Ultrasensitive QRS made by supramolecular assembly of functionalized cyclodextrins and graphene for the detection of lung cancer VOC biomarkers

Information for Authors .

View Journal

Journal of

Materials Chemistry B

Accepted Manuscript

This article can be cited before page numbers have been issued, to do this please use: S. NAG, L.

DUARTE, E. BERTRAND, V. Celton, M. Castro, V. Choudhary, P. Guegan and J. FELLER, J. Mater. Chem. B,

2014, DOI: 10.1039/C4TB01041H.

This is an Accepted Manuscript, which has been through the

Royal Society of Chemistry peer review process and has been

accepted for publication.

Accepted Manuscripts are published online shortly after

acceptance, before technical editing, formatting and proof reading.

Using this free service, authors can make their results available

to the community, in citable form, before we publish the edited

article. We will replace this Accepted Manuscript with the edited

and formatted Advance Article as soon as it is available.

You can find more information about Accepted Manuscripts in the

Please note that technical editing may introduce minor changes

to the text and/or graphics, which may alter content. The journal’s

standard T erms & Conditions and the Ethical guidelines still

apply. In no event shall the Royal Society of Chemistry be held

responsible for any errors or omissions in this Accepted Manuscript

or any consequences arising from the use of any information it

contains.

www.rsc.org/materialsB

Page 1 of 10

Journal of Materials Chemistry B

DOI: 10.1039/C4TB01041H

Graphical Abstract

A novel approach to control the multiscale architecture of quantum chemo-resistive

transducers is introduced by the supramolecular assembly of graphene and cyclodextrin (CD)

derived hybrids. Tailorable selectivity to cancer biomarkers could be achieved due to the

insertion of CD with varying functionality between graphene foils.

Keyword: anticipated diagnosis; functional cyclodextrins; graphene; cancer biomarkers;

electronic nose

Sananda Nag,^a,bLisday Duarte,^cEmilie Bertrand,^cVéronique Celton,^cMickaël Castro,^aVeena

Choudhary,^bPhilippe Guegan, *^d,eJean-François Feller*^a

Ultrasensitive QRS made by supramolecular assembly of functionalized cyclodextrins and

graphene for the detection of lung cancer VOC biomarkers

Journal of Materials Chemistry B

Page 2 of 10

Dynamic Article Links ►

Journal Name

DOI: 10.1039/C4TB01041H

Cite this: DOI: 10.1039/c0xx00000x

ARTICLE TYPE

Ultrasensitive QRS made by supramolecular assembly of functionalized

cyclodextrins and graphene for the detection of lung cancer VOC

biomarkers

Sananda Nag,^a,bLisday Duarte,^cEmilie Bertrand,^cVéronique Celton,^cMickaël Castro,^aVeena

Choudhary,^bPhilippe Guegan*^d,eand JeanꢀFrançois Feller*^a

Received (in XXX, XXX) Xth XXXXXXXXX 20XX, Accepted Xth XXXXXXXXX 20XX

DOI: 10.1039/b000000x

A novel electronic nose system comprising functionalized βꢀcyclodextrin wrapped reduced graphene

10 oxide (RGO) sensors with distinct ability of discrimination of a set of volatile organic compounds, has

been developed. Nonꢀcovalent modification of chemically functionalized cyclodextrin with RGO is

carried out by using pyrene adamantan as a linker wherever necessary, in order to construct a supra

molecular assembly. The chemical functionality on cyclodextrin is varied utilising the principle of

selective chemical modification of cyclodextrin. In the present study, the combined benefits of hostꢀguest

15 inclusion complex formation ability and tunable chemical functionality of cyclodextrin, as well as high

surface area and electrical conductivity of graphene are utilized for the development of a set of highly

selective quantum resistive chemical vapour sensors (QRS), which can be assembled in an electronic

nose.

high consumption. On the contrary, arrays of nonspecific sensors,

i.e., electronic noses (eꢀnose), have demonstrated their

effectiveness in the detection of VOC.¹⁸Eꢀnoses present in

particular the advantage of being nonꢀinvasive, cost effective,

⁵⁰quick and portable. Additionally they are allowing real time

monitoring¹⁹and provide almost directly a pattern of exhaled

biomarkers in the form of a VOC breath print.²⁰Moreover, a

wide choice of vapour sensors can be assembled in the array,

1. Introduction

Enhancing the quality of life by decreasing the impact

of severe diseases is still a global challenge. According to the

World Health Organization (WHO), cancers which are killing

each year 7.6 million people can be considered as a global

priority.¹The WHO has also projected that without immediate

25 action, the global number of deaths from cancer will increase by

nearly 80 % by 2030, with most occurring in lowꢀ and middleꢀ

income countries. Cancers could be more efficiently treated in

case of an anticipated diagnosis, and from this point of view

metabolomics can bring very promising solutions. This new and

30 active field of research is studying the volatile organic

compounds (VOC) produced by a biological system to investigate

the metabolite differences between natural and perturbed systems

(cells, organs and tissues). Thus, metabolomics allows profiling

diseases such as cancer from a VOC fingerprint found in breath,

35 urine, faeces, saliva, nasal mucus or gaseous excretions of the

skin²or combination of them.³Among these, exhaled breath is an

excellent source containing several hundreds of VOC (including

water) but at the ppm or ppb level, which makes their

identification difficult. However, analyzing VOC profiles with

⁴⁰pattern recognition algorithms allows to efficiently discriminate

between cancerous and healthy subjects.^4–13Nevertheless, the

commonly used techniques for breath VOC analysis such as GCꢀ

MS,¹⁴infrared spectroscopy,¹⁵ion flow tube mass

spectrometry,¹⁶optical spectroscopy,¹⁷suffer from several

⁴⁵shortcomings such as high cost, lowꢀportability, low sensitivity or

depending on the nature of the transducer: metal oxides (MO),^20ꢀ

intrinsically conductive polymers (ICP^),22ꢀ23and functionalized

carbon nanomaterials or nanocomposites.^25–29However, eꢀnoses

have also some drawbacks, such as processing and performance

reproducibility, drift in baseline and response of sensors requiring

sometimes recalibration. MO sensors often have a low sensitivity,

⁶⁰a lack of selectivity, a high operating temperature (several

hundred degrees) imposed by their too high sensitivity to water,³⁰

whereas ICP sensors are too sensitive to humidity, irradiation and

oxidoꢀreduction.³¹Most of these problems can be overcome when

using quantum resistive sensors (QRS), which are versatile

65 nanomaterials obtained by the structuring of conductive

nanofillers like carbon nanotubes (CNT) into an insulating matrix

like a polymer of suitable functionality, provided that a control of

30ꢀ31

both the conductive architecture

and the nanoswitching at

junctions of percolated nanofillers^32ꢀ33is ensured. In addition, the

70 selectivity of QRS can be easily adjusted by changing the

chemical nature of the thin organic layer separating the

conductive junctions.^34ꢀ35The first problem can be fixed by using

a process such as spray layer by layer (sLbL), which allows a step

by step building of the transducers from the nano to the macro

This journal is © The Royal Society of Chemistry [year]

[journal], [year], [vol], 00–00 | 1

CREATED USING THE RSC ARTICLE TEMPLATE (VER. 3.0) - SEE WWW.RSC.ORG/ELECTRONICFILES FOR DETAILS

Page 3 of 10

ARTICLE TYPE

Journal of Materials Chemistry B

DOI: 10.1039/C4TB01041H

scale,^36–38whereas the second can find a solution through the

biological molecules and polymers showing high molecular

tailoring of junction’s gap.^39,40This last parameter will determine 60 selectivity.⁶³The main driving forces for the host/guest inclusion

both the selectivity (depending on the nature of intermolecular

interactions between the organic layer and the biomarker) and the

sensitivity (depending on the initial value and amplitude of the

gap) of QRS.

complex formation are hydrophobic interactions, van der WAALS

interactions or hydrogen bonding. Various fields of applications

of chemically modified CD include drug delivery,⁶⁴catalysis,⁶⁵

artificial channels⁶⁶and sensing.⁶⁷

Several strategies have already been experimented to

control the nanoswitching of QRS involving exclusion

volumes,^26,33,41functionalization of carbon nanofillers such as

A synergistic effect between RGO (due to high

conductivity and high surface area) and native CD (thanks to

hostꢀguest recognition and enrichment) has been anticipated in

2010 by GUO et al.,⁶⁸who found that RGOꢀCD hybrids simply

prepared by dispersing CD in a graphene oxide (GO) solution

10 CNT^41,42or graphene³⁸with a polymer,^43–45an oligomer⁴²or

nonpolymer organic/inorganic molecules^46–49

which could

effectively modify the selectivity and sensitivity of chemoꢀ ⁷⁰followed by the reduction of GO, led to interesting

resistive vapour sensors. However, nonꢀcovalent electrochemical response towards biomolecules and drugs, unlike

unmodified graphene or carbon nanotubes.

functionalization of nanofillers, which can also be obtained

¹⁵through secondary interactions such as πꢀπ stacking, chargeꢀ

charge interaction and hydrogen bonding, is probably the most

LEGER et al.,

demonstrated that triazole group of functionalized CD may

interact with MWCNT by πꢀπ interaction, thus acting as a good

attractive approach to preserve the graphitic chemical structure, ⁷⁵dispersing agent that could potentially introduce new recognition

and hence, provide a positive synergy of properties.⁵⁰

However, none of these strategies really allow

20 controlling simultaneously the architecture of the conducting

network and the physicoꢀchemical properties of the junction’s

sites on the surface of MWCNT.⁶⁹Additionally, KONG et al. have

successfully grafted monofunctionalꢀCD derivatives onto

SWCNT (Single wall carbon nanotube) by in situ diazonium

reaction, to synthesize novel SWCNTꢀCD with excellent

gap, optimizing at the same time sensitivity, stability and 80 sensitivity and selectivity towards organic pollutants, thanks to

selectivity of the QRS. The objective of the present work is to

demonstrate that these parameters can be optimized, through the

²⁵development of hybrid transducers, fabricated by supramolecular

assembly of graphene and cyclodextrin. These two singular

molecular recognition.⁷⁰MWCNT (multi walled carbon

nanotubes) was also functionalized with monofunctional CD,

dangling on the MWCNT and thus providing an easy access to

the cavity of the CD.⁷¹

YANG et al. also reported adsorption of a

nanofillers, which combine steric and chemical effects, will allow 85 protein, bovine serum albumin, on these functionalized MWCNT

reaching a positive synergy thanks to their complementary

properties that will be shortly described in the following.

through nonꢀcovalent interactions with CD derivatives

functionalized aligned carbon nanotubes for the electrochemical

sensing of DNA via host–guest recognition.⁷²Their study recalls

that the diazonium reaction does not allow the formation of a

³⁰Since its discovery in 2004 by GEIM et al.⁵¹graphene has

continuously demonstrated exceptional intrinsic properties such

as: high electrical and thermal conductivity (respectively 7200 ⁹⁰single layer of cyclodextrin on the carbon nanotubes, which may

S.m^ꢀ1and 5000 W.m^ꢀ1.K^ꢀ1)^52,53, high aspect ratio (surface area of

2630 m².g^ꢀ1for a thickness of 1nm),⁵⁴mechanical strength (130

³⁵GPa),⁵⁵optical transmittance, high carrier mobility and charge

limit the performances of the detection. Finally, GUO et al.

modified native CD with a pyrene residue to allow the adsorption

of the pyrene labeled cyclodextrin on SWCNT to prepare lightꢀ

switchable SWCNT.⁷³However, the reported modifications of

carrier concentration at room temperature (250,000 cm².V^ꢀ1.s^ꢀ1).

In particular the combination of high electrical conductivity 95 graphene or CNT for sensing applications are not versatile, and

and large specific surface has made graphene an attractive

nanomaterial for chemical sensing applications,⁵⁶since the

40 adsorption of individual gas molecules onto a graphene based

transducer brings about significant changes in electrical

can only be dedicated to one kind of detection. Furthermore, no

chemoꢀresistive sensing application has been investigated with

cyclodextrin functionalized conductive nanocomposites so far.

The originality of the present work is to take benefit

resistance.⁵⁷Moreover, the high carrier mobility of graphene,^58,59¹⁰⁰from the multiscale architecture of quantum chemoꢀresistive

allows room temperature detection and high signal to noise ratio

of the resistive sensors.⁶⁰Moreover, It has been found that the

45 nonꢀcovalent modification of reduced graphene oxide (RGO)

with different functional molecules was helping to prevent the

transducers resulting from the supramolecular assembly of

graphene and CDꢀderivate hybrids, to get a high sensitivity of the

nanoswitching at junctions due to the insertion of CD between

graphene foils and to obtain a tailorable selectivity to cancer

restacking of individual graphene sheets, and was improving both 105 biomarkers due to CD functionalization and geometry. This

sensitivity and selectivity of the graphene based QRS to different

VOC.⁶¹

βꢀcyclodextrin (CD) is a cyclic oligosaccharide having

a toroidal shape. βꢀCD is composed of 7 Dꢀglucopyranosidic

strategy includes the effectiveness of spray layer by layer to

maintain the nanomaterial’s properties across the different scales

(from nano to micro) and the versatility of nonꢀcovalent bonding

of functionalizedꢀCD onto graphene via a pyrene spacer (by πꢀπ

units containing a hydrophobic internal cavity and a hydrophilic ¹¹⁰stacking), which preserve at the same time the integrity of

outer side with primary and secondary hydroxyl groups on

narrower and wider side respectively. The presence of hydroxyl

conducting foils and manages to some extent junctions’ gap.

Moreover, RGOꢀCD is expected to combine two types of

55 groups on the outer surface not only makes CD water soluble but

also makes it eligible for regioꢀselective chemical modification,⁶²

molecular recognitions, nonꢀspecific via van der

interactions and specific reversible chemical reactions with CD

WAALS

while the hydrophobic internal cavity provides an ability to form 115 functions, coming for instance from a free amino group that can

host/guest inclusion complexes with various organic, inorganic, react with acetone.

Journal Name, [year], [vol], 00–00

This journal is © The Royal Society of Chemistry [year]

Journal of Materials Chemistry B

Dynamic Article Links^{Page 4 of 10}

►

Journal Name

DOI: 10.1039/C4TB01041H

Cite this: DOI: 10.1039/c0xx00000x

ARTICLE TYPE

ꢀ

Perbenzylated cyclodextrin linked RGO (RGO@PBCD) is

2. Results and Discussion

providing a strong nonꢀpolar character to the junction due to its

numerous aromatic rings.

2.1. Selection of VOC Lung Cancer Biomarkers

It is quite tricky to sense lung cancer biomarkers at the

ppb level in real conditions, considering that they are blended

with many other VOC including water at high concentration

(more than 80 %). Still, detection of pure molecules must first be

conducted before addressing a discriminating detection. Thus we

have restricted the present study to the analysis of chemoꢀ

resistive properties of an array of RGOꢀCD based QRS in the

The variety of CD functionalization is a strong advantage, which

⁵⁵makes this molecule an important brick in the sensors’ design,

allowing tailoring the junction’s physicoꢀchemical properties, i.e.,

the gap and the intermolecular interactions. Moreover, the

association of functionalized CD with RGO foils is expected to

preserve the accessibility of analytes to functional sites by

⁶⁰maintaining a distance between foils that can be varied as a

function of substituents’ size and conformation. Additionally, an

efficient connection between CD and RGO can be obtained with

PYAD, resulting on one end from the fitting of the globular shape

of adamantan (about 0.7 nm in diameter) inside the CD cage,⁷⁹

⁶⁵and on the other end from the πꢀπ stacking of pyrene with

graphene’s surface.⁸⁰Graphene oxide (GO) was synthesized from

natural graphite powder by a modified HUMMER’s method.⁸¹

Thereafter, GO was nonꢀcovalently modified with PYAD, and

then reduced to give RGO@PYAD, so that pyrene gets adsorbed

⁷⁰on to the surface of graphene by πꢀπ interaction. In the final step

of synthesis, the functionalized cyclodextrin was allowed to

interact with RGO@PYAD to induce a hostꢀguest connection

between the adamantan group and the cavity of cyclodextrin as

shown in route a) of Fig. 1. For perbenzylated CD modified

⁷⁵graphene, the step of adsorption of the linker was skipped, taking

advantage of the πꢀπ interaction between graphene and the

aromatic residue of the CD according to route b) of Fig. 1. The

characterizations of successful synthesis of functional

cyclodextrins, and nonꢀcovalent modifications of graphene with

⁸⁰functionalized cyclodextrins are available in the supporting

information part (see Fig.S5 to Fig. S20).

¹⁰presence of synthetic biomarkers produced under controlled

conditions in the laboratory, being aware of the remaining

possible obstacles until the final validation with cancerous

patients. Among the several hundreds of cancer biomarkers

produced by the body, we have selected the following ones,

¹⁵which are present in the breath of patients with lung cancer:

acetone,⁷⁴benzene,^74–76methanol,⁷⁶ethanol,⁷⁶formaldehyde,⁷⁴

toluene,^74–76xylene,⁷⁵propanol,^74,77isopropanol.⁷⁶Relatively

high contents of acetone (several ppm) are found because of a

higher rate of oxidation of fatty acid under the conditions of

²⁰oxidative stress. Aromatic hydrocarbons like benzene, toluene are

also prominent lung cancer biomarker. Alcohols are abundant in

the breath extract of both healthy subjects and lung cancer

patients, but 1ꢀpropanol is found in the breath extract of lung

cancer patients at higher concentration than healthy subjects.⁷⁴

25 This is the reason why an efficient detection and discrimination

of alcohols, ketone bodies like acetone and aliphatic or aromatic

hydrocarbons is required. Since breath extracts contain a huge

amount of water tending to mask the trace amounts of VOC,⁷⁸

this molecule had also to be included into the selected set of

³⁰VOC.

2.2. Synthesis and Fabrication of Sensors

The consideration of the wide range of polar/nonꢀpolar

interactions possible with the selected biomarkers is justifying the

varieties of transducers used (Table. 1):

ꢀ

Reduced graphene oxide (RGO) will exhibit nonꢀpolar

interactions and low disconnection capability, it is used as a

reference,

ꢀ

Pyrene adamantan linked RGO (RGO@PYAD) will increase

40 junctions’ gap due to PYAD spacer but is not expected to change

selectivity,

ꢀ

Native cyclodextrin linked RGO (RGO@PYADꢀCD) is

providing OH polar groups susceptible of interaction with polar

analytes,

ꢀ Mannose functionalized cyclodextrin linked RGO

(RGO@PYAD–MCD) providing OH polar groups with a longer

spacer capable of easier disconnection,

Fig. 1 Different routes of synthesis of functionalized cyclodextrin linked

to reduced graphene oxide by supramolecular assembly. a) Route for the

85 synthesis of RGO@PYADꢀCD RGO@PYADꢀMCD & RGO@PYADꢀ

NCD (R is a hydroxyl group for CD, an amino group for NCD, and a

mannose sugar for MCD), b) Route for the synthesis of RGO@PBCD.

ꢀ

Amino

functionalized

cyclodextrin

linked

RGO

(RGO@PYADꢀNCD) is bringing reversible covalent bonding of

50 its free amino group with acetone for instance,

This journal is © The Royal Society of Chemistry [year]

Journal Name, [year], [vol], 00–00 | 3

Page 5 of 10

ARTICLE TYPE

^{CREATED USING THE RSC ARTI}J^CLo^Eu^TEr^Mn^Pa^Ll^ATo^Ef^(VM^ERa^.t³e^.0r^{) -}ia^SEls^{E W}C^Wh^We^.Rm^SCi^.Os^Rtr^Gy^/ELB^{ECTRONICFILES FOR DETAILS}

DOI: 10.1039/C4TB01041H

The fabrication of QRS is a key point to ensure a good

adjust the initial resistance of all the sensors to comparable

reproducibility of sensing performances. Firstly, the RGO 10 values, so that only the chemical functionality of CD and not the

decorated with different kinds of functionalized cyclodextrins

needs to be efficiently dispersed in acetone with the aid of ultraꢀ

sonication for 1 h at 50 ^◦C with a BRANSON 3510 device (100 W,

40 kHz), and secondly this well dispersed solution can be sprayed

layerꢀbyꢀlayer (sLbL) onto interꢀdigitated electrodes.⁸²Both the

solutions’ content and the number of sprayed layers allow to

transducer’s percolation level,⁸³will have an influence on sensors

sensitivity to VOC. After fabrication, all sensors were

conditioned in controlled atmosphere for one night before chemoꢀ

resistive tests.

Table 1 List of the fabricated sensors and their main characteristics

Sensor

number

Description of

the sensor

Designation of

sensor

Structure of CD attached

Number

of layers

sprayed

Initial

resistance of

the sensor

(R₀) kꢁ

Reduced

graphene oxide

Pyrene

RGO

N.A

120±4

RGO@PYAD

130±6

adamantan

linked RGO

Native

RGO@PYADꢀCD

150±7

cyclodextrin

linked RGO

Mannose

functionalized

cyclodextrin

linked RGO

RGO@PYAD–

MCD

150±3

Amino

RGO@PYADꢀNCD

155±4

functionalized

cyclodextrin

linked RGO

Perbenzylated

cyclodextrin

linked RGO

RGO@PBCD

125±5

Bn=

2.3. Morphological Characterization

measure (resistance variation). Atomic force microscopy (AFM)

can give a good idea of the nanoꢀstructure obtained after sLbL in

ambient conditions, using light tapping mode (TMꢀAFM) on a

The technique consisting in the superimposition of

30 calibre multimode scanning probe microscope from BRUKER

ꢀ

layers of several tens of nm up to a final film thickness of some

µm showed in Fig. 2a, has proved to allow the building of

hierarchically structured transducers, which presents the

25 advantage of bridging the original nanostructures developed by

self assembly of RGO with CD, to the macroscopic parameter to

EECO, France. By comparing the images of Fig. 2b and Fig. 2c,

one can notice a three times increase of graphene sheets’

thickness in RGO@PBCD (around 3 nm) compared to GO

(around 1.2 nm), which is comforting the idea that the decoration

Journal Name, [year], [vol], 00–00

This journal is © The Royal Society of Chemistry [year]

^{CREATED USING THE RSC ARTI}J^CLo^Eu^TEr^Mn^Pa^Ll^ATo^Ef^(VM^ERa^.t³e^.0r^{) -}ia^SEls^{E W}C^Wh^We^.Rm^SCi^.Os^Rtr^Gy^/ELB^{ECTRONICFILES FOR DETAILS}

Page 6 of 10

ARTICLE TYPE

DOI: 10.1039/C4TB01041H

of RGO with functionalized CD has been effective and useful to

prevent the restacking of graphene sheets.

2.4.1. Chemoꢀelectrical behaviour

The bar graph of Fig. 3a gives a good overview of the

45 eꢀnose’s fingerprint (assembled with RGO, RGO@PYAD,

RGO@PYADꢀCD, RGO@PYADꢀMCD, RGO@PYADꢀNCD,

RGO@PBCD sensors, see Table. 1 for details about the sensors

composition), when exposed to a set of 10 VOC (methanol,

ethanol, propanol, isopropanol, benzene, toluene, pꢀxylene,

⁵⁰acetone, formaldehyde which have been selected among lung

cancer biomarkers, plus water which is present in breath at more

than 90 %, (see Table S2 in supplementary information for details

about solubility parameters of these solvents). Looking at Fig. 3a

in more details shows the selectivity of the different sensors and

⁵⁵also allows noticing their nonꢀspecificity (they all respond to all

VOC but not with the same segregation ability, as it is the case

for the olfactory receptors of mammal nostrils). As shown in

supplementary information (Fig. S21) the resistance variation of

CCDC sensors is sharp upon exposure to VOC (some seconds

⁶⁰only are necessary to reach signals’ saturation), stable and

reversible (recovery of the initial value of the resistance R₀,

before the end of the dry nitrogen purge cycle).

Fig. 2 (a) Scheme showing spray LbL deposition and AFM images of (b)

GO and (c) RGO@PBCD

2.4. Sensing Performance of QRS

As postulated before, the sensitivity/selectivity of each

The chemoꢀresistive properties of the different CD

sensor can be interpreted as a function of the modification of

⁶⁵junctions’ physicoꢀchemical properties. In this study two main

parameters can significantly affect junctions to promote tunneling

to the detriment of ohmic conduction in the CCDC network, the

steric hindrance of linkers and their ability to adsorb analytes that

will expand the junctions. This last parameter is in turn related to

⁷⁰the functionality of the CD. For instance the slightly higher

sensitivity of RGO@PYAD than RGO must be due to a better

exfoliation of the graphene sheets due to PYAD spacing that is

preventing at the same time the restacking of foils. However it

can be noticed that PYAD is not changing the selectivity of

⁷⁵pristine graphene, meaning that this molecule is not specifically

interacting with any of the studied VOC. The best selectivity of

RGO@PYADꢀCD towards polar VOC (although of low

magnitude) can be explained by the presence of several hydroxyl

groups on native CD. Moreover the fact that RGO@PYADꢀ

⁸⁰MCD, RGO@PYADꢀNCD and RGO@PBCD have different

selectivity than RGO@PYADꢀCD is a confirmation of the

effectiveness of the strategy chosen in this work to tailor the

sensors’ selectivity, i.e., changing its van der WAALS interactions

with the analytes by CD functionalization. Nevertheless, the

85 strong sensitivity of RGO@PYADꢀNCD sensors to acetone

vapour is more likely coming from its free amino group in

RGO@PYADꢀNCD that might have undergone a reversible

imine formation reaction with acetone, and not from specific van

der WAALS interactions. The ranking of sensitivity to VOC of

90 RGO@PBCD deduced from Fig. 3a is: benzene> toluene > pꢀ

xylene> isopropanol> propanol> ethanol> acetone> methanol>

water> formaldehyde is consistent with the fact the

perbenzylation of βꢀCD enhances the affinity of RGO@PBCD

for nonꢀpolar VOC. Reversely the ranking of sensitivity to VOC

95 of RGO@PYADꢀMCD sensors is demonstrating a higher affinity

for polar vapours, i.e., methanol > ethanol > propanol >

isopropanol > water > benzene > toluene > pꢀxylene > acetone >

formaldehyde. In this case the presence of triazole rings and free

hydroxide groups (of the sugar residue) in the MCD can explain

functionalized RGO@PYAD sensors were then investigated by

10 submitting them to a standard protocol developed in our group,

which consists in successive expositions to rectangular pulses of

5 min in saturated VOC followed by 5 min in pure nitrogen. The

control of the electrical valves adjusting the gas flows and data

acquisition are both driven by a program developed under

LABV

IEW environment as described in a previous publication.⁸⁴

The chemoꢀresistive responses of conductive CD nanocomposite

(CCDC) sensors can be easily expressed by calculating the

relative amplitude of electrical signals (A_r) towards solvents using

Equation 1:

ꢅ

ꢀꢁ ꢂ ^ꢃꢄꢃ……………………… (1)

ꢃ_ꢅ

where R₀and R are the initial resistance of the sensor in pure

nitrogen and the resistance of the sensor in the presence of

solvent vapour, respectively.

The same volume of each vapour under saturated condition does

25 not contain the same number of molecules due to differences in

saturation vapour pressure of each analyte. Hence, in order to get

rid of the influence of the number of molecules, response

amplitude was normalized by a factor related to number of

analyte molecules, following Equation 2, when partially selective

³⁰response of each sensor to different VOC was compared.

ꢒꢓ

ꢆꢇ_{ꢈꢉꢊꢋꢌꢍꢎꢏꢐꢑ}

ꢂ

_ꢟ,ꢠ∗ 10^ꢡ………….. (2)

ꢔꢒꢕꢖꢗꢘꢙꢚꢛ_ꢜꢝꢞ

Where [Analyte]_satis the concentration of analyte molecules at

saturation, T = 25 °C and p = 0.098 MPa

For experiments in the ppmꢀppb range another device was used to

35 control VOC concentrations by combining LABVIEW and OVGꢀ4

softwares. This accurate chemical vapour generator is composed

of an oven and a controller of both sample and split flows. The

sample flow was kept at 100 cm³.min^ꢀ1whereas the split flow was

varied to change the VOC concentration at a fixed temperature of

40 25 °C.

Journal Name, [year], [vol], 00–00

This journal is © The Royal Society of Chemistry [year]

Page 7 of 10

Journal of Materials Chemistry B

Dynamic Article Links

►

Journal Name

DOI: 10.1039/C4TB01041H

Cite this: DOI: 10.1039/c0xx00000x

ARTICLE TYPE

the enhancement of affinity of this sensor for polar solvents,

especially alcohols, thanks to the possibility of hydrogen

bonding. Furthermore, it can be noticed distinctly that

RGO@PYADꢀMCD sensors are able to discriminate alcohols,

depending on their number of carbon atoms. Hence, in this case

the analytes’ size might also contribute to the selectivity of

sensors.

according to Equation 3, but is it the same in CCDC in which the

polymer phase is replaced by functionalized CD?

ꢤ

ꢆꢇ_{ꢈꢉꢊꢋꢌꢍꢎꢏꢐꢑ}ꢂ ꢢ. ꢣ^ꢥꢦꢧ…………..(3)

With χ₁₂the FloryꢀHuggins intermolecular interaction parameter

which can be determined using Equation 4, a and b are constants

ꢨ_ꢩꢪꢂ ꢫ_ꢬꢃꢭꢮꢯ_ꢭꢰꢱꢲꢳ ꢯ_ꢭꢴꢱꢲꢵ² …... (4)

where V_mis molar volume of solvent, T is temperature in Kelvin,

35 R is universal gas constant and δ_{T pol}and δ_{T sol}are polymer and

solvent total solubility parameter respectively

To check that functionalized cyclodextrins, which can be

considered as oligomers, satisfy also the

FLORYꢀHUGGINS

approach, the global solubility parameter δ_Tof four of them, i.e.,

40 CD, MCD, PBCD and NCD, have been evaluated by the FEDORS

’

method of groups’ contribution.⁸⁶The following values of δ_T

were found respectively 27.6, 28.3, 20.29 and 25.9 J .cm^ꢀ3/2

allowing to calculate the corresponding χ₁₂parameters using

Equation 4 (detailed values are given in Table S2 of supporting

⁴⁵information). The validity of the model can then be evaluated by

plotting the evolution of the normalized relative amplitude Ar of

RGO@PBCD, RGO@PYADꢀNCD, RGO@PYADꢀMCD and

RGO@PYADꢀCD transducers with 1/χ₁₂using Equation 3. The

resulting curves plotted in Fig. 3b, show that three over the four

50 sensors,

i.e.,

RGO@PBCD,

RGO@PYADꢀMCD

and

RGO@PYADꢀCD are well fitting the exponential law, whereas

RGO@PYADꢀNCD evolution seems to be better described by a

logarithm. Comparing the curves of RGO@PYADꢀCD and

RGO@PYADꢀMCD shows that they can almost be superimposed

⁵⁵by translation, which means that they have quite the same

discrimination ability due to similar interactions with analytes,

RGO@PYADꢀMCD being always more sensitive probably

because of the larger size of its spacer, making more easy the

disconnection of junctions. The larger slope of RGO@PBCD

⁶⁰compared to that of RGO@PYADꢀMCD demonstrates the

superior discrimination ability of the former, which appears to

offer a better compromise between selectivity and sensitivity. The

peculiar behaviour of RGO@PYADꢀNCD, which is not

following the same law as the other sensors, seems to confirm the

⁶⁵possibility of a chemoꢀresistive response based on molecular

recognition by the reversible reaction of the amine function of

NCD that can form an imine with analytes such as acetone. Thus,

this chemoꢀresistive behaviour must not be mainly based on

weaker van der WAALS interactions. Therefore it can be

70 concluded that the decoration of graphene platelets with different

types of cyclodextrins is an effective strategy to tailor the

sensors’ selectivity to VOC, simply by changing the chemical

nature of the functions present on CD.

Fig. 3. a) Normalized average maximum relative amplitude of an array of

10 6 sensors towards 10 VOC and b) correlation between χ₁₂parameter and

transducer’s selectivity for RGO@PBCD, RGO@PYADꢀNCD,

RGO@PYADꢀMCD and RGO@PYADꢀCD

2.4.2. Origin of selectivity

As intuited in an early work ⁸⁵and confirmed later, ^35,43

the van der WAALS interaction between the chemical groups

present in the CCDC and in the VOC can be correlated to the

amplitude of their chemoꢀresistive response Ar, thus explaining

20 the selectivity of CCDC sensors provided that no stronger effect

interferes, such as size of analytes or stronger interactions (ionic,

or covalent bonding). When these constraints are satisfied,

secondary interactions between molecules can be well described

by the

FLORYꢀHUGGINS interaction parameter (χ₁₂). χ₁₂is

75 2.4.3. Performance of the sensor array in an electronic nose

25 expected to tend to zero in case of strong secondary interactions

that correspond also to high values of A_r. Normally for CPC

vapour QRS, A_rincreases exponentially with the inverse of χ₁₂

To illustrate the discrimination ability of functionalized

cyclodextrin decorated graphene sensors, an array of three CCDC

This journal is © The Royal Society of Chemistry [year]

Journal Name, [year], [vol], 00–00 | 3

^{CREATED USING THE RSC ARTI}J^CLo^Eu^TEr^Mn^Pa^Ll^ATo^Ef^(VM^ERa^.t³e^.0r^{) -}ia^SEls^{E W}C^Wh^We^.Rm^SCi^.Os^Rtr^Gy^/ELB^{ECTRONICFILES FOR DETAILS}

Page 8 of 10

ARTICLE TYPE

DOI: 10.1039/C4TB01041H

sensors, namely RGO@PYADꢀMCD, RGO@PYADꢀNCD and

RGO@PBCD responses shown in Fig. 4c corresponding to pulses

RGO@PBCD, was assembled in an eꢀnose and exposed to the 60 of 400 ppb of benzene vapour was found to give 11 and 88

same set of 10 VOC biomarkers as previously. For each analyte,

five VOC/nitrogen successive sequences were recorded and all

signals’ maxima were collected into an m by n matrix (m being

the number of measurements and n the number of sensors), and

subsequently treated with a principal component analysis (PCA)

respectively. This demonstrates that, the noise density of RGO

sensor has reduced significantly after being wrapped by

perbenzyle cyclodextrin (PBCD) along with enhancement of

sensitivity of the sensor.

algorithm implemented in the TANAGRA software PCA is a

simple and effective method that makes a multiꢀdimensional

¹⁰mapping of data onto 3 axes with a minimum loss of information.

As shown by the pattern of Fig. 4a, the two first principal

components (PC1, PC2) obtained scores of 60.75 and 25.55

respectively, accounting for 86.28 % of the total variance. This is

a quite good result considering that only three sensors were used

¹⁵in this eꢀnose compared to several tens in some devices.

Additionally, the clusters of points corresponding to target

analytes are clearly well separated, which confirms that the

CCDC sensors are effective to distinguish the ten VOC

biomarkers. The PCA map is also showing the ability of the eꢀ

²⁰nose to discriminate specific lung cancer biomarker as 2ꢀpropanol

from other alcoholic VOC and even water, which are abundant in

breath extract.

2.4.4. Limit of detection

The organic vapour flow rate was varied by blending

mixtures of saturated VOC and dry nitrogen at different ratios. In

this way the analyte flow rate was varied from 20 to 100 cm³.min^ꢀ

¹, which corresponds to variation or the solvent fraction from 0.2

30 to 1 (the concentration of benzene at saturated is 147088 ppm),

keeping the total flow rate constant at 100 cm³.min^ꢀ1. It can be

clearly seen in Fig. 4b that the amplitude of the chemoꢀresistive

responses Ar of the set of CCDC sensors is regularly increasing

with the amount of benzene molecules in their surrounding. None

³⁵of the curves are found to cross each others, which attests that

their ranking of selectivity is kept in the whole range of analyte

concentration. Thus all sensors can be considered as operational

over a wide range of concentration of VOC from the some

hundred of ppb to slightly more than a thousand hundred ppm.

Moreover, since all the lung cancer VOC biomarker

found in breath extract are usually present at some ppm (part per

million) or even some ppb (part per billion), it is important to

check that the limit of detection of CCDC sensors is compatible

with such low concentrations. In order to assess the efficiency of

⁴⁵the transducers to sense traces of analyte molecules in their

surrounding, their signal to noise ratio (SNR) was calculated with

Fig.4. a) Pattern recognition of the electronic nose exposed to a set of

Equation 5⁸⁸

lung cancer biomarker VOC after principal component analysis (PCA), b)

Chemoꢀresistive responses of a set of CCDC sensors to benzene vapour

for varying analyte flow rates (lines have been drawn to guide the eyes),

ꢸꢃ_ꢬꢹꢺ

ꢶꢷꢃ ꢂ _ꢻ

………….. (5)

ꢼꢹꢴꢽꢲꢾꢿꢽ

70 c) Responses of RGO@PBCD and RGO sensors to benzene vapour in the

ppmꢀppb concentration range

⁵⁰Where, ꢀR_maxor A_{r max}is the steadyꢀstate resistance change of the

sensor upon exposure to solvent vapour molecules (analyte), i.e.,

ratio between the maximum resistance and the baseline. σ_baseline

corresponds to the standard deviation in baseline resistance

before analyte delivery, calculated using 10 data points as shown

55 in supplementary information (Fig. S22).

The sensors’ signals can be considered as valid if, for a defined

concentration of VOC (analyte), their SNR is higher or equal to

2.0 (SNR ≥ 2.0). The calculation of S/N ratios for the RGO and

III. Conclusion

Our study is opening a novel approach to control the

75 multiscale architecture of quantum chemoꢀresistive transducers

by the supramolecular assembly of graphene and CDꢀderivate

Journal Name, [year], [vol], 00–00

Page 9 of 10

Journal of Materials Chemistry B

Dynamic Article Links

►

Journal Name

DOI: 10.1039/C4TB01041H

Cite this: DOI: 10.1039/c0xx00000x

ARTICLE TYPE

hybrids. This strategy allows to get a high sensitivity of the

ability towards the ten lung cancer VOC biomarkers studied,

nanoswitching at junctions due to the insertion of CD between 10 which could interestingly be predicted by a model based on van

graphene foils and to obtain a tailorable selectivity to cancer

biomarkers due to CD functionalization and geometry. The

resulting CCDC QRS have demonstrated a high sensitivity,

evidenced by SNR up to 88 at a concentration as low as 400 ppb

der WAALS interactions, excepted in case of a covalent bonding

used for molecules’ recognition (acetone). These findings are

promising to switch to the next step, i.e., the confrontation of eꢀ

noses to real VOC biomarkers present in patients’ breath in order

without any preconcentration of vapours or amplification of 15 to make an anticipated diagnosis of cancers.

signals. These sensors also proved to have good discrimination

19. H.T.Nagle, R.G.Osuna, and S.S.Schiffman, IEEE Spectr., 1998, 35,

Notes and references

22–31.

70 20. S. Dragonieri, M. P. van der Schee, T. Massaro, N. Schiavulli, P.

Brinkman, A. Pinca, P. Carratú, A. Spanevello, O. Resta, M. Musti,

and P. J. Sterk, Lung Cancer, 2012, 75, 326–31.

21. M. M. Arafat, B. Dinan, S. A. Akbar, and A. S. M. A. Haseeb,

Sensors, 2012, 12, 7207–58.

⁷⁵22. C. Wang, L. Yin, L. Zhang, D. Xiang, and R. Gao, Sensors, 2010, 10,

2088–106.

23. J. Lu, B. J. Park, B. Kumar, M. Castro, H. J. Choi, and J. F. Feller,

Nanotechnology, 2010, 21, 1–10.

^aSmart Plastics Group, European University of Brittany (UEB), LIMAT^Bꢀ

UBS, Lorient 56321, France; Eꢀmail:.jeanꢁfrancois.feller@univꢁubs.fr

^bCentre for Polymer Science & Engineering, Indian Institute of

Technology, Delhiꢁ110016, India

^cLab. of Analysis & Modelling for Biology & Environment (LAMBEꢁ

UMR 8587), University of Evry, 91025 Evry Cedex, France;

25 ^dSorbonne Universités, UPMC Univ. Paris 06, 75005 Paris Cedex,

France ; Eꢀmail:.philippe.guegan@upmc.fr

24. Y. Dan, Y. Cao, T. E. Mallouk, S. Evoy, and A. T. C. Johnson,

^eCNRS, UMR 8232, IPCM, Chimie des Polymères, 75005 Paris, France

† Electronic Supplementary Information (ESI) available: [details of

synthesis and characterizations of functionalized cyclodextrins and

³⁰reduced graphene oxide wrapped functionalized cyclodextrins are

available in supplementary information]. See DOI: 10.1039/b000000x/

Nanotechnology, 2009, 20, 434014.

25. B. Kumar, M. Castro, and J. F. Feller, Sensors Actuators B Chem.,

2012, 161, 621–628.

26. J. Lu, J. F. Feller, B. Kumar, M. Castro, Y. S. Kim, Y. T. Park, and J.

C. Grunlan, Sensors Actuators B Chem., 2011, 155, 28–36.

⁸⁵27. A. Bouvree, J. F. Feller, M. Castro, Y. Grohens, and M. Rinaudo,

Sensors Actuators B Chem., 2009, 138, 138–147.

28. G. Peng, E. Trock, and H. Haick, Nano Lett., 2008, 8, 3631–3635.

29. G. Peng, U. Tisch, and H. Haick, Nano Lett., 2009, 9, 1362–1368.

30. C. Wang, L. Yin, L. Zhang, D. Xiang, and R. Gao, Sensors, 2010, 10,

1. Cent. Dis. Control Prev., 2014.

2. B. de Lacy Costello, A. Amann, H. AlꢀKateb, C. Flynn, W. Filipiak,

T. Khalid, D. Osborne, and N. M. Ratcliffe, J. Breath Res., 2014, 8,

014001.

3. Y. Y. Broza, L. Zuri, and H. Haick, Sci. Rep., 2014, 4, 1–6.

4. A. Manolls, Clin. Chem., 1983, 29, 5–15.

5. M. Phillips, K. Gleeson, J. M. Hughes, J. Greenberg, R. N. Cataneo,

L. Baker, and W. P. McVay, Lancet, 1999, 353, 1930–1933.

6. J. W. Gardner, H. W. Shin, and E. L. Hines, Sensors Actuators B

Chem., 2000, 70, 19–24.

7. G. Peng, U. Tisch, O. Adams, M. Hakim, N. Shehada, Y. Y. Broza,

S. Billan, R. AbdahꢀBortnyak, A. Kuten, and H. Haick, Nat Nano,

2009, 4, 669–673.

8. N. G. Hockstein, E. R. Thaler, D. Torigian, W. T. Miller, O.

Deffenderfer, and C. W. Hanson, Laryngoscope, 2004, 114, 1701–

1705.

9. R. Ionescu, Y. Broza, H. Shaltieli, D. Sadeh, Y. Zilberman, X. Feng,

L. GlassꢀMarmor, I. Lejbkowicz, K. Müllen, A. Miller, and H. Haick,

ACS Chem. Neurosci., 2011, 2, 687–93.

2088–2106.

31. H. Bai and G. Shi, Sensors, 2007, 7, 267–307.

32. J. Lu, B. Kumar, M. Castro, and J. F. Feller, Sensors Actuators B

Chem., 2009, 140, 451–460.

33. J. F. Feller, J. Lu, K. Zhang, B. Kumar, M. Castro, N. Gatt, and H. J.

Choi, J. Mater. Chem., 2011, 21, 4142.

34. B. Kumar, M. Castro, and J. F. Feller, Carbon N. Y., 2012, 50, 3627–

3634.

35. B. Kumar, M. Castro, and J. F. Feller, J. Mater. Chem., 2012, 22,

10656.

100 36. B. Kumar, M. Castro, and J. F. Feller, Chem. sensors, 2013, 3, 41–

44.

37. S. Chatterjee, M. Castro, and J. F. Feller, J. Mater. Chem. B, 2013, 1,

4563.

38. T. T. Tung, M. Castro, T. Y. Kim, K. S. Suh, and J. F. Feller, J.

10. M. McCulloch, T. Jezierski, M. Broffman, A. Hubbard, K. Turner,

and T. Janecki, Integr. Cancer Ther., 2006, 5, 30–9.

105

Mater. Chem., 2012, 22, 21754–21766.

39. B. Kumar, M. Castro, and J. F. Feller, J. Mater. Chem., 2012, 22,

11. G. Peng, M. Hakim, Y. Y. Broza, S. Billan, R. AbdahꢀBortnyak, A.

Kuten, U. Tisch, and H. Haick, Br. J. Cancer, 2010, 103, 542–551.

12. C. L. Silva, M. Passos, and J. S. Câmara, Talanta, 2012, 89, 360–8.

13. A. P. F. Turner and N. Magan, Nat Rev Micro, 2004, 2, 161–166.

14. H. J. O’Neill, S. N. Gordon, M. H. O’Neill, R. D. GIbbons, and J. P.

Szldon, Clin. Chem., 1988, 34, 1613–1618.

60 15. G. Giubileo, Proc. SPIE, 2002, 4762, 318–325.

16. D. Smith, T. Wang, J. SuléꢀSuso, P. Spanel, and A. El Haj, Rapid

Commun. Mass Spectrom., 2003, 17, 845–50.

17. M. J. Thorpe, D. BalslevꢀClausen, M. S. Kirchner, and J. Ye, Opt.

Express, 2007, 16, 6–15.

65 18. A. D’Amico, G. Pennazza, M. Santonico, E. Martinelli, C. Roscioni,

G. Galluccio, R. Paolesse, and C. Di Natale, Lung cancer, 2010, 68,

170–6.

10656–10664.

40. B. Kumar, M. Castro, and J. F. Feller, Carbon N. Y., 2012, 50, 3627–

3634.

110 41. J. F. Feller, N. Gatt, B. Kumar, and M. Castro, ChemoSensors, 2014,

2, 26–40.

42. M. Castro, J. Lu, S. Bruzaud, B. Kumar, and J. F. Feller, Carbon N.

Y., 2009, 47, 1930–1942.

43. B. Kumar, J. F. Feller, M. Castro, and J. Lu, Talanta, 2010, 81, 908–

115

15.

44. A. Bouvrée, J. F. Feller, M. Castro, Y. Grohens, and M. Rinaudo,

Sensors Actuators B Chem., 2009, 138, 138–147.

45. B. Kumar, M. Castro, and J. F. Feller, Carbon N. Y., 2012, 50, 3627–

3634.

Journal Name, [year], [vol], 00–00 | 5

Journal of Materials Chemistry B

Page 10 of 10