Synthesis, structure, and complexation properties of hydroxybenzyl analogs of diethylenetriaminepentaacetic acid

Article abstract of DOI:10.1080/00958972.2010.499936

As part of continuous search for new chelating agents for transition metal ions, a practical method for the preparation of hydroxybenzyl derivatives of diethylenetriaminepentaacetic acid (DTPA) has been developed in this study. N,N''-bis(2-hydroxybenzyl)diethylenetriamine-N,N',N''-triacetic acid (HBDTTA) and four other hydroxybenzyl derivatives of DTPA have been synthesized. The structures and chelating capacity of three ligands with Fe³⁺ have been predicted using density functional theory-based calculations at the BP86/TZVP level within a previously developed and tested computational procedure. The results show that like DTPA, HBDTTA can adopt several different six- and seven-coordinate complex structures. The good stability of Fe ³⁺ complexes of two HBDTTA derivatives indicates that HBDTTA-type ligands can be modified by removing some of their functional groups without greatly affecting their metal-binding efficiency.

Full text of DOI:10.1080/00958972.2010.499936

This article was downloaded by: [University of Arizona]
On: 04 December 2012, At: 01:15
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Coordination Chemistry
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/gcoo20
Synthesis, structure, and complexation
properties of hydroxybenzyl analogs of
diethylenetriaminepentaacetic acid
Henna Pesonen a , Anna Wuorimaa b , Reija Jokela b , Reijo Aksela
c & Kari Laasonen a
a Department of Chemistry, University of Oulu, PO Box 3000,
90014 Oulu, Finland
b Laboratory of Industrial Chemistry, Helsinki University of
Technology (TKK), PO Box 6100, 02015 Espoo, Finland
c Espoo Research Centre, Kemira Oyj, PO Box 44, 02271 Espoo,
Finland
Version of record first published: 13 Jul 2010.
To cite this article: Henna Pesonen, Anna Wuorimaa, Reija Jokela, Reijo Aksela & Kari
Laasonen (2010): Synthesis, structure, and complexation properties of hydroxybenzyl analogs of
diethylenetriaminepentaacetic acid, Journal of Coordination Chemistry, 63:12, 2026-2041
To link to this article: http://dx.doi.org/10.1080/00958972.2010.499936
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-
conditions
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.

Journal of Coordination Chemistry
Vol. 63, No. 12, 20 June 2010, 2026–2041
Synthesis, structure, and complexation properties of
hydroxybenzyl analogs of diethylenetriaminepentaacetic acid
HENNA PESONENy, ANNA WUORIMAAz, REIJA JOKELAz,
REIJO AKSELAx and KARI LAASONEN*y
yDepartment of Chemistry, University of Oulu, PO Box 3000, 90014 Oulu, Finland
zLaboratory of Industrial Chemistry, Helsinki University of Technology (TKK),
PO Box 6100, 02015 Espoo, Finland
xEspoo Research Centre, Kemira Oyj, PO Box 44, 02271 Espoo, Finland
(Received 22 December 2009; in final form 22 April 2010)
As part of continuous search for new chelating agents for transition metal ions, a practical
method for the preparation of hydroxybenzyl derivatives of diethylenetriaminepentaacetic acid
(DTPA) has been developed in this study. N,N00-bis(2-hydroxybenzyl)diethylenetriamine-
N,N0,N00-triacetic acid (HBDTTA) and four other hydroxybenzyl derivatives of DTPA have
been synthesized. The structures and chelating capacity of three ligands with Fe3þ have been
predicted using density functional theory-based calculations at the BP86/TZVP level within a
previously developed and tested computational procedure. The results show that like DTPA,
HBDTTA can adopt several different six- and seven-coordinate complex structures. The good
stability of Fe3þ complexes of two HBDTTA derivatives indicates that HBDTTA-type ligands
can be modified by removing some of their functional groups without greatly affecting their
metal-binding efficiency.
Keywords: DTPA; HBDTTA; Complexation; Iron; DFT
1. Introduction
For environmental reasons, pulping methods based on oxygen chemicals, for example
hydrogen peroxide and peroxy acids have become popular. Transition metal ions,
especially Fe3þ and Mn2þ, catalyze the decomposition of hydrogen peroxide [1, 2].
Thus, their removal from bleaching liquors by chelation is an essential step in the
bleaching sequences. Commonly used complexing agents in such chelation steps are
ethylenediaminetetraacetic acid (EDTA) and diethylenetriaminepentaacetic acid
(DTPA) (figure 1) [3]. These ligands are well-known and effective chelating agents
for transition metal ions. However, during the past decades, the environmental fate of
such ligands has become an issue. The possible ability of chelating agents to mobilize
heavy metals from the sediments is one of their environmental drawbacks.
As a result of continuous search for environmentally benign complexing agents,
we have introduced a series of green alternatives for EDTA and DTPA [4–11].
*Corresponding author. Email: kari.laasonen@oulu.fi
Journal of Coordination Chemistry
ISSN 0095-8972 print/ISSN 1029-0389 online ß 2010 Taylor & Francis
DOI: 10.1080/00958972.2010.499936

Diethylenetriaminepentaacetic acid
2027
Figure 1. Structures of commonly known complexing agents.
However, despite their high stability constants, the performance of complexing agents is
not always satisfactory in deactivating iron and manganese ions in hydrogen peroxide
solutions. In fact, some metal complexes are able to catalyze the decomposition of the
peroxy compounds [12]. At this point, it becomes very important to know the detailed
structure of each complex. Furthermore, by studying the relative stability of different
conformations, the performance of the ligands can be better understood both in
bleaching processes and other applications.
Replacement of the carboxymethyl groups of EDTA with 2-hydroxybenzyl
groups has increased the affinities of the ligands for some trivalent metal ions.
Examples of such compounds are N,N0-bis(2-hydroxybenzyl)ethylenediamine-N,N0-
diacetic acid [13] (HBED, figure 1) and ethylenediamine di(o-hydroxyphenylacetic acid)
[14] (EDDHA, figure 1), which are known for their strong ability to bind metal ions
like Fe3þ. Also, DTPA is known to be an excellent chelating agent. Recently, an
analog of DTPA, N,N00-bis(2-hydroxybenzyl)diethylenetriamine-N,N0,N00-triacetic acid
(HBDTTA, figure 1) and its stability constants with several metal ions have been
introduced [15, 16].
Molecular modeling has become an efficient tool in ligand study and design. By
utilizing powerful modeling tools, the development time of new ligands can be
significantly reduced. Moreover, the volume of chemicals can be decreased due to fewer
syntheses and application tests. In the previous work [17], we studied the complexation
properties of various amino polycarboxylic acids, including, for example, EDTA,
DTPA, AES, and ISA using density functional methods (DFT), a continuum solvation
model (COSMO, conductor-like screening model) and Car–Parrinello ab initio
molecular dynamics. As a result, a fast and straightforward computational method
for evaluating the performance of complexing agents was developed. The method has
been further tested with oligomeric ligands [17, 18] as well as with other amino
polycarboxylic acid ligands [19]. The results (binding trends and complexation
geometries) were promising compared to the available experimental data. Recently,
we have applied the method to study the complexation properties of cysteine [20].
Here, we report the syntheses of HBDTTA, N,N00-bis(2-hydroxybenzyl) diethylene-
triamine-N,N0,N00-tri-2-propionic acid (HBDTTP), and three other hydroxybenzyl
derivatives of DTPA. For three ligands, complexation geometries and energies with
Fe3þ have been determined with DFT (BP86/TZVP) and COSMO solvation model.
The results have been compared to DTPA as well as to other commonly known

2028
H. Pesonen et al.
chelating agents, such as EDTA. In summary, in this study, molecular modeling has
been used to predict the complexation behavior of completely new ligands within the
previously developed and tested computational procedure [17]. Applicability of the
ligands is discussed with special attention to pulp bleaching processes.
2. Experimental
Infrared spectra were recorded on a Perkin–Elmer Spectrum One FT-IR. NMR spectra
were recorded on a Bruker AV-400 spectrometer. Chemical shifts (ꢀ) are reported in
ppm relative to TMS (1H-NMR; ꢀH ¼ 0.00 ppm) and CDCl3 (13C-NMR;
ꢀC ¼ 77.000 ppm). Signal assignments are based on standard HSQC and 1H-1H
COSY experiments. Mass spectra (EI and HRMS, 70 eV, m/z) were obtained on a
Jeol DX 303/DA 5000 instrument. Mass spectra of compounds 3*b, 4*c, e, and 5*d
were obtained on a Micromass LCT TOF MS (ESþ) instrument.
All compounds were purified by column chromatography (CH2Cl2 : MeOH,
99.5 : 0.5) using Merck Kieselgel 60 (230–400 mesh).
2.1. Preparation of N,N00-bis(2-hydroxybenzyl)diethylene triamine (1)
Salicylaldehyde (0.2 mol) was added to a solution of diethylenetriamine (0.1 mol) in
99.5% ethanol (120 mL). After stirring for 3 h at room temperature, NaBH4 (0.22 mol)
was added in small portions during 30 min at 0ꢀC. After 4 h, the mixture was poured
into H2O (110 mL) and extracted with CH2Cl2. The extracts were washed with 5%
NaHCO3, dried over Na2SO4, and evaporated in vacuo.
2.1.1. N,N00-bis(2-hydroxybenzyl)diethylenetriamine
(1). Amorphous;
1H-NMR
(400 MHz; CDCl3) ꢀH 2.65–2.71 (8H, m), 3.90 (4H, s), 6.74 (2H, t, J ¼ 7.5 Hz), 6.77
(2H, d, J ¼ 7.5 Hz), 6.95 (2H, d, J ¼ 7.5 Hz), 7.11 (2H, t, J ¼ 7.5 Hz); 13C-NMR
(100 MHz; CDCl3) ꢀC 47.6 (2C), 48.1 (2C), 51.9 (2C), 115.9 (2C), 118.7 (2C), 122.4 (2C),
128.2 (2C), 128.3 (2C), 157.7 (2C); m/z (EI) 315 (Mþ), 136, 107(100%). Exact mass:
315.1950 (Calcd for C18H25N3O2: 315.1947).
2.2. General procedure for the N-alkylation
N,N00-bis(2-hydroxybenzyl)diethylenetriamine (1) (1.0 mmol) and N,N-diisopropylethy-
lenediamine (DIPEA, 6.4 mmol) in DMF (20 mL) were mixed under argon and cooled
to 0ꢀC. The alkylating agent 2a–e (3.5 mmol) was added during 45 min keeping the
solution at 0ꢀC. The reaction mixture was stirred at room temperature for 2 h and then
at 50ꢀC for 18 h. The reaction mixture was poured into H2O and extracted with CH2Cl2.
The extracts were washed with H2O, dried over Na2SO4, and evaporated in vacuo.
The N-alkylation with 2a (683 mg, 3.5 mmol) gave 526 mg of crude product
comprising 368 mg of compound 3*a.

Diethylenetriaminepentaacetic acid
2029
2.2.1. N,N00-bis(2-hydroxybenzyl)diethylenetriamine-N,N0,N00-triacetic acid tert-butyl
ester (3*a). Amorphous, FT-IR 1731 (CO); 1H-NMR (400 MHz; CDCl3) ꢀH 1.41
(9H, s), 1.46 (18H, s), 2.69 (4H, t, J ¼ 7.0 Hz), 2.78 (4H, t, J ¼ 7.0 Hz), 3.20 (2H, s), 3.24
(4H, s), 3.78 (4H, s), 6.74 (2H, t, J ¼ 7.5 Hz), 6.81 (2H, d, J ¼ 7.5), 6.94 (2H, d,
J ¼ 7.5 Hz), 7.15 (2H, t, J ¼ 7.5 Hz); 13C-NMR (100 MHz; CDCl3) ꢀC 27.9 (9C), 50.9
(2C), 51.3 (2C), 55.3, 55.4 (2C), 56.8 (2C), 80.9, 81.5 (2C), 116.2 (2C), 118.8 (2C), 121.8
(2C), 128.8 (2C), 129.0 (2C), 157.4 (2C), 169.9 (2C), 170.1; m/z (EI) 657 (Mþ), 556, 301,
189 (100%), 107. Exact mass: 657.4000 (Calcd for C36H55N3O8: 657.3989).
The N-alkylation with 2b (732 mg, 3.5 mmol) gave 429 mg of crude product
comprising 265 mg of compound 3*b. According to 13C spectrum, there are four
diastereomeric structures of compound 3*b. Assignment (1C) in spectral data includes
13C chemical shifts for four diastereomers.
2.2.2. N,N00-bis(2-hydroxybenzyl)diethylenetriamine-N,N0,N00-tripropionic acid tert-butyl
ester (3*b). Amorphous; FT-IR 1725 (CO); 1H-NMR (400 MHz; CDCl3) ꢀH 1.10–1.18
(3H, m), 1.19–1.25 (6H, m), 1.41 (9H, s), 1.48 (18H, s), 2.52–2.79 (8H, m), 3.27 (1H, q,
J ¼ 7.0 Hz), 3.46 (2H, q, J ¼ 7.0 Hz), 3.73–3.85 (4H, m), 6.74 (2H, t, J ¼ 7.5 Hz), 6.79
(2H, d, J ¼ 7.5), 6.93 (2H, d, J ¼ 7.5 Hz), 7.13 (2H, t, J ¼ 7.5 Hz); 13C-NMR (100 MHz;
CDCl3) ꢀC 12.7–13.1 (2C), 15.0–15.8 (1C), 28.1 (9C), 48.5–49.2 (2C), 49.5–50.3 (2C),
54.5–54.9 (2C), 58.1–58.4 (2C), 58.6–59.4 (1C), 80.8–80.9 (1C), 81.5–81.6 (2C), 116.2
(2C), 119.0 (2C), 121.9 (2C), 128.8–129.2 (4C), 157.7–157.8 (2C), 171.9–172.0 (2C),
172.7 (1C); m/z (ESþ) 701 (Mþ þ 1), 320 (100%). Exact mass: 699.4497 (Calcd for
C39H61N3O8: 699.4459).
The N-alkylation with 2c (429 mg, 3.5 mmol) gave 320 mg of crude product
comprising 179 mg of compound 4*c.
2.2.3. Piperazinone 4*c. Amorphous; FT-IR 1737 (CO), 1654 (CON); 1H-NMR
(400 MHz; CDCl3) ꢀH 1.27 (3H, t, J ¼ 7.0 Hz), 2.72–2.80 (4H, m), 3.14 (2H, dd,
J ¼ 6.0 Hz), 3.24 (2H, s), 3.45–3.47 (2H, m), 3.49 (2H, s), 3.74 (2H, s), 3.80 (2H, s), 4.19
(2H, q, J ¼ 7.0 Hz) 6.73–7.19 (8H, m); 13C-NMR (100 MHz; CDCl3) ꢀC 14.0, 42.4, 45.4,
48.8, 49.6, 53.8, 56.4, 58.1, 60.0, 61.0, 116.2 (2C), 119.1, 119.4, 120.2, 121.6, 128.9,
129.1, 129.2, 129.5, 157.0, 157.2, 166.1, 171.0; m/z (ESþ) 442 (Mþ þ 1), 336 (100%),
230. Exact mass: 441.2243 (Calcd for C24H31N3O5: 441.2264).
The N-alkylation with 2d (585 mg, 3.5 mmol) gave 481 mg of crude product
comprising 192 mg of compound 3*d and 289 mg of compound 5*d.
2.2.4. N,N00-bis(2-hydroxybenzyl)diethylenetriamine-N,N0,N00-tripropionic acid ethyl
ester (3*d). Amorphous; FT-IR 1739 (CO); 1H-NMR (400 MHz; CDCl3) ꢀH 1.21
(3H, t, J ¼ 7.0 Hz), 1.25 (6H, t, J ¼ 7.0 Hz), 2.72 (4H, t, J ¼ 7.0 Hz), 2.80 (4H, t,
J ¼ 7.0 Hz), 3.31 (2H, s), 3.34 (4H, s), 3.79 (4H, s), 4.09 (2H, q, J ¼ 7.0 Hz), 4.15 (4H, q,
J ¼ 7.0 Hz), 6.75 (2H, t, J ¼ 7.5 Hz), 6.82 (2H, d, J ¼ 7.5 Hz), 6.95 (2H, d, J ¼ 7.5 Hz),
7.16 (2H, t, J ¼ 7.5 Hz); 13C-NMR (100 MHz; CDCl3) ꢀC 13.7 (3C), 50.6 (2C), 50.9 (2C),
53.2 (2C), 54.2 (1C), 56.6 (2C), 59.9 (1C), 60.4 (2C), 115.8 (2C), 118.6 (2C), 121.5 (2C),
128.6 (2C), 128.9 (2C), 157.1 (2C), 170.3 (2C), 170.4 (1C); m/z (EI) 573 (Mþ), 500, 245
(100%), 222, 107. Exact mass: 573.3007 (Calcd for C30H43N3O8: 573.3050).

2030
H. Pesonen et al.
2.2.5. Piperazinone 5*d. Amorphous; FT-IR 1739 (CO), 1650 (CON); 1H-NMR
(400 MHz; CDCl3) ꢀH 1.23 (3H, t, J ¼ 7.0 Hz), 1.24 (3H, t, J ¼ 7.0 Hz), 2.69 (2H, t,
J ¼ 6.0 Hz), 2.80 (2H, t, J ¼ 6.0 Hz), 3.10 (2H, t, J ¼ 6.0 Hz), 3.23 (2H, s), 3.24 (2H, s),
3.44 (2H, t, J ¼ 6.0 Hz), 3.47 (2H, s), 3.77 (2H, s), 4.15 (2H, q, J ¼ 7.0 Hz), 4.16 (2H, q,
J ¼ 7.0 Hz), 6.72 (1H, t, J ¼ 7.5 Hz), 6.78 (1H, d, J ¼ 7.5 Hz), 6.93 (1H, d, J ¼ 7.5 Hz),
7.12 (1H, t, J ¼ 7.5 Hz); 13C-NMR (100 MHz; CDCl3) ꢀC 14.1 (2C), 42.9, 46.0, 49.0,
49.1, 54.1, 56.6, 57.8, 58.1, 60.9, 61.1, 116.3, 119.2, 121.8, 129.2, 129.5, 157.3, 166.5,
169.2, 170.7; m/z (ESþ) 422 (Mþ þ 1), 316 (100%), 213. Exact mass: 421.2206 (Calcd
for C21H31N3O6: 421.2213).
The N-alkylation with 2e (634 mg, 3.5 mmol) gave 458 mg of crude product
comprising ca 300 mg of compound 4*e (mixture of diastereomers). Assignment (1C)
in spectral data includes 13C chemical shifts for two diastereomers.
2.2.6. Piperazinone 4*e. Amorphous; FT-IR 1730 (CO), 1646 (CON); 1H-NMR
(400 MHz; CDCl3) ꢀH 1.30 (3H, t, J ¼ 7.0 Hz), 1.37 (3H, d, J ¼ 7.0 Hz), 1.51 (3H, d,
J ¼ 7.0 Hz), 2.56–2.67 (1H, m), 2.72–2.84 (2H, m), 3.04–3.21 (3H, m), 3.27–3.32
(1H, m), 3.38–3.46 (1H, m), 3.54–3.59 (1H, m), 3.60–3.69 (1H, m), 3.75–3.90 (2H, m),
3.96–4.05 (1H, m), 4.14–4.18 (1H, m), 4.21 (2H, q, J ¼ 7.0 Hz), 6.74–6.86 (4H),
6.96–7.02 (2H), 7.13–7.20 (2H); 13C-NMR (100 MHz; CDCl3) ꢀC 11.2–11.8 (1C), 14.1
(1C), 15.1–15.3 (1C), 43.7–43.9 (1C), 44.3–44.6 (1C), 44.7–44.8 (1C), 46.3–46.6 (1C),
53.7 (1C), 56.4–56.5 (1C), 57.0 (1C), 60.1–60.2 (1C), 60.9–61.0 (1C), 116.1–116.2 (2C),
119.0–119.3 (2C), 121.6–121.7 (2C), 128.8–129.0 (2C), 129.4–129.5 (2C), 157.2–157.3
(2C), 169.4–169.6 (1C), 172.9–173.2 (1C); m/z (ESþ) 470 (Mþ þ 1) 386, 364 (100%).
Exact mass: 469.2577 (Calcd for C26H35N3O5: 469.2577).
2.3. Computational procedure
All geometry optimizations have been performed using the program TURBOMOLE
6.0 [21]. The optimizations have been carried out with DFT using the BP86 functional
[22, 23] and all-electron triple zeta valence polarization (TZVP) basis set for all atoms.
The RI approximation [24, 25] has been used to speed up the calculations. Two other
density functionals, PBE0 [26] and B3LYP [27], have been tested for selected structures.
All the geometries have been optimized using the COSMO solvation model [28].
Apart from the radius for Fe3þ ion, default options of the program TURBOMOLE
have been used. The COSMO radius used for Fe3þ is 1.391 A and it has been
˚
determined using a hexaaquo Fe3þ complex. For details of the determination of the
metal radius, we refer the reader to references [17, 18]. All complexes have preferred
high-spin configurations.
The complexation energies have been calculated using equation (1).
DEC0 ¼ DECSM ¼ ECSMðMLnꢁmÞ þ 6ECSMðH2OÞ ꢁ ECSMðMðH2OÞn6þÞ ꢁ ECSMðLmꢁÞ,
ð1Þ
ECSM (X) is the COSMO-corrected total energy of species (X). Both the complex
structures and the free ligands were optimized using the COSMO solvation model.
Complexation energies DEC0 offer a useful tool for comparing the relative stabilities
of different structures. However, if one wants to evaluate the absolute binding strength

Diethylenetriaminepentaacetic acid
2031
of the ligands, DEC0 values calculated need to be corrected as they have been shown [17]
to deviate quite a lot from the experimental DGexperimental derived from log K values. In
this study, we apply the previously [17] introduced effective parameterization to correct
for missing contributions in complexation energies. Corrected complexation energy
DGC1 is defined as
~
DGC1ðMi, Lj Þ ¼ DEC0ðMi, Lj Þ þ GMLi
,
ð2Þ
~
GMLi is a metal-specific correction parameter, determined for metal complexes of amino
polycarboxylic acid ligands, including, for example, EDTA and DTPA. These
parameters were determined by fitting the complexation energies obtained by the
DFT/COSMO approach to reproduce experimental data. The method has been
successfully applied to metal complexes of oligomeric ligands [18, 29] as well as to other
amino polycarboxylic acids [19] including, for example, different isomers of EDDHA.
~
The value of GMLi depends on the used density functional. Recently [20], we tested three
other density functionals for metal complexes of cysteine, and DEC0 were found to be
highly dependent on the chosen functional. For EDTA, on the other hand, differences
were smaller [30].
The complexation energy calculations do not include any explicit thermal or internal
entropy corrections. Other missing contributions include at least corrections for ZPE,
nonelectrostatic solvation effects, and basis set superposition error (BSSE). However, as
accurate calculation of these contributions is very demanding and, on the other hand, as
the applicability of this kind of approach has already been demonstrated in previous
studies [17–20], this omission is not a concern. The magnitude of some of these
contributions has been estimated for selected cases by Sillanpaa et al. [17].
¨
¨
3. Results
3.1. Synthesis of hydroxybenzyl analogs of DTPA
We applied a commonly known procedure to synthesize N,N00-bis(2-hydroxybenzyl)
diethylenetriamine (1) for the synthesis of salenides [31]. Addition of diethylenetriamine
to salicylaldehyde followed by reduction with NaBH4 (scheme 1) resulted in the
formation of 1 in good yield (97%). The product was used without purification in the
next N-alkylation step.
Five different alkylating agents (scheme 2, compounds 2a–e) were chosen for the
N-alkylation step. The influences of the ester moiety, the leaving group X and an added
methyl group were studied. All the experiments were performed with equimolar ratios
of starting material and identical reaction conditions (time and temperature).
In the literature [13], an inorganic base has been used in the N-alkylation step.
However, sodium carbonate in DMF solution was unsuccessful in our hands, whereas
the use of an organic base DIPEA gave good results in most cases.
In the SN2-reactions of N,N00-bis(2-hydroxybenzyl)diethylenetriamine (1) with
alkyl halides 2a–e, formation of the corresponding trialkylated products was assumed.
Indeed, thistookplacewithalkylhalides2a, 2b, and2d. Thetrialkylatedproducts3*a, 3*b,
and 3*d were formed. In addition, small amounts of dialkylated products were formed.

2032
H. Pesonen et al.
Scheme 1. Preparation of N,N00-bis(2-hydroxybenzyl)diethylenetriamine (1).
Scheme 2. Synthesis of the N-alkylated 2-hydroxybenzyl derivatives of diethylenetriamine.
Sincethereactionofthesecondaryalkylhalide2bwassignificantlyslowerthanthereaction
of the primary alkyl halide 2a, more of the dialkylated product was formed.
With dialkylated products additional ring formation was possible. Ring formation
was observed with alkyl halides 2c and 2e. The fact that alkyl halides 2a and 2b yielded
no cyclic products, 4*a and 4*b can be explained by comparing the two leaving groups.
In an SN2-reaction, a t-butoxy group is a poorer leaving group than an ethoxy group.
When ethyl bromoacetate 2d was used, the cyclic compound 5*d where one of the
hydroxybenzyl groups was replaced by an alkyl group was formed. It can be explained
by the formation of a quaternary ammonium salt of 4*c from which the hydroxybenzyl
group is removed more readily than the alkyl group. The formation of a quaternary salt
is due to better reactivity of ethyl bromoacetate 2d compared to ethyl chloroacetate 2c.

Diethylenetriaminepentaacetic acid
2033
As a result, an effective method for preparation of glycine and methylglycine
derivatives of N,N00-bis(2-hydroxybenzyl)diethylenetriamine was developed. The reac-
tions gave good yields when t-butyl esters of the alkylating agents were applied.
In summary, we were able to shorten the known syntheses of HBDTTA [15, 16]
by ignoring either the protection step or the amide reduction. In addition, the
products were hydrolyzed in aqueous alkaline conditions to yield salts of 3a, 3b, 3d, 4c,
4e, and 5d.
3.2. Geometries and relative stabilities
As described above, the synthesis products (figure 2) may have either a methyl group or
a hydrogen attached to the carbons next to the amino groups (R2, respectively).
However, as the methyl group does not coordinate with the metal, it does not directly
participate in the complexation reaction. Thus, we replaced it by hydrogens simply for
computational reasons.
We optimized Fe3þ complexes for three synthesis products. The studied ligands were
HBDTTA and its two incompletely alkylated derivatives (4c and 5d). In order to find
the minimum energy structures, we optimized 5–8 different structures for each ligand
with different coordination modes. In addition to complexes of new ligands, previously
studied [Fe–DTPA]2ꢁ complexes were reoptimized with COSMO. All the carboxylic
Figure 2. Synthesis products in their acid form.

2034
H. Pesonen et al.
acid and phenolic groups were considered completely deprotonated. The most stable
structures for [Fe–HBDTTA]2ꢁ, [Fe–4c]ꢁ, [Fe–5d]ꢁ, and [Fe–DTPA]2ꢁ are shown in
figure 3. As the ligands were relatively flexible, figure 3 includes several structures for
each complex.
In general, we found six- and seven-coordinate structures for all the ligands. Despite
the fact that HBDTTA and DTPA have in total eight possible metal-coordinating
atoms, eight-coordinated structures seem to be clearly higher in energy than the most
stable structures [17] and in fact, no such structures were found for [Fe–HBDTTA]2ꢁ
.
For [Fe–HBDTTA]2ꢁ, six-coordinate structure 6b is the most stable one with an
energy difference of 23 kJ molꢁ1 with respect to the second best six-coordinate structure
6a and 22 kJ molꢁ1 to the most stable seven-coordinate structure. For [Fe–DTPA]2ꢁ
,
all three structures are within 7 kJ molꢁ1 and thus, considered equally stable. However,
if the absolute energies are considered (see section 3.4), all modeled HBDTTA
structures have clearly lower complexation energies than the lowest DTPA complexes.
Note that for HBDTTA, there are several relatively stable structures with only one
metal-binding phenol group and thus, the expected increase in stability with respect to
DTPA does not necessarily originate from the fact that both phenol groups bind to
the metal.
Figure 3. The most stable structures for [Fe–HBDTTA]2ꢁ, [Fe–4c]1ꢁ, [Fe–5d]1ꢁ, and [Fe–DTPA]2ꢁ
The structures have been optimized at the BP86/TZVP level using the COSMO solvation model.
.

Diethylenetriaminepentaacetic acid
2035
For 4c, the most stable structure is identical to six-coordinate structure 6b of
HBDTTA with the exception of one free COOꢁ group of HBDTTA complex which
is absent in [Fe–4c]ꢁ. The energy difference between the six- and seven-coordinate
structures is almost identical to HBDTTA. With 5d, the most stable structure is
six-coordinate with one COOꢁ group pointing out to the solvent. The structures
of two DTPA and HBDTTA complexes are compared in more detail in the following
section.
In general, all the ligands were shown to be quite flexible and we found several
structures within some dozens of kilojoules per mol for each complex. This means that
it is probable that different conformations of the complexes are present simultaneously,
especially at high temperatures like in the bleaching processes in which temperatures
above 80ꢀC are often used.
3.3. Structural comparisons
Six-coordinate structures 6b of DTPA and HBDTTA are very close to octahedral
geometry and the differences in structural parameters (bond lengths and angles) of
Fe–DTPA and Fe–HBDTTA are relatively small. From the structural point of view,
seven-coordinate structures are more interesting. Figure 4 shows a structural compar-
ison for seven-coordinate [Fe–DTPA]2ꢁ and [Fe–HBDTTA]2ꢁ. Metal–ligand bond
lengths (in Angstrom) and selected bond angles (in degrees) between the structures
¨
are shown.
Both complexes are in distorted pentagonal bipyramidal geometry. As seen in
figure 4, Fe–N bonds are slightly longer in [Fe–HBDTTA]2ꢁ with respect to
[Fe–DTPA]2ꢁ. Fe–O2 bond is clearly longer in [Fe–HBDTTA]2ꢁ and the bond angle
N2–Fe–O2 smaller than the corresponding value in [Fe–DTPA]2ꢁ, probably due to two
effects: first, the large phenolate groups at both ends of the ligand cause general
distortion to the ligand with respect to DTPA. For the same reason, the O1–Fe–O4
Figure 4. Metal–ligand bond lengths (in Angstrom) and bond angles (in degrees) in seven-coordinate
¨
.
[Fe–DTPA]2ꢁ and [Fe–HBDTTA]2ꢁ

2036
H. Pesonen et al.
angle is more than 5ꢀ larger in Fe–HBDTTA than in Fe–DTPA. Second, the
phenol–oxygen bonds to Fe3þ are shorter than Fe–O1, and Fe–O4 bonds in Fe–DTPA,
indicating stronger bonding, which weakens (i.e., lengthens) the other metal–ligand
bonds. Overall Fe–HBDTTA is more distorted than Fe–DTPA, which partly explains
the lower binding energy of seven-coordinate HBDTTA compared to DTPA.
In general, COSMO model produces slightly longer bonds than those measured
experimentally using crystal structures as seen previously, for example with ZnCys [20]
and Fe–EDDHA [19] complexes. The ligand strain energies for these complexes have
been calculated and the results are presented in section 3.5.
3.4. Complexation energies
Complexation energies have been calculated using equation (1). Table 1 shows the
complexation energies for the most stable [Fe–HBDTTA]2ꢁ, [Fe–4c]ꢁ, and [Fe–5d]ꢁ
complexes compared to the corresponding values of DTPA, HBED, EDTA, and o,o-
EDDHA. Experimental values DGexperimental are shown (where available). DEC0 values
indicate that the ligands o,o-EDDHA, 4c, and HBED are the most effective chelators
for Fe3þ. With respect to EDTA and DTPA, HBDTTA is expected to be a very efficient
chelating agent. DEC0 for [Fe–5d]ꢁ is ca 64 kJ molꢁ1 higher than DEC0 for [Fe–4c]ꢁ but
still ca 30 kJ molꢁ1 lower than the complexation energy for [Fe–EDTA]ꢁ, and
17 kJ molꢁ1 lower than DEC0 for [Fe–DTPA]2ꢁ
.
The two ligand pairs HBDTTA–DTPA and HBED–EDTA have one thing in
common, the first ligand is obtained by replacing two COOH groups of the latter by
phenol. In the case of EDTA and HBED, this modification leads to an improvement of
85 kJ molꢁ1 (experimentally) and 95 kJ molꢁ1 (computationally) in complexation free
energy. For DTPA and HBDTTA, on the other hand, the difference in experimental DG
is only 14 kJ molꢁ1. Computational results predict clearly larger enhancement
(63 kJ molꢁ1). To us, this discrepancy is strange, as in our calculations the two ligand
Table 1. Calculated (BP86/TZVP/COSMO) and experimental complexation free
energies. DGexperimental have been derived from log K1 values for complexation at 0.1
ionic strength and 298 K using DGexperimental ¼ ꢁRT ln 10 log K1. All values are expressed
in kilojoules per mol. Note that in experiments, the coordination number is not known
and thus we used the same DGexperimental for all binding geometries.
Ligand
Coordination
DEC0
DGC1
DGexperimental
HBDTTA
DTPA
HBDTTA
DTPA
HBDTTA
DTPA
4c
6a
6a
6b
6b
7
7
6
7
6
ꢁ467
ꢁ420
ꢁ490
ꢁ422
ꢁ468
ꢁ427
ꢁ508
ꢁ485
ꢁ444
ꢁ507c
ꢁ413c
ꢁ520c
ꢁ183
ꢁ136
ꢁ206
ꢁ138
ꢁ184
ꢁ143
ꢁ224
ꢁ201
ꢁ160
ꢁ223
ꢁ129
ꢁ236
ꢁ174a
ꢁ160b
ꢁ174a
ꢁ160b
ꢁ174a
ꢁ160b
–
4c
5d
–
–
HBED
EDTA
o,o-EDDHA
6
6
6
ꢁ227d
ꢁ143b
ꢁ200e
aRef. [15]; bRef. [32]; cRef. [19]; dRef. [33]; and eRef. [34].

Diethylenetriaminepentaacetic acid
2037
pairs behave analogously. This clearly demonstrates that the difficulty of predicting
complexation behavior of a new ligand is just based on earlier experience on
similar cases.
In general, calculated DEC0 deviate quite a lot from the DGexperimental and thus,
further correction (in addition to the correction made by COSMO) is needed. In this
study, we have applied the previously developed empirical correction parameter to
correct for omitted contributions in ‘‘raw’’ DEC0 values. The parameter has been fitted
with DEC0 and DGexperimental with several aminopolycarboxylate ligands, including, for
example, EDTA and DTPA. For Fe3þ, correction is 284 kJ molꢁ1 [17]. Note that the
designation of DGC1 has been changed with respect to the study of Sillanpaa et al.
¨
¨
The value of DGC1 for [Fe–HBDTTA]2ꢁ is overestimated with respect to the
experimental value. The same tendency has been seen with o,o-EDDHA and thus, also
the DGC1 values for complexes of 4c and 5d can be considered optimistic. Aware of this
probable but rather small inaccuracy, we rely more on the relative energies and on the
general trend, which has been proven to reproduce the experimental binding trends.
To evaluate the possible inaccuracies that originate from the used GGA functional
(BP86), we tested two hybrid functionals PBE0 and B3LYP for selected structures.
The tested structures were six- and seven-coordinate structures of [Fe–DTPA]2ꢁ and
[Fe–HBDTTA]2ꢁ. DEC0 for the complexes obtained with different functionals are
shown in table 2. All values are expressed in kilojoules per mol and they refer to
BP86/TZVP/COSMO optimized geometry.
As seen in table 2, the calculated DEC0 are strongly dependent on the used functional.
With respect to BP86, the values obtained with PBE0 are lower and those obtained with
B3LYP are visibly higher (¼less negative). Two important conclusions can be made
based on data shown in table 2: (1) The trends in relative stabilities of the structures
obtained with BP86 and PBE0 are equal. B3LYP, on the other hand, slightly favors the
six-coordinate structures. (2) Despite the large differences in DEC0, for all tested
functionals, the deviation from DGexperimental (table 1) is notable and thus, further
correction of the ‘‘raw’’ DEC0 is necessary.
3.5. Strain energies
An interesting observation about the energies presented in table 2 is that with BP86,
the six- and seven-coordinate structures of DTPA and HBDTTA complexes
Table 2. Calculated complexation energies DEC0 (kJ molꢁ1) for six- and seven-
coordinate [Fe–HBDTTA]2ꢁ and [Fe–DTPA]2ꢁ complexes (6a, 6b, and 7) with different
density functionals. All values refer to BP86/TZVP/COSMO optimized geometries
(figure 3).
DEC0
Ligand
Coordination
BP86
PBE0
B3LYP
HBDTTA
DTPA
HBDTTA
DTPA
HBDTTA
DTPA
6a
6a
6b
6b
7
ꢁ467
ꢁ420
ꢁ490
ꢁ422
ꢁ468
ꢁ427
ꢁ476
ꢁ459
ꢁ541
ꢁ464
ꢁ517
ꢁ473
ꢁ323
ꢁ264
ꢁ346
ꢁ266
ꢁ280
ꢁ258
7

2038
H. Pesonen et al.
(except [Fe–HBDTTA]2ꢁ 6a) are almost equally stable. It is very likely that in the seven-
coordinate structures, the electrostatic metal–ligand interactions are stronger and so
there has to be another effect to compensate this. To study this, we calculated ligand
strain energies for the selected complex structures of DTPA, HBDTTA, 4c, and 5d. For
the first two ligands, both six- and seven-coordinate structures were considered. Strain
energy Estrain is defined as
Estrain ¼ ECSMðLcomplexÞ ꢁ ECSMðLfreeÞ,
ð3Þ
ECSM(Lfree) refers to the total energy of the COSMO optimized free ligand, while
ECSM(Lcomplex) is the total energy of the ligand in the complex structure. Strain energies
for HBDTTA, DTPA, 4c, and 5d are presented in table 3.
In the first two six-coordinate complexes (6a) Fe–HBDTTA and Fe—DTPA, ligand
strain energy is almost equal. In the 6b structures, the difference in ligand strain energies
is huge, ca 300 kJ molꢁ1 favoring the DTPA complex. Also, the seven-coordinate
Fe–HBDTTA is more strained by 60 kJ molꢁ1 than the corresponding DTPA complex.
The seven-coordinate structures of Fe–HBDTTA and Fe–4c are identical from the
point of view of metal–ligand bonding pattern and thus, Estrain in these complexes are
within 10 kJ molꢁ1. The ligand 5d has only one phenol ring and has strain energy
comparable to Estrain of six-coordinate Fe–HBDTTA and Fe–DTPA.
The study of strain energies clearly shows that there are two competing effects related
to complexation. First, there is the electrostatic attraction between the metal and donor
groups of the ligand and second, there is the ligand strain energy. The calculation of the
strain energies is straightforward. However, the study of the electrostatic attraction is
more complicated because different donor atoms seem to have fairly different
interaction energies with the metal. In order to increase our knowledge related to this
phenomenon, we are currently studying the metal–ligand interactions more in detail.
4. Discussion
The members of the studied HBDTTA ligand family seem to be very effective
complexing agents for Fe3þ. Their binding capacity is better than that of EDTA and
DTPA and comparable to that of o,o-EDDHA or HBED. As HBED is obtained by
Table 3. Ligand strain energies (kJ molꢁ1) in selected
Fe–HBDTTA, Fe–DTPA, Fe–4c, and Fe–5d complexes.
Ligand
Coordination
Estrain
HBDTTA
DTPA
HBDTTA
DTPA
HBDTTA
DTPA
4c
6a
6a
6b
6b
7
7
6a
7
370
366
618
310
463
402
346
455
366
4c
5d
6

Diethylenetriaminepentaacetic acid
2039
substituting two COOH groups of EDTA by phenolic rings, one could expect that
similar effect can be seen when changing from DTPA to HBDTTA. Computationally,
this effect is clear even if the experimental results predict smaller enhancement in log K
values. Also the ‘‘incomplete’’ forms of HBDTTA, i.e., 4c and 5d, are predicted to be
very effective chelators for Fe3þ. This makes HBDTTA-type ligands very interesting as
these ligands can be modified by removing some of their functional groups without
greatly affecting their binding capacities. For other ligands, modification can lead to
more drastic changes on their binding behavior. This is the case for EDDHA, where
repositioning of the phenolic OH group from ortho to para causes a change of ca
40 kJ molꢁ1 in complexation energy [19].
Iron is a very effective catalyst for many reactions, for example for the dissociation
of hydrogen peroxide. Chelating agents are often used to prevent the unwanted
dissociation of H2O2. To be able to deactivate the metal ion, the ligand must shield the
metal ion completely. Our calculations on the HBDTTA ligand family show that all
the ligands (HBDTTA, 3b, 4c, 4d, and 5d) provide enough coordination sites to do
this. This is in agreement with preliminary peroxide stabilization test results [35].
The results mean that the performance of a certain chelating agent cannot be predicted
by considering only the log K1 values of its metal complexes. Possible impurities,
synthesis side products, and isomerism are also factors that may have an effect on the
ligands’ overall behavior as stabilizer. This is the case for EDDHA, the most common
synthesis route of which [36, 37] produces also o,p- and p,p-isomers of the ligand.
These two isomers are rather effective chelating agents for Fe3þ but due to their
under-coordinated complex structures [19], they are probably not able to inactivate
the metal ion.
Although the absolute energies are inaccurate due to the used empirical correction,
the relative energies are reliable and can be used to compare the binding capacity of
different ligands for one metal ion. The study of geometries is of special importance in
the case of Fe3þ, which can form metal complexes with coordination numbers from six
to eight and often has more than one stable complexation geometry. As seen in this
study, this kind of approach is more than helpful when the applicability of a certain
ligand is studied.
5. Conclusions
Amino polycarboxylic acids are well-known for their capacity to form stable complexes
with transition metal ions. Experimental methods provide techniques for determining
the stability constants for complexation reactions. Some geometrical information can be
obtained, for example from NMR experiments, and crystal structures can be measured
with X-ray technique. However, also these methods have their limitations as several
equilibrium reactions may occur simultaneously, and the crystal structures do not
necessarily represent the solution structures of the complexes. Molecular modeling
provides a fast approach for structural analysis without complicated experimental
procedures. In this study, we have applied a previously developed and tested method,
which has been shown to be suitable and powerful for this kind of applications.
The ligands have been synthesized and their complexation properties predicted by
DFT calculations at the BP86/TZVP level. The results indicate that like [Fe–DTPA]2ꢁ
,

2040
H. Pesonen et al.
[Fe–HBDTTA]2ꢁ also has several complex structures that are energetically close to each
other, which may be a reason for its performance in peroxide stabilization tests. Other
members of the HBDTTA ligand family are predicted to be efficient chelators for Fe3þ
as long as they have enough (i.e., at least six) well-positioned coordination sites for
complexation. Generally, based on geometry optimizations and complexation energy
calculations, the following conclusions can be made for complexation of Fe3þ with
HBDTTA-type ligands. The groups that coordinate to the metal ion include at least one
phenolic group, two amino groups, and two carboxyl groups. Thus, the presence of
these groups in the ligand is essential. The sixth (and seventh, if possible) coordination
site of metal may be occupied by carboxyl, amino, or phenolic group. The study shows
the power of molecular modeling when the structural properties of metal complexes are
concerned.
Acknowledgments
The financial support of Academy of Finland, grant no. 116839 (HP), Finnish
National Foundation of Technology and Kemira Oyj is gratefully acknowledged. CSC
(the Finnish IT Center for Science) is acknowledged for providing computational
resources.
References
[1] Z. Yang, M. D’Entremont, Y. Ni, A.R.P. Heiningen. Pulp Pap. Can., 98, T408–413 (1997).
[2] J. Basta, L. Holtinger, W. Hermansson, P. Lundgren. Metal management in TCF/ECF bleaching. In
Proceedings of the International Pulp Bleaching Conference, Vancouver, Canada, 13–16 June, pp. 29–32
(1994).
[3] L. Lapierre, J. Bouchard, R.M. Berry, B. Van Lierop. J. Pulp Pap. Sci., 21, 269 (1995).
[4] M. Orama, H. Hyvonen, H. Saarinen, R. Aksela. J. Chem. Soc., Dalton Trans., 4644 (2002).
¨
[5] H. Hyvonen, M. Orama, H. Saarinen, R. Aksela. Green Chem., 5, 410 (2003).
¨
[6] H. Hyvonen, M. Orama, P. Ale
´
[7] H. Hyvonen, M. Orama, R. Arvela, K. Henriksson, H. Saarinen, R. Aksela, A. Pare
n, H. Saarinen, R. Aksela, A. Pare
´
n. J. Coord. Chem., 58, 1115 (2005).
n, J. Jakara,
¨
´
¨
I. Renvall. Appita J., 59, 142 (2006).
¨
¨
¨
[8] H. Hyvonen, R. Aksela. J. Coord. Chem., 60, 901 (2007).
¨
[9] H. Hyvonen, P. Lehtinen, R. Aksela. J. Coord. Chem., 61, 984 (2008).
¨
[10] H. Hyvonen, R. Aksela. J. Coord. Chem., 61, 2515 (2008).
¨
[11] R. Aksela, A. Paren, J. Jakara, I. Renvall. The overall performance of new diethanolamine derivatives
´
¨ ¨ ¨
as complexing agents in peroxide and peracetic acid bleaching of TCF pulp. In Proceedings of
the 11th ISWPC International Symposium on Wood and Pulping Chemistry, Nice, France, 11–14 June,
pp. 481–484 (2001).
[12] D. Swern. Organic Peroxides, Vol. II, pp. 497–503, Wiley-Interscience, New York (1970).
[13] F. L’Eplattenier, I. Murase, A.E. Martell. J. Am. Chem. Soc., 89, 837 (1967).
[14] A.E. Frost, H.H. Freedman, S.J. Westerback, A.E. Martell. J. Am. Chem. Soc., 80, 530 (1958).
[15] R. Ma, I. Murase, A.E. Martell. Inorg. Chim. Acta, 223, 109 (1994).
[16] R.D. Hancock, I. Cukrowski, E. Cukrowska, G.D. Hosken, V. Iccharam, M.W. Brechbiel,
O.A. Gansow. J. Chem. Soc., Dalton Trans., 18, 2679 (1994).
[17] A.J. Sillanpaa, R. Aksela, K. Laasonen. Phys. Chem. Chem. Phys., 5, 3382 (2003).
¨
¨
[18] H. Pesonen, A. Sillanpaa, R. Aksela, K. Laasonen. Polymer, 46, 12653 (2005).
¨
¨
[19] H. Pesonen, R. Aksela, K. Laasonen. J. Mol. Struct. (Theochem), 804, 101 (2007).
[20] H. Pesonen, R. Aksela, K. Laasonen. J. Phys. Chem. A, 114, 46 (2010).
[21] R. Ahlrichs, M. Bar, M. Haser, H. Horn, C. Kolmel. Chem. Phys. Lett., 162, 165 (1989).
¨
¨
[22] A.D. Becke. Phys. Rev. A, 38, 3098 (1988).
¨

Diethylenetriaminepentaacetic acid
[23] J.P. Perdew. Phys. Rev. B, 33, 8822 (1986).
2041
¨
[24] K. Eichkorn, O. Treutler, H. Ohm, M. Haser, R. Ahlrichs. Chem. Phys. Lett., 240, 283 (1995).
¨
¨
[25] K. Eichkorn, O. Treutler, H. Ohm, M. Haser, R. Ahlrichs. Chem. Phys. Lett., 242, 652 (1995).
¨
[26] J.P. Perdew, M. Ernzerhof, K. Burke. J. Chem. Phys., 105, 9982 (1996).
[27] A.D. Becke. J. Chem. Phys., 98, 5648 (1993).
[28] A. Klamt, G. Schuurmann. J. Chem. Soc., Perkin Trans., 2, 799 (1993).
¨
[29] H. Pesonen, A. Sillanpaa, R. Aksela, K. Laasonen. Polymer, 46, 12641 (2005).
¨
¨
¨
[30] H. Pesonen, R. Aksela, K. Laasonen. In preparation (2010).
[31] W.M. Coleman, L.T. Taylor. Inorg. Chem., 10, 2195 (1971).
[32] A.E. Martell, R.M. Smith (Eds), Critically Selected Stability Constants of Metal Complexes, Version 4.0,
Texas A&M University, College Station, TX (1997).
[33] R.J. Motekaitis, A.E. Martell, M.J. Welch. Inorg. Chem., 29, 1463 (1990).
´ ´
[34] F. Yunta, S. Garcia-Marco, J.J. Lucena, M. Gomez-Gallego, R. Alcazar, M.A. Sierra. Inorg. Chem., 42,
5412 (2003).
[35] R. Aksela. Unpublished results.
[36] H.E. Petree, H.L. Myatt, A.M. Jelenevsky. Preparation of phenolic ethylene diaminepolycarboxylic
acids. US Patent 4130582.
[37] J.A.L. Julien, A. Aymard. Nouveau proce
bis(o-hydroxyphenylacetique) et de derives de celui-ci. European Patent 0331556 A2.
de de preparation de la
´ ´ ´
cide ethylenediamine N,N0-
´
´