Restoring the validity of MS sensitivity specifications
Why Agilent is using IDL to replace signal-to-noise
By Tom Doherty Agilent Triple Quadrupole GC/MS Product Manager
and Na Pi Parra Agilent Triple Quadrupole LC/MS Product Manager
Dividing by zero never yields a meaningful result. Yet dividing by zero is a fundamental problem that challenges the validity of signal-to-noise (S/N; SNR) specifications as ultra-low MS noise approaches zero. Ultra-low noise is the result of a combination of factors:
Noise has steadily decreased with advances in instrument design.
Noise is lower in new, ultra-clean systems.
Noise is reduced by the selectivity of MS/MS and high resolution MS.
Noise from sample matrix is absent in pure check samples.
Noise algorithms vary; there is no industry standard that defines the duration of the noise segment or the position of the noise segment relative to the signal peak.
Noise reduction through digital filtering also varies among vendors.
The validity of S/N has been further compromised by steady increases in MS signal without a proportional decrease in the amount injected on-column. The net result has been an escalation of S/N from 10:1 in the 1980-90s and now to greater than 100,000:1. Because many international regulatory agencies define an appropriate S/N ratio as < 10:1 for determining detection limits, these very high S/N ratios are problematic at best.
A final consideration is the fact that the S/N sensitivity specification is not statistically based and as such does not represent good science. As an example, a S/N “pass” condition is acknowledged when a single, random injection exceeds the published specification, but did the previous injections almost pass? If not, can you be confident that the next three injections will also pass spec? Maybe the pass injection was just one lucky analysis that finally exceeded the specification. It is obvious that change was needed.
Figure 1. GC/MS IDL test with overlay of 8 consecutive injections of 2 fg OFN (LOD = 300 attograms).
IDL – a statistically based standard
Rather than trying to ‘fix’ S/N measurements, Agilent’s GC/MS and LC/MS teams have agreed that it is time for a new sensitivity metric (Figure 1). While this metric differs from S/N, it is a time-tested, statistically based calculation that has been promoted by the US EPA since the early 1990s and replicated by many regulatory agencies worldwide. Because the new metric applies to the instrument performance and not to an entire method, Agilent adopted the term Instrument Detection Limit (IDL) rather than using the US EPA’s term, Method Detection Limit (MDL).
IDL is more effective than S/N ratio because IDL is statistically derived (Student’s t test with 99% confidence limit). IDL is based on consecutive injections, not a single, random injection and allows you to confirm the performance of all components (autosampler, chromatograph, and MS). IDL requires a low amount to be injected (very low fg, not pg), and more importantly, IDL directly correlates with ion count.
LC/MS and GC/MS sensitivity specifications
In Table 1 we show the LC/MS and GC/MS sensitivity specifications with IDL results using varying samples.
Triple Quad LC/MS Specification
6495A Introduced June 2014
Positive, using reserpine
12.5 fg measured @ 20 fg
0.75 fg measured @ 1 fg
Negative, using chloramphenicol
12.5 fg measured @ 20 fg
0.75 fg measured @ 1 fg
Triple Quad GC/MS Specification
7010 Introduced June 2014
EI, using octafluoro-naphthalene
4 fg measured @ 10 fg
0.5 fg measured @ 2 fg
Table 1. IDL test specifications using reserpine, chloramphenicol, and octafluoro-naphthalene.
Response precision is an essential metric
For all MS systems, the fundamental goal is to produce as many ions as possible and transmit the maximum number of ions to the detector. Ion count is the true measure of MS sensitivity. Furthermore, higher ion count is essential to good ion statistics and optimal precision. Precision, in turn, holds benefits for both quantitative analysis (peak area or height) and qualitative analysis (stable spectral response, stable SIM, or MS/MS ion ratios). For S/N, dividing by an ultra-low noise value approaching zero can hide problems of low ion count. Precision and IDL are better sensitivity tests at installation, and precision continues to provide valuable information about performance during normal operation.
Figure 2. LC/MS IDL test with overlay of 10 consecutive injections of 1 fg reserpine in positive mode (left) and chloramphenicol in negative mode (right) using the new 6495 Triple Quadrupole LC/MS. The IDL of reserpine is 200 attograms and the IDL of chloramphenicol is 270 attograms.
A higher standard to confirm the highest performance
S/N will continue to be a valuable metric for the analysis of real samples where there is significant matrix contribution to the baseline noise. This will even hold true for highly selective MS/MS and hi-res methods with complex matrices. But for installation checks, Agilent recommends the IDL specification to guarantee the best performance for your new GC/MS or LC/MS system – a guarantee with a statistical foundation, as shown in Figure 2. IDL is not just a ‘factory spec’. IDL should be demonstrated at your installation to confirm the performance of every component in the system: the lowest IDL from best systems to support your best science.
GC/MS IDL test with overlay of 8 consecutive injections of 2 fg OFN (LOD = 300 attograms).
LC/MS IDL test with overlay of 10 consecutive injections of 1 fg reserpine in positive mode (left) and chloramphenicol in negative mode (right) using the new 6495 Triple Quadrupole LC/MS. The IDL of reserpine is 200 attograms and the IDL of chloramphenicol is 270 attograms.