The dissociation chemistry of low-energy N-formylethanolamine ions: Hydrogen-bridged radical cations as key intermediates

Article abstract of DOI:10.1016/j.ijms.2011.05.011

Tandem mass spectrometry experiments show that N-formylethanolamine molecular ions HOCH₂CH₂NHC(H)O⁺ (FE1) lose C₂H₃O, CH₂O and H₂O to yield m/z 46 ions HC(OH)NH₂⁺, m/z 59 ions CH₂N(H)CHOH ⁺, and m/z 71 N-vinylformamide ions CH₂C(H)N(H)CHO ⁺. A detailed mechanistic study using the CBS-QB3 model chemistry reveals that the readily generated 1,5-H shift isomer HOCHCH₂N(H)C(H) OH⁺ (FE2) and hydrogen-bridged radical cations (HBRCs) act as key intermediates in a 'McLafferty + 1′ type rearrangement that yields the m/z 46 ions. The co-generated C₂H₃O neutrals are predicted to be vinyloxy radicals CH₂CHO in admixture with CH₃CO generated by quid-pro-quo (QPQ) catalysis. A competing C-C bond cleavage in FE1 leads to HBRC [CH₂N(H)C(H)O-...H...OCH₂] ⁺, which serves as the direct precursor for CH₂O loss. In addition, ion FE2 also communicates with a myriad of ion-molecule complexes of vinyl alcohol and formimidic acid whose components may recombine to form distonic ion FE3, HOCH(CH₂)N(H)C(H)OH⁺, which loses H ₂O after undergoing a 1,5-H shift. Further support for these proposals comes from experiments with D- and ¹⁸O-labelled isotopologues. Previously reported proposals for the H₂O and CO losses from protonated N-formylethanolamine are briefly re-examined.

Full text of DOI:10.1016/j.ijms.2011.05.011

International Journal of Mass Spectrometry 306 (2011) 9–26
Contents lists available at ScienceDirect
International Journal of Mass Spectrometry
journal homepage: www.elsevier.com/locate/ijms
The dissociation chemistry of low-energy N-formylethanolamine ions:
Hydrogen-bridged radical cations as key intermediates
Karl J. Jobsta, Richard D. Bowenb, Johan K. Terlouwa,∗
a Department of Chemistry and Chemical Biology, McMaster University, 1280 Main Street West, Hamilton, ON, L8S 4M1, Canada
b School of Life Sciences, University of Bradford, Richmond Road, Bradford, West Yorkshire, BD7 1DP, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 19 April 2011
Received in revised form 17 May 2011
Accepted 17 May 2011
Available online 26 May 2011
Tandem mass spectrometry experiments show that N-formylethanolamine molecular ions
HOCH2CH2NHC(H) O•+ (FE1) lose C2H3O•, CH2O and H2O to yield m/z 46 ions HC(OH)NH2+, m/z
59 ions •CH2N(H)CHOH+, and m/z 71 N-vinylformamide ions CH2 C(H)N(H)CHO•+
.
A detailed mechanistic study using the CBS-QB3 model chemistry reveals that the readily generated
1,5-H shift isomer HOCHCH2N(H)C(H)OH•+ (FE2) and hydrogen-bridged radical cations (HBRCs) act as
key intermediates in a ‘McLafferty + 1ꢀ type rearrangement that yields the m/z 46 ions. The co-generated
C2H3O• neutrals are predicted to be vinyloxy radicals CH2 CHO• in admixture with CH3C O• generated
by quid-pro-quo (QPQ) catalysis.
In memory of Professor Dudley H. Williams,
whose pioneering studies on the role of
ion–neutral complexes in gas-phase ion
chemistry have set the stage for this
contribution.
A competing C−C bond cleavage in FE1 leads to HBRC [CH2N(H)C(H) O-· · ·H· · ·O CH2]•+, which serves
as the direct precursor for CH2O loss.
In addition, ion FE2 also communicates with a myriad of ion–molecule complexes of vinyl alcohol and
formimidic acid whose components may recombine to form distonic ion FE3, HOCH(CH2)N(H)C(H)OH•+
,
Keywords:
Tandem mass spectrometry
CBS-QB3 model chemistry
C3H5NO ion structures
Conformers
which loses H2O after undergoing a 1,5-H shift. Further support for these proposals comes from experi-
ments with D- and 18O-labelled isotopologues.
Previously reported proposals for the H2O and CO losses from protonated N-formylethanolamine are
briefly re-examined.
Quid-pro-quo catalysis
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
debate over the mechanism. Nevertheless, these studies provided
crucial mechanistic information, viz., the structures and energies of
The well-known McLafferty rearrangement is arguably the most
important hydrogen-shift rearrangement in the structure analysis
of organic compounds by electron ionization (EI) mass spectrom-
etry [1]. The reaction was discovered in the early 1950s [1c], but
its detailed mechanism has been a subject of animated debate for
almost 40 years. The issue of whether the ␥-H transfer and ␣–␤
bond cleavage reactions occur in a concerted or a step-wise manner
was not finally settled until the isotopic labelling study of Derrick
and Bowie [2] left little doubt that the 1,5-H shift yields a distonic
ion [3] as an intermediate.
Early investigators recognized that integrating theory and
experiment would be an ideal approach to study reaction mech-
anisms, but this only became feasible in the late 1980s following
spectacular advances in computer technology and software. The
pioneering computational studies [4] of the prototypal McLafferty
rearrangement in ionized n-butanal (using simple Hartree–Fock
calculations) therefore appeared at a relatively late stage in the
the key intermediates and transition-states, which cannot readily
be ascertained by experiment. More recently, the Schwarz group
has published a trilogy of papers [5] dealing with the dissociation
of ionized valeramide. By bringing to bear sophisticated DFT (den-
sity functional theory) based calculations, the authors provided
an elegant (mechanistic) rationale of a stepwise McLafferty rear-
rangement and three other competing dissociations, in excellent
agreement with the experimental observations.
The presently available CBS [6a], Gaussian [6b] and Weizmann
[6c] model chemistries often make it possible to study mechanisms
with chemical accuracy ( 1 kcal mol−1 [6d]).
In our research we have exploited these powerful tools to
study dissociation mechanisms of low-energy radical cations for
which experiment alone can at best provide tentative proposals [7].
This approach has led to the growing realization that rearrange-
ments of solitary radical cations containing heteroatoms often
involve hydrogen-bridged radical cations (HBRCs) [8], a subclass
of ion–molecule complexes [1a], as key intermediates. HBRCs are
often more stable than the incipient molecular ions and at elevated
energies they may catalyze further H-transfers leading to energet-
ically more favourable product ions or neutrals. Proton-transport
∗
Corresponding author. Tel.: +1 905 525 9140; fax: +1 905 522 2509.
E-mail address: terlouwj@mcmaster.ca (J.K. Terlouw).
1387-3806/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijms.2011.05.011

10
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
catalysis (PTC) [9a], where, in a complex, a neutral molecule induces
an ion to isomerize via a HBRC, is a prime example. An interesting
variant is the catalyzed transformation of a neutral species by an
ion [9b].
In the present study, we use the CBS-QB3 model chemistry
[6d] to probe the dissociation of HOCH2CH2NHC(H) O•+ (FE1),
ionized N-formylethanolamine. This challenging ion, whose three
heteroatoms make it prone to extensive hydrogen-shift rear-
rangements via HBRCs, has not been previously studied by either
experiment or theory. It undergoes three competing fragmen-
tations that feature in both the conventional EI and the MI
(metastable ion) mass spectra: loss of C2H3O•, CH2O and H2O.
The C2H3O• loss from FE1 points to a ‘McLafferty + 1ꢀ rearrange-
ment in which a second H-transfer occurs following the ␣–␤ bond
+
cleavage to produce oxygen-protonated formamide, HC(OH)NH2
.
There is compelling evidence for the involvement of ion–molecule
complexes in this type of rearrangement [10] and in the present
system these are expected to adopt the configuration of HBRCs.
This opens the possibility that PTC also plays a role in this rear-
rangement. Indeed, theory predicts that the neutrals of this reaction
are vinyloxy radicals CH2 CHO• in admixture with acetyl radicals
CH3C O• generated by catalysis.
The competing loss of CH2O from FE1 could involve a 1,6-H shift
to yield the distonic ion •CH2N(H)CHOH+ but, here too, a very stable
HBRC appears to act as a key intermediate.
Fig. 1. EI mass spectrum of N-formylethanolamine HOCH2CH2NHCHO (M•+ = m/z
89).
was synthesized using ethanolamine-d4 (CDN Isotopes) in the
above procedure.
The
HOCH2CH2NHCDO, HOCH2CH2NHCH18O and HOCH2CH2NHCD18
were prepared from the labelled methyl formates CH3OCDO
(Aldrich) CH3OCH18 and CH3OCD18O.
5:2:1 molar mix-
D-,
18O-,
and
mixed
D/18O-isotopologues
O
O
A
ture of MeOH, H218O, and (D-labelled) formic acid was kept at
room temperature for ∼5 days and thereafter distilled to obtain
CH3OCH18O and CH3OCD18O. N-formyl-O-methylethanolamine
and its isotopologue CH3OCH2CH2NHCH18O were synthesized
from O-methylethanolamine (Aldrich > 99%) in a procedure analo-
gous to that used for the preparation of NFE. All compounds were
used without further purification.
The calculations were performed with the CBS-QB3 model
chemistry [6a]. Most of the calculations were run with the Gaussian
09, Rev B.01 suite of programs [15] on the SHARCNET computer
network at McMaster University. In the CBS-QB3 model chem-
istry the geometries of minima and connecting transition states
are obtained from B3LYP density functional theory in combination
with the 6-311G(2d,d,p) basis set (also denoted as the CBSB7 basis
set). Computational data pertaining to the dissociation products
are found in Table 1. The resulting total energies and enthalpies
of formation for minima and connecting transition states (TS)
in the N-formylethanolamine systems of ions are presented in
Tables 2a and 2b. Selected optimized geometries are shown in Fig. 5.
Spin contaminations (ꢁS2ꢂ values of Tables 2a and 2b) were accept-
able. Unless stated otherwise, all enthalpies presented in the tex◦t
and in the Schemes (numbers in square brackets) refer to ꢀfH298
values in kcal mol−1 derived from the CBS-QB3 calculations. The
complete set of computational results is available from the authors
upon request.
Perhaps not surprisingly [7], the loss of H2O appears to be a fairly
complex multi-step reaction. It yields ionized N-vinylformamide,
CH2 C(H)N(H)CHO•+, in admixture with its more stable cyclic dis-
tonic isomer. The time-honoured approach of isotopic labelling [11]
was also used to study this reaction. Analysis of various D- and
18O-labelled isotopologues supports our mechanistic proposal but
it also challenges theory to evaluate a bewildering array of potential
exchange reactions in the key intermediates.
2. Experimental and theoretical methods
The experiments were performed with the VG Analytical ZAB-R
mass spectrometer of BEE geometry (B, magnet; E, electric sector)
[12a] using an electron ionization (EI) source at an accelerating volt-
age of 8 kV. Metastable ion (MI) and collision induced dissociation
(CID) mass spectra were recorded in the second field free region
(2ffr). In all collision experiments O2 was used as the collision gas.
All spectra were recorded using a PC-based data system developed
by Mommers Technologies Inc. (Ottawa). Kinetic energy releases
(corrected T0.5 values) were measured according to standard pro-
cedures [13]. The N-formylethanolamine samples were introduced
into the source (kept at 120 ◦C) via a standard insertion probe held
at ∼60 ◦C.
The HeI photoelectron (PE) spectrum of N-vinylformamide was
obtained using a home-built spectrometer [12b].
3. Results and discussion
The synthesis of N-formylethanolamine (NFE) from
ethanolamine (Aldrich, > 99%) was performed using methyl
formate (Aldrich > 99%), rather than ethyl formate as stipulated in
the procedure of Tip et al. [14]. Methyl formate (100 mg, 1.7 mmol)
was added slowly to a solution of ethanolamine (100 mg, 1.7 mmol)
dissolved in methanol (1.8 mL) while cooling the mixture with
an ice water bath. The solvent was removed under reduced pres-
sure but the prescribed [14] vacuum distillation of the resulting
high-boiling residue was not performed because it appears that
2-oxazoline (4,5-dihydro-1,3-oxazole) is generated as a persistent
impurity by thermal decomposition of the sample. This point
is discussed further in Section 3.2. Repeated exchanges of the
labile hydrogens of NFE with MeOD were performed to prepare
DOCH2CH2NDCHO. The D-labelled compound HOCD2CD2NHCHO
3.1. The EI mass spectrum of N-formylethanolamine and the MI
and CID mass spectra of the (pseudo) molecular ions
The mass spectrum of N-formylethanolamine (NFE), see Fig. 1,
displays a weak molecular ion signal (m/z 89) relative to the peaks
at m/z 71, 59 and 46. These fragment ions correspond to the losses
of H2O, CH2 O and C2H3O• and they are also observed in the MI
mass spectrum of Fig. 2a. Fig. 1 also displays prominent peaks at
m/z 58 and 30, which likely result from the direct bond cleavage
reaction HC( O)NHCH2CH2OH•+ (FE1) → HC( O)NH CH2 (m/z
+
58) + CH2OH• and a subsequent partial decarbonylation of the m/z
+
58 ions in CH2NH2 (m/z 30). These ions are absent in the MI
spectrum but they reappear in the CID spectrum of Fig. 2b. There-

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
11
Table 1
Energetic data derived from CBS-QB3 calculationsa used to probe the structure and reactivity of dissociation products of ionized N-formylethanolamine (Fig. 4 and Scheme 3
or Scheme 4).
Species and mass
CBS-QB3 E (total) [0 K]
QB3 ꢀfH◦
Species and mass
CBS-QB3 E (total) [0 K]
QB3 ꢀfH◦
298
298
Fig. 4 ion 1a
Fig. 4 ion 1b1
Sch. 3 ion 1b2
Fig. 4 ion 1c
Fig. 4 ion 1d
Fig. 4 ion 1e
Fig. 4 ion 1f
Fig. 4 ion 2a
Fig. 4 ion 2b
Fig. 4 ion 3a
Fig. 4 ion 3b
Fig. 4 ion 4a
Fig. 4 ion 4b
Fig. 4 ion 5a
Fig. 4 ion 5b
Fig. 4 ion 5c
71
71
71
71
71
71
71
71
71
71
71
71
71
71
71
71
−246.57683
−246.56491
−246.56198
−246.55627
−246.51574
−246.53003
−246.54533
−246.56802
−246.52221
−246.58663
−246.55412
−246.52992
−246.53251
−246.53566
−246.53980
−246.55398
174.2
182.3
184.2
188.0
212.8
204.4
195.3
179.9
209.1
167.9
189.0
203.6
202.1
200.6
198.7
189.3
Fig. 4 ion 6a
Fig. 4 ion 6b
Fig. 4 ion 6c
Fig. 4 ion 6d
Fig. 4 ion 7a
Fig. 4 ion 7b
Fig. 4 ion 8a
Fig. 4 ion 8b
Fig. 4 ion 9
71
71
71
71
71
71
71
71
71
−246.51284
−246.51551
−246.51742
−246.54710
−246.52469
−246.54593
−246.46343
−246.50370
−246.50699
−246.52039
−246.55441
−246.54987
−246.45400
−246.50773
−246.51782
−246.53291
214.7
213.4
212.3
194.7
207.3
194.5
245.5
221.2
218.0
209.6
188.6
191.0
251.8
218.9
211.4
203.2
TS 1a → m/z 70
TS 1b1 → 1b2
TS 1b1 → 1a
TS 1b1 → m/z 43
TS 1b1 → m/z 43b
TS 1b2 → 1f
TS 1f → m/z 43
•+
Scheme 3 ion 1a-H
70
59
46
45
45
44
−246.02895
−208.54478
−169.96226
−169.65356
−169.63475
−153.22216
153.2
159.2
124.7
−45.7
−34.2
186.2
HC( NH)CH3
43
43
43
30
28
18
−133.35001
−152.94205
−152.93284
−114.34417
−113.18201
−76.33750
230.9
−2.1
•CH2N(H)CHOH+
CH3C O•c
+
HC(OH)NH2
CH2 CHO•c
3.4
HC( O)NH2
HC( NH)OH
CH2 CHOH•+
CH2
CO
O
−27.0
−26.6
−57.0
H2O
a
E (total) in Hartrees, 298 K enthalpies in kcal mol−1
.
b
c
“Ion dipole” TS of Scheme 4.
The tabulated values are in close agreement with the experimental values of −2.3 and 3.1 kcal mol−1 found in Refs. [19b/c].
Table 2a
Energetic dataa from CBS-QB3 calculations of stable isomers involved in the dissociation chemistry of N-formylethanolamine ions FE1.
Species
B3LYP/CBSB7 E (total)
CBS-QB3 E (total) [0 K]
ZPE
QB3 ꢀfH◦
QB3 ꢀfH◦
ꢁS2ꢂ
0
298
NFE1a
NFE1b
NFE1c
NFE1d
NFE1e
(Scheme 5)
(Scheme 5)
(Scheme 5)
(Scheme 5)
(Scheme 5)
−323.82528
−323.82706
−323.82142
−323.82683
−323.82232
−323.24071
−323.24156
−323.23703
−323.24147
−323.23745
67.0
67.3
66.8
67.1
66.7
−83.4
−83.9
−81.1
−83.9
−81.4
−89.4
−90.1
−87.0
−89.9
−87.2
–
–
–
–
–
FE1b
FE1d
FE1e
FE2a
FE2b
FE2c
FE3
FE4
FE5
FE6
FE7
(Scheme 5)
(Scheme 5)
(Scheme 5)
(Scheme 7a/b)
(Scheme 7a)
(Scheme 7a)
(Scheme 7b)
(Scheme 7a)
(Scheme 6)
(Scheme 6)
(Scheme 9)
(Scheme 10)
(Scheme 10)
(Scheme 10)
−323.50004
−323.49489
−323.48986
−323.50756
−323.50953
−323.50920
−323.50794
−323.49180
−323.47327
−323.48281
−323.49109
−323.52560
−323.51812
−323.46636
−322.89040
−322.89483
−322.88961
−322.91103
−322.91367
−322.91312
−322.91411
−322.89800
−322.87645
−322.88688
−322.89120
−322.92649
−322.92581
−322.87536
66.9
66.0
65.8
67.1
66.8
66.7
65.9
66.9
66.7
67.1
65.3
67.0
66.3
65.2
136.4
133.6
136.9
123.5
121.8
122.1
121.5
131.6
145.2
138.6
135.9
113.8
114.2
145.8
130.1
127.7
131.1
117.2
115.9
116.3
115.7
125.6
139.0
132.7
130.0
107.4
108.2
139.8
0.77
0.94
0.96
0.76
0.76
0.76
0.76
0.76
0.76
0.78
0.76
0.80
0.84
0.77
FE8
FE9
FE10
IDC
LB1a
LB1b
(Scheme 7b)
(Scheme 8)
(Scheme 8)
−323.51318
−323.51981
−323.52181
−322.91348
−322.90252
−322.90477
65.0
64.3
64.4
121.9
128.8
127.4
116.6
123.6
122.2
0.77
0.77
0.77
HBRC1
HBRC2
HBRC3a
HBRC3b
HBRC4a
HBRC4b
HBRC4c
HBRC5a
HBRC5b
HBRC6
HBRC7
HBRC8
HBRC9
(Scheme 7a)
(Scheme 7a)
(Scheme 7b)
(Scheme 7b)
(Scheme 7a)
(Scheme 8)
(Scheme 8)
(Scheme 6)
(Scheme 6)
(Scheme 6)
(Scheme 9)
(Scheme 9)
(Scheme 9)
−323.51259
−323.49539
−323.52366
−323.53146
−323.52832
−323.54258
−323.53891
−323.49940
−323.49096
−323.49910
−323.51895
−323.49969
−323.50921
−322.91384
−322.89580
−322.92885
−322.94155
−322.92920
−322.93982
−322.93843
−322.90789
−322.90211
−322.90335
−322.92015
−322.90551
−322.91527
64.0
64.4
63.7
64.8
64.4
65.0
64.9
61.8
62.4
64.1
63.4
63.2
61.8
121.7
133.0
112.3
104.3
112.1
105.4
106.3
125.4
129.1
128.3
117.7
126.9
120.8
116.5
127.6
107.5
98.9
107.0
100.0
101.3
121.2
124.0
123.6
112.9
121.9
116.4
0.82
0.77
0.86
0.84
0.86
0.82
0.76
0.76
0.91
0.78
0.90
0.87
0.78
a
E(total) in Hartrees, all other components, including the ZPE scaled by 0.99, are in kcal mol−1
.

12
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
Table 2b
Energetic dataa from CBS-QB3 calculations of the connecting transition states involved in the dissociation chemistry of N-formylethanolamine ions FE1.
Species
B3LYP/CBSB7 E (total)
CBS-QB3 E (total) [0 K]
ZPE
QB3 ꢀfH◦
QB3 ꢀfH◦
ꢁS2ꢂ
0
0
TS NFE1a → b
TS NFE1b → c
TS NFE1c → d
TS NFE1c → e
(Scheme 5)
(Scheme 5)
(Scheme 5)
(Scheme 5)
−323.82081
−323.82041
−323.82136
−323.79310
−323.23558
−323.23648
−323.23738
−323.21054
67.4
66.5
66.7
66.5
−80.2
−80.8
−81.3
−64.5
−86.8
−87.0
−87.7
−71.0
–
–
–
–
TS FE1b → d
TS FE1b → e
(Scheme 5)
(Scheme 5)
(Scheme 7a)
(Scheme 6)
(Scheme 10)
(Scheme 10)
(Scheme 7a)
(Scheme 6)
(Scheme 7a)
(Scheme 7b)
(Scheme 10)
(Scheme 7a)
(Scheme 7a)
(Scheme 7a)
(Scheme 9)
(Scheme 10)
(Scheme 7b)
(Scheme 7a)
(Scheme 6)
(Scheme 6)
(Scheme 6)
(Scheme 9)
(Scheme 10)
−323.47975
−323.47121
−323.48963
−323.40192
−323.46028
−323.45826
−323.48772
−323.46057
−323.48601
−323.48746
−323.46304
−323.50061
−323.49387
−323.47197
−323.48386
−323.46103
−323.48684
−323.47133
−323.47395
−323.45533
−323.48119
−323.47504
−323.46838
−322.88220
−322.87311
−322.89026
−322.81344
−322.86744
−322.86223
−322.88549
−322.86376
−322.89432
−322.88894
−322.87033
−322.90787
−322.89520
−322.88049
−322.89105
−322.87167
−322.89275
−322.87227
−322.88016
−322.84521
−322.88215
−322.88060
−322.87663
65.5
64.8
65.8
63.0
62.5
63.0
64.3
66.4
65.7
64.8
63.0
65.9
64.2
64.1
63.2
65.4
64.4
64.5
65.1
63.3
65.5
63.9
65.9
141.6
147.3
136.5
184.7
150.8
154.1
139.5
153.1
133.9
137.3
149.0
125.4
133.4
142.6
136.0
148.2
134.9
147.8
142.8
164.8
141.6
142.6
145.0
135.2
141.3
130.3
178.7
144.6
147.5
133.4
147.0
127.9
131.5
142.7
119.4
127.6
136.3
129.4
141.6
128.6
141.8
136.4
159.3
135.7
136.2
138.6
0.76
0.78
0.78
0.79
0.80
0.76
0.77
0.76
0.76
0.77
0.76
0.76
0.77
0.76
0.79
0.79
0.76
0.93
0.77
0.76
0.76
0.78
0.76
TS FE1b → 2a
TS FE1b → 6
TS FE1b → 7
TS FE1b → 9
TS FE1b → HBRC1
TS FE1e → 5
TS FE2a → b
TS FE2a → 3
TS FE2a → 10
TS FE2b → c
TS FE2b → HBRC2
TS FE2c → 4
TS FE3 → 7
TS FE3 → 10
TS FE3 → IDC
TS FE4 → HBRC4
TS FE5 → 6
TS FE5 → HBRC5a
TS FE6 → HBRC6
TS FE7 → HBRC9
TS FE8 → 9
TS IDC → HBRC3a
TS IDC → FE8
(Scheme 7b)
(Scheme 10)
(Scheme 8)
(Scheme 8)
−323.50905
−323.51190
−323.50549
−323.51333
−322.91927
−322.91028
−322.89145
−322.89953
63.6
66.0
63.4
63.0
118.3
123.9
135.7
130.7
113.1
117.6
130.3
125.3
0.82
0.77
0.78
0.78
TS LB1a → 1b
TS LB1b → HBRC4c
TS HBRC2 → 4
(Scheme 7a)
(Scheme 7b)
(Scheme 9)
(Scheme 9)
(Scheme 8)
(Scheme 8)
(Scheme 8)
(Scheme 6)
−323.49204
−323.50544
−323.50662
−323.47406
−323.51051
−323.49055
−323.47228
−323.48681
−322.89453
−322.91285
−322.90980
−322.87975
−322.91368
−322.89100
−322.87312
−322.89611
64.1
63.8
63.1
64.0
63.3
60.2
62.3
62.6
133.8
122.3
124.2
143.1
121.8
136.0
147.3
132.8
128.3
116.8
118.2
137.7
116.5
130.3
141.9
127.7
0.77
0.78
0.91
0.82
0.85
0.79
0.77
0.83
TS HBRC3a → b
TS HBRC3a → 7
TS HBRC3a → 8
TS HBRC4a → b
TS HBRC4b → LB1a
TS HBRC4b → c
TS HBRC5a → b
a
E(total) in Hartrees, all other components, including the ZPE scaled by 0.99, are in kcal mol−1
.
fore, an important precondition imposed by experiment on theory
is that enthalpies of stable intermediates, connecting transition
states and dissociation thresholds of proposed dissociation mech-
CID spectrum of Fig. 2d. That these peaks do not originate from
the naturally occurring 13C-isotopologue of FE1 follows from the
fact that the CID spectrum of Fig. 2d does not change with the m/z
90:89 peak ratio. Their presence suggests that PFE may lose H• to
produce an m/z 89 ion having dissociation characteristics akin to
those of FE1.
anisms of the metastable ions should not exceed 157 kcal mol−1
,
the calculated ꢁꢀfH value of the direct bond cleavage products
HC( O)NH CH2+ + CH2OH•.
The conspicuous peak at m/z 90 in Fig. 1 represents [M + H]+
ions resulting from the remarkably efficient self-protonation
of NFE by chemical ionization: raising the source pressure
from c. 3 × 10−7 to 1 × 10−6 Torr, by gently heating the sam-
ple, causes the m/z 90: 89 peak intensity ratio to increase to
∼10:1! In line with this, our calculations show that the various
proton transfer reactions between FE1 and its neutral NFE are
exothermic: the proton affinity (PA) of NFE is 214 kcal mol−1
The next Section (3.2) deals with the structures of the m/z 71,
59 and 46 product ions generated from (metastable) ions FE1. Sec-
tions 3.3–3.5 present a conformational and mechanistic analysis
of the dissociation chemistry of ions FE1 and D- and 18O-labelled
isotopologues. Section 3.6 re-examines the previously reported dis-
sociation mechanisms [14] of protonated N-formylethanolamine.
but those of
O
C•–NHCH2CH2OH, HC( O)NH–C•HCH2OH,
3.2. Identification of the product ions generated in the
dissociation of ions FE1
HC( O)NH–CH2C•HOH,
HC( O)NHCH2CH2O•
and
HC( O)N•CH2CH2OH are only 190, 187, 187, 203 and
204 kcal mol−1, respectively.
A comparison of the CID mass spectra of the m/z 59 and 46 prod-
uct ions, see Fig. 3a and b with reference spectra [16,17] leaves little
doubt that the distonic ion HC(OH)NH CH2•+ (m/z 59) and oxygen-
The MI and CID mass spectra of protonated N-
formylethanolamine (PFE), see Fig. 2c and d, are dominated
by peaks at m/z 62 and 72, corresponding to the losses of CO
and H2O. The associated reaction mechanisms have been briefly
addressed by Tip et al. [14] in their study of N-formylethanolamine
as a matrix for the Fast Atom Bombardment analysis of non-
polar compounds, including saccharides. One point that was not
addressed concerns the notable peaks at m/z 71, 59 and 46 in the
+
protonated formamide HC(OH)NH2 (m/z 46) are generated.
Using the values from Table 1, we find that the
•+
sum of the enthalpies of HC(OH)NH CH2
and CH2 O
(ꢁꢀfH = 132 kcal mol−1
) is well below the upper limit of
157 kcal mol−1 imposed by experiment. Using this energy cri-
terion, we can safely conclude that formaldehyde is lost rather

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
13
46
62
59
71
72
61.3
(a)
(c)
59
62
46
72
58
44
59
71
71
30
28
30
46
(b)
(d)
Fig. 2. (a) MI and (b) CID mass spectra of the N-formylethanolamine radical cation (FE1); (c) MI and (d) CID mass spectra of protonated N-formylethanolamine (PFE); note:
a weak peak (c. 5%) at m/z 89 in Fig. 2d is not shown.
than its elusive hydroxycarbene isomer HCOH [18], which lies
considerably higher in energy.
induced and spontaneous dissociations. When the unimolecular
component (whose intensity strongly depends on the internal
energy of the precursor ions) is removed by applying an external
voltage (−1 kV) to the collision cell [22], the spectra of Fig. 3e and
f are obtained. It is gratifying to observe that the m/z 42: 43 peak
ratios are now virtually the same. In fact, the more intense m/z 41
peak of Fig. 3f, which may consist of a narrow and a broad compo-
nent, and the more prominent charge-stripping peak (++) are the
only distinguishing features.
Second, we propose that the narrow part of the m/z 41 peak
of Fig. 3f results from thermal decomposition because its inten-
sity appeared to vary during the evaporation of the sample. Indeed,
the CID spectrum of the m/z 71 ions obtained from a sample that
had been subjected to prolonged heating (at 100 ◦C for 8 h), see
Fig. 3d, is dominated by a narrow m/z 41 peak. A plausible thermal
decomposition product, see Scheme 1, is 2-oxazoline (4a), whose
CID spectrum [21b] is very close to that of Fig. 3d.
In contrast, the structure of the mass 43 neutral is not
so easily ascertained because the enthalpies of CH3C O• and
CH2 CHO• are close, see Table 1. Moreover, the combined
enthalpies ꢁꢀfH [HC(OH)NH2+ + CH3C O•] = 124 kcal mol−1 and
ꢁꢀfH [HC(OH)NH2+ + CH2 CH–O•] = 129 kcal mol−1 both satisfy
the energy criterion discussed in Section 3.1.
A recent computational study [19a] predicts that the other iso-
mers CH2 C–OH• and CH CH–OH• lie 29 and 33 kcal mol−1 higher
than the acetyl radical global minimum and thus can be excluded
from our analysis. Unfortunately, a CIDI experiment [20] to probe
the structure of the C2H3O• radical could not be realized due to
insufficient signal intensity. The intriguing possibility that both
CH3CO• and CH2 CHO• isomers are co-generated is discussed in
Section 3.4.
Establishing the structure of the m/z 71 product ion is not
straightforward because there is a dearth of information on the
C3H5NO•+ system of ions [13] and very few CID mass spectra have
been reported [21a]. Fig. 4 shows potentialm/z 71 product ion struc-
tures and their associated CBS-QB3 derived 298 K enthalpies. These
will be used as a guide in the analysis of the CID mass spectrum of
Fig. 3c.
In this context, we note that experiments with
HOCH2CH2NHCH18O indicate that the thermal decomposition
involves a specific loss of H218O: the CID spectrum of the resulting
m/z 71 ions (not shown) displays an intense, narrow peak at m/z
41 whereas the CID spectrum of the m/z 73 ions shown in Fig. 3h,
does not.
As pointed out in Section 3.1, the combined enthalpy of
C3H5NO•+ + H2O must not exceed 157 kcal mol−1. Using ꢀfH
(H2O) = −58 kcal mol−1, an upper limit of 215 kcal mol−1 is obtained
for ꢀfH (C3H5NO•+), so that ions 8a, 8b and 9 can be ruled out as
viable candidates.
This leads us to conclude that the m/z 73 ions are generated
by the specific loss of H216O from the 18O-labelled radical cations
FE1. Indeed, a comparison of Fig. 3e and h provides strong evidence
that ions 1b are generated because the m/z 41–43 peak clusters
The most attractive of the remaining structures is the very stable
N-vinylformamide ion 1b. Its CID spectrum appears to be close to
that of Fig. 3c, but it does not display the prominent, narrow m/z 41
peak and its m/z 43 peak is considerably more intense.
We are confident, however, on the basis of the arguments pre-
sented below, that a mixture of ions 1b and its cyclic distonic isomer
1a (see Fig. 4), is generated in the H2O loss.
O
H
H
O
H
H
C
N
C
N
O
Δ
CH2
CH2
+ H2O
CH2
CH2
2-oxazoline
NFE1
First, the discrepancy in the relative intensity of the m/z 43 peaks
appears to be due to the fact that they represent both collision-
Scheme 1.

14
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
++
42
29
OH
O
C
CH2
C
+
+
28
C
H
NH2
H
N
H
45
H
43
41
17
18
44
43
29
28
27
30
16
++
15
(a)
(e)
++
28
42
30
OH
H
C
+
C
41
H
N
H
H
43
58
28
29
++
41
44
15
15
(b)
(f)
42
43
42
41
41
43
28
29
28
29
++
15
++
15
(c)
(g)
42
41
41
43
31
43
++
28
29
28
15
++
15
(d)
(h)
Fig. 3. Left column: (a), (b) and (c) are CID mass spectra of the source generated m/z 46, m/z 59 and m/z 71 ions from N-formylethanolamine; (d) CID spectrum of the m/z 71
ions generated by thermal decomposition.
Right column: CID spectra with the collision chamber at −1 kV of m/z 71 ions generated from (e) N-vinyl-formamide; (f) N-formylethanolamine and (g) N-formyl-O-
methylethanolamine CH3OCH2CH2NHCHO; (h) CID spectrum of the m/z 73 ions generated from 18O-labelled N-formylethanolamine HOCH2CH2NHCH18O.
are almost identical and the shift of m/z 29–31 confirms the HCO
structure motif.
Complementary evidence that ions 1b are generated comes
from an analysis of N-formyl-O-methylethanolamine (NOE). As
shown in Scheme 2, ion NOE readily loses the methoxy oxygen
atom (in the form of CH3OH) to generate ions of m/z 71.
The CID spectrum of the m/z 71 ions, see Fig. 3g, is virtually
the same as those of Fig. 3e and f, indicating that ions FE1 and
NOE dissociate into 1b via analogous mechanisms. Note that the
analogous thermal decomposition of Scheme 1 does not occur in
NOE as witnessed by the spectrum of Fig. 3g, which shows no trace
of a narrow component in the m/z 41 peak.

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
15
N-vinyl formamide
H
H
C
HO
H
H
H
C
C
H
H
H
C
N
C
C
H
H
H
H
N
C
O
O
O
N
N
N
C
r.c.
O
C
CH2
C
CH2
CH2
H
H
CH2
CH2
1a
[174]
1b
1c
1d
1e
[204]
[182]
[188]
[213]
H
C
H
C
H
H
O
CH2
C
H
H
C
H
H
H
H
N
C
N
O
O
C
N
O
N
N
O
C
C
H2C
C
C
H2C
C
H
CH3
H
H
H
H
H
1f
[195]
2a
[180]
2b
[209]
3a
[168]
3b
[189]
2-oxazoline
H
2-oxazolidinylidene
2-azetidinone
*
*
*
O
C
H
H
O
O
H
N
C
C
H
N
C
N
C
O
N
O
N
H2C
H2C
CH2
H2C
CH2
CH2
H2C
CH2
H2C
CH2
4a
[204]
4b
[202]
5a
[201]
5b
[199]
5c
[189]
NH
NH
NH
O
C
O
C
+
O
H2C
C
CH2=O--H-N=C=CH2
(HBRC)
CH2
H2C
H2C
CH2
CH2
6a
[215]
6b
[213]
6c
[212]
6d
[195]
NH
H
H
NH
CH2
O
N
C
CH2
N
CH2
C
CH2
O
C
H2C
CH2
C
CH2
C
CH2
O
CH2
O
CH2
H
N
O
7a
[207]
7b
[195]
8a
[230]
8b
[221]
9
[219]
Fig. 4. C3H5NO+ (m/z 71) ion structures and their 298 K enthalpies of formation in kcal mol−1 derived from CBS-QB3 calculations; * For these ions, the CID mass spectra have
been reported [21a].
A subtle difference between the CID spectra of ion 1b (Fig. 3e)
and the FE1 derived ions (Fig. 3f) is the enhanced charge-stripping
peak at m/z 35.5 in Fig. 3f.
This prompted us to consider the co-generation of the cyclic
counterpart of ion 1b, the remarkably stable distonic ion 1a of Fig. 4.
Distonic ions are often more prone to charge-stripping, especially
with O2 as collision gas, and in line with this, our calculated vertical
IE of 1a (13.6 eV) is considerably lower than that of 1b (16.3 eV).
Nevertheless, the proposal that the spectrum of Fig. 3f is com-
patible with a mixture of ions 1a and 1b hinges on the question
whether the two isomers have closely similar CID characteristics.
The CID spectrum of pure ions 1a is not accessible, but the following
computational analysis indicates that this is indeed the case.
O
O
CH3
H
H
C
N
C
N
O
- CH3OH
CH2
CH2
H
H
CH2
CH
NOE
1b
Scheme 2.

16
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
Scheme 3 summarizes our computational results on metastable
+CH2−CH N−CH O or CH2 C+−NH−CH O have ꢁꢀfH values
of 236, 234 and 245 kcal mol−1 and are thus far more energy
demanding!
The competing formation of the m/z 43 ion involves loss
of CO: the identity of the neutral was established by a CID
experiment on the isotopologue HC( O18)NHCH CH2. A pre-
vious theoretical study [23] indicates that ionized formamide
ions 1b, which generate both m/z 43 and m/z 70 ions. We propose
that the H• loss does not occur from ion 1b directly, but rather from
its cyclic distonic isomer 1a: the transformation 1b → 1a involves
a facile ring-closure via a TS (at 191 kcal mol−1) that lies well
below the dissociation threshold at 205 kcal mol−1. In contrast,
+
direct H• loss reactions from ion 1b to form CH2 CH−NH− C O,
Fig. 5. Optimized geometries of selected minima and transition states involved in the dissociation chemistry of ionized N-formylethanolamine. Numbers in brackets are
CBS-QB3 derived 298 K enthalpies in kcal mol−1
.

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
17
Fig. 5. (Continued )
undergoes decarbonylation via an ion–dipole complex accessi-
ble from the first excited state, rather than a classical 1,2-H
shift. Although the energy difference between the first two
bands of the PE spectrum of N-vinylformamide is larger than
that of formamide (1.0 vs. 0.2 eV), it seemed worth consider-
ing whether ions 1b also decarbonylate via an excited state.
Scheme 4 shows the optimized geometries of the classical 1,2-H
shift and ion–dipole type transition states. Both are quite energy
demanding with enthalpies of 252 and 220 kcal mol−1 respec-
tively.

18
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
Fig. 5. (Continued )
A more attractive decarbonylation route involves the 1,4-H shift
connecting ions 1b2 and 1f of Scheme 3. Loss of CO from 1f yields
the m/z 43 acetimine ion CH3C( NH)H•+
type isomerism of the amide functionality [24], whose bar-
rier may greatly exceed that of
shift.
a McLafferty-type 1,5-H
.
An important result from the above computational analysis is
that the TS connecting ions 1a and 1b lies ∼15–20 kcal mol−1 below
the dissociation thresholds for the losses of H• and CO. This implies
that a significant fraction of the stable ions 1a/b may interconvert
and that, apart from the charge-stripping peak, their CID mass spec-
tra may not be significantly different. Thus, the CID spectrum of
Fig. 3f may well represent a mixture of 1a and 1b.
The calculations of Scheme 5 (bottom) indicate that the O–H–O
and N–H–O bridged NFE conformers 1b and 1d are a few kcal mol−1
more stable than the non-bridging neutrals 1a, 1c and 1e. The bar-
riers connecting 1a, 1b, 1c and 1d are marginal, but rotation of the
C–N amide bond NFE1c → 1e, and the associated disturbance of
the ␲-system [24], requires 16 kcal mol−1! This would imply that
1b and 1d are the principal species among the freely equilibrating
Z-conformers and that the less stable E-conformer NFE1e may also
be present in the sample.
The corresponding ionic conformers are shown in the top part
of Scheme 5. Ion FE1a is not a minimum but instead it rearranges to
the distonic ion FE2 via a 1,5-H shift. This reaction will be discussed
in the next section. Ion FE1c is not a minimum either: it rotamerizes
to FE1b.
3.3. Conformational analysis of neutral and ionized
N-formylethanolamine
Before addressing the proposed mechanisms, we will first
consider the various conformers of neutral and ionized N-
formylethanolamine. This is relevant in view of the E/Z-

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
'Classical' 1,2-H [252]
19
CH2=CH-NH2 + + CO
= [176]
Σ
'Ion-dipole' 1,2-H [220]
m/z 43
H
C
H
C
H
C
H
H
H
H
O
N
C
O
O
r.c.
N
N
dbc
+ H
[191]
C
CH2
C
C
CH2
H
H
H
m/z 70
1b1 [182]
1a [174]
= [205]
Σ
rot
[189]
O
O
C
H
C
N
H
H
H
N
C
N
C
1,4-H
[211]
dbc
+ CO
CH3
C
H
H
H
CH2
CH3
m/z 43
1b2 [184]
1f [195]
= [204]
Σ
Scheme 3. Mechanistic proposals for the loss of CO and H• from N-vinylformamide ions 1b. The numbers are CBS-QB3 derived 298 K enthalpies in kcal mol−1
.
Prima facie, it is surprising that the O–H–O bridge is broken
upon ionization of NFE1b. However, a charge-distribution analy-
sis indicates that the amide moiety is ionized, which would repel
the bridging proton.
Ions FE1b and FE1d are separated by a negligible barrier but,
as with the neutrals, the rotation of the amide bond FE1b → 1e is
energy demanding. This raises the possibility that the dissociation
chemistry of FE1e differs from that of conformers FE1b/d.
Scheme 6 explores the a priori plausible H2O loss from con-
former FE1e. A 1,5-H shift in FE1e would produce distonic ion FE5,
+
which may further rearrange into HBRC5a by cleaving the C–OH2
bond. The subsequent loss of H2O from HBRC5a, or its ring-closed
isomer HBRC5b, would produce ions 5a and 5b respectively. How-
ever, the C–OH2 bond cleavage is surprising high, which makes
+
Scheme 4. Optimized geometries of the transition states for the loss of CO from the
Aꢀ and Aꢀꢀ states of the N-vinylformamide ion 1b.
this an unrealistic proposal.
Scheme 5. Conformers of neutral (NFE) and ionized (FE) N-formylethanolamine. Numbers in brackets are CBS-QB3 derived 298 K enthalpies of formation in kcal mol−1. IEv
is the calculated vertical ionization energy.

20
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
O
O
O
C
H
C
C
H
H
O
H
H
N
O
N
H
N
H
CH2 CH2
H2C
CH2
CH2 CH2
START
FE-1e
[131]
1,5-H
[147]
[141]
5b + H2O
HBRC5a
[121]
FE5
[159]
[138]
[136]
H
r.c.
[128]
1,4-H
O
H
rotation
[143]
O
H
H
O
O
C
N
C
N
C
H
H
H2C
CH2
H2C
CH2
N
O
H
[124]
5a + H2O [143]
HBRC5b
H2C
CH2
1,2-H
via HBRC 6 (Figure 5)
[136]
FE-1b
[130]
FE6
[133]
[132]
NH2CH2CH2OH (m/z 61) + CO
[179]
Scheme 6. Plausible mechanisms for the H2O loss from the ionized N-formylethanolamine conformer FE1e, in conjunction with the non-observed decarbonylation. The
numbers refer to 298 K enthalpies (in kcal mol−1) derived from CBS-QB3 calculations.
requirement (12 kcal mol−1) is so much higher than that of the
CH2O loss (3 kcal mol−1) that competition becomes unlikely.
Inspired by the proposal that ion–molecule complexes can play
an important role in the McLafferty + 1 mechanism [10], we have
explored the possibility that HBRCs are generated from FE2a.
Indeed, the calculations of Scheme 7a show that conformer FE2b
may cleave at the C–N bond to yield the remarkably stable ion
HBRC2. Transfer of a proton from the vinyl alcohol moiety of
HBRC2 to the amide nitrogen leads to the even more stable N–H–O
Experiment supports this conclusion because the CID spectrum
of ion 5a, the 2-azetidinone ion, is dominated by m/z 28 [21a] and
the same is undoubtedly true for 5b.
Scheme
6 also predicts that the decarbonylation route
FE5 → FE6 → NH2CH2CH2OH•+ (m/z 61) + CO lies well below the
critical energy for the 1,5-H shift FE1e → FE5. However, the MI
spectrum of Fig. 2a shows that decarbonylation does not occur. The
spectrum does contain a minor peak at m/z 61 but its non-integral
position on the energy scale (m/z 61.3) betrays its identity as an
interference peak [25] originating from the protonated precursor
molecule.
+
bridged species HBRC4a, which may dissociate into HC(OH)NH2
(m/z 46) + CH2 CHO•.
That the decarbonylation does not occur is because the prereq-
uisite 1,5-H shift requires some 4 kcal mol−1 more energy than the
rotation of FE1e to FE1b/d. In the next Section it is shown that
conformer FE1b initiates the losses of H2O, CH2O and C2H3O•.
We note that the computed transition-state energies involved
in this HBRC mediated reaction lie below the enthalpy of FE1 so
that the reaction is without a net barrier!
The above proposal implies that the vinyloxy radical CH2 CHO•
is lost in the dissociation, but an intriguing question is whether
the more stable isomer CH3CO• is also generated, as is the case
with the 1-methoxy-2-propanol ion [26]. As mentioned above, it
was not possible to probe this by experiment and so we have
turned to the calculations of Scheme 8, which uses HBRC4a
of Scheme 7 as the starting point, to shed light on this ques-
tion.
3.4. The proposed mechanisms for the losses of CH2O, C2H3O•
and H2O from ionized N-formylethanolamine
The energy diagrams of Scheme 7 summarize our proposed
mechanisms starting from conformer FE1b.
The calculations predict that vertical ionization of NFE1b yields
Scheme 8 indicates that the 1,2-H shift in the CH2 CHO•
component of HBRC4b that leads to HBRC4c is not feasible
because its TS lies 13 kcal mol−1 above the dissociation threshold
HC(OH)NH2+ + CH2 CHO•. However, the 1,2-H shift barrier may be
lowered significantly if protonated formamide acts as a catalyst.
In this scenario, the protonated formamide component of HBRC4b
transfers a hydrogen to the methylene carbon of CH2 CHO•, to
produce ion LB1a, a two center, 3-electron bonded complex [27]
comprising formamide and acetaldehyde. The subsequent trans-
formation LB1a → LB1b may be viewed as a simple rotation of the
CH3C(H) O component. This orients the aldehydic hydrogen so
vibrationally excited ions FE1b at an energy level (144 kcal mol−1
)
that lies well above the low TS (133 kcal mol−1) for the dissocia-
tion route FE1b → HBRC1 → CH2N(H)CHOH•+ (m/z 59) + CH2O, of
Scheme 7a. This scenario is consistent with the fact that the molec-
ular ion in the EI mass spectrum of Fig. 1 is very weak relative to
the m/z 59 base peak. Likewise, the deep potential well occupied
by HBRC1 would explain the high relative intensity of the m/z 59
peak in the MI spectrum, c. 10% of the precursor ion beam. Since the
m/z 46 and 71 peaks of this spectrum have comparable intensities,
the key transition states of their proposed generation should not
significantly exceed the TS for the CH2O loss at 133 kcal mol−1. It
will be seen that this is indeed the case.
that it may be transferred back to the formamide moiety to gen-
erate HBRC4c, which decomposes into CH3C O• and HC(OH)NH2
,
+
The loss of a C2H3O• radical from ion FE1 formally results from
the consecutive transfer of two hydrogens in a process coined the
“McLafferty + 1ꢀꢀ rearrangement [1]. In the first step of the mecha-
nism, FE1b rotamerizes to FE1a, which spontaneously undergoes a
␥-H shift to produce the distonic ion FE2a. A reasonable proposal for
the second H-transfer involves the classical 1,4-H shift FE2c → FE4,
shown in the inset of Scheme 7a. However, its minimum energy
the ‘regenerated’ catalyst.
This route reduces the barrier for the transformation
CH2 CHO•
→
CH3C O• from 42 to 30 kcal mol−1 so that
it now lies within 1 kcal mol−1 of the dissociation threshold
HC(OH)NH2+ + CH2 CHO•. As a result, a fraction of the C2H3O•
radicals lost may be CH3C O•.

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
21
(a) Loss of CH2O and C2H3O•
H
H
H
C
H
O
H
O
H
O
H
C
N
C
N
O
H
HBRC4a
[126]
H
H
CH2
CH2
C
FE2c
FE4
[126]
1,4-H [136]
[142]
[116]
CH2=CHOH•+ + HC(=NH)OH
vertical ionization
m/z 44
149
144
H
142
•
OH
H
C
O
CH2
CH2=CHOH•+ + HC(=O)NH2
H
N
C
H
N
m/z 44
H
H
137
CH2
C
+ CH2=O
+
OH
HC(OH)NH2
133
see part (b)
132
128
+
m/z 71 + H2O
CH2=C(H)O•
132
m/z 59
FE1a
130
128
m/z 46
FE1b
129
HBRC2
H
H
C
H
C
O
O
H1.0
O
1.6
H
H
H
N
O
C
CH2
CH2 CH2
N
117
117
116
H
START
HBRC1
FE2a
FE2b
H
O
H
H
O
H
OH
N
C
107
H
C
H
C
H
C
O
HBRC4a
H
H
N
H
H
H
H
N
H
O
C
CH2
C
O
C
N 1.0 H 1.6
O
C
H
CH2
C
OH
H
H
CH2
H
H
(b) Loss of H2O
H
OH
C
H
N
H
H
H
C
CH2
C
C
H
H
H
OH
see part (a)
FE1b
N
C
O
N
O
CH2
1,5-H
129
H
C
132
C
130
H
CH2
H
N
C
O
1b + H2O
H
m/z 71
O
124
H
CH2
2.2 Å
H
ring-close
117
117
117
1a + H2O
116
m/z 71
FE2a
IDC
116
FE3
H
C
H
OH
108
HBRC3a
H
C
H
O
H
N
C
H
N
H
H
O
H
CH2
START
C
H
OH
CH2
O
H
H
C
C
O
H
H
H
O
O
1.6 Å
99
N
C
H
N
H
CH2
HBRC3b
C
CH2
H
Scheme 7. Mechanistic proposals for the losses of CH2O and CH2 CHO• (top), and H2O (bottom) from N-formylethanolamine ions FE1. The numbers refer to 298 K enthalpies
derived from CBS-QB3 calculations.

22
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
CH3
H
START
CH2
C
C
(Scheme 7)
HBRC4a
H
O
O
O
O
H
C
C
H
H
N
H
rot
[117]
N
H
H
H
H-transfer
[130]
HBRC4b
[100]
LB1a
[124]
- CH2=CHO•
[129]
1,2-H [142]
rot [130]
OH
C
H
H
NH2
CH3
C
CH3
C
[124]
H
O
O
O
O
C
C
H
H
N
H
N
H
H
H
- CH3C=O•
H-transfer
[125]
HBRC4c
LB1b
[122]
[101]
Scheme 8. The proposed quid-pro-quo catalysis mechanism that leads to the loss of CH3C O• from ionized N-formylethanolamine. The numbers refer to 298 K enthalpies
(in kcal mol−1) derived from CBS-QB3 calculations.
Overall, the mechanism is characterized by the catalyst donating
a proton to one site of the substrate, and subsequently abstract-
ing a proton from a different site. This variant of proton-transport
catalysis has been coined as quid-pro-quo catalysis [9b].
lar ions of Table 3. For the sake of clarity, the Table expresses the
normalized MI peak intensities as neutral losses.
First we note that the proposals of Scheme 7 for the losses of
CH2O and C2H3O• are entirely compatible with the labelling data
of Table 3 as are the CID spectra of the labelled product ions (not
shown). It is also obvious that D isotope effects are operating in
these losses but, given the complexity of the system, these were
not further analyzed.
Returning to Scheme 7, it is seen in part (b) that the H2O loss
mechanism is initiated from distonic ion FE2a. Lengthening of its
N–C bond leads to an ion–molecule complex of a vinyl alcohol ion
and formimidic acid, which, at an energy level of 132 kcal mol−1
,
may recombine to form distonic ion FE3. In essence, the reaction
FE2 → FE3 involves transfer of the formimidic acid component of
FE2 to the adjacent carbon in the vinyl alcohol moiety. A 1,5-H shift
in FE3, followed by elongation of the incipient C–OH2+ bond, results
in the formation of ion–dipole complex IDC at 117 kcal mol−1. Ion
IDC may either dissociate into the N-vinylformamide ion 1b + H2O,
In contrast, the data pertaining to the water loss suggest that a
partial exchange of the H and O atoms occurs or that this reaction
involves additional (minor) dissociation pathways.
The aldehydic hydrogen does not participate in the proposal of
Scheme 7b and indeed ions HOCH2CH2NHCDO•+ predominantly
lose H2O. However, ∼25% of the ions lose HDO and mechanisms
that could account for this observation are presented in Scheme 9.
One option is that ion FE3 of Scheme 7b undergoes a competing
1,5-H shift of the same TS-energy: the route FE3 → FE7 → 1f + HDO
of Scheme 9. This proposal is not attractive, considering that the CID
spectrum of the 3-methylaziridinone ion [21b], the cyclic isomer of
distonic ion 1f, is characterized by an intense CH3 loss and that this
loss is absent in the CID spectrum of Fig. 3c.
or adopt the configuration of HBRC3a at 108 kcal mol−1
.
As discussed in Section 3.1, solitary ions 1b may cyclize to the
more stable distonic ions 1a and the same reaction may also occur
in HBRC3a to produce HBRC3b, with the water molecule acting as
a spectator. Since the associated TS lies well below the dissociation
threshold 1b + H2O, a mixture of ions 1a and 1b may be generated.
This is consistent with the analysis of the CID spectrum of the m/z
71 ions in Section 3.2.
Finally, we note that the mechanisms of Scheme 7 have no
reverse barriers, in line with the observed modest T0.5 values of
23, 18 and 36 meV for the formation of the m/z 46, 59 and 71 ions.
An intriguing alternative is the H/D exchange reaction depicted
at the bottom of Scheme 9. Here, the H2O molecule of HBRC3a
abstracts the aldehydic deuteron and then donates a proton back
to the carbonyl oxygen atom, yielding HBRC8 in a QPQ type reac-
tion. Loss of HDO from HBRC8 is likely too energy demanding but
TS HBRC3a → 8 may be sufficiently accessible (138 kcal mol−1) to
account for a partial exchange of the H2O hydrogens and the alde-
hydic deuterium of ions HBRC3a by a back-and-forth QPQ reaction
with HBRC8.
3.5. The dissociation behaviour of D- and 18O-isotopologues of
N-formylethanolamine
The minor loss of D2O from DOCH2CH2NDCHO•+ may also arise
from a QPQ type reaction: Scheme 9 shows that TS HBRC3a → 7 lies
To seek further support for the proposals of Scheme 7, we have
examined the MI mass spectra of the D- and 18O-labelled molecu-

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
23
Table 3
The neutral losses in the MI spectra of various isotopologues of N-formylethanolamine.
Ion
Neutral lost
C2H3O•
C2D3O•
CH2O
CD2O
H2O
HDO
H2O
HDO
D2O
HOCH2CH2NHCHO•+
HOCH2CH2NHCDO•+
DOCH2CH2NDCHO•+
HOCD2CD2NHCHO•+
HOCH2CH2NHCHO•+
HOCH2CH2NHCDO•+
100
100
100
–
100
100
–
–
–
100
–
–
77
92
53
–
80
83
–
–
–
47
–
–
45
38
5
13
40
30
–
13
40
23
–
–
–
–
–
10
6
–
–
–
–
–
5
–
–
10
–
–
–
4
The 18-O and D isotopes are printed in bold face; intensities relative to the most intense loss = 100%.
so low in energy, at 118 kcal mol−1, that the hydronium hydrogens
of HBRC3a are expected to exchange prior to dissociation. Loss of
D2O may also occur from HBRC7 to generate 1c, a tautomer of the N-
vinylformamide ion. The minor H2O loss from DOCH2CH2NDCHO•+
may similarly be explained by the QPQ transformation HBRC3a → 8
mentioned above. Finally, by invoking the above exchange reac-
tions, the losses of HDO and H2O from HOCD2CD2NHCHO•+ can
also be rationalized.
Table 3 shows, in agreement with the mechanism of Scheme 7b,
that most of the ions HOCH2CH2NHCH18O•+ lose their hydroxyl
oxygen as H2O16. However, ∼25% of the ions loses the labelled
amide oxygen. We further note that ions HOCH2CH2NHCD18O•+
lose H2O, HDO, H218O and HD18O, which indicates that the loss of
the 18O- and D-labels involves different (exchange) mechanisms.
A likely rationale for the partial loss of the 18O-label is that a key
intermediate of Scheme 7 communicates with an isomer whose
oxygen atoms are equivalent. This scenario was probed by many
exploratory calculations that yielded the two pathways depicted
in Scheme 10 as plausible candidates. The first one involves ion
FE9 generated by the route IDC → FE8 → FE9. In this reaction, the
water molecule of IDC attacks the terminal methylene group and
synchronously transfers one of its hydrogens to the carbonyl oxy-
gen atom. The resulting ion FE8 may then undergo a 1,4-OH shift
to generate ion FE9. The second option is slightly more energy
O
D
C
C
H
H
O
D
O
N
C
N
C
C
H
N
C
H
H
H
+ HDO
via HBRC9
(Figure 5)
O
CH3
CH2
O
H
H
FE7
[130]
CH3
H
FE3
[120]
1,5-H + rot
[129]
1,4-H
[136]
1f
Σ = [137]
[129] Scheme 7
1,5-H
H
H
O
HO
D
D
O
H
C
C
N
H
H
O
N
C
+ H2O
H
N
D
H
C
C
C
H
H
CH2
CH2
O
CH2
QPQ
HBRC3a
HBRC7
1c
[118]
H
[108]
[112]
Σ = [130]
H
N
C
D
C
O
O
H
H
CH2
QPQ
1e + HDO
Σ = [146]
HBRC8
[122]
X
[138]
Scheme 9. Potential pathways for the minor HDO loss from the N-formylethanolamine isotopologue HOCH2CH2NHCDO. Numbers in brackets are CBS-QB3 derived 298 K
enthalpies in kcal mol−1
.

24
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
H
H
OH
H
HO
C
H
H
H
C
C
C
O
H
H
O
N
C
C
O
H
C
H
H
N
C
OH
CH2
H
N
C
HN
H
H
H
H
N
O
O
O
CH2
CH2
C
CH2
H
H
HO
FE10
CH2
[108]
H
H
1,4-OH
[139]
[140]
[120]
[107]
FE3
IDC
[117]
FE8
FE9
[142]
[118]
loss of H2O of Scheme 7
Scheme 10.
demanding: it involves cyclization of ion FE3 to the symmetric ion
FE10.
Instead, we propose that the water loss involves a concerted
cyclization and direct bond cleavage in PFE1c. The resulting proton-
bridged ion PFE2 may then lose water (as D2O) to produce
protonated 2-oxazoline. An analogous mechanism has been pro-
posed for the loss of H2O from serine residues in protonated
peptides and proteins [29].
The energy of the TS of these exchange routes lies below that of
the incipient molecular ions FE1 of Scheme 7 (144 kcal mol−1), but
well above the dissociation threshold 1 + H2O. This may be the rea-
son why only a fraction of the 18O-labelled ions loses the positional
identity of the oxygen atoms prior to dissociation. In the same vein,
the proposed H/D exchange reactions of the D-labelled ions do not
lead to a statistical distribution of the label [28].
A direct 1,2-H shift in PFE1b [14] cannot account for the com-
peting loss of CO because it requires 24 kcal mol−1 more energy
than our proposed H2O loss mechanism. However, a search of
the potential energy surface yielded the remarkable structure “TS
[95]” of Scheme 11, whose energy lies 2 kcal mol−1 below the
rate-determining TS for the water loss. In this decarbonylation,
the amide N–C bond of “TS [95]” elongates as the aldehydic H is
transferred to the hydroxyl group. The resulting transient complex,
3.6. The losses of H2O and CO from protonated
N-formylethanolamine
N-formylethanolamine may be protonated at three different
sites: at the amide oxygen, the amide nitrogen, or the hydroxyl
oxygen. The resulting ions, PFE1a–c, are shown in Scheme 11. As
expected, the carbonyl protonated ion PFE1a is calculated to have
a much greater stability than the other tautomers.
Tip et al. [14] propose that the prominent loss of H2O (see Fig. 2c
and d) is initiated by a 1,3-H shift in N-protonated ions PFE1b, yield-
ing N-protonated N-vinylformamide. This, however, is at odds with
our observation that D-labelled ions DOCH2CH2NDCHOD+ specifi-
cally lose D2O. Moreover, the proposed 1,3-H shift is undoubtedly
quite energy demanding, much like the 1,3-H shift connecting
PFE1a and PFE1b in Scheme 11.
+
comprising the O-protonated species NH2CH2CH2OH2 and CO,
then rearranges via a 1,4-H shift to produce +NH3CH2CH2OH + CO.
As suggested in Section 3.1, the notable peaks at m/z 71, 59 and
46 in the CID spectrum of PFE ions generated by self-protonation
(Fig. 2d) arise from PFE ions that have lost a H• radical to pro-
duce m/z 89 ions having dissociation characteristics akin to those
of ionized N-formylethanolamine. We have not computationally
examined the myriad of NFEA isomers that can be envisaged to be
generated upon H• loss. An attractive possibility is that distonic ion
FE2 of Scheme 7, a key intermediate in the dissociation of ion FE1,
is generated.
H
H
H
H
O
D
O
C
C
C
C
D
D
1,6-H + rot
[89]
r.c.
O
O
D
N
D
D
N
O
D
N
+ D2O
D
O
N
O
[97]
CH2
CH2
CH2
CH2 CH2
CH2
CH2
CH2
D
O
D
m/z 72
[73]
PFE1a
[62]
PFE1c
[88]
PFE2
[57]
protonated
2-oxazoline
1,4-H [87]
1,3-H
[121]
X
H
H
1.3 Å
C
O
C
O
2.3 Å
H
H
N
C
1.3 Å
D
D
D
D
D
D
N
OD
N
OD
CH2 CH2
PFE1b [81]
CO
+
HDO
D
+
CH2
D
N
CH2 CH2
CH2 CH2OD
C
H
[95]
TS
[70]
m/z 62
m/z 72
N-protonated
N-vinylformamide
X
X
protonated
1,3-H
[121]
1,2-H
ethanolamine
Scheme 11. Mechanistic proposals for the losses of H2O and CO from protonated N-formylethanolamine (PFE1a). Numbers in brackets are CBS-QB3 derived 298 K enthalpies
in kcal mol−1
.

K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
25
4. Summary
his influence on scientists of many persuasions and interests, is
acknowledged with thanks
The fruitful interplay of theory and experiment of this study
reveals that distonic ions and hydrogen-bridged radical cations
(HBRCs) play a key role in the fascinating dissociation chemistry of
the N-formylethanolamine radical cation, HOCH2CH2NHC(H) O•+
(FE1).
JKT and KJJ thank the Natural Sciences and Engineering Research
Council of Canada (NSERC) for financial support and Prof. N.H.
Werstiuk for obtaining the UV photoelectron spectrum of N-
formylethanolamine. The assistance of Mr Tariq Mahmood in
optimising conditions for the condensation of ethanolamine with
methyl formate is gratefully acknowledged.
As discussed in Section 3.2, the primary fragmentations
involving the loss of CH2O and C2H3O• yield the distonic
ion •CH2N(H)CHOH+ and O-protonated formamide HC(OH)NH2
+
References
respectively. The competing H2O loss yields a mixture of N-
vinylformamide ions 1b and its cyclized distonic isomer 1a: the
CID spectrum is very close to a reference spectrum of 1b, but
its enhanced charge-stripping peak betrays that ions 1a are also
present.
Rotation of the amide C–N bond of neutral and ionized N-
formylethanolamine involves a sizeable energy barrier. In Section
3.3 and Scheme 6, theory provides a rationale for the experimental
finding that the ions do not decarbonylate which implies that these
geometrical isomers have the same dissociation characteristics.
The mechanistic analysis of Section 3.4 and Scheme 7 indicates
that:
[1] (a) I. Howe, D.H. Williams, R.D. Bowen, Mass Spectrometry: Principles and
Applications, second ed., McGraw-Hill International Book Company, New York,
1981;
(b) F.W. McLafferty, F. Turecˇek, Interpretation of Mass Spectra, fourth ed., Uni-
versity Science Books, Mill Valley, 1993, p. 81;
(c) R.M. Smith, Understanding Mass Spectra: A Basic Approach, second ed., John
Wiley & Sons, Hoboken, 2004, p. 234;
(d) F. Turecˇek, in: N.M.M. Nibbering (Ed.), Encyclopedia of Mass Spectrometry,
vol. 4, Elsevier, Amsterdam, 2005, p. 396.
[2] M.B. Stringer, D.J. Underwood, J.H. Bowie, C.E. Allison, K.F. Donchi, P.J. Derrick,
Org. Mass Spectrom. 27 (1992) 270.
[3] D. Kuck, in: N.M.M. Nibbering (Ed.), Encyclopedia of Mass Spectrometry, vol. 4,
Elsevier, Amsterdam, 2005, pp. 110–111.
[4] (a) G. Bouchoux, Y. Hoppilliard, Int. J. Mass Spectrom. 90 (1989) 197;
(b) C.E. Hudson, L.L. Griffin, D.J. McAdoo, Org. Mass Spectrom. 24 (1989) 866;
(c) A.E. Dorigo, M.A. McCarrick, R.J. Loncharich, K.N. Houk, J. Am. Chem. Soc.
112 (1990) 7508.
[5] (a) J. Loos, D. Schröder, W. Zummack, H. Schwarz, R. Thissen, O. Dutuit, Int. J.
Mass Spectrom. 214 (2002) 105;
(i) cleavage of the C–C bond in the incipient molecular ions FE1
generates HBRC1, [CH2N(H)C(H) O· · ·H· · ·O = CH2]•+, which
serves as the direct precursor to CH2O loss.
(b) M. Semialjac, J. Loos, D. Schröder, H. Schwarz, Int. J. Mass Spectrom. 214
(2002) 129;
(c) D. Schröder, J. Loos, M. Semialjac, T. Weiske, H. Schwarz, G. Hohne, R. Thissen,
O. Dutuit, Int. J. Mass Spectrom. 214 (2002) 155.
(ii) a competing 1,5-H shift, the first step of
a
McLaf-
ferty
rearrangement,
generates
distonic
ion
FE2,
HOCHCH2N(H)C(H)OH•+,which serves as the precursor
for the loss of both C2H3O• and H2O.
[6] (a) J.A. Montgomery Jr., M.J. Frisch, J.W. Ochterski, G.A. Petersson, J. Chem. Phys.
112 (2000) 6532;
(iii) loss of C2H3O• from FE2 involves
a
“McLafferty + 1ꢀꢀ
(b) L.A. Curtiss, P.C. Redfern, K. Raghavachari, J. Chem. Phys. 126 (2007) 084108;
(c) J.M.L. Martin, S. Parthiban, in: J. Cioslowski, A. Szarecka (Eds.), Quantum
Mechanical Prediction of Thermochemical Data, Understanding Chemical Reac-
tivity Series, vol. 22, Kluwer Academic Publishers, Dordrecht, 2005, pp. 31–65;
(d) L.N. Heydorn, Y. Ling, G. de Oliveria, J.M.L. Martin, Ch. Lifshitz, J.K. Terlouw,
Z. Phys. Chem. 215 (2001) 141.
type reaction in which HBRCs act as key intermediates:
FE2 → HBRC2 → HBRC4a → HC(OH)NH2+ + CH2 CHO•. Prior
to dissociation, a fraction of ions HBRC4a may undergo the
quid-pro-quo (QPQ) catalysis of Scheme 8, which would
co-generate the more stable acetyl radical CH3C O•.
[7] (a) R. Lee, P.J.A. Ruttink, P.C. Burgers, J.K. Terlouw, Can. J. Chem. 83 (2005) 1847;
(b) K.J. Jobst, T.R. Khan, J.K. Terlouw, Int. J. Mass Spectrom. 42 (2007) 1024.
[8] P.C. Burgers, J.K. Terlouw, in: N.M.M. Nibbering (Ed.), Encyclopedia of Mass
Spectrometry, vol. 4, Elsevier, Amsterdam, 2005, p. 173.
[9] (a) R. Lee, P.J.A. Ruttink, P.C. Burgers, J.K. Terlouw, Int. J. Mass Spectrom. 255
(2006) 244, and references cited therein;
(b) K.J. Jobst, P.C. Burgers, P.J.A. Ruttink, J.K. Terlouw, Int. J. Mass Spectrom. 254
(2006) 127.
[10] (a) D.J. McAdoo, C.E. Hudson, M. Skyiepal, E. Broido, L.L. Griffin, J. Am. Chem.
Soc. 109 (1987) 7648;
(iv) loss of H2O from FE2 involves communication with
ion–molecule complexes of vinyl alcohol and formimidic
acid, whose components may recombine to form distonic
ion FE3, HOCH(CH2)N(H)C(H)OH•+. This ion loses H2O via
the 1,5-H shift reaction FE3 → HBRC3a → ion 1b + H2O. The
facile cyclization of HBRC3a to HBRC3c may account for the
co-generation of ion 1a.
(b) J. Loos, D. Schröder, H. Schwarz, J. Org. Chem. 70 (2005) 1073.
[11] J.L. Holmes, K.J. Jobst, J.K. Terlouw, J. Labelled Compd. Radiopharm. 50 (2007)
1088.
The behaviour of the various D- and 18O-labelled isotopologues
examined in Section 3.5 is basically in accord with the above mech-
anisms. The MI spectra of the labelled ions indicate that, in the loss
of water, the positional identity of the D and O atoms is not com-
pletely retained. A rationale provided by theory for this finding
involves the participation of the QPQ type reactions of Scheme 9
for H/D exchange. Scheme 10 presents two side reactions of ion
IDC, a key intermediate in the H2O loss of Scheme 7, that lead to
unreactive isomers whose O atoms are equivalent. These reactions
could account for the partial equilibration of the O-atoms.
[12] (a) H.F. van Garderen, P.J.A. Ruttink, P.C. Burgers, G.A. McGibbon, J.K. Terlouw,
Int. J. Mass Spectrom. Ion Process. 121 (1992) 159;
(b) N.H. Werstiuk, D.N. Butler, E. Shahid, Can. J. Chem. 65 (1986) 760.
[13] J.L. Holmes, C. Aubry, P.M. Mayer, Assigning Structures to Ions in Mass Spec-
trometry, CRC Press, Boca Raton, 2007.
[14] L. Tip, C. Versluis, J.W. Dallinga, W. Heerma, Anal. Chim. Acta 241 (1990) 219.
[15] M.J. Frisch et al. Gaussian 03 (Revision C.02), Gaussian, Inc., Wallingford CT,
2004.
[16] G. Schaftenaar, R. Postma, P.J.A. Ruttink, P.C. Burgers, G.A. McGibbon, J.K. Ter-
louw, Int. J. Mass Spectrom. Ion Process. 100 (1990) 521.
[17] G.A. McGibbon, P.C. Burgers, J.K. Terlouw, Int. J. Mass Spectrom. 136 (1994) 191.
[18] P.R. Schreiner, H.P. Reisenauer, F.C.I.V. Pickard, A.C. Simmonett, W.D. Allen, E.
Matyus, A.G. Csaszar, Nature 453 (2008) 906.
[19] (a) G. da Silva, C.-H. Kim, J.W. Bozzelli, J. Phys. Chem. A 110 (2006) 7925;
(b) E.A. Fogleman, H. Koizumi, J.P. Kercher, B. Sztáray, T. Baer, J. Phys. Chem. A
108 (2004) 5288;
(c) G. Bouchoux, J. Chamot-Rooke, D. Leblanc, P. Morgues, M. Sablier, Chem.
Phys. Chem. 2 (2001) 235.
[20] B.L.M. van Baar, P.C. Burgers, J.L. Holmes, J.K. Terlouw, Org. Mass Spectrom. 23
(1988) 355 (note that the CIDI mass spectra of CH3CO• (Fig. 2a) and CH2CHO•
(Fig. 2b) are transposed).
Finally, a re-examination of the dissociation of protonated
N-formylethanolamine proposes that the losses of H2O and CO
generate protonated 2-oxazoline and N-protonated ethanol-amine
respectively via the mechanisms of Scheme 11.
Acknowledgements
[21] (a) T. Wong, J. Warkentin, J.K. Terlouw, Int. J. Mass Spectrom. Ion Process. 115
(1992) 33;
(b) M.M. Cordero, J.J. Houser, C. Wesdemiotis, Anal. Chem. 65 (1993) 1594.
[22] J.K. Terlouw, P.C. Burgers, H. Hommes, Org. Mass Spectrom. 14 (1979) 387.
[23] P.J.A. Ruttink, P.C. Burgers, J.K. Terlouw, Int. J. Mass Spectrom. Ion Process. 145
(1995) 35.
The role of the late Professor Dudley Williams in developing and
popularising experimental approaches to studying and describ-
ing the reactions of ions [30], his enthusiasm for science, and

26
K.J. Jobst et al. / International Journal of Mass Spectrometry 306 (2011) 9–26
[24] (a) D.M. Pawar, A.A. Khalil, D.R. Hooks, K. Collins, T. Elliot, J. Stafford, L. Smith,
E.A. Noe, J. Am. Chem. Soc. 120 (1998) 2108;
(b) S. Humbel, I. Côte, N. Hoffmann, J. Bouquant, J. Am. Chem. Soc. 121 (1999)
5507.
(b) J.P. Terhorst, W.L. Jorgensen, J. Chem. Theory Comput. 6 (2010) 2762.
[25] W. Heerma, M.M. Sarneel, J.K. Terlouw, Org. Mass Spectrom. 16 (1981)
326.
[26] S. Nakajima, T. Asakawa, O. Sekiguchi, S. Tajima, N.M.M. Nibbering, Eur. J. Mass
Spectrom. 7 (2001) 47.
[28] R.G. Cooks, J.H. Beynon, R.M. Caprioli, G.R. Lester, Metastable Ions, Elsevier
Scientific Publishing Company, New York, 1973, p. 257.
[29] B. Paizs, S. Suhai, Mass Spectrom. Rev. 24 (2005) 508.
[30] (a) D.H. Williams, Acc. Chem. Res. 10 (1977) 280;
(b) R.D. Bowen, D.H. Williams, H. Schwarz, Angew. Chem. Int. Ed. Eng. 18 (1979)
451.
[27] (a) P.M.W. Gill, L. Radom, J. Am. Chem. Soc. 110 (1988) 4931;