Todays detection limits are very low. Using various methods such as amperometric detection , surface plasmon resonance , and laser–induced fluorescence detection  it is possible to sense few attomolar (10-18 mol·L-1) concentrations of analytes.
The latter paper  is a report from 1996. Hence, the question arises wether it is possible to go further down the road of metric prefixes. Recently, some paper mentioned detection limits of zeptomolar (10-21 mol·L-1) concentrations  . Is that even possible?
First, what is a zeptomolar concentration? Let us take one kilogram of table sugar, which is simply saccharose (M = 342 g·mol-1) , and create a solution out of it! One kilogram is equal to 2.9 mol of saccharose. If we put this amount in 2.9 L of water, we get a concentration of one mole per liter. For zeptomole per liter we need 1·1021 times more water, i.e. 2.9·1021 L or 2.9·1021 dm3 or 2.9·109 km3.
I searched Wikipedia for a body of water (on Earth) with this volume. There is none. In fact, the sea itself is the largest body of water with a volume of 1.4·109 km3 , which is approximately two times of the volume we would need for our zeptomole solution. Therefore, half a package of table sugar thrown into the sea (assuming of course no one did that before, sugar is not consumed by anything, and the sugar is magically well distributed over the whole sea) gets you a zeptomolar solution of sugar (about 600 molecules per liter).
Is it possible to sense such a low concentration? Technically, analytical chemistry works in dimension of few mL down to some µL. For instant, typical injection volumnes of HPLC are 25–100 µL. Even ‘classical’ (wet chemical) methods only require some µL or few mL of a sample. A mL of a zeptomolar solution may or may not contain one molecule (odds: 6 out of 10). Thus, a preconcentration step would be necessary. However, it is not feasible (and often not possible at all) to take liters(!) of a sample and reduce the volume to gain an appropriate sample solution. Conclusively, it seems near to impossible to detect low zeptomolar concentrations of anything. Reasonable detection limits lie somewhere around 500 zM in a 300 µL volume (i.e. a total of 4 molecules) .
One has to pay attention not to confuse zeptomolar detection with zeptomole detection (or single molecule detection), though. For instant, one of the paper mentioned before is a review about microbeads and reports ‘the sensitivity of microbead arrays is down to the zeptomolar range’ . Looking at the corresponding reference (no. 66) it becomes clear that the detection limit of the referred DNA microsensor is about 600 molecules (1 zeptomole) . However, this limit was determined in a volume of 10 µL (not one liter), i.e. the concentration was ‘only’ 100 aM (still very good!).
The other paper about TNT and DNT detection refers to another work (in the same area) in the following way: ‘There is one report with an LOD at the zeptomolar level for the detection of TNT.’ . In contrast, the reference (17e) states the detection of TNT at sub–zeptomole levels, i.e. the presented gold mesoflower-approach can (theoretically, they only discuss it) detect TNT in 34 fL of a 100 ppt (about 440 pM) solution (i.e. a total of 9 molecules) .
In summary, it is hardly possible to sense low zeptomolar concentrations, but it is possible to detect zeptomole quantities (single molecules) in low volumes! Yeah, we definitely went further down the road 🙂
- W. Gao, H. Dong, J. Lei, H. Ji, and H. Ju, "Signal amplification of streptavidin–horseradish peroxidase functionalized carbon nanotubes for amperometric detection of attomolar DNA", Chemical Communications, vol. 47, pp. 5220, 2011. http://dx.doi.org/10.1039/C1CC10840A
- J. Ferreira, M.J.L. Santos, M.M. Rahman, A.G. Brolo, R. Gordon, D. Sinton, and E.M. Girotto, "Attomolar Protein Detection Using in-Hole Surface Plasmon Resonance", Journal of the American Chemical Society, vol. 131, pp. 436-437, 2009. http://dx.doi.org/10.1021/ja807704v
- D.B. Craig, J.C.Y. Wong, and N.J. Dovichi, "Detection of Attomolar Concentrations of Alkaline Phosphatase by Capillary Electrophoresis Using Laser-Induced Fluorescence Detection", Analytical Chemistry, vol. 68, pp. 697-700, 1996. http://dx.doi.org/10.1021/ac950650z
- S. Rödiger, C. Liebsch, C. Schmidt, W. Lehmann, U. Resch-Genger, U. Schedler, and P. Schierack, "Nucleic acid detection based on the use of microbeads: a review", Microchimica Acta, vol. 181, pp. 1151-1168, 2014. http://dx.doi.org/10.1007/s00604-014-1243-4
- M. Mohan, and D.K. Chand, "Visual colorimetric detection of TNT and 2,4-DNT using as-prepared hexaazamacrocycle-capped gold nanoparticles", Anal. Methods, vol. 6, pp. 276-281, 2014. http://dx.doi.org/10.1039/C3AY41824C
- "Sucrose - Wikipedia", Wikipedia, 2017. http://en.wikipedia.org/wiki/Sucrose
- "Sea - Wikipedia", Wikipedia, 2017. http://en.wikipedia.org/wiki/Sea
- S. Lee, and S.H. Kang, "Gold-nanopatterned single interleukin-6 sandwich immunoassay chips with zeptomolar detection capability based on evanescent field-enhanced fluorescence imaging", The Analyst, vol. 138, pp. 3478, 2013. http://dx.doi.org/10.1039/C3AN36914E
- J.R. Epstein, M. Lee, and D.R. Walt, "High-Density Fiber-Optic Genosensor Microsphere Array Capable of Zeptomole Detection Limits", Analytical Chemistry, vol. 74, pp. 1836-1840, 2002. http://dx.doi.org/10.1021/ac0156619
- A. Mathew, P.R. Sajanlal, and T. Pradeep, "Selective Visual Detection of TNT at the Sub-Zeptomole Level", Angewandte Chemie International Edition, vol. 51, pp. 9596-9600, 2012. http://dx.doi.org/10.1002/anie.201203810