Graphene Biotransistor Interfaced with a Nitrifying Biofilm

Using a graphene field-effect transistor biosensor, we monitored the pH inside a living biofilm with fast temporal resolution (∼1 s) over multihour time periods. The atomically thin sensor is positioned between the biofilm and a supporting silicon oxide surface, providing noninvasive access to conditions at the base of the biofilm. We determine the transient changes in pH when the biofilm metabolizes substrate molecules and when it is exposed to biocide. The pH resolution is approximately 0.01 pH units when using 1 s time averaging. The sensor drift is approximately 0.01 pH units per hour. Our results demonstrate the potential of this technology to study biofilm metabolism and monitor biofilm health. ■ INTRODUCTION Quantitative tools to monitor living biofilms are important in fields ranging from medicine to environmental monitoring. In applications, such as wastewater treatment, biofilm reactor systems are typically monitored using bulk fluid measurements. This technique is powerful, because a range of traditional chemical assays are available and measurements are directly linked to process performance. However, it is well known that understanding the kinetics of growth and substrate utilization within a biofilm is essential for optimizing fixed-film processes. Microsensors have been used to measure chemical gradients in biofilms including specific chemicals, such as dissolved oxygen, sulfide, nitrate, nitrite, and pH. pH microelectrodes commonly used in biofilm studies have an H selective membrane made of special glass. Although they have a long lifetime, their spatial resolution is limited to about 20 μm and they are expensive. Liquid ion-exchange (LIX) pH microsensors have a higher spatial resolution of ∼5 μm, but have a short lifetime of only a few days. Thus, there is interest to develop inexpensive long-lived sensors that have high spatial resolution. Recent advances in bioelectronic sensors made from nanoscale materials offer exciting new ways to monitor biofilm activity. The goal of this study is to explore the suitability of field-effect transistor (FET) biosensors for monitoring the metabolic activity of a biofilm. We use graphene, an atomically thin sheet of sp-bonded carbon atoms, as the active material for our FET biosensor. The electrical resistance of graphene is sensitive to charged species adsorbed on the graphene surface. Graphene is remarkably biocompatible, as shown by recent tests with Escherichia coli and neural cells. While graphene FET (GFET) biosensors are being pursued by a number of research groups (reviewed in refs 5 and 8), efforts to interface GFETs with bacteria are just beginning. Previous authors have used chemically functionalized graphene to capture bacteria on GFET sensors, but there are no previous reports of naturally formed biofilms interfaced with GFET sensors. The biofilm used for this work is formed by ammonia oxidizing bacteria (AOB). AOB play a critical role in the global nitrogen cycle and in the removal of nitrogen during wastewater treatment. Because AOB are slow growing, biofilm-based processes are particularly important. In the biofilm form, AOB exhibit higher nutrient removal rates and higher resistance to washout when compared to planktonic bacteria. Techniques to monitor the health of AOB biofilms are desired, because AOB are considered to be some of the most sensitive microorganisms found in wastewater treatment plants. ■ MATERIALS AND METHODS The GFET was fabricated using graphene grown via chemical vapor deposition. Details of the fabrication process are described in the Supporting Information. To limit the contact between liquid and electrical connections, the electrode traces were covered by 70 nm of SiO2 and wires were sealed with silicone. Parasitic currents were 3 orders of magnitude smaller than the source drain current (Isd). For all sensing experiments, the GFET was biased with a source-drain voltage Vsd = 25 mV. The electric potential of the liquid was controlled by an Ag/ AgCl reference electrode attached to a voltage source, Vliq (see Figure 1c). Prior to biofilm growth, we characterized the sensitivity of the bare GFET to pH (Figure 2a). The device was operated with Vliq in the range 100−250 mV where the slope dIsd/dVliq = 39 μA/V. Changing the pH by one unit was equivalent to a Received: January 29, 2015 Revised: February 25, 2015 Accepted: February 26, 2015 Letter pubs.acs.org/journal/estlcu © XXXX American Chemical Society A DOI: 10.1021/acs.estlett.5b00025 Environ. Sci. Technol. Lett. XXXX, XXX, XXX−XXX changing Vliq by ΔV = 17 mV. Similar pH sensitivities have been observed by other authors and attributed to specific adsorption of hydroxyl and hydronium ions on the graphene surface. Mailly-Giacchetti et al. verified that the linear relationship between ΔV and pH extends from pH 4 to above pH 8. Therefore, for the pH conditions inside our AOB biofilms, the GFET can be operated at fixed Vliq to generate an Isd signal that is linearly proportional to pH. The AOB biofilm (Nitrosomonas europaea) was grown directly on the surface of the GFET device. The sensor was submersed in a bath of HEPES buffer (30 mM) with trace nutrients for growth. The volume of liquid was maintained at approximately 60 mL in a standard Petri dish. The GFET remained in this bath for the duration of the experiments. The N. europaea cells (ATCC strain 19718) used to inoculate the GFET surface were grown in batch, concentrated, and the cells and growth media, including biomolecules, were added to the Petri dish. The initial bacterial attachment phase lasted ∼8 h, during which bacteria, protein, and polysaccharides floating in solution gradually attached to the surface of the GFET. During this 8 h period we observed changes in the Isd−Vliq curve equivalent to ΔV = 60 mV (Figure S5, Supporting Information). Such changes in ΔV are consistent with charged molecules adsorbing on the graphene surface. After the initial bacterial attachment, the position of the Isd(Vliq) curve became stable (dΔV/dt ∼ ±0.2 mV/hour, corresponding to a baseline drift of 0.01 pH units per hour) and the device exhibited greater pH sensitivity (1.4 μA/pH) (Figure 2b). The augmented pH sensitivity of the GFET sensor is likely due to new ionizable groups on the graphene surface, as illustrated in Figure 2b inset. The pH-dependent charge state of moieties such as COOH and NH2 modifies the number of charge carriers in the graphene, causing changes in Isd. ■ RESULTS AND DISCUSSION Optical micrographs were taken to determine biofilm development on the GFET. These images show pillar formation, followed by the development of mature biofilms in a timeframe of approximately 2.5 weeks (see Figure 1a,b). Similar timescales for N. europaea biofilm development were observed by Lauchnor et al. To confirm that the biofilm was indeed an AOB, rather than an unwanted bacterium, we used traditional methods to monitor the nitrite and pH levels of the bulk fluid over 4 days (Figure S6, Supporting Information). We observed a continuous increase in nitrite levels and decrease in pH, consistent with the activity of an AOB and consistent with N. europaea batch tests at a similar buffer capacity. While establishing the biofilm, the medium was exchanged regularly to replace nutrients and buffer and to remove bacteria suspended in solution. After a mature biofilm was established, roughly 3 weeks, the GFET was used to monitor pH at the base of the biofilm with high temporal resolution. We first studied the response of the system to ammonia (NH3). N. europaea is known to convert NH3 to NH2OH and then NO2 − in a twostep process. The enzyme ammonia mono-oxygenase (AMO) catalyzes the first step, and hydroxylamine oxidoreductase (HAO) catalyzes the second step. + ⎯ → ⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯⎯ + + + − + NH 3/2O NO H H O 3 2 (AMO HAO)