We documented the details of fabricating, assembling and embedding DF plugs in our previous work [21]. In brief, the disc consists of five layers: three layers of 1.5-mm thick poly- (methylmethacrylate) (PMMA) sheets (Radionics, Dublin Ireland) and two layers of ˜ 90 μm thick pressure sensitive adhesives (PSA) from Adhesives Research, Limerick, Ireland. A series of design iterations was assisted by CAD software. The optimised disc design with chambers and access holes was machined using a Zing CO₂ laser cutter (Epilog Laser, Golden, CO, USA).

The chambers were tailored to the desired reagent volume. The pneumatic chambers are 4 mm in diameter. The exit channel from the blood separation (BS) chamber (Figure 1B) has been designed so as to fundamentally prevent leakage of whole blood into the pneumatic chamber. The microchannels were defined in the PSA using a Craft ROBO knife cutter (Graphtec Corp, Irvine, CA, USA). The microchannels are 0.6 mm wide. The hybrid DF valve tabs were fabricated by firstly cutting 1-mm diameter through holes in the PSA, and then rolling the ˜ 20 μm DFs (also supplied by Adhesives Research) on the PSA, in order to cut out a ∅ 3.5-mm outline. After placing the DF tabs at the designated valve locations (DFB (DF-blood;, DF1-DF4, Figure 1B) the 5-layer discs (Figure 1A) are assembled using a standard hydraulic laminator. The DF is a proprietary product of Adhesives Research.

In this study we employed fluorometric nitrate and nitrite assay kits purchased from Cambridge Biosciences, Cambridge, UK. Nitric oxide (NO) undergoes a series of reactions with several biological fluids. The final products of NO in vivo are nitrate (NO³⁻) and nitrite (NO²⁻). Yet, since their proportion in whole blood is small, it is quite common in clinical chemistry to independently determine the total concentration of (NO³⁻ + NO²⁻), and subsequently NO²⁻, and then calculate the NO³⁻ concentration from their difference. Using this assay kit the respective concentrations of total (NO³⁻ + NO²⁻) and NO²⁻ were determined on two separate discs. The step-by-step sequence of the on-disc assay protocol is illustrated in Figure 2. In order to determine the total (NO³⁻ + NO²⁻) concentration; nitrate reductase was added to the extracted blood plasma, followed by the addition of an enzyme cofactor. Subsequently, an acidic solution of diaminonapthalene (DAN) was introduced to the reaction chamber, eventually producing naphthotriazole (Equation 1).

(1)

The nitrate reductase reagent converts NO³⁻ in blood plasma to NO²⁻. After this reduction reaction, the enzyme cofactor crosslinks with the formed NO²⁻ and makes it an active enzyme which is fluorescently tagged by the DAN reagent.

Following addition of NaOH solution, the enzymatic reaction stops and the fluorescence of the product (total NO³⁻ + NO²⁻) is enhanced (Figure 2A). Addition of the NaOH solution enhances the fluorescence and the analyte is ready for detection. Read-out is then performed off-chip on a fluorescence analyser at an excitation wavelength λ = 360–365 nm and emission at λ = 430 nm.

In order to determine the NO²⁻ concentration, a second disc was used. The assay procedure is similar to that of total NO³⁻ + NO²⁻ described above; however, the enzyme and cofactor were not added (Figure 2B). Fluorescence measurement of the formed product yields an accurate blood plasma NO²⁻ concentration. The NO³⁻ level was then obtained by subtracting the NO²⁻ from the total (NO²⁻ + NO³⁻) content. Calibration curves were generated according to the user manual of the commercial assay protocol.

The microfluidic discs are designed to concurrently run eight simultaneous tests (see Figure 1A), e.g., seven calibration standards, and one biological sample. All samples and reagents are pre-loaded on the disc and then taken to the centrifugal test stand. For disc 1, 70 μL of whole blood and assay buffer were loaded in BS and RC respectively (Figure 1B). Ten μL of nitrate reductase, enzyme cofactor and DAN reagents are loaded into RS1-RS3 and finally 20 μL of NaOH solution is loaded in RS4 (Figure 1B). The sequential release of plasma from BS and reagents from RS1-RS4 into RC is controlled by the DF valves DFB and DF1-DF4, respectively. For disc 2, 70 μL of whole blood and assay buffer were loaded in BS and RC, respectively (Figure 1B), while 10 μL of DAN reagent and NaOH were loaded in RS1 and RS2.

The centrifugal test stand, which has been fully described in previous work [21,26] comprises of a computer-controlled motor and a stroboscopic light for imaging a chosen area on the disc during rotation. The steps involved in this assay are sample metering, blood separation, plasma metering and extraction, sequential reagent delivery, mixing and, again, reagent delivery. The centrifugal force provides the pumping pressure for the liquid reagents.

The normally-closed DF valve follows a centrifugo-pneumatic principle [21]. While the disc is at rest, liquid reagent in the reservoir is retained by a combination of the capillary and surface tension forces at the entrance of the microchannel leading to the dynamic formation of the pneumatic chamber. In other words, the entrance to the microchannel that connects the reservoir and the pneumatic chamber acts as a capillary stop. Rotation at the angular velocity ω induces a pressure head: (2) Δ p ω = ρ r ¯ Δ r ω 2(Figure 1C) where ρ is the liquid density, ω is the angular velocity of the rotating disc; r̄ =(R_max + R_min)/2 is the characteristic distance of the liquid element from the centre of rotation, while Δr= R_max − R_min, is the radial height of the liquid column. The advancing meniscus stops as Δp_ω is balanced by pressure in the valve chamber: (3) p = p 0 1 1 − Δ V / V 0as obtained by Boyle's law. In Equation (3), p₀ is the ambient pressure, V₀ is the (full) chamber volume and ΔV is the gas volume reduction due to the protruding liquid meniscus [27].

The meniscus in this metastable, inverted-layer configuration is stabilized by the surface tension. As the centrifugal pressure head Δp_ω Equation (2) increases, the liquid plug can proceed farther into the compression chamber, thus making it more difficult for the surface tension to sustain the “hanging” liquid volume ΔV. At the so-called burst frequency, the liquid plug disrupts to invert the metastable liquid-gas configuration. Liquid then penetrates unimpeded into the pneumatic chamber to dissolve the DF and hence fully open the valve.

For a given, rotationally induced pressure head Δp_ω Equation (2), the distance the plug can protrude into the compression chamber (and the displaced air volume ΔV) until Δp_ω Equation (2) balances the counter pressure p Equation (3), which increases with the dead volume of the chamber V₀. As an alternative explanation, the fluidic capacitance defined by volume change ΔV induced by a given pressure change p Equation (3) increases with V₀. As the destabilisation is induced towards increasing ΔV, the burst frequency tends to shrink with increasing chamber volume V₀.

All valves are initially in a closed state until their individual critical burst pressure is reached, at which the valve yields and liquid is pumped onward for further processing. The measurements in Figure 3 verify the three main experimental impact parameters on the burst frequency: (a)

Mean radial position r̄ of the liquid plug (and thus also the valve) on the disc. According to Equation (2), the centrifugal force increases with the radial distance from the centre of rotation, thus the valves which are located more radially outward are scheduled to be actuated first (Figure 3A). At a constant pneumatic chamber volume and exit channel length (see Figure 1C), when the radial position of the DF valves increased from 15 mm to 50 mm, the average centrifugal burst frequency decreased from 4,350 to 1,800 RPM. The R² of the linear fit which is 0.969 demonstrate the predictability of this behaviour.

Consequently, the rotational burst frequency as the central control parameter for all (liquid bearing) valves on the centrifugal platform can be specifically tailored over a wide range. Figure 4 for the first time demonstrates the spin protocol of a fully integrated and rotationally actuated, 7-step bio-assay protocol. Each valve is colour-coded in order to highlight the time course of the rotational burst frequency and the pressure head as deduced from Equation (2). The plot in Figure 4A references the pressure heads of each valve Equation (2) to their respective burst pressures. The sequential crossing of the horizontal (y = 1) -line corresponds to a valve opening, i.e., the point where p Equation (2) exceeds the critical pressures p_i,crit of each valve i.

Figure 4A demonstrates the sequential actuation of the DF valves DFB and DF1-4 to fully integrate and rotationally automate the homogeneous nitrate and nitrite bioassays including blood separation, plasma extraction and sequential release of four different reagents. Figure 4B shows the rotationally induced (absolute) pressure heads of the liquid column as a function of the spin protocol expressed in terms of rotations per minute (RPM) displayed in Figure 4C. The critical burst pressures for the DF valves for each unit operation are indicated in the graph: plasma extraction (P_Bcrit), DF valves 1-4 (P_1crit – P_4crit). The lighter shades of this colour-coded graph indicate the points at which the critical pressures are exceeded and the DF valves are actuated. The absolute pressure head is proportional to the pressure exerted by the liquid reagent in the reservoir at its specific radial position on the disc. The first step is to meter the whole blood sample, by spinning the disc at 600 RPM for 1 min, leaving 65 μL of blood in BS chamber (Figure 1B). The disc is then rotated at 1,200 RPM for 3 min to sediment the red blood cells and separate plasma. After this plasma separation step, the rotational frequency is increased to 2,100 RPM in order to open valve #1 (DFB, Figure 1B), thus releasing 10 μL of plasma into the reaction chamber (RC), which has already been preloaded with 70 μL of assay buffer. The enzyme is released into the RC at 2,700 RPM, while the cofactor is added to the same chamber at 2,850 RPM. At this stage the assay requires an incubation step to sufficiently mix plasma + assay buffer + enzyme cofactor + nitrate reductase. To this end the disc was alternatingly spun in a clockwise and anticlockwise sense of rotation at an amplitude of 600 RPM for 3 min [12].

Upon increasing the rotational frequency to 3,300 RPM, the DAN reagent was released from reagent storage (RS3) chamber into the RC. Subsequently, NaOH was added to the same RC at a frequency of 3,900 RPM from RS4 chamber. The NaOH solution terminates the enzymatic reaction and also amplifies the fluorescence signal, as schematically illustrated in Figure 2 and Equation (1).

The procedure for the nitrite assay is very similar. In this case, only 10 μL of the DAN reagent and 20 μL of NaOH are used. The DAN reagent and NaOH solution are sequentially released into the RC while the disc rotated at 2,700 RPM and 2,900 RPM, respectively, after which 3 min of on-disc incubation is carried out by using the vigorous, “shake-mode” mixing at ±600 RPM.

A schematic representation of the liquid handling procedure and the image frame time sequence of this process are detailed in Figure 5. In order to enhance the contrast for this demonstration, coloured food dye was used. The full video of this centrifugally automated bioassay with all assay reagents including blood separation and plasma extraction processes can be found in ESI.

As this paper focuses on the fluidic automation aspect of a multi-step assay protocol, we decided to leave separate issue of the integration of an optical detector onto the instrument to a later stage. So for now, a 2-μL sample of the end product is taken from the disc-based reaction chamber, placed on the NanoDrop 3300 (Thermo Fisher Scientific Inc., Dublin, Ireland) pedestal for off-chip analysis.

As shown in Figure 6, the unknown concentration of nitrate and nitrite in the blood of two anonymous donors has been determined. A fresh whole blood sample was obtained and used on the same day. The control assays using the donated blood were initially carried out on a 96-well plate; and later implemented on the automated disc platform (Figure 6). For blood donor A, the levels of nitrate and nitrite are 20.7 μM and 8.7 μM, respectively. For blood donor B; the nitrate and nitrite levels are 10.7 μM and 5.4 μM, respectively. These levels of nitrate and nitrite are well within the normal reported clinically relevant range of 4–45.4 μM for nitrate and 1.3–13 μM for nitrite [27]. Figure 6 compares the results obtained from the 96-well plate and the automated centrifugal platform. As further shown in Figure 6; the high R² of the linear regression indicates a good correlation between the data obtained from the standard well plate and that obtained from this automated disc platform.

In order to validate the potential of the DF barriers for longer-term reagent storage, 30 μL of coloured water were loaded in the reagent chambers; the discs were enclosed in a heat seal bag and stored in the fridge at ˜ 4 °C for a period of 7 days (see Figure 7). The aim is to assess the integrity of the valving technique after this storage period. After 7 days of storage, the disc was taken out, examined and rotated on the spin stand. It was observed that liquid droplets had leaked into the pneumatic chamber of the section of the disc without the centrifugo-pneumatic DF valves, while the sections with DF valves remained sealed. This observation provides evidence that the DF valves provided enough vapour barriers to prevent reagents from creeping into the pneumatic chamber by capillary movement.