More Related Content Similar to Kalafut optimizing srm - r2 - asms2016 (20) Kalafut optimizing srm - r2 - asms20161. Bennett S. Kalafut, Harald Oser. Thermo Fisher Scientific, 355 River Oaks Parkway, San Jose, CA 95134
RESULTS
Figure 5. Sensitivity of Q2 RF tuning to CID gas pressure
Collision cell RF amplitude tuning curves for CF3
- (m/z=69)
varying the collision energy with no argon in the collision
cell (top) and the collision gas pressure while holding the
collision energy fixed (bottom). The “collision energy “is the
portion of kinetic energy imparted to the ion by a DC offset
of the collision cell regardless of the presence or absence of
collision gas in the collision cell.
Table 2. Results of tandem optimization
At right (Figure 4) are
collision cell RF amplitude
tuning curves collected for
the same compound, with no
gas in the collision cell.
We observe (qualitatively)
the expected rightward shift
of the low-amplitude cutoff in
a curved quadrupole. This
shifts the optimum Q2 tuning
by tens of volts.
Figure 4. Sensitivity of Q2 RF tuning to collision energy
Figure 6. Optimization of Q2 RF amplitude for MS/MS Table 1. Gains from tuning collision cell RF amplitude for MS/MS
Signal intensity improvement from tuning the collision cell RF amplitude. Comparison is between SRM signal
intensity for a freshly tuned and calibrated TSQ Endura™ mass spectrometer using its standard determination of the
Q2 RF amplitude and the signal intensity when Q2 RF amplitude is tuned to optimize the transition of interest.
Improvements larger than a 50% increase of signal have been highlighted.
For each measurement CID gas was set to 3 mTorr, and collision energy set to its optimum value (not recorded.)
The MS/MS Q2 RF tuning curve for the 69 to 42
amu transition from Figure 6 is plotted here (green)
together with a simple heuristic generated by
multiplying the 69 amu and 42 amu tuning curves
from Figure 6 together and rescaling (blue). This is
the simplest possible model for the relationship
between MS/MS tuning and Q1 or Q3 MS tuning.
Figure 8. A simple approximation
ABSTRACT
The ion-optical properties of a collision cell are a significant determinant of the quantity of
product ions produced by collision-induced dissociation (CID) for MS2 analysis. The number of
product ions exiting the collision cell increase with both the quantity of precursor ions on stable
trajectories and the quantity of product ions on stable trajectories after dissociation.
To optimize Selected Reaction Monitoring (SRM) signals for desired transitions, we find the
collision cell RF amplitude which optimally balances precursor and product transmission, both
alone and in tandem with collision energy and collision gas pressure.
INTRODUCTION
For m/z larger than 100 amu, in TSQ Endura™ or Quantiva™ mass spectrometers (collision
cell frequency≈2700 kHz, radius r0=5mm) the high-amplitude cutoff is high enough to be of
little practical importance. In practice there is also a low-amplitude cutoff. The use of curved
collision cells to reduce the flux of neutral ions at the detector introduces a need for higher
(pseudo-)potentials to steer ions around the curve. For low q (𝑞 = 2𝑒𝑉
𝜔2 𝑟0
2(
𝑚
𝑧
)
) Syka and Schoen
[1] approximate the curve-induced motion away from the center of a collision cell as a function
of time as
𝑋𝑐 =
16𝐸 𝑧
𝑚𝑅𝑞2 𝜔2 1 − cos
𝑞𝜔𝑡
2 2
.
The inverse square dependence on q implies an inverse square dependence on the RF
amplitude V and (considering the m in the denominator) a linear dependence on m. We note
also the linear dependence on collision energy Ez .
Interaction with collision gas, non-quadrupole motion due to the use of flat electrodes, and
(mass-dependent) acceptance into the second quadrupole mass filter add further deviation
from idealized stability considerations resulting in single-ion (SIM) tuning curves similar to
those presented below.
The low-mass ion presented above (tetramethylpiperadine+H+, m/z=142.15 amu) optimizes at
an RF amplitude at which the high mass ion (Ultramark 1522+H+, m/z=1522 amu) is not
transmitted through the quadrupole at all.
Optimal transmission of product ions through the quadrupole collision cell is intuitively
expected to involve a tradeoff between precursor ion penetration into the cell, optimizing at
higher RF amplitude, and product ion exit, optimizing at lower RF amplitude. In the plots above
the tuning of the low mass ion (142 amu) appears relatively flat but there is a factor of 2
difference between optimal transmission and the plateau of the tuning curve for m/z=1522
amu.
The curves collected above were collected with no collision energy applied and no gas in the
collision cell. When collision energy is applied and collision gas is present, the tuning curves
narrow, reducing regions of overlap between high mass ion and low mass ion collision cell RF
and increasing the importance of per-transition optimization.
CONCLUSIONS
Given the potential for twofold or greater increase of the SRM intensity at the detector due to increased
flux of product ions out of the collision cell we recommend adjusting the collision cell RF amplitude per
CID transition to optimize MS/MS assays.
Figure 8 suggests the prospect of computation of the near-optimal collision cell tuning for given
precursor and product ions, based on a one-time calibration of the collision cell RF amplitude across
the instrument’s mass, collision gas pressure, and collision energy ranges using unfragmentable or
difficult-to-fragment ions. It is not clear that the heuristic approach generalizes to cases where there is
greater difference between precursor and product ion tunings or where the Q1 MS and Q3 MS tuning
curves are more complicated. Further study is necessary before a calibration procedure can be
developed.
Automatic tandem optimization of the collision cell RF amplitude and collision energy is being
considered for addition to the compound optimization feature in a future release of TSQ Endura™ and
TSQ Quantiva ™ instrument control software.
REFERENCES
1. Syka, J.E.P. and Schoen, A. “Characteristics of linear and non-linear R.F.-only quadrupole collision
cells.” Int. J. Mass Spectrom. Ion Processes 96, 97-109 (1990)
2. Trieber, Marco Alexander. Optimization for Computer Vision: An Introduction to Core Concepts and
Methods. Springer, London (2013)
3. Press, William H., Teukolsky, Saul A., Vetterling, William T., and Flannery, Brian P. Numerical
Recipes in C, 2nd Edition. pp. 413-414 Cambridge University Press, Cambridge (1992)
4. Powell, M.J.D. “An efficient method for finding the minimum of a function of several variables
without calculating derivatives.” The Computer Journal 7,155-162 (1964)
5. Kalafut, Bennett. “Automatic optimization of MS and MS/MS assays for dilute samples or weak
transitions” 63rd ASMS Conference Proceedings (2015)
TRADEMARKS/LICENSING
© 2016 Thermo Fisher Scientific Inc. All rights reserved. All trademarks are the property of Thermo
Fisher Scientific and its subsidiaries. This information is not intended to encourage use of these
products in any manner that might infringe the intellectual property rights of others.
Optimizing production of selected product ions for SRM analysis in a quadrupole collision cell
OPTIMIZATION FOR MS/MS
Plotted above in Figure 6 are collision cell RF tuning curves for the 69 to 42 amu CID transition
of imidazole (left) and the 1522 to 248.8 amu transition of Ultramark 1522 (right). Q2 RF
amplitude tunings were performed with the collision energy fixed to 20 and 50 eV, respectively,
which were the optima found using the compound optimization feature of the TSQ Quantiva™
instrument control software.
Due to the presence of difficult-to-fragment ions of nearly identical mass to precursor and
product in the same standard mix in the first case and the inefficient fragmentation of the
precursor ion in the second, these transitions are instructive. Plotted for comparison are the
Q2 RF tunings for acetonitrile (m/z=42) and CF3
- (m/z=69) and for Ultramark 1522, obtained
with collision gas pressure and collision energy identical to the MS/MS experiments.
comparison. The optimum tuning falls somewhere between that for the precursor and product
mass (with the tuning for the product mass assumed in the second case--compare to Figure
2.) trading off precursor stability in the collision cell for product stability through the rest of its
length and entry into Q3.
In the first case, the difference in tunings appears small—15 V between precursor and MS/MS
optimum—but the difference in efficiency is significant, with the MS/MS signal falling off to 80%
of its peak at the precursor ion optimum and 75% of its peak at the product ion optimum. In
the second case, 50% of the MS/MS signal is lost at the optimum tuning for the precursor ion.
TANDEM OPTIMIZATION
Collision cell RF amplitude tuning curves for MS/MS analysis. Left: 69 to 42 amu CID transition of
imidazole+H+ , with 42 amu (acetonitrile+H+) Q3 MS and 69 amu (CF3
-) Q1 MS tunings plotted for comparison.
Right. 1522 to 248.8 amu transition of Ultramark 1522, with parent ion Q1 MS tuning plotted for comparison.
SRM SIGNAL IMPROVEMENT
Representative improvements over standard determination of the Q2 RF amplitude are presented
in Table 1. As expected from heuristic considerations, the improvement is greatest when there is
a large difference between the precursor and product ion mass, and more pronounced in general
when parent, product, or both are low mass ions.
In normal operation, in MS/MS modes of operation (e.g. SRM), TSQ Endura™ and TSQ
Quantiva™ mass spectrometers apply the collision cell RF amplitude calibrated for the precursor
ion m/z, except when this RF amplitude is near the Mathieu stability boundary (q≈0.908) for the
product ion, in which case the product ion calibrated value is applied. Using the optimal value for
Q3 MS analysis at the product ion m/z in many other cases degrades the signal (data not shown.)
Routine mass-dependent tuning of the Q2 amplitude is done in the absence of collision gas and
with no extra DC offset (collision energy) applied.
Improvement by greater than a factor of two is possible if the experiment conditions (collision gas
pressure, collision energy) and tradeoff between precursor and product ion transmission are taken
into account by transition-specific tuning.
In an RF quadrupole, there is a maximum RF
amplitude above which there is no stable
trajectory for an ion of a given mass to charge
ratio. This corresponds to the point at which the
right edge of the Mathieu stability diagram meets
the x (RF amplitude) axis, and scales linearly with
m/z.
MATERIALS AND METHODS
Measurements presented here were taken on a Thermo Scientific™ TSQ Quantiva™ triple
quadrupole mass spectrometer, excepting those of Table 1, which were taken on a TSQ
Endura™. The primary relevant difference between the Endura and Quantiva in this investigation
is between r0 for the quadrupole mass filters: 4 mm for Endura and 6 mm for Quantiva.
CID transitions were selected from those of precursor ions present in Pierce™ Extended Mass
Range Calibration Solution (Thermo Fisher Scientific catalog number 88340), infused at a rate of
5 µL/min., using argon as collision gas.
Tuning curves for individual system voltages were established by continuously varying the voltage
while monitoring an SRM transition, at the rate of one full scan of the voltage range per second.
All curves presented here are averages of 60 scans, taken over the course of a minute.
Tandem tunings were performed using the “taxi cab” fixed direction set method [2-3], which
converged after 3-4 iterations, eliminating any need to update the direction set (as in Powell’s
Method [4]) to speed up convergence. At each iteration, the bounds of the line scan were fixed to
the intersection of the scan vector and instrument voltage limits. The line optimization at each
step was performed not on the raw data but on a kernel intensity function estimate [5] using the
“oversmooth” bandwidth selection rule.
Figure 3
.
Figure 2.
Precursor Mass
(amu)
Product Mass
(amu)
Untuned
TIC
Tuned
TIC
Improvement Factor
1522 248 180 380 2.1
322 183 39000 39000 1
322 154 48000 49000 1.02
322 109 67000 70000 1.04
142 69 27000 44000 1.63
142 41 6400 11000 1.71
142 39 5300 9400 1.77
Figure 1.
Qualitative features of Figure 6 suggest a naïve
model in which we treat the effects of Q2
detuning as a multiplicative loss for both the
precursor and product. This approach disregards
any differences in the rate of precursor production
per unit length, amount of time spent in the
metastable state prior to dissociation, or any ion-
optical details of where and how precursors or
products on unstable trajectories are lost.
When applied to the 69 amu and 42 amu tuning
curves, this approach recovers the peak location
and some of the qualitative features of the 69 to
42 amu tuning curve. The domain in which this
approximation is useful is not clear, and merits
further study. If it is generally reliable, it provides
a means to compute a reasonably good collision
cell RF amplitude using Q2 MS and Q3 MS Q2
RF tuning curves collected and stored while
tuning and calibrating the instrument, reducing
the need for time-consuming compound
optimization.
Cooling and retarding of the
ion beam due to interaction
with collision gas has a more
complicated effect on the
collision cell RF tuning. In
Figure 5 at right the location
of the original optimum stays
fixed even as the intensity
(normalized here for clarity)
varies by 5 orders of
magnitude. A second local
optimum at a very low RF
amplitude emerges at 2.5
mTorr and becomes the
global maximum at higher
collision gas pressures.
Figure 7. Sensitivity to collision energy
change at high CID gas pressureThe result plotted in Figure 4 and
the formula provided in the
Introduction both suggest that the
tuning of collision energy and
collision cell RF amplitude are
coupled. Insufficient RF amplitude
for confinement and steering of the
ion beam in the curved collision
cell, for example, may cause the
collision energy that optimizes the
MS/MS signal to be de-tuned from
the best value for fragmentation.
Figure 7, at right, shows that at
relatively high gas pressures the
relationship between Q2 RF tuning
and collision energy becomes
complicated, with the shape of the
tuning curve changing as more collision energy is applied, in addition to a gradual drift of the
local optima.
To account for the coupling between collision energy and collision cell RF optimization and the
potential tradeoffs between CID efficiency and ion-optical transmission, collision energy and Q2
RF amplitude should be optimized in tandem. Table 2 presents representative results,
comparing tandem and solo tuning of collision energy.
Precursor Mass (amu) Product Mass (amu)
Collision energy,
tuned alone (eV)
Collision energy,
tandem tuning (eV)
1.5 mTorr
1522 1077 55 55
1522 249 52 55
3.5 mTorr
1522 1077 44 55
1522 249 41 55
322 228 35 35
322 154 45 45
322 109 55 54
69 42 22 21
113 69 10 5
Comparison of tandem tuning of collision energy and collision cell RF amplitude with tuning of the collision energy
alone. Tandem tunings were obtained by direction set optimization (see Materials and Methods). Solo tuning was
performed using the compound optimization feature of TSQ Quantiva™ instrument control software, version 2.0.
Major differences have been highlighted. 10 eV and 5 eV are the respective lower limit of the standard compound
optimization software and practical instrument operation; the results for the 113-69 CID transition are considered
equal.
Use of the correct RF amplitude for beam steering and confinement allows the application of another 11-14 eV of
collision energy to the difficult-to-fragment fluorocarbon molecule Ultramark 1522 without reduction of signal intensity
due to interaction with the collision gas or losses caused by the curved collision cell.