Water treatment membranes embedded with a stable and bactericidal nanodiamond material

With the rapidly expanding world population, exploitation of natural resources and extensive pollution, the quality and availability of potable water represent a growing threat to human health (Jackson et al. 2001; Ogunlela 2010; Xie et al. 2015); 2.2 billion people lack access to safely managed drinking water services and 297,000 children under the age of five die every year from diarrheal diseases due to poor sanitation, poor hygiene or unsafe drinking water (World Health Organization 2019). The constant expansion of impervious surfaces in urban environments is adding nonpoint source (NPS) pollutants to the watershed and groundwater catchment areas thereby changing the quality and quantity of available potable water (Sun et al. 2016). The recorded deterioration of public health due to waterborne outbreaks caused by this increment of impervious surfaces and improper watershed management in point catchment areas is an important issue that scientists, public health officials, politicians and community leaders need to address (Dubinsky et al. 2016; Kirschner et al. 2017).

Biocompatible applications of nano materials have been an active research area in recent years because of their unique structure and physicochemical properties. According to previous reports, nanodiamonds (NDs) in the size range 2–20 nm are biocompatible with in vitro human cells (Helland et al. 2007; Simate et al. 2012; Villalba et al. 2012) and have a large surface area leading to a high affinity for biomolecules (Solarska et al. 2012). Recently, ultradispersed diamonds (UDDs) have gained worldwide attention due to their inexpensive large-scale synthesis based on the detonation of carbon-containing explosives (Future Markets, Inc. 2019). These semi-crystalline nanoparticles consist of diamond nanocrystals embedded within a graphite-like carbon matrix forming large aggregates of particulates with some graphitic carbon content (Michel & Lukehart 2015; Ashek-I-Ahmed et al. 2019). The detonation produces UDD with a small primary particle size (ca. 4–5 nm), high biocompatibility (Schrand et al. 2009) and the capacity for facile functionalization. Medina et al. (2012) investigated the ND's surface interaction with P. aeruginosa Gram-negative bacteria and concluded that its bactericidal and anti-adhesive properties are due to its semiconducting properties. The electrically active surface causes membrane damage (Etemadi et al. 2016) and oxidative stress to the bacteria thereby inducing its death (Medina et al. 2012). The bactericidal properties and stability of the UDD upon usage and cleaning sparked interest in the nanoparticles for use in water treatment (Upadhyayula et al. 2009; Viet Quang et al. 2013; Yin & Deng 2015).

Membrane separation processes are increasingly utilized methods for the treatment of water and wastewater. Pressure-driven membrane technology is a common process for water purification in manufacturing and pharmaceutical industries, which need to meet water quality standards (Kumar et al. 2014). Currently, this methodology confronts key challenges, e.g. membrane selectivity and permeability, and fouling and membrane lifetime, all of which need to be overcome for this technology to become a leading water treatment option. Fouling resistance is one of the biggest challenges for microfiltration (MF) and nanofiltration (NF) membranes since most of them are hydrophobic. Functional nanomaterials, incorporated into the membranes, may be the solution to these challenges by changing permeability, fouling resistance, as well as their mechanical and thermal stability (Kunduru et al. 2017).

Lower membrane fouling allows higher potable water productivity, less cleaning and longer membrane life, leading to reduced capital and operational costs. There are different types of membrane fouling, such as inorganic, organic and biofouling. To reduce fouling, classical solutions are available such as membrane pre-treatment, operation optimization and chemical cleaning (Beyer et al. 2017). Poor and ineffective pre-treatment can lead to higher rates of fouling and all of these treatments have the potential to damage the structural composition of the membrane (Sun et al. 2013). The chemical structure and morphology of the membrane, i.e. functional groups, charge and hydrophobicity, pore size, surface roughness and/or surface pattern are required knowledge to be able to reduce or increase fouling. The need for new cost-effective membrane materials, capable of overcoming the trade-off between anti-fouling capacity and permeability, as well as simple advanced methods of membrane modification are areas of research that need to be further understood (Qin et al. 2020; Zhao et al. 2021). The research presented here demonstrates how current water purification membranes can be enhanced by incorporating ND particles to reduce biofouling resistance problems, strengthening their mechanical stability and increasing their useful lifetime for water purification.

The polluted water source used in this research was collected from the Rio Piedras River in San Juan, Puerto Rico (18°24′8.91″N, 66° 3′54.44″W), which contains high fecal bacteria concentrations (Lugo et al. 2011) mainly caused by anthropogenic outputs, such as household septic tanks and agricultural practices upstream of the river (Garcia-Montiel et al. 2014; Laureano-Rosario et al. 2017). A water sample of 750 cm³ was collected and stored in a 1,000 cm³ bottle covered in aluminum foil for low-light interaction. UDD interaction with the polluted water was obtained by mixing the UDD with 100 mL of the polluted water in a 250 cm³ Erlenmeyer flask. Two 250 cm³ Erlenmeyer flasks, one containing 100 cm³ of polluted water from the river and the other with 100 cm³ of polluted water with UDD, were placed in the incubator at 37 °C with a shaker operating at 110 RPM for UDD interaction promotion. Fecal and total coliforms bacteria analyses of the microbial polluted water were done using Coliscan Easygel petri dishes at 10-min intervals from 0 to 40 min. Bacteria colony forming units (CFU) characterization of the petri dishes was done after 24 h by using arithmetic sample mean and standard deviation analysis. The calculation of the coefficient of variation (CV) was done using sample mean and standard deviation of each time interval. Each microbial characterization was done in triplicate to reduce variability and increase precision.

UDD usage for biological applications has been limited due to its dispersibility properties caused by particle aggregation in aqueous solutions ranging from 0.1 to 1 μm in size (Stehlik et al. 2016; Whitlow et al. 2017). Recent research has demonstrated that a decrease in the degree of agglomeration is observed when pristine UDD is treated with strong acids (Pedroso-Santana et al. 2017).

UDD powder was obtained from Adamas Nano Inc. (Rayleigh, NC, USA) The NDs have an average aggregate size of 200 nm, a Z potential of +20 mV in deionized water (DI H₂O) and 1.7 wt.% ash content. The NDs were first sonicated with hexane to remove nonpolar impurities, followed by acetone, isopropanol and DI water to remove polar impurities. Each UDD solution was centrifuged after sonication, and the supernatant was removed. Once the UDD was cleaned, the precipitate obtained from the centrifuge was dried using Labconco FreeZone 2.5 for the lyophilization process.

The lyophilized UDD was treated with a 3:1 sulfuric acid (H₂SO₄) (CAS number 7664-93-9)/hydrogen peroxide solution 30% (H₂O₂) CAS number 7722-84-1) solution for surface functionalization and reduce particle's aggregation. The solution was heated to 120 °C for 30–40 min and then cooled to room temperature. The solution was centrifuged to separate the functionalized UDD precipitate. This process was repeated three times using nano pure water and the precipitate was dried using the Labconco FreeZone 2.5 for the lyophilization process.

Characterization of the clean UDD and functionalized UDD FTIR particles was done using a Nicolet IS 50 FT-IR Continuum IR Microscope to confirm UDD surface functional groups. The FTIR parameters used were: 32 scans in % Transmittance format with automatic atmospheric suppression. Both powders were placed in a KBr IR card with a15 mm aperture, which was inserted into the Continuum IR Microscope.

To investigate the interaction between UDD nanoparticles and commercial membranes, a combination of electron microscopy (JSM-7500F (JEOL, Tokyo, Japan) field-emission scanning electron microscope (FE-SEM) operated at 200 kV) and tensile strength measurements (Brookfield CT3 texture analyzer) were performed to characterize the mechanical properties, i.e. the membrane's Young Modulus (the stress the membrane can withstand without deforming), yield point (the limit of the material's elastic region before changing into its plastic region) and how the elastic and plastic regions of the membrane change upon UDD incorporation. The Young modulus value was obtained by calculating the slope of the elastic region having an R² value of 0.9 and the yield point resulted from the elastic region stress increase before moving towards the plastic region. Commercially available membranes, polyether sulfone (PES) and polyvinylidene (PVFD), each one with different surface morphologies and composition, were used. The PES membrane has a symmetric pore size of 0.5 μm with dimensions of 150 μm thickness and a 47 mm diameter. The PVDF membrane has an asymmetric pore size of 0.45 μm, with dimensions of 125 μm of thickness and a diameter of 47 mm. UDD-embedded membranes were fabricated by doing a dead-end filtration of 0.10 wt./vol% UDD solution in 125 mL DI water. The formed UDD paste was removed by sonicating the fabricated membrane, and then leaving the material with embedded UDDs at the membrane's surface and in its porous matrix. Three membrane samples were obtained, one with the UDD paste, another with 1 min of sonication for UDD paste removal and the third one with 2 min of sonication for UDD paste removal. SEM characterization for all samples was performed to verify the UDD presence at the surface of the membrane.

Sample preparation for the FE-SEM was performed by depositing 10 nm of gold, from a 99.999% Au target, onto the sample using the PELCO SC-7 Auto Sputter Coater. For membrane comparison, we sonicated the UDD-embedded membranes for either 1 or 2 min to remove the UDD excess formed above the membrane. Tensile strength stress vs. strain characterization was performed by cutting the studied membrane into 3 × 1 mm rectangles (0.125 mm thick) and placing them in Brookfield cT3 texture analyzer.

Bacteriological studies between the commercially available membranes and the UDD-embedded ones were done using the Micrology Laboratories Coliscan membrane filtration method, a U.S. Environmental Protection Agency approved method for bacteria CFU enumeration (Taylor et al. 2015). The polluted water source used in this research was collected from the Rio Piedras river in San Juan, Puerto Rico watershed (18°24′8.91″N, 66° 3′54.44″W), which contains high fecal bacteria concentrations (Lugo et al. 2011) mainly caused by anthropogenic outputs, such as household septic tanks and agricultural practices upstream of the river (Garcia-Montiel et al. 2014; Laureano-Rosario et al. 2017). The water samples were collected in sterile bottles, stored in an ice cooler and tested within 4 h of collection. A 1:10 dilution was performed to the collected water using DI water to maintain the CFU enumeration within the limits specified by the Coliscan method. Plate counting was performed and T-test analysis of triplicate samples was done to identify fecal E. coli CFU changes between the membrane-filtered samples. The fecal E. coli death rate % comparison between the control and studied membranes was analyzed to see the reduction in CFU counting when the studied water was filtered with UDD-embedded membranes.

A study to verify UDD antimicrobial properties was achieved by the treatment of polluted water with different UDD concentrations. Table 1 shows UDD-treated water fecal E. coli CFU and control polluted comparison at different time intervals. Results indicated a significant reduction of CFU when polluted water was mixed with UDD. At 0 min, the control water and UDD-treated water showed fecal E. coli bacteria concentrations of 950 ± 100 and 875 ± 98 Col/cm³, respectively. After 40 min, control plates had a sample mean of 2,425 ± 96 Col/cm³ and UDD-treated water plates had a sample mean of 150 ± 57 Col/cm³. Fecal coliforms death rate results for each time interval showed a constant reduction of CFU as time passed.

In order to increase the fecal E. coli death rate, we increased the UDD concentration to 0.02 g/cm³ and the results were similar. The CV % shown in Table 1 for 0.02 g/cm³ UDD suggests a higher variability in UDD treated plates in comparison to untreated water as time passes. This high variability is likely due to a decrease of UDD-treated water (Col/cm³). When the mean value approaches 0, the CV will approach infinity and is therefore sensitive to small changes in the mean. A 97% death rate was achieved compared to the 94% at 0.01 g/cm³ UDD concentration. These results demonstrate UDD's bactericidal properties suggesting the use of these nanoparticles as a disinfectant agent for water microbial treatment.

The PVDF membrane's structure also plays a role when UDDs are embedded into its porous matrix (Table 3). UDD Wt.% on UDD Funct. PVDF membranes are 27% less compared to UDD PVDF. UDD deagglomeration of clusters and the asymmetric structure of PVDF compared to the symmetric PES can be a possible reason for the UDDs to pass through the porous matrix and not be embedded above the surface.

From an operational point of view, the leaching of biocompatible nanoparticles, i.e. UDD, may be part of the sieving effect phenomena during the membrane filtration process (Drioli et al. 2017). The particles larger than the pore size are retained, and the smaller size nanoparticles are leached into the effluent water.

The PES membrane's ability to withstand stress without deforming between commercial PES (control) and UDD-embedded PES samples is shown in the Young Modulus (Table 4) values. The results show that PES’ ability to withstand stress does not significantly change when UDD nanoparticles are embedded into the membrane's porous matrix. These findings suggest that the incorporation of UDD into the porous matrix does not change the membrane's ability to withstand stress but does increase its yield point in the elastic region, enhancing its ability to support higher stress before deformation.

Table 4

Young Modulus value comparison between the pristine UDD and functionalized UDD-embedded PES membranes

Young Modulus (Pa) .	PES .	PES UDD .	PES UDD + 1 min sonication .	PES UDD + 2 min sonication .
	20,878	21,271	20,120	20,505
Young Modulus (Pa) .	PES .	PES UDD Funct. .	PES UDD Funct. + 1 min sonication .	PES UDD Funct. + 2 min sonication .
	20,878	21,421	20,795	19,192

Young Modulus (Pa) .	PES .	PES UDD .	PES UDD + 1 min sonication .	PES UDD + 2 min sonication .
	20,878	21,271	20,120	20,505
Young Modulus (Pa) .	PES .	PES UDD Funct. .	PES UDD Funct. + 1 min sonication .	PES UDD Funct. + 2 min sonication .
	20,878	21,421	20,795	19,192

View Large

Bactericidal properties of studied membranes are illustrated in Tables 6 and 8 showing the CFU death rate % comparison between the commercial and UDD-embedded membranes. When microbially polluted water was filtered by the UDD-embedded PES membrane, the filtered water showed a death rate of 89% compared to the 88% shown by the commercially available PES membrane. At a 5% confidence interval, T-test analysis between the UDD-embedded PES and the commercial PES indicates no significant reduction on fecal E. coli CFU after the filtration was performed. Different from this, a significant difference in bacteria removal was seen, with a value of 0.02, between the functionalized UDD-embedded PES membrane and commercial PES indicating that the excess particle clusters above the membrane do not improve bactericidal properties and can lead to poor bacteria removal.

Like the previous analysis, UDD-embedded PES membranes that were sonicated to remove the excess of particle clusters above the membrane also showed that there is no significant difference when the water was filtered. Table 6 also shows the death rate % values when the water was filtered with sonicated funct. UDD-embedded PES membranes. Contrary to previous results, sonicating the UDD functionalized membranes enhances its bactericidal properties by significantly reducing fecal E. coli CFU as compared to commercial PES membranes. The incorporation of these nanoparticles into the membrane's porous matrix increases the bacterial death rate percentage by 5–8% depending on the sonication time for UDD cluster removal. T-test analysis of CFU values dependent on sonication time showed a p-value of 0.20 (Table 7) between PES UDD Funct. 1 min and PES UDD Funct. 2 min meaning that there is no significant difference of bacterial removal if the membrane is sonicated for 1 or 2 min.

For PVDF membranes, Table 8 shows the CFU values of UDD-embedded PVDF membranes in filtered water. Compared to the control sample, commercial and studied UDD PVDF membranes significantly reduce bacteria concentration with a death rate of 80 and 83%, respectively. T-test analysis from these two filtrations gave the value of 0.26 suggesting no significant difference. For UDD functionalized membrane comparison, a result of 0.0003 demonstrated that the incorporation of functionalized membranes into the PVDF porous matrix significantly reduces bacteria concentration compared to commercial ones. The insertion of a more dispersed UDD significantly improves PVDF bactericidal properties by reducing CFU by 17% more producing water with minimum bacteria CFU.

T-test analysis of CFU values dependent of sonication time showed a p-value of 0.14 (Table 9) between PVDF UDD Funct. 1 min and PVDF UDD Funct. 2 min meaning that there is no significant difference of bacterial removal if the membrane is sonicated for 1 or 2 min. These improvements in both membranes’ bactericidal properties are consistent with the ones reported in the literature when this incorporation of carbon nanoparticles into organic membranes is done (Etemadi et al. 2016).

This research has demonstrated the potential use of UDD as a new disinfection material for water treatment. The results show that UDD effectively reduces fecal E. coli bacteria CFU concentration in polluted surface water. A death rate between 94 and 97% was observed in UDD-treated water plates depending on UDD concentration.

UDD FTIR spectra from functionalized UDD showed prominent peaks at the UDDs characteristic region (from 1,000 to 1,500 cm⁻¹) compared to commercial UDD, indicating the addition of oxygen-containing functional groups into the UDDs surface. SEM images of PES and PVDF membranes showed UDD particles at the membranes’ surface and symmetric (PES) and asymmetric (PVDF) porous matrix. Compared to commercial UDD, functionalized UDD looked more spread throughout the membrane surface and UDD cluster sizes were smaller. The increase in oxygen-containing functional groups shown by the FTIR spectra promotes UDDs deagglomeration and enhances UDDs hydrophilic properties causing these nanoparticles to be more dispersed throughout the membrane surface and inside the porous matrix.

Tensile strength characterization for PES membranes demonstrated no change in its Young Modulus values and an increase in the membranes’ yield point when UDDs were embedded. A higher yield point was observed when UDD nanoparticles are embedded into the porous matrix and the membrane is sonicated to remove the excess nanoparticles at the top of the surface. A decrease in tensile strength yield point was detected in UDD-embedded PES membranes that contained a high concentration of UDDs at the porous matrix. For PVDF membranes, a different result was seen compared to PES. UDDs dispersion throughout its asymmetrical porous matrix makes the membranes less elastic, reducing its yield point and Young Modulus values. The changes in the membranes’ plastic and elastic regions tailor towards the findings reported in the literature on how the incorporations of NDs change the membranes mechanical.

Coliscan membrane filtration characterization was performed for commercial and UDD-embedded membranes. The insertion of functionalized UDDs into the PES and PVDF commercial membranes significantly enhances bacteria removal by providing a better water quality, enhancing its current bactericidal properties and demonstrating the potential these membranes have to improve current water filtration technology for bacteria removal properties.

The development of these organic/carbon nanoparticle membranes has the potential to enhance current membrane lifetime usage by changing the membrane's plastic and elastic properties, enhancing microbial removal properties thereby having the potential to produce better drinking water quality and quantity for membrane filtration treatment systems.

This research project was supported by NASA Training Grant NNX15AI11H (PR Space Grant). The authors gratefully acknowledge the Molecular Science Research Center of the University of Puerto Rico, and the technical support from the Materials Characterization Center. The authors also acknowledge the undergraduate students Ms. Swizel Teneval Ms. Adneliz Cátala and Mr. Saúl Sandoval and the laboratory administrative staff Mr Jose Guzmán.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.