Supporting Information:
Infrared Photodissociation Spectroscopy of
Water-Tagged Ions with a Widely Tunable
Quantum Cascade Laser for Planetary Science
Applications
Tyler M. Nguyen,
†
Douglas C. Ober,
†
Aadarsh Balaji,
†
Frank W. Maiwald,
‡
Robert P. Hodyss,
‡
Stojan M. Madzunkov,
‡
Mitchio Okumura,
∗
,
†
and Deacon J.
Nemchick
∗
,
‡
†
Division of Chemistry and Chemical Engineering, California Institute of Technology,
Pasadena, California 91125, United States
‡
NASA Jet Propulsion Laboratory, California Institute of Technology, Pasadena,
California 91109, United States
E-mail: mo@caltech.edu; Deacon.J.Nemchick@jpl.nasa.gov
S-1
Contents
Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S-3
Figure S1: Experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S-9
Figure S2: Additional experimental IR action spectra and fitted spectra of
alanine isomer mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . S-10
Figure S3: IR action spectra compared to predicted IR spectra . . . . . S-11
Table S1: Calculated binding energies of amino acids . . . . . . . . . . . . . . S-12
Figure S4: Amino acid water tagging fraction versus binding energy . S-13
S-2
Materials and Methods
Materials
O-phospho-L-tyrosine (
≥
95.0 %), glycine (
≥
99.0 %), L-
α
-alanine (
≥
98.0 %), L-phenylalanine
(
≥
98.5 %),
β
-alanine (
≥
99 %), sarcosine (
≥
98 %), glycylglycine (
≥
99 %), and L-proline (
≥
99 %) were purchased from Sigma-Aldrich and used without further purification. Methanol
with 0.1 vol % formic acid of LC-MS grade and water with 0.1 vol % formic acid of LC-MS
grade were obtained from Fluka Analytical and Fisher Chemical, respectively.
Experimental Method
Liquid samples used with the electrospray ionization (ESI) source were prepared by dissolving
each amino acid in a water/methanol (1:1) mixture with 0.1 vol % formic acid. The final
concentration of each amino acid solution was 100
μ
M
. For mixtures of amino acid isomers,
the total sample concentration was 100
μ
M
.
The experimental setup depicted in Figure S1 consists of a continuous-wave (CW) MIRcat-
QT quantum-cascade laser (QCL, Daylight Solutions) and a Thermo-Finnigan LTQ mass
spectrometer. Radiation generated from the QCL source was focused into the center of the
linear quadrupole ion trap using an optical train and 250 mm focal length zinc selenide
(ZnSe) lens. The mass spectrometer was outfitted with a custom flange that hosted a ZnSe
window to allow radiation to enter the ion trap region of the mass spectrometer system. A
beam pick off was incorporated into the beam bath for monitoring laser powers with a power
meter (Thorlabs PM100USB with S401C head).
Ions were prepared with the commercial ESI source with an injection rate of 4.0
μ
L/minute
and entered the mass spectrometer through a heated metal capillary inlet (358 K). Gener-
ated parent amino acid ions were guided to the linear ion trap (room temperature) with ion
optics and ion guides native to the commercial LTQ mass spectrometer. To assist in the for-
mation of water tagged amino acids, the helium buffer gas required by the commercial mass
S-3
spectrometer to stabilize ion trajectories was seeded with water vapor using a bubbler. The
approximate pressure of the buffer gas within the ion trap region of the mass spectrometer
was
∼
1 mtorr, with water vapor content constituting a fraction of that pressure. Trapping
of parent ions, water tagging, irradiation, and subsequent mass analysis of the parent and
water tagged ions all occurred within a single linear ion trap stage.
The MIRcat-QT is operated through its own software package provided by the vendor
(Daylight Solutions). This work deployed the step-and-measure mode over the operational
frequency range of 925 - 1670 cm
−
1
using a step size of 5 cm
−
1
. The LTQ mass spectrometer is
operated through the commercial XCalibur software package in the mass selective MS
n
mode.
Each photodissociation mass spectrum is
<
640 ms in duration and includes a prescan/trap
filling time (
∼
1-100 ms; dynamically varied depending on incoming ion flux), trapping time
(500 ms; fixed), ion ejection time (20 ms; fixed), and dead time (20 ms; fixed). Infrared
(IR) radiation is not modulated and continuously irradiates the ion trap during the fill, trap,
ejection and dead periods. The relative ion counts of the parent and water tagged species are
the outcome of competing processes (complex formation, photodissociation) and are analyzed
after each photodissociation mass spectrum. A total of 30 individual photodissociation mass
spectra were collected at each frequency set point with each IR action spectrum composed
of
∼
4000 total photodissociation spectra. Synchronization of the mass spectrometer and
stepping of the laser source is accomplished by TTL pulse communication, triggered through
the built-in "contact closure" connection from XCalibur and the LTQ (originally for HPLC
autosampler interfacing) and feedback from the laser source.
S1
Photodissociation yields can
be calculated from the experimental data after converting the commercial mass spectrometer
raw binary files to readable mascot generic format (mgf) files using MSConvert, a command
line program associated with the ProteoWizard software package.
S2
S-4
Photodissociation Yield and Spectrum Processing
Photodissociation (PD) yield can be represented either as the growth of a parent ion signal
(S
P
) or as a depletion of a water tagged complex ion signal (S
C
). For favorable tagging
conditions, where S
C
/(S
P
+S
C
) > 0.1 observed in the absence of IR radiation, yields are
presented by plotting integrated parent ion signal as a function of wavelength with response
normalized to laser power.
For adverse tagging conditions, where S
C
/(S
P
+S
C
) < 0.1 observed in the absence of IR
radiation, this work adopts a modified version of the yield used in messenger assisted infrared
photodissociation techniques
S3–S5
that is shown in Equation 1.
yield
PD
=
−
ln
1
−
S
P
S
P
+
S
C
(1)
Standard messenger assisted infrared techniques deploy independent ion traps supporting
complex formation with subsequent mass analysis stages. This allows for untagged parent
ions to be discarded or otherwise isolated before irradiation/analysis. These photodissoci-
ation mass spectra provide unobscured views of both the photodissociation product parent
ions (S
P
in Equation 1) and remaining complex ions (S
C
in Equation 1). In contrast, this
work deploys a single stage for complex formation, irradiation, and analysis meaning parent
ions produced from photodissociation events are obscured by large counts of untagged ions.
This work alternatively defines the value for parent ions formed from dissociation events in
terms of the complexed ions observed in the presence and absence of infrared radiation.
S
P
=
S
C,
[
sans Radiation
]
−
S
C
(2)
This definition reasonably assumes the only photodissociation product of a messenger
tagged ion is the parent ion species and that the ion counts for the complex species in the
absence of IR radiation would remain fixed/constant over the course of the experiments.
Monitoring the levels of the complex species before and after experimental trials confirmed
S-5
the second case to be true. Substitution of Equation 2 into Equation 1 yields:
yield
PD
=
−
ln
1
−
S
C,
[
sans Radiation
]
−
S
C
S
C,
[
sans Radiation
]
.
(3)
Which can be reduced to:
yield
PD
=
−
ln
S
C
S
C,
[
sans Radiation
]
.
(4)
Equation 4 represents the depletion form of the PD yield equation, which can then be
normalized to laser power at each frequency set point. Infrared action spectra are generated
by plotting normalized yield as a function of frequency. For each presented spectrum a
smoothed version is also plotted to aid in visualization (Savitzky-Golay filter with a window
size of 8 and a polynomial order of 3). For any quantitative results, such as determination of
relative concentrations of components of a mixture, analysis was executed on un-smoothed
infrared action spectra.
Results from the mixture analysis study are presented in Figure S2. Five different mixture
samples, with different relative fractions of
α
-AlaH
+
and
β
-AlaH
+
, were prepared with their
respective infrared action spectrum recorded. The method of least squares was used to fit
observed mixture spectra to a linear combination of pure experimental spectra shown below:
y
model
(
ν
) =
f
α
-
AlaH
+
(H
2
O)
(
ν
)
·
x
+
f
β
-
AlaH
+
(H
2
O)
(
ν
)
·
(1
−
x
)
(5)
Where
ν
is the frequency in wavenumbers (cm
−
1
) and
f
α
-
AlaH
+
(H
2
O)
(
ν
)
and
f
β
-
AlaH
+
(H
2
O)
(
ν
)
are the pure experimental spectra. The relative fractions of the two compounds in the
mixture are denoted as
x
and
1
−
x
, respectively.
To determine the relative fractions
x
and
1
−
x
, the method of least squares was used to
minimize the sum of the squares of the differences between the observed mixture spectrum,
y
obs
(
ν
)
, and the linear combination of the individual pure experimental spectra,
y
model
(
ν
)
.
Minimization is achieved by differentiating with respect to
x
, and setting the derivative equal
S-6
to zero:
d
dx
N
X
i
=0
[
y
obs
(
ν
i
)
−
y
model
(
ν
i
)]
2
= 0
(6)
Insertion of Equation 5 yields:
d
dx
N
X
i
=0
y
obs
(
ν
i
)
−
f
α
-
AlaH
+
(H
2
O)
(
ν
i
)
·
x
+
f
β
-
AlaH
+
(H
2
O)
(
ν
i
)
·
(1
−
x
)
2
= 0
(7)
Where
N
= 120
and
ν
0
= 1000 cm
−
1
, corresponding to a spectral range of
1000 cm
−
1
to
1600 cm
−
1
with 5
cm
−
1
steps.
Solving this equation gives the values of
x
and
1
−
x
that minimize the sum of squared
differences, providing the relative fractions of
α
-AlaH
+
and
β
-AlaH
+
, respectively, in the
initial liquid sample.
Computational Methods
Protonation sampling and a conformational search with non-covalent interactions was per-
formed using the Conformer-Rotamer Ensemble Sampling Tool (CREST).
S6,S7
A conforma-
tional search was performed for the bare protonated amino acids and their single water-tagged
complex counterparts. In the case of O-phospho-L-tyrosine, an optimized protonated struc-
ture, tagged with a single water molecule, was obtained from the literature.
S8
The protonated, lowest energy conformer of each water-tagged amino acid obtained from
CREST (or the literature) was selected for geometry optimization using density functional
theory (DFT) in ORCA 5.0.4.
S9
Each protonated amino acid-water complex was optimized
using the PBEh-3c composite method,
S10
followed by optimization at the
ω
B97X-V/def2-
QZVPP (auxiliary basis set def2/j)
S11–S15
level of theory. Electronic energies were refined
using coupled cluster theory (DLPNO-CCSD(T1)-F12/cc-pVTZ-F12 with auxiliary basis
sets cc-pVQZ/jk, cc-pVTZ-F12-MP2Fit, and cc-pVTZ-F12-CABS),
S16–S20
using optimized
geometry from DFT (
ω
B97X-V/def2-QZVPP).
S-7
Harmonic oscillator vibrational frequencies and intensities were calculated from DFT,
using a numerical Hessian calculation and the geometry output from the
ω
B97X-V/def2-
QZVPP optimization. Predicted IR spectra (for both the water-tagged, protonated amino
acids and bare protonated amino acids) were obtained from these DFT calculations and
were compared to experimental spectra (see Figure S3). Vibrational frequencies were scaled
by 0.965 and each frequency was plotted along with a 20
cm
−
1
full width half maximum
Gaussian to simulate experimental line widths.
The binding energy for each amino acid-water complex was calculated using DFT and
coupled cluster theory. The binding energy is formally defined as the negative enthalpy of
binding at absolute zero. It is the difference in absolute enthalpy of two quantities: (1) the
complex and (2) the sum of the absolute enthalpies of the water and amino acid individually.
The absolute enthalpy of a molecular species is defined as the sum of its single point (elec-
tronic) energy and its zero-point energy. Binding energies with thermal corrections were
also calculated. For this study, each DFT-derived binding energy was obtained from the
single point (electronic) energy calculated at the
ω
B97X-V/def2-QZVPP level of theory and
zero-point energy from numerical frequency calculations at the same level of theory (includ-
ing auxiliary basis sets). Each coupled-cluster derived binding energy was obtained from
the single point (electronic) energy calculated at the DLPNO-CCSD(T1)-F12/cc-pVTZ-F12
level of theory (including auxiliary basis sets), with geometry optimized from
ω
B97X-V/def2-
QZVPP level of theory (including auxiliary basis sets), and zero-point energy from numerical
frequency calculations at the
ω
B97X-V/def2-QZVPP level of theory (including auxiliary ba-
sis sets). Table S1 lists the binding energies of the water-tagged amino acids from this study.
Figure S4 shows the water tagging fraction of each amino acid versus their water binding
energy in kcal/mol and wavenumbers (
cm
−
1
).
S-8
Figure S1: The experimental setup showing the QCL and water vapor-seeded helium buffer
gas coupled to the commercial mass spectrometer. General connections between the QCL,
mass spectrometer, and instrument control computers are shown and additional components
are labeled.
S-9