Bismuth Phosphinates in Bi-Nanocellulose Composites and their Efficacy towards Multi-Drug Resistant Bacteria

Article abstract of DOI:10.1002/chem.201801803

A series of poorly soluble phenyl bis-phosphinato bismuth(III) complexes [BiPh(OP(=O)R¹R²)₂] (R¹=R²=Ph; R¹=R²=p-OMePh; R¹=R²=m-NO₂Ph; R¹=Ph, R²=H; R¹=R²=Me) have been synthesised and characterised, and shown to have effective antibacterial activity against Escherichia coli (E. coli), Staphylococcus aureus (S. aureus), methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant Enterococcus (VRE). The bismuth complexes were incorporated into microfibrillated (nano-) cellulose generating a bismuth-cellulose composite as paper sheets. Antibacterial evaluation indicates that the Bi-cellulose materials have analogous or greater activity against Gram positive bacteria when compared with commercial silver based additives: silver sulfadiazine loaded at 0.43 wt % into nanocellulose produces a 10 mm zone of inhibition on the surface of agar plates containing S. aureus whereas [BiPh(OP(=O)Ph₂)₂] loaded at 0.34 wt % produces an 18 mm zone of inhibition. These phenyl bis-phosphinato bismuth(III) complexes show potential to be applied in materials in healthcare facilities, to inhibit the growth of bacteria capable of causing serious disease.

Full text of DOI:10.1002/chem.201801803

A Journal of
Accepted Article
Title: Bismuth Phosphinates in Bi-Nanocellulose Composites and their
Efficacy towards Multi-Drug Resistant Bacteria
Authors: Philip C. Andrews, Melissa Werrett, Megan Herdman, Rajini
Brammananth, Uthpala Garusinghe, Warren Batchelor, Paul
Crellin, and Ross Coppel
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: Chem. Eur. J. 10.1002/chem.201801803
Link to VoR: http://dx.doi.org/10.1002/chem.201801803
Supported by

Chemistry - A European Journal
10.1002/chem.201801803
Bismuth Phosphinates in Bi-Nanocellulose Composites and their
Efficacy towards Multi-Drug Resistant Bacteria
a
a
b
Melissa V. Werrett, Megan E. Herdman, Rajini Brammananth, Uthpala
c
c
b
b
Garusinghe, Warren Batchelor, Paul K. Crellin, Ross L. Coppel and Philip C.
a,
Andrews *
a
School of Chemistry, Monash University, Clayton, Melbourne, VIC 3800, Australia,
b
Department of Microbiology, Monash University, Clayton, Melbourne, VIC 3800
Australia.
c
Bioresource Processing Research Institute of Australia (BioPRIA), Department of
Chemical Engineering, Monash University, Clayton, Melbourne, VIC 3800, Australia.
Corresponding author email: phil.andrews@monash.edu
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Abstract
A series of poorly soluble phenyl bis-phosphinato bismuth(III) complexes
1
2
1
2
1
2
1
2
1
[
BiPh(OP(=O)R R )
2
] (R = R = Ph; R = R = p-OMePh; R = R = m-NO
2
Ph; R =
2
1
2
Ph, R = H; R = R = Me) have been synthesised and characterised, and shown to
have effective antibacterial activity against Escherichia coli (E. coli), Staphylococcus
aureus (S. aureus), methicillin-resistant Staphylococcus aureus (MRSA) and
vancomycin-resistant Enterococcus (VRE). The bismuth complexes were
incorporated into microfibrillated (nano-) cellulose generating a bismuth-cellulose
composite as paper sheets. Antibacterial evaluation indicates that the Bi-cellulose
materials have analogous or greater activity against Gram positive bacteria when
compared with commercial silver based additives: silver sulfadiazine loaded at 0.43
wt% into nanocellulose produces a 10 mm zone of inhibition on the surface of agar
2 2
plates containing S. aureus whereas [BiPh(OP(=O)Ph ) ] loaded at 0.34 wt%
produces an 18 mm zone of inhibition. These phenyl bis-phosphinato bismuth(III)
complexes show potential to be applied in materials in healthcare facilities, to inhibit
the growth of bacteria capable of causing serious disease.
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Introduction
The likelihood of being infected by multi-drug resistant bacteria is highest in hospitals
[
1,2]
and healthcare facilities.
The rise of antimicrobial resistance (AMR) means that
formerly easily treatable infections are becoming, once-again, life threatening.
Evidence indicates that contaminated surfaces in hospitals and healthcare facilities
[
3]
significantly contribute to the transmission of infection. Furthermore, invasive
medical devices or surgical procedures are common causes of hospital-acquired
[
2]
infections.
Disinfectants are used to help maintain a sterile environment and to pre-treat medical
devices; however disinfectants only have short term benefits and so these measures
[
4,5]
are not always sufficient.
Alternatively, the act of rendering specific surfaces and
[
6]
medical devices resistant to microbial colonisation, has been proposed. This has
given rise to increased interest and activity in the development of antibacterial
[
7]
polymers, leading to significant advances in the production of natural and synthetic
[
4,8–13]
antimicrobial materials and surface coatings.
Metal(loids) and metal(loid)-based antimicrobial agents using Te, Mg, As, Cu, Hg
and Ag have a rich history in medicine, and their use was prevalent well before the
[
14]
discovery of modern antibiotics in the 1920s.
The emergence of increasing
numbers of multidrug resistant bacteria, and the ever shortening times in which
resistance develops towards new natural and synthetic drugs, means research into
metal-based antimicrobials has become once again of great interest.
Thus, in recent times a large range of metal-based antimicrobial compounds have
been explored, each displaying promising activities against a range of bacteria. Some
[
15]
[16,17]
[17]
[18,19]
[18,20]
of these include compounds derived from Ru, Rh,
Ir, Sc,
In,
Cu,
[
21,22]
Co, Ni and Zn.
Gallium has been explored and utilised as an antimicrobial due to
3
+
[14,23]
its ‘trojan horse’ nature, mimicking Fe in the body,
with Ga-siderophore
Gallium nitrate has
[
14,24]
complexes showing broad-spectrum bactericidal activity.
[
25,26]
proven antimicrobial efficacy
3 3
and intravenous Ga(NO ) is currently undergoing
clinical trials in adults with cystic fibrosis, who are chronically infected with
[
27]
Pseudomonas aeruginosa.
There have also been recent reports describing the
[
28,29]
development of Ga-containing antimicrobial materials.
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Silver has been used extensively throughout recorded history and still remains one of
the most common antimicrobials used today, in both the healthcare and consumer
[
14,30]
sectors.
Ag and its compounds are often used as additives in antibacterial
[
31–36]
materials since they display broad spectrum activity, at relatively low loadings.
However, the predominance of Ag, and in particular silver nanoparticles (AgNP), in a
wide range of broad-spectrum antimicrobial products, ranging from consumer goods
to hospital grade products, has generated significant discussion recently. Concerns
exist surrounding the major knowledge gaps in AgNP toxicity, environmental
[
37],[38]
accumulation and acquired bacterial resistance.
An increasing body of opinion
[
39]
favours the need for tighter regulation of the use of AgNPs, and it may be phased
out in many industries. Consequently, there is a crucial need to find new, safe
alternatives to silver-based antimicrobial additives.
Our ongoing interest in this field involves investigations of the development and
application of bismuth complexes as antimicrobials. Despite being a heavy metal, the
[
40]
use of bismuth in medicine spans more than two centuries, and today is generally
[
41]
considered to be of low toxicity in humans. However, many bismuth complexes
[
40,42–45]
also exhibit good antibacterial activity,
including bismuth carboxylates (used
to treat Helicobacter pylori (H. pylori) infection), and certain ‘bismuth thiol’
[
40,45]
complexes which are known to inhibit biofilm formation.
Our recent studies have
shown that bismuth complexes of arene-sulfonates, amino-arenesulfonates and
[
46–48]
hydroxamates
act against H. pylori in low µM - high nM concentrations, which
is orders of magnitude better than current carboxylate based medications, while
bismuth(III) thiolates derived from small heterocyclic thiones display remarkable
broad-spectrum antibacterial activity (including against MRSA and VRE) at low µM
[
49–52]
concentrations.
1
2
Organophosphorus compounds, in particular phosphinic acids (R R P(=O)OH), are
[
53–55]
known to have antibacterial properties.
Phosphinic acids, like sulfonic acids,
have low pKa values, generate ligands which complex through their O atoms, and
would be expected to generate similarly labile complexes. While there have been
[
56–58]
several reports on organophosphinato bismuth compounds in the literature,
there
have been no reports exploring their biological activity. Since the antibacterial
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
[
59]
activity of bismuth(III) complexes are known to be profoundly ligand dependent,
there is significant potential in exploring the synergistic effects of Bi(III) and
phosphinato ligands towards multi-drug resistant bacteria, with a hope that the
activity would be similar to, if not better than, analogous sulfonato complexes.
Herein, we report the synthesis, characterisation and antibacterial activity of a series
of homo- and hetero-leptic bismuth(III) phosphinato complexes. Five phosphinic
1
2
1
2
1
acids, R R P(=O)OH: (i) diphenyl-, (R = R = Ph); (ii) bis(4-methoxyphenyl)-, (R =
2
1
2
1
R = p-OMePh); (iii) bis(3-nitrophenyl)-, (R = R = m-NO
2
Ph); (iv) phenyl-, (R =
2
1
2
Ph, R = H); (v) dimethyl-, (R = R = Me), were used to generate the heteroleptic
mono-phenyl bismuth(III) bis-phosphinato complexes: [BiPh(OP(=O)Ph
2
)
2
] (1),
(3);
[BiPh(OP(=O)(p-OMePh)
2
)
2
]
(2);
2 2 2
[BiPh(OP(=O)(m-NO Ph) ) ]
[
BiPh(OP(=O)(Ph)(H)) ] (4); [BiPh(OP(=O)Me
2
2
)
2
] (5), and the corresponding
] (6), [Bi(OP(=O)(p-
] (8), [Bi(OP(=O)(Ph)(H)) ] (9) and
2 3
) ] (10). The phopshinic acids used are shown in Figure 1 and the full
bismuth(III) tris-phosphinato complexes: [Bi(OP(=O)Ph
OMePh) ] (7), [Bi(OP(=O)(m-NO Ph)
Bi(OP(=O)Me
2 3
)
2
)
3
2
2
)
3
3
[
list of complexes and numbers are given in the Supporting Information, Table S1.
Figure 1: Phosphinic acids (i-v) used to generate the phenyl bis-phosphinato bismuth(III) complexes;
1
2
[
(
BiPh(OP(=O)R R )
2
]
and the corresponding tris-phosphinato bismuth(III) complexes [Bi
1
2
3
OP(=O)R R ) ].
All complexes were studied for their activity towards the Gram positive bacteria;
Staphylococcus aureus, (S. aureus), Methicillin-resistant S. aureus, (MRSA), and
Vancomycin-resistant Enterococcus, (VRE), and the Gram negative bacterium
(
1
2
2
Escherichia coli (E. coli). The heteroleptic complexes [BiPh(OP(=O)R R ) ] 1-5
displayed superior antibacterial activity, so they were subsequently incorporated into
microfibrillated (nano-) cellulose matrices and the antibacterial activity of these
materials established. The efficacy of the bismuth complexes was then able to be
directly compared with analogous silver preparations.
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Results and Discussion
Synthesis and Characterisation
t
3 3
Two different bismuth precursors; BiPh and Bi(O Bu) , were reacted with the
1
2
phosphinic acids to generate the series of target homo- [Bi(OP(=O)R R )
3
] and
1
2
hetero-leptic [BiPh(OP(=O)R R )
2
] bismuth(III) phosphinato complexes (Scheme
1
).
Scheme 1: General synthesis of two series of phosphinato bismuth(III) complexes,
1
2
1
2
[
2 3
BiPh(OP(=O)R R ) ] and [Bi(OP(=O)R R ) ]. SF = solvent free.
Reaction of each phosphinic acid with BiPh
3
at reflux in ethanol (or heating in
DMSO) results in the protonation and elimination to two equivalents of benzene
1
2
(
PhH) generating the heteroleptic bis-phosphinato complexes [PhBi(OP(=O)R R )
2
]
as poorly soluble white solids, (1, 3 – 5), in moderate to high yields (Table 1).
However, following this simple solvent-based approach the reaction with bis(4-
methoxyphenyl)phosphinic acid (ii) invariably produces mixtures of the mono and
bis-substituted phosphinato complexes. Therefore, the target complex
[BiPh(OP(=O)(p-OMePh)
2 2
) ] 2 was synthesised using a solvent-free approach
whereby BiPh and the acid in a 1:2 stoichiometry were ground together and heated in
3
a flask in a Kugelrohr oven at 130 °C for 4 hours. The optimal temperature of the
reaction was determined from the thermochemical trace obtained by differential
scanning calorimetry (DSC) on a small sample of the reaction mixture. The resulting
solid was washed with a small amount of hot ethanol to remove any unreacted
materials.
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
t
3
Reaction of each acid with Bi(O Bu) (from -78 °C to room temperature) produced the
1
2
homoleptic complexes [Bi(OP(=O)R R )
3
] (6 – 10) in moderate to high yields (Table
1
). Complexes 6 and 10 have been reported previously, but were synthesised using a
3
metathesis approach, using BiCl and the sodium salt of the corresponding phosphinic
[
57]
acid.
Using our approach all the products are easily isolated, as white powders or
microcrystalline solids, by simple filtration and subsequently washed with a small
amount of ethanol to remove unreacted material. The physical properties of all
complexes are shown in Table 1.
Thermogravimetric analysis and differential scanning calorimetry (TGA-DSC) was
used to establish the melting and decomposition temperatures of the complexes.
Compounds 1-5 melt in the range of 197 °C (4) to ≥ 300 °C (1,5) whereas 6-10 have
melting points in the range of 130 °C (9) to >300 °C (6, 10). Temperatures for the
onset of thermal decomposition of all complexes are outlined in Table 1.
Table 1: Summary of physical properties of the synthesised bismuth complexes 1-10. Melting and
decomposition temperatures determined by differential scanning calorimetry and thermal gravimetric
analysis.
Yieldꢀ
%)ꢀ
Meltingꢀ
Decompositionꢀpointꢀ
(°C)ꢀ
BismuthꢀComplexꢀ
Colourꢀ
(
pointꢀ(°C)ꢀ
1
2
3
4
5
6
7
8
9
1
ꢀ
[BiPh(OP(=O)Ph
[BiPh(OP(=O)(p-OMePh)
[BiPh(OP(=O)(m-NO Ph)
[BiPh(OP(=O)(Ph)(H)) ]ꢀ
[BiPh(OP(=O)(CH ]ꢀ
[Bi(OP(=O)Ph ]ꢀ
[Bi(OP(=O)(p-OMePh)
[Bi(OP(=O)(m-NO Ph)
[Bi(OP(=O)(Ph)(H)) ]ꢀ
[Bi(OP(=O)(CH ]ꢀ
2
)
2
]ꢀ
90ꢀ
72ꢀ
88ꢀ
78ꢀ
57ꢀ
69ꢀ
76ꢀ
87ꢀ
67ꢀ
58ꢀ
whiteꢀ
whiteꢀ
creamꢀ
whiteꢀ
whiteꢀ
whiteꢀ
whiteꢀ
creamꢀ
whiteꢀꢀ
whiteꢀ
>300ꢀ
250ꢀ
330ꢀ
420ꢀ
320ꢀ
197ꢀ
300ꢀ
450ꢀ
400ꢀ
260ꢀ
170ꢀ
330ꢀ
ꢀ
2
)
2
]ꢀ
ꢀ
2
2
)
2
]ꢀ
280ꢀ
ꢀ
2
197ꢀ
ꢀ
3
)
2
)
2
>300ꢀ
>300ꢀ
230ꢀ
ꢀ
2 3
)
ꢀ
2
)
3
]ꢀ
ꢀ
2
2
)
3
]ꢀ
260ꢀ
ꢀ
3
130ꢀ
0ꢀ
3
)
2
)
3
>300ꢀ
Complexes 1-10 display limited solubility in common organic solvents including;
ethanol, dichloromethane, tetrahydrofuran, toluene, and dimethylsulfoxide (DMSO).
They are also poorly soluble in water with the exception of complexes 5 and 10 which
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
show a higher level of water solubility, with evidence that complex 5 undergoes a
degree of hydrolysis. Given our understanding of the chemistry of bismuth
compounds, it is almost certain that this generalised very low solubility results from
[
57]
the complexes being polymeric.
Due to their sparingly soluble nature the composition of each complex was confirmed
using solid-state NMR spectroscopy and elemental analysis (C, H, N, P and Bi).
These results are provided in the Experimental section and in the ESI, Table S1.
Deprotonation of the phosphinic acid and coordination of the phospinato ligand to the
3
1
Bi(III) centre leads to a general high field shift in the P signals in the HPDEC-MAS
[
57]
spectra, which is in agreement with reports of other phosphinato Bi(III) species.
3
1
For example, the P signal for bis(4-methoxyphenyl)phosphinic acid (ii) is 25.2 ppm,
whereas complexes 2 and 7 give chemical shifts of 24.6 ppm and 23.8 ppm
respectively.
The FT-IR spectra of complexes 1-10 are all consistent with coordination of the
phosphinato ligands to the bismuth centre. The broad, medium intensity O-H bands
-
1
present above 2000 cm in the spectra of the phosphinic acids are no longer observed
in either series of the bismuth complexes. The characteristic broad band around 1660
-
1
[60]
cm in (most) of the phosphinic acids is also absent in the bismuth complexes, and
are instead replaced by the appearance of characteristic PO asymmetric and
symmetric stretching bands typically observed for salts of phosphinic acids (R R PO
2
1
2
-
2
[
57,60]
-1
)
.
For example, the (P)O-H stretching band at 1587 cm in diphenylphosphinic
acid (i) is not observed in either [BiPh(OP(=O)Ph ] 1 or [Bi(OP(=O)Ph ] 6.
Instead, very strong signals are observed at 1112 and 1013 cm
2
)
2
2 3
)
-
1
-
1
(
2 2 2 3
[BiPh(OP(=O)Ph ) ]) and 1091 and 1009 cm ([Bi(OP(=O)Ph ) ]) and are likely the
[
57]
2
asymmetric and symmetric PO stretching bands; indicative of bismuth binding.
Solubility and Stability
On first inspection, all the complexes appear insoluble in a large range of typical
solvents. However, this is not the case and they have been found to have some very
limited solubility in some solvents, particularly water, DMSO, and ethanol.
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
To establish the solubility of the complexes in water approximately 50 mg was added
to water, vortexed and maintained at 25 ºC overnight. The samples which appear as
cloudy white suspensions were then vortexed again and filtered through 0.2 µm
syringe filters. Approximately 3 mL aliquots of the solution were transferred into vials
and the solvent was removed. The mass of the remaining residue was determined and
the solubility of 1 g of complex in water was determined by extrapolation from the
initial aliquot mass. The results are summarised in Table 2.
1
2
It can be clearly seen that the bis-phosphinato complexes [BiPh(OP(=O)R R )
2
] are
less water soluble when compared to their analogous tris-phosphinato complexes
1
2
[
Bi(OP(=O)R R )
g of 1 ([BiPh(OP(=O)(Ph
([Bi(OP(=O)(Ph ]) (9.4 kg).
3
] (Table 2). For example, the mass of water required to dissolve 1
2 2
)
]) is greater (16.3 kg) than that required to dissolve 1 g of
6
2 3
)
Table 2: Experimental solubility data and calculated clogP values for bismuth complexes 1-10. clogP
Values were calculated using ChemDraw Professional 15.0. Bismuth complexes listed in order of least
to most water soluble, as determined experimentally.
Bismuthꢀ
complexꢀ
Massꢀ(kg)ꢀofꢀwaterꢀneededꢀtoꢀ
clogPꢀvalues*ꢀ
dissolveꢀ1ꢀgꢀofꢀcomplexꢀ
2
1
7
6
3
4
8
9
5
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
18.4ꢀ
16.3ꢀ
13.0ꢀ
9.4ꢀ
6.14ꢀ
6.46ꢀ
6.51ꢀ
7.00ꢀ
5.43ꢀ
1.94ꢀ
5.46ꢀ
0.22ꢀ
-0.48ꢀ
-3.87ꢀ
7.3ꢀ
3.9ꢀ
1.6ꢀ
1.5ꢀ
0.64ꢀ
0.22ꢀ
1
0ꢀ
silverꢀ
1
2.0ꢀ
-1.32ꢀ
sulfadiazineꢀ
*
Note: these complexes likely exist in a polymeric form, which increases their overall calculated clogP
value, however their relative order is expected to remain consistent.
3
Another aliquot (3 mL) was also taken and diluted in 3 % HNO and analysed for Bi
content using ICP-OES. The ICP data revealed that for complexes 1-3 and 6-8 the
amount dissolved in the aliquot was extremely low, < 5 ppm (µg/mL). The bismuth
content detected for complexes derived from phenylphosphinic and
9
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
dimethylphosphinic acid (4, 5, 9 and 10) was notably higher. The phenylphosphinato
complexes (4 and 9) had bismuth content averaging 150 ppm and 350 ppm
respectively. It is noted that there is a difference in solubility between the two forms
1
2
of complex, with the homoleptic [Bi(OP(=O)R R )
3
] complex having greater
1
2
solubility than the heteroleptic [BiPh(OP(=O)R R )
2
] form. This trend was also
observed for the dimethylphosphinato complexes 5 and 10, with average bismuth
concentrations being 2000 ppm and 4000 ppm respectively. It appears that as the
steric bulk of the complex decreases, the complexes show greater solubility. The
observed solubility trends correlate with the calculated clogP values for the
complexes and are relatively consistent with the experimental solubility data (Table
2
). The relative clogP values as well as the observed solubility data indicate that 4 and
would exhibit higher hydrophilicity, compared to 1-3.
5
1
The residues from the solubility experiments were also analysed by H NMR. It must
be noted that ‘solubility’ of the complexes, and hence higher Bi concertation in water,
1
may be due to their instability in water. H NMR spectral analysis on the solubilised
component from the addition of complex 5 to water shows more than one species
present. Rather than observing the single expected bis-phosphinato complex
[
BiPh(OP(=O)(CH
also be identified (ESI, Figure S1). And based on the increased integration of the CH
signals relative to the aromatic protons, it is also likely that there is some tris-
3 2 2 2 3 2
) ) ], the mono-phosphinato species [BiPh (OP(=O)(CH ) )] could
3
3
1
dimethylphosphinato bismuth 10 present. P NMR confirms that the increased
integration of the CH signals is unlikely to be from free phosphinic acid.
In the case of the 10 ([Bi(OP(=O)(CH ]), which has the highest degree of water
O both show that the complex remains intact,
3
3 2 3
) )
1
31
solubility, H and P NMR spectra in D
2
with no evidence of rearrangement or decomposition (ESI, Figures S2 and S3).
Antibacterial activity
The complexes were screened for their antibacterial activities against a range of Gram
negative and Gram positive bacteria including E. coli, S. aureus, MRSA and VRE.
Zone of inhibition tests were undertaken to determine the susceptibility of the bacteria
to the suite of bismuth complexes. Due to the very low solubility of the complexes (as
well as hydrolytic instability of 5), a qualitative solid-state antibacterial assay was
employed as a first analysis.
1
0
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Bacteria were prepared as described in the Experimental section. Samples (100 µL) of
turbid, overnight cultures of bacterial cells (E. coli, S. aureus, MRSA and VRE) were
spread onto agar plates. Sterile pipette tips were used to transfer a small quantity (< 1
mg) of the solid complex onto the agar; the pipette tip (with the complex) was gently
pierced into the agar. The plates were incubated at 37 °C overnight to allow bacterial
replication and any zone of inhibition visualised the next day. These tests were carried
1
2
out in triplicate for each bacterial strain. It was clear that the [BiPh(OP(=O)R R )
2
]
complexes 1-5 displayed significant antibacterial activities while the
1
2
[
Bi(OP(=O)R R )
are not susceptible to the homoleptic tris-phosphinato bismuth complexes. Figure 2
panels 1 and 2) shows representative agar plates from the triplicate analyses carried
3
] complexes 6-10 did not. This indicated that these bacterial strains
(
out across four bacterial strains and the ten bismuth complexes (1-10). (ESI, Figures
S4 and S5 for additional images).
MRSA: 1-5
MRSA: 6-10
MRSA: i-v
S. aureus
Figure 2: Zone of inhibition test against MRSA, using <1 mg of the complexes, 1-5, 6-10 and
phosphinic acids i-v (panels 1-3 respectively). Zone of inhibition test against S. aureus using known
3 3 3
bismuth compounds; BiPh , Bi(NO ) and complex 1 (panel 4).
To ensure antibacterial activity was not simply due to the phosphinato ligand (or free
3
+
+
acid) or the Bi /BiO ion alone, control assays were conducted on the parent
phosphinic acids, BiPh and Bi(NO . Figure 2 (panels 3 and 4) shows representative
3
3 3
)
agar plates, showing no evidence of inhibitory activity for either the phosphinic acids
or bismuth salts at comparable masses.
Taken together, these results provide some insight into the necessary structure-
1
2
2
property relationship of the bismuth(III) complexes. Since the [BiPh(OP(=O)R R ) ]
1
2
3
complexes display superior antibacterial activity to [Bi(OP(=O)R R ) ], it appears
1
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
vital for the active complexes to have either one aryl group connected directly to the
bismuth centre, or for the complexes to be simply heteroleptic. It must be noted that
the antibacterial activity cannot simply be related to solubility as the homoleptic
complexes are more water soluble than the heteroleptic (Table 2), however display
inferior antibacterial activity. It is not yet clear why this is the case but it mirrors
results we have obtained on the comparative antibacterial activity of a large number
[
50]
of bis-thiolato [BiPh(SR)
2
] and tris-thiolato [Bi(SR)
3
] complexes.
Further, those
studies have shown that while the phenyl group proved important, similar
antimicrobial activity can be obtained by creating heteroleptic tris-thiolato complexes
[
50]
2
[Bi(SR’)(SR) ].
The mechanism of action of Bi(III) complexes on bacteria is not well known.
However, recent investigations have started to uncover some mechanisms by which
bismuth is acting on the bacteria, particularly through studying the effects of typical
bismuth drugs on H. pylori. It has been found that bismuth can interfere with a range
of proteins, in particular Zn(II) and Fe(III) regulating proteins, cause cytoplasmic
degradation, lead to the formation of Bi-glycoproteins, and act on metallo-enzymes
[
61]
such as urease and alcohol dehydrogenase.
A recent breakthrough study by the
group of Sun and co-workers describes how some known bismuth antimicrobial drugs
[
62]
can act as broad-spectrum metallo-β-lactamase inhibitors.
Mammalian Cell Toxicity
Preliminary toxicity testing on the parent phosphinic acids (i-v) and complexes 1-5
1
2
(
[BiPh(OP(=O)R R )
kidney-derived). Cytotoxicity was assessed using the MTT-based toxicology assay
Sigma–Aldrich; TOX1 assay kit) according to the manufacturer’s instructions.
Compounds were however added as solid powders to cultured COS-7 cells at 0.5 and
.0 mg/mL. Preliminary results indicate that complexes 1, 4 and 5 displayed low
2
]) was performed against COS-7 eukaryotic cells (monkey
(
1
toxicity at 1.0 mg/mL. COS-7 cells which had been incubated with those complexes
had a normalised cell viability of at least 60 % that of blank or control COS-7 cells to
which no complex had been added. Complexes 1-5 showed no cytotoxicity when
incubated at a concentration of 0.5 mg/mL (ESI, Figure S6).
Bismuth nanocellulose composites
1
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Due to the low solubility of the complexes, we propose that they can be used as
additive antibacterial agents, to a range of different polymeric materials with potential
applications in surface coatings, packaging and/or medical devices. To assess the
viability of this, we have incorporated the active bismuth complexes 1-5 into
microfibrillated cellulose (MFC), a form of nanocellulose.
Generally, when preparing additives for MFC incorporation they are suspended in
water. However, due to the aggregation of the complexes in water, they were
suspended in a 20 % isopropanol/water solvent mixture and stirred. This was
combined with a cationic polyacrylamide (CPAM) suspension, to promote retention
[
63,64]
of the bismuth complexes into the cellulose nanofibre matrix.
Finally, the
bismuth-CPAM suspension was combined with the MFC and poured through a
British hand-sheet maker and filtered by gravity. The sheet was removed from the
filter paper and dried at 105 °C, to produce a sheet of paper with a consistent diameter
of 15.9 cm. Complex 1, as a representative Bi agent, was used to make a range of
paper sheets with different %Bi loadings, targeted bismuth complex to MFC ratios
ranging from 0.1- 20 wt%.
The sheets were analysed using scanning electron microscopy (SEM) to establish the
presence of bismuth in the sheet (in particular at the surface) and to gain an
understanding of the dispersion of the complex within the cellulosic matrix.
It is evident from the SEM images that the bismuth complexes are entangled in the
cellulose network, Figure 3 (A1 and B1). The bismuth phosphinate particles are
clearly distinguishable by their needle like shape however backscattered electron
images were used to probe the dispersion of the particles in the sheet. Figure 3 (A2
and B2) clearly show that the bright bismuth-containing particles are fairly well
dispersed throughout the cellulose.
1
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Figure 3: SEM images of a sample of the bismuth paper; 15.94 wt% loading of complex 1. Panels A1
and B1 are secondary electron images. Panels A2 and B2 are backscattered electron images, where the
bright spots are indicative of the complex 1. Scale bars: 10 µm.
Energy dispersive X-ray spectroscopy (EDX) was used to further confirm the
presence and identity of the bismuth complex in the paper. The analysis revealed that
the small crystalline solids (~ 1µm in diameter) observed in the paper contained both
Bi and P, (Figure 4 and S7). Since SEM/EDX analysis only provided qualitative data,
ICP-OES was used to quantify the amount of Bi retained in the paper, as well as
confirm dispersion.
1
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Figure 4: EDX spectrum of a sample of bismuth paper: 15.94 wt% loading of complex 1.
Signals of interest labelled. Ir is also present in the spectrum due to sample coating. EDX
spectrum: x-axis label, keV and y-axis label, cps/eV. Spectrum taken from the area shown in the SEM
image inset. Scale bar 1 µm.
3
The cellulosic based Bi-paper is not readily decomposed in concentrated HNO so
samples were ashed in an oven at 600 °C for two hours to remove the organic
cellulose and leave any remaining bismuth containing residue. After ashing, the solid
3
residue was dissolved in concentrated HNO and diluted for ICP-OES analysis.
Table 3: Comparison of the targeted versus the actual bismuth loading for several bismuth-cellulose
sheets. The calculated loadings were determined using ICP-OES. Final values are averages (where
n=3) showing the (sample) standard deviation.
Actualꢀloadingꢀofꢀbismuthꢀcomplexꢀ
Bismuthꢀ
Complexꢀ
Targetꢀloadingꢀofꢀbismuthꢀ
complexꢀ(wt%)ꢀ
usingꢀICP-OESꢀ±ꢀstandardꢀdeviationꢀ
ꢀ
(
wt%)
1
1
1
1
1
1
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
0.1ꢀ
0.49ꢀ
0.96ꢀ
4.87ꢀ
9.10ꢀ
13.04ꢀ
0.05ꢀ±ꢀ0.02ꢀ
0.34ꢀ±ꢀ0.00ꢀ
0.64ꢀ±ꢀ0.00ꢀ
4.06ꢀꢀ±ꢀ0.19ꢀ
8.20ꢀ±ꢀ0.13ꢀ
12.31ꢀ±ꢀ0.15ꢀ
1
ꢀ
16.67ꢀ
15.94ꢀ±ꢀ0.06ꢀ
The ICP-OES analysis indicates that all papers produced have loadings consistently
lower than expected (Table 3). Detailed calculations can be found in the ESI, Table
S2. This can be attributed to experimental error and loss of compound upon transfer to
the hand sheet maker. The relatively low standard deviations between samples taken
from different areas of the same sheet indicates the dispersion of the bismuth complex
within the sheet. We can therefore calculate the amount of Bi-complex in each disc
used for antibacterial testing, discussed below.
Antibacterial Testing of Bi-cellulose composites
Zone of inhibition tests were carried out on 5 mm discs taken from the bismuth
sheets, with loadings ranging from 0 - 16 wt% complex 1. Each of the bismuth-loaded
discs produced a clear zone, as shown in Figures 5, 6 and S8 (ESI). The Gram
1
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
positive bacteria are more susceptible to the bismuth impregnated paper, as inferred
by the relatively larger zones of inhibition.
It is interesting to note that as the bismuth loading increases up to 16 wt%, the zone of
inhibition (mm) does not proportionally increase, meaning that at a lower wt% a
saturation point is reached. This must correlate with the rate of diffusion of the
bismuth complex into the agar.
MRSA
E. coli
Figure 5: Zone of inhibition tests against MRSA and E. coli using the bismuth paper with calculated
loadings (wt%) of complex 1: 0% (0, blank), 4.06% (A), 8.20% (B), 12.31% (C) and 15.94% (D).
Figure 6: Zone of inhibition tests against S. aureus, VRE, MRSA and E. coli using the bismuth paper
with calculated loadings (wt%) of complex 1: 0% (0, blank), 4.06% (A), 8.20% (B), 12.31% (C) and
1
5.94% (D). Measurement values are averages ± (sample) standard deviation, where n=3. Anywhere
on the graph where zero (0) appears, no zone was observed.
1
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
To confirm the presence of Bi and establish its concentration within the ‘zone of
inhibition’ ICP-OES was used to analyse samples of agar taken from within and
outwith the zones for a range of samples. Although the concentrations detected were
at the edge of the detection limits of the instrument (0.01 ppm Bi), it was clear that
there was a difference in the signal for Bi detected in- and outside the zones. The Bi
intensities detected at 223.061 nm for samples taken within the zones average 7783
cps (approx. 0.02 ppm) compared to the intensities detected for samples taken from
areas on the agar plate where bacteria grew (4166 cps, <0.01 ppm). This confirms
diffusion of the complex into agar and the presence of bismuth as being the cause of
the antibacterial activity. Figures S9 and S10 in the Supporting Information, show the
detailed results of this study.
Sampleꢀ
Zoneꢀ(mm)ꢀ
0
ꢀ
0ꢀ
9ꢀ
Agꢀ
Aꢀ
Bꢀ
7ꢀ
11ꢀ
14ꢀ
20ꢀ
Cꢀ
Dꢀ
Figure 7: Zone of inhibition tests against VRE using the bismuth paper with calculated loadings (wt%)
of complex 1: 0% (0, blank), 0.05% (A), 0.34% (B), 0.64% (C) and 4.06% (D). Ag: silver sulfadiazine
at 0.43 wt%.
To determine the lowest active loading, sheets containing 0.05, 0.34, 0.64, and 4.06
wt% of 1 were produced and tested against the range of bacteria. Zone of inhibition
experiments identified that there was still observable activity at the 0.34 wt% loading
of 1 (Figure 7, B). To allow a comparison with silver, the established antibacterial
[
65,66]
compound silver sulfadiazine,
was used to generate analogous silver-loaded
cellulose. ICP-OES on the Ag-paper indicates that the loading was 0.43 ± 0.04 wt%.
This equates to approximately 0.008 mg of silver sulfadiazine per disc. Figure 7
shows that VRE is more susceptible to the bismuth paper compared to that containing
the antimicrobial silver additive. For all bacterial strains except E. coli the bismuth
containing paper outperformed the silver standard, as show in the Supporting
Information (Figure S11). This agrees with results published by Luan et al. in
[
67]
2
2
012. At loadings of 23.6 mg per 100 cm of bacterial cellulose, samples of 6 mm
1
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
diameter produced zones of 11 mm against S. aureus. Our bismuth complex (1) was
2
loaded at just 6.2 mg per 198 cm of nanocellulose and produced zones of inhibition
of 18 mm against S. aureus (Figure S11, Supporting Information). The solubility of
silver sulfadiazine was determined (using the same method as employed for
complexes 1-10) to ensure that results observed in Figure 7 were not simply due to
differences in the solubility. In fact, the silver sulfadiazine had a similar solubility to
that of complex 1 (Table 2).
In 2014, Wu et al. reported the antibacterial activity of a series of silver containing
bacterial cellulose samples. The bacterial cellulose sample loaded with 1.01 wt% Ag
produced small zones of inhibition against E. coli, S. aureus and P. aeruginosa. The
zone sizes reported were all below 8 mm (6.5 mm cellulose sample size + 1.5 mm
[
68]
growth inhibition ring). Comparatively, our samples at loadings of 0.64 wt% of 1
produced larger zones of inhibition, 20 mm, against S. aureus (Figure S11, Supporting
Information). However, the silver bacterial cellulose samples outperformed our
bismuth loaded samples against E. coli. Complex 1 is required at loadings of at least
4
.06 wt% in cellulose for any zone of inhibition to be observed against the Gram
negative bacteria, E. coli (Figure 5 and Figure S11, Supporting Information).
The lowest active loading of 1 in the cellulose samples was identified as 0.34 wt%,
therefore a series of cellulose papers were made using the active bismuth complexes
2
-5 at a targeted bismuth complex:MFC ratio of 0.5 wt%. ICP-OES was again used to
confirm the actual loading of the bismuth complexes, summarised in Table 4.
As can be seen from Table 4 complexes 4 and 5 did not incorporate into the cellulose
as effectively as 1-3, most likely due to their slightly higher solubilities in the large
amounts of water used in the sheet making process. We can also attribute some loss of
compound upon transfer of the cellulosic-bismuth mixture to the hand sheet maker.
1
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Table 4: Comparison of the targeted versus actual bismuth loading for several bismuth-cellulose
sheets. The calculated loadings were determined using ICP-OES. Final values are averages (where
n=3) showing the (sample) standard deviation.
ꢀ
Bismuthꢀcomplexꢀ
Targetꢀloadingꢀofꢀ
Actualꢀloadingꢀofꢀbismuthꢀ
complexꢀusingꢀICP-OESꢀ±ꢀ
standardꢀdeviationꢀ(wt%)ꢀ
0.34ꢀ±ꢀ0.00ꢀ
1
2
[
BiPh(OP(=O)R R )
2
]ꢀ
bismuthꢀcomplexꢀ
(wt%)ꢀ
0.49ꢀ
ꢀ
1
2
3
4
5
ꢀ
ꢀ
ꢀ
ꢀ
ꢀ
0.49ꢀ
0.37ꢀ±ꢀ0.06ꢀ
0.50ꢀ
0.31ꢀ±ꢀ0.02ꢀ
0.50ꢀ
0.21ꢀ±ꢀ0.01ꢀ
0.50ꢀ
0.18ꢀ±ꢀ0.01ꢀ
Paper containing complex 4 was remade, using a higher targeted loading (1 wt%),
which resulted in paper with 0.43 ± 0.03 wt% 4. Paper incorporating 5 at a higher
loading was not targeted due to the significant hydrolytic instability observed.
Antibacterial testing was again carried out using the zone of inhibition assay. Results
show that the low loading of complex 5 (0.18 wt%) is not enough to produce an
antibacterial effect, with no zones of inhibition observed (Figures 8 and S12).
Figure 8: Antibacterial activity of different bismuth papers against, S. aureus, VRE, MRSA and E.
coli. Bismuth paper loaded with complexes 1-5 (as indicated by the numbers on the x-axis). Numbers
in brackets indicate the wt% loading of that particular complex in the cellulose. Measurement values
are averages ± (sample) standard deviation, where n=3. Anywhere on the graph where zero (0)
appears, no zone was observed.
1
9
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Based on the ICP-OES analysis above, the weight (mg) of bismuth complex within
each of the 5 mm paper discs used in the assays was calculated. The calculations
assume a homogenous distribution of the complex within the cellulose network. For
the cellulose sheets made with complexes 1-4 it is estimated that each disc contains
0
.005-0.007 mg of the bismuth complex. To have significant zones of inhibition at
such low loadings is an indication of the potency of these complexes against Gram
positive bacteria.
That the bismuth complexes presented here demonstrate greater antibacterial
efficiency against the Gram positive bacteria over Gram negative (E. coli) is not
unexpected. Our previous studies, as well those from Kotani et al. on cyclic
[
69]
organobismuth(III) compounds have shown similar trends and it is often thought
that the double membrane of Gram negative bacteria can translate to lower
[
69]
permeability and therefore lower activity.
Although these lower loadings of the
1
2
[
BiPh(OP(=O)R R ) ] complexes were not enough to give zones of inhibition against
2
E. coli, Figures 5 and 6 show that E. coli is susceptible to the complexes at
significantly higher loadings of 4.06 wt% which corresponds to approximately 0.051
mg of complex 1 per disc, or about ten times more than required for activity against
the Gram positive bacteria.
Homoleptic versus heteroleptic complexes
1
2
2
It is still unclear why the heteroleptic [BiPh(OP(=O)R R ) ] type complexes
outperform the homoleptic analogues as antibacterial agents. Although the initial
qualitative stab tests clearly indicated no activity for the homoleptic complexes, paper
samples containing complexes 6 and 9 were prepared for a direct quantitative
comparison.
2 3 3
Complexes 6 [Bi(OP(=O)Ph ) ] and 9 [Bi(OP(=O)(Ph)(H)) ] were incorporated into
cellulose in the same manner as their corresponding bis-phosphinato bismuth
complexes 1 and 4 (details in the experimental section). ICP-OES determined the
overall loading of the complex to be 3.87 wt% 6 and 0.42 wt% 9.
2
0
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Preliminary results showed that although paper containing 6 gave small haloes against
MRSA and S. aureus, no clear zones could be observed (Figure S13, Supporting
Information). Comparatively, cellulose loaded with 1 at 4.06 wt% gave clear zones of
1
6 mm against MRSA and 25 mm against S. aureus. Likewise, the paper loaded with
(0.42 wt%) elicited no effect. The same loading of the analogous bis-phopshinato
9
complex 4 (0.43 wt%) in cellulose, produced zones of 9.67 and 11.67 mm against
MRSA and S. aureus, respectively.
The activity of the tris complexes 6 and 9 are not equivalent to their bis-phosphinato
analogues. The difference in activity observed between the homo- and hetero-leptic
complexes is not simply a diffusion or solubility issue, as the less active homoleptic
1
2
3
complexes ([Bi(OP(=O)R R ) ]) demonstrate increased hydrophilicity (Table 2).
Conclusion
1
2
3
We have synthesised and characterised a series of ten homoleptic [Bi(OP(=O)R R ) ]
1
2
and heteroleptic [BiPh(OP(=O)R R )
2
] phosphinato bismuth(III) complexes, derived
t
from treatment of a series of phosphinic acids with either Bi(O Bu)
3
or BiPh
3
. All the
complexes have low solubility in common organic solvents and water. The
heteropleptic complexes proved to have potent antibacterial activity against a range of
bacteria (S. aureus, MRSA, VRE, E. coli), while the homoleptic complex analogues
proved to be inactive. The low solubility of the heteroleptic complexes led to their
incorporation into microfibrillated (nano-) cellulose to generate bismuth-impregnated
paper composites. Using a disc diffusion assay the bismuth paper was found to have
antibacterial activity at Bi-complex loadings of 0.34 wt % (0.005-0.007 mg of
bismuth complex per disc) for Gram positive bacteria and at 4.06 wt% (0.051 mg
bismuth complex per disc) for E. coli (Gram negative). In our study, this was more
effective than analogous silver impregnated sheets. ICP-OES analysis of the agar
zone-of-inhibition indicated the presence of Bi within the zone, at very low levels of
0
.02 ppm, and not outside, thereby defining Bi as the antibacterial agent.
These studies confirm the ability of the complexes to maintain antibacterial activity
within a foreign matrix, in this study cellulose - a natural polymer. The high thermal
2
1
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
stability of the heteroleptic complexes means there is a real potential for them to be to
be incorporated into high-melt synthetic polymers, thereby generating new
antimicrobial materials and coatings.
2
2
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Experimental
General Considerations
Infrared spectra were recorded on an Agilent Technologies Cary 360 FTIR
-
1
spectrometer in the range of 4000-700 cm . IR frequencies are reported in
wavenumbers with the relative intensities as strong (s), medium (m) or weak (w).
Shoulders (sh) and broad bands (br) are also reported. Mass spectrometry (ESI) was
performed on a Micromass Platform QMS spectrometer with an electrospray source
and a cone voltage of 35 eV. Elemental microanalyses were performed by the
Campbell Microanalytical Laboratory, Department of Chemistry, University of
Otago, Dunedin, New Zealand. Melting points were measured on a Stuart Scientific
SMP10 digital melting point apparatus. Differential scanning calorimetry (DSC)
measurements carried out using a TA Q100 DSC between 40 °C and 250 °C, with a
temperature ramp rate of 10 °C per minute. Thermogravimetric analysis (TGA) and
DSC was carried out on was carried out using a Mettler Toledo TGA/DSC 1 Star
System.
Solid state NMR spectra were recorded on a Bruker Avance 300 (7.05 Tesla magnet)
with a 4 mm multinuclear solid state probe at room temperature (ca. 24 °C). Solid
samples were packed into 4 mm ZrO
2
rotors with KeI-F caps and spectra were
1
3
recorded using cross polarization with magic angle spinning; CP-MAS ( C) or high-
3
1
power decoupling with magic angle spinning; HPDEC-MAS ( P) techniques.
1
3
C CP at 100.6 MHz. 10000 Hz spin rate, 50000 Hz SW, acquisition time of 20 ms,
992 data points collected with a 4 s delay and a 2 ms contact time. Spectra were
1
referenced to an external glycine standard.
3
1
P HPDEC at 121.5 MHz. 10000 Hz spin rate, 100000 Hz SW; acquisition time of 41
ms, 8192 data points collected with a 5 s delay. Spectra were referenced to an external
ammonium phosphate standard.
1
13
31
Solution state NMR spectra (consisting of H, C, P and 2D COSY experiments)
were recorded using a Bruker Avance III 400 spectrometer with a 9.4 Tesla magnet
1
13
31
(
400.13 MHz for H, 100 MHz for C and 161.98 MHz for P) at room temperature.
The spectra were recorded with either a 5 mm broadband autotunable probe with Z-
gradients and a BACS 60 tube autosampler or, with a 5 mm broadband inverse probe
2
3
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
1
13
with Z-gradients and a 5 mm Quad switchable probe. H and C chemical shifts were
3
1
referenced to residual solvent resonances and for P spectra, an external standard of
5% H PO was used.
8
3
4
ICP-MS for Bi analysis
Bismuth analyses were carried out using a GBC Optimass 9500 ICP Time of Flight
MS at Monash University’s School of Earth, Atmosphere and Environment. Stock
sample solutions were prepared by dissolving a known amount of complex (1-10) in
concentrated HNO
3
and diluted. Samples for analysis were prepared by further
. Raw count rates from the analyses were
dilution of the stock, using 3% HNO
3
externally standardised by means of a calibration curve constructed using a
commercially available Bi stock solution. Drift corrections were applied by the use of
In as internal standard and the variability of standard solutions interspersed with the
unknowns. Precision is of the order of 1.3% RSD from the single analyses. The final
%Bi calculation was averaged from three replicates and total error calculated.
ICP-OES for determination of Bi/Ag content in the cellulose samples
Determination of the Bi content of the cellulose sheets was carried out using a Perkin
Elmer Avio 200 ICP-OES, measuring Bi at 223.06 nm. A hole punch was used to
collect small discs from the cellulose sheets. A known number (and mass) of discs
were placed into crucibles and heated in an oven from 25-600 °C over four hours and
then held at 600 °C for two hours. This process was used to remove the bulk of the
organic material, enabling the bismuth based residue to be easily dissolved in
3
concentrated HNO . The dissolved residue was diluted using a known mass of 3%
HNO . Ag samples were prepared in the same way and measured at 328.068 nm. Raw
3
count rates from the analyses were externally standardised by means of a calibration
curve constructed using a commercially available Bi stock solution or the Instrument
Calibration Standard 2 (containing Ag, part number N9301721). Precision is of the
order of 0.6% RSD from the single analyses. The final wt% Bi or wt% Ag calculation
was averaged from three samples and sample standard deviation calculated.
2
4
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy
EDX)
(
Samples analysed by SEM and EDX were mounted, using carbon adhesive discs, onto
an SEM specimen stub and coated with iridium. Analyses were carried out on a JEOL
JSM-7001F FEGSEM with an Oxford Instruments AZtec X-ray analysis system.
Samples were studied using 15 kV accelerating voltage and the data was processed
using Oxford AZtec software.
Synthesis
THF and diethyl ether were dried prior to use by M-Braun-SPS-800 solvent
purification system and molecular sieves (4Å) were used to store. All moisture
sensitive reactions were conducted with oven dried glassware under an atmosphere of
dry nitrogen using a vacuum/nitrogen line and Schleck techniques. n-pentane was
distilled over sodium.
BiPh
3
was synthesised through a standard Grignard metathesis reaction by the
with PhMgBr in dried diethyl ether at 0 °C, and subsequently
treatment of BiCl
3
[
70]
recrystallised from ethanol.
t
Bi(O Bu)
3
was synthesised through a standard metathesis reaction from the treatment
t
of BiCl
3
with KO Bu in dried THF at 0 °C. The product was extracted with dried n-
[
71]
pentane using a Soxhlet extractor.
All phosphinic acids were purchased from
Sigma-Aldrich and used as received without any further purification, unless otherwise
stated.
bis(3-Nitrophenyl)phosphinic acid (iii)
[
72]
The synthesis was adapted from previously published methods.
Diphenylphosphinic acid (0. 712 g, 3.26 mmol) was slowly added to a solution of
concentrated HNO (2 mL) and concentrated H SO (4 mL) stirring in an ice bath.
3
2
4
After the addition, the yellow solution was allowed to stir at room temperature
overnight. The reaction was poured over ice to yield a white precipitate in a yellow
solution. The solid was collected by filtration over a frit and washed with water until
[
73]
the filtrate became colourless. Yield: 0.442 g, 44 %, m.p. 268 °C (Lit. 265, dec).
2
5
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
1
6
H NMR (δ, ppm, DMSO-d ): 8.51 (2H, d, J = 12.4 Hz, CH), 8.38 (2H, d, J = 8Hz, N-
C-CH-CH), 8.19 (2H, t, J = 9.4 Hz, P-C-CH-CH), 7.79 (2H, td, J = 8, 3.2 Hz, N-C-
1
3
6
CH-CH), 5.29 (1H, s (br), OH). C NMR (δ, ppm, DMSO-d ): 147.7 (JCP =15 Hz, N-
C), 137.4 (JCP =10 Hz, P-C-CH-CH), 136.6 (JCP = 135 Hz, C-P), 130.7 (JCP = 13 Hz,
N-C-CH-CH), 126.51 (JCP = 2.6 Hz, N-C-CH-CH), 125.4 (JCP = 11 Hz, C-CH-C)
3
1
-1
6
P NMR (δ, ppm, DMSO-d ): 17.69. νmax (FT-IR)/cm 3087 w, 2871 w, 2111 w,
2
9
088 w, 1989 w, 1607 m, 1528 s, 1468 m, 1420 m, 1349 s, 1285 m, 1140 m, 1090 m,
-
65 (914 sh) s, 884 s, 813 s, 772 s, 730 s. ESI-MS (solvent; DMSO): m/z = 307
-
(100%, C12
H
8
N
2
PO
6
).
3
General Procedure 1 (GP 1): BiPh (1 eq.) and the phosphinic acid (2 eq.) were each
separately dissolved in hot ethanol (unless otherwise stated), then combined and
heated at reflux, for twelve to 24 hours. The final product was collected by gravity
filtration and washed with hot ethanol and diethyl ether to yield the desired
1
2
[
2
BiPh(OP(=O)R R ) ] complex.
General Procedure 2 (GP 2): All reactions were carried out under standard Schlenk
t
conditions using THF. Bi(O Bu)
3
(1 eq.) dissolved in THF was added to a THF
solution (or suspension) of the phosphinic acid (3 eq.) which was already cooled to
80 °C. This temperature was maintained for approximately one hour and
-
subsequently warmed to room temperature while stirring overnight. The final product
was collected by gravity filtration and washed with hot ethanol and diethyl ether to
1
2
3
yield the desired [Bi(OP(=O)R R ) ] complex.
[BiPh(OP(=O)Ph
2
)
2
], 1
1
was prepared in accordance with GP 1 using diphenylphosphinic acid. Yield: 0.140
1
3
31
g, 69 %, m.p. >300 °C. C CP-MAS (δ, ppm): 138.9, 131.3, 129.2. P HPDEC-MAS
-
1
(
δ, ppm): 20.47. νmax (FT-IR)/cm 3050 w, 1981 w, 1484 w, 1437 m (P-Ph), 1180 w,
), 1112 s νasym(PO ), 1068 m νasym(PO ), 1036 m νsym(PO ), 1013 s
), 995 s νsym(PO ), 926 m, 849 w, 752 m, 733 m, 720 s, 691 s. elemental
analysis calcd (%) for C30 : C 50.01, H 3.50, Bi, 29.0; found C 49.72, H
.45, Bi 30.4.
1
122 s νasym(PO
2
2
2
2
ν
sym(PO
2
2
H
4 2
25BiO P
3
2
6
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
[BiPh(OP(=O)(p-OMePh)
2
)
2
], 2
2
was prepared via solvent free reaction of BiPh
3
(1 eq.) and bis(4-
methoxyphenyl)phosphinic acid (2 eq.) in a Kugelrohr oven at 130 °C for 4 hours.
The temperature of the reaction was determined by DSC (ESI, Figure S14). The white
solid was collected and washed with hot ethanol. Yield: 0.159 g, 72 % yield of
1
3
complex 2. m.p. 250 °C. C CP-MAS (δ, ppm): 161.3, 136.8, 131.7, 129.1, 127.2,
3
1
1
23.9, 118.1, 112.1, 110.3, 55.5, 54.3. P HPDEC-MAS (δ, ppm): 24.60. νmax (FT-
-
1
IR)/cm 3058 w, 2837 w, 2548 w, 2108 w, 2007 w, 1597 m, 1570 m, 1501 m, 1463
m, 1430 m, 1292 m, 1247 s, 1175 m, 1121 s νasym(PO ), 1017 s
), 909 w, 826 s, 802 s, 732 s. elemental analysis calcd (%)
: C 48.58, H 3.96, P 7.37, Bi 24.9; found C 48.82, H 3.82, P 7.13, Bi
2
), 1083 s νasym(PO
2
ν
sym(PO
2
), 991 s νsym(PO
2
for C34
H33BiO
8
P
2
2
3.8.
2 2 2
[BiPh(OP(=O)(m-NO Ph) ) ], 3
3
was prepared in accordance with GP 1 using bis(3-nitrophenyl)phosphinic acid.
1
3
Yield: 0.165 g, 88%, m.p. 280 °C. C CP-MAS (δ, ppm): 148.6, 147.1, 138.3, 130.3,
3
1
-1
1
26.9. P HPDEC-MAS (δ, ppm): 15.52. νmax (FT-IR)/cm 2227 w, 1608 m, 1531 s
N-O), 1472 w, 1430 w, 1346 s (N-O), 1304 w, 1277 w, 1202 w, 1132 s νasym(PO ),
), 916 w, 876 m, 809 m, 770 m,
: C 40.02, H 2.35, N 6.22;
(
2
1
7
073 m νasym(PO
2
), 1028 s νsym(PO
2
), 996 m νsym(PO
2
36 m. elemental analysis calcd (%) for C30
H21BiN O P
4 12 2
found: C 40.36, H 2.32, N 6.10.
[BiPh(OP(=O)(Ph)(H))
2
], 4
4
7
was prepared in accordance with GP 1 using phenylphosphinic acid. Yield: 0.130 g,
1
3
31
8%, m.p. 197 °C (dec.). C CP-MAS (δ, ppm): 140.5, 132.0, 130.7, 127.2.
P
-
1
HPDEC-MAS (δ, ppm): 20.05, 13.64. νmax (FT-IR)/cm 3058 w, 2366 w, 2338 w,
110 w, 1891 w, 1812 w, 1591 w, 1475 w, 1438 m, 1429 m, 1135 m νasym(PO ), 1091
), 1034 m, 1015 s νsym(PO ), 970 s, 861 w, 747 s, 726 s, 707 m, 689 w.
elemental analysis calcd (%) for C18 : C 38.05, H 3.02, Bi 36.8; found: C
8.01, H 2.97, Bi 37.0.
2
2
s νasym(PO
2
2
H
4 2
17BiO P
3
3 2 2
[BiPh(OP(=O)(CH ) ) ], 5
2
7
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
5
was prepared in accordance with GP 1 using dimethylphosphinic acid and
dimethylsulfoxide (DMSO) as the solvent. Yield: 0.063 g, 57%, m.p. 300 °C (dec.).
1
3
31
C CP-MAS (δ, ppm): 176.3, 43.5. P HPDEC-MAS (δ, ppm): 46.3 (sh), 42.4. νmax
-
1
(
FT-IR)/cm 3052 w, 2982 w, 2916 w, 2344 w, 2114 w, 2086 w, 1475 w, 1429 w,
415 m, 1293 m, 1290 m, 1144 m νasym(PO ), 1083 m νsym(PO ), 1061 m, 1017 s
), 913 w, 860 s, 730 s, 689 s. elemental analysis calcd (%) for C10
C 25.44, H 3.36, Bi 44.3; found: C 25.48, H 3.57, Bi 43.2.
1
2
2
ν
sym(PO
2
4 2
H17BiO P :
[Bi(OP(=O)Ph
2
)
3
], 6
6
was prepared in accordance with GP 2 using diphenylphosphinic acid. Yield: 0.282
[
57] 13
g, 69%, m.p. >300 °C (Lit 237 °C) . C CP-MAS (δ, ppm): 142.2, 135.0, 133.3,
3
1
-1
1
2
30.1, 127.3. P HPDEC-MAS (δ, ppm): 24.22 νmax (FT-IR)/cm 3051 w, 2113 w,
058 w, 1906 w, 1591 w, 1484 w, 1436 m (P-Ph), 1256 w, 1183 w, 1122 m
ν
asym(PO
2
), 1091 s νasym(PO
2
), 1066 m, 1031 s, 1009 s νsym(PO
2
), 990 s νsym(PO
2
), 746
m, 728 s. elemental analysis calcd (%) for C36
found C 50.26, H 3.56, Bi 23.5.
H30BiO P : C 50.25, H 3.51, Bi 24.3;
6 3
[Bi(OP(=O)(p-OMePh)
2
)
3
], 7
7
was prepared in accordance with GP 2 using bis(4-methoxyphenyl)phosphinic acid.
1
3
Yield: 0.393 g, 76%, m.p. 230 °C. C CP-MAS (δ, ppm):137.4, 135.9, 133.3, 131.3,
3
1
-1
1
2
26.6, 124.7. P HPDEC-MAS (δ, ppm): 23.82. νmax (FT-IR)/cm 3064 w, 3002 w,
950 w, 3901 w, 2834 w, 2544 w, 2299 w, 2105 w, 1595 m, 1570 m, 1500 m, 1458
m, 1403 w, 1292 m, 1248 m, 1178 m, 1122 s νasym(PO
), 993 s νsym(PO ), 827 s, 800 s, 722 s, 668 s. elemental analysis calcd (%) for
: C 48.47, H 4.07, Bi 20.1; found C 48.53, H 4.09, Bi, 19.6.
2 2
), 1082 s νasym(PO ), 1018 s
ν
sym(PO
2
2
C
42
H42BiO12P
3
2 2 3
[Bi(OP(=O)(m-NO Ph) ) ], 8
8
was prepared in accordance with GP 2 using bis(3-nitrophenyl)phosphinic acid.
1
3
Yield: 0.253 g, 87%, m.p. 260 °C (dec.). C CP-MAS (δ, ppm): 148.4, 136.6, 129.8,
3
1
-1
1
1
1
8
26.4. P HPDEC-MAS (δ, ppm): 19.26. νmax (FT-IR)/cm 3082 w, 2869 w, 2113 w,
608 m, 1575 m, 1525 s (N-O), 1470 w, 1421 w, 1348 s (N-O), 1305 w, 1277 w,
173 w, 1146 w, 1091 s νasym(PO
2 2 2
), 1073 s νsym(PO ), 1017 s νsym(PO ), 992 s, 904 w,
80 m, 811 m, 767 m, 729 s, 676 s, 662 s. elemental analysis calcd (%) for
C
36
H24BiN
6
O
18
P
3
: C 38.25, H, 2.14, N 7.43; found: C 38.04, H 2.09, N 7.19.
2
8
This article is protected by copyright. All rights reserved.

Chemistry - A European Journal
10.1002/chem.201801803
[Bi(OP(=O)(Ph)(H))
3
], 9
9
6
1
was prepared in accordance with GP 2 using phenylphosphinic acid. Yield: 0.22 g,
1
3
7 %, m.p. 130-136 °C. C CP-MAS (δ, ppm): 136.9, 133.1, 130.9, 128.9, 127.8,
3
1
-1
26.3. P HPDEC-MAS (δ, ppm): 17.28. νmax (FT-IR)/cm 3143 w, 3074 w, 3049
w, 2377 w, 2113 w, 1897 w, 1591 w, 1482 w, 1436 m, 1377 w, 1309 w, 1211 w, 1185
), 1031 s νsym(PO ), 1016 s νsym(PO ), 974 (924 sh) s, 740 s, 705
: C 34.20, H 2.87, Bi 33.05;
w, 1098 s νasym(PO
2
2
2
s, 688 s. elemental analysis calcd (%) for C18
found: C 34.39, H, 2.93, Bi 34.1
6 3
H18BiO P
3 2 3
[Bi(OP(=O)(CH ) ) ], 10
1
0 was prepared in accordance with GP 2 using dimethylphosphinic acid. Yield: 0.11
1
g, 58 %, m.p. >300 °C. NMR data H NMR (δ, ppm, CDCl
3
): 1.49 (18H, d, J = 14.2
1
3
31
Hz, CH
CDCl
299 m, 1290 m, 1200 w, 1080 s νasym(PO
3 3
). C NMR (δ, ppm, CDCl ): 18.70 (d, JPC = 92.85 Hz). P NMR (δ, ppm,
-
1
3
): 47.65. νmax (FT-IR)/cm 2986 w, 2919 w, 2120 w, 1928 w, 1647 w, 1425 w,
), 1013 s νsym(PO ), 863 s, 742 m, 713 w.
18BiOP .H O: C 14.24, H, 3.98, Bi 41.3; found:
1
2
2
elemental analysis calcd (%) for C
6
H
3
2
C 14.24, H 3.58, Bi 39.6.
Preparation of cellulose based paper samples
The MFC sample was Celish grade KY-100G from Daicel Corporation, Japan. The
measured average diameter was 73 nm and the aspect ratio estimated by
[
74,75]
sedimentation was within the range 125-142.
The never-dried material, received
at 25% solids content, was dispersed in deionised water at 0.2 wt% by agitating using
a 3l disintegrator.
Suspensions of complexes 1-6 and 9 were prepared by dispersing each complex in
2
0% isopropanol/water at 0.2 wt% and stirring for 24 hours at room temperature.
Note: water alone was not able to disperse the complex well enough. The bismuth
papers were filtered over Whatman hardened ashless filter paper of grade 541,
diameter 185 mm.
Cationic dimethylaminoethyl methacrylate poly-acrylamide (CPAM) polymer with
high molecular weight (13 MDa) and charge density of 40 wt% (F1, SnowFlake
2
9
This article is protected by copyright. All rights reserved.