For sample volumes <1 mL, a series of glass receptacles was used, with thin polypropylene tubing and disposable pipettor tips (Cat. no. 1235A68; Thomas Scientific, Swedesboro, NJ, USA) connecting to syringes with Luer fittings (Figure 1, B and C). The following four additional refinements were developed for small volumes.Using reflected light to evaluate fluid level
For measuring relatively small changes in fluid levels, observation of a meniscus was unreliable. More precise results were obtained by defining the criterion level instead at the top edge of the receptacle, with the fluid surface nearly flat. The degree of flatness was visualized by illuminating the surface from an angle of ~45° with a bright white light-emitting diode (LED; TW00 series; Agilent Technologies, Santa Clara, CA, USA) positioned a few centimeters from the receptacle and observing the pattern of reflected light under magnification (Figure 1C). To align the path of the reflected LED light with the optical axis of a dissecting microscope, the LED was mounted on a micromanipulator (MM33; Narishige, East Meadow, NY, USA; or equivalent), and the LED angle and position were adjusted. If the fluid surface is either convex or concave (Figure 2A), the image of the LED appears as a small bright spot. As the fluid surface becomes increasingly flat, the bright spot expands to fill the entire surface.
Aqueous solutions adhered to the top edge of a glass container much more reliably than to plastic. For ~0.1- to 1-mL volumes (Figure 1B), a small (2.5-mL) glass vial with a hole drilled in the bottom served as the receptacle. For smaller volumes, a series of receptacle sizes (Figure 1C, inset) was fashioned from glass Pasteur pipets by scoring and breaking the tapered portion at the desired diameter, then grinding the broken end flat with a spinning abrasive disk (Cat. no. 413; Dremel, Racine, WI, USA). The hydrophobicity of the ground-glass surface was increased by periodically coating it with a trace of silicone grease. The smallest receptacles were made from capillary glass tubing with inner diameters of 1.0 or 0.5 mm (Cole-Parmer, Vernon Hills, IL, USA). Receptacle ends were bent in a flame to a 90° angle and connected to the syringe via a pipettor tip.
For each receptacle and measurement sequence, a criterion level was chosen in advance by selecting a specific local reflected light pattern. As in the basic procedure above, fluid was first pushed up to produce a slightly convex meniscus. Next, the fluid was brought down close to the preselected level, causing the spot of reflected light to expand until small “fingers” of reflected light extended outward. The point at which one of the fingers first contacted the glass edge (Figure 2A) could be chosen as the criterion level. Slight changes in the color of the reflected light were also very useful for identifying a criterion level.Accounting for evaporation
When using small receptacles in low humidity, sample volumes can be underestimated due to evaporation from the fluid reservoir, which alters the reflected light pattern within seconds. Rather than attempting to eliminate evaporation, this source of bias was accounted for by taking a time-series of measurements. After adjusting the fluid to a point slightly above the criterion level, the fluid was permitted to reach the criterion level passively, the time was recorded to the nearest second, and then the micrometer position was noted. This procedure was repeated four to five times before and after introducing the sample. The resulting series of micrometer positions (Figure 2B) defines two straight lines, which theoretically share the same slope. The slope of the lines reflects the evaporation rate, and the vertical distance between the pre-addition and post-addition lines corresponds to the true volume of the sample. Linear regression methods (12) were used to estimate the slope of these lines and the distance between them.Matching receptacle to sample dimensions
Given that the smallest volume measurements are based on judging the flatness of a circular fluid-air interface, the ability to resolve a given change in volume depends not on the total volume of the receptacle, but on the diameter of the receptacle opening. This was confirmed by examining errors in measured volumes as a function of the ratio of reference ball-to-receptacle diameters.