Towards the development of synthetic routes using theoretical calculations: An application of in silico screening to 2,6-dimethylchroman-4-one

Article abstract of DOI:10.3390/molecules15118289

This study describes an attempt to develop a synthetic route using theoretical calculations, i.e., in silico synthesis route development. The KOSP program created four potential synthetic routes for generating 2,6-dimethylchroman-4-one. In silico screenin

Full text of DOI:10.3390/molecules15118289

Molecules 2010, 15, 8289-8304; doi:10.3390/molecules15118289
OPEN ACCESS
molecules
ISSN 1420-3049
www.mdpi.com/journal/molecules
Article
Towards the Development of Synthetic Routes Using
Theoretical Calculations: An Application of In Silico Screening
to 2,6-Dimethylchroman-4-one
Kenji Hori 1,*, Hirotaka Sadatomi 2, Atsuo Miyamoto 3, Takaaki Kuroda 2, Michinori Sumimoto 1
and Hidetoshi Yamamoto 1
1
Department of Material Science, Graduate School of Science and Technology, Yamaguchi
University, Tokiwadai, Ube 755-8611, Japan
2
Transition State Technology, Tokiwadai, Ube 755-8611, Japan; E-Mail: sadatomi@tstcl.jp (H.S.)
3
NARD Institute, Ltd., Nishinagasu, Amagasaki 660-8050, Japan; E-Mail: miyamoto@nard.co.jp (A.M.)
* Author to whom correspondence should be addressed; E-Mail: kenji@yamaguchi-u.ac.jp;
Tel.: +81-836-85-9238; Fax: +81-836-85-9601.
Received: 17 September 2010; in revised form: 9 November 2010 / Accepted: 11 November 2010 /
Published: 15 November 2010
Abstract: This study describes an attempt to develop a synthetic route using theoretical
calculations, i.e., in silico synthesis route development. The KOSP program created four
potential synthetic routes for generating 2,6-dimethylchroman-4-one. In silico screening of
these four synthetic routes was then performed. In silico screening involves theoretical
analysis of synthetic routes prior to actual experimental work. A synthetic route using the
Mitsunobu reaction had already been reported by Hoddgets et al. Theoretical investigations
were also conducted on two SNAr reactions as well as a Michael reaction before they were
examined experimentally. In silico screening using DFT calculations indicated that only
the Michael reaction was likely to produce the target. Experimental work confirmed that
the target was obtained in a yield of 76.4% using the Michael reaction. The other two
routes, except for the Mitsunobu reaction, failed to generate the target. Our results
demonstrate that theoretical calculations can be used to narrow down the number of
experiments that need to be conducted when developing novel synthetic routes.
Keywords: In silico screening; synthetic route development; theoretical calculations;
Mitsunobu reaction; Michael reaction

Molecules 2010, 15
8290
1. Introduction
Synthetic route design systems (SRDSs), such as LHAS [1], EROS [2] and AIPHOS [3], have been
developed that make possible the creation of new synthetic routes for target compounds. The KOSP
(Knowledgebase-Oriented Synthesis Planning) program [4,5], one of the AIPHOS family of programs,
is now commercially available and has been used to create synthetic routes for a large number of
targets. It is, therefore, possible to obtain synthetic routes without having detailed prior knowledge of
the organic synthesis. However, there is still a significant degree of uncertainty using the current
approaches, including the KOSP program.
Theoretical calculations have been a very powerful tool to investigate mechanisms of chemical
reactions. Transition state (TS) searches are the key to understanding reaction mechanisms in detail
[6-8]. Using this approach we have been able to determine transition states (TSs) for existing chemical
reactions. It is, therefore, likely that we can find TSs for synthetic reactions that have never been
examined, i.e., those reactions created by the SRDS. If there is a TS for a synthetic route, then it is
feasible to use this information to design a method for synthesizing the target. However, if exhaustive
efforts do not locate the TS then it is unlikely that the proposed reaction will produce the desired target.
These computational analyses can be performed before starting any experimental work to synthesize
the target compounds.
It is generally difficult to analyze reaction mechanisms from a theoretical perspective for synthetic
routes that have not been examined experimentally, mainly due to a lack of reference data. Moreover,
all possible reactions have to be analyzed, even though some of them are unwanted side reactions. In
particular, it is necessary to check whether side reactions are likely to proceed by comparison with the
main reaction. We have been investigating these computational procedures to analyze synthetic routes,
which are referred to as “in silico screenings” [9-11].
The present paper describes an in silico screening for synthetic routes with 2,6-dimethyl chroman-4-
one (1) as the target compound. Comey originally extracted (S)-1 from the roots of edelweiss
(Leontopodium alpinum) [12]. The KOSP program generated more than twenty possible synthetic
routes, which included using Pd-BINAP, the Mannich reaction of 1-(2-hydroxy-5-
methylphenyl)ethanone with acetaldehyde, and other potential reactions [13]. Because it is difficult to
perform in silico screening for such a large number of potential routes, we chose four such routes, as
described in Scheme 1. Routes A and B use the SNAr mechanism for the reactants with a hydroxyl
group in the side chain and a halogen substituent on the aromatic ring.
Scheme 1. Synthetic Routes for 2,6-Dimethylchroman-4-one from the TOSP/KOSP program.
Route A (SNAr Reaction)
Route C (Mitsunobu Reaction)
O
OH
O
OH
TBAF,
THF
O
O
Ph3P,
DEAD
F
O
3
OH
O
2a
2b
OH
1
NaH,
NaOMe,
MeOH
2,6-dimethylchroman-4-one
DME
Cl
Route B (SNAr Reaction)
Route D (Michael Reaction)
OH
4

Molecules 2010, 15
8291
Route C adopts the Mitsunobu reaction of a reactant 3 with aliphatic and aromatic hydroxyl groups
[14]. The last synthetic route (Route D) uses the Michael reaction of 4 with a double bond in the side
chain. The detailed mechanisms of these reactions will be discussed later. Although Route C had
already been attempted experimentally, to the best of our knowledge the other routes had not been
previously examined either theoretically or experimentally.
2. Results and Discussion
2.1. Routes A and B
In Route A, a fluorine anion from TBAF (tetrabutylammonium fluoride) acts as a base to extract a
proton from the hydroxyl group of 2a to form the alkoxide 5a. Route B uses NaH to extract a proton
from the alkoxyl group of 5b. The target is then produced from the alkoxide under a SNAr mechanism.
Therefore, the reaction mechanism of Route A is almost the same as that of Route B shown in Scheme
2. The KOSP program suggested not only these reactions, but also the experimental conditions from
papers describing similar reactions [15-18].
Scheme 2. Plausible reaction pathways related to producing the target under a SNAr mechanism.
O
O
O
Aldol
Condensation
O
X
OH
H
+
O
5
X
X
2
O
X
OH
O
O
X
Route A: X=F,TBAF
Route B: X=Cl,NaH
a: F
X
b: Cl
1
6
According to the pKa of the hydrogen of the alcohol group compared with the α-hydrogen of the
carbonyl group, 6 is expected to be thermodynamically more stable than 5. DFT calculations
confirmed this expectation, i.e., 6a is more stable by 9.7 kcal mol-1 than 5a, and 6b by 4.1 kcal mol-1
than 5b. Therefore, 6 is the intermediate from the reaction of 2 and the base in both the fluoro and
chloro compounds. Given that 5 has to be generated to form the target, we explored a likely TS
structure of TS6-5 connecting 6 and 5 (Appendix 1). The activation energies (Ea) were calculated to be
more than 30 kcal mol-1 (39.4 kcal mol-1 for X=F and 33.0 kcal mol for X=Cl) as shown in Figure 1.
Based on these calculations no target product will form under the proposed reaction conditions. Thus,
our analysis concluded that Routes A and B are unlikely to generate the desired product.
Although the routes from the KOSP program adopted precursor 5 generated from 2, it is also
possible to produce the ion by use of the aldol condensation of 1-(2-halo-5-methylphenyl)ethanone and
acetaldehyde using lithium diisopropylamide in THF, as also shown in Scheme 2, via an alternative
method. This is a better route than Routes A and B to generate the alkoxide ion 5 because it avoids the
synthesis of 2.

Molecules 2010, 15
8292
The C-O bond distances in TS5-1, where X=F or Cl, were 2.205 and 2.214 Å, respectively, and the
activation energies were 9.1 and 12.1 kcal mol-1, respectively. However, 5 does not form in the
+
reaction mixture because precursor 5 generates a complex 5’ with Li(THF)2 as shown in Figure 1.
Two reactions are considered for 5’: one is the SNAr reaction to form the target, and the other is a
proton transfer from the α-hydrogen atom of the carbonyl group to the alkoxyl oxygen atom to form 6’.
In both reactions, the Li(THF)2+ fragment reduces the nucleophilicity of the alkoxide oxygen. The
alkoxide form 5’ was calculated to be more stable than 6’ with the anionic carbon (5.1 and
2.4 kcal mol-1 for X=F and Cl, respectively) because the coordination of the alkoxide oxygen to the
Li(THF)2+ fragment releases a stabilization energy larger than that of the alcohol oxygen.
Figure 1. Energy correlation diagram (kcal mol-1) and TS structures (lengths in Å) related
to Routes A and B.
TS'5-6
a:36.2
b:33.7
TS'5-1
a:23.7
b:36.2
THF
THF
O
THF
THF
OH
O
Li
H
Li
O
TS"5-2
a:2.8
(2.1)
H
O
6'
2'
a:-1.4
(-0.7)
a:5.1
b:2.4
X
5'
0.0
5"
X
THF
THF
O
0.7
Li
THF
O
O
THF
OH2
1'
Li
O
Li
a:-24.3
b:-39.4
O
O
H
THF
THF
X
X
H
X
We succeeded in locating the TS for the reaction producing target 1’ as shown in Appendix 1. The
C-O bond distances in TS’5-1 for X=F and Cl were calculated to be 1.907 and 2.028 Å, respectively,
and the activation energies to be 23.7 and 36.2 kcal mol-1, respectively. The corresponding barrier
heights increase by 14.6 and 24.1 kcal mol-1 in comparison with those for TS6-5. The coordination of 5
to Li+ greatly reduces the nucleophilicity of the alkoxide oxygen atom. The H-C distance in TS’5-6 was
lengthened by 0.022 Å over that for TS6-5 in both cases. TS’5a-1a is more stable by 12.5 kcal mol-1 than
TS’5a-6a. Moreover, 1a’ is much more stable than 6a’. These results suggest that the target is more
likely to be synthesized using the alternative route. However, according to the calculated barrier
heights the reaction will not take place at temperatures as low as -75 ºC.
Water should be present in the THF solution of lithium diisopropylamide so that a water molecule
may coordinate to Li+ in 5’ to form 5’’ (as shown in Appendix 1). The intermediate 5’’ was calculated
to be slightly less stable (by 0.7 kcal mol-1) compared with 5’. We found TS”5a-2a, the TS structure in
which the water molecule donates its proton to the alkoxide oxygen to form 2’. In this structure, Li-
OH2 and Li-O bond lengths were calculated to be 1.918 and 2.437 Å, respectively. Two O-H bond
lengths were calculated to be 1.151 and 1.219 Å indicating a proton of the OH2 ligand is moving
towards the alkoxide oxygen. The Ea of the process was calculated to be only 2.1 kcal mol-1 and the

Molecules 2010, 15
8293
product 2’ was more stable by 0.7 kcal mol-1 than 5’. It is considered that equilibration between 5’ and
2’ takes place in the reaction mixture and the SNAr reaction of 5a’ does not proceed at a reaction
temperature higher than 0 ºC. Thus, the alternative route also fails to produce target 1.
It was confirmed that the aldol condensation proceeded to form 5’ at -75 ºC because the reaction
yielded 2a, as mentioned in the Experimental section. The calculations, including the counter ion effect,
produced activation energies of more than 23.7 kcal mol-1. This barrier is high enough to prevent 5a’
from proceeding to generate the target at temperatures as low as -75 ºC. Reactions conducted at 0 and
50 ºC also produced 2a, and other reaction mixtures generated no detectable target compound as
anticipated from theoretical considerations.
2.2. Route C using the Mitsunobu Reaction
Schenk and his coworkers investigated the mechanism of the Mitsunobu reaction from a theoretical
perspective [19]. This study used PH3 rather than PPh3, which eliminates any steric effect of the large
substituents for the reaction. However, the steric effect is critical in order to avoid the side reactions of
the α-hydrogen atoms in 3 of Route C. In the present study, we used the 6-31+G(d) level of theory
because of the sizes of the molecules involved in the reaction. According to textbooks, the mechanism
of the Mitsunobu reaction is very similar to Path B [20-23]. However, on the basis of functional groups
in 3, there are two other possible reactions.
Scheme 3. Plausible routes producing the target using Mitsunobu reaction.
O
OH
Ph3P
CO2Et
N N
EtO2C
N N
PPh3
H
EtO2C
CO2Et
OH
7
3
CO2Et
CO2Et
HN
HN
O
OH
O
O
Path A
12
O
O
O
OH
CO2Et
CO2Et
CO2Et
HN
N
HN
N
O
O
Ph3P
CO2Et
14
O PPh3
13
PPh3
PPh3
8(0.0)
10
11
9
PPh3
PPh3
Path B
O
O
O
O
O
O
O
O
12
11
9
OH
OH
1
14
15(not optimized)
17
16
O
OH
9
Path C
18(30.8)
OH
In the Mitsunobu reaction, PPh3 first reacts with diethyl azodicarboxylate (DEAD) to form an
intermediate 7. Figure 2 displays an energy diagram related to this mechanism. Our calculations show
that this reaction has a 8.2 kcal mol-1 barrier and is exothermic by 6.4 kcal mol-1. Almost all of the
molecules in Route C consist of two fragments. The notation such as 8+9 represents the optimized

Molecules 2010, 15
8294
structure in which 8 interacts with 9, in this case via hydrogen bonding. Hereafter, similar notations
will be used for intermediates.
There are three reactive positions in 3. The first is the hydroxyl group attached to the aromatic ring.
Intermediate 7 extracts a proton to form 8 in Path A. The second is the hydroxyl group of the side
chain, and the proton extraction from this group leads to the formation of the alkoxide ion 15 in Path
B. It is necessary to consider the third possibility involving the α-hydrogen atom of the carbonyl group
of the side chain. The loss of one α-proton leads to the formation of 18. All extractions produce 9 as
one of the products. In order to confirm which proton is extracted, geometry optimizations of the three
anions were performed. Although we made an initial geometry of 15, geometry optimization gave 8 as
the most stable structure. Therefore, Path B cannot be practical. It was calculated that 8 is more stable
by 30.8 kcal mol-1 than 18. These results indicate that the reaction of 3 with 7 yields the intermediate
8+9.
Figure 2. Energy correlation diagram (kcal mol-1) of the Mitsunobu reaction of 3. Values
in the diagram are energies relative to that of 3+7. The numbers in parentheses are
activation energies for each path.
TS20-1
22.2(25.0)
TS8-18
29.9
(27.6)
TS7
14.6
(8.2)
TS10-20
14.4
(6.7)
18
TS8-21
12.1
(9.7)
26.1
TS3-8
3.1
19
7.1
TS17-4
0.3
(3.1)
10+11
(13.4)
PPh3 + DEAD
6.4
7.7
21
1.1
8 + 9
2.3
20 + 12
3+7
0.0
-2.8
4+14
-37.3
16 + 11
-7.6
TS17-1
-7.7
(5.4)
17 + 12
-13.1
1+14
-61.6
Three reactions were investigated to determine the likely outcome following the formation of 8 as
shown in Eq. 2. The numbers in the equation indicate energies relative to 3+7 (8+9). TS structures
obtained are shown in Appendix 2. The first is the reaction of the phenol anion extracting one of the α-
hydrogen atoms to form 18 via TS8-18 as shown in Scheme 4. It was calculated that this reaction has to
overcome a barrier of 27.6 kcal mol-1 and 18 is less stable by 23.8 kcal mol-1 than 8+9. The structure of
TS8-18 is shown in Appendix 2. This reaction is not considered any further because the other two
reaction paths discussed below do not require overcoming such a high energy barrier.
The second reaction is between the phenol anion of 8 and the PPh3 fragment of 9 to form 19 with a
hypervalent P atom and an O-P bond. This intermediate was calculated to be less stable by 4.8 kcal
mol-1 than 8+9.

Molecules 2010, 15
8295
Scheme 4. Reaction pathway from 8 to 1 via 19 and to 18.
H
Ph
P
N
Ea=27.6
Ph
Ph
OH
Ph
+
O
N
O
O
9 +
O P N
O
O
O
O
O
Ph
Ph
N
18
OH
H
HO
8
19
9
OH
7.1(4.8)
2.3(0.0)
26.1(23.8)
O
O
H
O
O
Ea=25.0
O
N
+
HN
HN
O
O
P
7.7(5.4)
+
14
O
NH
Ph
O
Ph
O
1
O
Ph
Ph3PO
10+11
12
20
-61.6(-63.9)
-2.8(-5.1)
Before formation of 13 in Scheme 2, the P-N bond has to be cleaved to form 10 and 11, and then
fragment 11 successively changes its position adjacent to the hydroxyl group to form 10+11. Fragment
11 extracts a proton from the hydroxyl group to form fragment 13. However, it was confirmed that the
reaction does not pass through 13, but rather intermediate 20 with an eight membered ring. The H-N
and H-O bond lengths in TS10-20 in Appendix 2 were calculated to be 1.187 and 1.345 Å, respectively
although the O-P bond length is rather long at 3.525 Å. We also found TS20-1 formed 1+14, the target 1
interacting with triphenylphosphine oxide (14). The calculated activation energy was rather high at
25.0 kcal mol-1. This is not a typical Mitsunobu reaction, and the reaction has an energy barrier higher
than that for the third path.
The third reaction to be considered occurs via 16 as shown in Scheme 5. The numbers in the
equation indicate energies relative to 3+7 (8+9). Because 15 is too unstable to be formed from 3, 16
cannot be generated directly from 15. However, an alternative route is feasible from 8+9. Two changes
in geometry occur simultaneously; the oxygen atom of the hydroxyl group of the side chain attacks the
P atom of the PPh3 fragment of 9 and the anionic oxygen atom of 8 extracts the proton from the
hydroxyl group. The O-H, H-O and O-P bond lengths were calculated to be 1.304, 1.123 and 2.448 Å,
respectively. The IRC calculation using TS8-21 indicated that the product of the reaction was not 16 but
21 with the P atom adopting a trigonal bipyramidal geometry. The barrier height for the TS was
calculated to be only 9.7 kcal mol-1.
Due to a strong hydrogen bond between the hydroxyl group on the aromatic ring of 16 and the
anionic nitrogen atom of 11, 16+11 was calculated to be more stable by 8.7 kcal mol-1 than 21.
Therefore, the intermediate is considered to be easily decomposed to 16+11. The succeeding reaction
is an extraction of a phenol proton by 11 to form 17+12. This reaction results in changing the hydrogen
bond from O-H･･N to O･･H-N, and stabilizing the system by 5.5 kcal mol-1. As the barrier height for
TS8-21 is lower by 2.3 kcal mol-1 than that of TS10-20, the reaction via TS8-21 mainly proceeds to form
17+12. The intermediate 17 is a typical intermediate of a Mitsunobu reaction.

Molecules 2010, 15
8296
Scheme 5. Reaction pathway from 8 to 17.
H
O
CO2Et
O
O
HN
N
Ea=9.7
CO2Et
Ph
HN
Ph3P
CO2Et
+
N
O P
OH
8
9
21
O
CO Et
1.1(-1.2)
Ph
Ph
2
2.3(0.0)
CO2Et
CO2Et
N
CO2Et
N
H
O
16 + 11
17 + 12
CO2Et
N
H
N
H
H
O
O
O
Ph
O
P
Ph
P
Ph
O
Ph
Ph
-7.6(-9.9)
-13.1(-15.4)
Ph
The anionic oxygen of 17 attacks the carbon atom adjacent to the OPPh3 moiety in 17 and this
reaction leads to the formation of 1 via TS17-1 as shown in Scheme 6. Values in parentheses are
energies relative to 17. In the TS, the C--O and C--O=PPh3 bond lengths were calculated to be 2.298
and 1.834 Å, respectively, and the central carbon is almost flat, as shown in Appendix 2. Although the
anionic oxygen attacks the tertiary carbon, TS17-1 has typical geometrical features of TSs for SN2
reactions. This reaction accompanies inversion at the carbon atom. Such a change in geometry is
usually seen in the Mitsunobu reaction. The activation barrier was calculated to be only 5.4 kcal mol-1
and a very large heat of reaction (such as 48.5 kcal mL-1).
Scheme 6. Reaction pathway forming the target 1 and the subproduct 4 from 17.
O
H
OH
O
Ea=13.4
Ph
P
(-24.2)
Ph
O
Ph
or
4
17
O
+
O PPh3
O
O
Ea=5.4
O
14
OPPh3
(-48.5)
1
O
It is noteworthy that there is another TS, TS17-4, in which the phenol oxygen anion extracts an α-
hydrogen atom with concomitant loss of the O=PPh3 fragment to form 4. The O-H, H-C and C-O bond
lengths in the TS were calculated to be 1.460, 1.219 and 1.917 Å, respectively. The structure has the
typical features of TSs in geometry for the E2 mechanism. The TS17-4 is less stable by 8.0 kcal mol-1
than the TS17-1
.

Molecules 2010, 15
Scheme 7. Yields of the main and the sub products using Mitsunobu reaction.
8297
O
OH
PPh3 (131 mg, 0.5 mmol)
DEAD (88 mg, 0.5 mmol)
O
O
+
in THF (3 mL) at 0 °C
OH
O
OH
3
1
4
(97 mg, 0.5 mmol)
(71 mg, 81 %)
(7 %)
Hodgetts et al. had already examined this synthetic route and succeeded in producing the target 1 in
good yield (81%). Enone 4 was observed to be a subproduct of the synthetic experiments, although the
energy difference between the two TSs is larger than that expected from the experiments (7%).
Baldwin and coworkers have previously reported this subproduct [24].
2.3. Route D Using the Michael Reaction
The addition of a nucleophile to activated double bonds, as seen in Route D, is referenced in the
Michael reaction. A methoxide ion in dry methanol extracts a phenol proton of 4 to form 22, followed
by the addition of this anion to the double bond to form 23 in Scheme 8. Numbers in the equation
indicate energies relative to 4 (22). The keto-enol equilibrium leads to the formation of 1. To our
knowledge, this is a novel synthetic route to produce the target.
Although only the phenol fragment of 4 has a proton that the base extracts, two more functional
groups exist for the reaction with the methoxide ion. One is the carbonyl group, in which the anion
attacks its carbon to form the tetrahedral intermediate 24. The other is the double bond on the side
chain, and its reaction with the anion produces 25 or 26. Theoretical calculations optimized only one
stable structure 25. In these calculations, two methanol molecules interacting with the lone pair orbitals
of the methoxide oxygen were included, as shown in Appendix 3. As 25 was calculated to be less
stable by 10.5 kcal mol-1 than 22 at the B3LYP/6-31+G(d) level of theory as shown in Figure 3,
extracting a proton from the phenol fragment occurs more readily than addition of the methoxide ion to
the carbonyl carbon.
Scheme 8. Plausible reactions considered for the Michael reaction.
Ea=20.4
O
O
O
O
O
OMe
MeO
MeO
MeOH
MeO
1
OH
O
OH
4
22
-8.6
23
-1.8(6.8)
24
MeO
O
O
O
O
OMe
O
OMe
OH
28
61.3(69.9)
O
(-1.9)
25
26
OH
27

Molecules 2010, 15
8298
There are three feasible reactions for 22 to create a ring. The first one generates 27 with a four
membered ring by a reaction of the carbonyl carbon with the anionic oxygen charge. However, the
small ring is expected to cause this product to be very unstable. Thus, no stable structure was obtained.
Secondly, the double bond reacts with the anion fragment to produce 28 with a five membered ring. It
was calculated that 28 is less stable by 69.9 kcal mol-1 than 22. By contrast, the third reaction generates
23 with a six membered ring, which is the precursor for 1, with a barrier of 20.4 kcal mol-1 and is
endothermic by 6.8 kcal mol-1. A C-O bond is attempting to form in the TS22-23 as shown in Appendix
2 but the distance between the atoms was calculated to be 1.948 Å. Enolate 23 has to extract a proton
from MeOH, and accept it at the carbon adjacent to the carbonyl group to form the target. The low
activation energy, as little as 2.0 kcal mol-1, suggests this process proceeds smoothly. Product 1 is
more stable by 6.6 kcal mol-1 than 23. Therefore, the DFT calculations revealed that the Michael
reaction using 4 is a promising strategy for synthesizing target 1. A synthesis experiment was then
performed in dry MeOH, which gave 1 in an overall yield of 76.4%, albeit as a mixture of enantiomers
(Scheme 9).
Scheme 9. Yields of the target using the Michael reaction.
O
O
28% NaOMe (5.25 g, 27.2 mmol)
in MeOH (50 g) at rt.
OH
O
4
1
(2.2 mg, 12.5 mmol)
(1.68 g, 76.4 %)
Figure 3. Energy correlation diagram (kcal mol-1) and structures related to the Michael
reaction for the target 1.
28
61.3
TS4-25
3.3
(3.3)
TS22-23
9.8
(20.4)
TS23-1
0.2
(2.0)
25
-1.9
4
+ MeO-(MeOH)2
0.0
1
+ MeO-(MeOH)2
-8.4
22
-8.6
23
-1.8
3. Experimental
3.1. Density Functional Theory (DFT) Calculations
Geometry optimizations including TSs were carried out using Density Functional Theory (DFT)
calculations from the Gaussian03 program [25] at the B3LYP/6-31+G(d) level of theory for Routes A,
B and D, or the B3LYP/6-31G(d) level of theory [26-28] for Route C. The GaussView03 [29] program
was used for molecular modeling. The intrinsic reaction coordinates (IRCs) [30] were calculated to

Molecules 2010, 15
8299
analyze all of the mechanisms in detail. The vibration frequencies were calculated to confirm that the
obtained geometries would be stable for the TS structures.
3.2. Synthesis of 1-(2-fluoro-5-methylphenyl)-3-hydroxy-1-butanone (2a)
A solution of 2’-fluoro-5’-methylacetophenone (152.2 mg, 1.0 mmol) in THF (0.5 mL) was added
to a solution of diisopropylamine (111.3 mg, 1.1 mmol) and butyllithium (1.60 M in hexane, 0.69 mL,
1.1 mmol) in THF (1 mL), and then stirred for 1 h at –78 ºC. The mixture was added to a solution of
acetaldehyde (84.0 µL, 1.5 mmol) in THF (0.5 ml), stirred for 2 h at –78 ºC, quenched with saturated
ammonium chloride (5 mL), and then extracted with dichloromethane (10 mL × 5). The combined
extracts were dried over anhydrous magnesium sulfate and concentrated under reduced pressure. The
residue was chromatographed on a silica gel with hexane – ethyl acetate (1/1 v/v) as an eluant to give
1-(2-fluoro-5-methylphenyl)-3-hydroxy-1-butanone (155.6 mg, 79%) as a colorless oil.
3.3. Synthesis of 1-(2-hydroxy-5-methylphenyl)but-2-en-1-one (4)
Anhydrous aluminium chloride (3.92 g, 29.4 mmol) was added in small portions to a solution of p-
methylanisole (3.66 g, 30 mmol) and crotonoyl chloride (3.06 g, 29.3 mmol) in carbon disulfide
(146 mL) at 20~30 ºC. After stirring for 2 h, additional anhydrous aluminium chloride (3.92 g,
29.4 mmol) was added all at once, and the mixture was stirred at 35~40 ºC until the evolution of
hydrogen chloride had ceased (about 3 h). The reaction mixture was then poured slowly into a mixture
of concentrated hydrochloric acid (10 mL) and ice water (150 g). The organic layer was separated,
washed with water and saturated sodium chloride, dried, and then concentrated under reduced
pressure. The residue was chromatographed on a silica gel with hexane–ethyl acetate (10/1 v/v) as an
eluant to give 1-(2-hydroxy-5-methylphenyl) but-2-en-1-one (4.4 g, 83%) as a yellow oil. The spectral
data for this compound matched those previously reported.
3.4. 2,6-dimethylchroman-4-one (1)
A solution of 1-(2-hydroxy-5-methylphenyl)but-2-en-1-one (2.20 g, 12.5 mmol) and 5% aqueous
sodium hydroxide (147 mL) was stirred at room temperature for 16 h. The precipitate was filtered,
washed with water, and dried to give a pale brown solid (2.18 g) in which the unreacted material was
detected by TLC. The precipitate was dissolved in a mixture of 28% sodium methoxide (5.25 g,
27.2 mg) and dry methanol (50 g), and stirred for 5 h at room temperature. The mixture was acidified
to pH 5 with 1 N hydrochloric acid and concentrated under reduced pressure. The residue was
chromatographed on silica gel with hexane–ethyl acetate (10/1 v/v) as an eluant to give 2,6-
dimethylchroman-4-one (1.68 g, 76%) as a yellow solid. The spectral data for this compound matched
those previously reported.
4. Conclusions
In the present study, we have analyzed the reaction mechanisms for four synthetic routes for 2,6-
dimethylchroman-4-one using the B3LYP/6-31+G(d) or B3LYP/6-31G(d) level of theory. Four
reactions were selected from among the 21 synthetic routes created by the program KOSP. The DFT

Molecules 2010, 15
8300
calculations suggested that two routes using the SNAr mechanism would not generate the desired
product. It was also confirmed from theoretical calculations that an alternate route using the aldol
condensation could not produce the target compound. These findings are consistent with our
experimental data which have done after the present calculations.
The theoretical analysis clearly showed that the target should be obtained from Route C. This route
has already been used to synthesize the target with a good yield of 81%. However, the mechanism of
this Mitsunobu reaction has not been clarified from a theoretical standpoint. The DFT calculations
indicated the possibility of generating a byproduct, which was observed in Hodgetts experiments. Our
proposed mechanism explains the formation of both the main product as well as the
observed byproduct.
Theoretical examinations confirmed that Route D using the Michael reaction is also a feasible route
for obtaining the desired target. Our synthesis experiments produced the target with an overall yield of
76.4%, although no attempt was made to fully optimize the reaction conditions.
Scheme 10. Results of in silico synthesis development of the target.
Synthesis Routes
from KOSP
in silico Screening
using DFT Calculations
Experimental
Results
Target
O
Route A: SNAr reaction
Route B: SNAr reaction
Route C: Mitsunobu reaction
Route D: Michael reaction
Other 17 routes
X Route A
X Route B
O Route C
O Route D
X Route A
X Route B
O Route C
O Route D
O
2,6-dimethylchroman-
4-one
Scheme 10 displays the results of “in silico synthesis development” of the target including the
comparison between the results of in silico screenings and synthesis experiments. We performed a
theoretical analysis of the reaction mechanisms for Routes A, B and D, and concluded that only the last
route was worth attempting experimentally. It is important to point out that the KOSP program created
these synthetic routes investigated here, and no experiments had been conducted prior to our
theoretical investigations. The KOSP program also created Route C, which Hodgetts et al. used to
synthesize 1 prior to our calculations. However, the mechanism of this reaction had not been
investigated.
Synthetic organic chemists can create synthetic routes for targets, but cannot give answers to their
feasibility without performing the actual experiments. Theoretical analyses of the reaction mechanisms
for these synthetic routes are considered to be a valuable tool for examining the potential means of
obtaining the desired product. We can use theoretical calculations to narrow down the number of
experiments that have to be conducted (e.g., from potential routes generated using the KOSP program).
Thus, the combination of computational chemistry and cheminformatics offers a new way of helping to
develop novel synthetic routes for target compounds. Nonetheless, real experiments still have to be
performed and optimized in order to create new synthetic schemes for industrial applications.
Acknowledgements
This study was supported by a Grant-In-Aid from Japan Science and Technology Agency (JST,
Project No. 1912).

Molecules 2010, 15
Appendix 1. 3D structures related to Figure 1. Lengths in the Figure are in Å units.
8301
Li-OH2 1.918
C-O
a:1.907
b:2.028
1.151
1.219
Li-O: 2.437
O-C
a:2.205
b:2.214
TS5-1
a:9.1
b:12.1
C-X
a:1.894
b:1.980
TS'5-1
TS"5-2
H-C
a:1.371
b:1.366
O-H
a:1.348
b:1.354
O-H
a:1.371
b:1.376
TS'5-6
a:36.2
b:33.7
TS6-5
a:39.4
b:33.0
H-C
a:1.349
b:1.344
Appendix 2. 3D structures related to Figure 2. Lengths in the Figure are in Å units.
1.345
1.187
1.922
2.486
3.525
1.187
1.482
TS8-18
TS10-20
TS20-1
1.304
1.123
1.917
1.834
2.448
1.219
1.460
2.298
TS8-21
TS17-1
TS17-4

Molecules 2010, 15
8302
Appendix 3. 3D structures related to Figure 3. Lengths in the Figure are in Å units.
1.692
2.052
1.948
1.610
1.595
TS22‐23
1.675
1.915
1.753
1.186
1.765
1.461
TS23‐1
TS4‐25
4+ MeO‐(MeOH)2
References and Notes
1. Corey, J.E.; Cheng, X.-M. The Logic of Chemical Synthesis; John Wiley & Sons: New York, NY,
USA, 1989.
2. Gasteiger, J. Computer-assisted synthesis design. Present state and future perspectives. Chim. Ind.
(Milan) 1982, 64, 714-721.
3. Funatsu, K.; Sasaki, S. “AIPHOS”, Computer Chemistry Series 2; Kyoritsu Shuppan: Tokyo,
Japan, 1994.
4. Satoh, K.; Funatsu, K. A novel approach to retrosynthetic analysis using knowledge bases derived
from reaction databases. J. Chem. Inf. Comput. Sci. 1999, 39, 316-325.
5. AIPHOS/KOSP V1.0 (2003) was commercially available from Fujitsu Limited (Tokyo, Japan).
http://www.fujitsu.com/global/, accessed on 15 November 2010.
6. Hori, K.; Yoshimura, K.; Ohno, H.; Onimura, K.; Ohishi, T. Theoretical study on the
polymerization mechanism of substituted maleimides by using a chiral catalyst with Zn2+.
Tetrahedron 2003, 59, 6301-6309.
7. Hori, K.; Sonoda, T.; Harada, M.; Yamazaki-Nishida, S. Theoretical study on the reactivity of
phenyl cation with a propyl group at ortho-position. Tetrahedron 2000, 56, 1429-1436.
8. Musashi, Y.; Sakaki, S. Theoretical study of rhodium(III)-catalyzed hydrogenation of carbon
dioxide into formic acid. Significant differences in reactivity among rhodium(III), rhodium(I), and
ruthenium(II) complexes. J. Am. Chem. Soc. 2002, 124, 7588-7603.
9. Hori, K.; Sadatomi, H.; Okano, K.; Sumimoto, M.; Miyamoto, A.; Hayashi, S.; Yamamoto. H. An
attempt method for developing new synthetic routes by fusing computational chemistry and
chemoinformatics: Syntheses of ethyl and benzyl methacrylates. J. Comput. Aided Chem. 2007, 8,
12-18.

Molecules 2010, 15
8303
10. Hori, K.; Okano, K.; Yoshimura, K.; Nishida, A.; Yamamoto. H. An attempt judging possibility of
synthesis routes from the TOSP program using DFT calculations. Application to synthesis routes
for substituted fran-2,3-dione. J. Comput. Aided Chem. 2005, 6, 30-36.
11. Hori, K.; Yamaguchi, T.; Okano, K. A system fusing computational and information chemistry for
developing new synthesis routes of compounds. J. Comput. Aided Chem. 2004, 5, 26-34.
12. Comey, N.; Hook, I.; Sheridan, H.; Walsh, J. Isolation of (S)-(-)-2,3-dihydro-2,6-dimethyl-4H-
benzopyran-4-one from roots of Leontopodium alpinum. J. Nat. Prod. 1997, 60, 148-149.
13. Asolkar, R.N.; Kamat, V.P.; Wagner-Dobler, I.; Laatsch, H. Limnazine, the first bacterial azine
derivative from Bacillus sp. GW90a. J. Nat. Prod. 2002, 65, 1664-1666.
14. Hodgetts, K.J. Asymmetric synthesis of (S)-2,6-dimethylchroman-4-one. ARKIVOC 2001, xi,
74-79.
15. Beswick, P.J.; Widdowson, D.A. The synthesis of 3-methylene-2,3-dihydrobenzofuran stabilized
as its tricarbonylchromium complex. Synthesis 1985, 492-493.
16. Benarab, A.; Guillaumet, G. Synthesis and Serotonergic Activity of 8-[4-(2,3-Dihydro-1,4-
dioxino[2,3-b]pyridine-2-yl-methylamino)butyl]-8-azaspiro[4,5]decane-7,9-dione. Heterocycles
1993, 36, 2327-2333.
17. Sanchez, I.; Pujol, M.D. A convenient synthesis of pyrrolo[2,1-c][1,4]benzoxazines. Tetrahedron
1999, 55, 5593-5598.
18. Moukri, M.; Pujol, M.D.; Akss ira, M.; Leger, J.M.; Jarry, C.; Guillaumet, G. Synthesis of new
compounds containing the 2,3-dihydro[1,4]dioxino[2,3-b]pyridine heterocyclic system as a
substructure. Tetrahedron 2003, 59, 3665-3672.
19. Schenk, S.; Weston, J.; Anders, E. Density functional investigation of the Mitsunobu reaction. J.
Am. Chem. Soc. 2005, 127, 12566-12576.
20. Mitsunobu, O. The use of diethyl azodicarboxylate and triphenylphosphine in synthesis and
transformation of natural products. Synthesis, 1981, 1-28.
21. Kato, K.; Mitsunobu, O. Oxidation of mercaptans with diehyl azodicarboxylate and trivalent
phosphorus compounds. J. Org. Chem. 1970, 35, 4227-4229.
22. Mitsunobu, O.; Yamada, M. Preparation of esters of carboxylic and phosphoric acid via
quaternary phosphonium salts. Bull. Chem. Soc. Jpn. 1967, 40, 2380-2382.
23. Kürti, L.; Czakó, B. Stragtegic Application of Named Reactions in Organic Synthesis; Elsevier:
Amsterdam, The Netherlands, 2006.
24. Baldwin, J.E.; Thomas, R.C.; Kruse, L.; Dilberman, L. Rules for ring closure: ring formation by
conjugate addition of oxygen nucleophiles. J. Org. Chem. 1977, 42, 3846-3852.

Molecules 2010, 15
8304
25. Frisch, M.J.; Trucks, G.W.; Schlegel, H.B.; Scuseria, G.E.; Robb, M.A.; Cheeseman, J.R.;
Montgomery, J.A., Jr.; Vreven, T.; Kudin, K.N.; Burant, J.C.; Millam, J.M.; Iyengar, S.S.;
Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G.A.;
Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J.E.; Hratchian, H.P.; Cross, J.B.;
Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R.E.; Yazyev, O.; Austin, A.J.;
Cammi, R.; Pomelli, C.; Ochterski, J.W.; Ayala, P.Y.; Morokuma, K.; Voth, G.A.; Salvador, P.;
Dannenberg, J.J.; Zakrzewski, V.G.; Dapprich, S.; Daniels, A.D.; Strain, M.C.; Farkas, O.;
Malick, D.K.; Rabuck, A.D.; Raghavachari, K.; Foresman, J.B.; Ortiz, J.V.; Cui, Q.; Baboul,
A.G.; Clifford, S.; Cioslowski, J.; Stefanov, B.B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi,
I.; Martin, R.L.; Fox, D.J.; Keith, T.; Al-Laham, M.A.; Peng, C.Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P.M.W.; Johnson, B.; Chen, W.; Wong, M.W.; Gonzalez, C.; Pople, J.A.
Gaussian 03, Revision C.02; Gaussian, Inc.: Wallingford, CT, USA, 2004.
26. Ditchfield, R.; Hehre, W.J.; Pople, J.A. Self-consistent molecular-orbital methods. IX. An
extended Gaussian-type basis for molecular-orbital studies of organic molecules. J. Chem. Phys.
1971, 54, 724-728.
27. Dunning, T.H., Jr.; Hay, P.J. Modern Theoretical Chemistry; Schaefer III, H.F., Ed.; Plenum:
NewYork, NY, USA, 1976; Volume 3.
28. Hay, P.J.; Wadt, W.R. Ab initio effective core potentials for molecular calculations. Potentials for
the transition metal atoms Sc to Hg. J. Chem. Phys. 1985, 82, 270-283.
29. GaussView 3.0 was commercially available from Gaussian Inc. http://www.gaussian.com/,
accessed online 2 October 2003.
30. Fukui, K. The path of chemical reactions - the IRC approach. Acc. Chem. Res. 1981, 14, 363-368.
Sample Availability: Samples of the compounds are available from the authors.
© 2010 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/3.0/).