Chemical and spectroscopic studies related to the Lewis acidity of lithium perchlorate in diethyl ether

Article abstract of DOI:10.1021/jo981042x

Polarimetric studies on camphor (2) as well as IR studies on crotonaldehyde (CA;1) and benzonitrile (BN;3) confirm the conclusion of a previously published NMR study on crotonaldehyde that lithium perchlorate (LP) weakly binds to probe bases in diethyl ether (DE). The weak binding is a consequence of the fact that the lithium ion (actually the LP ion pair and higher aggregates), a powerful Lewis acid in the gas phase, competitively binds to ether and the added base. Methylene camphor (5), (E)-1,3-pentadiene (4), camphene, and phenylacetylene (6) do not bind to LP in DE. Shifts to lower energy of the C=O modes of CA in ether solutions containing increasing amounts of LP are consistent with moderate increases in solvent polarity. Only small or no shifts are seen in the C≡N modes of BN and its 1:1 complex with added LP. Because the C≡N and especially C=O modes are blue shifted under external applied pressure, the large internal pressures of LP/DE do not mimic external applied pressure. Likewise, the small or no changes observed in λ(max) for the absorption and emission spectra of anthracene (9) and azulene (8) in ether as a function of LP concentration do not conform to what is observed under external applied pressure. Studies of the Diels-Alder reaction of (E)-1,3-pentadiene with methyl acrylate show that the reaction is entirely catalyzed in LP/DE; polarity and internal pressure do not influence product selectivity in this reaction.

Full text of DOI:10.1021/jo981042x

2
202
J . Org. Chem. 1999, 64, 2202-2210
Ch em ica l a n d Sp ectr oscop ic Stu d ies Rela ted to th e Lew is Acid ity
of Lith iu m P er ch lor a te in Dieth yl Eth er
Gerald Springer, Chanda Elam, Anna Edwards, Craig Bowe, David Boyles, J ohn Bartmess,
Martin Chandler, Kevin West, J an Williams, J ames Green, Richard M. Pagni,* and
George W. Kabalka*
Department of Chemistry, University of Tennessee, Knoxville, Tennessee 37996-1600
Received J une 1, 1998
Polarimetric studies on camphor (2) as well as IR studies on crotonaldehyde (CA; 1) and benzonitrile
(BN; 3) confirm the conclusion of a previously published NMR study on crotonaldehyde that lithium
perchlorate (LP) weakly binds to probe bases in diethyl ether (DE). The weak binding is a
consequence of the fact that the lithium ion (actually the LP ion pair and higher aggregates), a
powerful Lewis acid in the gas phase, competitively binds to ether and the added base. Methylene
camphor (5), (E)-1,3-pentadiene (4), camphene, and phenylacetylene (6) do not bind to LP in DE.
Shifts to lower energy of the CdO modes of CA in ether solutions containing increasing amounts
of LP are consistent with moderate increases in solvent polarity. Only small or no shifts are seen
in the CtN modes of BN and its 1:1 complex with added LP. Because the CtN and especially
CdO modes are blue shifted under external applied pressure, the large internal pressures of LP/
DE do not mimic external applied pressure. Likewise, the small or no changes observed in λmax for
the absorption and emission spectra of anthracene (9) and azulene (8) in ether as a function of LP
concentration do not conform to what is observed under external applied pressure. Studies of the
Diels-Alder reaction of (E)-1,3-pentadiene with methyl acrylate show that the reaction is entirely
catalyzed in LP/DE; polarity and internal pressure do not influence product selectivity in this
reaction.
In tr od u ction
Lithium perchlorate in diethyl ether (LP/DE) has
the solvent,11 which then mimics the effect that P
on reactions. Kumar determined P for LP/DE from
thermodynamic and other data and showed that Pi
a
has
i
become a widely used medium in organic synthesis,
particularly for what are formally pericyclic reactions.1
The mode by which this solvent does its work, however,
has become a source of controversy and uncertainty.
Consider the following points. (1) Pocker has shown that
indeed goes up significantly as the concentration of LP
in DE goes up. Kumar also suggested on theoretical
12
grounds that P does mimic the effect that P has on the
i
a
rates of DA reactions. In fact the rates of the Diels-Alder
reaction of N-methylmaleimide with anthracene-9-carbinol
in aqueous salt solutions correlate nicely with internal
pressure, but the internal pressure has only a minimal
effect on rates.13 Likewise, the dimerization of cyclopen-
tadiene correlates with internal pressure, but the cor-
relation is not very good.14 Others have also shown
LP in DE has a very large accelerating effect on reactions
that yield ions, e.g., S
N
1 reactions.2 He attributed these
-8
rate enhancements to electrostatic catalysis in which the
large Coulomb fields of the ions stabilize polar transition
states. Much of these data can also be rationalized by
mechanisms in which the lithium ion functions as a
Lewis acid catalyst. (2) On the basis of how LP/DE and
experimentally that P
cycloaddition reactions as P
i
does not have the same effect on
has;1
5-17
its effect is at best
a
quite small. (3) Braun and Sauer, on the other hand, have
proposed that changes in product selectivity for the
reactions of cyclopentadiene (CP) with methyl acrylate
hydrostatic (applied) pressure (P
a
) influence Diels-Alder
(DA) reactions in general and the DA reaction of furan
with 2,5-dihydrothiophene-3,4-dicarboxylic anhydride in
particular, which is used in the total synthesis of
1
8
(MA) in LP/DE are due to changes in solvent polarity,
as one might expect when salts are dissolved in liq-
9
10
cantharidin, Grieco made the novel suggestion that
adding LP to DE increases the internal pressure (P ) of
uids.1
1,19
Braun and Sauer in fact calculated Ω, a solvent
i
2
0
polarity index devised by Berson which is based on the
CP/MA reaction, for the LP/DE salt solutions. Kiselev and
(1) (a) Grieco, P. A. Aldrichim. Acta 1991, 24, 59. (b) Waldmann,
H. Angew. Chem., Int. Ed. Engl. 1991, 30, 1306. (c) Grieco, P. A. In
Organic Chemistry: Its Language and Its State of the Art; Kisak u¨ rek,
M. V., Ed.; VCH: Weinheim, 1993; p 133. (d) Flohr, A.; Waldmann,
H. J . Prakt. Chem. 1995, 37, 609.
(11) Reichardt, C. Solvents and Solvent Effects in Organic Chemistry,
2nd ed.; VCH: Weinheim, 1988.
(12) Kumar, A. J . Org. Chem. 1994, 59, 4612.
(13) Kumar, A. Pure Appl. Chem. 1998, 70, 615.
(14) Reference 11, p 193.
(15) Dauben, W. G.; Lam, J . Y. L.; Guo, Z. R. J . Org. Chem. 1996,
61, 4816.
(16) Faita, G.; Righetti, P. P. Tetrahedron 1995, 51, 9091.
(17) J enner, G.; Salem, R. B. Tetrahedron 1997, 53, 4637.
(18) Braun, R.; Sauer, J . Chem. Ber. 1986, 119, 1269.
(19) Loupy, A.; Tchoubar, B. Salt Effects in Organic and Organo-
metallic Chemistry; VCH: Weinheim, 1992.
(20) Berson, J . A.; Hamlet, Z.; Mueller, W. A. J . Am. Chem. Soc.
1962, 84, 297.
(2) Pocker, Y.; Buchholz, R. F. J . Am. Chem. Soc. 1970, 92, 2075.
(3) Pocker, Y.; Buchholz, R. F. J . Am. Chem. Soc. 1970, 92, 4033.
(4) Pocker, Y.; Buchholz, R. F. J . Am. Chem. Soc. 1971, 93, 2905.
(5) Pocker, Y.; Ellsworth, D. L. J . Am. Chem. Soc. 1977, 99, 2276.
(6) Pocker, Y.; Ellsworth, D. L. J . Am. Chem. Soc. 1977, 99, 2284.
(7) Pocker, Y.; Ciula, J . C. J . Am. Chem. Soc. 1988, 110, 2904.
(8) Pocker, Y.; Ciula, J . C. J . Am. Chem. Soc. 1989, 111, 4728.
(9) Dauben, W. G.; Kessel, C. R.; Takemura, K. H. J . Am. Chem.
Soc. 1980, 102, 6894.
10) Grieco, P. A.; Nunes, J . J .; Gaul, M. D. J . Am. Chem. Soc. 1990,
12, 4595.
(
1
1
0.1021/jo981042x CCC: $18.00 © 1999 American Chemical Society
Published on Web 03/11/1999

Lewis Acidity of Lithium Perchlorate
J . Org. Chem., Vol. 64, No. 7, 1999 2203
co-workers arrived at a conclusion similar to that of
Braun and Sauer based on the kinetics of several (4+2)-,
(
3+2)-, and (2+2)-cycloaddition reactions.2 In one (3+2)-
1
cycloaddition the rate was actually slower in LP/DE than
in DE because one of the polar reactants was stabilized
by the polar LP/DE more than the transition state was.21
Neither of these structuressand others for the larger
aggregatessshould be envisioned as static, however.
There is clearly a rapid interchange of the ions in the
various structures as well as the complexed ethers with
those in the solvent. LP in ether must contain therefore
an admixture of ion pairs and higher aggregates, all in
rapid equilibrium and whose composition depends on the
initial concentration of LP.
(4) A still different perspective was given by Pagni,
Kabalka, et al., who showed using chemical, spectro-
scopic, and theoretical data that LP in DE functions as
a mild Lewis acid, i.e., catalyst, with the strong Lewis
+
acidity of Li being abated in part by competitive com-
2
2
plexation to DE. Kinetic studies are also consistent with
LP functioning as a catalyst.1
6,23
We have now carried out a new set of experiments,
largely different in character from those reported in our
Several other properties of LP in DE should be
mentioned, as they may also have a bearing on the
various mechanisms of “catalysis” described above. On
the basis of the measurements of conductivity and
viscosity of LP/DE, Ekelin and Sill e´ n in the 1950s
proposed that LP in ether is aggregated (ion pairs and
higher aggregates) and the lithium ions are, in addition,
22
earlier work, to assess the catalytic power of LP in DE.
We wish to report the results of these experiments at this
time. We hope these new results, and those published
earlier, will give a new perspective on the competing roles
of catalysis, polarity, and internal pressure of LP/DE in
organic reactions.
2
4
complexed to an unknown number of ether molecules.
1
In the early 1970s Pocker demonstrated by H NMR
spectroscopy that each lithium ion is in fact bound to
Resu lts a n d Discu ssion
2
One way to assess the importance of catalysis to
reactions occurring in LP/DE is to see if LP binds to
species with lone pairs of electrons, and how strongly.
To see why this is important consider the LP-catalyzed
DA reaction of MA with CP to form product (P) (actually
endo and exo adducts) which requires three steps: (1)
the reversible binding of LP to MA; (2) the DA reaction
of the complex with CP; and (3) the reversible binding of
the product to LP.
two ether molecules in solutions up to 4.25 M in LP, the
concentration at which all the ether is complexed and
the solution becomes a molten salt;25 between 4.25 and
6
.00 M, where saturation is reached, the molten salt
becomes progressively richer in lithium ions bound to one
2
ether. Somewhat later M e´ nard and Chabanel proved
using dielectric measurements that LP ion pairs exist
primarily in dimers (eq 1) at low concentrations of LP in
-1
2
LiClO h (LiClO )
K ) 650 M
(1)
4
4 2
Ke
MA + LP y\ z MA‚LP
DE (<0.1 M) at 20 °C.26 Presumably at higher LP
concentrations, higher aggregates of LP exist. Thus, if
each lithium ion is complexed to two ether molecules and
also exists in an LP dimer at low LP concentration, a
picture emerges in which each lithium ion in the dimer
exists in a tetrahedral environment, as shown below.
k2
MA‚LP + CP
98 P‚LP
P‚LP y z P + LP
Ke′
The rate of formation of the product and its complex is
given by28
d[P] d[P‚LP]
+
) kobs[CP][MA][LP]
dt
dt
where kobs ) k
2
K
e
.29,30 Thus the magnitude of the binding
of the Lewis acid to the dienophile directly influences any
rate acceleration for the DA reaction. The rate accelera-
2
tion is also influenced by the magnitude of k , whose
The monomeric LP ion pair can be detected as well by
IR and Raman spectroscopy. On the basis of how the
complexing lithium ion perturbs the tetrahedral perchlo-
rate anion away from ideal symmetry, Chabanel and co-
workers showed in 1996 that the anion functions as a
monodentate ligand.27 Thus, the structure of the ion pair,
at least at low LP concentration, is
value depends on how effectively the Lewis acid lowers
the LUMO of the dienophile,31 and also the intrinsic
reactivity of the diene.
The molecules initially chosen for spectroscopic study
were (E)-2-butenal (1; CA ) crotonaldehyde), a model
(28) This expression is preferred to that for d[P]/dt because the latter
e
one will contain terms in K ′ which are not relevant to the present
discussion.
(
21) Shtyrlin, Y. G.; Murzin, D. G.; Luzanova, N. A.; Ishkakova, G.
G.; Kiselev, V. D.; Konovalov, A. I. Tetrahedron 1998, 54, 2631.
22) (a) Pagni, R. M.; Kabalka, G. W.; Bains, S.; Plesco, M.; Wilson,
(29) If the uncatalyzed Diels-Alder reaction competes with the
catalyzed reaction, an unusual situation but one that occurs here when
CP is the diene (this is addressed later in the paper), the rate
expression will contain an additional term. This does not occur, even
in LP/DE, when less reactive dienes are used.
(30) There are, of course, other limiting cases. For instance, if the
binding of Lewis acid and dienophile is fast and essentially irreversible,
(
J .; Bartmess, J . J . Org. Chem. 1993, 58, 3130. (b) Bains, S.; Pagni, R.
M.; Kabalka, G. W.; Pagni, R. M. Tetrahedron Asymmetry 1994, 5, 821.
(
(
(
23) Forman, M. A.; Dailey, W. P. J . Am. Chem. Soc. 1991, 113, 2762.
24) Ekelin, K.; Sill e´ n, L. G. Acta Chem. Scand. 1953, 7, 987.
25) Solutions below 4.25 M may also be considered a molten salt if
+
-
one considers Li(Et
(
(
2
O)
2
ClO
4
to be the solvent and ether the solute.
2
a common situation, kobs ) k because one is measuring the rate for
26) M e´ nard, D.; Chabanel, M. J . Phys. Chem. 1975, 79, 1081.
27) Chabanel, M.; Legoff, D.; Touaj, K. J . Chem. Soc., Faraday
the reaction of the complex with the diene.
(31) Isaacs, N. S. Physical Organic Chemistry; Longman: Essex,
1987; p 704.
Trans. 1996, 92, 4199.

2
204 J . Org. Chem., Vol. 64, No. 7, 1999
Ta ble 1. Equ ilibr iu m Con sta n ts for th e Rea ction of Ba ses w ith LP in DEa
nature of complex
B‚LPb B2‚LPc
Springer et al.
base (B)
Reichardt’s dye
method
UV/vis
ref
6
5.0 × l0
15.8d
16.8
8
8
8
8
8
8
22a
4
this work
this work
this work
this work
this work
this work
d
7.10e
7.77e
8
8
8
8
,9-diacetylsesquifulvene
18.2
UV/vis
UV/vis
UV/vis
UV/vis
UV/vis
H NMR
polarimetry
polarimetry
polarimetry
IR
IR
IR
IR
,9-dibenzoylsesquifulvene
,9-diformylsesquifulvene
,9-dicarbomethoxysesquifulvene
7.39
N,N-diethylindoaniline
crotonaldehyde
menthone
camphor
methylene camphor
crotonaldehyde
f
3.20g
7.3
4.95
0
1
8.7
0
0.71
0
(E)-1,3-pentadiene
benzonitrile
phenylacetylene
a
Determined at room temperature. b Keq/M-1. c Keq/M-2. d Seen at high [LP]. e Seen at low [LP]. f Keq could not be determined.
g
Determined at -20 °C.
base used in the study of Lewis acidity,32 optically active
Ta ble 2. Op tica l Rota tion s of Ca m p h or in LP /DE
initial concn rotations
camphor]0
camphor (2), and benzonitrile (3; BN), all of which are
weakly basic. Similar studies were also carried out on
the all-carbon analogues of these bases: (E)-1,3-penta-
diene (4; trans-piperylene), methylene camphor (5),33 and
phenylacetylene (6). Additional experiments were also
carried out on acetone (7), azulene (8), and anthracene
observeda
theoryb
[
[LP]0
0
0
0
.534
.535
.535
+4.24
+4.05
+3.56
+3.08
+2.43
+2.25
+4.24
+4.06
+3.56
+3.07
+2.57
+2.29
0.0981
0.250
0.503
1.001
1.904
0.534
0
0
.529
.539
(
9).
a
Taken in a 1 dm cell at +21 °C at the sodium D line. b Assumes
Keq ) 4.95 and the rotation of camphor minus the rotation of
complex ) -2.22°.
rotation of methylene camphor, the all-carbon analogue
of camphor, remained constant at -48.4 ( 0.3° in
solutions from 0.125 to 1.50 M in LP. Clearly LP does
not bind to the alkene. Surprisingly, there is also no
measurable solvent effect on the optical rotation of this
compound.34
Several prior attempts have been made to study
quantitatively) the binding of probe bases (Bs) to LP in
(
4
,8,22a
DE by various spectroscopies (Table 1).
In none of
the reported cases was the complex seen directly; it was
only “detected” by inference based on assumptions used
to fit the spectral data to a model. Considering the
complexity of LP in DE, as described earlier, the majority
of cases, remarkably, fit the simplest model: B + LP h
B‚LP.
Direct detection of the contributing components would
verify the UV/vis, NMR, and chirooptical results and give,
in addition, information on how the species respond to
changes in solvent properties. IR spectroscopy is well
suited for these purposes. One, of course, can detect
chromophores such as nitrile and carbonyl groups by
their distinctive stretching frequencies. Complexation of
these groups often leads to new stretching frequencies,
and species in dynamic equilibrium are often seen by IR
when they are not by other spectroscopies.
Similar to that reported previous for optically active
4
menthone, the binding of camphor to LP in DE was
studied polarimetrically (Table 2), fitting the changes in
optical rotation to a mathematical model reported in a
previous publication in which the binding of crotonalde-
hyde to LP was studied by 1H NMR.22a The fit was
surprisingly good considering that optical rotations are
solvent dependent, and there is a large difference in
solvent properties on going from pure ether to 1.90 M
LP in ether. The fit yielded Keq(camphor) ) 4.95, a value
similar to that for menthone (Table 1). By contrast, the
CA (0.100 M) in DE has a carbonyl stretching vibration
-
1
at 1699 cm . When LP at low concentration (0.0400-
.100 M) is incorporated into this solution, a new band
0
-
1
at 1687 cm grows in at the expense of the CA band
(32) Childs, R. F.; Mulholland, D. L.; Nixon, A. Can. J . Chem. 1982,
6
6
0, 801.
(33) Clase, J . A.; Li, D. L. F.; Lo, L.; Money, T. Can. J . Chem. 1990,
(34) The optical rotations of camphene also remained constant in
DE containing 0-2M LP.
8, 1829.

Lewis Acidity of Lithium Perchlorate
J . Org. Chem., Vol. 64, No. 7, 1999 2205
F igu r e 1. IR spectra of crotonaldehyde in LP/DE, carbonyl region at low LP concentration.
Ta ble 3. Cu r ve-F ittin g Resu lts (Lor en tzia n Lin e Sh a p e)
Ta ble 4. Ca lcu la tion of Keq for CA + LP Rea ction a
a
for CA in LP /DE, Low -Con cen tr a tion Regim e
at equilibrium
CA band
WHHb,c
CA‚LP band
WHHb,c
[
LP]0
[CA]b
[CA‚LP]c
[LP]d
Keq
[
LP]0
λmaxb
area
λmaxb
area
0
.0400
0.0723
0.0686
0.0646
0.0614
0.0279
0.0314
0.0362
0.0378
0.0121
0.0286
0.0438
0.0622
31.9
16.0
12.8
9.9
0
0
0
0
0
1699
1699
1699
1699
1699
8.38
7.57
7.56
7.59
7.79
l8.53
13.41
12.72
11.98
11.39
0.0600
0.0800
0.100
.0400
.0600
.0800
.100
1687
1686
1686
1685
28.87
24.33
23.15
20.70
7.27
8.19
9.44
9.86
a
[CP]0 ) 0.100 M; ꢀCA ) 23780 and ꢀCA‚LP ) 33445; d ) 0.00780
cm. b [CA] ) ACA/ꢀCAd. c [CA‚LP] ) ACA‚LP/ꢀCA‚LPd. d [LP] ) [LP]0
- [CA‚LP].
a
CA]0 ) 0.100 M. b In cm-1. c Width at half-height.
[
(
Figure 1). The new band at lower energy is due to the
concentration of CA, and ꢀCA and ꢀCA‚LP are the respective
integrated extinction coefficients, is obeyed (r ) 0.993).
3
9
carbonyl stretch of the CP‚LP complex. It is well-known
that complexation of a Lewis acid to an aldehyde or
ketone results in a carbonyl vibration at lower energy.35
MNDO calculations on CA and CA‚LP yield the expected
shift, although the calculated magnitude is much larger
ACA‚LP ACA
ꢀ
CA‚LP
)
ꢀ
CA‚LPd -
(2)
(
)(
)
[CA]0
ꢀCA [CA]0
-
1
-1 36
than what is observed (90 cm vs 12 cm ).
The slope and intercept of eq 2 yield the two extinction
coefficients (ꢀCA‚LP > ꢀCA) and thus Keq (Table 4). Surpris-
Keq
CA + LP y z CA‚LP
0
ingly, Keq goes down as [LP] goes up (log Keq ) 1.19 (
0
.12). This behavior often arises when a third (or more)
3
9
component is present in the system. This is not the case
here, however, because factor analysis,40 which is a
mathematical procedure designed specifically to identify
the number of components in a system, shows that only
two components, CA and CA‚LP, are required to fit the
data for the titration of CA by LP in DE, with log Keq
0
because the titration is only 30% complete. One cannot
Because the system is well behaved at low [LP], i.e.,
yielding an isosbestic point between the two bands, it is
possible to calculate Keq for the reaction. One way to do
this requires knowledge of the areas under the two bands.
This was accomplished by a curve-fitting routine assum-
ing the bands had Gaussian, Lorentzian, or Pearson line
shapes. Only two bands were required to give very good
fits (r ≈ 0.99) for all three line shapes. The results, which
are shown in Table 3 for the Lorentzian line shape, reveal
that λmax(CA) remains constant while λmax(CA‚LP) changes
very slightly, with the difference in peak positions being
)
.94 ( 0.11.4
1-43
This value is subject to some uncertainty
(
38) Nakanishi, K. Infrared Absorption Spectroscopy; Holden-Day:
San Francisco, 1964.
39) (a) Bulmer, J . T.; Shurvell, H. F. In Vibrational Spectra and
Structure; Durig, J . R., Ed.; Elsevier: Amsterdam, 1977; Vol. 6; p 91.
b) McBryde, W. A. E. Talanta Rev. 1974, 21, 979.
40) Harman, H. H. Modern Factor Analysis, 3rd ed.; University of
(
-
1 37
1
2-14 cm
.
As expected for a well-behaved two-
(
component system, eq 2, where ACA and ACA‚LP are the
areas under the bands, d is the cell length (determined
(
Chicago: Chicago, 1976.
by the interference fringe method 38), [CA]
is the initial
0
(41) It is not clear why the more direct method gives a seemingly
less precise estimate of Keq than factor analysis even though the data
fit eq 1 nicely. In fact the Keq values and the error limits are not
appreciably different.
(
(
35) Cassimatis, D.; Susz, B. P. Helv. Chim. Acta 1960, 43, 852.
36) A better model for a complex might have been an adduct of the
LP ion pair with CA. One would also expect to see a shift in the
carbonyl stretching frequency to lower energy, but not as much as for
(42) These Keq values, and many in Table 1, assume that all the LP
is in the form of the LP ion pair, which is not the case. If one takes
-
1
2 0
into account the fact that 2LP h (LP) , with Keq ) 650 M at the LP
+
the CA‚Li complex.
concentrations used here, the calculated equilibrium constant comes
out about an order of magnitude larger.
(
37) The results were essentially the same for all three line shapes.

2
206 J . Org. Chem., Vol. 64, No. 7, 1999
Springer et al.
Ta ble 6. IR Da ta for Aceton ea in LP /DE
Ta ble 5. IR Da ta for CAa in LP /DE, High er
Con cen tr a tion s Regim e
1
[
LP]0
λmax (cm)-
amplitude
CA
CA‚LP
0
1719.5
1718.0
1717.1
1716.4
1716.0
1715.7
1715.2
1714.3
1713.9
0.394
0.421
0.436
0.465
0.481
0.518
0.572
0.637
0.743
[LP]0
λmaxb
WHHb
λmaxb
WHHb
0.0992
0
0
0
0
0
0
1
.151
.207
.247
.297
.401
.750
.001
0
0
0
1
2
3
.100
.300
.600
.00
.00
.00
1699
1696
1694
1692
1688
1687
7.79
10.06
11.06
11.96
12.27
11.65
1686
1682
1680
1680
1679
1680
20.70
16.37
15.08
14.70
13.90
11.75
a
CA]0 ) 0.100 M. b In cm-1
[
.
a
[
acetone]0 ) 0.100 M.
use data beyond this point in the calculation for reasons
that will be come apparent shortly. Interestingly Keq(CA)
is not appreciably different in value from those obtained
for other bases in LP/DE using other spectroscopic
techniques.
The change in λmax for the CO mode of the complex is
more difficult to interpret. The shifts here may also be a
reflection of the change in the polarity of the medium.
They may also be due to the build up of new complexes
with CO stretching vibrations at lower energy and the
synchronous decay of the CO mode for LP‚CA at higher
energy. Although there is no evidence for this from the
shapes of the peaks, this would be consistent with
changes in the widths of the peaks. In an attempt to
resolve this issue, acetone’s CO stretching vibration was
also studied by IR in LP/DE (Table 6; Figure 2). Surpris-
ingly, acetone yielded only a single peak in LP/DE whose
λmax went to lower energy as the LP concentration
increased. Peak amplitudes and widths simultaneously
increased. One may surmise that acetone is not binding
to LP, but this is unlikely. A more reasonable interpreta-
tion, one that is generally believed by researchers in this
area,8 is that there are undetected, but overlapping
bands for acetone and its LP complex. With this premise
in mind it was possible to fit acetone’s spectral data to
As the solutions become still richer in LP, ACA contin-
ues to decrease and ACA‚LP increases, as expected for the
reversible reaction of CA with LP. However, one is no
longer justified in using the areas to calculate Keq. First
of all, the Beer-Lambert law breaks down because band
areas now also correlate with increasing solvent polarity,
which we believe is occurring as LP is added to DE (vide
infra). Published data show that the integrated absorp-
tion intensity (area) for the CdO mode of CA (at constant
concentration) increases significantly as solvent polarity
goes up.44 Second, although no new CdO modes are
observed at higher LP concentrations, the isosbestic point
disappears between the CA and CA‚LP bands because
,46
λ
max(CA) and λmax(CA‚LP) shift to lower energy and the
line widths change as well (Table 5).
The causes for the changes in line widths are unknown
but may be due to environmental influences on lifetime
and dephasing of the vibrational excited states.45 Line
width may also change when new LP-CA complexes are
formed at higher LP concentrations and old complexes
disappear.
The λmax’s of both CO stretching modes shift to lower
energy as the LP concentration in DE increases The shift
in CA’s mode is a consequence of an increase in the
polarity of LP/DE, as reflected, for example, in the
increase in the dielectric constant of the medium. Similar
8
the model used by Pocker to analyze the visible spectra
of dyes in LP/DE. The analysis yielded K ) 1.9 ( 0.3
eq
for A + LP h A‚LP.48 Thus, because the A and A‚LP
bands are not resolved, the spectral data are not helpful
in resolving the origin of the λmax shifts of the CO
stretching mode of LP‚CA.
Unlike the CdO mode of CA, the CdC mode at 1640
cm- does not change position or line width in DE as the
1
4
9
concentration of LP is increased. Likewise, the two
CdC stretching modes of (E)-1,3-pentadiene (4) at 1611
-
1
behavior has been observed for Ba(ClO
4
)
2
in N,N-di-
and 1647 cm do not undergo changes in line position
and minimal changes in line width and amplitude in LP/
DE. There are thus no bonding or through-space interac-
tions between the diene and LP in DE.
methylacetamide.46 As further proof that this is so, the
CO stretching mode of acetone (A) in 18 solvents cor-
relates nicely with the solvents’ E
the mode goes down in energy as the polarity of the
solvent goes up.47
(30) is an empirically derived
T
(30) values; that is,
Benzonitrile (BN) also binds to LP in DE. The CtN
-1
E
T
stretching mode for BN at 2228 cm is systematically
1
1
50
measure of solvent polarity. On the basis of this
correlation and the behavior of CA’s CO mode in LP/DE,
replaced by a new CtN mode at higher energy (∼2255
-1
cm ) as the concentration of LP in DE is increased. This
5
0
1
.00 M LP/DE has a polarity similar to that for CH
3
CN
new mode arises from the BN‚LP adduct. Unlike the
CdO mode for CA in LP/DE, the CtN mode for BN does
not change position and minimally changes line width
in LP/DE (Table 7).51 The CtN mode for BN‚LP, on the
other hand, changes position (to lower energy) and line
and DMF, while 3.00 M LP/DE has a polarity similar to
that of ethanol and formamide.
(
43) It is not clear if Et
take this fact into account in calculating the equilibrium constant.
However, because the concentration of Et O in LP/DE is so much
higher than the concentration of Et O displaced in the reaction, one
can assume that [Et O] is constant and, thus, can be ignored in the
equilibrium expression.
2
O is displaced by CA. If it is, one should
2
2
(48) Our model assumes (1) that the A and A‚LP carbonyl bands do
not change wavelengths and (2) that the spectrum in 1 M LP/DE
largely is due to the complex.
2
-1
(
(
(
44) Brunn, J .; Beck, C. Z. Chem. 1984, 24, 222.
45) Bradley, M. S.; Bratu, C. J . Chem. Educ. 1997, 74, 553.
46) Baron, M. H.; J aeschke, J .; Moravie, R. M.; De Loz e´ , C.; Corset,
(49) MNDO calculations predict a shift of 48 cm for the CdC mode
when CA is complexed to Li+
.
(50) Complexation of Lewis acids to nitriles in solution, solid state,
and crystal results in an adduct having a CtN mode at higher energy
due to shortening of the C-N bond length: Beattie, I. R.; J ones, P. J .
Angew. Chem., Int. Ed. Engl. 1996, 35, 1527. J iao, H. Schleyer, P. v.
R. J . Am. Chem. Soc. 1994, 116, 7429.
J . In Metal-Ligand Interactions in Organic Chemistry and Biochem-
istry; Pullman, B., Goldblum, N., Eds.; Reidel: Dortrecht, 1977; Part
1
; p 171.
47) (a) Pagni, R. M. Unpublished results based on data taken from
ref 11. (b) The correlation is E ) -1.593λmax + 2771 (r ) 0.943). (c) If
one includes PhNH and D O, both of which can donate two H bonds,
in the analysis, the correlation is not quite as good (r ) 0.907).
(
T
(51) Phenylacetylene, the all-carbon analogue of benzonitrile, has
-
1
2
2
a CtC stretching mode at 2110 cm , whose position changes by no
more than 1 cm- on going from DE to 1.50 M LP in DE.
1

Lewis Acidity of Lithium Perchlorate
J . Org. Chem., Vol. 64, No. 7, 1999 2207
F igu r e 2. IR spectra of acetone in LP/DE, carbonyl Region.
Ta ble 7. Cu r ve-F ittin g Resu lts (Lor en tzia n Lin e Sh a p e)
of binding, as measured by equilibrium constants, is
a
for BN in LP /DE
+
relatively small even though Li is a superior Lewis acid.
+
BN band
BN‚LP band
This is a consequence of the facts that Li is ion paired
-
λmaxb
WHHb,c
λmaxb
WHHb,c
to ClO
4
in ether and the probe bases compete for LP.
[LP]0
area
area
Changes in λmax as a function of LP concentration are
consistent with modest increases in solvent polarity as
the medium becomes richer in LP. The IR studies provide
no evidence that the well-established internal pressures
0
0
0
0
0
1
2
2228
2228
2228
2228
2228
2228
2228
11.63
12.26
12.01
12.33
11.79
11.46
11.52
4.67d
3.07
4.32
3.46
3.91
3.37
2.18
.200
.300
.450
.600
.00
2264
2257
2256
2256
2254
2253
29.40
22.21
22.19
20.40
19.24
18.40
4.13
3.63
4.12
5.71
8.42
11.79
12
of LP/DE solutions mimic applied external pressure.
Carbonyl and nitrile stretching modes are blue shifted
.00
5
3
under applied pressure, which is not seen in LP/DE.
a
BN]0 ) 0.285 M. b In cm-1. c WHH ) width at half-height.
BN ) ABN/[BN]0d ) 2100.
-1
[
Shifts of up to 15 cm are seen for ketones under 10 kbar
d
ꢀ
of external pressure, a value numerically similar to the
internal pressure of 3 M LP in DE.
width. It is because of these changes that a genuine
isosbestic point is not observed between the two CtN
bands (Figure 3). Nonetheless, it was still possible to
obtain Keq from the peak areas.
There are even more dramatic data that show that
internal pressure does not mimic applied or external
pressure. It is well-known that the positions of absorption
and emission bands for benzenoid and nonbenzenoid
hydrocarbons are often very sensitive to external pres-
sure. Thus, if the internal pressure of LP/DE mimics
BN + LP h BN‚LP
1
0
To adequately curve-fit the two CtN bands, it was
necessary to include in the procedure two ether peaks
that flanked the two CtN bands (Figure 4). The data
applied pressure, as suggested by Grieco and Ku-
mar,1
2,13
one will be able to predict where the absorption
and fluorescence spectra will occur for the hydrocarbons
in LP/DE if the behavior of the hydrocarbons in solution
under external pressure has been established.
obtained at [LP] ) 0.100 M were excluded from the
0
calculations because the BN‚LP and ether modes are
weak and broad and, as a result, difficult to deconvolute,
and they give anomalous results.52 From the resultant
areas (Table 8), the integrated extinction coefficients for
BN and BN‚LP were derived, and thus Keq (Table 8). One
thus gets a consistent set of Keq values all the way up to
To test this idea the first absorption band of an-
thracene and its fluorescence band and the second
absorption band of azulene and its fluorescence band
were examined in DE as a function of LP concentration.
These data, as well as the predicted band positions, are
recorded in Table 8. Although the emission behavior of
azulene is nicely predicted by the internal pressure
argument, its absorption behavior is not. Furthermore,
the experimental and predicted absorption bands of
anthracene are dramatically different. Experimentally
the 0-0 band shifts 2 nm to the red on going from DE to
5.0 M LP in DE; the predicted shift is 52 nm. Even
though the predicted shifts may be lower than listed in
2
.00 M LP in DE, an extraordinary result (Keq ) 0.71;
log Keq ) -0.147 ( 0.014).
The previously reported NMR studies on CA2 and the
current polarimetry studies on camphor and IR studies
on CA and BN (and acetone) confirm that LP in ether
binds to aldehydes, ketones, and nitriles. Alkenes, on the
other hand, show no evidence of complexation. The extent
2a
(
52) Note, for example, the anomalous λmax(BN‚LP) value at [LP]
1.00 M. Calculation of Keq from the [LP] ) 0.100 M numbers also
afforded a value of Keq out of line with the other Keq values.
0
)
0
(53) Wiederkehr, R. R.; Drickamer, H. G. J . Chem. Phys. 1958, 28,
311.

2
208 J . Org. Chem., Vol. 64, No. 7, 1999
Springer et al.
F igu r e 3. IR spectra of benzonitrile in LP/DE, nitrile stretching region.
Ta ble 9. Absor p tion a n d Em ission Da ta for An th r a cen e
a n d Azu len e in LP /DE
anthracene (0-0 band in nm) azulene (0-0 band in nm)
absorption
emission
absorption
emission
[
LP] expt theorya expt theoryb expt theoryc expt theoryd
0
.0 376
.5 376
.0 376
.0 376
.0 378
.0 378
376
378
381
388
410
428
381
383
380
382
382
382
352
352
352
352
352
352
352
353
354
356
358
360
354
355
355
356
357
357
354
354
354
355
356
357
0
1
2
4
5
a
Based on -203 cm-1 kbar shift in pentane (refs 54 and 55).
c
b
-1
No suitable data available. Based on a -40 cm kbar shift in
pentane (refs 54 and 55). d Based on a -15 cm kbar shift in
-1
F igu r e 4. Curve fitting of two CtN and two ether bands:
solid line) experimental; (9) computer fit; (dashed line)
computer-generated spectra of individual bands.
methyl methacrylate polymer (refs 54 and 56).
(
observed in fact. This demonstrates again that internal
pressure does not mimic applied pressure.
Ta ble 8. Ca lcu la tion of Keq for BN + LP Rea ction a ,b
Finally, let us return to Braun’s and Sauer’s conten-
tion, mentioned earlier in the paper, that changes in the
polarity of LP/DE are responsible for changes in the
diastereoselectivity of Diels-Alder reaction of cyclopen-
at equilibrium
[BN]c
[BN‚LP]d
[LP]a,c
[LP]0
Keq
0
0
0
0
1
2
.100
.300
.450
.600
.00
0.239
0.251
0.201
0.227
0.195
0.126
0.0425
0.0502
0.0570
0.0790
0.117
0.0575
0.250
0.393
0.521
0.883
1.84
(3.1)
0.80
0.72
0.69
0.68
0.70
18
tadiene with methyl acrylate. These results tacitly
assume that this reaction is uncatalyzed, which seems
unlikely on the basis of the results presented in this
paper. Had internal pressure mimicked applied pressure,
the changes in diastereoselectivity could have been
explained as well because the two product-forming reac-
tions have different activation volumes, but this idea has
now been discredited. In an earlier publication, Pagni,
Kabalka, and co-workers provided a more credible and
unique explanation.2 In their model the catalyzed and
uncatalyzed reactions, each giving a different ratio of
products, compete in LP/DE, with the catalyzed reaction
becoming more dominant as the LP concentration goes
up. The competition is a consequence of having a reactive
diene (CP)57 and a small catalytic acceleration, i.e.,
.00
0.163
a
BN]0 ) 0.285 M. b ꢀBN ) 2210 and ꢀBN‚LP ) 9260. These were
[
determined by the method described in the text using all data
except that at [LP]o ) 0.100 M. [BN] ) ABN/ꢀBNd. [BN‚LP] )
ABN‚LP/ꢀBN‚LPd. [LP] ) [LP0] - [BN‚LP].
c
d
e
2b
Table 9 because the shifts under applied pressed are
5
4
solvent dependent and some shifts level off at higher
applied pressure,5 there can be no doubt that the
calculated shifts would still be much larger than those
5a
(
54) Isaacs, N. S. Liquid-Phase High-Pressure Chemistry; Wiley:
Chichester, 1981; Chapter 6.
55) (a) Robertson, W. W.; Weiging, O. E.; Matsen, F. A. J . Mol.
Spectros. 1957, 1, 1. (b) Robertson, W. W.; King, A. D. J . Chem. Phys.
961, 34, 1511.
(
(56) Mitchell, D. J .; Drickamer, H. G.; Schuster, G. B. J . Am. Chem.
Soc. 1977, 99, 7489.
(57) Kiselev, V. D.; Konovalov, A. I. Russ. Chem. Rev. 1989, 58, 230.
1

Lewis Acidity of Lithium Perchlorate
J . Org. Chem., Vol. 64, No. 7, 1999 2209
k
cat/kuncat. For the reaction of CP with dimenthyl fuma-
rate, for example, which fits the model nicely, kobs(cat)/
obs(uncat) ) 2.58.58 Braun’s and Sauer’s data for the MA
The polarimetric data were collected using a Perkin-Elmer
M421 polarimeter. This is an automatic polarimeter with
optical null balance and automatic gain control. Readings were
taken at 21 °C. A continuous sodium discharge lamp was used
as a light source. The instrument was set at an integration
time of 5 s for all readings. A standard glass cell with a path
length of 1 dm and a capacity of 6.2 mL was used for all work.
Ch em ica ls. The chemicals were of the highest quality and
used as received. Methylene camphor was prepared by a
literature procedure.33 Lithium perchlorate obtained from
Aldrich Chemical Co. was dried under a high-vacuum over
k
+
CP reaction, which also fit the Pagni-Kabalka model
well, yield kobs(cat)kobs(uncat) ) 14.59
If one had a Diels-Alder reaction in which the catalytic
accelerating effect is large, product selectivity would be
invariant, provided that polarity and internal pressure
were not important. This may occur by “slowing down”
the uncatalyzed reaction by using unreactive dienes such
as furan and (E)-1,3-pentadiene. The published reaction
of (aromatic) furan with 2,5-dihydrothiophene-3,4-dicar-
boxylic anhydride10 does not occur thermally but does in
LP/DE, yielding a constant ratio of adducts from 1 to 5
M LP in DE. Likewise, the reaction of (E)-1,3-pentadiene
P
°
2
O
5
for 24 h in a drying piston. Ethylene glycol (bp 196-198
C) was the refluxing solvent in the pot. The LiClO was kept
in a desiccator over calcium chloride until it was ready for use.
5
7
4
All manipulations involving the use of LiClO
rapidly under open air conditions.
4
were done
Crotonaldehyde solutions were prepared by first preparing
a 2.85 M stock solution of crotonaldehyde in diethyl ether. This
6
0
with MA is exceptionally slow at room temperature but
in LP/DE is greatly accelerated,61 yielding essentially the
solution was prepared by adding 2.36 mL of crotonaldehyde
3
(
density ) 0.856 g/cm ) by syringe to a 10 mL volumetric flask
same product distribution from 1 to 6 M LP in DE.
and then diluting to the mark with diethyl ether. From this
stock solution 1 mL aliquots were taken via a 1 mL pipet and
then transferred to a 10 mL volumetric flask to which was
subsequently added the appropriate molar amount of LiClO
Diethyl ether was then added to the mark with periodic
shaking. After all the LiClO dissolved, the solutions were then
4
.
4
transferred to 10 mL Wheaton vials, which were then septum
sealed and crimped. The solutions were kept in these vials
until they were ready for FTIR analysis. Other solutions were
prepared in a like manner.
Acqu isition a n d Sp ectr a l Ma n ip u la tion of IR Da ta .
Before filling, the sealed NaCl cells were first cleaned once
with spectroscopic grade acetone and then with anhydrous
diethyl ether. The cell was then dried using a clean inert air
duster (Enviro. Ro. Tech, Duster 1671) from Tech Spray Inc.
The solution to be analyzed was removed by syringe through
the Wheaton vial’s septum seal. After allowing a copious
amount of the solution to flow through the cell, it was capped
and the outside thoroughly cleaned with acetone. This was
necessary in order to remove traces of LiClO
deposited on the cell and its windows.
4
which became
All data were acquired on either a Bio-Rad FTS-60 A or a
FTS-7 Fourier transform infrared spectrometer. Each spec-
trometer is controlled by a computer with an Idris operating
system, which is a UNIX-like operating system, and the FTS
3
200 data station, which is the software installed on the
Winchester hard drive. The software acquires and stores all
the data into files. Data were acquired at a resolution of 2 cm
Con clu sion s
-1
for 64 scans. The IR cell used to measure the absorbancies of
the other solutions was a precision-amalgamated flow through
cell which consisted of a lead spacer between sodium chloride
plates. Luer-Lock fittings with Teflon seals permitted filing
and draining the cell via Luer-Lock syringes. The cell path
length, b, was computed by the interference fringe method for
the empty cell using the following equation:
These studies confirm our earlier published hypothesis
that LP functions as a weak Lewis acid in DE. There is
no evidence that the large internal pressures of LP/DE
mimic applied external pressure.
Exp er im en ta l Section
In str u m en ts. Gas chromatography (GC) was performed
using a Hewlett-Packard 5890 GC instrument equipped with
a 30 m × 1/8 in. column packed with 30% SE-30 by weight on
a Chromosorb W support. Gas chromatography with mass
spectrometry was performed using a Hewlett-Packard 5890
GC with a 5970 mass spectrometer.
1
n
b )
(
)
2
D ν - ν
1
2
where n is the number of fringes (successive maxima or
minima) between two wavenumbers ν and ν and D is the
1
2
Nuclear magnetic resonance (NMR) spectra were obtained
on Bruker AC 250 and 400 MHz spectrometers. Samples were
refractive index of the sample material (1.5443 for NaCl).
The addition of lithium perchlorate to ether solutions of
crotonaldehyde resulted in the appearance of a new band in
the carbonyl stretching region which overlapped with the
uncomplexed band. Similarly, a new band overlapping the
nitrile stretching band appeared when lithium perchlorate was
added to ether solutions of benzonitrile. To curve-fit the bands,
portions of the original files were transferred to MS-DOS
format and imported into DOS software packages on an IBM
dissolved in CDCl
3
with 1% tetramethylsilane (TMS) as the
internal standard (δ ) 0.0 ppm).
(58) The reaction of the very reactive 9,10-dimethylanthracene with
acrylonitrile in LP/DE shows a small catalytic acceleration of 2.2:
Forman, M. A.; Dailey, W. J . Am. Chem. Soc. 1991, 113, 2761. This is
based on the linear plot of kobs versus [LP], where kobs ) kuncat + kcat
[LP].
-
3
86 computer. Additional Bio-Rad software was acquired for
(
59) Pagni, R. M. Unpublished results.
-
1
this purpose. First, a 240 cm interval centered about the
band of interest was extracted from the original data file using
the “zap” programm which creates a temporary truncated
(60) (a) Inukai, T.; Kojima, T. J . Org. Chem. 1967, 32, 869. (b)
Inukai, T.; Kojima, T. J . Org. Chem. 1966, 31, 1121.
61) The catalytic rate acceleration is ∼140.
(

2
210 J . Org. Chem., Vol. 64, No. 7, 1999
Springer et al.
spectrum. To create the temporary file, the following command
was typed at the % command prompt: % Zap-trunc-startHINO-
stopLONO DATAFILE.DT temp, where HINO refers to the
higher wavenumber of the interval, LONO refers to the lower
wavenumber of the interval, DATAFILE.DT is the data file
being truncated, and temp is the name of the temporary
truncated file. A second Bio-Rad program, “dtdumb2”, converts
the temps file into an ASCII text file. A third Bio-Rad program,
MO Ca lcu la tion s. The molecular orbital (MO) calculations
were carried out on a SPARC server 1000 workstation using
MOPAC 6 (QCPE 581). The input geometries were obtained
from molecular mechanics calculations using PCMODEL (Ser-
ena Software), 4th edition.
Absor p tion a n d F lu or escen ce Mea su r em en ts a n d Ca l-
cu la tion s. Absorption spectra were recorded on a Hewlett-
Packard 8452A diode array spectrophotometer, and fluores-
cence spectra were recorded on an Aminco-Bowman Series 2
luminescence spectrometer. Internal pressures of LP/DE were
calculated by the method of Kumar.12 To predict how a band
should shift if internal pressure mimics external pressure, the
“cp2pc”, converts the Bio-Rad text file into a MS-DOS text file
on floppy disk. Once stored on the floppy disk, the file was
imported into a J andel Scientific Software package called
“Peakfit”. In Peakfit, the absorbance bands were first baseline
-
1
corrected and then curve fitted. Several fit options were
available for the data required here; both the Lorentzian and
Pearson routines appeared to give the best correlation coef-
ficients and the lowest standard deviations.
Diels-Ald er Rea ction of (E)-1,3-P en ta d ien e a n d Meth -
4
yl Acr yla te in th e P r esen ce of 6 M LiClO . To a 10 mL
following equation was used: Shift in cm ) (known shift
1
-
-1
under external pressure in cm kbar ) (internal pressure of
LP/DE in kbar - internal pressure of pure ether in kbar). The
reported shifts for the absorption spectra of azulene in pentane
under applied pressure were reported against density of the
solvent.5 To correct these data into spectral shifts versus
kbars of applied pressure, the published data on the density
of pentane versus applied pressure were used.62
5b
flame-dried volumetric flask containing 6.40 g of lithium
perchlorate (60.0 mmol) was added 3 mL of anhydrous diethyl
ether. Methyl acrylate (0.33 mL, 3.7 mmol) was added by
syringe. The volumetric flask was filled with ether to its mark.
The solution was shaken to dissolve the lithium perchlorate,
then transferred via a double-ended needle to a septum-sealed,
flame-dried test tube equipped with a magnetic stirrer. The
mixture was left to stir for 12 h at room temperature. The
solution was then first washed with water (2 × 10 mL), then
neutralized with 5% sodium bicarbonate solution (2 × 10 mL),
and finally concentrated in vacuo. Products were compared to
independently made compounds. Analysis was accomplished
Ack n ow led gm en t. This work was supported by the
Department of Energy and the National Science Foun-
dation. M.C. thanks the University of Tennessee Science
Alliance for summer support. K.W. and J .W. thank the
National Science Foundation for summer support. The
authors thank Professor J akob Wirz for carrying out the
factor analysis. The authors also thank Professor Gary
Schuster for suggesting the absorption and emission
experiments.
1
13
via H and C NMR spectroscopy and GC. GC yield: 40% yield
of ortho-endo; 11% yield of ortho-exo. No meta-endo and meta-
exo adducts were formed.
J O981042X
Reactions with different concentrations of LP in DE were
run similarly.
(62) Danforth, W. E., J r. Phys. Rev. 1931, 38, 1224.