Instrument Detection Limit (IDL)

as a Measurement of Intrinsic MS Sensitivity

Internal Training for Account Managers, Product
Specialists, and Applications Engineers

Patrick Batoon, Ph.D

Product Manager of Triple Quadrupole LC/MS, MassHunter Acquisition for
LC/TQ, LC/MS Ion Sources, and Nitrogen Generators
Mass Spectrometry Division

1 Agilent Restricted - DE44193.5966435185

Learning Objectives

1. Understand the concept of Instrument Detection Limit (IDL)

• Understand the definition of Signal to Noise (SNR)
• Recognize the difference between sensitivity and ion signal

2. Learn various methods in which SNR can be calculated

• Realize the drawbacks of SNR as a descriptor of instrument performance

3. Understand how IDL is calculated

• Be able to educate your customers through Agilent application notes and resources

2 Agilent Restricted - DE44193.5966435185

Noise in MS Instrumentation Noise can be
measured by hand
Dramatic reduction in noise over the past 20-30 years

Interference from electronics and lab equipment

Mass spectral and chromatographic interference
originated from electronic & chemical noise

Improvements in HW science & engineering

Higher sensitivity → Greater detection of chemical
interferences

Tandem mass filtering & Improved Ionization

Drastic reduction in chemical interferences

Application of digital signal processing

Realtime smoothing, burst detection, software
enhancements…
Ultra low noise baseline

3 Agilent Restricted - DE44193.5966435185

Signal & Sensitivity – Defined by IUPAC
Increased Slope = Increased Sensitivity

It’s important to remember that sensitivity Response vs Amount of Analyte

has a strict IUPAC definition based on
“the slope of the calibration curve”. 200

175
In Mass Spectrometry, more ions = greater signal.
150

MS/MS techniques results in less ions, but quieter 125

Response
baselines. Higher Signal-to-Noise (SNR), but lower 100
IUPAC sensitivity.
75

50

High SNR ≠ High Sensitivity 25

When people refer to sensitivity, 0 25 50 75 100

Amount of Analyte
125 150 175

they usually mean lower detection limits.

4 Agilent Restricted - DE44193.5966435185

SNR an Industry Standard with Even More Variables
Widely used in GC/MS and LC/MS analysis

Noise
Signal Height

Signal Height
SNR =
Noise

Noise calculations
Zero-to-Peak, Peak-to-Peak, Peak-to-Peak
from Drift, Root-Mean-Square (RMS),
Auto-RMS, ASTM, etc…

Method of Noise calculation is not standardized and often not reported

5 Agilent Restricted - DE44193.5966435185

Various SNR Algorithms
SNR varies by Noise calculation algorithm

Same SNR approach, but various calculations for noise

Algorithm Noise SNR
Zero-to-Peak 8.1 counts 7.5
Analyte Signal = 61 counts Peak-to-Peak 16.5 counts 3.7
Peak-to-Peak from drift 15.6 counts 3.9
Root-Mean-Square 3.7 counts 16.6

Noise Region

6 Agilent Restricted - DE44193.5966435185

Various Noise Widths
SNR is dependent on noise region and width

Root Mean Square Algorithm

Analyte Signal = 61 counts

SNR = 6.1
SNR = 11.5

SNR = 14.3
SNR = 38.7
Noise = 5.3 counts
Width = 0.5 min Noise = 4.3 counts
Width = 0.25 min Noise = 9.9 counts
Noise = 1.6 counts
Width = 0.1 min Width = 1 min

7 Agilent Restricted - DE44193.5966435185

SNR as an Instrument “Sensitivity” Descriptor
Unreasonable to predict true detection limits of analytes
6495 Triple Quadrupole LC/MS
Manufacturers often report SNR values to
demonstrate the sensitivity of the instrument. 0.0029 2875896.5

“SNR for 1pg of Reserpine” 1pg Reserpine

Height = 8340.1 counts
Noise = 0.0029 counts
Can extrapolate down to 3:1 ??
SNR = 2,875,896
(1pt calibration not recommended!!)

502,415:1 = 1 picogram
200
Actual 6495 spec
Signal-to-Noise (SNR)

175
is <0.6 fg
150 3:1 = 1.04 attograms
125
100
75
50
Limit of Detection
25 10:1 = 3.48 attograms
0
0 25 50 75 100 125 150 175
Amount of Analyte

8 Agilent Restricted - DE44193.5966435185

Limits of Detection (LOD) Determined by SNR
Consensus and Regulatory bodies have provided definitions for minimum acceptable SNR ratios for
analytes

European Commission
Guidance Document on the Estimation of
LOD and LOQ for Measurements in the Field
of Contaminants in Feed and Food

“different instrument manufacturers do

not necessarily use the same
Minimum LOD SNR Values by Agency
algorithms in their data processing
US EPA 2.5-10 : 1
software for calculating S/N ratios
US FDA 2-3 : 1
despite applying the same noise
Eu. Medicines Agency 2-3 : 1
approach.”

9 Agilent Restricted - DE44193.5966435185

What is a meaningful metric for the performance of an instrument?

SNR was useful when noise was easily measured

• Many confounding factors defining noise
• MS has undergone an incredible evolution, our metrics have not

Use an alternate measure based on these criteria:

• Not parameter dependent
• Reliable & reproducible signal
Instrument Detection
• Confident that signal ≠ noise Limit (IDL)

10 Agilent Restricted - DE44193.5966435185

Precision worsens at lower concentration
Measurement precision reveals confidence of measurement

Peak Area %RSD vs On-Column Concentration

Reproducibility and peak area deteriorates rapidly as the
50%
instrument’s detection limit is approached.
45%
40% Precision decreases explosively at
35% lower concentrations

Peak Area %RSD

2.5 fg Reserpine 5 fg Reserpine 10 fg Reserpine 30%
%RSD = 48.5% %RSD = 26.0% %RSD = 12.5% 25%
20%
15%
N = 10 injections

10%
5%
0%
0 50 100 150 200
Conc (fg/uL)

IDL determines the most minimally tolerable

point of uncertainty

11 Agilent Restricted - DE44193.5966435185

 Instrument Detection Limit (IDL)
Measurement based on Student’s t-test at 99% confidence

Known low-level St.Dev

IDL = ∙ Conc ∙ t(α=0.01)
concentration experiment Avg Area
Instead of measuring noise, response statistics are measured.

IDL
concentration
99% Confident that Signal ≠ Noise

Baseline Noise
Baseline Noise

12 Agilent Restricted - DE44193.5966435185

IDL is based on the US EPA’s – Method Detection Limit (MDL)
Purely based on overall instrument performance – No matrix effects

St.Dev
IDL = · Conc · t(α=0.01)
Avg Area
99% confident that Signal ≠ Noise

Not an obscure Agilent spec – based on US EPA’s Method Detection Limits (MDL, Rev 1.11)

"The method detection limit (MDL) is defined as the minimum concentration … reported with 99% confidence
that the analyte concentration is greater than zero and is determined from analysis of a sample in a given
matrix containing the analyte."

The instrument detection limit (IDL) is defined as the minimum concentration of a substance that
can be measured and reported with 99% confidence that the analyte concentration is greater than
zero and is determined from analysis of a sample in solvent containing the analyte.

13 Agilent Restricted - DE44193.5966435185

Example IDL Determination

General Technique & Practical Considerations

1. Determine amount to inject on-column
2. Obtain response statistics
3. Verify calculated IDL value

14 Agilent Restricted - DE44193.5966435185

Example IDL Determination
Ultivo Triple Quadrupole LC/MS w/ 1290 Infinity II LC

Determine the appropriate amount to inject on-column

RSD=10-15% is a good place to start

Method Parameters: 100%

Peak Area %RSD vs On-column Concentration

90%
Samples Injected 80%
2.5, 5, 10, 20, 40, 80 fg Reserpine On-column

Peak Area %RSD

70%
60%
1290 Infinity II LC 50%
ZORBAX RRHD Eclipse Plus C18, 2.1x50mm, 1.8µm 20 fg on-column used for
40%
H2O and Acetonitrile w/ 0.1% FA IDL determination
30%
20%
Ultivo LC/MS Time (min) %B
MRM 609.3→195.1 10%
0.1 10
Fragmentor = 150 V 1.0 98 0%
CE = 41 V 2.0 98 0.0 20.0 40.0 60.0 80.0 100.0
Dwell Time = 200 ms 2.01 10 On-column Concentration (fg/uL)
Stoptime 3 min

15 Agilent Restricted - DE44193.5966435185

Example IDL Determination
Ultivo Triple Quadrupole LC/MS w/ 1290 Infinity II LC

Obtain peak area Average and St.Dev

At least N=8 injections

St.Dev
IDL = · Conc · t(α=0.01)
Avg Area 20 fg Reserpine
N = 8 Injections
t(α=0.01) = 2.998
0.71 counts
IDL = · 20 fg · 2.998 Avg Area = 16.75 counts
16.75 counts St.Dev = 0.71 counts

IDL = 2.54 fg Reserpine on-column

Noise was not explicitly considered

Result is based purely on sample measurements
and is less subject to manipulation.

16 Agilent Restricted - DE44193.5966435185

Example IDL Determination
Ultivo Triple Quadrupole LC/MS w/ 1290 Infinity II LC

Validate calculation with actual experiment

IDL = 2.54 fg Reserpine on-column 20 fg Reserpine

N = 8 Injections
t(α=0.01) = 2.998
Avg Area = 16.75 counts
St.Dev = 0.71 counts
A series of 8 injections were carried out at
2.5 fg Reserpine on-column.

Actual 2.5 fg on-Column

Demonstrated that Signal ≠ Noise and IDL
can be reproducibly & confidently detected.

17 Agilent Restricted - DE44193.5966435185

What’s the more informative and meaningful metric?
Evaluate what’s more important and informative

20 fg Reserpine 1 pg Reserpine
N = 8 Injections N = 8 Injections
t(α=0.01) = 2.998 Avg Height = 195.5 counts
Avg Area = 16.75 counts St.Dev Height
St.Dev = 0.71 counts = 4.0 counts
Avg SNR = 20,559:1

Actual 2.5 fg on-Column

≥ 99% confidence Extrapolated to 0.2 fg
Unknown confidence

18 Agilent Restricted - DE44193.5966435185

Endorsements & articles
Experts have recognized that SNR has gotten out of hand

“signal increased, but there was no increase in the

number of ions detected, and therefore, no change
to the real detection limit. This allows S/N to be
adjusted to any arbitrary value without changing the
real detection limit.”

Greg Wells, Harry Prest, Charles William Russ IV

“Disconnect between SNR for routine method and SNR for

an installation test has increased significantly over the last
decade and led to vendor SNR sensitivity specifications
that do not accurately represent the instrument’s routine
performance.”

Terry L. Sheehan and Richard A. Yost

19 Agilent Restricted - DE44193.5966435185

Deployable IDL Calculator
Excel worksheet is available on the field portal

Agilent IDL Calculator Worksheet

Input analyte response:

• Automatically calculates measurement

statistics, critical t-value, and IDL

• Includes tips for good characterization

20 Agilent Restricted - DE44193.5966435185

Summary – Why Instrument Limit of Detection (IDL)?
SNR as a “sensitivity” & performance metric is no longer usable
• Very useful when noise floor was measured by reasonable means
• MS techniques have evolved dramatically to filter noise – essentially “zero noise” baseline

IDL is an informative metric on the intrinsic performance of the instrument

• Based on peak area statistics of replicate injections – 99% confident that Signal is detectable
• Gives user the bottom limit for the instrument’s true limits of detection & quantitation – “sensitivity”
• Similar to US EPA’s MDL determination, commonly used in GC/MS

The ultimate goal is to produce scientifically rigorous results and end the misleading SNR
“games”. Mass spectrometry technology has changed and our metrics need to adapt with it.

View the frequently asked questions webpage at https://www.agilent.com/en/support/liquid-chromatography-mass-spectrometry-lc-

ms/instrument-detection-limit or search “Agilent IDL” online

21 Agilent Restricted - DE44193.5966435185

Acknowledgements

LC/MS Quadrupole HW R&D Visit the Agilent booth – Exhibit 800

• Shane Tichy, Anabel Fandino MP 387 – Dynamically Adjusted Mass Axis Delay Times for
Optimized MRM Data Acquisition
• Laura Pollum, Haopeng Wang, Behrooz
Zekavat TP 292 – Improved Peak Detection for Mass Spectrometry
via Augmented Dominant Peak Removal
• James Bertsch, Layne Howard, Tim Woods
WP 405 – Ion Transmission through dual Field Tapered
• Jessica Schubert, Mike Schoessow, Glen Multipoles: Cyclone Ion Guide and Vortex Collision Cell
Ingle, Rick Carlquist
• Daniel Abramovitch, Huy Bui, Ken
Krzeczowski, Huy Nguyen

IDL Contributors Video and Technical Note at

https://www.agilent.com/en/products/mass-spectrometry/idl
• Terry Sheehan, Harry Prest or search “Agilent IDL” online

22 Agilent Restricted - DE44193.5966435185