An Investigation into the Stephens–Castro Synthesis of Dehydrotriaryl[12]annulenes: Factors Influencing the Cyclotrimerization

Article abstract of DOI:10.1002/ejoc.201901053

Detailed investigation into the CuX/base/phosphine modified Stephens–Castro syntheses of dehydrotriaryl[12]annulenes 1–3 have shown that cyclization is suppressed by excess CuX, strong donor ligands, high dilution conditions, and thermally unstable ethyny

Full text of DOI:10.1002/ejoc.201901053

DOI: 10.1002/ejoc.201901053
Full Paper
Cyclotrimerization
An Investigation into the Stephens–Castro Synthesis of
Dehydrotriaryl[12]annulenes: Factors Influencing the
Cyclotrimerization
Paul N. W. Baxter,*[a] Abdelaziz Al Ouahabi,[a] Lydia Karmazin,[b] Alexandre Varnek,[c]
Jean-Marc Strub,[d] and Sarah Cianferani[d]
Abstract: Detailed investigation into the CuX/base/phosphine of 2 and 4. Overall, our findings are consistent with a cyclization
modified Stephens–Castro syntheses of dehydrotriaryl[12]an- pathway governed more by the nature of prior self-association
nulenes 1–3 have shown that cyclization is suppressed by ex- of the ethynylcuprate monomers. Crystallographic characteriza-
cess CuX, strong donor ligands, high dilution conditions, and tion of tetrameric by-product 4, is also reported and revealed
thermally unstable ethynylcuprate monomers. Surprisingly, in- that it assembles into tubular stacks in the solid state.
termediate dimer 16 plays only a minor role in the formation
Introduction
unusual Stephens–Castro-mediated cyclotrimerization reaction
from 2-iodoarylethyne precursors (Scheme 2).[2,4,10,11] It is also
the smallest substructure of a potentially large class of multiply
fused triangular [12]DBA hydrocarbons of increasing intercon-
nectivity and crosslinking, leading ultimately to the polymer
graphyne, a hypothetical expanded carbon network predicted
to possess many intriguing mechanical and physicochemical
properties.[12–17]
Dehydrotribenzo[12]annulene[1] ([12]DBA, 1, Scheme 1) has
grown in popular recognition since its first reported syntheses
in 1966,[2–4] to become a structural classic within the field of
hydrocarbon and macrocyclic chemistry.[5–7]
Scheme 2. Stephens–Castro cyclotrimerization synthesis of 1. (i) CuSO4/
NH2OH·HCl in EtOH/H2O; or CuX (X = Cl, I) and base (e.g. tBuOK, K2CO3) in
N-donor solvents at ambient temperature; (ii) Δ (140–170 °C).
Scheme 1.
However, despite these considerations, 1 remained very
much a laboratory curiosity until the mid-1980s when it was
found to form metal sandwich complexes with potential molec-
ular electronics applications.[18–20] Motivated by renewed inter-
est in carbon allotropes, multiply fused [12]DBAs were also pre-
pared as model systems for the study of graphyne.[12,21–28]
In line with the novel conjugated structure of 1, [12]DBAs
have since been found to function as new types of high spin
magnetic,[4,29,30] 2-photon absorbers,[21,31] liquid crystalline[32]
and vesicular materials,[33] gels,[34] platforms for chiral recogni-
tion[35] and inclusion,[36] as well as affording stacked charge
transfer arrays,[37] novel charge transport materials with poten-
tial for use as organic semiconductors,[21,34,38,39] and porous
networks which absorb CO2 gas.[40,41]
The interest stems from its unique triangular conjugated
architecture,[8,9] and that it can be directly prepared via an
[a] Institut Charles Sadron, UPR 22 (CNRS-UdS)
23 rue du Loess, 67034 Strasbourg, France
E-mail: baxter@unistra.fr
http://www-ics.u-strasbg.fr
[b] Service de Radiocristallographie, Fédération de Chimie “Le Bel” FR2010,
Tour de Chimie,
1 rue Blaise Pascal, 67008 Strasbourg, France
[c] Laboratoire de Chémoinformatique, UMR 7140 CNRS,
Université de Strasbourg,
4 rue Blaise Pascal, 67000 Strasbourg, France
[d] Laboratoire de Spectrométrie de Masse Bio-Organique, Département des
Sciences Analytiques IPHC, UMR 7178 (CNRS-UdS) ECPM,
25 rue Becquerel, 67087 Strasbourg, France
In accordance with the wide ranging interest and potential
that [12]DBAs offer for materials science applications, an ex-
panded range of synthetic methods have been developed for
Supporting information and ORCID(s) from the author(s) for this article are
available on the WWW under https://doi.org/10.1002/ejoc.201901053.
Eur. J. Org. Chem. 2019, 6783–6795
6783
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
the construction of the [12]DBA ring system. However, the cyclization yields using the procedure of Eglinton et al.[2] We
Stephens–Castro cyclotrimerization synthesis of 1 (Scheme 2) therefore carried out a detailed investigation into determining
remained the preferred method for the syntheses of symmetri- the factors necessary for optimizing the Stephens–Castro cy-
cally trisubstituted [12]DBAs,[42–44] until it was superseded by clotrimerization process and then applied the optimized condi-
the adaptation of Iyoda et al.,[25] using K2CO3/CuI/PPh3 in tions discovered for 2 to the attempted generation of isomeric
DMF.[36,37,39,45] Sonogashira–Hagihara conditions[46,47] have also dehydrotripyridyl[12]annulene 3, as well as an improved syn-
been successfully utilized with sensitive substrates.[21,37,48,49] Al- thesis of the tribenzo system 1. In conjunction with this work, a
though of less general utility,[44] the latter approach to [12]DBAs reinvestigation into the catalytic modification of the Stephens–
has been successfully demonstrated under phase transfer con- Castro mediated cyclotrimerization of 6 to 1 (Scheme 2) was
ditions,[26,35,50,51] and in ionic liquids,[52] as well as via the in- also undertaken, in order to throw more light onto the possible
situ generation of the 1-halo-2-ethynylbenzene.[53] Ethyne me- mechanism of the cyclization process. Details of these studies
tathesis cyclotrimerizations have also been successfully em- and resulting conclusions are described below.
ployed for [12]DBA syntheses,[54,55] although this approach is
restricted to symmetric [12]DBAs that do not have regiochemi-
Results and Discussion
cal issues.
Since the pioneering work of Staab et al., using Wittig meth-
odology,[3,4] various alternative intramolecular ring-closing strat-
egies have also been employed using Sonogashira–Hagihara
coupling,[28,32] ethyne metathesis[21,23] and aldehyde coupling
approaches.[22,24] Miscellaneous syntheses of [12]DBAs via
[2+2+2] phenyl-ethyne cycloreversion[56] and a 6-component
one-pot cyclization with o-diiodobenzenes and acetylene
gas,[57] have also been reported, but have not so far found gen-
eral usage.
Generation of 2 by Transition Metal Catalysis
[58]
Our initial synthesis of 2
utilized the classical Stephens–
Castro reaction conditions described for the synthesis of 1, [2,43]
whereby aqueous NH2OH·HCl reduction of CuSO4 in the pres-
ence of the 2-iodoarylethyne directly yields the ethynylcuprate,
which is then heated in pyridine to effect the cyclotrimerization.
However, this procedure only afforded 2 in low (5 %) yields. As
palladium catalytic methods have been reported to be a suc-
cessful alternative to the Stephens–Castro route for the prepara-
tion of [12]DBAs, [37] we investigated this approach for the gen-
eration of 2 (Table 1, entries 1–3). However, with solvents and
temperatures that are so generally successful for the Sonoga-
shira–Hagihara coupling, these conditions were actually detri-
mental to the cyclization of 8, affording either zero or only very
low yields of 2. Furthermore, application of the modified cata-
lytic Stephens–Castro methodology as described by Iyoda et al.,
Overall, the Stephens–Castro methodology as originally re-
ported,[2] still remains one of the most general and reliable
methods of syntheses of regiosymmetric [12]DBAs. In view of
the importance of this reaction, it is therefore rather surprising
that comparatively little is understood about its mechanism,
and how the substrate structure and reaction conditions influ-
ence the cyclotrimerization yield.
In the course of our studies on the generation of coordina-
tion networks with dehydrotripyridyl[12]annulene 2 (Scheme 1
and Scheme 3), progress was severely hampered by the low
[25]
yielded only a trace of 2 (entry 4, Table 1).
One possible reason why the catalytic experiments gave at
best only poor yields of 2 may lie in the instability of the ortho-
iodoethynylpyridine 7 towards N-donor solvents. The light sen-
sitive 7 was found to undergo extensive decomposition to an
as yet unidentified black material upon standing in deoxygen-
ated N-donor solvents for a period of days to weeks. This proc-
ess is accelerated by heating, and suggests that the decomposi-
tion-sequestration rate of 7 is more rapid than coupling under
the experimental conditions investigated.
Effect of [CuI] on the CuX/tBuOK Mediated Stephens–
Castro Cyclotrimerization of 8 to 2
Scheme 3. Stephens–Castro mediated cyclizations to give 2 and 4, and at-
tempted preparation of 3. (i) CuX (X = Cl, I) and base (e.g. tBuOK, K2CO3) in
N-donor solvents at ambient temperature; (ii) Δ (120–170 °C).
In contrast to the instability of 7 in N-donor solvents, the
ethynylcuprate 8 is indefinitely thermally stable under identical
Table 1. Attempted synthesis of 2 from 7 using catalytic conditions.
Entry
Catalytic system[a]
Solvent
Δ
[°C]
Reaction
time
Isolated yield
of 2 [%]
1
2
3
4
Cat. PdCl2(PPh3)2/CuI
Cat. PdCl2dppf/CuI
Cat. PdCl2dppf/CuI
1:8 Et3N/THF
1:20 Et3N/Py
1:5 Et3N/toluene
DMF
20
20
55
140
7 d
6 d
6 d
24 h
0
0
1.5
< 1
[b]
CuI/PPh3/K2CO3
[a] 3–10 mol-% in Pd cat.; [7] = 3 × 10–2 – 2 × 10–1 mol L–1. [b] 1:0.3:0.3:3 stoichiometric ratio of 7/CuI/PPh3/K2CO3; [7] = 1 × 10–1 mol L–1
.
Eur. J. Org. Chem. 2019, 6783–6795 www.eurjoc.org © 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6784

Full Paper
conditions. We therefore reasoned that a more detailed investi- rior to the classical Stephens–Castro conditions, the presence
gation into the Stephens–Castro-mediated cyclotrimerization of a relatively large excess of CuI in the form of CuX halides has
of 8 would afford the best possibility for optimizing the yield a detrimental effect upon the yield of 2.
of 2.
We next investigated the effect of excess tBuOCu on the
Ethynylcuprates are known to be aggregated in the solid formation of 2, in which increasing quantities of tBuOCu were
state via a variety of T-shaped CuI–ethyne, CuI–CuI and ethynes generated in situ in the presence of 8, with the following stoi-
bridging multiple CuI centres, forming a range of polymeric and chiometric ratios: 8/tBuOCu of 1:1 (entries 6–13, Table 2),
cluster structures.[59] We therefore initially reasoned that the 1:2 (entries 14–16, Table 2) and 1:3 (entry 17, Table 2). With the
presence of excess CuI may saturate and break up such poly- exception of entry 12, the yields of 2 in pyridine were signifi-
mers to generate soluble species that are more amenable cycli- cantly improved relative to the previous CuX additive experi-
zation. The fact that Eglinton et al.[2] generated the ethynyl- ments, varying between 21 and 36 %. No correlations were
cuprate under aqueous conditions was also of concern, in that observed between yields of 2 and the nature of the anion (X =
possible presence of a trace of water in 8 may be detrimental Cl, I), the ageing time of 8 for 8/tBuOCu ratios of 1:1, and minor
to cyclization. We therefore investigated the effect of [CuX] on changes in heating time (16–18 h). The presence of PPh3 (entry
a modified Stephens–Castro cyclotrimerization in which tBuOCu 11, Table 2) exerted no significant effect upon the yield of 2
was used as the in-situ metallation reagent under anhydrous compared to the natural yield variation of the reaction per-
conditions.[42] The results of these investigations are shown in formed repeatedly under similar conditions in the absence of
Table 2.
phosphine (entry 6, Table 2). Ultrasonication (entry 9, Table 2)
and dilution (entry 10, Table 2) shifted the yield of 2 to the
lower end of the range (23 and 21 % respectively), while DMF
was revealed to be an inferior solvent for the reaction, reducing
the yield of 2 to 17 % (entry 13, Table 2). The presence of
P(nOct)3 however, completely blocked the formation of 2 (entry
12, Table 2); in parallel to experimental observations discussed
later in the text.
In entries 1–4 of Table 2, a 1:3:1 molar ratio of 7/CuX/tBuOK
was employed, which ensured the generation of a 1:2 molar
ratio of ethynylcuprate 8/CuX, i.e. 8 in the presence of excess
CuX. In all four cases, the yield of 2 was improved (9–19 %)
compared to that of our initial preparation of 2 (5 % yield) using
classical Stephens–Castro conditions,[58] and showed no overall
correlation with respect to changes in anion X (X = Cl, I) and
the ageing time of 8 (entries 1, 2 and 4. Table 2). The most
Further increasing the amount of free tBuOCu in the reaction
dilute reaction (entry 3, Table 2) afforded the highest yield of 2 from molar ratios of 8/tBuOCu of 1:4 (entry 18, Table 2) to 1:5
at this stage in the investigations of 19 %.[60] However, not a (entry 20, Table 2) to 1:9 (entry 21, Table 2), with 3 h ageing
trace of 2 was isolated from the reaction in which the amount times for 8, completely suppresses the formation of 2, mirroring
of CuX was further increased to a 1:5 molar ratio of 8/CuX (entry the effect of excess CuX. However, increasing the ageing time
5, Table 2). Thus, although the tBuOCu cupration route is supe- of 8 from 3 to 20 h, in the experiment employing a 1:4 molar
Table 2. Influence of excess CuI upon the Stephens–Castro mediated cyclization yields of 2, using tBuOCu as the metallating agent.
Entry
7:CuX:
X
Solvent
Ageing time of
8 [h]
Δ
[°C]
Heating time
[h]
Isolated yield
of 2 [%]
tBuOK molar ratio[a]
1
2
3
4
5
6
7
8
1:3:1
1:3:1
1:3:1
1:3:1
1:6:1
1:2:2
1:2:2
1:2:2
Cl
Cl
Cl
I
Cl
Cl
I
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Py
Py
19
7
1
0.25
0.42
3–18
5
3
140
140
140
140
140
140
140
120
140
140
140
140
140
140
140
140
140
140
140
140
140
18
18
18
18
18
16
18
18
17
18
18
18
18
19
16
19
18
18
20
17
18
13
14
19
9–10
0
24–36
26
30
23
21
28
0
17
24
29
33
26
0
Py (dilute)[b]
Py
Py
Py
Py
Py
Py
9
1:2:2
1:2:2
4 h ultrasonicn.[c]
10
11
12
13
14
15
16
17
18
19
20
21
Py (dilute)[d]
2
3
4
5
2
3
16
4
3
1:2:2 + 2eq PPh3
1:2:2 + 2eq P(nOct)3
Py
Py
DMF
Py
Py
Py
Py
Py
Py
Py
Py
1:2:2
1:3:3
1:3:3
1:3:3
1:4:4
1:5:5
1:5:5
1:6:6
1:10:10
20
3
3
24
0
0
[a] General experimental procedures are detailed in the Supporting Information. [b] Cyclization conducted in 45 mL of solvent. [c] After 0.3 h ultrasonication,
a further 5 mL of pyridine added to give total reaction volume of 13 mL, in order to fluidify and homogenize suspension. Ultrasonication performed at a
water bath temp. of 25–30 °C. [d] 7 was added as a solution in 10 mL of pyridine to the tBuOCu in 160 mL of pyridine, to give a total reaction volume of
170 mL.
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6785
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
ratio of 8/tBuOCu (entry 19, Table 2), resulted in an improved optimal for the generation of 2 were also applied to the cycliza-
yield of 2, which was isolated in 24 %. This behavior parallels tion of ethynylcuprate 10. The results of these investigations
the trend observed in the 1:2 mol ratio experiments of are detailed in Table 3.
8/tBuOCu (entries 14–16, Table 2), in which increasing the age-
ing time of 8 from 2–16 h causes an increase in yield of 2 from
24–33 %.
In entries 1 and 2, we investigated the influence of added
phosphines on the yields of 2, employing a 1:1 molar ratio of
8/PR3. In entries 3–8 of Table 3, the reactions were performed
under related conditions in the absence of PPh3, and afforded
similar yields of 2 (26–36 %), showing that the presence of PPh3
had no significant effect upon the yield of 2. In contrast, addi-
tion of P(nOct)3 completely suppressed the cyclotrimerization.
This behavior was identical to that observed for parallel experi-
ments with excess tBuOCu (entries 11 and 12, Table 2) de-
scribed above. For entries 3–8 of Table 3, no correlation was
observed between the yield of 2 and the ageing and heating
times of 8, except in entry 3 of Table 3. In this latter case, the
comparatively short heating time at the lower temperature limit
for cyclization resulted in a slightly lower yield of 2 due to in-
complete consumption of 8. The effect of inverse addition i.e.
the addition of the tBuOCu solution to 7 (entry 8, Table 3) also
had no significant effect upon the yield of 2. In entries 9–11 of
Thus the negative effect of excess CuI (in the form of CuX
and tBuOCu) on the yield of 2 is consistent with breakup or
reorganization of the organometallic species responsible for
subsequent cyclization. The fact that the yield of 2 increases
with the ageing time of 8 in the presence of relatively high
concentrations of tBuOCu, suggests however that a structural
reorganization of associations of 8 in the presence of excess
tBuOCu is occurring, that enables the species necessary for the
successful cyclization to be slowly regenerated over time.[61]
Influence of Reaction Conditions on the Stoichiometric
Stephens–Castro Mediated Cyclotrimerization Yield of
2 and 3
We then investigated the effect of experimental conditions on Table 3 we explored the effect of reaction medium upon the
the yield of 2 employing stoichiometric molar ratios of 2-iodo- yield of 2. 4-tBu-pyridine and benzonitrile were found to be
arylethyne/CuI/base. In an attempt to prepare 3, the conditions marginally deleterious to the yield of 2. However, in 1-methyl-
Table 3. Influence of reaction conditions upon the yields of 1–3, using a 1:1:1:1 equivalence ratio of 2-iodoarylethyne/CuI/base/PR3 when present.
Entry
2-iodoaryl
ethyne
CuI/base/
(PR3)[a]
Solvent
Ethynylcuprate
ageing time [h]
Δ
[°C]
Heating time
[h]
Isolated yield
[%]
1
2
3
4
5
6
7
8
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
9
9
5
5
5
5
5
5
5
5
5
5
CuI/tBuOK/PPh3
CuI/tBuOK/P(nOct)3
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
Py
Py
Py
Py
Py
Py
Py
Py
6
6
120
120
120
120
130
130
130
130
130
130
130
130
130
130
140
150
150
170
140
140
140
140
140
140
140
140
140
140
140
170[i]
13
13
2
18
1
18
20
17
18
18
18
14
14
14
1
1
33 (2)
0 (2)
18
18
18
4
68
4
6
4
19
4
4
26[b](2)
34 (2)
34 (2)
36 (2)
34 (2)
34[c] (2)
27 (2)
27 (2)
8[b] (2)
36 (2)
35 (2)
19 (2)
31 (2)
36 (2)
8[b] (2)
33 (2)
0 (3)
9
4-tBu-Py
PhCN
1-methyl-morpholine
Py
Py
DMF
Py
Py
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
CuI/tBuOK
CuI/tBuOK
[d]
CuI/LiN(iPr)2
[e]
CuI/K2CO3
CuI/K2CO3
[e]
4
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
18
18
18
18
4
4
4
4
4
Py
Py
Py
Py
DMF
Py
12(in p-xylene)[f]
1
12
12
18
18
18
18
18
18
18
18
18
18
[e]
CuI/K2CO3
0 (3)
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK
CuI/tBuOK/PPh3
CuI/tBuOK
50 (1)
61 (1)
38 (1)
58 (1)
51 (1)
53 (1)
62 (1)
53 (1)
48 (1)
0 (1)[j]
Py (dilute)[g]
Py
4
Py/10 %H2O
Py
DMF
DMF
DMF/10 %H2O
DMF
7[h]
4
[e]
CuI/K2CO3
CuI/K2CO3
[e]
4
4
CuI/K2CO3[e]/PPh3
CuI/K2CO3
7[h]
20
[e]
Ag2CO3
[a] See General Experimental Methods. [b] Reaction incomplete. Yield based upon 2-iodoarylethyne consumed. [c] Inverse addition; tBuOCu reaction was
syringed onto a stirred solution of 2-iodoarylethyne in 1 mL of pyridine to give a total volume of 10 mL. [d] The LiN(iPr)2 was syringed into a stirred solution
of CuI in pyridine. [e] A 1:1:0.6 molar ratio of 2-iodoarylethyne/CuI/K2CO3 was employed. [f] After stirring for 18 h, the pyridine was removed from the
ethynylcuprate by distillation under vacuum, and replaced with an equivalent volume of p-xylene under argon. [g] 200 mL total reaction volume. [h] After
stirring for 4 h, deoxygenated water was added and stirring continued for a further 3 h before heating. [i] No observable reaction at 140 °C. [j] ≈ 10 % yield
1,4-bis(2-iodophenyl)buta-1,3-diyne was isolated.
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6786
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
morpholine, the yield of 2 was reduced to 8 %, and the progress metallating system, the reaction conducted in DMF gave a
of the reaction significantly retarded.[62] The use of the alterna- higher yield of 1 compared to that in pyridine (entries 26 and
tive bases LiN(iPr)2 and K2CO3 in pyridine exerted little effect 27, Table 3). This latter finding however was the reverse of the
upon the yield of 2 (entries 12 and 13, Table 3) compared to solvent effect seen for the generation of 2, where pyridine was
the use of tBuOK (cf. for example entry 6, Table 3). However, found to be a superior to DMF for the CuI/K2CO3 metallating
DMF caused a significant lowering in the yield of 2 when K2CO3 system (entries 13 and 14, Table 3). Solvent effects in the
was used as base (entry 14, Table 3).
In entries 15, 16 and 18 of Table 3, the effect of higher reac-
tion temperatures showed no significant deleterious effect
Stephens/Castro mediated cyclotrimerization appear therefore
to be substrate specific.
When in its preferred solvent medium, the nature of the met-
upon the yield of 2, highlighting its thermal stability. In the allating system has little effect upon the yield of 1, which
case of entry 17, the ethynylcuprate 8 was prepared in an iden- reaches a common optimum of 61–62 % (entries 22 and 27,
tical way as in entries 15, 16 and 18 of Table 3, but then the Table 3). Performing the reaction under medium dilution condi-
pyridine removed and the heating stage conducted in o-xylene. tions causes a reduction in the yield of 1 (cf. entries 22 and 23,
As with 1-methylmorpholine, o-xylene appears to be a non- Table 3),[64] as does the presence of water in the thermal cycliza-
solvent for 8, whereas in pyridine, 4-tBu-pyridine and DMF, it is tion step (cf. entries 22/25 and 27/29, Table 3). Addition of PPh3
sparingly soluble, imparting a slight coloration to the surround- to the tBuOCu metallating system exerted little effect upon the
ing medium.
yield of 1 (cf. entries 22 and 24, Table 3), mirroring that seen
for 2 (cf. entries 6 and 11, Table 2). However, the presence of
PPh3 in the CuI/K2CO3/DMF system may be detrimental to the
yield of 1, when added in stoichiometric quantities (cf. entries
27 and 28, Table 3).
The experiments in entries 11 and 17 of Table 3 confirm that
solubilization of 8 is essential for successful thermal generation
of 2. In the case of the CuI/tBuOK systems, it therefore appears
to be the action of pyridine on the sparingly soluble 8 that is
the main factor generating and solubilizing the structural form
By contrast, the use of Ag2CO3 in place of CuI/K2CO3 in DMF
of 8 responsible for assisting the cyclotrimerization step. Fur- completely suppressed the cyclotrimerization (entry 30,
ther support for the conclusion that a soluble form of 8 is the Table 3), again emphasizing the crucial role that the structure
most likely candidate directing the cyclization process was ob- of the CuI oligomeric and/or polymeric aggregates must play in
served for entry 13 of Table 3, which employed CuI/K2CO3 in directing the course of the cyclization.
pyridine as the metallating reagent. This latter reaction ap-
peared as a pale yellow solution throughout the entire ethynyl-
cuprate ageing and initial heating steps, yet the yield of 2 was
similar to that obtained using CuI/tBuOK in pyridine, in which
the bulk of 8 was always present as a precipitated solid (entries
4–8, Table 3).[63]
A Reinvestigation into the Formation of 1 via
the Catalytic Modification of the Stephens–Castro
Mediated Cyclotrimerization
The finding that stoichiometric excesses of copper reagents
were unfavorable to the cyclization of 7 to 2, suggested that
catalytic quantities of CuI may be more advantageous. However,
investigations into this possibility were thwarted by the thermal
instability of ortho-iodoethynylpyridine 7, which decomposed
at a faster rate than its conversion to the more stable ethynyl-
cuprate 8 by the recycled CuI catalyst (entry 4, Table 1). We
therefore decided to reinvestigate the CuI/PPh3/K2CO3 cata-
lyzed Stephens–Castro mediated cyclization of 6 to 1[25] verses
the use of CuI/PPh3/tBuOK, both with and without phosphine,
in order to obtain more insight into the influence of the nature
of the catalyst on the cyclization yield. The results of these stud-
ies are shown in Table 4.
With a method in hand for preparing 2 in acceptable and
reproducible yields, we employed similar reaction conditions for
the attempted synthesis of the isomeric 3 from 9 (Scheme 3).
In this case, the addition of 9 to mixtures of CuI/tBuOK and CuI/
K2CO3 under comparable conditions generated ethynylcuprate
10 in situ, which was observed as a khaki precipitate in the
former reaction (entries 19 and 20, Table 3). However subse-
quent heating and workup of each reaction yielded not a trace
of 3, or any other clearly characterizable by-products.
Thus, the presence of metal coordinating pyridine nitrogens
ortho- to the site of cyclization appears to suppress the cyclotri-
merization of 10 to 3.
As a control, we repeated the exact procedure of Iyoda et
al., and obtained a 59 % yield of 1 (entry 1, Table 4), which
compared favorably with the reported yield of 1 of 55 %.[25] We
then repeated the reaction in the absence of added PPh3 and
under milder conditions, i.e. longer ageing time for the forma-
tion of 6 and lower heating temperature (entry 2, Table 4).
These modifications to the reported conditions resulted in an
increase in the yield of 1 to 64 %. Reducing the amount of CuI
from a molar ratio of 5/CuI of 1:0.3 to 1:0.1 in the phosphine-
Influence of Reaction Conditions on the Stoichiometric
Stephens–Castro Mediated Cyclotrimerization Yield of 1
In order to assess the broader generality of the reaction condi-
tions found optimal for the generation of 2, we also reinvesti-
gated the Stephens–Castro mediated cyclization of ethynyl-
cuprate 6 to 1, the results of which are shown in the lower part
of Table 3.
As in the synthesis of 2, the solvent was found to influence free reaction caused a significant drop in the yield of 1 to 27 %,
the cyclization yield of 1. In the tBuOCu metallating system, the and required a much longer heating time to effect complete
use of pyridine rather than DMF was found to afford superior consumption of 5 (entry 3, Table 4). When the former reaction
yields of 1 (cf. entries 21 and 22, Table 3). For the CuI/K2CO3 was conducted using KHCO3 in place of K2CO3, a comparable
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6787
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Table 4. An exploration of the effect of phosphines and solvent on the yields of 1 using catalytic quantities of CuI.
Entry
Reaction components[a]
(molar ratio)
Solvent
Cuprate ageing
time [h]
Δ
[°C]
Heating
% isolated
yield of 1
time [h][b]
1
5/CuI/PPh3/K2CO3
DMF
DMF
DMF
DMF
DMF
DMF
DMF
Py
0
4
4
4
4
4
4
0
4
4
4
2
2
160
140
140
140
140
140
140
160
140
140
140
140
140
24
26
72[d]
18
26
18
18
24
72
96
96
18
18
59
64
27
60
35
35
21
44
41
23
28
24
23
(1:0.3:0.3:3)
5/CuI/K2CO3
[c]
2
(1:0.3:0.6)
5/CuI/K2CO3
[c]
3
(1:0.1:0.6)
4
5/CuI/KHCO3
(1:0.3:1.2)
5
5/CuI/MeCO2K[e]
(1:0.3:1.2)
[c]
6
5/CuI/P(nBu)3/K2CO3
(1:0.3:0.3:0.6)
5/CuI/P(nOct)3/K2CO3
[c]
7
(1:0.3:0.3:0.6)
5/CuI/PPh3/K2CO3
(1:0.3:0.3:3)
8
[c]
9
5/CuI/K2CO3
Py
(1:0.3:0.6)
10
11
12
13
5/CuI/PPh3/tBuOK
(1:0.3:0.3:1.2)
5/CuI/tBuOK
(1:0.3:1.2)
5/CuI/PPh3/tBuOK
(1:0.3:0.3:1.2)
5/CuI/tBuOK
(1:0.3:1.2)
DMF
DMF
Py
Py
[a] The cupration reagents were generated as described in the General Experimental Methods, as were the phosphine and 2-iodoarylethyne addition proce-
dures, to give total reaction volumes of 10 mL. [b] Time at which 2-iodoarylethyne is completely consumed, as determined by TLC sampling. [c] A 1:0.6 molar
ratio of 2-iodoarylethyne/K2CO3 was employed. [d] Reaction rate much slower at a 10 mol-% catalyst loading. [e] Dried under vacuum at 100 °C directly
before use.
yield of 60 % of 1 was obtained (entry 4, Table 4), whereas the the CuI in the catalytic cyclization process, with DMF as the
use of MeCO2K as base resulted in a yield reduction of 1 to 35 % favored solvent, to give a process that is self-catalytic with re-
(entry 5, Table 4). The use of the more strongly coordinating, spect to CuI alone (i.e. independent of phosphine).
comparatively electron rich phosphines, P(tBu)3 and P(nOct)3
reduced the cyclization yields of 1 to 35 and 21 % respectively
(entries 6 and 7, Table 4), an effect that has been observed in
An Investigation into the Nature of Oligomer Growth
Leading to Cyclization
the acyclic catalytic Stephens–Castro coupling.[65] A reduction
in the yield of 1 also occurred when the cyclization conditions
of Iyoda et al.,[25] were repeated in pyridine, and in the same
solvent under milder conditions in the absence of PPh3 (entries
8 and 9, Table 4). In the latter entry, performed under milder
conditions, the reaction appeared to proceed much more
slowly than in DMF (cf. entry 2, Table 4), requiring a 72 h heat-
ing time in order for all the 5 to be consumed.
In a classical cyclization process, the formation of 2 and 4 would
be expected to proceed via a stepwise process in which mono-
mer 8 first combines with another monomer 8 to afford 16
(Scheme 4).
Intermediate 16 may then react further with a molecule of
8 to yield a trimer oligomer that can intramolecularly cyclize to
2. Dimer 16 may also be expected to intermolecularly couple
with another molecule of 16 to give a tetramer oligomer that
can cyclize to give 4; the latter of which may also be expected
to result from the stepwise reaction of 16 with a further two
molecules of 8 (Scheme 4). Alternatively, in a situation in which
three or four molecules of 8 are juxtapositioned within some
type of supramolecular aggregate or template, then a con-
certed process may be possible in which the appropriate num-
ber of monomer units of 8 may simultaneously cyclize to give
2 and 4, and with little or no intermediacy of 16 and higher
oligomer intermediates.
In contrast, the tBuOCu mediated cyclization with catalytic
CuI conducted both in DMF and pyridine in the presence and
absence of PPh3 afforded 1 in comparatively moderate yields
of 23–28 % (entries 10–13, Table 4). Thus, if a catalytic process
is operative, it is significantly less efficient compared to the re-
actions with carbonate as base.[66] The reaction was also ob-
served to proceed much more slowly in DMF than in pyridine,
a reversal of the solvent rate effect seen for the analogous
CuI/K2CO3 catalyzed cyclizations (cf. entries 2 and 9, Table 4).
As in the stoichiometric reactions described in Table 3, the
effect of the phosphine on the catalytic cyclization yield is neg-
ligible[67] or even detrimental, depending upon its electronic
In order to try to throw more light on the reaction pathway
leading to cyclization, we independently synthesized 16 from
2–
properties. The results in Table 4 show that it is the CO3
/
HCO3– anions that play a major role in stabilizing and recycling 11 and 12 via 13, 14 and 15 as shown in Scheme 4, and
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6788
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
1:1 reaction between 16 and 8 therefore represents a significant
contrast to that of the cyclotrimerization of 8 alone, which
afforded higher yields of 2 with only traces of 4 (Entries 4–8,
15–16, Table 3). These findings suggest that 16 plays at most a
minor role in the cyclotrimerization of 8 to 2, although it may
be involved in the generation of tetramer 4. Although the exact
cyclization mechanism is unknown, the factors enabling the cy-
clotrimerization of 8 to 2 may therefore more probably associ-
ated with the structure of intermolecular associations of 8 and
how they are affected by the presence of structurally different
ethynylcuprates such as 16.
Thermochemical Investigations on Dehydrotriaryl[12]-
annulenes 1 and 2, and Ethynylcuprates 6, 8, 10 and 16
Scheme 4. Synthesis of 16, and its 1:1 stoichiometric reaction with 8 to give
a mixture of 2 and 4, which may form via couplings with respectively one
and two equivalents of 8. (i) PdCl2(PPh3)2/CuI cat./toluene/Et3N; (ii) a) nBuLi/
Et2O/–78 °C; b) I2; (iii) 1 equiv. of. nBu4NF/THF; (iv) a) CuI/tBuOK/py, 1 h;
b) 15/18 h; (v) a) CuI/tBuOK/py, 1 h; b) addn. 7 + 15 to generate 8 + 16 in
situ; c) Δ (135 °C); (vi) Δ (135–150 °C).
In order to determine if differences in thermochemical stabili-
ties of the ethynylcuprate precursors and/or the macrocyclic
products were influencing the cyclization yields, we investi-
gated the thermochemical stability of 1, 2, 6, 8, 10 and 16.
Their thermogravimetric analysis (TGA) decomposition profiles
are shown in Figure 1.
performed a 1:1 stoichiometric reaction of 16/8 under identical
reaction conditions and concentration as Entry 4, Table 3, but
with an elevated cyclization temperature of 135 °C. These con-
ditions were chosen as they fell within the range of those that
previously afforded the highest yields of 2 (Entries 4–8, 15–
16, Table 3), while providing an organocuprate ageing time of
sufficient length to ensure that all internal changes were essen-
tially complete before heating to effect cyclization.
However, instead of the higher yield of 2, which would be
expected of a stepwise reaction involving 16 as an intermedi-
ate, we obtained a much lower than expected yield of 2 and,
although all starting materials were consumed, a much lower
overall conversion to cyclic products (12 % and 6 % yields of
2 and 4 respectively; see Supporting Information). The crude
reaction product also contained an unusually elevated ratio of
4/2, the former of which was normally observed in only trace
amounts starting from monomer 8 alone.
Figure 1. TGA decomposition profiles of 1 and 2 from ambient – 800 °C (left).
TGA decomposition profiles of ethynylcuprates 6, 8, 10 and 16 from ambient
– 800 °C (right). Heating was performed at a rate of 10 °C/min under N2.
The TGA experiments revealed that both 1 and 2 underwent
two main mass loss events upon heating from ambient temper-
ature (≈ 25 °C) to 800 °C. In 2, these two principle mass loss
events were preceded by a ≤ –0.5 % change in mass up to 90 °C
Surprisingly, when 16 alone was heated at 135 °C under corresponding to expulsion of included CH2Cl2. In 1, the first
identical conditions to the 1:1 reaction between 16 and 8 de- mass loss of –1.3 % occurred from 145–240 °C, and in the case
scribed above, it was found to be almost completely unreactive. of 2, the mass decreased by –7 % from 115–225 °C due to initial
A H NMR of the crude reaction product isolated after further partial sublimation of the material out of the crucible.[68] The
1
heating at 150 °C for 12 h, showed it to comprise almost en- second and greatest change in mass of –24 % for 1, and of
tirely of unreacted 15, along with only a trace of 4, although –22 % for 2 occurred from 240–330 °C due to chemical decom-
35 % of the 15 had been consumed. As the expected product position. In 2, this was found by differential scanning calorime-
4 was found to be thermally unstable above 140 °C, it is possi- try (DSC) to be a strongly exothermic reaction involving the
ble that any 4 generated during the course of the reaction was liberation of 275 kJ mol–1 of energy.
decomposing as it was formed at 150 °C. This suggests that the
self-coupling of 16 does not constitute the principle source of
the 4 generated during the 1:1 reaction between 16 and 8 as
it would be unreactive at the reaction temperature of 135 °C.
Thus, although 2 is rather more volatile than 1, they both
exhibit similar thermal stabilities in the solid state, and are com-
pletely stable up to the maximum temperature employed for
the solution cyclization reaction investigations (170 °C, entries
The fact that the less reactive 16 was completely consumed 18 and 30, Table 3). In a parallel experiment, 2 was refluxed in
during the 1:1 reaction between 16 and 8 at 135 °C, suggest pyridine for 3 days in the presence of 3 equivalents of CuI under
that the product cycles 2 and 4 arise from reactions between argon, and subsequently found to have undergone minimal de-
16 and the more reactive 8 (Scheme 4). Thus, the low yield composition,[69] confirming that the product yield losses ob-
of conversion to cyclic products and the suppression of the served during the synthetic investigations were not due to ther-
generation of 2 and enhancement of the yield of 4 in the mal instabilities of 1 and 2 or to CuI mediated decompositions.
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6789
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
The TGA heating profiles of the ethynylcuprates 6, 8, 10 and
“Fine tuning” solvent effects (pyridine vs. DMF) specific to
16 were also recorded, and are shown in Figure 1. Compounds base and 2-iodoarylethyne also appear to be operative. Thus in
6 and 8 were found to undergo an initial explosive reaction at the case of the stoichiometric reactions, optimal yields for 1 are
175 °C (–4 % mass) and 202 °C (–6 % mass) respectively. By obtained with 2-iodoarylethyne/CuI/tBuOK in pyridine and
contrast, 10 underwent a steady slight loss in mass up to 2-iodoarylethyne/CuI/K2CO3 in DMF, and 2-iodoarylethyne/CuI/
135 °C, then decomposed more smoothly over a temperature tBuOK or K2CO3 in pyridine for 2. In the catalytic generation
range of 135–180 °C (–10 % mass). These comparatively small of 1, optimal yields were obtained using 2-iodoarylethyne/CuI/
losses in mass are consistent with some type of thermal polym- K2CO3 in DMF.
2–
erization process. However, at ≥ 420 °C extensive chemical de-
composition occurs with mass losses of 53 %, 57 % and 42 %
for 6, 8 and 10 respectively, the last-mentioned decomposition
of which was still underway at the heating limit of 800 °C. Eth-
ynylcuprate 16 was the most thermally stable of the series stud-
ied. However, it underwent a sudden explosive decomposition
at 235 °C with a 51 % mass loss.
Lastly, the results in Table 4 clearly show that it is the CO3
and HCO3 anions that are responsible for catalytic recycling of
–
the CuI in the cases where CuI/K2CO3 was used as the metallat-
ing reagent, with the result that the reaction is self-catalytic
with respect to CuX alone, i.e. without the need for ancillary
ligands such as triphenylphosphine.
Overall, the ethynylcuprate stabilities increase along the se-
ries: 10 < 6 < 8 < 16, whereas the cyclization yields increased
as follows from the corresponding ethynylcuprates: 10 < 16 <<
8 < 6. Thus within the series studied, the least and most stable
ethynylcuprates 10 and 16 gave the lowest cyclization yields
(0 % 3 and <1 % 4 respectively); a finding that is consistent
with incompatibilities between the reaction cyclization temper-
ature and the macrocycle/ethynylcuprate stability. For example,
10 is clearly of much lower thermal stability than 6 and 8, and
is already undergoing some mass loss and internal chemical
transformations well below 140 °C, which was the temperature
used for the unsuccessful synthesis of 3 (entries 19 and 20,
Table 3). The enhanced thermal instability of 10 may therefore
be a causative factor in its failure to cyclize to give 3. The much
more stable 16 showed little propensity to react in pyridine at
140 °C, whereas melting point studies clearly showed that 4
was already chemically transforming or decomposing at this
temperature (see Supporting Information). This suggests that
the elevated temperatures necessary for the cyclization of 16
result in the decomposition of 4 as it is formed.
Discussion of the Stephens–Castro Mediated
Cyclotrimerization Reaction
Although the reaction mechanism and identity of the organo-
metallic species responsible for the cyclization pathway are
currently unknown, examination of the crystal structures of eth-
ynylcuprates show them to be highly aggregated polymers in
the solid state, in the absence of additional ligands, which may
exist as cyclic organometallic CuI-ethyne bridged oligomers in
the case of [tBuC≡ CCu]6 and [tBuC≡ CCu]8.[59] With additional
ligands such as PR3, the non-polymeric structures [Cu2-
(PPh2Me)4(μ-η1-C≡ CPh)2][71] and [Cu4(PPh3)4(μ3-η1-C≡ CPh)4] [72–74]
are obtained, in the latter of which the ligands are arranged
around a central Cu4-tetrahedron. Ethynylcuprates may also un-
dergo structural rearrangements upon coordination to addi-
tional CuI, either via ethyne-binding[75] or coordination to ortho-
pyridine nitrogens if present.[76] These crystallographic reports
raise the intriguing hypothesis that the formation of the de-
hydrotriaryl[12]annulene and higher cyclic oligomers may result
from preorganized oligonuclear supramolecular aggregates/
clusters that assist in positioning the reactive groups into an
arrangement favorable to thermal cyclization. Whereas excess
CuI, PR3, a pyridine nitrogen ortho- to the coupling site, and
high dilution conditions, as well as the presence of structurally
different ethynylcuprates, would be expected to cause breakup
and rearrangement of such entities into structures unfavorable
for cyclization coupling, i.e. consistent with the observed find-
ings.
Summary of Factors Influencing the Cyclotrimerization
Yields of 1–3
Thus, the above investigations revealed a range of factors that
negatively influence the cyclization reaction which may be
summarized as follows: 1) stoichiometric excesses of CuI; 2) the
replacement of CuI by AgI; 3) the presence of strong donor
ligands, e.g. P(tBu)3 and P(nOct)3, 4) conducting the thermal
cyclization step in non- or poor solvents for the ethynylcuprate
Heating the solid ethynylcuprate generated from the tBuOCu
e.g. p-xylene and 1-methylmorpholine; 5) the presence of H2O metallation in a non-solvent such as p-xylene causes suppres-
in the thermal cyclization step; 6) reaction dilution; 7) hetero- sion of cyclization showing that the cyclization reaction is not
atoms adjacent to the coupling site, which may influence the occurring in the solid state via reactions within the precipitated
thermal instability of the ethynylcuprate,[70] and 8) the presence ethynylcuprate.[77] Furthermore, 1 and 2 were obtained in rea-
of structurally different ethynylcuprates such as intermediate 16 sonable yields from the CuI/K2CO3 reactions performed in pyrid-
in the presence of 8. Overall, the stoichiometric reactions were ine; reactions in which all components remained in solution.
insensitive to the anion (chloride vs. iodide), the ethynylcuprate The organometallic species responsible for the cyclization is
ageing time and duration of thermal cyclization, as long as suf- therefore a solution entity, and thus more likely to be oligomeric
ficient time was afforded for both processes, and that the latter rather than polymeric in nature.[78] No polymeric organic mate-
was conducted within the temperature window of 120–170 °C. rials were identified during reaction workup. Apart from 1 and
PPh3 has a negligible effect on the cyclotrimerization yield or 2, tetramer macrocycles 4 and 17 were the only other products
is even slightly detrimental to the cyclization in the case of the isolated from this reaction, consistent with a mechanism direct-
catalytic reactions.
ing the reaction towards oligomerization rather than polymeri-
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6790 © 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
zation, despite the relatively concentrated reaction condi- affording the highest yields of 2 were conducted in pyridine in
tions.[79]
the absence of triarylphosphines (Table 3).
In [83Cu(Py)4]+, the iodine atoms of each ethynylpyridine 8
are blocked from closest approach to the Cu atoms of each
adjacent 8 by the coordinated pyridine ligands. We therefore
applied two methods for calculating the distance of closest
Cu···I approach. Firstly, the 3-iodopyridine rings of the calcu-
lated DFT model of [83Cu(Py)4]+ were rotated about the pyrid-
ine N/C4 axis in order to orient each iodine closest to the near-
est Cu atom. Secondly, we performed a DFT calculation on
structure [83Cu(Py)]+ in which the three meridionally coordi-
nated pyridines were removed, and the iodines constrained to
lie at closest approach to the nearest Cu atoms during minimi-
zation (Figure 2). From the rotation procedure performed on
[83Cu(Py)4]+, we obtained an average Cu···I distance of closest
approach of 2.95 Å. The DFT calculation on structure
[83Cu(Py)]+, afforded a Cu···I closest approach distance of
2.63 Å. The sum of the Van der Waals radii of atoms of Cu
(2.00 Å) and iodine (2.20 Å) is 4.20 Å, and the sum of the ionic
radii of Cu+ (0.46 Å) and iodide (2.20 Å) is 2.66 Å.[82] The latter
value being very close to the average Cu-I bond length of
2.641 Å in the X-ray crystal structure of [Cu2(μ-I)2Py4] (Support-
ing Information). The shortest Cu···I distances calculated for
both [83Cu(Py)4]+ and [83Cu(Py)]+ are therefore within Van der
Waals contact, and in the latter calculation, similar to the sum
of ionic radii. Therefore from a theoretical perspective, the most
proximal copper and iodine atoms in the cyclic tri-ethynyl-
copper(I) core of [83Cu(Py)4]+ and [83Cu(Py)]+ do indeed lie in
close enough proximity for possible chemical reaction with ex-
pulsion of CuI.
In the case of the catalytic reactions resulting in 1, in which
the CuI is recycled by the CO32– and HCO3– anions, it cannot be
ruled out that these anions may be incorporated into and stabi-
lize the organometallic species involved in the cyclization proc-
ess.[80]
Theoretical Investigations
Based upon the known bonding modes of ethynylcuprates dis-
cussed above, we reasoned that if an aggregated solution entity
is responsible for holding the ethynylcuprate monomers into an
arrangement preorganized for cyclotrimerization, it would most
likely incorporate three ethynylcopper(I) core units in close
proximity.
Indeed, experimental evidence for such a core unit is to be
found in the X-ray crystal structure of the tetranuclear cation
[Cu4{P(p-MeC6H4)3}4(μ3-η1,η1,η2-C≡ CR)3]+ (where R = p-MeO-
C6H4), which comprises a trimeric [Cu3{P(p-MeC6H4)3}3(μ2-η1,η2-
C≡ CR)3] ring capped on one face by a [Cu{P(p-MeC6H4)3}]+
unit.[81] This prompted us to perform DFT theoretical modeling
on [Cu4(C5H5N)4(μ3-η1,η1,η2-4-C≡ C,3-I-C5H3N)3]+, i.e. [83Cu(Py)4]+
(Figure 2), a hypothetical analogue of the abovementioned
tetranuclear cationic complex in which the p-MeOC6H4-C≡ CCu
and triarylphosphine ligands were replaced by respectively 8,
the precursor of 2, and coordinating pyridines, as the reactions
Pertinent to these investigations is the recently reported
Stephens–Castro mediated cyclotrimerizations of p-MeO2CPh-
difunctionalized analogues of 6 possessing; 1) an ethynylcopper
substituent as in 6, and; 2) an elongated p-ethynylphenyleth-
ynylcopper substituent in place of the shorter ethynylcopper
substituent of 6, which afforded a 32 % yield of dehydrotri-
aryl[12]annulene product in the former, and a 6 % yield of cy-
clotrimer macrocyclic product in the latter reaction.[40] These
comparative cyclotrimerization reactions differ only in the
length of the ethynylcopper substituent, yet result in a signifi-
cant yield disparity. If preorganization into an entity such as
for example one incorporating a tri-ethynylcopper(I) core unit
discussed above, is operative in these reactions, the copper(I)
and iodine reactive sites in the latter reaction would be posi-
tioned at a distance that is out of reach for an intra-aggregate
cyclization, due to the increased length of the p-ethynylphenyl-
ethynylcopper substituents.
The feasibility that preorganized oligonuclear ethynylcuprate
structures/aggregates may direct the Stephens–Castro reaction
towards macrocyclization rather than polymerization, is there-
fore supported by theoretical investigations and by experimen-
tal yield differences such as that in the example discussed
above.[40] Future experimental investigations into the Stephens/
Castro mediated cyclization of oligonulear o-iodoarylethynyl-
cuprates based upon [Cu4{P(p-MeC6H4)3}4(μ3-η1,η1,η2-C≡ CR)3]+
(where R = o-IC6H4) for example, would therefore be of interest
in order to throw more light on these considerations.
Figure 2. Density Functional Theory (DFT) molecular modeling at the B3LYP
level of theory (employing the 6-31G* and LANL2DZ>Kr basis sets) of
[83Cu(Py)4]+ and [83Cu(Py)]+. Upper left: [83Cu(Py)4]+; view down axis of the
capping [Cu(py)]+. Upper right: lateral view of [83Cu(Py)4]+. Lower left: Ventral
view of [83Cu(Py)]+ with Cu and I atoms constrained at closest approach.
Lower right: Schematic of computed model [83Cu(Py)]+ (pyridine removed
for clarity) showing cuprophilic and Cu-ethyne bonding, where d = 2.63 Å.
Computed model color scheme C (brown), H (white), N (blue), Cu (red),
I (purple).
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6791
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
X-Ray Crystal Structure of Dehydrotetrapyrido[16]annulene 4
This situation would normally be expected to be repulsive
and counterproductive to columnar stacking. Although the
stacking of dehydrobenzo[n]annulenes and related macrocycles
into tubular columnar arrays has been reported, they comprise
entirely hydrocarbon structures with no potentially repulsive
intermolecular effects.[85] In 4 however, the overlying pyridine
rings are slipped and offset with respect to each other such
that only edge-to-edge π–π contacts exist between the inner
edge C1 and C5 and outer edge C3 and N1 atoms of each ring.
This gives an infinite stepped ladder arrangement of inter-
pyridine π–π interactions, with C1–C3 and C5–N1 distances of
3.525 and 3.529 Å respectively,[86] that mitigates any repulsive
dipole alignments that would arise if the center of the pyridine
rings were more closely aligned.
Careful chromatographic purification of the combined yields of
2 and the following eluates yielded small quantities of a prod-
uct that X-ray crystal structural analysis confirmed to be that of
the macrocyclic tetramer 4 (Figure 3).
Conclusion
Overall, our findings suggest that the expected stepwise reac-
tions of molecules of ethynylcuprate 8 via intermediate 16 is
not the principal reaction pathway leading to the formation of
trimer 2 and tetramer macrocycle 4. The dominant cyclization
process must therefore involve a more direct route from 8 to 2
and 4, which is suppressed by the presence of strong donor
ligands such as P(nOct)3, excess CuI and high dilution condi-
tions. These findings, coupled with a cyclization yield depend-
ency upon the structural reorganization of the ethynylcuprate
observed for the 1:4, 8/tBuOCu reaction of entry 19, Table 2,
is consistent with a pathway involving self-association of the
ethynylcuprate monomers into structures preorganised for cy-
clization, and compatible with the known structural bonding
modes reported for ethynylcuprates.
On the other hand, PPh3 was found to have a negligible
effect upon the yields of 2, and 1 in both stoichiometric and
catalytic modification of the Stephen–Castro reaction, contrary
to the findings of Iyoda and Miura et al.[25,65] Our results there-
fore point to the CO32– anion, and not PPh3, as the agent prima-
rily responsible for stabilizing the catalytic species and recycling
the CuI.[87]
TGA studies revealed that 1 and 2 were stable to the temper-
atures employed for the cyclization reactions showing that yield
variations were not originating from simple thermal decomposi-
tion of these products. Ethynylcuprate 10 was less stable than
6 and 8, suggesting that this may be a contributory factor to
the failed generation of 3. Tetrameric by-product 4 was success-
fully isolated and crystallographically characterized, and shown
to crystallize into a porous solid comprising tubular columns
with parallel alignment.
Although the exact nature of the cyclotrimerization mecha-
nism to afford dehydrotriaryl[12]annulenes remains unclear, our
results point towards a process in which the nature of monomer
ethynylcuprate aggregation in solution is influencing the suc-
cess or failure of cyclization. A conclusion further supported by
published X-ray crystallographic evidence and theoretical calcu-
lations described herein.
Figure 3. Space-filling representation of a single columnar stack of molecules
of 4 that propagate along the c-axis; left. Space-filling representation of crys-
tal packing of 4 in a–b plane; right.
Cycle 4 is a non-planar molecule which is folded along the
axes that pass through the center of the cavity and bisect the
pyridines on either side, to afford a V-shaped pocket of suitable
dimensions for the inclusion of simple gases, metal ions and
aromatic rings. The intramolecular N–N distances of 4 are
8.827 Å between adjacent pyridines, and 9.860 Å between pyr-
idines lying on opposite sides of the cavity along the axes of
folding. The diameter of the hole within the V-shaped pocket
is 1.94 × 2.45 Å, as defined by the inner Van der Waals surfaces.
The cycle shows some degree of crystal packing-induced ring
strain, as evidenced by the slight curvature of the alkynes to-
wards the interior of the cavity as viewed down the c-axis. Fur-
ther distortions are apparent when the macrocycle is viewed
from its side along the a- and b-axes, whereby pyridines posi-
tioned on opposite sides of the cycle are not in complete align-
ment as expected but slightly twisted relative to each other.
However, the tetrameric structure of 4 is essentially identical to
that of the hydrocarbon analogue 17 (Scheme 1), reported ear-
lier by Youngs and co-workers.[83]
Depending upon the solvent of crystallization, 17 was iso-
lated as two crystal polymorphs, which both comprised inter-
locked networks of cycles held together by various edge-to-
face and face-to-face aromatic π–π interactions. In contrast, the
molecules of 4 are stacked on top of each other into infinite
tubular columns with intermolecular N–N distances of 4.664 Å
within each column. The columns in turn are packed into infi-
nite parallel arrays which contact each other by inter-columnar
N1–H2 hydrogen bonds of 2.710 Å in length.[84] Each macro-
cycle makes eight such hydrogen bonds with its nearest neigh-
bors, resulting in a comparatively large number of such con-
tacts that must play a significant role in the overall stabilization
of the columnar packing. No solvent of crystallization is in-
cluded in any part of the crystal lattice. Furthermore, all the
pyridine nitrogens within each columnar stack directly overlay
In the light of the continually expanding interest in de-
each other, with the result that the pyridine dipoles point in the hydrotriaryl[12]annulenes, further investigations into the verac-
same direction along each of the four edges of every column. ity of our hypothesis of ethynylcuprate preorganization will lead
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6792
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[6] H. Hopf, Classics in Hydrocarbon Chemistry: Syntheses, Concepts, Perspec-
tives, Wiley-VCH, Weinheim, 2000, pp. 226–227.
[7] M. M. Haley, J. J. Pak, S. C. Brand in Topics in Current Chemistry: Carbon
Rich Compounds II (Ed.: A. De Meijere), Springer-Verlag, 1999, vol. 201,
pp. 81–130.
to an increased understanding of the Stephens–Castro reaction
and contribute to the further development of dehydrotri-
aryl[12]annulenes and their higher cyclic oligomers as a well-
spring of new materials.
[8] J. Jusélius, D. Sundholm, Phys. Chem. Chem. Phys. 2001, 3, 2433–2437.
[9] A. J. Matzger, K. P. C. Vollhardt, Tetrahedron Lett. 1998, 39, 6791–6794.
[10] T. A. Schaub, M. Kivala in Metal-Catalyzed Cross-Coupling Reactions and
More (Eds.: A. de Meijere, S. Bräse, M. Oestreich), Wiley-VCH, Weinheim,
2014, Ch. 9, pp. 665–667.
[11] R. D. Stephens, C. E. Castro, J. Org. Chem. 1963, 28, 3313–3315.
[12] K. Tahara, T. Yoshimura, M. Sonoda, Y. Tobe, R. V. Williams, J. Org. Chem.
2007, 72, 1437–1442.
[13] N. Narita, S. Nagai, S. Suzuki, Phys. Rev. B 2001, 64, 245408–1–245408–7.
[14] N. Narita, S. Nagai, S. Suzuki, K. Nakao, Phys. Rev. B 2000, 62, 11146–
11151.
[15] U. H. F. Bunz, Y. Rubin, Y. Tobe, Chem. Soc. Rev. 1999, 28, 107–119.
[16] N. Narita, S. Nagai, S. Suzuki, K. Nakao, Phys. Rev. B 1998, 58, 11009–
11014.
Supporting Information. General experimental methods, General
procedures for the syntheses of 1 and 2 and attempted synthesis
of 3, isolation and purification of 4, preparation and characteriza-
tion of 6, 8 and 10 for the thermochemical and TEM studies. Physi-
cal Methods: UV/Vis, Luminescence Spectroscopic Infrared and TEM
experimental, 1H NMR spectra of 1 and 2, and 1H and 13C NMR
spectra of 4 with peak assignments, General X-ray experimental,
Crystallographic data, refinements and ORTEP representations for 4
and [Cu2(μ-I)2Py4] with atom numbering, Calculated Structure Carte-
sian Coordinates of [83Cu(Py)4]+ and [83Cu(Py)]+, and references
(continued).
CCDC 1951505 (for 4), 1951506 {for [Cu2(μ-I)2Py4]} contain the sup-
plementary crystallographic data for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic Data
Centre.
[17] R. H. Baughman, H. Eckhardt, M. Kertesz, J. Chem. Phys. 1987, 87, 6687–
6699.
[18] W. J. Youngs, C. A. Tessier, J. D. Bradshaw, Chem. Rev. 1999, 99, 3153–
3180.
[19] J. D. Ferrara, A. A. Tanaka, C. Fierro, C. A. Tessier-Youngs, W. J. Youngs,
Organometallics 1989, 8, 2089–2098.
[20] J. D. Ferrara, C. A. Tessier-Youngs, W. J. Youngs, J. Am. Chem. Soc. 1985,
107, 6719–6721.
Acknowledgments
The Centre National de la Recherche Scientifique and the Insti-
tut Charles Sadron is acknowledged for financial support
(P. N. W. B), and the Ministère de l′Enseignement Supérieur et
de la Recherche for a PhD fellowship (A. A.). M. Legros and
C. Saettel of the ICS are thanked for recording the TGA, and
Dr B. Heinrich of the IPCMS, Strasbourg, is thanked for the DSC
measurement, and Dr L. Allouche and B. Vincent of the Service
RMN, GDS 3648-CNRS-UdS, are thanked for recording the 13C
NMR of 4, and Dr E. Wasielewski of the Cronenbourg NMR Core
Facility, UMR 7042, for the 2-D correlation NMR of 13-15.
Noémie Bourgeois of the Service d'analyses, de mesures phy-
siques et de spectroscopie optique, Fédération de Chimie "Le
Bel" - FR 2010 (Unistra - CNRS)
[21] K. Tahara, Y. Yamamoto, D. E. Gross, H. Kozuma, Y. Arikuma, K. Ohta, Y.
Koizumi, Y. Gao, Y. Shimizu, S. Seki, K. Kamada, J. S. Moore, Y. Tobe, Chem.
Eur. J. 2013, 19, 11251–11260.
[22] K. Tahara, T. Yoshimura, M. Ohno, M. Sonoda, Y. Tobe, Chem. Lett. 2007,
36, 838–839.
[23] C. A. Johnson II, Y. Lu, M. M. Haley, Org. Lett. 2007, 9, 3725–3728.
[24] T. Yoshimura, A. Inaba, M. Sonoda, K. Tahara, Y. Tobe, R. V. Williams, Org.
Lett. 2006, 8, 2933–2936.
[25] M. Iyoda, S. Sirinintasak, Y. Nishiyama, A. Vorasingha, F. Sultana, K. Nakao,
Y. Kuwatani, H. Matsuyama, M. Yoshida, Y. Miyake, Synthesis 2004, 1527–
1531.
[26] M. Sonoda, Y. Sakai, T. Yoshimura, Y. Tobe, K. Kamada, Chem. Lett. 2004,
33, 972–973.
[27] O. Š. Miljanić, K. P. C. Vollhardt, G. D. Whitener, Synlett 2003, 29–34.
[28] J. M. Kehoe, J. H. Kiley, J. J. English, C. A. Johnson, R. C. Petersen, M. M.
Haley, Org. Lett. 2000, 2, 969–972.
[29] H. Nishide, M. Takahashi, J. Takashima, Y.-J. Pu, E. Tsuchida, J. Org. Chem.
1999, 64, 7375–7380.
[30] Y. J. Pu, M. Takahashi, E. Tsuchida, H. Nishide, Chem. Lett. 1999, 28, 161–
162.
is thanked for the elemental analyses, and Corinne Bailly of the
Service de Radiocristallographie, Fédération de Chimie “Le Bel” -
FR2010, for the Cambridge Crystallographic Database searches.
Professor J. D. Wallace of Nottingham Trent University, G. Britain,
is kindly thanked for insightfull discussions.
[31] K. Kamada, L. Antonov, S. Yamada, K. Ohta, T. Yoshimura, K. Tahara, A.
Inaba, M. Sonoda, Y. Tobe, ChemPhysChem 2007, 8, 2671–2677.
[32] S. H. Seo, T. V. Jones, H. Seyler, J. O. Peters, T. H. Kim, J. Y. Chang, G. N.
Tew, J. Am. Chem. Soc. 2006, 128, 9264–9265.
Keywords: Alkynes · Annulenes · Cross-coupling ·
Cuprates · Macrocycles
[33] S. H. Seo, J. Y. Chang, G. N. Tew, Angew. Chem. Int. Ed. 2006, 45, 7526–
7530; Angew. Chem. 2006, 118, 7688–7692.
[34] H. Shigemitsu, I. Hisaki, E. Kometani, D. Yasumiya, Y. Sakamoto, K. Osaka,
T. S. Thakur, A. Saeki, S. Seki, F. Kimura, T. Kimura, N. Tohnai, M. Miyata,
Chem. Eur. J. 2013, 19, 15366–15377.
[35] N. Saito, R. Terakawa, M. Yamaguchi, Chem. Eur. J. 2014, 20, 5601–5607.
[36] I. Hisaki, H. Senga, H. Shigemitsu, N. Tohnai, M. Miyata, Chem. Eur. J.
2011, 17, 14348–14353.
[37] K. Tahara, T. Fujita, M. Sonoda, M. Shiro, Y. Tobe, J. Am. Chem. Soc. 2008,
130, 14339–14345.
[38] M. Kiguchi, K. Tahara, Y. Takahashi, K. Hasui, Y. Tobe, Chem. Lett. 2010,
39, 788–789.
[39] I. Hisaki, Y. Sakamoto, H. Shigemitsu, N. Tohnai, M. Miyata, S. Seki, A.
Saeki, S. Tagawa, Chem. Eur. J. 2008, 14, 4178–4187.
[40] I. Hisaki, S. Nakagawa, N. Ikenaka, Y. Imamura, M. Katouda, M. Tashiro, H.
Tsuchida, T. Ogoshi, H. Sato, N. Tohnai, M. Miyata, J. Am. Chem. Soc. 2016,
138, 6617–6628.
[1] Named variously in the literature as tribenzocyclyne and tribenzocyclo-
triyne (TBC), benzo[12]annulene, dehydrobenzo[12]annulene and tris-
dehydrotribenzo[12]annulene ([12]DBA), tribenzohexadehydro[12]an-
nulene, hexadehydrotribenzo[12]annulene, hexadehydrotribenzo[a,e,i]-
[12]annulene, 5,6,11,12,17,18-hexadehydrotribenzo[a,e,i]cyclododecene
(TBC), 1,2:5,6:9,10-tribenzocyclododeca-1,5,9-triene-3,7,11-triyne and
1,2:5,6:9,10-tribenzo-3,7,11-tridehydro[12]annulene. For the sake of brev-
ity and due to frequent use in the literature, this class of compound is
herein referred to as dehydrotribenzo[12]annulene or the abbreviation
[12]DBA.
[2] I. D. Campbell, G. Eglinton, W. Henderson, R. A. Raphael, J. Chem. Soc.,
Chem. Commun. 1966, 4, 87–89.
[3] H. A. Staab, F. Graf, Tetrahedron Lett. 1966, 7, 751–757.
[4] H. A. Staab, F. Graf, Chem. Ber. 1970, 103, 1107–1118.
[5] Y. Tobe, M. Sonoda in Modern Cyclophane Chemistry (Eds.: H. Hopf, R.
Gleiter), Wiley-VCH, Weinheim, 2004, Ch. 1, pp. 1–40.
[41] I. Hisaki, S. Nakagawa, Y. Suzuki, N. Tohnai, Chem. Lett. 2018, 47, 1143–
1146.
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6793
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[42] D. Zhang, C. A. Tessier, W. J. Youngs, Chem. Mater. 1999, 11, 3050–3057.
[43] D. Solooki, J. D. Ferrara, D. Malaba, J. D. Bradshaw, C. A. Tessier, W. J.
Youngs, Inorg. Synth. 1997, 31, 122–128.
[44] J. D. Kinder, C. A. Tessier, W. J. Youngs, Synlett 1993, 149–150.
[45] K. Tahara, S. Furukawa, H. Uji-i, T. Uchino, T. Ichikawa, J. Zhang, W. Mam-
douh, M. Sonoda, F. C. De Schryver, S. De Feyter, Y. Tobe, J. Am. Chem.
Soc. 2006, 128, 16613–16625.
[46] S. Takahashi, Y. Kuroyama, K. Sonogashira, N. Hagihara, Synthesis 1980,
627–630.
[47] K. Sonogashira, Y. Tohda, N. Hagihara, Tetrahedron Lett. 1975, 16, 4467–
4470.
Table 4). However these small yield differences (3 % and 1 % respec-
tively) may lie within the natural yield variation of each reaction, so
that a slight positive phosphine effect cannot yet be unambiguously
ascertained for these reactions.
[68]
[69]
The onset and termination of the two principal mass loss processes oc-
curred 15–20 °C lower when the TGA heating was performed at 2 °C/min,
compared to 10 °C/min. The TGA mass loss profiles were independent
of whether argon or nitrogen was employed as the carrier gas.
As determined by 1H NMR integrations of the macrocycle proton reso-
nances compared to that of the internal CHCl3 in the CDCl3 solvent, at
a 1 × 10–2 mol L–1 [2] before and after reaction. After refluxing with CuI
and removal of the pyridine by warming under reduced pressure, the 2
was decomplexed from the CuI by extraction between CH2Cl2 and 10 %
aq. KCN.
[48] K. Tahara, S. Lei, D. Mössinger, H. Kozuma, K. Inukai, M. Van der Auweraer,
F. C. De Schryver, S. Höger, Y. Tobe, S. De Feyter, Chem. Commun. 2008,
3897–3899.
[70]
This effect mirrors the Stephens/Castro cyclotrimerization reactivity pat-
tern of ortho-ethynyliodothiophenes: D. Solooki, V. O. Kennedy, C. A. Tes-
sier, W. J. Youngs, Synlett 1990, 427–428.
V. W.-W. Yam, W.-K. Lee, K.-K. Cheung, H.-K. Lee, W.-P. Leung, J. Chem.
Soc., Dalton Trans. 1996, 2889–2891.
V. W.-W. Yam, W.-K. Lee, K.-K. Cheung, J. Chem. Soc., Dalton Trans. 1996,
2335–2339.
M. P. Gamasa, J. Gimeno, E. Lastra, X. Solans, J. Organomet. Chem. 1988,
346, 277–286.
[49] K. Tahara, H. Yamaga, E. Ghijsens, K. Inukai, J. Adisoejoso, M. O. Blunt, S.
De Feyter, Y. Tobe, Nat. Chem. 2011, 3, 714–719.
[50] T. Nishinaga, H. Nakayama, N. Nodera, K. Komatsu, Tetrahedron Lett.
1998, 39, 7139–7142.
[51] C. Huynh, G. Linstrumelle, Tetrahedron 1988, 44, 6337–6344.
[52] Y. Li, J. Zhang, W. Wang, Q. Miao, X. She, X. Pan, J. Org. Chem. 2005, 70,
3285–3287.
[53] M. Csékei, Z. Novák, A. Kotschy, Tetrahedron 2008, 64, 975–982.
[54] H. Yang, Y. Du, S. Wan, G. D. Trahan, Y. Jin, W. Zhang, Chem. Sci. 2015, 6,
4049–4053.
[71]
[72]
[73]
[74]
[75]
L. Naldini, F. Demartin, M. Manassero, M. Sansoni, G. Rassu, M. A. Zo-
roddu, J. Organomet. Chem. 1985, 279, C42–C44.
[55] O. Š. Miljanić, K. P. C. Vollhardt, G. D. Whitener, Synlett 2003, 1, 29–34.
[56] J. W. Barton, M. K. Shepherd, Tetrahedron Lett. 1984, 25, 4967–4970.
[57] M. Iyoda, A. Vorasingha, Y. Kuwatani, M. Yoshida, Tetrahedron Lett. 1998,
39, 4701–4704.
For a crystallographically characterized CuI-phenylacetylide complex,
[Cu3(μ2-dppm)3(μ3,η1-C≡ CPh)2]PF6, incorporating extra CuI, see: V. W.-W.
Yam, W.-K. Lee, T.-F. Lai, Organometallics 1993, 12, 2383–2387.
M. C. Blanco, J. Camara, M. C. Gimeno, P. G. Jones, A. Laguna, J. M. Lopez-
de-Luzuriaga, M. E. Olmos, M. D. Villacampa, Organometallics 2012, 31,
2597–2605.
[58] P. N. W. Baxter, R. D.-Youcef, J. Org. Chem. 2005, 70, 4935–4953.
[76]
[77]
[59]
S. S. Y. Chui, M. F. Y. Ng, C.-M. Che, Chem. Eur. J. 2005, 11, 1739–1749
and references cited therein.
[60]
In this case (entry 3, Table 2), the increase in yield of 2 upon moderate
dilution of the reaction is consistent with a dilution-induced dissociation
of CuI from the reactive species, thereby reversing the negative effect of
excess CuI, discussed later.
All reaction components remained in solution throughout the course of
the ageing time and heating stage of the medium dilution reaction of
entry 23 of Table 3. The absence of ethynylcuprate precipitates that are
normally observed for the more concentrated reactions involving
tBuOCu as metallating agent, reinforces the conclusion that the cycliza-
tion step occurs by way of a non-polymeric solution species.
Attempts were also undertaken to characterize some of the entities pro-
duced upon ambient temperature ageing of the reactions described in
entries 6 and 13 of Table 3, both by crystallization of solution species
from the reaction medium via diethyl ether vapor diffusion and by ESI
and MALDI mass spectral analyses. However, the crystallization experi-
ments yielded only [Cu2(μ-I)2Py4], which was identified by X-ray crystal-
lography and found to have a different packing to that described in G. C.
Dyason, L. M. Engelhardt, P. C. Healy, A. H. White, Aust. J. Chem. 1984,
37, 2201–2205. The structural details of this modification are therefore
included in the Supporting Information. Furthermore, the peaks in the
mass spectra of the reaction solutions did not correspond to any identifi-
able CuI containing species.
The cyclic tetramers 4 and 17 were searched for in all table entries, but
were found to be produced in only very low yields, contributing little
to the total mass balance of each reaction. For example, isolation of a
multimilligram quantity of cycle 4 was only possible upon combination
of all the chromatographic eluates containing traces of this compound
from all the table entries affording 2, as detailed in the Supporting Infor-
mation. Unidentified by-products were also observed in the 1H NMRs of
the crude reaction products, as well as insoluble black solids that some-
times hampered the solvent extractive stage of workups, but all were
formed in only very small/trace amounts. However, we cannot rule out
the possibility of competing dehalogenation reaction pathways that
would afford volatile products, lost upon workup. Starting 7 was isolated
only in the cases of entries 3, 11 and 17 of Table 3. In all reactions, the
yields of 1 and 2 were based upon 5 and 7 consumed.
[61]
This process is indicated visually by the initial production of a dark
brown solution, in the presence of excess CuX and tBuOCu (entries 5,
18, 20, 21; Table 2), which result in zero yields of 2 after heating. How-
ever, with longer ageing times prior to heating, the normally observed
red precipitate of sparingly soluble 8 reappears, and subsequent heating
affords improved yields of 2 (entries 14–16 and 19; Table 2). It must be
noted however that subsequent experiments (entry 17, Table 3) showed
that the reactive precursor leading to 2 was a solution species rather
than the precipitate itself.
[78]
[62]
After removal of the reaction solvent in the initial phase of workup, the
residue was partitioned between aqueous KCN and CH2Cl2, during which
8 if present, is demetallated to 7. The combined organic extracts were
then dried, and the solvent distilled off at ambient temperature. Extrac-
tion of the residue with n-pentane selectively dissolves out the 7, which
can be subsequently isolated and weighed in order to determine the
actual yield of 2 based upon 7 consumed.
[79]
[63]
[64]
In the systems employing 1:1 to 1:3 stoichiometric ratios of 7/tBuOCu,
the 8 always appears during ambient temperature ageing as a red or
red-orange precipitate in pyridine, 4-tBu-pyridine, PhCN and DMF, and
as an orange solid in 1-methyl-morpholine and p-xylene. During subse-
quent heating, the precipitate gradually changes color to black.
Excepting entry 23, the [5], [7] and [9] = 6 × 10–2 mol L–1 in all other
entries of Table 3. By comparison, the [7] and [5] = 3 × 10–3 mol L–1 in
the medium dilution reactions of entry 10 of Table 2 and that of entry
23 of Table 3 respectively.
[65]
[66]
K. Okuro, M. Furuune, M. Enna, M. Miura, M. Nomura, J. Org. Chem. 1993,
58, 4716–4721.
If the reactions in Table 4 were stoichiometrically quantitative but en-
tirely non-catalytic, the maximum yield of 1 that could be obtained from
0.3 mol CuI would be 29.3 %. As none of the yields of the reactions
described in entries 10–13 of Table 4 exceed this value, the tBuOK/CuI
mediated cyclizations cannot be unambiguously shown to be catalytic.
It may be noted that in the catalytic reactions performed in pyridine,
the yields of 1 were marginally higher when the cyclizations were per-
formed in the presence of PPh3 (cf. entries 8/9 and entries 12/13,
[80]
The coordination complexes (PPh3)2Cu2CO3 and [(PPh3)2CuCO3H]2 have
been crystallographically characterized, see: D. J. Darensbourg, M. W.
Holtcamp, B. Khandelwal, J. H. Reibenspies, Inorg. Chem. 1995, 34, 5390–
5394.
C.-L. Chan, K.-L. Cheung, W. H. Lam, E. C.-C. Cheng, N. Zhu, S. W.-K. Choi,
V. W.-W. Yam, Chem. Asian J. 2006, 1–2, 273–286.
[81]
[82]
[67]
S. S. Batsanov, Inorg. Mater. 2001, 37, 871–885.
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6794
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[83]
[84]
[87] The positive effect of PPh3 on the yields of the Stephens–Castro reaction
reported by Miura et al., in reference [65] may in fact originate from a
rate enhancement rather than a yield enhancement. The 5 h reaction
time stated for the reactions in Table 1 of Miura et al., may have been
insufficient for the reaction without PPh3 to go to completion, consistent
with the low yield of diphenylacetylene coupling product reported for
this entry.
D. Solooki, J. D. Bradshaw, C. A. Tessier, W. J. Youngs, R. F. See, M. Church-
ill, J. D. Ferrara, J. Organomet. Chem. 1994, 470, 231–236.
The H2 hydrogens are those meta- to pyridine nitrogens.
[85] See for example: M. Suzuki, A. Comito, S. I. Khan, Y. Rubin, Org. Lett.
2010, 12, 2346–2349.
[86] These distances lie within the accepted range for aromatic π–π interac-
tions of 3.4–4.8 Å. See for example, C. A. Hunter, J. Singh, J. M. Thornton,
J. Mol. Biol. 1991, 218, 837–846.
Received: July 19, 2019
Eur. J. Org. Chem. 2019, 6783–6795
www.eurjoc.org
6795
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim