(d-f) show line cuts for different frequencies, where we can see that the oscillation period with water thickness is different for each S-parameter, and varies with microwave frequency.

We can understand the S-parameter oscillations by considering microwave interference effects in a thin dielectric film, as described in classical optics [22]. The incident microwave undergoes multiple transmission and reflection events at the air-water and water-container boundaries (see Fig. 5), producing waves which interfere with one-another. In the most simple picture, the net reflection and transmission coefficients oscillate sinusoidally with a frequency proportional to nd cos(θ)/λ, where n is the complex refractive index of water, d is the water thickness, θ is the microwave angle of incidence, and λ is the microwave wavelength. This qualitatively explains the observed S-parameter oscillations as a function of microwave frequency (∝1/λ) and water thickness, and also the faster oscillations as a function of water thickness for higher microwave frequencies, where the d/λ ratio is larger. The amplitude of the S-parameter oscillations as a function of water thickness decays faster at higher microwave frequencies (Fig. 4c), which is due to the absorptive component of the refractive index increasing with microwave frequency [23]. Accurate modelling of the quantitative features of the S-parameter oscillations, such as how the oscillation frequency is different for S21, S11 and S22, and for different measurement setups, is beyond the scope of this work. These features may be explained through explicit consideration of factors such as microwave attenuation in the water, integration over a range of θ for each microwave horn, and the material-dependence of reflection and transmission at the various water-(wet/dry) concrete, water-aluminium, and water-Pyrex boundaries. In future setups, these factors may be best accounted for by performing 3D holographic reconstruction of spatially resolved measurements [24].

We propose that improved detection and quantification of hidden water in buildings would enable more efficient and effective building design and remediation leading to improved public health. Better data on the state of water in all its forms over time might improve the relevance of microbiome analysis to the health of building occupants. This paper includes a demonstration that microwave sensing offers one approach for the problem of detecting hidden water in the built environment. The most effective realization would be best accomplished via an interdisciplinary research program including the healthy building disciplines, microwave engineering or physics, and microbiology as related to epidemiology.