Divide the two scalings and the argument fits on one line. Linear chiroptical response grows roughly with the plasmonic enhancement factor P, while the nonlinear circular dichroism used here grows roughly with P squared, according to Govorov [6]; the ratio of the two is therefore P itself [13]. The advantage over a linear measurement tracks the field enhancement rather than sitting at some fixed multiple, and it does so only inside the small volume where the enhancement is real. That volume is what the spiral geometry is for: the structures amplify both the electric and the so-called superchiral near fields, and the signal appears as circularly polarized second-harmonic generation when chiral molecules interact with the metal surface [2].
The demonstrated sensitivity is about 11 picomolar for adsorbed bovine serum albumin [3], which is 1.1 x 10^-11 moles per litre [14], with a figure of merit reaching 3,260 per micromolar [4]. BSA is a convenient test protein. Whether a small chiral drug molecule behaves the same way on these spirals is the open question, and the account does not report a pharmaceutical substance run on the platform [15].
The source's own framing of the problem is worth keeping in view: conventional circular dichroism and optical rotatory dispersion are the established linear methods in research and industry [7], and their signals are often weak enough to make sensitive detection difficult [8]. The consequence of getting handedness wrong is the thalidomide case, where one enantiomer sedates and the other can cause birth defects [9]. For a quality control bench, though, the more interesting result is not the picomolar figure but the mirror-image pair of chips: left- and right-handed spirals respond preferentially to matching molecular forms, so enantiomer composition can be read out of a racemic mixture without a chromatographic step first [5].
The thing this does not tell you is how well that readout holds up. The writeup says composition can be quantified but gives no accuracy figure and no range of enantiomeric excess over which the response stays usable [12]. It also reports no same-sample comparison against a commercial CD or ORD instrument, no acquisition time, and nothing on chip-to-chip reproducibility [11]. A spiral array is a fabricated object, and a sensor whose response depends on how this week's array came out needs a calibration story before it competes with a column. I would take the scaling relationship as the durable result here, conditional on P staying large where the molecules actually bind; the path to a production instrument runs entirely through the numbers the account omits [11].
