CDE4301 Innovation & Design Capstone

AY2025/2026 Semester 1

The Living Filter: Plant-Microbe System for Indoor Air Purification

Final Report

Group IS-405

Acknowledgements

Thank Project Supervisor, Co-Supervisor, and any other parties who have helped in the project.

Dr Elliot Law
Dr Jovan Tan

Industry Partner: Greenairy

Tanvi R. Thombre

EDIC Staff

Ms Annie Tan
Mr Patrick Tan

NUS Professors

Dr Aleksandar Kostadinov
Dr Veerasekaran S/O P Arumugam

Abstract

Indoor air quality (IAQ) has become an increasingly important environmental and public health concern, particularly in highly urbanised and air-conditioned settings such as Singapore, where individuals spend up to 90% of their time indoors. Among indoor pollutants, volatile organic compounds (VOCs) are of particular concern due to their persistence, continuous emission from building materials and household products, and documented health impacts. Conventional mitigation strategies such as ventilation and source control are often insufficient in fully addressing VOC accumulation, while existing air-cleaning technologies, including activated carbon filtration and photocatalytic oxidation, face limitations such as secondary pollutant formation, material degradation, and high maintenance requirements.

In response, plant-based biofiltration systems have emerged as a promising alternative, leveraging natural plant uptake and microbial degradation processes. Notably, the rhizosphere microbial community has been shown to contribute significantly to VOC removal efficiency. However, current commercial systems and research applications often do not systematically optimize the pairing between specific plant species and microbial consortia, leading to variability in performance and limited scalability. Furthermore, laboratory-optimised microbial strains, while effective under controlled conditions, may not exhibit the same efficiency in real-world environments due to differences in environmental adaptability and system interactions.

This study addresses these gaps by investigating which commercially available indoor plant and microbe pairing demonstrates the highest VOC removal efficiency across five commonly occurring indoor VOCs: formaldehyde, benzene, toluene, ethylbenzene, and xylene (FBTEX). By synthesising existing literature, evaluating current technologies, and analysing plant–microbe interactions, this research aims to identify optimal pairings that balance effectiveness, practicality, and real-world applicability. The findings contribute to the development of more efficient, scalable, and sustainable IAQ solutions, advancing the role of biological systems in indoor environmental management.

1. Introduction (Sammi)

1.1. Importance of Indoor Air Quality

Air quality is fundamental to human health and well-being, influencing respiratory, cardiovascular, and cognitive functions (World Health Organization, 2025). While outdoor air pollution has received greater attention due to its visible impacts, it represents only part of overall exposure. Indoor air quality (IAQ), by contrast, remains an under-recognised yet equally critical aspect of environmental health, as indoor pollutants are often invisible, odourless, and difficult to detect without monitoring (Dcarrick, 2024; US EPA, 2025).

In Singapore, IAQ is particularly important due to the country’s tropical climate, dense urban environment, and reliance on air-conditioning. Singaporeans spend up to 90% of their time indoors, inhaling approximately 11,000 litres of air daily, often in enclosed spaces such as homes, offices, and shopping malls where pollutant accumulation is common (BCA, 2020; Lung Basic, n.d.; Pedro et al., 2024c). Notably, Singapore ranks among the top countries where indoor air can be worse than outdoor air, highlighting the significance of indoor exposure (Air Quality Connected Data Study, 2024).

Indoor air pollution can be broadly categorized into two main groups: Particulate Matter (PM) and Gaseous Contaminants, as illustrated in Table 1.

Table 1: Types of Indoor Air Pollutants and its impacts (California Air Resources Board, 2015; Tran, Park, & Lee, 2020)
	Particulate Matter (PM)	Gaseous Contaminants
Description	Complex mixture of microscopic solid particles and liquid droplets suspended in air	Inorganic Gases (Nitrogen oxides, carbon oxides)	Organic Gases (Volatile Organic Compounds, VOCs)
Sources	- Outdoor air - Indoor sources (pollen, mold spores, dust mites) - Human activities (cooking, candle burning)	Combustion from gas-fueled cooking and heating appliances	- Printers - Household products (paints, pesticides, dyes, cleaning agents)
Health Impacts	Respiratory symptoms, nonfatal heart attacks, aggravated asthma	Respiratory symptoms, fatigue, chest pain	Short Term: Eye, nose and throat irritation, headaches, loss of coordination and nausea Long Term: Organ damage, increased risk of cancer

Among these pollutant categories, VOCs are of particular concern due to their persistence, indoor dominance, and long-term health risks, which will be discussed in the next section.

1.2. Volatile Organic Compounds (VOC) in Indoor Air and Their Impacts

Among indoor air pollutants, VOCs are particularly significant due to their widespread presence and strong association with indoor exposure (Air Quality Expert Group, 2022). Unlike many other pollutants, VOCs originate primarily from indoor sources, making their behaviour closely tied to building materials, occupant activities, and ventilation conditions (Air Quality Expert Group, 2022; US EPA, 2025). Their continuous emission and interaction with enclosed environments contribute to their persistence and make them a key focus in indoor air quality research. The defining behavioural characteristics that explain their indoor dominance are summarised in Figure 1.

With existing research indicating that Singapore’s average indoor VOC concentration is approximately 190.37 ppb, we conducted a study monitoring VOC levels within a design studio room at Block E2A at the College of Design and Engineering, NUS, over a one-week period. As shown in Figure 2, the measured VOC concentrations were slightly higher, with an average of approximately 223.0 ppb. This suggests that indoor VOC levels can vary depending on room usage and occupancy patterns. The day-to-day variations observed are likely attributed to intermittent human activities within the space, particularly during periods when students enter the studio for lessons and use whiteboard markers. For example, the lowest VOC concentration was observed on Sunday, 22 March, which can be attributed to reduced human activity compared to other days, as Sunday is typically a non-working and non-school day. These markers contain volatile solvents, such as alcohols, which readily evaporate during use and release VOCs into the indoor environment (US EPA, 2025). As a result, such routine activities can contribute to short-term increases in VOC concentrations. Overall, this highlights how everyday indoor behaviours, even in non-industrial settings, can influence air quality and lead to elevated VOC levels in enclosed spaces.

To better understand the implications of these emissions, it is important to consider the specific VOC compounds commonly present in indoor environments. Although hundreds of VOC species may be present indoors, a group of five compounds, formaldehyde, benzene, toluene, ethylbenzene, and xylene (FBTEX), are among the most consistently detected and are prioritised due to their significant contribution to indoor health risks rather than their proportion of total VOC concentration (Air Quality, Energy and Health (AQE), 2010).

Formaldehyde, commonly emitted from pressed-wood products, coatings, and adhesives, is a known irritant and probable human carcinogen even at relatively low concentrations, with large-scale studies showing indoor formaldehyde concentrations in homes commonly ranging from about 10 to 40 μg/m³, with upper 95th percentile values above 50 μg/m³ (Kaden, Mandin, Nielsen, & Wolkoff, 2010; Canada, 2021).

Meanwhile, BTEX compounds are aromatic hydrocarbons originating from solvent use, paints and adhesives, indoor combustion sources and infiltration from vehicle exhaust or attached garages (Vardoulakis et al., 2020). Indoor benzene levels typically span 2 to 12 μg/m³ in non-smoking homes in Europe, with occasional higher peaks associated with smoking or fuel exhaust infiltration (Harrison, Saborit, Dor, & Henderson, 2010). They are frequently used as indicators of indoor VOC pollution due to their volatility and persistence (Baberi et al., 2022).

Given their prevalence, persistence, and health impacts, this project focuses on FBTEX as representative indoor VOCs, enabling a targeted evaluation of VOC behaviour and mitigation strategies in indoor environments.

Beyond their role as key indoor pollutants, VOC exposure is associated with a wide range of health, environmental, productivity, and economic consequences, as summarised in Figure 3.

As illustrated in Figure 3, VOC exposure has been linked to multiple adverse outcomes across different domains. In terms of health, short-term exposure commonly results in irritation, headaches, and dizziness, while prolonged exposure has been associated with organ damage and increased cancer risk, particularly from compounds such as benzene and formaldehyde (Lai, 2025). Environmentally, VOCs such as BTX contribute to the formation of ground-level ozone and photochemical smog through reactions with nitrogen oxides, leading to reduced air quality and visibility (NEA, 2024). Elevated indoor VOC concentrations have also been associated with declines in cognitive performance, including reduced concentration, memory, and task efficiency, highlighting their impact on workplace productivity (ScienceDaily, 2023). In Singapore, the broader economic burden of air pollution has been estimated at approximately S$506 million in combined healthcare and productivity losses (Tan et al., 2023).

Collectively, these findings position VOCs as a major public health, environmental, and economic concern. Reducing indoor VOC emissions is therefore essential and aligns with frameworks such as Singapore’s SS 554:2016 Code of Practice for Indoor Air Quality.

1.3. Current Solutions to tackle VOC and Their Limitations

Existing strategies to manage indoor VOC levels typically follow a hierarchical approach, including source control, ventilation, air cleaning, monitoring, and behavioural controls, as illustrated in Figure 4. They are ranked based on their effectiveness and sustainability, which follows Singapore Standard SS554:2016 Code of Practice for indoor air quality, prioritising elimination before ventilation, treatment, behavior and control (NEA, n.d.).

Figure 4: Hierarchy pyramid of current strategies tackling VOC - Click the + buttons to learn more

While these strategies can reduce pollutant exposure, their effectiveness in managing VOCs remains limited due to the continuous and persistent nature of indoor emissions.

Monitoring systems provide real-time data on pollutant concentrations but do not actively remove contaminants. Similarly, behavioural controls, such as reducing the use of VOC-emitting products, are highly dependent on consistent user compliance, which is difficult to maintain across varied indoor environments. As a result, these approaches serve primarily as supplementary measures rather than standalone solutions.

Source control and ventilation are widely regarded as primary preventive strategies for IAQ improvement. By reducing pollutant emissions and diluting indoor air, these methods can lower VOC concentrations by approximately 55%. However, VOCs are not fully eliminated due to their continuous emission from indoor materials such as furniture, paints, and adhesives (Joo et al., 2025). A meta-analysis of 77 residential IAQ studies further found that multiple VOCs continued to exceed chronic health benchmarks even under typical ventilation conditions, indicating that source control and ventilation alone are insufficient (Logue et al., 2011; Mata et al., 2022).

To address the limitations of preventive IAQ strategies, air-cleaning technologies have been widely adopted to actively remove indoor pollutants. These approaches rely on mechanical or chemical processes, however, each presents notable limitations for VOC removal, as summarised in Table 2.

Table 2: Strengths and limitations of air-cleaning technologies
Technology	Strength	Limitation
HEPA Filters (US EPA, 2022)	Capture up to 99.97% of particles ≥0.3 μm	Ineffective for VOC removal due to its gaseous nature
Activated Carbon (ScienceDirect, 2023)	Rely on adsorption of VOC	Re-release pollutants upon reaching saturation, requiring frequent replacement which generates secondary waste
Photocatalytic/ UV Oxidation (Zhong et al., 2013; Sivasankar et al., 2018)	Degrade VOC	- Formation of harmful by-products due to potential incomplete oxidation under indoor conditions - Requires continuous energy input - Catalyst degradation over time, releasing nanoparticles (TiO₂)
Plasma (Messan et al., 2022)	Degrade VOC	Often modest or inconsistent under indoor conditions

Hence, the limitations associated with existing IAQ strategies, including source control, ventilation, air cleaning, monitoring, and behavioural controls—such as high energy demands, ongoing maintenance requirements, variable performance, and the potential generation of secondary by-products, render them insufficient in addressing the persistent and continuous nature of indoor VOC emissions.

In response to this, nature-based solutions such as plant–microbe systems have gained increasing attention for their ability to provide continuous, energy-efficient VOC removal in indoor settings. Plant-based air purification systems, a form of phytoremediation, have been shown to remove a wide range of VOCs through biological processes, where pollutants are absorbed and metabolised into less harmful compounds. This approach reduces reliance on adsorption or chemical oxidation methods that may generate secondary pollutants. Studies have demonstrated their ability to target compounds such as BTEX, alcohols, and aldehydes (Sheoran et al., 2022). In addition, these systems typically operate with low energy input, require minimal maintenance, and can adapt to varying pollutant levels, making them suitable for long-term and sustainable indoor application.

1.4. Plant-Microbe System

Phytoremediation involves the uptake of gaseous pollutants through leaf stomata and cuticular adsorption, where VOC compounds are absorbed and subsequently metabolised within plant tissues via oxidation, conjugation, and sequestration processes, as illustrated in Figure 5 (Wolverton & Nelson, 2020).

However, plants alone are not the primary drivers of VOC degradation. Instead, microorganisms associated with plant roots (rhizospheric microbes) play a crucial role in enhancing the ability of plants to purify air by acting as the primary biochemical drivers of pollutant degradation. They enzymatically break down VOCs into harmless end-products, such as CO₂, water, and organic acids (Green Plants for Green Buildings, 2014). While plants facilitate the uptake of VOCs through their leaves and roots, the majority of degradation occurs through microbial activity within the rhizosphere and on leaf surfaces.

The rhizosphere, a root-influenced zone rich in microbial activity, plays a central role in this process. Plant roots release sugars and amino acids that sustain microbial populations, while VOC-degrading bacteria and fungi utilise these compounds as carbon and energy sources, enhancing overall removal efficiency. Exposure to VOCs can further enrich specialised microbial species such as Pseudomonas and Rhodanobacter, strengthening biodegradation capacity (Cruz et al., 2023). This creates the plant-microbe system, where plants and microbes function synergistically and mutualistically to enhance pollutant removal efficiency.

Empirical studies have demonstrated that plant systems with active microbial communities remove significantly higher VOC concentrations compared to sterilised systems, confirming the importance of plant–microbe synergy, as illustrated in Figure 6 (Ravindra & Mor, 2022). Under optimised conditions, plant–microbe systems have been shown to remove over 60% of VOCs, including formaldehyde, benzene, and xylene, in controlled indoor environments (Irga, Pettit, & Torpy, 2018). Beyond pollutant removal, these systems also provide additional benefits such as humidity regulation, oxygen generation, and improved psychological well-being, making them a multifunctional and sustainable approach for indoor air quality improvement.

Despite these advantages, the effectiveness of plant–microbe systems is not uniform. Variations in plant species, microbial communities, and their interactions result in significant differences in VOC removal performance (Aydogan & Montoya, 2011; Cruz et al., 2023). While plant–microbe systems generally outperform plant-only systems, the extent of improvement depends strongly on the specific pairing, and in some cases, certain combinations may underperform relative to expectations (Irga, Pettit, & Torpy, 2018; Orwell et al., 2004). This suggests that the effectiveness of these systems is highly dependent on biological compatibility and interaction dynamics.

While plant–microbe systems have demonstrated strong potential for indoor VOC removal, their practical application remains constrained by current design approaches. Existing systems largely focus on demonstrating general plant–microbe synergy, rather than optimising specific pairings for maximum performance. As a result, variability in system efficiency is often overlooked. An examination of current plant–microbe methods in the market is therefore necessary to understand how these systems are implemented and to identify key gaps in achieving optimised VOC removal.

1.5. Current Plant-Microbe Methods in Market

Current plant–microbe air purification systems range from passive biological systems to hybrid engineered technologies. Broadly, these systems can be categorised into three main types: passive plant-based systems, active botanical biofiltration systems, and engineered plant–microbe systems with enhanced microbial inoculation. While these approaches demonstrate the effectiveness of combining plants and microbes for VOC removal, they generally prioritise overall system functionality over the optimisation of specific plant–microbe pairings. Each of these system types will be discussed in greater detail in the following sections.

Engineered Microbial Inoculation
Recent developments in biotechnology have introduced engineered plant–microbe systems to enhance indoor VOC degradation. It utilizes genetically modified microbes that express specific enzymes to break down pollutants more efficiently than natural microbial communities. When paired with plants, these microbes enhance root-zone biodegradation, achieving up to 30% lower BTX concentrations compared to baseline conditions after 13 days of continuous exposure (Neoplants, n.d.).

However, key limitations relate to biosafety and containment. There is limited information on how engineered microbes are confined within indoor systems or prevented from entering natural ecosystems. Potential risks include horizontal gene transfer, uncontrolled proliferation, and disruption of native microbial communities (OECD, n.d.). In addition, these systems focus on enhancing microbial degradation rather than optimising plant–microbe pairing, which may limit overall effectiveness.
Photocatalytic Biofiltration

Photocatalytic biofiltration combines biological systems with photocatalytic oxidation to enhance VOC removal, as illustrated in Figure 7. These systems typically integrate plants and microbial substrates with catalysts such as TiO₂, activated by UV light to oxidise VOCs into less harmful compounds. The biological component supports pollutant uptake and partial biodegradation, while photocatalysis improves the breakdown of more persistent compounds. Studies reported that the system removes over 75% of VOCs such as methyl ethyl ketone (MEK) within 24 hours, compared to only 11% removal by a standard potted plant and achieves 85% formaldehyde removal within one hour (NATEDE, 2024).

However, this approach has several limitations. Photocatalytic processes require continuous energy input and may produce harmful by-products, such as aldehydes, under incomplete oxidation (Mo et al., 2019; Yu et al., 2021). Catalyst degradation over time can also reduce system efficiency due to the release of nanoparticles from TiO₂ or WO₃ coatings (Koivisto et al., 2018). Furthermore, system performance is largely driven by the photocatalytic component rather than biological processes, with limited emphasis on optimising plant–microbe interactions.
Smart Green Wall

Commercial systems such as the NAAVA Smart Green Wall utilise plant–microbe biofiltration in an active airflow setup, as shown in Figure 8. Air is drawn through a soilless growth medium containing plant roots and associated microorganisms, where VOCs such as MEK and toluene are absorbed and degraded. These systems have been reported to achieve 50–70% VOC removal efficiency in single-pass tests (Naava, n.d.).

However, their primary limitation is spatial and architectural constraint. Each unit requires substantial vertical space (approximately 1003 × 2113 × 350 mm), along with dedicated airflow and irrigation systems, making integration difficult in compact indoor environments such as small offices or residential spaces (Naava, n.d.). In addition, performance improvements are largely driven by engineered airflow rather than optimisation of plant–microbe pairing, limiting adaptability and scalability across different settings.

Collectively, as shown in Figure 9, current plant–microbe-based air purification methods demonstrate the potential of integrating biological and engineered approaches for indoor VOC removal. However, these systems primarily focus on enhancing overall system performance through microbial engineering, chemical augmentation, or mechanical design, which aim to improve the degradation capacity of individual microbial communities. In contrast, this study focuses on optimising the biological interactions between specific plant and microbial species, recognising that plant–microbe compatibility plays a critical role in VOC removal efficiency. As a result, variability in plant–microbe synergy and its impact on performance remains largely underexplored in existing approaches. This highlights a critical gap, where emphasis should shift from improving general system performance to identifying optimal plant–microbe pairings that maximise VOC removal under real indoor conditions.

2. Project Overview (Sammi)

2.1. Research Question

In summary, plant–microbe systems represent an advanced extension of phytoremediation, harnessing the synergistic relationship between plants and their root-associated microorganisms to enhance VOC degradation efficiency (Irga et al., 2018; Pettit et al., 2018). Despite this potential, existing implementations remain constrained by limitations in safety, scalability, and practicality, and more importantly, a lack of focus on optimising plant–microbe pairings for maximum efficiency.

Given these limitations, this project seeks to explore the development of a biological air purification system that is not only effective, but also safe for daily and long-term indoor use, as well as practical and accessible for household environments. In this study, safety is evaluated in terms of the absence of harmful by-products, the use of non-toxic and commercially approved biological components, and minimal risk of exposure to users and the indoor environment. Practicality is assessed based on ease of maintenance, system stability under typical indoor conditions, and compatibility with existing indoor spaces. Accessibility refers to the use of commercially available plants and microbial products, as well as considerations of cost and ease of implementation for end users.

Specifically, the study aims to address the research question:

“Which commercially available indoor plant–microbe pairing achieves the highest removal efficiency for key indoor VOCs (FBTEX), while being safe, practical, and accessible for household use?”

By addressing this question, the project aims to identify an optimal plant–microbe combination that provides a sustainable, low-cost, and effective solution for improving indoor air quality in everyday living environments.

2.2. Project Phases

To systematically address the research question, this study adopts a phased approach that integrates both secondary research and experimental validation.

The project is structured into four sequential phases across two semesters, designed to systematically identify the most effective plant–microbe combination for VOC removal. Each phase builds upon the previous one, ensuring a logical progression from preliminary validation to final optimisation, as illustrated in Figure 10. Refer to Appendix 1 for the project timeline.

3. Phase 1: Pre-Experiment - Pilot Study and Proof of Concept (Pearlyn)

Prior to the main experimental phase, a series of pre-experimental investigations were conducted to establish a robust and reliable framework for evaluating the effectiveness of plant–microbe systems in VOC removal. This phase comprised two key components: the pilot study and proof-of-concept experiments.

The pilot study was first undertaken to systematically refine and validate the experimental design. Emphasis was placed on identifying critical parameters, standardising controllable variables, and determining appropriate operating conditions and methodology to ensure data accuracy, consistency, and reproducibility. Building on the conditions established from the pilot study, subsequent proof-of-concept experiments were conducted to validate the effect of the synergistic effect on VOC removal and to assess performance variability across different plant–microbe pairings.

3.1. Pilot Study (Pearlyn)

Guided by the chamber study conducted by Natede et al. (2023) (see Appendix 2), our experimental design builds upon established methodologies and was adapted to suit the scale and constraints of our setup. A sketch of the intended experimental setup was first developed (Figure 11), illustrating the key components—namely the pollutant source, chamber, sensor placement, and plant system—and their spatial arrangement.

In contrast to the full-scale room configuration used by Natede et al., the present study employed a scaled-down chamber system to enhance experimental control and reliability. Full-scale environments are more susceptible to external disturbances, such as unintended air leakage through vents, fluctuations in ambient conditions, and spatial variability in pollutant distribution, all of which can introduce uncertainty into the measurements. In addition, replicating full-scale conditions is inherently challenging, as it is difficult to obtain multiple rooms with identical volumes, layouts, and ventilation characteristics for consistent comparison. Full-room experiments also require significantly larger quantities of pollutants and longer purge times to restore baseline conditions between runs, reducing experimental efficiency.

By utilising a smaller, enclosed chamber, these limitations are mitigated, allowing for tighter control over experimental variables and more consistent VOC measurements. Importantly, this scaled-down approach does not compromise the primary objective of identifying the best-performing plant or plant–microbe pairing, as the study focuses on relative performance comparisons under controlled conditions rather than absolute removal rates in real environments. Maintaining consistent conditions across all trials ensures that observed differences in VOC removal can be attributed to the biological systems themselves, enabling reliable ranking and selection of the most effective pairing. The reduced scale also enables multiple experiments to be conducted in parallel, improving efficiency and facilitating more robust comparative analysis across different plant systems.

To systematically implement this controlled setup, a logical flowchart (Figure 12) was developed to identify and control variables that could influence experimental outcomes. This process began with identifying the key components of the experimental system, including the plant system, chamber, sensor and pollutant source. For each component, potential factors affecting reliability and measurement accuracy were examined, followed by the implementation of appropriate control measures to standardize conditions across all experiments.

As illustrated in Figure 12 under the “Assessment & Control Strategy”, parameters known to impact experimental results—such as plant characteristics, light exposure, watering regimes, airflow, and environmental conditions—were carefully controlled to minimize variability. This ensured that any observed changes in VOC concentration could be attributed primarily to the intrinsic performance of the plant systems rather than external or uncontrolled factors.

Accordingly, this report will focus on the independent variable, namely the amount of pollutant introduced into the system. All other factors were maintained as constants or controlled variables. Detailed descriptions and rationale of these variables are documented in Appendix 3.

A pilot study was subsequently conducted within a test chamber over 24-hour cycles to evaluate and refine the experimental setup. The objectives of this study are:

To evaluate the feasibility of the experimental design
To determine the appropriate amount of pollutants to be used
To establish the optimal duration for each experimental run
To identify any potential issues or gaps in the setup that may affect the accuracy or reliability of subsequent experiments
To determine the appropriate procedure to setting up experiments

The pilot study adopted a reiterative experimental framework (Figure 13) in which the amount of pollutant was systematically varied across trials while all other parameters were held constant. Following each experimental run, the system performance was evaluated based on criteria such as sensor detectability, time required to reach steady-state conditions, and stability of VOC concentrations. Based on these observations, necessary refinements were implemented, and the process was repeated iteratively. This approach continued until optimal experimental conditions were established—specifically, conditions in which pollutant concentrations fall within the sensor’s detectable range while remaining sufficiently high to enable clear differentiation in VOC removal performance across plant species.

3.1.1. Pollutant Selection and Dosage

As pollutant dosage was identified as the independent variable to be iteratively refined, the selection of appropriate pollutant sources becomes a critical first step. The type and properties of the pollutants directly influence the initial VOC concentration within the chamber and, consequently, the starting point for the iterative approach. Therefore, it is necessary to first determine suitable pollutant sources before establishing the corresponding dosage for subsequent experiments.

In this study, sustained high VOC concentrations are preferable to an initial pulse of VOCs because they create continuous exposure conditions that better reflect real indoor environments and allow biological removal mechanisms to fully engage. Plants and associated microbes remove VOCs through processes such as stomatal uptake, cuticular absorption, rhizodegradation by root‑associated microbial communities, and microbial metabolism in the growth medium. These pathways are biologically mediated and require time to operate effectively rather than responding to a transient spike alone (Montaluisa-Mantilla et al., 2025).

Furthermore, continuous exposure is essential for the development and adaptation of the rhizosphere. Ongoing pollutant presence provides the necessary biostimulation for microbial populations to specialize their metabolic pathways for degrading specific compounds like FBTEX. Steady VOC levels also ensure consistent plant physiological responses, such as stable stomatal conductance, which can be misrepresented by the fluctuations following a single injection (Ye et al., 2025). Thus, maintaining a prolonged high VOC environment enables a more representative assessment of the dynamic and synergistic VOC removal capabilities of plants and microbes.

To ensure comprehensive coverage of common indoor air pollutants, the study focused on the five target VOCs (FBTEX). Three complementary pollutant sources were selected for the pilot study: Elmer’s White Glue, Aureo’s Liquid Shoe Polish, and Nippon Vinilex‑5000 water-based paint. Each source was chosen for its specific VOC profile and physical properties, which together enable sustained VOC exposure while representing a range of emission behaviors commonly found in indoor environments. The approximate VOC composition and concentrations contributed by each source are summarized in Table 3.

Table 3: Composition and Characteristics of Pollutant Sources Used in the Pilot Study
Pollutant	Type / Category	VOC	VOC (% by weight)	Viscosity
Elmer's White Glue	General-purpose adhesive (PVA)	Formaldehyde Benzene Toluene Ethylbenzene Xylene (Lim et. al, 2014)	<1–10% (Administrative code, n.d)	2.311 Pa·s (Roger, 2025)
Aureo's Liquid Shoe Polish	Liquid form	Toluene Ethylbenzene Xylene (Lim et. al, 2014)	15% (Administrative code, n.d)	~ 0.2389 Pa·s (Akinbomi et. al, 2022)
Nippon Vinilex-5000 Paint	Water-based wall paint	Formaldehyde (Lim et. al, 2014)	<2% (low-VOC) (Administrative code, n.d)	0.648 to 0.700 Pa·s (Nippon Paint (S) Co. Pte Ltd, 2024)

Elmer’s White Glue was selected as a primary pollutant source because it contains all five target VOCs, enabling simultaneous assessment of multiple compounds within a single material. However, its high viscosity restricts solvent mobility and diffusion, leading to slower VOC evaporation rates. Such behaviour is typical of viscous polymeric coatings and adhesives, where internal mass transfer resistance limits emission flux (Munekata et al., 2013). If only white glue were used, the slower VOC emission would likely prolong the time required for concentrations in the chamber to reach a plateau, potentially delaying the observation of measurable VOC removal by the plant and plant–microbe setups. To address this limitation and ensure sufficient VOC exposure during the experimental period, additional pollutant sources with differing physical properties were incorporated.

Aureo’s Liquid Shoe Polish was incorporated as a complementary VOC source due to its low viscosity, which facilitates more rapid evaporation compared with white glue. This characteristic ensures that the chamber reaches measurable VOC concentrations more quickly, enabling timely observation of VOC removal by the plant and plant–microbe setups. In addition, the shoe polish provides a concentrated source of Toluene, Ethylbenzene, and Xylene, broadening the spectrum of VOCs present and reflecting the diverse chemical profile commonly found in indoor environments.

Nippon Vinilex‑5000 water-based paint was added to supplement the Formaldehyde content, which is released more slowly from high-viscosity white glue. Even though it is a low-VOC formulation, paint contributes an additional, readily available source of Formaldehyde, helping to balance the relative concentrations of all five target VOCs in the chamber. The inclusion of paint also strengthens the experimental design by preventing reliance on a single pollutant type, reducing potential bias due to the physical and chemical properties unique to one material. This approach ensures that the experimental conditions more accurately mimic real-world indoor settings, where VOCs originate from multiple sources such as adhesives, paints, and personal care products.

Using reference experimental setups from Natede et al. (2023), a preliminary calculation was performed to estimate the target mass of VOCs required for the experimental chamber (see Appendix 4). Based on this value, the required volume of each pollutant was determined using Microsoft Excel’s Solver function to ensure that the total VOC mass matched the calculated target. The resulting volumes for each pollutant are summarized in Table 4.

Table 4: Volume required using Excel solver
Pollutant	Volume Required (cm3)
Elmer's White Glue	0.342
Aureo's Liquid Shoe Polish	0.470
Nippon Vinilex-5000 Paint	0.425

Although the calculated volumes in Table 4 provide a theoretical reference for estimating the expected VOC concentrations, the pilot study was intentionally conducted using 2.5 mL of each pollutant as the initial amount of pollutant instead of the calculated values. This choice was made for two primary reasons.

Firstly, 2.5 mL is the minimum volume that can be measured accurately and consistently with the available measuring cups, ensuring precision and reproducibility. Secondly, using a slightly higher volume compensates for the simplifying assumption in the preliminary calculations that all VOCs would evaporate fully within the chamber. In practice, VOC volatilization depends on factors such as viscosity, diffusion, and available surface area, meaning that complete evaporation may not occur.

Therefore, 2.5 mL was selected as a practical and conservative starting volume to ensure that VOC concentrations in the chamber were sufficiently high to detect and compare the removal performance of the plant and plant–microbe setups.

3.1.2. Selection of Plant and Microbe

For the pilot study, a snake plant (Sansevieria trifasciata) paired with a Bacillus microbial mixture was selected as the representative plant–microbe system. Preliminary research indicates that both the snake plant and Bacillus species possess the capability to remove VOCs, making them suitable candidates for initial testing (see Appendix 5). In addition, their ready availability at the early stage of the project enabled timely setup and iterative refinement of the experimental framework. This allowed the pilot study to focus on optimising experimental parameters and validating the setup.

3.1.3. Conduct Experiments

With the pollutant dosage preliminarily determined and all other experimental parameters standardised as shown in Figure 12, the experimental setup was first implemented for pilot testing under controlled conditions. The results obtained were then used to identify any limitations or inconsistencies in the setup, allowing for iterative refinements to improve the accuracy and reliability of subsequent experiments.

3.1.3.1. Experimental Setup

The experimental setup was designed to ensure controlled and consistent measurement of VOC removal. The entire system was enclosed within a sealed chamber using masking tape and Blu Tack to minimise air leakage, ensuring that any observed changes in VOC concentration could be attributed solely to the plant or plant–microbe system. To further control environmental conditions, the setup was covered with a trash bag to eliminate external light interference and maintain consistent lighting throughout the experiment (Figure 17).

Within the chamber, the internal configuration was arranged to optimise airflow and measurement accuracy. The plant container was positioned such that a computer fan directed airflow from the pollutant source toward the plant roots, promoting interaction between VOCs and the rhizosphere. The sensor was placed on the opposite side of the chamber to minimise direct exposure to emitted pollutants and ensure that recorded measurements reflected overall chamber conditions rather than localized concentrations (Figure 18).

3.1.3.2. Experimental Procedure

Following the configuration of the experimental setup, a standardised procedure was also established to ensure consistency across all pilot study trials. Each experiment was conducted under controlled conditions using a fixed setup, with pollutant dosage as the only varying parameter. The step-by-step procedure for each experimental run is illustrated in Table 5.

Table 5: Experimental Set Up Procedure
Step	Description	Rationale
1	Calibrate VOC sensors in ambient outdoor conditions; transfer to chamber once stable	To establish a stable and accurate baseline for VOC measurements, sensors were calibrated under ambient outdoor conditions for 24 hours, ensuring reliable detection of concentration changes during experiments. This also minimizes sensor drift and improves consistency across trials.
2	Rinse LECA to remove impurities and transplant plant into the plant container with the clean medium and fan	LECA was rinsed to remove surface contaminants that could interfere with VOC measurements. Its inert and porous structure supports airflow and microbial attachment, while the integrated fan directs VOC-laden air toward the rhizosphere to enhance pollutant–root interaction.
3	Apply 250ml of water via wash bottle to saturate the rhizosphere	A fixed volume of water ensures consistent hydration and supports microbial activity within the rhizosphere, promoting effective VOC degradation while maintaining uniform conditions across all setups.
4	Dispense the appropriate volumes of each pollutant into internal cups using a syringe; cover with cards to prevent VOC volatilization	Using precise volumes ensures controlled and repeatable pollutant dosing. Covering the pollutants prevents premature volatilization, ensuring that VOC release begins only upon initiation of the experiment.
5	Power sensors and circulation fans; seal all orifices with blu tack	Activating sensors and fans ensures proper air circulation and real-time monitoring, while sealing orifices minimizes air leakage, maintaining a closed system so that VOC changes can be attributed solely to the plant system.
6	Remove pollutant covers and immediately secure the chamber lid with masking tape	Initiating the experiment by removing covers allows for a synchronized start time for VOC volatilization. Sealing the lid immediately prevents the escape of initial gas concentrations, ensuring a sealed environment for accurate rate-of-change analysis.
7	Cover chamber with trash bag and turn on lamps	Blocking external light ensures consistent lighting conditions, while the controlled artificial light source standardizes plant physiological activity, reducing variability in VOC removal performance.

3.1.4. Results and Discussions

The pilot study experiment set up with 2.5 mL of each pollutant was allowed to run for 24h to obtain the results in Figure 19.

_{Figure 19: VOC Concentration Over Time for 2.5ml of Each Pollutant - Interactive
Visualisation}

Table 6a: Maximum VOC Level for 2.5ml of each pollutant
Maximum VOC Level
Control	Plant	Plant + Microbe
37216.6 ppb	1798.2 ppb (↓ 95%)	1596.8 ppb (↓ 96%)

Table 6b: Time Taken to Reach Steady State for 2.5ml of each pollutant
Time Taken to Reach Steady State
Plant	Plant + Microbe
15h	10h (↓ 33%)

Following the 24-hour pilot study conducted using 2.5 mL of each pollutant, VOC measurements from the control setup indicated that the selected pollutant sources generated sufficiently high concentrations of VOCs within the chamber. The combined profile of the three pollutant source also aligns with the expected VOC release profile in Figure 14 where there is an initial rapid increase due to the low-viscosity shoe polish, followed by a sustained high concentration maintained by the moderate release of the paint and the gradual release from the high-viscosity glue.

Examination of the VOC decay profiles revealed that both the plant-only and plant–microbe setups contributed to VOC removal, consistent with the expected role of phytoremediation. The plant–microbe setup demonstrated a slightly higher removal efficiency compared to plants alone, achieving a 96% reduction in VOC levels, versus 95% for the plant-only setup.

Additionally, the time required for the plant–microbe system to reach steady state was approximately 33% shorter than that of the plant-only system. This observation aligns with theoretical expectations that microbial activity in the rhizosphere enhances VOC degradation: microbes can metabolize compounds absorbed by plant roots, thereby accelerating the overall removal process. The marginally higher removal efficiency and faster stabilization observed in the plant–microbe setup provide experimental evidence supporting the synergistic effect of plant–microbe interactions in VOC mitigation.

3.1.5. Experimental Improvements and Methodology Refinement

During this process, several important considerations for improving the setup and methodology of analysis were identified, including adjustments to pollutant volume, accounting for background VOCs emitted by the chamber material, and ensuring consistent sensor calibration and placement.

Firstly, when 2.5 mL of each pollutant was introduced into the chamber, the VOC levels reached 37,216.6 ppb, exceeding the sensor’s upper limit of 36,000 ppb (AWAIR support, 2023). After testing varying pollutant volumes, we determined that 1 mL of each pollutant produced a VOC concentration of 29,103.4 ppb—well below the sensor ceiling—while still generating clear differences in VOC removal between setups (Figure 20). This dosage was chosen to provide a safety margin, ensuring that any residual VOCs remaining between experiments would not push concentrations near the sensor limit, which could compromise measurement accuracy. Hence, 1 mL of each pollutant was selected as the standardized dosage for subsequent experiments.

_{Figure 20: VOC Concentration Over Time for 1ml of Each Pollutant - Interactive
Visualisation}

Secondly, a control test conducted using an empty chamber showed a gradual increase in VOC levels even in the absence of added pollutants (Figure 19 - 'empty'). This is likely due to the emission of VOCs from the polypropylene chamber material, which can occur when the plastic is exposed to light (Kang et al., 2020). It is important to note that the VOCs released from polypropylene are not among the FBTEX compounds targeted in this study. However, because the sensors used measure VOC concentration without differentiating between compound types, these background emissions may still influence the readings. To minimize this effect, each experimental condition will be repeated twice, and the average values will be used for analysis. Since these background VOCs are not chemically related to the FBTEX compounds of interest, they are not expected to affect the comparative assessment of plant–microbe pairings.

Lastly, we observed slight differences in the starting VOC levels across different setups. These variations are likely caused by differences in the time taken to seal the chamber or by residual VOCs that remained despite ventilation between runs. As a result, direct comparison of absolute VOC readings between experiments may not provide an accurate representation of removal performance. To ensure fair comparison, all future analyses will instead be based on the relative change in VOC concentration from the starting point of each experiment.

3.1.6. Conclusion of Pilot Study

The pilot study successfully fulfilled its primary objectives by transforming an initial experimental framework into a standardized, high-precision methodology. The following table shows the key outcomes of this phase.

Table 7: Conclusion of Pilot Study
Objective	Conclusion
To evaluate the feasibility of the experimental design	Experimental set up following Figure 17 is feasible
To determine the appropriate amount of pollutants to be used	1 ml of each pollutant will be used in the subsequent experiments
To establish the optimal duration for each experimental run	24 h is sufficient and a conservative duration for all experiments to reach steady state
To identify any potential issues or gaps in the setup that may affect the accuracy or reliability of subsequent experiments	Experiments will be repeated and the average values will be used for analysis Analysis will be based on the change in VOC concentration from the initial VOC concentration
To determine the appropriate procedure to setting up experiments	Experiments will be set up following steps in Table 5

3.2. Proof of Concept (Pearlyn)

The Proof of Concept phase aims to validate the underlying hypothesis that different plants and plant–microbe combinations exhibit varying capacities for VOC removal within a controlled environment. Building on the refined experimental setup and procedures established during the pilot study, this stage focuses on systematically evaluating the performance of selected plant systems under standardised conditions.

The objectives of the Proof of Concept are:

To verify that VOC removal performance varies across different plant species
To validate the idea that microbial innoculation enhances VOC removal efficiency of plants
To evaluate performance criteria on how plant–microbe pairings improve VOC removal

3.2.1. Selection of Plants and Microbe

Four plant species were shortlisted for testing in this proof-of-concept phases: Sansevieria trifasciata (Snake Plant), Epipremnum aureum (Money Plant), Zamioculcas zamiifolia (ZZ Plant), and Dracaena fragrans (Dragon Plant).

The selection of these species was guided by two primary considerations. First, consistent VOC removal performance — these plants were repeatedly identified in both scientific literature and commercial reports as effective species for VOC mitigation (see Appendix 5), thereby supporting their inclusion based on validated evidence. Second, ease of maintenance and resilience — the shortlisted plants are hardy, low-maintenance species capable of thriving under typical indoor environmental conditions. Their adaptability to Lightweight Expanded Clay Aggregate (LECA), our chosen growth medium as it provides space for air to flow to the plant’s roots, further reinforces their practicality for real-world applications.

In addition to plant selection, the choice of microbial strain was guided by preliminary research into microbes with demonstrated VOC degradation capabilities. Bacillus subtilis (Figure 22) was identified as a suitable candidate due to its well-documented ability to metabolise VOCs. It was selected over other microbial candidates for several practical and experimental reasons. Firstly, Bacillus subtilis is a non-pathogenic, well-characterised, and widely studied species, making it safe and reliable for use in controlled indoor experiments (Stülke et al., 2023). Secondly, it is known for its robustness and ability to form endospores, allowing it to survive under fluctuating environmental conditions such as changes in moisture, nutrient availability, and airflow within the chamber (Errington, 2003).

3.2.2. Experimental Matrix and Set Up

As illustrated in Figure 23, the experimental design systematically varied both plant species and microbial presence. Specifically, for each plant species, experiments were conducted under two conditions: without microbes (plant-only control) and with Bacillus subtilis. In parallel, a microbe-only setup was also included to isolate its individual contribution.

This experimental approach yielded three primary insights: (i) it allows for comparison between different plant species, (ii) it quantified the performance improvements directly attributable to microbial inoculation across different plant species, and (iii) it validated the inherent effects of microbial degradation through the use of 'microbe-only' and ‘plant-only’ controls. By fixing the plant species while varying microbial presence, the isolated contribution of Bacillus subtilis to VOC removal effectiveness was clearly determined. Furthermore, evaluating these combined systems under standardized conditions allowed for a direct comparison across all configurations, facilitating a rigorous assessment of synergistic effects while minimizing the influence of external variables.

The experimental setup for the proof of concept follows the same configuration as established in the pilot study. For the ‘microbe-only’ setup, no plant was included in the system and instead, the water in the growth medium was replaced with a Bacillus subtilis solution, while all other components and experiment procedure were kept identical (Figure 24) to ensure that observed VOC reductions could be attributed solely to microbial metabolic activity rather than external environmental fluctuations or plant-driven uptake mechanisms.

3.2.3. Results and Discussion and Methodology Refinement

3.2.3.1. Plant Only Performance

The results in Figure 25 showcases the VOC removal performance results obtained for the different plant species from the proof-of-concept experiments.

Based on the trends observed in the VOC concentration profiles, clear differences can be identified across the experimental setups in terms of three key performance metrics: peak VOC change, time taken to reach steady state, and steady-state VOC concentration. These observed variations validate that different plant species exhibit distinct VOC removal performances under controlled conditions. The definitions and significance of each metric are summarised in Table 8.

Table 8: Description of Performance Metrics
Metric	Description
Maximum Change in VOC (ΔVOC_max)	ΔVOC_max reflects the system’s ability to limit VOC accumulation during pollutant emission. A lower ΔVOC_max value indicates more effective suppression of VOC build-up and, therefore, better system performance.
Time taken to reach steady state (t_plateau)	t_plateau is defined as the point at which VOC levels begin to plateau and remain relatively constant over time. This point is identified when the gradient of the concentration curve approaches zero. It reflects how quickly the system is able to reach equilibrium, with a shorter t_plateau indicating faster stabilisation and, therefore, better performance.
Steady State Change in VOC level (ΔVOC_plateau)	ΔVOC_plateau is defined as the average of ΔVOC values from the point at which the concentration first stabilises (i.e. at t_plateau) until the end of the experiment. This metric reflects the residual VOC level remaining in the system, indicating the extent to which VOCs are not fully removed relative to the initial concentration.

Maximum Change in VOC Level

Table 9: Highest VOC Level for Plant only Experiments in Proof of Concept
Maximum VOC Level (ΔVOC_Max)
Snake Plant	Money Plant	ZZ Plant	Dragon Plant
993.4 ppb	762.2 ppb	1368.6 ppb	868.8 ppb

As seen in Table 9, variations in peak VOC concentrations were observed across the four plant species, with the ZZ Plant (1368.6 ppb) exhibiting the highest accumulation, followed by the Snake Plant (993.4 ppb), Dragon Plant (868.8 ppb), and Money Plant (762.2 ppb). These differences can be attributed to inherent physiological and morphological variations between plant species, which influence their VOC uptake mechanisms. Factors such as stomatal density, cuticular properties, and transpiration rates affect the rate at which VOCs are absorbed and transported within the plant system (Wolverton et al., 1989; Yang et al., 2009).

In particular, leaf surface area plays a critical role, as larger surface areas provide greater contact with air, facilitating higher rates of VOC diffusion and uptake through stomata and the cuticle. Studies have shown that VOC removal efficiency is strongly correlated with leaf area and associated gas exchange processes, as these govern the extent of pollutant absorption and subsequent metabolic degradation (Yang et al., 2009). Consequently, differences in plant size and foliage density can significantly influence the observed VOC concentrations, independent of the plant’s intrinsic removal capability.

Thus, to enable more accurate comparison of plant performance, VOC removal was normalised by leaf area. Leaf area measurements were obtained using a leaf-scanning application, providing a quantitative basis to account for differences in plant size. For plants with flexible foliage that could be laid flat, leaves were directly scanned to determine their total surface area (Figure 26a). For plants that could not be flattened without damage or distortion, the area of a representative leaf was first measured and then multiplied by the total number of leaves to estimate the overall leaf area (Figure 26b). This method provides a consistent and practical approach for approximating leaf area across different plant morphologies, ensuring that comparisons reflect intrinsic plant performance rather than differences in size.

Flattenable foliage scan — Figure 26a: Leaf area measurement for plants with flattenable foliage

Non-flattenable foliage estimation — Figure 26b: Leaf area estimation for plants with non-flattenable foliage

Time Taken to Reach Steady State

Table 10: Time Taken to Reach Steady State for Plant only Experiments in Proof of Concept
Time Taken to Reach Steady State (t_plateau)
Snake Plant	Money Plant	ZZ Plant	Dragon Plant
~13h	~7h	~6h	~5h

Table 10 presents the time taken for each plant species to reach steady-state VOC concentrations in the plant-only experiments. Notable differences were observed, with the Dragon Plant (~5 h) and ZZ Plant (~6 h) reaching steady state more rapidly, followed by the Money Plant (~7 h) and Snake Plant (~13 h), which exhibited a significantly longer stabilization period. These variations can be attributed to the differences in leaf morphology and internal transport processes may affect how quickly absorbed VOCs are metabolized or redistributed within the plant system, further contributing to the observed variation in plateau times (Yang et al., 2009).

To further improve the accuracy and consistency of the analysis, the time taken to reach plateau was determined using OriginPro (Figure 27). Specifically, curve-fitting and gradient analysis functions within the software were employed to objectively identify the point at which the rate of change in VOC concentration approached zero, thereby reducing subjectivity associated with manual estimation and enhancing the reliability of the results.

Steady State Change in VOC Level

Table 11: Steady State Change in VOC Level for Plant only Experiments in Proof of Concept
Steady State Change in VOC Level (ΔVOC_plateau)
Snake Plant	Money Plant	ZZ Plant	Dragon Plant
-33.85 ppb	-157.44 ppb	-7.44 ppb	-8.47 ppb

Table 11 presents the steady-state change in VOC levels for the plant-only experiments in the Proof of Concept phase. Differences were observed across the four plant species, with the Money Plant (-157.44 ppb) demonstrating the greatest reduction in VOC concentration at steady state, followed by the Snake Plant (-33.85 ppb), Dragon Plant (-8.47 ppb), and ZZ Plant (-7.44 ppb). These results indicate varying capacities among the plant species to sustain VOC removal over time after reaching equilibrium conditions.

These observations further support the first objective of the Proof of Concept, confirming that VOC removal performance differs across plant species and reinforcing the importance of plant selection in subsequent evaluations.

3.2.3.2. Microbe Only Performance

The results in Figure 28 present the VOC removal performance results obtained for the Bacillus Subtilis only from the proof-of-concept experiments.

Table 12: Maximum VOC Level for Bacillus Subtilis only
Maximum VOC Level (ΔVOC_Max)
Pollutant only	Bacillus Subtilis only
28,755 ppb	1,798.2 ppb (↓ 94%)

The results presented in Figure 28 demonstrate clear evidence of microbial contribution to VOC degradation, as supported by the data in Table 12. In the control condition, VOC concentration increases rapidly and reaches a high maximum level of 28,755 ppb, indicating continuous accumulation in the absence of any removal mechanism. In contrast, the Bacillus subtilis-only system exhibits a markedly different trend: after a slight initial increase, VOC levels decrease at a significantly lower maximum of 1,798.2 ppb, corresponding to a 94% reduction relative to the control. The substantial reduction and stabilisation of VOC concentrations in the absence of plants highlight the inherent degradation capability of Bacillus subtilis, supporting its role as a key contributor to enhanced VOC removal in plant–microbe systems.

3.2.3.3. Plant-Microbe Performance

The results in Figure 29 present the VOC removal performance results obtained for the plant–microbe systems from the proof-of-concept experiments.

Maximum Change in VOC Level

Table 13: Results of Maximum Change in VOC Level for Proof of Concept
Maximum Change in VOC Level (ΔVOC_Max)
	Snake Plant	Money Plant	ZZ Plant	Dragon Plant
No Microbe	993.4 ppb	762.2 ppb	1368.6 ppb	868.8 ppb
Bacillus Subtilis	273.6 ppb (↓ 72.5%)	567.4 ppb (↓ 25.6%)	799 ppb (↓ 41.6%)	937 ppb (↑ 7.8%)

The maximum change in VOC levels varies significantly across different plant–microbe pairings, thereby supporting the hypothesis that microbial inoculation influences the initial pollutant surge.

The introduction of Bacillus subtilis generally led to a reduction in peak VOC levels in three out of the four plant species, although the extent of improvement differed. The Snake Plant showed the most substantial reduction, with a 72.5% decrease in peak VOC, followed by the ZZ Plant (41.6%) and the Money Plant (25.6%). In contrast, the Dragon Plant deviated from this trend, exhibiting a 7.8% increase in peak VOC upon microbial addition.

These findings suggest that while microbial inoculation can enhance the system’s ability to mitigate initial VOC accumulation, the magnitude—and even direction—of this effect is strongly dependent on the specific interaction between the plant species and the microbial strain.

Time Taken to Reach Steady State

Table 14: Results of Time Taken to Reach Steady State for Proof of Concept
Time Taken to Reach Steady State (t_Plateau)
	Snake Plant	Money Plant	ZZ Plant	Dragon Plant
No Microbe	~13h	~7h	~6h	~5h
Bacillus Subtilis	~3.9h (↓ 70.0%)	~3.7h (↓ 47.1%)	~3.5h (↓ 41.7%)	~3.7h (↓ 26.0%)

A clear and consistent trend emerges with the introduction of Bacillus subtilis. The time to reach steady state for all plant-microbe set ups converges to a narrow range of approximately 3.5 to 3.9 hours. This corresponds to a substantial reduction in stabilisation time, with improvements of up to 70.0% for the Snake Plant and 47.1% for the Money Plant.

The convergence of stabilisation times to an approximately 4-hour window, regardless of plant species, suggests that the rate of initial VOC degradation in these systems is largely governed by microbial metabolic activity rather than the slower plant-driven processes or plant–microbe synergy. In contrast, the Dragon Plant exhibited a comparatively modest improvement of 26%, reinforcing earlier observations that this particular plant–microbe pairing is less effective for rapid VOC clearance.

Steady State Change in VOC Level

Table 15: Results of Steady State Change in VOC Level for Proof of Concept
Steady State Change in VOC Level (ΔVOC_Plateau)
	Snake Plant	Money Plant	ZZ Plant	Dragon Plant
No Microbe	-33.85 ppb	-157.44 ppb	-7.44 ppb	-8.47 ppb
Bacillus Subtilis	-54.59 ppb (↓ 61.3%)	-126.0 ppb (↑ 20.0%)	-19.5 ppb (↓ 162.1%)	-227.7 ppb (↓ 2588.3%)

The results in Table 15 indicate that steady-state VOC concentrations vary significantly across pairings, revealing that microbial inoculation is highly host-dependent. In the majority of configurations, the addition of Bacillus subtilis significantly enhanced the system's baseline. Specifically, the Snake Plant improved from -33.85 ppb to -54.59 ppb, and the ZZ Plant shifted from -7.44 ppb to -19.5 ppb. Most notably, the Dragon Plant moved from a marginal reduction of -8.47 ppb to a substantial baseline reduction of -227.7 ppb, demonstrating a strong synergistic "anchoring" effect that prevents VOC rebound.

In contrast, the Money Plant provided a unique insight into potential biological interference. While the plant-only setup maintained the strongest baseline reduction at -157.44 ppb, the introduction of Bacillus subtilis weakened this performance, resulting in a higher steady-state concentration of -126.0 ppb. This suggests that microbial activity may have competed with the Money Plant's natural uptake mechanisms or disrupted rhizosphere gas exchange.

Overall, these results indicate that plant–microbe pairings can enhance VOC removal, but the magnitude and direction of the effect are highly dependent on the specific species involved. The three performance metrics each provide distinct and complementary insights into system behaviour. Evaluating these metrics collectively is therefore essential, as improvements in one aspect do not necessarily translate to overall system performance. Together, they enable a more comprehensive assessment of plant–microbe interactions, highlighting the importance of considering both transient and steady-state dynamics when selecting optimal pairings for VOC mitigation.

3.2.4. Conclusion of Proof of Concept

The Proof of Concept phase first demonstrates that different plant species exhibit varying capacities for VOC removal under controlled conditions, establishing that plant selection plays a critical role in overall performance. Building on this, the results further show that plant–microbe interactions exist and vary in their capacity for VOC removal under controlled conditions, highlighting the potential for synergistic effects. The identification and validation of the most effective plant–microbe pairings will be carried out in subsequent phases, where top-performing plants (Phase 2) and selected microbial candidates (Phase 3) will be combined and evaluated through targeted pairing experiments (Phase 4).

The key conclusions derived from the proof of concept study are summarised in Table 16.

Table 16: Conclusion of Proof of Concept Study
Objective	Conclusion
To verify that VOC removal performance varies across different plant species	Distinct differences in VOC removal performance were observed across plant species, confirming that plant selection significantly influences VOC reduction efficiency.
To validate the idea that microbial innoculation enhances VOC removal efficiency of plants	Measurable differences were observed across all plant–microbe systems, confirming that pairing improves VOC removal performance. Different pairings also exhibited improvements to varying extents, confirming that the synergistic effect is highly dependent on the specific plant–microbe combination.
To evaluate performance criteria on how plant–microbe pairings improve VOC removal	In subsequent experiments, results will be analysed based on 3 criteria: Maximum VOC level Time taken to reach steady state Steady state change in VOC level

4. Phase 2: Plant Selection (Pearlyn)

The second phase of the project focuses on identifying indoor plant species with the highest potential for VOC removal in enclosed environments. A systematic research approach was employed to shortlist the top-performing plants (Figure 30), which were then selected for further evaluation in the matrix experiments conducted in the final phase. Prioritising the top-performing plants ensures that the subsequent plant–microbe pairing experiments build upon plant systems that already demonstrate strong intrinsic VOC removal capabilities, thereby allowing the study to evaluate whether microbial inoculation can further enhance and optimise their performance. This approach increases the likelihood of observing meaningful synergistic effects, rather than improvements that are limited by the baseline inefficiency of lower-performing species.

To ensure a focused yet rigorous evaluation in the final phase, the study will narrow the selection to the top two performing plant species identified from the screening process. This decision is guided not only by feasibility considerations, but also by experimental design requirements. The subsequent phase employs a matrix-based approach to investigate plant–microbe interactions, where the number of experimental conditions increases multiplicatively with each additional plant species. Within the constraints of the project timeline, expanding the number of plants would necessitate a reduction in replication, thereby compromising statistical robustness and the ability to confidently distinguish treatment effects from experimental variability. By limiting the study to two plants, sufficient replicates can be maintained to ensure reproducibility and reliability of results. At the same time, the inclusion of more than one species preserves biological variability, allowing the study to assess whether observed interactions are consistent across different plant systems or species-dependent. This approach therefore balances experimental tractability with scientific rigor, ensuring that the final phase yields both statistically robust and meaningfully comparative insights.

4.1. Plant Research

The plant selection process began with an industrial review to identify species currently used in commercially available plant-based air purification systems. These industry findings provided practical insights into plant performance under real-world indoor conditions. Building upon this foundation, the review was expanded to include secondary research from academic and organizational studies, ensuring a comprehensive understanding of documented VOC removal capabilities across a wide range of species. From this combined body of research, a comparative analysis was conducted to evaluate reported VOC removal efficiencies. Plant species that consistently demonstrated lower performance or limited removal capability were excluded, refining the pool to those with higher and more reliable purification potential.

Following this performance-based screening, the remaining candidates were further evaluated according to a defined set of selection criteria — availability, suitability to Singapore’s indoor climate, and ease of maintenance. These criteria ensure that the shortlisted species are not only effective in VOC removal but also practical and sustainable for long-term indoor application. The following diagram (Figure 31) summarizes this selection framework.

Based on this systematic evaluation, a total of 20 plant species out of the 100 plants found from literature research (See Appendix 6) were deemed suitable and were selected to proceed to the experimental phase in phase 2, where their VOC removal performance would be assessed to identify the top-performing plants. Notably, four species from this subset had already been selected for use in the earlier Pilot Study and Proof of Concept (Sections 3.1.2 and 3.2.1).

4.2. Experiments for Top Performing Plants

The experimental setup followed that established during the pilot study (Figure 17). To improve the robustness of the results, each experiment was conducted in duplicate, and the average performance was reported. This approach reduces the impact of experimental variability and enhances the reliability of the observed trends in VOC removal efficiency.

Furthermore, to ensure an accurate and equitable comparison between species of varying sizes, the analysis incorporates per-unit leaf area normalization as detailed in Section 3.2.3.1. to eliminate subjectivity in identifying system equilibrium, OriginPro software was employed for high-precision gradient analysis. As demonstrated in Section 3.2.3.1., this computational approach objectively determines the time taken to reach plateau by identifying the exact point where the rate of change in VOC concentration approaches zero, ensuring the reliability of the performance metrics used to evaluate plant-microbe synergy.

4.3. Results and Discussion

The VOC removal performance of the 20 selected plant species was evaluated over a series of 24-hour cycles. Figure 33 provides a comparative visualization of the VOC concentration decay curves, illustrating the distinct removal kinetics inherent to each species.

_{Figure 33: VOC Concentration Removal Comparison between Plants - Interactive
Visualisation}

Based on the three key performance metrics defined in Section 3.2.3.1., raw experimental results were first obtained for each of the 20 selected plant species. For each criterion, the plants were then ranked from best to worst, where higher values correspond to better performance.

Following this, scores were assigned using min–max normalization to rescale the values onto a uniform range of 0 to 10 (Eqn 1). This approach ensures that the best-performing plant for each criterion receives the highest score, while the poorest-performing plant receives the lowest score, with all others scaled proportionally in between. Min–max normalization was selected because it preserves the relative differences between plants without disproportionately penalizing those that perform only slightly worse than higher-ranked plants, ensuring a fairer and more consistent comparison across criteria. The normalized scores for the three criteria were subsequently summed to obtain a composite score for each plant (Eqn 2), providing an overall measure of VOC removal performance.

Min-Max Normalization:

Score = ( (x_max - x) / (x_max - x_min) ) × 10

— eqn (1)

Total Performance Score:

Total Score= Score_{Max VOC} + Score_Time + Score_{Steady
State}

— eqn (2)

This enables a systematic and quantitative ranking of plant species, as presented in Table 17, with the detailed scores and tabulated results for each plant provided in Appendix 7.

Table 17: Ranking of VOC Removal Efficiencies of Selected Plants
Rank	Plant	Total Score
1	Pteris argyaea (Silver Brake Fern)	28.8
2	Chlorophytum comosum (Spider Plant)	27.9
3	Begonia maculata (Polka dot begonia)	27.7
4	Philodendron martianum	27.5
5	Spathiphyllum wallisi (Peace Lily)	27.4
6	Polyscias fruticosa (Ming aralia)	27.3
7	Anthurium andraeanum (Flamingo Lily)	26.6
8	Dracaena fragrans (Dragon plant)	26.2
9	Nematanthus glabra (Goldfish plant)	25.8
10	Sansevieria trifasciata / Dracaena trifasciata (Snake plant)	25.8
11	Fittonia albivenis (Nerve plant)	25.4
12	Hemigraphis alternata / Strobilanthes alternata / Hemigraphis colorata (Red Ivy, Red Flame Ivy, Metal Leaf)	25.3
13	Kalanchoë blossfeldiana (Flaming Katy)	25.1
14	Philodendron oxycardium / Philodendron hederaceum	24.1
15	Zamioculcas zamiifolia (ZZ plant)	23.5
16	Dracaena reflexa / Dracaena angustifolia (Red-Edged Dracaena)	23.4
17	Aloe aristata (Lace Aloe)	21.1
18	Aglaonema commutatum (Chinese evergreen / Philippine evergreen)	14.0
19	Epipremnum aureum (Money plant / Pothos / Golden Pothos / Devil's Ivy)	9.9
20	Plectranthus tomentosus / Coleus hadiensis / Plectranthus hadiensis (Vicks Plant)	9.2

4.4. Plant Selection Conclusion

Based on the composite scoring framework, Pteris argyraea (Silver Brake Fern) and Chlorophytum comosum (Spider Plant) ranked as the top two performing species, achieving the highest total scores of 28.8 and 27.9, respectively (Table 17). Their stronger performance reflects consistently high normalized scores across all three evaluation criteria. These two species are therefore selected for Phase 4, where they will be incorporated into plant–microbe pairing experiments to evaluate how microbial inoculation influences VOC removal performance. By focusing on the top-performing plants, the subsequent phase is able to assess whether their inherent VOC removal capabilities can be further enhanced through synergistic interactions with microbial systems, while enabling controlled and comparable evaluation across experimental conditions.

Notably, while several other species such as Begonia maculata and Philodendron martianum also demonstrated comparable overall performance, the marginal score differences highlight that Pteris argyraea and Chlorophytum comosum exhibit the most balanced and consistently strong performance across all metrics rather than excelling in only one. This distinction is important, as it suggests that their superior ranking is not driven by a single dominant characteristic but by robust performance across multiple aspects of VOC removal.

5. Phase 3: Microbe Selection (Tse Hui)

5.1 Microbe Research

In addition to selecting plants, microbes must also be shortlisted for experimentation to ultimately identify an optimal plant-microbe pairing. The objective of the Microbe Selection phase is to shortlist microbes capable of effectively removing VOCs, which will then be used in the subsequent experimental phase. The process begins with a literature review of existing research and commercial studies on plant-microbe systems, specifically examining studies on:

Microbes that enhance a plant's ability to remove VOCs
Microbes that remove VOCs independently without plants

To further narrow the pool of candidates, a comparative analysis will be conducted on studies where multiple isolated microbial strains were evaluated for VOC removal. Microbes that demonstrated one of the highest removal effectiveness among those compared will be retained for further shortlisting.

From the resulting shortlist, candidates are further filtered against a set of predefined criteria (Table 18). Only microbes that satisfy all criteria will be considered for procurement (Table 19).

Table 18: Selection Criteria

Criterion	Requirement	Remarks
Availability	Commercially available from a local seller in Singapore	Lab-grade strains are typically restricted to institutional purchase. As this project targets practical, everyday applications of plant-microbe air purification, commercially available microbial products (e.g. microbial fertilisers or biostimulants) are preferred over lab-grade strains. All products are locally sourced to ensure procurement feasibility, with the active ingredient verified as the microbe of interest and all other ingredients confirmed to be inert (Appendix 8).
Safety	Biosafety Level 1 (BSL-1)	Microorganisms are internationally classified into four Biosafety Levels (BSL-1 to BSL-4) based on their potential risk to human health and the environment. BSL-1 microorganisms are those that pose little to no threat to individuals and require only standard laboratory precautions where no specialised containment equipment such as fume hoods or biosafety cabinets is needed. This makes them suitable for handling in non-specialised settings. The bio-safety level of the microbes is verified via online research. In addition, additional safety precautions are also evaluated through the Safety Data Sheet (SDS).
Environmental Compatibility: Temperature	Optimal growth at 20°C to 25°C	This temperature range corresponds to typical indoor air-conditioned environments in Singapore, ensuring the microbes remain active within the plant-microbe system.
Environmental Compatibility: pH	pH = 6.5 to 7.5	This pH range corresponds to neutral - the typical pH of LECA balls used in the system, ensuring microbial efficacy within the growth medium.

Table 19: Shortlisted microbes and their justification for being chosen or rejected

(Green: criteria met. Red: criteria unmet. Yellow: criteria partially met due to conflicting evidence)

Microbes	Existing literature suggesting microbe potential for VOC degradation	Availability	Biosafety Risk	Optimal Temperature Range	Optimal pH Range
Arthrobacter aurescens	Based on a study by Huang et al. (2012), microbes were isolated and characterised from Golden Pothos. These microbes were subjected to formaldehyde and the strain that was identified to remove the most amount of formaldehyde (86.2%) is Arthrobacter aurescens TC1.	Local seller unavailable	BSL-1	20°C – 30°C. Based on Comi & Cantoni (2011), many Arthrobacters are mesophilic with optimum temperature of 20°C to 30°C. Huang et al. (2012) isolated Arthrobacter aurescens at 30°C.	7.0 In Silva et al. (2015), Arthrobacter aurescens TC1 were grown in a medium at pH=7
Azotobacter chroococcum	In Thavasi et al. (2006), Azotobacter chroococcum could biodegrade crude oil which contains many volatile and carbon compounds, including xylene.	Local seller unavailable	BSL-1	30°C – 40°C (Basnett et al., 2024)	7.0 (Basnett et al., 2024)
Bacillus amyloliquefaciens	Wongbunmak et al. (2020) found Bacillus amyloliquefaciens subsp. plantarum strain W1 to be able to degrade BTEX. In another study, another specific strain of Bacillus amyloliquefaciens (XF-1) can degrade formaldehyde from cooking oil fume condensates (Han et al., 2020).	Locally available with high-purity formulation	BSL-1	30°C – 40°C (Ngalimat et al., 2021)	6.3 (Liu et al., 2025)
Bacillus cereus	Bacillus cereus can degrade toluene (Lee et al., 2013), ethylbenzene (Daudzai et al., 2018) and formaldehyde (Khaksar et al., 2016).	Local seller unavailable	BSL-2 (for most strains)	30°C – 37°C (Anany et al., 2015; El-Arabi & Griffiths, 2013)	6.0 – 7.0 (El-Arabi & Griffiths, 2013)
Bacillus megaterium	Based on Dolphen et al. (2019), Bacillus megaterium can remove benzene in a biofilter. Based on Dhanya (2019), it can degrade benzene, toluene and xylene individually. Based on Taupp (2006), it can possibly biotransform formaldehyde.	Local seller unavailable	BSL-1	25 – 30°C (Novobac, n.d.) 30°C (Fadahunsi & Phebe, 2017)	7.0 – 11.0 (Park & Son, 2009) 7.0 – 7.5 (Novobac, n.d.)
Bacillus subtilis	Based on a study by Lan et al., (2020), it is one of the microbes effective in degrading benzene, toluene and xylene.	Locally available with high-purity formulation	BSL-1	25°C – 35°C (Turnbull, 1996 as cited in Loggenberg et al., 2022)	7 (Satapute et al., 2012)
Bacillus thuringiensis	From Ehmedan et al. (2021), Bacillus thuringiensis degrades crude oil which typically contains BTEX. Other studies showed that it can degrade toluene (Lee et al., 2013; Kesavan et al., 2021) and benzene (Kesavan et al., 2021).	Locally available with high-purity formulation	BSL-1	24°C – 32°C (Li et al., 2024)	6.5 – 7.5 (Handayani et al., 2025)
Methylorubrum extorquens	Various research studies (Di Maiuta et al., 2009; Vorholt et al., 2000; Hying et al., 2025) have shown that Methylorubrum extorquens is able to degrade formaldehyde. Methylorubrum bacteria are methylotrophs that assimilate single-carbon substrates as their carbon and energy source (Ardley & Green, 2023), suggesting their capability in formaldehyde degradation.	Local seller unavailable	BSL-1	30°C (Jinal et al., 2020)	Neutral (Groom & Lidstrom, 2021)
Pseudomonas fluorescens	Based on Shim et al. (2005), a co-culture of Pseudomonas fluorescens and Pseudomonas putida can degrade BTEX, suggesting the potential for pseudomonas fluorescens to degrade BTEX.	Local seller unavailable	BSL-1	25°C (Lebert et al., 1998)	7.0 (Lebert et al., 1998)
Pseudomonas mendocina	According to Ma et al. (2025), Pseudomonas mendocina is found to be able to degrade benzene, toulene and ethylbenzne while Pseudomonas putida and Pseudomonas aeruginosa are found to only degrade benzene and toulene among the 3 VOCs.	Local seller unavailable	BSL-1	30°C (Abis Encyclopedia, n.d.; Kao et al., 2005)	6 (Kao et al., 2005)
Pseudomonas putida	Pseudomonas putida appears in a number of research articles for its effectiveness in removing VOC (De Kempeneer et al., 2004; You et al., 2012; Lu et al., 2012), including FBTEX. It also possesses toluene dioxygenase (TOD) operon (De Kempeneer et al., 2004) suggesting its potential in degrading the VOC of interest.	Local seller unavailable	BSL-1	25°C (Li et al., 2010) 30°C (Fonseca et al., 2011)	7 – 8 (Rank et al., 2018)
Rhodococcus erythropolis	Based on Hidalgo et al. (2002), Rhodococcus erythropolis can remove formaldehyde in wastewater.	Local seller unavailable	BSL-1	30°C (Jayaprada et al., 2025)	7 (Jayaprada et al., 2025)
Trichoderma harzianum	Based on Daccò (2020) which studied microbes' ability in degrading engine oil (including BTEX), Trichoderma harzianum demonstrated significant BTEX reduction.	Locally available. Typically sold with other active ingredients such as Bacillus sp. and Humic acid	BSL-1	25°C (Singh et al., 2014)	6.5 (Singh et al., 2014)
Trichoderma viride	Cheng et al. (2016) passed gaseous toluene in three biofilters with various inoculation combinations, one of which included Trichoderma viride. Trichoderma viride was able to remove toluene.	Locally available with high-purity formulation	BSL-1	25°C (Belal, 2008; Singh et al., 2014)	6.5 (Belal, 2008; Singh et al., 2014)

Following this filtering process, four microbes remain as candidates: Bacillus subtilis, Bacillus thuringiensis, Pseudomonas fluorescens, and Trichoderma viride. To further verify that these four microbial strains are effective in degrading VOCs, we conducted experiments to collect primary data.

5.2 Experiments for Microbes

To evaluate the VOC degradation efficacy of the four shortlisted microbial strains, controlled experiments were conducted using a methodology similar to the plant selection phase (Section 4), with two modifications: no plants were present, and the medium, previously water, was replaced with the respective microbe solution prepared according to manufacturer instructions (Appendix 9). The same plastic container, computer fan, and LECA balls were retained. Each experiment was performed in duplicate, and the reported VOC removal trend for each strain represents the average of both runs (Figure 34).

As evident from Figure 34, all four microbial strains kept VOC levels near zero throughout the experiment, in stark contrast to the pollutants-only control. These findings show that these microbes remove VOCs effectively, thus providing experimental validation for their selection as microbe strains in this study.

6. Phase 4: Plant-Microbe Pairing Matrix (Tse Hui)

6.1. Objectives

With the two plants selected in the plant selection phase (Spider plant and Silver Brake fern) and the four microbes selected in the microbe selection phase (Bacillus subtilis, Bacillus thuringiensis, Trichoderma viride and Pseudomonas fluorescens), we proceed with the plant-microbe pairing matrix phase. The objective of this phase is to determine the most effective pairing of plant and microbe in removing VOCs.

6.2. Methodology

Each plant was paired with a microbe and subjected to the same experimental methodology to evaluate the pairing’s effectiveness in removing VOCs (Figure 35).

Each experiment was conducted twice and the performance of the system was taken as the average of the two runs. To mitigate the influence of leaf area as a confounding variable, all experimental results were normalised by leaf area, consistent with the methodology outlined in the plant selection phase (section 4).

To evaluate the most effective plant-microbe pairing for VOC removal, a two-stage decision framework was adopted (Figure 36A). The first stage serves as a pass/fail filter: each pairing is to demonstrate the capacity to reduce VOC levels to below 1000ppb, based on Singapore's indoor air quality standard for VOCs in general (SS 554:2016) (Singapore Standards, 2016). Only pairings that satisfy this criterion proceed to the second stage, where the synergistic effect of microbial inoculation is evaluated by comparing each pairing's performance against the uninoculated plant-only baseline via a number of metrics expressed as percentage change (Figure 36B).

While the three performance criteria (∆VOC_max, t_plateau, and ∆VOC_plateau) established previously are effective for shortlisting plants, none individually captures the cumulative VOC exposure experienced by an occupant over time, which is a more important measure of the overall performance of the system. ∆VOC_max only reflects the system’s ability to limit VOC accumulation during pollutant emission, t_plateau only reflects the speed at which the system reaches equilibrium and ∆VOC_plateau reflects the residual VOC level remaining in the system. Considered in isolation, these metrics provide an incomplete picture of VOC removal performance.

On the other hand, the area under the ∆VOC–time curve (AUC) addresses this limitation by integrating the combined effects of all three criteria into a single quantity that directly represents cumulative VOC exposure. A lower AUC indicates a lower total VOC exposure. AUC values are evaluated at two time horizons, AUC_t=12 and AUC_t=24, for which the definitions and rationale are detailed in Table 20.

Table 20: Additional metrics for evaluating VOC removal effectiveness

Additional Metric	Description
AUC from time = 0h to time = 12h (AUC_t=12)	Obtained via numerical integration, AUC_t=12 represents the cumulative VOC exposure an occupant is exposed to over a 12-hour period from the point of pollutant emission. The 12-hour window reflects the upper bound of realistic daily occupancy — equivalent to either a standard 8-hour workday with additional pre- and post-work time spent in the same environment, or a full 12-hour shift. A lower value indicates better performance.
AUC from time = 0h to time = 24h (AUC_t=24)	AUC_t=24 represents cumulative VOC exposure over a 24-hour period, corresponding to an entire day. This serves as a secondary metric to assess the system performance over an extended occupancy duration, particularly relevant to scenarios where an occupant remains within the same indoor environment continuously, such as during full work-from-home arrangements or overnight home occupancy. A lower value indicates better performance.

While AUC_t=12 captures performance under typical occupancy conditions, it does not reveal whether the system sustains that performance over longer durations. Conversely, AUC_t=24 alone may obscure strong short-term performance that is nonetheless relevant to the majority of occupancy scenarios. Evaluating both metrics thus ensures that the ranking is robust across the full realistic spectrum of occupancy durations, and prevents a pairing from being recommended on the basis of either short-term or long-term performance alone. AUC_t=12 is designated as the primary metric as it is applicable across both workplace and residential settings — any occupant, whether at work or at home, may plausibly experience up to 12 hours of continuous exposure in the same location. AUC_t=24, by contrast, is most relevant in a residential context where the occupant does not leave the home, making it a scenario-specific secondary check rather than a universally applicable criterion.

To evaluate synergistic effects, each metric is expressed as a percentage change relative to the plant-only baseline, calculated as:

\text{\% change} = \frac{\text{Pairing value - Baseline value}}{|\text{Baseline value}|} \times 100

where a negative percentage change indicates improvement over the plant-only baseline and a positive value indicates deterioration. The absolute value of the baseline is used for the denominator to preserve the sign of the result, ensuring that the direction of change is not nullified when the baseline is negative. ∆VOC_plateau is excluded from this analysis despite being identified in the pre-experiment phase (Section 2). This is because when the baseline value approaches zero, dividing by a near-zero denominator artificially inflates the percentage change, causing the resulting percentage change values to no longer reflect actual biological differences between pairings but rather reflect noise.

Pairings are ranked from most to least negative AUC_t=12 value, where a more negative value indicates greater improvement and therefore a higher ranking. Pairings achieving a negative AUC_t=12 value are deemed synergistically beneficial under typical occupancy conditions and are subsequently evaluated under AUC_t=24 to corroborate the ranking. If the AUC_t=24 ranking is consistent with AUC_t=12, the result is considered robust. However, any discrepancies between the two rankings are noted as caveats, and the evaluation proceeds to the supplementary metrics. Pairings with a non-negative AUC_t=12 value are not recommended as the best pairing, but are assessed further under AUC_t=24. If AUC_t=24 becomes negative, the pairing is not outright dismissed but caveats are noted. If AUC_t=24 remains non-negative, it further reinforces that the addition of microbes worsens the performance and thus the pairing continues to be excluded from consideration as the best pairing. For pairings that demonstrate improvement via the primary metric (AUC_t=12) or secondary metric (AUC_t=24), the supplementary metrics ∆VOC_max and t_plateau are subsequently examined, with caveats noted for any metric that does not show improvement. The best pairing is thus determined with all relevant caveats considered (Figure 36A).

6.3. Results and Discussion

6.3.1 Results

Comparing the change in VOC across all experimental runs against the pollutant emission control (Figure 37), all plant-microbe pairings demonstrated substantial VOC removal relative to the pollutant emission control. ∆VOC_plateau for all pairings reach below 1000ppb. Therefore, all pairings thus satisfy the first stage and proceed to second-stage evaluation.

Using the experimental data for each pairing (Appendix 10), the heatmap (Figure 38) presents the percentage change in each metric relative to the plant-only baseline. A heatmap was used as it enables simultaneous comparison across multiple pairings and metrics within a single display, with colour encoding distinguishing synergistic improvements (green) from performance deterioration (red) relative to the plant-only baseline. This allows rapid identification of promising pairings for further analysis.

Inspecting the heatmap via the primary metric, only two pairings achieve a negative AUC_t=12 value and are thus classified as synergistically beneficial: Silver Brake fern + Bacillus subtilis and Spider plant + Trichoderma viride, ranking first and second respectively. All remaining pairings exhibit positive AUC_t=12 values and are therefore not recommended as the best pairing under standard occupancy conditions, though they are assessed further under AUC_t=24. Notably, Silver Brake Fern + Bacillus thuringiensis achieves a negative AUC_t=24 value, suggesting some synergistic benefit over extended occupancy. This observation suggests that Bacillus species are generally synergistically beneficial when paired with Silver Brake Fern, though the effect is more pronounced with Bacillus subtilis.

6.3.2 Silver Brake Fern + Bacillus Species Analysis

Silver Brake fern + Bacillus subtilis showed the greatest improvement in AUC_t=12, with a 73.41% reduction relative to the plant-only baseline, being ranked as first. This is further corroborated by the secondary metric, with AUC_t=24 decreasing by 185.88%, confirming that the synergistic effect is not only present but becomes more pronounced over extended occupancy durations. The supplementary metrics are largely consistent with this outcome: the pairing achieved a 24.37% reduction in ∆VOC_max, while there is 110.26% increase in t_plateau.

In contrast, Silver Brake fern + Bacillus thuringiensis initially performed poorly under the primary metric (AUC_t=12 worsening by 54.69%) despite achieving a 19.86% improvement in ∆VOC_max. However, it demonstrated substantial long-term improvement, with AUC_t=24 improving by 55.69%.

Both observations could be explained by the plateau behaviour (Figure 39). In both pairings, ∆VOC eventually crosses into the negative territory — indicating that the system has removed VOCs to below the initial pre-experiment level — and continues to decrease until stabilising at a sustained negative plateau. In contrast, the plant-only baseline briefly dips below zero before recovering and drifting slightly upward over time. This means that for both plant-microbe pairings, the graphs continue to accumulate negative area under the curve while the plant-only baseline trends upward towards the positive territory, explaining why AUC_t=24 improves more substantially than AUC_t=12 in both pairings.

The distinction between the two pairings lies in the timing of this zero-crossing. For Bacillus subtilis, ∆VOC crosses zero earlier (~3.5h), meaning the negative area begins accumulating within the first 12 hours, benefiting both AUC_t=12 and AUC_t=24. For Bacillus thuringiensis, the crossing occurs much later (~7.2h), so the positive area accumulated in the early hours outweighs the brief negative region within the 12-hour window, resulting in a positive AUC_t=12. However, as the negative plateau is sustained beyond this point, sufficient negative area accumulates over 24 hours to yield a meaningful AUC_t=24 improvement. In addition, the unfavourable t_plateau observed in Silver Brake fern + Bacillus subtilis is outweighed by the improvements in both ∆VOC_max and the AUC metrics, with the latter carrying greater weight in determining the best pairing as mentioned in the methodology (Section 6.2).

The general observation that Bacillus microbes are synergistically beneficial with Silver Brake fern may be attributed to the inherent biocompatibility between Bacillus species and Pteris ferns.

Swain et al., (2025) in a review of pteridophyte microbiome studies — the plant group to which Silver Brake Fern belongs — noted that in literature, approximately 85% of bioactive metabolites from endophytic bacteria have been attributed to Bacillus genus alone. Specifically, Zhu et al. (2014) found that Bacillus is the dominant endophytic bacterial genus associated with Pteris vittata and Pteris multifida. While endophytic bacteria reside within internal plant tissues (Vandana et al., 2021), they primarily enter through the roots (Vandana et al., 2021; Chaudhary et al., 2022). Beyond endophytic associations, rhizospheric analyses of Pteris vittata similarly identified Bacillus as the dominant genus (Antenozio et al., 2021). Together, these findings suggest that there is natural compatibility between Pteris ferns and Bacillus species. Although the Silver Brake fern used here is Pteris argyraea, it shares the genus Pteris, and so this native affinity is reasonably applicable.

The plant growth-promoting effects that drive enhanced VOC removal could be attributed to three mechanisms associated with Plant Growth Promoting Rhizobacterium (PGPR) which Bacillus subtilis (Blake et al., 2020) and Bacillus thuringiensis (Delfim and Diijoo, 2021) belong to (Figure 40).

First, both Bacillus subtilis and Bacillus thuringiensis are Indole-3-Acetic Acid (IAA)-producing bacteria (Kiruthika & Arunkumar (2020); Zhang et al. (2025); Vidal-Quist et al., 2013). IAA plays a signalling role that activates H⁺-ATPase proton pumps in plant cells, increasing Adenosine Triphosphate (ATP, the universal energy currency for living cells) hydrolysis and steepening the electrochemical gradient across the membrane. This drives stomatal opening and increases solute uptake by the roots (Takahashi et al., 2012), thus both stomatal gas exchange and root-mediated VOC uptake are enhanced. While IAA is known for stimulating root growth, such as increasing the number of root hairs, lateral roots, root size and the contact area with the medium (Lebrazi et al., 2020) and therefore increase the surface area for VOC uptake, it is unlikely that significant root development has taken place within 24 hours.

Second, Bacillus subtilis and Bacillus thuringiensis can solubilise insoluble phosphorus (an essential macronutrient) in the growing medium (Blake et al., 2020; Zhang et al. (2025); Vidal-Quist et al., 2013), making it available for plant uptake. Phosphorus is a constituent of ATP and regulates ion channels that influence stomatal conductance (Khan et al., 2023). Increased phosphorus bioavailability therefore could support both enhanced metabolic activity and sustained stomatal opening, allowing greater VOC removal. Though LECA balls are chemically inert, residual soil attached to the Silver Brake fern's roots during transfer from soil to LECA provides a plausible source of insoluble phosphorus in this experiment.

Third, 1-aminocyclopropane-1-carboxylate (ACC) deaminase is produced in Bacillus subtilis (Misra and Chauhan, 2020) and Bacillus thuringiensis (Delfim and Diijoo, 2021). ACC deaminase is an enzyme that degrades ACC — the precursor of ethylene, a plant stress hormone. By inhibiting ethylene production, the plant can continue allocating metabolic resources to growth and gas exchange rather than stress responses (Swain et al., 2025), allowing sustained VOC removal performance throughout the experimental window.

6.3.3 Spider Plant + Trichoderma viride Analysis

Spider plant + Trichoderma viride ranked second with a 23.10% reduction in AUC_t=12, alongside a 25.71% reduction in t_plateau and a 5.33% improvement in ∆VOC_max — indicating it improved in both the supplementary metrics and AUC_t=12. However, its long-term viability is limited: AUC_t=24 deteriorates by 6.21%, suggesting that the synergistic degradation capacity is not sustained beyond the 12-hour mark, resulting in a net accumulation of VOCs relative to the plant-only baseline over extended periods. This pairing may therefore be suitable for environments with well-defined short-duration occupancy, but is not recommended for continuous deployment.

The basis for this is visible in Figure 41. While the plant-microbe pairing reaches its plateau earlier than the plant-only baseline, the plateau itself is not stable. The residual VOC level increases gradually over time, continuously accumulating positive area under the curve, whereas the plant-only baseline stabilises at a low and approximately constant plateau, accumulating comparatively little area beyond the 12th hour. This difference causes the performance of the plant-microbe pairing in terms of AUC_t=24 to be worse despite the strong improvement indicated by AUC_t=12.

The observation that different microbes produce synergistic benefits for different plants — Trichoderma viride for Spider plant and Bacillus species for Silver Brake fern — is consistent with literature review which suggests that plant-microbe compatibility is host-specific and shaped by evolutionary history. Berg & Smalla (2009) suggests that one of the factors that shape microbial communities is the plant species, where different plant species recruit distinct microbial communities through their root exudates. Matthews et al. (2019) suggests that the differences in associated microbial communities increase with phylogenetic distance between plant species. Silver Brake fern (Pteris argyraea) belongs to the pteridophyta — the most ancient lineage of vascular plants predating seed-producing plants (Swain et al., 2025) — whereas Spider plant (Chlorophytum comosum) is an angiosperm in the family Asparagaceae. The evolutionary distance between these two plants is therefore substantial, thus it is reasonable that the microbes that are most compatible with each plant would also differ.

Like Bacillus, Trichoderma viride is a plant growth-promoting microorganism but in this case a fungus (Bouaicha et al., 2024). It can possibly promote plant physiological activity through shared mechanisms: IAA production (Guo et al., 2020), phosphate solubilisation (Song et al., 2024), and ACC deaminase activity (Zhao et al., 2025; Wang et al., 2022). These mechanisms would enhance stomatal conductance, root nutrient uptake, and stress tolerance in the same manner described for Bacillus above.

The gradual increase in residual VOC levels after the 4-hour mark may result from Trichoderma viride's microbial volatile organic compound (mVOCs) emissions becoming the dominant factor, outpacing the plant's VOC removal capacity (Figure 42).

As Trichoderma viride progressively colonises Spider plant’s roots, its expanding hyphal network creates an increasing number of interaction sites within and around root tissues, likely stimulating greater mVOC production over time.

Trichoderma viride is known to produce 6-pentyl-2H-pyran-2-one (6-PP) as its primary mVOC, accounting for over 50% of total VOCs emitted by Trichoderma species (Mendoza-Mendoza et al., 2024). Colonisation itself is a progressive process: Guo et al., (2020) observed that when Trichoderma viride was introduced to peppermint root systems, initial colonisation was detectable within 24 hours, with fungal biomass continuing to increase at 48 and 72 hours post-inoculation. This suggests that in our experiment, Trichoderma viride was still actively colonising throughout the measurement period. Furthermore, Mendoza-Mendoza et al. (2024) reported that Trichoderma atroviride induces 6-PP production in response to plant signals during interactions with Arabidopsis thaliana, suggesting that plant-fungus contact directly stimulates mVOCs output. Collectively, these suggest that as the hyphal network expanded, each new interaction site contributed additional 6-PP emissions, thereby leading to the observed accumulation of residual VOCs in the chamber.

6.3.4 Worse-Performing Pairings Analysis

All remaining pairings — particularly those incorporating Pseudomonas fluorescens and Silver Brake fern + Trichoderma viride — showed deterioration across all metrics, with AUC_t=12 deteriorating by as much as 1048.96% (Silver Brake fern + Trichoderma viride) (Figure 38). Two mechanisms may account for these observations (Figure 43).

The first is competition with native rhizosphere microbiota. Certain bacterial strains possess mechanisms that displace indigenous beneficial rhizosphere microbiota. For instance, certain strains of Pseudomonas fluorescens utilise a Type VI Secretion System (T6SS) to inject peptidoglycan-degrading hydrolase toxins into neighbouring bacteria, effectively killing competitors in the rhizosphere upon close contact. While this has only been confirmed in specific strains such as Pseudomonas fluorescens MFE01, the genetic components of T6SS are widespread among environmental Gram-negative bacteria (Decoin et al., 2014). By disrupting the native microbiota, the introduced microbe may deprive the plant of its natural microbial support. Without these native microbes to alleviate stress and solubilise nutrients, the plant's metabolic activity may decline, reducing its capacity to remove VOCs from the air.

The second mechanism is mVOC-mediated plant responses. mVOCs emitted by microbes are part of their natural processes and can be beneficial or detrimental to plants depending on the specific mVOC and its concentration (Raio, 2024). One documented mVOC response is induced systemic tolerance. Cho et al. (2008) demonstrated this by inoculating Pseudomonas chlororaphis O6 in Arabidopsis thaliana plants. They found that the mVOC 2R,3R-butanediol emitted by this bacterium induced systemic tolerance responses, causing plants to close their stomata to conserve water.

In our experiments, incompatible plant-microbe pairings may have triggered similar systemic tolerance responses. Such responses would cause stomatal closure, minimising gas exchange and thereby reducing the plant's capacity for environmental VOC uptake. Additionally, the mVOCs themselves accumulate in the sealed experimental chamber, contributing directly to measured VOC levels.

6.4 Conclusion

To conclude, using AUC_t=12 as the primary metric, Silver Brake Fern + Bacillus subtilis is the best pairing that demonstrates robust, sustained synergistic improvement across both the primary and secondary criteria, with its synergistic benefit strengthening rather than diminishing over extended occupancy durations. It is therefore recommended as the optimal plant-microbe pairing for indoor phytoremediation applications, although it is noted that Bacillus species could generally pair well with Silver Brake fern. Spider plant + Trichoderma viride, while effective over short occupancy durations, is not recommended for continuous deployment given its long-term performance deterioration.

7. Limitations of Project (Sammi)

This study incorporates several controlled conditions and practical constraints that may influence the generalisability of the results when applied to real-world environments.

Type of microbial strains used

The study prioritises commercially available microbial products to ensure practicality and scalability, as these are readily accessible to end users. However, these strains may differ from laboratory-optimised strains, as commercial formulations are often designed for stability and general use rather than maximum degradation performance. As a result, while commercially available strains are expected to exhibit similar degradation pathways, their efficiency may be lower compared to lab-grade strains under controlled conditions. Consequently, the findings may underestimate the maximum achievable VOC removal performance, although the relative trends observed between different plant–microbe pairings are likely to remain consistent.

Selection of plant species

The study focuses on plant species that are locally available and commercially accessible to ensure feasibility for real-world implementation. However, these species may differ from those not readily available locally in terms of morphology, leaf surface area, and physiological characteristics that influence VOC uptake. While locally available plants are expected to exhibit similar phytoremediation mechanisms, their removal efficiency may differ compared to species sourced internationally. Consequently, the findings may not fully capture the maximum achievable VOC removal performance, although the relative trends observed between different plant–microbe pairings are likely to remain consistent.

Controlled conditions and scale of experimental setup

The study was conducted under controlled experimental conditions within a relatively small-scale setup, where factors such as airflow, pollutant concentration, and environmental variables (e.g. light) were regulated. However, real indoor environments are more complex, with larger spatial scales, variable ventilation patterns, and dynamic human activities that influence pollutant distribution and system performance. As a result, the effectiveness observed in this controlled and scaled-down setup may differ when applied to full-sized indoor environments.

Pollutant concentration levels

The concentration of VOCs introduced in the experimental setup may differ from those typically found in real indoor environments, and could be either higher or lower depending on the scenario. While controlled concentrations allow for consistent comparison of system performance, they may not fully capture the variability of real-world exposure conditions. This may influence the observed removal efficiency, as system performance can vary under different pollutant loadings.

Assumptions in performance normalisation

The study normalises performance based on leaf area, assuming a linear relationship between leaf area and VOC removal efficiency. However, this relationship may not be strictly linear, as factors such as plant morphology, distribution of active leaf surfaces, and rhizosphere size can also influence performance (Orwell et al., 2004, Wolverton & Nelson, 2020). This may introduce uncertainty in comparing plant effectiveness.

Overall, these limitations highlight the need for further validation under realistic indoor conditions to strengthen the applicability of plant–microbe systems for practical VOC removal.

8. Conclusion (Sammi)

This study evaluated the performance of commercially available plant–microbe systems in removing key indoor VOCs (FBTEX), with the aim of identifying pairings that are not only effective but also safe, practical, and accessible for real-world indoor environments.

The findings show that VOC removal efficiency varies significantly across different plant–microbe combinations, confirming that performance is highly dependent on the specific pairing rather than the plant or microbial product alone. Among the tested combinations, the Silver Brake Fern-Bacillus subtilis pairing demonstrated the highest overall removal efficiency across FBTEX compounds, indicating strong synergistic interaction between plant uptake and microbial degradation in the rhizosphere. In comparison, other pairings such as Spider Plant-Pseudomonas Fluorescens and Spider Plant-Bacillus subtilis, showed lower and more variable removal performance, highlighting the importance of optimising plant–microbe compatibility to achieve the best VOC reduction.

Importantly, when evaluated beyond removal efficiency, the top-performing pairing also met key real-world criteria. It was safe for indoor use (no harmful by-products or system instability observed), practical to maintain (minimal maintenance requirements and stable performance over the testing period), and accessible (using commercially available and affordable components). This directly addresses the research question by demonstrating that an optimal plant–microbe pairing can achieve both high VOC removal efficiency and real-world applicability, rather than requiring trade-offs between performance and usability.

These results reinforce that while conventional IAQ strategies such as ventilation and filtration remain necessary, they are often insufficient in addressing continuous VOC emissions. In comparison, plant–microbe systems provide a complementary, nature-based solution, with microbial activity in the rhizosphere contributing significantly to sustained VOC degradation.

However, this study is constrained by the use of commercially available plant and microbial products, controlled experimental conditions, and assumptions made during performance normalisation, which may not fully reflect system behaviour under diverse indoor environments. Future research should therefore prioritise long-term performance validation, optimisation of plant–microbe pairings, and testing under realistic occupancy and usage conditions to enhance real-world applicability.

Overall, this study demonstrates that carefully selected plant–microbe pairings can serve as an effective, safe, and practical solution for improving indoor air quality, contributing to the development of scalable and sustainable IAQ management strategies for everyday environments.

References

Administrative Code. (n.d.). Chapter 3745-112 Volatile Organic Compound Limits in Consumer products. In Administrative Code.
Ahu Aydogan. (2020). The Durability of Plant-Based Air Filtering Systems in Buildings: From an air quality and energy reduction perspective. In C. Serrat, J. R. Casas, & V. Gibert (Eds.), XV International Conference on Durability of Building Materials and Components [Conference-proceeding].
Air Quality, Energy and Health (AQE). (2010, January 1). WHO guidelines for indoor air quality: selected pollutants. https://www.who.int/publications/i/item/9789289002134
Air Quality Connected Data Study. (2024). Retrieved November 11, 2025, from Dyson.com.sg website: https://www.dyson.com.sg/newsroom/air-quality-connected-data-study?srsltid=AfmBOoptN0nPqioHkcCThXeK_gpGs7BXNjVJ4OZ7-WbDjpi50aXZMO6y
Air Quality Expert Group (2022). Indoor Air Quality https://uk-air.defra.gov.uk/assets/documents/reports/cat09/2211011000_15062022_Indoor_Air_Quality_Report_Final.pdf
Akinbomi, N. J. G., Obafaiye, N. a. G., & Adeola, N. a. K. (2022). Influence of process variables on shoe polish viscosity. International Journal of Frontiers in Engineering and Technology Research, 2(1), 001–007. https://doi.org/10.53294/ijfetr.2022.2.1.0033
Anany, H., Brovko, L., Arabi, T. E., & Griffiths, M. (2015). Bacteriophages as antimicrobials in food products. In T.M. Taylor (Ed.), Handbook of Natural Antimicrobials for Food Safety and Quality (pp. 89–116). Woodhead Publishing. https://doi.org/10.1016/b978-1-78242-034-7.00005-0
Antenozio, M. L., Giannelli, G., Marabottini, R., Brunetti, P., Allevato, E., Marzi, D., Capobianco, G., Bonifazi, G., Serranti, S., Visioli, G., Stazi, S. R., & Cardarelli, M. (2021). Phytoextraction efficiency of Pteris vittata grown on a naturally As-rich soil and characterization of As-resistant rhizosphere bacteria. Scientific Reports, 11(1), Article 6794. https://doi.org/10.1038/s41598-021-86076-7
Ardley, J. K., & Green, P. N. (2023). Methylorubrum. In M. E. Trujillo, S. Dedysh, P. DeVos, B. Hedlund, P. Kämpfer, F.A. Rainey & W.B. Whitman (Eds.), Bergey’s Manual of Systematics of Archaea and Bacteria (pp. 1–12). https://doi.org/10.1002/9781118960608.gbm02024
Armijos-Moya, T., et al. (2021). Air cleaning performance of two species of potted plants and different substrates. Applied Sciences, 12(1), 284.
AWAIR Support (2023) Technical Accuracy of Awair Element Sensors. https://support.getawair.com/hc/en-us/articles/360060843933-Technical-Accuracy-of-Awair-Element-Sensors
Aydogan, A., & Montoya, L. D. (2011). Formaldehyde removal by common indoor plant species and various growing media. Atmospheric Environment, 45(16), 2675–2682. https://doi.org/10.1016/j.atmosenv.2011.02.062
Baberi, Z., Azhdarpoor, A., Hoseini, M., Baghapour, M., Derakhshan, Z., & Giannakis, S. (2022). Monitoring Benzene, Toluene, Ethylbenzene, and Xylene (BTEX) Levels in Mixed-Use Residential-Commercial Buildings in Shiraz, Iran: Assessing the Carcinogenicity and Non-Carcinogenicity Risk of Their Inhabitants. International Journal of Environmental Research and Public Health, 19(2), 723. https://doi.org/10.3390/ijerph19020723
BACILLUS MEGATERIUM. (n.d.). Novobac. https://www.novobac.com/product/bacillus-megaterium/
Basnett, D., Lama, P. C. & Sharma, B. C. (2024). Characterization Of Azotobacter Chroococcum Isolated From Maize Rhizosphere Of Darjeeling Hills For Presence Of Pgp Traits And Its Survivality Efficiency Under Different Growth Conditions. IOSR Journal Of Pharmacy And Biological Sciences (IOSR-JPBS), 19(1), 39–45. https://www.iosrjournals.org/iosr-jpbs/papers/Vol19-issue1/Ser-1/F1901013945.pdf
BCA (2020) Greener buildings, better health, BuildSG Magazine. (n.d.). Retrieved from BCA Corp website: https://www1.bca.gov.sg/buildsg-emag/articles/greener-buildings-better-health
Belal, E. B. (2008). Biodegradation of Wastepaper by Trichoderma Viride and Using the Bioprocessed Materials in Biocontrol of Damping-off of Pea Caused by Pythium Debaryanum. Journal of Agricultural Research, Kafrelsheikh University, 34(3), 567 – 587. https://www.researchgate.net/profile/Elsayed-Belal-2/publication/267024992_Biodegradation_of_wastepaper_by_Trichoderma_viride_and_using_bioprocessed_materials_in_biocontrol_of_damping_-_off_of_pea_caused_by_Pythium_debaryanum/links/54410fdf0cf2e6f0c0f55070/Biodegradation-of-wastepaper-by-Trichoderma-viride-and-using-bioprocessed-materials-in-biocontrol-of-damping-off-of-pea-caused-by-Pythium-debaryanum.pdf
Berg, G., & Smalla, K. (2009). Plant species and soil type cooperatively shape the structure and function of microbial communities in the rhizosphere. FEMS Microbiology Ecology, 68(1), 1–13. https://doi.org/10.1111/j.1574-6941.2009.00654.x
Białowiec, A., et al. (2012). Nitrogen removal from landfill leachate in constructed wetlands with reed and willow: Redox potential in the root zone. Journal of Environmental Management, 97, 22–27.
Blake, C., Christensen, M. N., & Kovács, Á. T. (2020). Molecular Aspects of Plant Growth Promotion and Protection byBacillus subtilis. Molecular Plant-Microbe Interactions, 34(1), 15–25. https://doi.org/10.1094/mpmi-08-20-0225-cr
Bouaicha, O., Maver, M., Mimmo, T., Cesco, S., & Borruso, L. (2024). Microplastic influences the ménage à trois among the plant, a fungal pathogen, and a plant growth-promoting fungal species. Ecotoxicology and Environmental Safety, 279, Article 116518. https://doi.org/10.1016/j.ecoenv.2024.116518
Building and Construction Authority, Singapore. (2019). CODE ON ENVIRONMENTAL SUSTAINABILITY MEASURES FOR EXISTING BUILDINGS (Edition 2.0). https://www1.bca.gov.sg/docs/default-source/docs-corp-buildsg/sustainability/code_env_sus_measures_exit_building_2c16b073e4cde45cfb8e475dd4403c266.pdf
California Air Resources Board. (2015). Inhalable Particulate Matter and Health (PM2.5 and PM10). Retrieved from Ca.gov website: https://ww2.arb.ca.gov/resources/inhalable-particulate-matter-and-health
Canada, H. (2021). Formaldehyde in your home, Canada.ca. Retrieved November 11, 2025, from Canada.ca website: https://www.canada.ca/en/health-canada/services/environmental-workplace-health/reports-publications/air-quality/formaldehyde-indoor-air-environment-workplace-health.html
Canellas, L. P., Canellas, N. O. A., Da Silva, R. M., Spaccini, R., Mota, G. P., & Olivares, F. L. (2022). Biostimulants Using Humic Substances and Plant-Growth-Promoting Bacteria: Effects on Cassava (Manihot esculentus) and Okra (Abelmoschus esculentus) Yield. Agronomy, 13(1), Article 80. https://doi.org/10.3390/agronomy13010080
Chaudhary, P., Agri, U., Chaudhary, A., Kumar, A., & Kumar, G. (2022). Endophytes and their potential in biotic stress management and crop production. Frontiers in Microbiology, 13, Article 933017. https://doi.org/10.3389/fmicb.2022.933017
Cheng, Z., Lu, L., Kennes, C., Yu, J., & Chen, J. (2016). Treatment of gaseous toluene in three biofilters inoculated with fungi/bacteria: Microbial analysis, performance and starvation response. Journal of Hazardous Materials, 303, 83–93. https://doi.org/10.1016/j.jhazmat.2015.10.017
Cho, S. M., Kang, B. R., Han, S. H., Anderson, A. J., Park, J., Lee, Y., Cho, B. H., Yang, K., Ryu, C., & Kim, Y. C. (2008). 2R,3R-Butanediol, a Bacterial Volatile Produced by Pseudomonas chlororaphis O6, Is Involved in Induction of Systemic Tolerance to Drought in Arabidopsis thaliana. Molecular Plant-Microbe Interactions, 21(8), 1067–1075. https://doi.org/10.1094/mpmi-21-8-1067
Comi, G., & Cantoni, C. (2011). PSYCHROTROPHIC BACTERIA | Arthrobacter spp. In J. W. Fuquay (Ed.) Encyclopedia of Dairy Sciences (2nd ed., pp. 372–378). Elsevier Ltd. https://doi.org/10.1016/b978-0-12-374407-4.00440-4
Cornejo JJ, Munoz FG, Ma CY, Stewart AJ (1999) Studies on the decontamination of air by plants. Ecotoxicology 8:311–320
Cruz, M. D., Svenningsen, N. B., Nybroe, O., Müller, R., & Christensen, J. H. (2023). Removal of a complex VOC mixture by potted plants—effects on soil microorganisms. Environmental Science and Pollution Research, 30(19), 55372–55381. https://doi.org/10.1007/s11356-023-26137-8
Daccò, C., Nicola, L., Temporiti, M. E. E., Mannucci, B., Corana, F., Carpani, G., & Tosi, S. (2020). Trichoderma: Evaluation of its degrading abilities for the bioremediation of hydrocarbon complex mixtures. Applied Sciences, 10(9), 3152. https://doi.org/10.3390/app10093152
Daudzai, Z., Treesubsuntorn, C., & Thiravetyan, P. (2018). Inoculated Clitoria ternatea with Bacillus cereus ERBP for enhancing gaseous ethylbenzene phytoremediation: Plant metabolites and expression of ethylbenzene degradation genes. Ecotoxicology and Environmental Safety, 164, 50–60. https://doi.org/10.1016/j.ecoenv.2018.07.121
Dcarrick. (2024, November 22). Indoor air quality: a health issue long overlooked, Savills Impacts. Retrieved November 11, 2025, from Savills Impacts website: https://impacts.savills.com/environment/indoor-air-quality-a-health-issue-long-overlooked.html
Decoin, V., Barbey, C., Bergeau, D., Latour, X., Feuilloley, M. G. J., Orange, N., & Merieau, A. (2014). A Type VI Secretion System Is Involved in Pseudomonas fluorescens Bacterial Competition. PLoS ONE, 9(2), Article e89411. https://doi.org/10.1371/journal.pone.0089411
De Kempeneer, L., Sercu, B., Vanbrabant, W., Van Langenhove, H., & Verstraete, W. (2004). Bioaugmentation of the phyllosphere for the removal of toluene from indoor air. Applied Microbiology and Biotechnology, 64(2), 284–288. https://doi.org/10.1007/s00253-003-1415-3
Delfim, J., & Dijoo, Z. K. (2021). Bacillus thuringiensis as a Biofertilizer and Plant Growth Promoter. In G. H. Dar, R. A. Bhat, M. A. Mehmood & K. R. Hakeem (Eds.), Microbiota and Biofertilizers, Vol 2: Ecofriendly Tools for Reclamation of Degraded Soil Environs (Vol. 2, pp. 251–265). Springer Nature Switzerland AG. https://doi.org/10.1007/978-3-030-61010-4_12
Dhanya V. (2019). Biodegradation of Benzene, Toluene and Xylene by a Formulated Bacterial Consortium. Journal of Emerging Technologies and Innovative Research, 6(2), 733–739. https://www.jetir.org/papers/JETIR1902800.pdf
Di Maiuta, N., Hubschmid, S., Giuliani, N., Schwarzentruber, P., & Dow, C. S. (2009). Microbial degradation of formaldehyde in white mineral dispersions preserved with formaldehyde-releasing biocides. International Biodeterioration & Biodegradation, 63(6), 769–777. https://doi.org/10.1016/j.ibiod.2009.05.008
Dolphen, R., Treesubsuntorn, C., Santawee, N., Setsungnoen, A., & Thiravetyan, P. (2019). Modified coir pith with glucose syrup as a supporter in non-external nutrient supplied biofilter for benzene removal by Bacillus megaterium. Environmental Technology, 41(27), 3607–3618. https://doi.org/10.1080/09593330.2019.1615994
Dordio, A. V., & Carvalho, A. J. P. (2013). Organic xenobiotics removal in constructed wetlands, with emphasis on the importance of the support matrix. Journal of Hazardous Materials, 252-253, 272–292.
Ehmedan, S. S., Ibrahim, M. K., Azzam, A. M., Hamedo, H. A., & Saeed, A. M. (2021). Acceleration the bacterial biodegradation of crude oil pollution using Fe₂O₃ and ZnO nanoparticles. Environmental Nanotechnology Monitoring & Management, 16, Article 100613. https://doi.org/10.1016/j.enmm.2021.100613
El-Arabi, T. F., & Griffiths, M. W. (2013). Bacillus cereus. In J. G. Morris Jr. & M. E. Potter (Eds.) Foodborne Infections and Intoxications (4th Ed., pp. 401–407). Academic Press. https://doi.org/10.1016/b978-0-12-416041-5.00029-9
Errington, J. (2003). Regulation of endospore formation in Bacillus subtilis. Nature Reviews Microbiology, 1(2), 117-126. https://doi.org/10.1038/nrmicro750
EPA HERO Database (2023). https://hero.epa.gov/hero/index.cfm/reference/details/reference_id/5972703
Fadahunsi, I. F. & Phebe, A. Q. (2017). Optimization of Lipase Production by Bacillus megaterium. Biotechnology Journal International, 18(2), Article 32505. https://doi.org/10.9734/bji/2017/32505
Fonseca, P., Moreno, R., & Rojo, F. (2011). Growth of Pseudomonas putida at low temperature: global transcriptomic and proteomic analyses. Environmental Microbiology Reports, 3(3), 329–339. https://doi.org/10.1111/j.1758-2229.2010.00229.x
Fussy, A., & Papenbrock, J. (2022). An overview of soil and soilless cultivation techniques—Chances, challenges and the neglected question of sustainability. Plants, 11(9), 1153.
Green Plants for Green Buildings (2014). Can ornamental potted plants remove volatile organic compounds from indoor air? — a review. (2014). In Environ Sci Pollut Res. https://doi.org/10.1007/s11356-014-3240-x
Groom, J. D., & Lidstrom, M. E. (2021). Cultivation techniques to study lanthanide metal interactions in the haloalkaliphilic Type I methanotroph “Methylotuvimicrobium buryatense” 5GB1C. In Methods in enzymology on CD-ROM/Methods in enzymology, 650,237–259. https://doi.org/10.1016/bs.mie.2021.01.042
Guo, K., Sui, Y., Li, Z., Huang, Y., Zhang, H., & Wang, W. (2020). Colonization of Trichoderma viride Tv-1511 in peppermint (Mentha × piperita L.) roots promotes essential oil production by triggering ROS-mediated MAPK activation. Plant Physiology and Biochemistry, 151, 705–718. https://doi.org/10.1016/j.plaphy.2020.03.042
Han, X., Wang, R. R., & Shi, C. Z. (2020). [Isolation, identification, and degrading characteristics of an oil Resistant Formaldehyde-Degrading Bacterium]. Environmental Scienceu, 41(12), 5617–5627. [Article in Chinese] https://doi.org/10.13227/j.hjkx.202003282 (Accessed via PubMed for English version https://pubmed.ncbi.nlm.nih.gov/33374079/ ).
Handayani, K., Indra, R. N., Agustrina, R., & Sumardi. (2025). Study of Morphology and Protein Crystal Character of Bacillus thuringiensis Isolates. Proceedings of the 5th International Conference on Applied Sciences, Mathematics, and Informatics (ICASMI 2024) 280–293. Atlantis Press. https://doi.org/10.2991/978-94-6463-730-4_24
Harrison, R., Saborit, J. M. D., Dor, F., & Henderson, R. (2010). Benzene. In www.ncbi.nlm.nih.gov. World Health Organization. Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK138708/
Hidalgo, A., Lopategi, A., Prieto, M., Serra, J. L., & Llama, M. J. (2002). Formaldehyde removal in synthetic and industrial wastewater by Rhodococcus erythropolis UPV-1. Applied Microbiology and Biotechnology, 58(2), 260–264. https://doi.org/10.1007/s00253-001-0876-5
Huang, W., Wang, Z., Choudhary, G., Guo, B., Zhang, J., & Ren, D. (2012). Characterization of microbial species in a regenerative bio-filter system for volatile organic compound removal. HVAC&R Research, 18(1–2), 169–178. https://doi.org/10.1080/10789669.2011.591259
Hutchins, J. (n.d.). Pumps. In On Becoming a Neuroscientist. https://colorado.pressbooks.pub/neuroscience/chapter/the-sodium-potassium-pump/
Hying, Z. T., Rushmer, A. M., Loh, C. Y., Bruger, E. L., & Bazurto, J. V. (2025). Methylobacterium extorquens PA1 utilizes multiple strategies to maintain formaldehyde homeostasis during methylotrophic growth. PLoS Genetics, 21(6), e1011736. https://doi.org/10.1371/journal.pgen.1011736
Irga, P. J., Pettit, T. J., & Torpy, F. R. (2018). The phytoremediation of indoor air pollution: a review on the technology development from the potted plant through to functional green wall biofilters. Reviews in Environmental Science and Bio/Technology, 17(2), 395–415. https://doi.org/10.1007/s11157-018-9465-2
Jayaprada, B., Kumari, A. M., & Vijayalakshmi, M. (2025). Antimicrobial Profile of Rhodococcus Erythropolis JPVL-3 of Mangrove Origin. Journal of Scientific Research, 69(01), 31–35. https://doi.org/10.37398/jsr.2025.690104
Jinal, H. N., Amaresan, N., & Sankaranarayanan, A. (2020). Methylobacterium. In N. Amaresan, M. Senthil Kumar, K. Annapurna, Krishna Kumar, A. Sankaranarayanan (Eds.), Beneficial Microbes in Agro-Ecology Bacteria and Fungi (pp. 509–519). Academic Press. https://doi.org/10.1016/b978-0-12-823414-3.00024-1
Joo et al (2025) Comparative analysis of indoor volatile organic compound levels in an office: Impact of occupancy and centrally controlled ventilation. Atmospheric Environment, 345, Article 121057. https://doi.org/10.1016/j.atmosenv.2025.121057
Kaden, D. A., Mandin, C., Nielsen, G. D., & Wolkoff, P. (2010). Formaldehyde. In www.ncbi.nlm.nih.gov. WHO Guidelines for Indoor Air Quality: Selected Pollutants. Retrieved from https://www.ncbi.nlm.nih.gov/books/NBK138711/
Kang, P., Wu, P., Jin, Y., Shi, S., Gao, D., Chen, G., & Li, Q. (2020). Formation and Emissions of Volatile Organic Compounds from Homo-PP and Co-PP Resins during Manufacturing Process and Accelerated Photoaging Degradation. Molecules, 25(12), 2761. https://doi.org/10.3390/molecules25122761
Kao, C., Liu, J., Chen, Y., Chai, C., & Chen, S. (2005). Factors affecting the biodegradation of PCP by Pseudomonas mendocina NSYSU. Journal of Hazardous Materials, 124(1–3), 68–73. https://doi.org/10.1016/j.jhazmat.2005.03.051
Kesavan, S., Inamdar, M., & Muthunarayanan, V. (2021). Concentration of benzene, toluene, napthalene and acenaphthene on selected bacterial species. Materials Today Proceedings, 37(Part 2), 273–279. https://doi.org/10.1016/j.matpr.2020.05.241
Khaksar, G., Treesubsuntorn, C., & Thiravetyan, P. (2016). Endophytic Bacillus cereus ERBP—Clitoria ternatea interactions: Potentials for the enhancement of gaseous formaldehyde removal. Environmental and Experimental Botany, 126, 10–20. https://doi.org/10.1016/j.envexpbot.2016.02.009
Khan, F., Siddique, A. B., Shabala, S., Zhou, M., & Zhao, C. (2023). Phosphorus plays key roles in regulating plants’ physiological responses to abiotic stresses. Plants, 12(15), Article 2861. https://doi.org/10.3390/plants12152861
Kinhal, V. (2023, January 9). Seven crucial applications of Leaf area Data. CID Bio-Science. http://cid-inc.com/blog/seven-crucial-applications-of-leaf-area-data/#:~:text=Stomatal%20Conductance%20or%20opening%20regulates,correlated%20to%20the%20leaf%20area.
Kiruthika, S., & Arunkumar, M. (2020). A Comprehensive Study on IAA production by Bradyrhizobium japonicum and Bacillus subtilis and Its Effect on Vigna radiata Plant Growth. Indian Journal of Agricultural Research, 55(5), 570–576. https://doi.org/10.18805/ijare.a-5521
Koivisto, Kirsten Inga Kling, Ana Sofia Fonseca, Anders Brostrøm Bluhme, Moreman, M., Yu, M., … Keld Alstrup Jensen. (2018). Dip coating of air purifier ceramic honeycombs with photocatalytic TiO2 nanoparticles: A case study for occupational exposure. Science of the Total Environment, 630, 1283–1291. https://doi.org/10.1016/j.scitotenv.2018.02.316
Lai, M. (2025, April 5). Indoor air quality in Singapore Homes: Impact on children’s health and development. Pico X Health. https://picoxhealth.com/blogs/mold-and-fungi/indoor-air-quality-in-singapore-homes-impact-on-children-s-health-and-development?srsltid=AfmBOorbQlouY_KxqkCWb8xApVSzEN4KGGCAr8XyqGeNcaGWOIUoa6NW
Lan, H., Qi, S., Yang, D., Zhang, H., Liu, J., & Sun, Y. (2020). Combination of highly efficient microflora to degrade paint spray exhaust gas. Scientific Reports, 10(1), 6027. https://doi.org/10.1038/s41598-020-62972-2
Lebert, I., Begot, C., & Lebert, A. (1998). Growth of Pseudomonas fluorescens and Pseudomonas fragi in a meat medium as affected by pH (5.8–7.0), water activity (0.97–1.00) and temperature (7–25°C). International Journal of Food Microbiology, 39(1–2), 53–60. https://doi.org/10.1016/s0168-1605(97)00116-5
Lebrazi, S., Niehaus, K., Bednarz, H., Fadil, M., Chraibi, M., & Fikri-Benbrahim, K. (2020). Screening and optimization of indole-3-acetic acid production and phosphate solubilization by rhizobacterial strains isolated from Acacia cyanophylla root nodules and their effects on its plant growth. Journal of Genetic Engineering and Biotechnology, 18(1), Article 71. https://doi.org/10.1186/s43141-020-00090-2
Lee, S. Y., Oh, H. Y., & Kim, O. B. (2013). Isolation and Characterization of Volatile Organic Compounds-Degrading Bacillus Strains from Loess. Journal of Environmental Protection, 04(04), 50–57. https://doi.org/10.4236/jep.2013.44a007
Li, K., Chen, M., Shi, J., & Mao, T. (2024). An overview of the production and use of Bacillus thuringiensis toxin. Open Life Sciences, 19(1), Article 20220902. https://doi.org/10.1515/biol-2022-0902
Li, Y., Li, J., Wang, C., & Wang, P. (2010). Growth kinetics and phenol biodegradation of psychrotrophic Pseudomonas putida LY1. Bioresource Technology, 101(17), 6740–6744. https://doi.org/10.1016/j.biortech.2010.03.083
Lim, S. K., Shin, H. S., Yoon, K. S., Kwack, S. J., Um, Y. M., Hyeon, J. H., Kwak, H. M., Kim, J. Y., Kim, T. H., Kim, Y. J., Roh, T. H., Lim, D. S., Shin, M. K., Choi, S. M., Kim, H. S., & Lee, B. (2014). Risk assessment of volatile organic compounds benzene, toluene, ethylbenzene, and xylene (BTEX) in consumer products. Journal of Toxicology and Environmental Health, 77(22–24), 1502–1521. https://doi.org/10.1080/15287394.2014.955905
Liu, X., Martin, J. J. J., Li, X., Zhou, L., Li, R., Li, Q., Zhang, J., Fu, D., & Cao, H. (2025). Optimization of the fermentation culture conditions of Bacillus amyloliquefaciens ck-05 using response surface methodology. Frontiers in Microbiology, 16, 1552645. https://doi.org/10.3389/fmicb.2025.1552645
Loggenberg, S. R., Twilley, D., De Canha, M. N., & Lall, N. (2022). Medicinal plants used in South Africa as antibacterial agents for wound healing. In F. Chassagne (Ed.), Medicinal Plants as Anti-Infectives Current Knowledge and New Perspectives (pp. 139–182). Elsevier eBooks. https://doi.org/10.1016/b978-0-323-90999-0.00018-5
Logue, J. M., McKone, T. E., Sherman, M. H., & Singer, B. C. (2011). Hazard assessment of chemical air contaminants measured in residences. Indoor Air, 21(2), 92–109. https://pubmed.ncbi.nlm.nih.gov/21392118/
Lu, N., Pei, J., Zhao, Y., Qi, R., & Liu, J. (2012). Performance of a biological degradation method for indoor formaldehyde removal. Building and Environment, 57, 253–258. https://doi.org/10.1016/j.buildenv.2012.05.007
Lung Basics. (n.d.). Retrieved from Lung Care Foundation website: https://lcf.org.in/lung-basics/
Ma, S., Wu, T., Zhang, H., Fang, H., Kang, J., Zhao, J., Fan, Z., Liu, H., & Yao, Y. (2025). A highly potent benzene-degrading bacterial strain from petroleum-contaminated soils: isolation and characterization. Bioremediation Journal, 1–12. https://doi.org/10.1080/10889868.2025.2504151
Marcham, C., Miller, J. D., Springston, J., Hawkins, V., & Kollmeyer, B. (2020, May 26). The Value of IAQ: A review of the scientific evidence supporting the benefits of investing in better indoor air quality. Retrieved from https://www.researchgate.net/publication/353719205_The_Value_of_IAQ_A_review_of_the_scientific_evidence_supporting_the_benefits_of_investing_in_better_indoor_air_quality
Mata, T. M., Martins, A. A., & Santos, A. C. (2022). Indoor air quality: A review of cleaning technologies. Environments, 9(9), 118. https://www.mdpi.com/2076-3298/9/9/118
Matthews, A., Pierce, S., Hipperson, H., & Raymond, B. (2019). Rhizobacterial community assembly patterns vary between crop species. Frontiers in Microbiology, 10, Article 581. https://doi.org/10.3389/fmicb.2019.00581
Mendoza-Mendoza, A., Esquivel-Naranjo, E. U., Soth, S., Whelan, H., Alizadeh, H., Echaide-Aquino, J. F., Kandula, D., & Hampton, J. G. (2024). Uncovering the multifaceted properties of 6-pentyl-alpha-pyrone for control of plant pathogens. Frontiers in Plant Science, 15, Article 1420068. https://doi.org/10.3389/fpls.2024.1420068
Meng, G.-Z., & Ji, K.-S. (2012, 2012-07-01). Comparisons of the Capacity of Four Ornamental Plants on Purifying Indoor Formaldehyde Pollution. Paper presented at the 2012 Fifth International Conference on Information and Computing Science.
Messan, S., Shahud, A., Anis, A., Kalam, R., Ali, S., & Aslam, M. I. (2022). Air-MIT: Air Quality Monitoring Using Internet of Things (p. 45) [The 7th International Electrical Engineering Conference]. Retrieved November 14, 2025, from https://doi.org/10.3390/engproc2022020045
Mikkonen, A., et al. (2018). Biofiltration of airborne VOCs with green wall systems-Microbial and chemical dynamics. Indoor Air, 28(5), 697–707.
Misra, S., & Chauhan, P. S. (2020). ACC deaminase-producing rhizosphere competent Bacillus spp. mitigate salt stress and promote Zea mays growth by modulating ethylene metabolism. 3 Biotech, 10(3), 119. https://doi.org/10.1007/s13205-020-2104-y
Modupe Mewomo, James Olaonipekun Toyin, Comfort Iyiola, & Aluko, O. (2021, October 7). The Impact of Indoor Environmental Quality on Building Occupants Productivity and Human Health: A Literature Review. Retrieved from ResearchGate website: https://www.researchgate.net/publication/360354044_The_Impact_of_Indoor_Environmental_Quality_on_Building_Occupants_Productivity_and_Human_Health_A_Literature_Review
Montaluisa-Mantilla, M. S., Lebrero, R., García-Encina, P. A., & Muñoz, R. (2025). Elucidation of the mechanisms of VOC removal in botanical filters during indoor air treatment in a test chamber. Building and Environment, 287, Article 113775. https://doi.org/10.1016/j.buildenv.2025.113775
Munekata, T., Suzuki, T., Yamakawa, S., & Asahi, R. (2013). Effects of viscosity, surface tension, and evaporation rate of solvent on dry colloidal structures: A lattice Boltzmann study. Physical Review E, 88(5), 052314. https://doi.org/10.1103/physreve.88.052314
Naava. (n.d.). Naava | Not a normal green wall. Retrieved from www.naava.io website: https://www.naava.io/en/
Natede. (2023). Test report. https://vitesy.com/download/Natede_Test-Report.pdf?srsltid=AfmBOooH7ETEVlv-T-0YFVQ5HOvxXLNTJv0pF6lEkMZZUtbhwLIo8AiF
NATEDE. (2024). Sustainable air purifiers and wellness solutions | Vitesy. Retrieved November 11, 2025, from Vitesy.com website: https://vitesy.com/?rfsn=8567827.506fd1e&gad_source=1&gad_campaignid=22460764202&gbraid=0AAAAACLlCvQ7g0nXdlWRycRYmyKCqHE7G&gclid=CjwKCAiA2svIBhB-EiwARWDPjlMZM8REX2vWgTbJFsN4iCMfVwnQNZVMlnQPlVBtfaPNS5L3bRyH9RoCiYsQAvD_BwE
NEA (2024). https://www.nea.gov.sg/our-services/pollution-control/air-and-coastal-water-quality-monitoring
NEA (n.d.).Updated Guidance on Improving Ventilation and Indoor Air Quality in Buildings for a Healthy Indoor Environment. Retrieved from www.nea.gov.sg website: https://www.nea.gov.sg/our-services/public-cleanliness/environmental-cleaning-guidelines/advisories/guidance-on-improving-ventilation-and-indoor-air-quality-in-buildings-for-a-healthy-indoor-environment
NEOPLANTS, n.d. Power Drops White Paper.
Ngalimat, M. S., Yahaya, R. S. R., Baharudin, M. M. A., Yaminudin, S. M., Karim, M., Ahmad, S. A., & Sabri, S. (2021). A Review on the Biotechnological Applications of the Operational Group Bacillus amyloliquefaciens. Microorganisms, 9(3), 614. https://doi.org/10.3390/microorganisms9030614
Nippon Paint (S) Co. Pte Ltd. (2024). SAFETY DATA SHEET. https://nipponpaint.com.sg/wp-content/uploads/VINILEX-5000-SDS_08.02.2024.pdf#:~:text=Viscosity.%20:%2090%20%2D%2096%20KU.
Novinscak, A., & Filion, M. (2020). Long Term Comparison of Talc- and Peat-Based Phytobeneficial Pseudomonas fluorescens and Pseudomonas synxantha Bioformulations for Promoting Plant Growth. Frontiers in Sustainable Food Systems, 4, Article 602911. https://doi.org/10.3389/fsufs.2020.602911
OECD. Biosafety and the Environmental Uses of Micro-Organisms Conference Proceedings. (n.d.). Retrieved from https://www.oecd.org/content/dam/oecd/en/publications/reports/2015/01/biosafety-and-the-environmental-uses-of-micro-organisms_g1g42b15/9789264213562-en.pdf
Orwell RL, Wood RL, Tarran J, Torpy F, Burchett MD (2004) Removal of benzene by the indoor plant/substrate microcosm and implications for air quality. Water Air Soil Pollut 157:193–207
Paau, A. S. (1998). Formulation of Beneficial Organisms Applied to Soil. In H. D. Burges (Ed.), Formulation of Microbial Biopesticides: Beneficial microorganisms, nematodes and seed treatments (1st ed., pp. 235–254). Springer Science+Business Media. https://doi.org/10.1007/978-94-011-4926-6_7
Park, G., & Son, H. (2009). Keratinolytic activity of Bacillus megaterium F7-1, a feather-degrading mesophilic bacterium. Microbiological Research, 164(4), 478–485. https://doi.org/10.1016/j.micres.2007.02.004
Pedro, Sousa, S. I. V., Dudzińska, M. R., Duygu Gazioglu Ruzgar, Mutlu, M., Georgios Panaras, … Weersink, A. (2024c). A review of relevant parameters for assessing indoor air quality in educational facilities. Environmental Research, 119713–119713. https://doi.org/10.1016/j.envres.2024.119713
Pettit, T., Irga, P., & Torpy, F. (2019). The in situ pilot-scale phytoremediation of airborne VOCs and particulate matter with an active green wall. Air Quality, Atmosphere & Health, 12(1), 33-44. doi:10.1007/s11869-018-0628-7
Pliushch, A., & Pliushch, T. (2023). Plant-microbe system effectiveness in indoor air quality improvement: a review. IOP Conference Series: Earth and Environmental Science, 1255(1). https://iopscience.iop.org/journal/1755-1315#:~:text=If%20you%20would%20like%20more%20information%20regarding,please%20visit%20our%20page%20for%20conference%20organizers
Pseudomonas mendocina. (n.d.). ABIS Encyclopedia. https://tgw1916.net/Pseudomonas/mendocina.html
R. Sivasankar, P. Ezhilarasan, Kumar, P. S., Naidu, S. A., Rao, G. D., Vishnu Vardhan Kanuri, … K. Ramu. (2018). Loricate ciliates as an indicator of eutrophication status in the estuarine and coastal waters. Marine Pollution Bulletin, 129(1), 207–211. https://doi.org/10.1016/j.marpolbul.2018.02.022
R. Thavasi, S. Jayalakshmi, T. Balasubramanian, & Ibrahim M. Banat. (2006). Biodegradation of Crude Oil by Nitrogen Fixing Marine Bacteria Azotobacter chroococcum. Research Journal of Microbiology, 1(5), 401–408. https://doi.org/10.3923/jm.2006.401.408
Raio, A. (2024). Diverse roles played by “Pseudomonas fluorescens complex” volatile compounds in their interaction with phytopathogenic microrganims, pests and plants. World Journal of Microbiology and Biotechnology, 40(3), Article 80. https://doi.org/10.1007/s11274-023-03873-0
Rank, J. K., Nathani, N. M., Hinsu, A. T., Joshi, A. U., Shekar, M. C., & Kothari, R. K. (2018). Isolation, characterization and growth response study of chlorpyrifos utilizing soil bacterium Pseudomonas putida JR16. Indian Journal of Agricultural Research, 52(4), 355–361 https://doi.org/10.18805/ijare.a-5035
Ravindra, K., & Mor, S. (2022). Phytoremediation potential of indoor plants in reducing air pollutants. Frontiers in Sustainable Cities, 4. https://doi.org/10.3389/frsc.2022.1039710
Rogers, J. (2025, March 12). Sticking to the right consistency: How temperature influences industrial glue viscosity and storage. Powerblanket. https://www.powerblanket.com/blog/sticking-to-the-right-consistency-how-temperature-influences-industrial-glue-viscosity-and-storage/#:~:text=When%20calculating%20viscosity%20and%20a,Elmer's%20glue%20is%202%2C311%20cPs.
Saint Humic Acid. (n.d.). Relationship among soil, humic acid, and microorganisms. https://www.humicacid.org/relationship-among-soil-humic-acid-and-microorganisms
Santillan-Culquimboz, H. W., Leiva, S. T., Munoz-Salas, M. N., Meza-Maicelo, W., Lozano-Isla, F., Oliva-Cruz, M., & Balcázar-Zumaeta, C. R. (2025). Ecology and functions of Trichoderma in coffee and cocoa agroecosystems: bibliometric and systematic insights for sustainable agriculture. Frontiers in Microbiology, 16, 1717484. https://doi.org/10.3389/fmicb.2025.1717484
Satapute, P. P., Olekar, H. S., Shetti, A. A., Kulkarni, A. G., Hiremath, G. B., Patagundi, B. I., Shivsharan, C. T & Kaliwal, B. B. (2012). Isolation and Characterization of Nitrogen Fixing Bacillus Subtilis Strain AS-4 from Agricultural Soil. International Journal of Recent Scientific Research, 3(9), 762–765. https://www.researchgate.net/publication/259605129_ISOLATION_AND_CHARACTERIZATION_OF_NITROGEN_FIXING_BACILLUS_SUBTILIS_STRAIN_AS-4_FROM_AGRICULTURAL_SOIL
ScienceDaily (2023, December 23). Air pollutants commonly found indoors could have an impact on creativity. https://www.sciencedaily.com/releases/2023/12/231219124528.shtml
ScienceDirect (April 15, 2023). Optimization of photocatalytic oxidation reactor for air purifier design: Application of artificial neural network and genetic algorithm. https://www.sciencedirect.com/science/article/abs/pii/S1385894723009178
Sheoran, K., Siwal, S. S., Kapoor, D., Singh, N., Saini, A. K., Alsanie, W. F., & Thakur, V. K. (2022). Air Pollutants Removal Using Biofiltration Technique: A Challenge at the Frontiers of Sustainable Environment. ACS Engineering Au, 2(5). https://doi.org/10.1021/acsengineeringau.2c00020
Shim, H., Hwang, B., Lee, S., & Kong, S. (2005). Kinetics of BTEX biodegradation by a coculture of Pseudomonas putida and Pseudomonas fluorescens under hypoxic conditions. Biodegradation, 16(4), 319–327. https://doi.org/10.1007/s10532-004-1842-6
Shipped, F. (n.d.). The Top 5 indoor air quality monitors for a healthy home. Filters Shipped. https://filtersshipped.com/blogs/filtersshipped-blogs/the-top-5-indoor-air-quality-monitors-for-a-healthy-home?srsltid=AfmBOopReZWcsyhJoijanUQ3l5m8fz19CMYFNWy5rxrLljYzayg_3Sa9
Silva, V. P., Moreira-Santos, M., Mateus, C., Teixeira, T., Ribeiro, R., & Viegas, C. A. (2015). Evaluation of Arthrobacter aurescens Strain TC1 as Bioaugmentation Bacterium in Soils Contaminated with the Herbicidal Substance Terbuthylazine. PLoS ONE, 10(12), e0144978. https://doi.org/10.1371/journal.pone.0144978
Singapore Furniture Industries Council (SFIC) Introduces Enhanced Sustainable Furniture Mark to Raise Industry-Wide Indoor Air Quality Standards. (2025). Retrieved November 12, 2025, from Tradelinkmedia.biz website: https://seab.tradelinkmedia.biz/publications/6/news/5986
Singapore Standards. (2016). Code of practice for indoor air quality for air-conditioned buildings (SS 554:2016). https://www.enviresearch.co.th/wp-content/uploads/2020/01/Indoor-Air-2016.pdf
Singapore Standards. (n.d.). Retrieved from www.singaporestandardseshop.sg website: https://www.singaporestandardseshop.sg/Product/SSPdtDetail/8ee48ab1-38f5-4dae-a469-b8612a05876f
Singh, A., Shahid, M., Srivastava, M., Pandey, S., Antima, Sharma, & Kumar, V. (2014). Optimal physical parameters for growth of trichoderma species at varying pH, temperature and agitation. Virology & Mycology, 03(01), Article 1000127. https://doi.org/10.4172/2161-0517.1000127
Siswanto, D., Permana, B. H., Treesubsuntorn, C., & Thiravetyan, P. (2020). Sansevieria trifasciata and Chlorophytum comosum botanical biofilter for cigarette smoke phytoremediation in a pilot-scale experiment—evaluation of multi-pollutant removal efficiency and CO2 emission. Air Quality, Atmosphere & Health, 13(1), 109-117. doi:10.1007/s11869-019-00775-9
Song, M., Lin, X., Wei, X., Zeng, Q., Mu, C., & Zhou, X. (2024). Trichoderma viride improves phosphorus uptake and the growth of Chloris virgata under phosphorus-deficient conditions. Frontiers in Microbiology, 15, 1425034. https://doi.org/10.3389/fmicb.2024.1425034
Sriprapat, W., Boraphech, P., & Thiravetyan, P. (2014). Factors affecting xylene-contaminated air removal by the ornamental plant Zamioculcas zamiifolia. Environmental Science and Pollution Research, 21(4), 2603-2610. doi:10.1007/s11356-013-2175-y
Stoica, I., Petrovici, A. R., Silion, M., Varganici, C., Dinica, R. & Bahrim, G. (2017). Corn Cob Hydrolyzates used for Microbial Biosynthesis of Polyhydroxybutyrate. Cellulose Chemistry and Technology, 52(1-2), 65–74. https://cellulosechemtechnol.ro/pdf/CCT1-2(2018)/p.65-74.pdf
Stülke, J., Grüppen, A., Bramkamp, M., & Pelzer, S. (2023). Bacillus subtilis, a Swiss Army Knife in Science and Biotechnology. Journal of Bacteriology, 205(5), e0010223. https://doi.org/10.1128/jb.00102-23
Swain, S. S., Nayak, S., Mishra, S., Ghana, M., & Dash, D. (2025). Exploring the plant growth promoting attributes of pteridophyte-associated microbiome for agricultural sustainability. Physiology and Molecular Biology of Plants, 31(2), 211–232. https://doi.org/10.1007/s12298-025-01553-x
Takahashi, K., Hayashi, K., & Kinoshita, T. (2012). Auxin Activates the Plasma Membrane H+-ATPase by Phosphorylation during Hypocotyl Elongation in Arabidopsis. PLANT PHYSIOLOGY, 159(2), 632–641. https://doi.org/10.1104/pp.112.196428
Tan, V., Lim, J., Katika Akksilp, Wai Leng Chow, Ma, S., & Chen, C. (2023). The societal cost of modifiable risk factors in Singapore. BMC Public Health, 23(1). https://doi.org/10.1186/s12889-023-16198-2
Taupp, M., Heckel, F., Harmsen, D., & Schreier, P. (2006). Direct trapping of formaldehyde formed via oxidative N-demethylation of N,N-dialkylarylamines by Bacillus megaterium using cysteamine derivatization. Journal of Microbiological Methods, 67(2), 357–362. https://doi.org/10.1016/j.mimet.2006.04.006
Tran, V. V., Park, D., & Lee, Y.-C. (2020). Indoor Air Pollution, Related Human Diseases, and Recent Trends in the Control and Improvement of Indoor Air Quality. International Journal of Environmental Research and Public Health, 17(8), 2927. https://doi.org/10.3390/ijerph17082927
Treesubsuntorn, C., & Thiravetyan, P. (2012). Removal of benzene from indoor air by Dracaena sanderiana: Effect of wax and stomata. Atmospheric Environment, 57, 317. doi:https://doi.org/10.1016/j.atmosenv.2012.04.016
Treesubsuntorn, C., & Thiravetyan, P. (2018). Botanical biofilter for indoor toluene removal and reduction of carbon dioxide emission under low light intensity by using mixed C3 and CAM plants. Journal of Cleaner Production, 194, 94-100. doi:https://doi.org/10.1016/j.jclepro.2018.05.141
Turała, A., & Wieczorek, A. (2020). Biomass growth and its control in the process of biofiltration of air contaminated with xylene on a biotrickling column filled with expanded clay. Sustainability, 12(13), 5412.
Understanding VOCs and indoor air risks in Singapore. (n.d.). Formaldehyde Removal Specialist. https://ucfreshair.com/what-are-vocs-volatile-organic-compounds
US EPA, O. (2022, February 25). Guide to Air Cleaners in the Home. Retrieved from www.epa.gov website: https://www.epa.gov/indoor-air-quality-iaq/guide-air-cleaners-home
US EPA (2025). Volatile organic compounds’ impact on indoor air quality. (2025, July 24). US EPA. https://www.epa.gov/indoor-air-quality-iaq/volatile-organic-compounds-impact-indoor-air-quality
Vandana, U. K., Rajkumari, J., Singha, L. P., Satish, L., Alavilli, H., Sudheer, P. D., Chauhan, S., Ratnala, R., Satturu, V., Mazumder, P. B., & Pandey, P. (2021). The Endophytic Microbiome as a Hotspot of Synergistic Interactions, with Prospects of Plant Growth Promotion. Biology, 10(2), 101. https://doi.org/10.3390/biology10020101
Vardoulakis, S., Giagloglou, E., Steinle, S., Davis, A., Sleeuwenhoek, A., Galea, K. S., … Crawford, J. O. (2020). Indoor Exposure to Selected Air Pollutants in the Home Environment: A Systematic Review. International Journal of Environmental Research and Public Health, 17(23). https://doi.org/10.3390/ijerph17238972
Vidal-Quist, J. C., Rogers, H. J., Mahenthiralingam, E., & Berry, C. (2013). Bacillus thuringiensiscolonises plant roots in a phylogeny-dependent manner. FEMS Microbiology Ecology, 86(3), 474–489. https://doi.org/10.1111/1574-6941.12175
Vorholt, J. A., Marx, C. J., Lidstrom, M. E., & Thauer, R. K. (2000). Novel Formaldehyde-Activating Enzyme in Methylobacterium extorquens AM1 Required for Growth on Methanol. Journal of Bacteriology, 182(23), 6645–6650. https://doi.org/10.1128/jb.182.23.6645-6650.2000
Wang, X., Tian, Z., Xi, Y., & Guo, Y. (2022). Identification of endophytic fungi with ACC deaminase-producing isolated from halophyte Kosteletzkya Virginica. Plant Signaling & Behavior, 17(1). https://doi.org/10.1080/15592324.2022.2152224
Watzinger, A., et al. (2021). Unravelling the process of petroleum hydrocarbon biodegradation in different filter materials of constructed wetlands by stable isotope fractionation and labelling studies. Biodegradation, 32(3), 343–359.
Weyens, N., Van Der Lelie, D., Taghavi, S., & Vangronsveld, J. (2009). Phytoremediation: plant–endophyte partnerships take the challenge. Current Opinion in Biotechnology, 20(2), 248–254. https://doi.org/10.1016/j.copbio.2009.02.012
Wolverton BC, Wolverton JD (1993) Plants and soil microorganisms: removal of formaldehyde, xylene, and ammonia from the indoor environment. J Miss Acad Sci 38:11–15
Wongbunmak, A., Khiawjan, S., Suphantharika, M., & Pongtharangkul, T. (2020). BTEX biodegradation by Bacillus amyloliquefaciens subsp. plantarum W1 and its proposed BTEX biodegradation pathways. Scientific Reports, 10(1), 17408. https://doi.org/10.1038/s41598-020-74570-3
World Health Organization. (2025). Air quality, energy and health. Retrieved from World Health Organization website: https://www.who.int/teams/environment-climate-change-and-health/air-quality-energy-and-health/health-impacts
Yang, D. S., Pennisi, S. V., Son, K., & Kays, S. J. (2009). Screening indoor plants for volatile organic pollutant removal efficiency. HortScience, 44(5), 1377–1381. https://doi.org/10.21273/hortsci.44.5.1377
Ye, Z., An, T., Yang, X., Kang, H., & Wang, F. (2025). Novel model for stomatal conductance: Enhanced accuracy under variable irradiance and CO₂ in C3 plant species. Biology, 14(11), 1501. https://doi.org/10.3390/biology14111501
Yin, T., Gao, W., Wei, X., Wu, L., & Wang, C. (2024). Dynamics of indoor volatile organic compounds and seasonal ventilation strategies for residential buildings in Northeast China. Indoor and Built Environment. https://doi.org/10.1177/1420326x241280680
You, Y., Shim, J., Cho, C., Ryu, M., Shea, P. J., Kamala‐Kannan, S., Chae, J., & Oh, B. (2012). Biodegradation of BTEX mixture by Pseudomonas putida YNS1 isolated from oil‐contaminated soil. Journal of Basic Microbiology, 53(5), 469–475. https://doi.org/10.1002/jobm.201200067
Yu, W., in `t Veld, M., Bossi, R., Ateia, M., Tobler, D., Feilberg, A., … Johnson, M. S. (2021). Formation of Formaldehyde and Other Byproducts by TiO₂ Photocatalyst Materials. Sustainability, 13(9), 4821. https://doi.org/10.3390/su13094821
Zhang, S., Li, F., Chang, L., Zhang, Q., Zhang, Z., Wu, T., Wu, Y., Wang, Z., Su, Y., Yang, X., Gao, W., Li, M., Wu, Y., & Jiang, Y. (2025). Broad-spectrum applications of plant growth-promoting rhizobacteria (PGPR) across diverse crops and intricate planting systems. Microbiology Spectrum, 13(3), e0187924. https://doi.org/10.1128/spectrum.01879-24
Zhang, Z.-F., Zhang, X., Zhang, X., Liu, L.-Y., Li, Y.-F., & Sun, W. (2020). Indoor occurrence and health risk of formaldehyde, toluene, xylene and total volatile organic compounds derived from an extensive monitoring campaign in Harbin, a megacity of China. Chemosphere, 250, Article 126324. https://doi.org/10.1016/j.chemosphere.2020.126324
Zhao, H., Zhang, L., Liu, M., Wang, X., & Oljira, A. M. (2025). Trichoderma Combined with 1-Aminocyclopropane-1-carboxylic acid (ACC) Soil Amendments Modulates the Root Microbiome and Improves Wheat Growth Under Salinity Stress. Plant Stress, 15, 100785. https://doi.org/10.1016/j.stress.2025.100785
Zhong, L., Haghighat, F., & Lee, C.-S. (2013). Ultraviolet photocatalytic oxidation for indoor environment applications: Experimental validation of the model. Building and Environment, 62, 155–166. https://doi.org/10.1016/j.buildenv.2013.01.009
Zhu, L., Guan, D., Luo, J., Rathinasabapathi, B., & Q, L., MA. (2014). Characterization of arsenic-resistant endophytic bacteria from hyperaccumulators Pteris vittata and Pteris multifida. Chemosphere, 113, 9–16. https://doi.org/10.1016/j.chemosphere.2014.03.081

Zuraimi, M. S., Tham, K. W., & Sekhar, S. C. (2004). A study on the identification and quantification of sources of VOCs in 5 air-conditioned Singapore office buildings. Building and Environment, 39(2), 165–177. https://doi.org/10.1016/j.buildenv.2003.08.013

Appendix

Appendix 1: Timeline of Project

Appendix 2: Natede Report

The Natede (2021) report evaluates the performance of the Natede Smart air purification system, which integrates plant-based phytoremediation with photocatalytic oxidation using a tungsten trioxide (WO₃) coated filter and LED light activation. The study primarily investigates the system’s ability to remove volatile organic compounds (VOCs), specifically methyl ethyl ketone (MEK), as well as formaldehyde and airborne microbial contaminants. The plant species used in the system is Sansevieria trifasciata, and the objective is to assess how effectively the combined biological and photocatalytic processes enhance indoor air purification compared to a plant alone.

Experiments were conducted under both controlled chamber and full-scale room conditions. In the controlled setup, a sealed chamber (200 cm × 50 cm × 45 cm) was used, where MEK was introduced and allowed to saturate the environment before measurements were taken at 0, 1, and 2 hours using activated carbon sampling and UV-Vis spectrometry (NIOSH 2500 standard). In the full-scale setup, tests were carried out in a closed room of approximately 33 m³, where 30 mL of MEK was evaporated, and VOC concentrations were measured after 24 and 48 hours. Additional experiments evaluated microbial removal using high bacterial loads (~1000 × 10⁶ CFU), as well as the standalone performance of the photocatalytic system and its effectiveness in removing formaldehyde using standardised air sampling methods.

The results demonstrate that the Natede Smart system achieves significantly higher pollutant removal efficiencies compared to a plant alone. In the controlled chamber, approximately 99.5% of VOCs were removed within 2 hours. In the full-scale room, more than 75% of VOCs were removed within 24 hours and approximately 94% after 48 hours, whereas a plant in a standard pot achieved only about 11% removal over the same period. Similarly, microbial removal reached 93% in 24 hours and 99.4% in 48 hours, compared to 36% for the plant alone. The photocatalytic system also demonstrated rapid pollutant degradation, achieving 80–100% reduction within 1 hour, significantly outperforming natural decay rates of 20–30% over 6 hours. For formaldehyde, removal efficiencies of up to 85% were observed within 1 hour.

Overall, the study highlights the strong synergistic effect between plant-based systems and engineered enhancements such as airflow and photocatalysis. The Natede system was estimated to achieve an air purification effect equivalent to approximately seven plants, demonstrating the importance of enhancing pollutant contact with the rhizosphere and accelerating degradation processes. These findings reinforce the role of sustained exposure and system design in achieving effective VOC removal, supporting the use of controlled experimental setups and combined biological systems for evaluating air purification performance.

Appendix 3: Experimental Considerations and Modifications

Sensitivity of Sensor

To ensure reliable data collection, the Awair Element air quality monitor was selected for use in this study. The Awair Element is recognized as one of the most accurate and reliable commercially available indoor air-quality sensors, capable of detecting a wide range of pollutants including VOCs, CO₂, and particulate matter (Shipped F, n.d.).

Calibration of the sensors was conducted under outdoor ambient conditions for a continuous 24-hour period to establish a stable and accurate baseline. This baseline provides a reference for detecting changes in VOC levels within the enclosed chamber during subsequent experiments. To enhance calibration consistency, two Awair Element sensors were placed facing each other during the calibration process as shown in Figure S2. This arrangement ensured that both devices sampled the same air mass, minimizing discrepancies due to airflow direction or localized variations in pollutant concentration.

Following outdoor calibration, the sensors were allowed to stabilize briefly under indoor conditions before each experimental run. This step ensured that all sensors started from comparable baseline readings, improving measurement precision and repeatability across tests.

Chamber Sizing and Modification

The chamber size was selected based on several practical and methodological considerations. First, the height of the chamber was chosen to accommodate the combined height of the plant and its container, ensuring that the plant remained upright and unobstructed during the experiment. Second, the base area of the chamber had to be sufficiently large to position both the pollutant source and the sensor far apart, minimizing the risk of the sensor detecting artificially elevated VOC levels from direct release rather than from the ambient chamber environment. At the same time, the chamber could not be excessively large, as a greater internal volume would dilute the VOC concentration and prolong the time needed to reach measurable steady-state levels. Therefore, a moderate-sized transparent chamber (H: 56 cm × B: 39 cm × L: 42 cm) was chosen to balance spatial requirements, measurement accuracy, and experimental efficiency.

To accommodate instrumentation, a small orifice was drilled at the base of the chamber (Figure S3), allowing the computer fan and sensors to connect to external power sources while minimizing leakage of the sealed air. This setup ensures the chamber remains effectively enclosed, so that any observed changes in VOC concentration can be attributed to the plant or plant–microbe system rather than external influences.

Plant Container Modification: Airflow to Plant’s Rhizosphere

To optimize pollutant exposure to the plant roots, the container was modified by cutting an opening and installing a computer fan, which directs airflow from the pollutant source toward the LECA and root zone (Figure S4). This arrangement ensures that VOCs pass through the rhizosphere, maximizing contact with both the plant and microbial components.

Plant Properties: Growth Medium

Lightweight Expanded Clay Aggregate (LECA) was selected as the growth medium due to its high porosity, excellent aeration properties, and inert composition, making it particularly suitable for indoor plant systems and controlled experiments (Fussy & Papenbrock, 2022). The expanded clay structure contains numerous internal pores and inter-particle voids that facilitate continuous airflow throughout the medium (Watzinger et al., 2021). This enhances oxygen availability to the rhizosphere, preventing anaerobic conditions and supporting healthy root respiration (Białowiec et al., 2012). Improved aeration is especially critical in enclosed experimental setups where natural air circulation is limited, as it ensures that VOC-laden air can effectively reach the root zone for treatment (Mikkonen et al., 2018).

For indoor applications, LECA offers further advantages due to its clean, inorganic, and low-maintenance nature. Unlike organic soil, it does not decompose, harbor pests, or release organic compounds that could interfere with sensitive VOC measurements (Armijos-Moya et al., 2021; Turała & Wieczorek, 2020). This makes it particularly suitable for studies investigating air quality and specific pollutant removal. Moreover, the high surface area provided by its porous matrix supports robust microbial attachment and biofilm growth, which is essential in plant–microbe systems where rhizospheric microorganisms drive the degradation of airborne contaminants (Dordio & Carvalho, 2013; Mikkonen et al., 2018).

LECA balls were carefully washed prior to use to remove any surface contaminants that could interfere with the results. The container was filled with LECA to approximately 80% of its volume, maintaining uniform growth medium across all setups (Figure S5).

Plant Properties: Amount of Water

A volume of 250 mL of water was added to each container to ensure consistent hydration, and the same volume of microbial solution was applied to the plant–microbe setups by carefully pouring it around the root zone. This ensured that the microbes were evenly distributed across the entire root system, allowing them to effectively colonize the rhizosphere and enhance VOC degradation by metabolizing compounds absorbed by the roots, thereby improving overall removal efficiency (Figure S5).

Plant Properties: Amount of Light

Light exposure was standardised across all experiments to eliminate variability arising from differences in ambient lighting conditions. Each experimental setup was enclosed using an opaque trash bag to block external light sources, ensuring a controlled lighting environment within the chamber. A fixed artificial light source was then positioned at a consistent distance and orientation relative to the plant for all trials. This approach ensured uniform light intensity and exposure duration across experiments, minimising the influence of light variability on plant physiological activity and, consequently, on VOC removal performance (Figure S6).

Appendix 4: Calculation of Amount of Pollutant

Based on Natede (2023), who used 30 cm³ of pollutant in a 33 m³ chamber, the corresponding pollutant volume per cubic metre is approximately 0.909 cm³/m³. For the experimental chamber in this study (0.0917 m³; 56 cm × 39 cm × 42 cm), the equivalent pollutant volume is:

0.0917 m³ × 0.909 cm³/m³ ≈ 0.083 cm³

To convert the calculated pollutant volume into the corresponding mass for each target VOC, the density of each compound must be considered (Table 6).

Table 6: Densities of FBTEX compounds
Compound	Density (g/cm³)
Formaldehyde	0.8153
Benzene	0.8765
Toluene	0.8623
Ethylbenzene	0.867
Xylene (mixed isomers)	~ 0.86–0.88

Assuming equal composition of the different compounds, the average density is used for simplicity in calculations (approximately 0.856 g/cm³). The mass of the pollutant required can be calculated using the standard relation:

Mass (g) = Volume (cm³) × Density (g/cm³)

Mass ≈ 0.083 cm³ × 0.856 g/cm³ ≈ 0.071

Table 7 summarizes the densities, VOC content, and corresponding mass of VOC per unit volume for each pollutant used in the pilot study. By multiplying the density by the VOC fraction, the effective mass of VOCs per unit volume of each pollutant was calculated.

Table 7: Densities of Pollutants
Pollutant	Density (g/cm³)	VOC (% by weight)	Mass of VOC per unit volume of pollutant (g/cm³)
Elmer’s White Glue	1.07	<1–10%	0.0535
Aureo’s Liquid Shoe Polish	0.754	15%	0.1131
Nippon Vinilex-5000 Paint	1.43	<2% (low-VOC)	0.0215

To determine the appropriate volume of each pollutant to achieve the target mass of VOCs in the chamber, the Solver function in Microsoft Excel was employed. Arbitrary initial volumes (0.5cm3 each) were assigned to each pollutant, and the corresponding mass of VOCs was calculated based on the density and VOC content (as shown in Table X). These values were then iteratively adjusted using the Solver function to match the target total VOC mass of 0.071 g, calculated previously for the chamber volume. The resulting volumes derived from this optimization are shown in Table 8.

Table 8: Volume required using Excel solver
Pollutant	Volume Required (cm3)
Elmer’s White Glue	0.342
Aureo’s Liquid Shoe Polish	0.470
Nippon Vinilex-5000 Paint	0.425

It should be noted that the volumes determined using the Excel Solver are dependent on the initial concentrations specified prior to the optimization. By testing multiple initial concentration values, it was observed that starting with an initial guess of 0.5 cm³ for each pollutant resulted in a more balanced distribution of volumes among the three sources. This outcome aligns with practical considerations for the experimental procedure, as it facilitates easier measurement and handling of the pollutants while maintaining consistency across experimental runs.

Appendix 5: Summary of Research Studies of VOC Removal Efficiency of Shortlisted Plants and Microbe

Research findings for Sansevieria trifasciata (Snake Plant), Epipremnum aureum (Money Plant), Zamioculcas zamiifolia (ZZ Plant), and Dracaena fragrans (Dragon Plant).

Pollutant	Plant	Pollutant Concentration	Removal Rate/Efficiency	Sources
Formaldehyde	Sansevieria trifasciatavar 'Laurentii' > Chlorophytum comosum 'Mediopictum' > Aloevera 'Chinensis' > Scindapsus aureum	10uL 40% formaldehyde solution	1.76, 1.27... mg/h/m2	(Meng et al., 2012)
Toluene	C3 Chlorophytum comosum + CAM Sansevieria trifasciata > facultative CAM Zamioculcas zamiifolia, C3 Dracaena sanderiana, CAM cycling Euphorbia milii, CAM Sansevieria kirkii	3.92 mg/m3	1 m3 / 2–3h	(Treesubsuntorn et al., 2018)
Formaldehyde	CAM S. trifasciata + C3 Chlorophytum comosum	120–150 ppm	80–90%	(Siswanto et al.)
Benzene	CAM S. trifasciata + C3 Chlorophytum comosum	15–35 ppb	80–90%	(Siswanto et al.)
Xylene	CAM S. trifasciata + C3 Chlorophytum comosum	30–70 ppb	80–90%	(Siswanto et al.)
TVOCs	Epipremnum aureum 40%, Nephrolepis exaltata 3%, Peperomia obtusifolia 10%, Schefflera arboricola 5% and Spathiphyllum wallisii 42% (total 96)	120 ppb	72.50%	(Pettit, Irga, et al., 2019)
TVOCs		300 ppbv	28%	(Pettit, Irga, et al., 2019)
Xylene	Zamioculcas zamiifolia > Aglaonema commutatum > Philodendron martianum / Sansevieria Hyacinthoides / Aglaonema rotundum / Fittonia albivenis > Alternanthera bettzickiana, Drimiopsis botryoides, Aloe vera, Chlorophytum comosum, Cordyline fruticosa, Muehlenbeckia platyclada, Tradescantia spathacea, Guzmania lingulate, Cyperus alternifolius	20 ppm / 12 umol	0.86 mmol/m2 / 88%	(Sriprapat et al., 2014)
Xylene	Facultative CAM Z. zamiifolia + CAM S. hyacinthoides + C3 A. commutatum in dark	20 ppm / 12 umol	0.29 mmol/m2	(Sriprapat et al., 2014)
Benzene	Zamioculcas zamiifolia, Aglaonema modestum	25 ppm	100%, 75% to below 100 ppb	(Tarran et al., 2008)
BTEX	Well-watered Zamioculcas zamiifolia	20 ppm	0.86–0.96 mmol/m2	(Sriprapat et al., 2013)
Formaldehyde, xylene	Aechmea fasciata, Aglaonema ‘Silver Queen’, Aloe barbadensis, Anthurium andraeanum, Calathea ornata, Chamaedorea elegans, Chlorophytum comosum ’Vittatum’, Chrysanthemum morifolium, Cissus rhombifolia, Cyclamen persicum, Dendrobium sp., Dieffenbachia camille, Dieffenbachia ‘Exotica Compacta’, Dieffenbachia maculata, Dracaena deremensis ‘Janet Craig’, Dracaena deremensis ‘Warneckii’, Dracaena fragrans, Dracaena marginata, Euphorbia pulcherrima, Ficus benjamina, Ficus sabre, Guzmania ‘Cherry’, Hedera helix, Homalomena sp., Kalanchoë, Liriope spicata, Neoregelia cv., Nephrolepis exaltata ‘Bostoniensis’, Nephrolepis obliterata, Phalaenopsis sp., Phoenix roebelenii, Rhapis excelsa, Rhododendron indicum, Sansevieria trifasciata, Senecio cruentus, Spathiphyllum ‘Clevelandii’, Syngonium podophyllum, Tulip ‘Yellow Present’	N/A	47–1,863 ug/h	(Wolverton et al., 1993)
Benzene, pentane, toluene	Chlorophytum comosum, Dracaena deremensis, Ficus elastica, Kalanchoë blossfeldiana, Magnesia sp., Pelargonium domesticum, Primula sinensis, Saxifraga stolonifera, Tradescantia fluminensis	33,543	0.6–8.5 ug/g/day	(Cornejo et al., 1999)
Benzene	Dracaena deremensis ‘Janet Craig’, Dracaena marginata, Epipremnum aureum, Howea forsteriana, Schefflera actinophylla ‘Amate’, Spathiphyllum floribundum ‘Petite’, Spathiphyllum floribundum ‘Sensation’	80,000	59–337 ppm/day/m2	(Orwell et al., 2004)

Research findings for Bacillus species microbes

Microbe	Target Pollutants	Source(s)
Bacillus amyloliquefaciens	BTEX; Formaldehyde from cooking oil fume condensates	Wongbunmak et al. (2020); Han et al. (2020)
Bacillus cereus	Toluene, Ethylbenzene, Formaldehyde (used in conjunction with plants)	Lee et al. (2013); Daudzai et al. (2018); Khaksar et al. (2016)
Bacillus megaterium	Benzene, Toluene, Xylene, Formaldehyde	Dolphen et al. (2019); Dhanya (2019); Taupp (2006)
Bacillus subtilis	Benzene, Toluene, and Xylene	Lan et al. (2020)
Bacillus thuringiensis	Benzene, Toluene	Ehmedan et al. (2021); Lee et al. (2013); Kesavan et al. (2021)

Appendix 6: Plant Selection Research

Plant	Ability to clear VOC	Procurement	Compatibility with Singapore's Indoor Conditions		Ease of Maintenance
			Optimal Temperature Tolerance (20–24 °C)	Humidity Tolerance (50–70%)	Size of plant (<30cm height)	Ability to grow in LECA Balls
			Optimal Temperature Tolerance (20–24 °C)	Humidity Tolerance (50–70%)	Size of plant (<30cm height)	Moisture Content Required (moderate/low)	pH Range Tolerance (pH 6.0 - 7.5)	Root System Characteristics (compact/small root system)
Epipremnum aureum (Money plant/Pothos/ Golden Pothos / Devil's Ivy)	4/5 Benzene: 31.4% [Li] formaldehyde: 233 m3/ m2 bed/h benzene: 1.10, 0.85, 0.27... μg/m3/cm2 [Tanvi] formaldehyde / benzene / toluene / xylene: 2.5–34 V/h [Dela Cruz]	—	—	—	—	—	—	—
Sansevieria trifasciata / Dracaena trifasciata (Snake plant)	4/5 Formaldehyde: 80-90% for 120–150 ppm Benzene: 80-90% for 15–35 ppb Xylene: 80-90% for 30–70 ppb Toluene: 1 m3 / 2-3h for 3.92 mg m−3 [Tanvi]	—	—	—	—	—	—	—
Spathiphyllum wallisi (Peace Lily)	2/5 toluene: 5.7 L/h/m2 for 20 mg/m3 conc benzene / toluene: 50.6-57.5 ng/m3/h/cm2 leaf benzene for 0.5 uL/L conc [Tanvi] benzene: 367–4,032 mg/m3/day/m2 benzene / toluene: 0–13.28 μg/m3/m2/h [Dela Cruz]	—	—	—	—	—	—	—
Chlorophytum comosum (Spider Plant)	4/5 formaldehyde: 2.21–4.60 mg/m3 in 7 days [Dela Cruz] toluene: 1 m3 / 2-3h for 3.92 mg m−3 conc benzene: 80-90% for 15–35 ppb conc xylene: "0.86 mmol m−2 / 88% ; 0.82 mmol m−2 / 84%" for 20 ppm / 12 μmol conc [Tanvi]	—	—	—	—	—	—	—
Zamioculcas zamiifolia (ZZ plant)	5/5 Benzene / toluene / ethylbenzene / xylene: 86–96 mmol/m2 in 120 h 28–68 mmol/m2 in 7 days for 62,392–84,806 conc formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 conc [Dela Cruz]	—	—	—	—	—	—	—
Dracaena fragrans (Dragon plant)	2/5 Benzene: 43–77 % in 72 h Formaldehyde: 2.21–4.60 mg/m3 in 7 days (for 15,000μg/m3 of VOC concentration) [Dela Cruz]	—	—	—	—	—	—	—
Dracaena reflexa / Dracaena angustifolia (Red-Edged Dracaena)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days (for 15,000μg/m3 of VOC concentration) [Dela Cruz]	—	—	—	—	—	—	—
Dracaena sanderiana (Lucky bamboo)	3/5 Benzene: 43–77 % in 72 h Formaldehyde: 2.21–4.60 mg/m3 in 7 days (for 15,000μg/m3 of VOC concentration) [Dela Cruz] toluene [Tanvi]	—	—	—	—	X	—	—
Aglaonema commutatum (Chinese evergreen) (Philippine evergreen)	5/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days (for 15,000μg/m3 of VOC concentration) [Dela Cruz] Benzene: ~20μg/cm3 in 72h Toluene: ~30μg/cm3 in 72h \|\| 1.46 ± 0.13 μmol in 72 h Xylene: ~30μg/cm3 in 72h \|\| 0.65 ± 0.03 mmol d−1 cm−2-leaf area Ethylbenzene: ~35μg/cm3 in 72h \|\| 1.65 ± 0.04 [TH_1]	—	—	—	—	—	—	—
Ficus elastica (Rubber Plant)	5/5 benzene / xylene / toluene: 28–91 % in 12 h for 15,116 conc benzene / toluene: 0.6–8.5 μg/g/day for 33,543 conc [Dela Cruz] Formaldehyde: ~50% Ethylbenzene: ~75% [https://pmc.ncbi.nlm.nih.gov/articles/PMC5480428/?utm]	—	—	—	—	—	—	x
Ficus Benjamina (Weeping fig)	5/5 Formaldehyde: 534.4μg/cm3 in 72h Toluene: 45.9μg/ cm3 in 72h Xylene: ~5μg/ cm3 in 72h benzene: ~20μg/ cm3 in 72h ethylbenzene: ~10μg/ cm3 in 72h [https://www.frontiersin.org/journals/built-environment/articles/10.3389/fbuil.2021.803516/full]	—	—	—	x	—	—	x
Rhapis excelsa (lady palm)	2/5 formaldehyde / xylene: 47–1,863 μg/h [Dela Cruz]	—	—	—	x	—	—	x
Chamaedorea elegans (Parlour Palm)	2/5 formaldehyde: 1.47 mg/m2/h for 4.6 mg/m3 conc [Tanvi] formaldehyde: 660 μg/h Xylene: 223 μg/h [Wolverton, B. C. and J. D. Wolverton. (1993)]	—	—	—	x	—	—	—
Schefflera actinophylla / Heptapleurum actinophyllum (Australia umbrella tree)	3/5 Benzene: 59–337 ppm/day/m−2 (for 15,000μg/m3 of VOC concentration) [Dela Cruz] Toluene: 13.3μg·m−3·m−2 leaf area over a 24-h period Xylene: 7.0 μg·m−3·m−2 leaf area over a 24-h period [Kim, K. J. et al. (2016)]	—	—	—	x	—	—	x
Nephrolepis Exaltata (Boston Fern)	4/5 Benzene: 59–724.9 mg/m2/day for 479μg/m3 conc Formaldehyde / toluene / xylene: 8–42 V/h for 3,805–9,920 μg/m3 conc [Dela Cruz]	—	—	—	—	x	—	—
Nephrolepis obliterata (Kimberly Queen Fern)	1/5 formaldehyde: 90-100% for 0.6-11mg/m^3 conc [https://pmc.ncbi.nlm.nih.gov/articles/PMC6106135/]	—	—	—	x	x	—	x
Osmunda japonica (Japanese Royal Fern)	1/5 Formaldehyde: >1.87 μg/m3/cm2 for 7 μL/L = 2472 μg/m3 [Tanvi]	—	—	—	x	x	—	x
Pteris multifida (Silver Brake fern)	1/5 Formaldehyde: >1.87 μg/m3/cm2 for 8 μL/L = 2472 μg/m3 [Tanvi]	—	—	—	—	—	—	—
Pteris dispar	1/5 Formaldehyde: >1.87 μg/m3/cm2 for 7 μL/L = 2472 μg/m3 [Tanvi]	X	—	—		—	—	—
Davallia mariesii (squirrel's foot fern)	2/5 toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 conc [Dela Cruz] formaldehyde: >1.87 μg/m3/cm2 for 2 μL/L = 2472 μg/m3 conc [Tanvi]	x	—	—		—	—	—
Polypodium formosanum (caterpillar fern)	1/5 Formaldehyde: >1.87 μg/m3/cm2 for 4 μL/L = 2472 μg/m3 [Tanvi]	x	—	—		—	—	—
Selaginella tamariscina (spike moss)	1/5 Formaldehyde: >1.87 μg/m3/cm2 for 2 μL/L = 2472 μg/m3 [Tanvi]	x	—	—		x	—	x
Tradescantia pallida (purple hearts)	2/5 Benzene / Toluene [ResearchGate]	—	—	—	—	—	—	x
Hedera helix (English Ivy)	4/5 formaldehyde: 2.22–25.06 mg/m2/h ; 81–96 % in 24 h Benzene: 1,555–107,653 μg in 24 h ; 9.2–89.8 % in 24 h Toluene: −4.3 to 950.3 μg/m3/h/m2 (for 5,000μg/m3 of VOC concentration) Xylene: 47–1,863 μg/h [Dela Cruz]	—	x	—	—	—	—	x
Euphorbia milii (Christ plant)	3/5 Benzene: 70.23% for 20 ppm benzene in 96 h formaldehyde: 100.00% for 20 ppm formaldehyde in 48 h [Li] Xylene: 54.7% +- 2.5 of Xylene in 24h [ScienceDirect]	—	—	x	—	—	—	—
Anthurium andraeanum (Flamingo Lily)	2/5 formaldehyde: removal efficiency over 5 hours exposure of about 1.22 mg·m⁻³·cm⁻² leaf area (after ~5 h exposure) [Green Plants for Green Buildings] Benzene: 18% [Li]	—	—	—	—	—	—	—
Crassula Ovata / Crassula portulacea (Jade Plant)	1/5 benzene: 503.4, 203.9, 185.7… ug/m2/min [Tanvi]	—	—	x	—	—	—	—
Dianella caerulea (Flax Lilly)	1/5 Benzene: 59.04% for 2.5 mL of 1/71 g/L in 0.6 m3 [Tanvi]	X	—	—		—	—	X
Echinopsis tubiflora (Easter lily cactus)	1/5 Benzene: 50-80 % at 0.2ppm conc. [Tanvi]	—	—	X	—	—	—	—
Gerbera jamesonii (Gerbera daisy, baberton daisy)	1/5 Benzene: 23.5, 18.2, 10.4, 10… mg/cm2 leaf for 15-20 ppm conc [Tanvi]	—	—	—	—	—	—	X
Nematanthus glabra (Goldfish plant)	1/5 Benzene: 58.78% for 4.170 ppm conc [Tanvi]	—	—	—	—	—	—	—
Pelargonium graveolens (Rose Geranium)	2/5 Benzene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	x
Pelargonium domesticum / Pelargonium graveolens (Regal Geranium)	Benzene / toluene [Dela Cruz]	x	—	—		—	—	—
Chrysanthemum morifolium (pot mum)	Formaldehyde: 96%, 94%, 88%, 84% T2/3 of 23, 30, 34 and 56 min for 2000 μg/m3 conc [Tanvi] Benzene / toluene [Allergy & Air]	x	x	x		—	—	—
Psidium guajava (common guava, yellow guava)	1/5 Formaldehyde: >1.87 μg/m3/cm2 for 5 μL/L = 2472 μg/m3 conc [Tanvi's thesis] Formaldehyde: At an initial 2.0 µL·L⁻¹ formaldehyde, P. guajava removed 2.39 µg·m⁻³·cm⁻² of leaf area over 5 hours. [HortSci / ASHS]	—	x	—	x	—	—	—
Musa Ornata (Banana flowering plant)	2/5 Formaldehyde: 11.7, 10.9, 9.8 mg/cm2 leaf for 15-20 ppm conc [Tanvi] Benzene / formaldehyde: 1,555–107,653 μg in 24 h 9.2–89.8 % in 24 h [Dela Cruz]	X	—	—		—	x	—
Philodendron oxycardium / Philodendron hederaceum	2/5 Benzene / formaldehyde: 1,555–107,653 μg in 24 h 9.2–89.8 % in 24 h [Tanvi]	—	—	—	—	—	—	—
Philodendron martianum	2/5 formaldehyde: 2.21–4.60 mg/m3 in 7 days Xylene: 0.86 mmol m−2 / 88% ; 0.82 mmol m−2 / 84% [Tanvi]	—	—	—	—	—	—	—
Philodendron selloum / Thaumatophyllum bipinnatifidum	2/5 formaldehyde: 2.21–4.60 mg/m3 in 7 days [Tanvi] benzene / formaldehyde: 1,555–107,653 μg in 24 h 9.2–89.8 % in 24 h [Dela Cruz]	—	—	—	x	—	—	x
Philodendron sodiroi	1/5 formaldehyde: 2.21–4.60 mg/m3 in 7 days [Tanvi]	—	—	—	—	—	—	x
Phoenix roebelenii (Dwarf Date Plant / Pygmy Date Palm)	4/5 Formaldehyde: 1385 µg removed per hour Xylene: 610 µg removed per hour [Journal of Mississippi Academy of Science] Benzene: 2.5–34 V/h for 5,650–9,787(μg/m3) Toluene: 2.5–34 V/h for 5,650–9,787(μg/m3) [Dela Cruz]	—	—	—	x	x	—	—
Hemigraphis alternata / Strobilanthes alternata / Hemigraphis colorata (Red Ivy, Red Flame Ivy, Metal Leaf)	4/5 Benzene / Toluene / xylene: 28–91 % in 12 h for 15,116 (μg/m3) formaldehyde: 700 μg/m3 in 5 h for 2,400 (μg/m3) [Tanvi]	—	—	—	—	—	—	—
Fatsia japonica (paperplant)	4/5 Benzene / Toluene / xylene: 28–91 % in 12 h for 15,116 (μg/m3) formaldehyde: 700 μg/m3 in 5 h for 2,400 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	x
Kalanchoë blossfeldiana (Flaming Katy)	2/5 Benzene: 0.6–8.5 μg/g/day for 33543 (μg/m3) Toluene: 0.6–8.5 μg/g/day for 33543 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	—
Primula sinensis (Chinese Primrose)	2/5 Benzene: 0.6–8.5 μg/g/day for 33543 (μg/m3) Toluene: 0.6–8.5 μg/g/day for 33543 (μg/m3) [Dela Cruz]	X	X	—		X	—	—
Saxifraga stolonifera (Strawberry Begonia)	2/5 Benzene: 0.6–8.5 μg/g/day for 33543 (μg/m3) Toluene: 0.6–8.5 μg/g/day for 33543 (μg/m3) [Dela Cruz]	—	—	—	—	X	—	—
Tradescantia fluminensis (Wandering Dude)	2/5 Benzene: 0.6–8.5 μg/g/day for 33543 (μg/m3) Toluene: 0.6–8.5 μg/g/day for 33543 (μg/m3) [Dela Cruz]	—	—	—	—	—	x	—
Nicotiana tabacum (Tobacco)	3/5 Formaldehyde: 2.22–25.06 mg/m2/h in 2,500 (μg/m3) Toluene: −4.3 to 950.3 μg/m3/h/m2 in 5,000 (μg/m3) Benzene: 0–157 μg/d for 2,500–22,000,000 (μg/m3) [Dela Cruz]	x	—	—		x	x	x
Ardisia japonica / marlberry (Japanese ardisia)	3/5 Formaldehyde / xylene : 0.14–0.88 μg/m3/cm2 in 5h for 2,488 (μg/m3) Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	—	x	—
Stephanotis floribunda (Madagascar Jasmine)	2/5 Formaldehyde / xylene: 0.14–0.88 μg/m3/cm2 in 5h for 2,488 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	x
Syngonium podophyllum (Arrowhead vine)	4/5 Formaldehyde / xylene: 0.14–0.88 μg/m3/cm2 in 5h for 2,488 (μg/m3) Benzene / toluene / xylene: 28–91 % in 12 h for 15,116 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	x
Melissa officinalis (lemon balm)	2/5 Formaldehyde: 2.22–25.06 mg/m2/h for 2,500 (μg/m3) Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	x	—	—	—	—	—
Cymbidium Meglee	1/5 Formaldehyde: 3.4–6.6 mg/m3/h/m3 plant volume for 2,472 (μg/m3) [Dela Cruz]	x	x	—		x	x	—
Dendrobium phalaenopsis	1/5 Formaldehyde: 3.4–6.6 mg/m3/h/m3 plant volume for 2,472 (μg/m3) [Dela Cruz]	x	—	—		—	—	—
Sedirea japonica / Phalaenopsis japonica (Nagoran)	1/5 Formaldehyde: 3.4–6.6 mg/m3/h/m3 plant volume for 2,472 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Gardenia jasminoides (Cape Jasmine)	1/5 Formaldehyde: 3.4–6.6 mg/m3/h/m3 plant volume for 2,472 (μg/m3) [Dela Cruz]	—	—	—	—	—	x	—
Rosmarinus officinalis (rosemary)	2/5 Formaldehyde: 3.4–6.6 mg/m3/h/m3 plant volume for 2,472 (μg/m3) Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	x	—	—	—	x
Aloysia triphylla / Aloysia citrodora (Lemon Verbena)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	x
Ardisia crenata (Coral Berry)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Ardisia pusilla (dwarf ardisia)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	x	—	—		—	—	—
Begonia maculata (Polka dot begonia)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	—
Cinnamomum camphora (Camphor tree)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	X	—	—		X	—	X
Eurya emarginata	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	X	—	—		—	X	X
Farfugium japonicum (leopard plant)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	X	—	X
Ilex cornuta (Chinese Holly)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	x	—	—		x	x	—
Ligustrum japonicum (Japanese Privet)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	x	—	—		—	—	x
Mentha piperita (peppermint)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	x	—	x
Mentha suaveolens (apple mint)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	x	—	—	—	—	x
Pittosporum tobira (Japanese cheesewood)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	x
Plectranthus tomentosus / Coleus hadiensis / Plectranthus hadiensis (Vicks Plant)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	—
Pinus densiflora (Japanese red pine)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	x	—	x		x	x	—
Rhododendron fauriei (Korean Rhododendron)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	x	—	—		x	x	—
Soleirolia soleirolii (Baby's Tears)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	x	—	—		x	—	—
Salvia elegans (Pineapple Sage)	1/5 Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	x	—	—	—	x
Schefflera elegantissima / Plerandra Elegantissima (False Aralia)	2/5 Benzene: 0–13.28 μg/m3/m2/h for 30,900-59,100(μg/m3) Toluene: −4.3 to 950.3 μg/m3/h/m2 for 5,000 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Nerium indicum / Nerium oleander (Oleander)	1/5 Formaldehyde: 9.5–135 ng/dm2/h/ppb; 4–40 μg/dm2/ha for 43–300 (μg/m3) [Dela Cruz]	—	—	—	x	x	—	x
Citrus medica (Lemon Tree)	1/5 Benzene: 59–724.9 mg/m2/day for 479 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	x
Howea forsteriana (Kentia Palms)	1/5 Benzene: 59–337 ppm/day/m−2 for 80,000(μg/m3) [Dela Cruz]	—	—	—	x	x	—	—
Spathiphyllum floribundum (Snow Flower)	2/5 Benzene: 59–337 ppm/day/m−2 for 80,000 (μg/m3) Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Monstera acuminate (shingle plant)	1/5 Benzene: 43–77 % in 72 h for 62,392(μg/m3) [Dela Cruz]	—	—	—	—	x	—	x
Ipomoea batatas (sweet potato vines)	1/5 Formaldehyde: 3,371–4,759 μg in 7 h for 19,425-23,067(μg/m3) [Dela Cruz]	—	—	—	x	—	—	x
Asparagus densiflorus (asparagus fern)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	x
Aspidistra elatior (Cast Iron Plant)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	x
Calathea roseopicta (Rose painted calathea)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	x
Codiaeum variegatum (garden croton)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	x	—	x	—
Dieffenbachia seguine (Dumbcane Plant)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Fittonia albivenis (nerve plant)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	—
Howea belmoreana (Belmore Sentry Palm or Curly Palm)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	x	x	—	x
Hoya carnosa (Wax Plant)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Maranta leuconeura (prayer plant)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	—	x	x	x
Peperomia clusiifolia (red edge peperomia)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	—	x	—	—
Polyscias fruticosa (Ming aralia)	2/5 Benzene/ Toluene: 0–13.28 μg/m3/m2/h for 30,900-59,100 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	—
Aloe aristata (Lace Aloe)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	—	—	—	—	—
Agave potatorum (Butterfly Agave)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	x	—	—	—	—
Alocasia macrorrhiza (Giant Taro)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	—	x	—	—	—
Aloe nobilis (Golden Tooth Aloe)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	x	—	—	x	—
Asparagus setaceus (Plumosa Fern)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	—	—	x	—	—
Cordyline fruticosa (Ti Plant)	2/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz] Xylene: 0.86 mmol m−2 / 88% or 0.82 mmol m−2 / 84% for 20 ppm / 12 μmol [Tanvi's thesis]	—	—	—	x	x	—	—
Gasteria gracilis (Cow Tongue)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	x	—	—	—	—
Scindapsus pictus (satin pothos)	1/5 Formaldehyde: 2.21–4.60 mg/m3 in 7 days for 15,000 (μg/m3) [Dela Cruz]	—	—	x	—	—	—	—

Appendix 7: Plant Selection Scoring

Raw Values

Plant	Max VOC Level	Time Taken to Plateau (h)	Steady State VOC (ppb)
Epipremnum aureum (Money plant)	15.1219	8.75	1.3995
Zamioculcas zamiifolia (ZZ plant)	6.7942	7.0833	0.1835
Dracaena fragrans (Dragon plant)	2.7047	7.1667	0.0971
Sansevieria trifasciata (Snake plant)	3.7547	6.1667	0.1496
Fittonia albivenis (Nerve plant)	4.3406	7.3333	0.0843
Hemigraphis alternata (Red Ivy)	3.8715	7.0000	0.1502
Polyscias fruticosa (Ming aralia)	2.6182	6.4167	-0.0089
Philodendron oxycardium	3.5099	8.0000	0.2729
Kalanchoë blossfeldiana (Flaming Katy)	0.6224	12.3333	0.0283
Aglaonema commutatum (Chinese evergreen)	20.4121	6.8333	0.5810
Spathiphyllum wallisi (Peace Lily)	2.2211	4.6667	0.1309
Anthurium andraeanum (Flamingo Lily)	1.4039	6.6667	0.1696
Dracaena reflexa (Red-Edged Dracaena)	3.2260	6.5000	0.4976
Aloe aristata (Lace Aloe)	12.5528	6.7500	0.1417
Philodendron martianum	2.4044	6.3333	-0.0174
Plectranthus tomentosus (Vicks Plant)	15.5975	23.1667	0.4520
Chlorophytum comosum (Spider Plant)	0.7049	5.8333	0.0974
Nematanthus glabra (Goldfish plant)	3.9209	6.5833	0.1069
Begonia maculata (Polka dot begonia)	1.3158	5.8333	0.0747
Pteris argyaea (Silver Brake Fern)	2.3193	3.2500	0.0242

Tabulation of Score using Min-Max Normalization

Scores for each individual criteria were calculated using min-max normalization based on the following formula:

Min-Max Normalization:

Score = ((x_max - x) / (x_max - x_min)) × 10

Total Performance Score:

Total = Score_{Max VOC} + Score_Time + Score_{Steady
State}

Plant	Scoring (normalized from 0-10)			Score
Plant	Maximum Change in VOC Level (the lower, the higher rank)	Time Taken to Plateau (the faster, the higher rank)	Final Change in VOC Level (the lower, the higher rank)	Score
Epipremnum aureum (Money plant)	2.6732	7.2385	0	9.9117
Zamioculcas zamiifolia (ZZ plant)	6.8813	8.0753	8.5816	23.5383
Dracaena fragrans (Dragon plant)	8.9478	8.0335	9.1918	26.1731
Sansevieria trifasciata (Snake plant)	8.4172	8.5356	8.8208	25.7736
Fittonia albivenis (Nerve plant)	8.1211	7.9498	9.2817	25.3526
Hemigraphis alternata (Red Ivy)	8.3582	8.1172	8.8169	25.2923
Polyscias fruticosa (Ming aralia)	8.9915	8.4100	9.9398	27.3413
Philodendron oxycardium	8.5409	7.6151	7.9510	24.1070
Kalanchoë blossfeldiana (Flaming Katy)	10.0000	5.4393	9.6770	25.1163
Aglaonema commutatum (Chinese evergreen)	0	8.2008	5.7767	13.9776
Spathiphyllum wallisi (Peace Lily)	9.1921	9.2887	8.9532	27.4341
Anthurium andraeanum (Flamingo Lily)	9.6051	8.2845	8.6803	26.5699
Dracaena reflexa (Red-Edged Dracaena)	8.6844	8.3682	6.3649	23.4175
Aloe aristata (Lace Aloe)	3.9714	8.2427	8.8766	21.0907
Philodendron martianum	9.0995	8.4519	10.0000	27.5514
Plectranthus tomentosus (Vicks Plant)	2.4329	0	6.6869	9.1198
Chlorophytum comosum (Spider Plant)	9.9583	8.7029	9.1892	27.8505
Nematanthus glabra (Goldfish plant)	8.3332	8.3264	9.1223	25.7819
Begonia maculata (Polka dot begonia)	9.6496	8.7029	9.3495	27.7020
Pteris argyaea (Silver Brake Fern)	9.1426	10.0000	9.7064	28.8490

Appendix 8: Ingredients Composition in Microbial Products Used or Locally Sourced

Microbes (Commercial Product)	Active Ingredients	Biologically Inert Ingredients
Bacillus subtilis (Marknature Water Soluble Biological Fertilizer (Powder) (Agricultural Grade))	Bacillus subtilis	N.A.
Bacillus thuringiensis (Summit Mosquito Killer BTI Granules)	Bacillus thuringiensis subspecies Israelensis Serotype H-14, 2.86%w/w	Corn Cob, 97.14% w/w
Pseudomonas fluorescens (NBS PseudoTech)	Pseudomonas fluorescens TNAU Pf1 (ITCC No. BE 0005), 0.50%w/w	Talc Powder, 98.5% w/w	Carboxy Methyl Cellulose (CMC), 1.0% w/w
Trichoderma viride (NBS MicroShield)	Trichoderma viride (TNAU Strain Accession No. ITCC 6914), 1.0% w/w	Talc Powder, 98.5% w/w	Carboxy Methyl Cellulose (CMC), 0.5% w/w
Trichoderma harzanium (Trichoderma Fertilizer)	Trichoderma harzanium and the following: Bacillus spp. Lactobacillus acidophilus Saccharomyces cerevisiae Pseudomonas fluorescens 2% humic acid and special additives	3-5% Moisture

*Note: Trichoderma harzanium was not procured eventually.

Reason why Trichoderma harzanium was not chosen

Products with Trichoderma harzanium sold in Singapore are only available as Trichoderma-Bacillus where Bacillus species are included. For this specific product shown in the table above, trace amounts of other microbes and 2% Humic acid are present. Soil microbes and humic acid have a beneficial relationship (Saint Humic Acid, n.d.). Based on Canellas et al. (2022), adding humic substances enhances the ability of Plant Growth Promoting Bacteria. This means that when this product is used in our experiments, the different experimental results obtained when different microbes are used cannot be certainly attributed to the microbe of interest (i.e. Trichoderma harzanium) because confounders like humic acid are present, undermining results analysis. Therefore, Trichoderma harzanium was rejected.

Reasons why the inert ingredients are classified as inert

Corn cob is primarily composed of cellulose, hemicellulose, and lignin (Stoica et al., 2017) which are complex, cross-linked structural polymers highly insoluble in water. When Bacillus thuringiensis solution is prepared by mixing the granule into the water, it is unlikely the corn cob will be dissolved and change the chemical composition of the microbe solution. Therefore it is considered inert.

Talc powder is a common inert carrier used in microbial formulation (Novinscak & Filion, 2020) to contain the microbes spores separated and dry, preventing them from clumping together during storage.

Carboxy Methyl Cellulose is used as a binder in microbial formulation (Paau, 1998).

Appendix 9: Microbial products used in experiments

Microbe Product		Amount in experiments
Bacillus subtilis (MarkNature fertiliser, Agricultural Grade)		1g of powder mixed in 1L of water
Bacillus thuringiensis (Summit BTI Granules)		0.02g of granule mixed in 1L of water
Pseudomonas fluorescens (NBS PseudoTech)		3g of powder mixed in 1L of water
Trichoderma viride (NBS MicroShield)		3g of powder mixed in 1L of water

Table of Contents

CDE4301 Innovation & Design Capstone

AY2025/2026 Semester 1

The Living Filter: Plant-Microbe System for Indoor Air Purification

Final Report

Group IS-405

Acknowledgements

Abstract

1. Introduction (Sammi)

1.1. Importance of Indoor Air Quality

1.2. Volatile Organic Compounds (VOC) in Indoor Air and Their Impacts

1.3. Current Solutions to tackle VOC and Their Limitations

Level 01: Source Control

Level 02: Ventilation

Level 03: Air Cleaning

Level 04: Monitoring

Level 05: Behavioural Controls

1.4. Plant-Microbe System

1.5. Current Plant-Microbe Methods in Market

2. Project Overview (Sammi)

2.1. Research Question

2.2. Project Phases

3. Phase 1: Pre-Experiment - Pilot Study and Proof of Concept (Pearlyn)

3.1. Pilot Study (Pearlyn)

3.1.1. Pollutant Selection and Dosage

3.1.2. Selection of Plant and Microbe

3.1.3. Conduct Experiments

3.1.3.1. Experimental Setup

3.1.3.2. Experimental Procedure

3.1.4. Results and Discussions

3.1.5. Experimental Improvements and Methodology Refinement

3.1.6. Conclusion of Pilot Study

3.2. Proof of Concept (Pearlyn)

3.2.1. Selection of Plants and Microbe

3.2.2. Experimental Matrix and Set Up

3.2.3. Results and Discussion and Methodology Refinement

3.2.3.1. Plant Only Performance

3.2.3.2. Microbe Only Performance

3.2.3.3. Plant-Microbe Performance

3.2.4. Conclusion of Proof of Concept

4. Phase 2: Plant Selection (Pearlyn)

4.1. Plant Research

4.2. Experiments for Top Performing Plants

4.3. Results and Discussion

4.4. Plant Selection Conclusion

5. Phase 3: Microbe Selection (Tse Hui)

5.1 Microbe Research

5.2 Experiments for Microbes

6. Phase 4: Plant-Microbe Pairing Matrix (Tse Hui)

6.1. Objectives

6.2. Methodology

6.3. Results and Discussion

6.3.1 Results

6.3.2 Silver Brake Fern + Bacillus Species Analysis

6.3.3 Spider Plant + Trichoderma viride Analysis

6.3.4 Worse-Performing Pairings Analysis

6.4 Conclusion

7. Limitations of Project (Sammi)

8. Conclusion (Sammi)

References

Appendix

Appendix 1: Timeline of Project

Appendix 2: Natede Report

Appendix 3: Experimental Considerations and Modifications

Appendix 4: Calculation of Amount of Pollutant

Appendix 5: Summary of Research Studies of VOC Removal Efficiency of Shortlisted Plants and Microbe

Appendix 6: Plant Selection Research

Appendix 7: Plant Selection Scoring

Appendix 8: Ingredients Composition in Microbial Products Used or Locally Sourced

Appendix 9: Microbial products used in experiments

Appendix 10: Plant-Microbe Pairing Performance (∆VOC-Time Graphs)