Microfluidics and Nanofluidics

Influence of geometry and surrounding conditions on fluid flow in paper-based devices

--Manuscript Draft--

Manuscript Number: MANO-D-15-00393

Full Title: Influence of geometry and surrounding conditions on fluid flow in paper-based devices

Article Type: Original Article

Abstract: Fluid flow behaviour in paper is of increasing interest due to the advantages and
expanding use of microfluidic paper-based analytical devices (known as PADs), for
applications such as biomedical diagnostic tests and environmental monitoring
devices. The rapid development of enhanced features in PADs, along with a need for
increased sensitivity and specificity in the embedded chemistry has necessitated a
deeper understanding of the passively-driven fluid motion in paper to enable precise
control and consistency of the devices. It is particularly important to understand which
parameters influence the flow behaviour and to quantify their impact. We
experimentally investigate the impacts of several properties during imbibition in paper,
including geometry and the surrounding conditions (humidity and temperature). Fluid
flow velocity in paper was found to vary with temperature and width, but not with length
of the paper strip and humidity for the conditions we tested. We observed post-wetting
flow for paper-strips in contact with a large fluid reservoir, a key consideration in
applications such as environmental testing where large sample volumes are available.
We also present a mathematical model for predicting flow behaviour in PADs based
on Darcy's Law and experimentally determined values for the permeability length
scale.

Powered by Editorial Manager and ProduXion Manager from Aries Systems Corporation
Manuscript Click here to download Manuscript Manuscript FINAL.pdf

Microfluid Nanofluid Page nr. 1 of 20

1
Influence of geometry and surrounding
2
3 conditions on fluid flow in paper-based
4
5
6 devices
7
8
9 Noosheen Walji, Brendan D. MacDonald*
10
*
11 Faculty of Engineering and Applied Science, University of Ontario Institute of
12 Technology, Oshawa, L1H 7K4 Canada
13
14
15 Corresponding Author:
16
17 Brendan D MacDonald,
18 e-mail: brendan.macdonald@uoit.ca
19
20 tel: +19057218668 ext. 5716
21
fax: +19057213370
22
23
24
25 The total number of words of the manuscript, including entire text from title page to figure legends:
26 6691
27 The number of words of the abstract: 196
28
29 The number of figures: 13
30 The number of tables: 3
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63 1
64
65
 Microfluid Nanofluid Page nr. 2 of 20

Abstract:
1
2 Fluid flow behaviour in paper is of increasing interest due to the advantages and expanding use of
3
4 microfluidic paper-based analytical devices (known as PADs), for applications such as biomedical
5 diagnostic tests and environmental monitoring devices. The rapid development of enhanced features
6
7 in PADs, along with a need for increased sensitivity and specificity in the embedded chemistry has
8 necessitated a deeper understanding of the passively-driven fluid motion in paper to enable precise
9
10 control and consistency of the devices. It is particularly important to understand which parameters
11 influence the flow behaviour and to quantify their impact. We experimentally investigate the
12
13 impacts of several properties during imbibition in paper, including geometry and the surrounding
14
conditions (humidity and temperature). Fluid flow velocity in paper was found to vary with
15
16 temperature and width, but not with length of the paper strip and humidity for the conditions we
17
tested. We observed post-wetting flow for paper-strips in contact with a large fluid reservoir, a key
18
19 consideration in applications such as environmental testing where large sample volumes are
20
available. We also present a mathematical model for predicting flow behaviour in PADs based on
21
22 Darcys Law and experimentally determined values for the permeability length scale.
23
24
25 Keywords:
26
27 Paper-based devices, paper-based microfluidics, Darcys law, imbibition, permeability
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63 2
64
65
 Microfluid Nanofluid Page nr. 3 of 20

1 1 Introduction
2
3 Microfluidic paper-based analytical devices (PADs) are a versatile technology capable of
4
5 facilitating a variety of complex detection and diagnostic processes (Martinez et al., 2010; Phillips &
6 Thom, 2013; Whitesides, 2006; Yetisen, Akram, & Lowe, 2013). Recently PADs have been
7
8 adapted for complex analytical tests including diagnostics for infectious diseases (Chin et al., 2011;
9
Gong et al., 2014; Veigas et al., 2012; Watkins et al., 2013; Yen et al., 2015), cancer detection and
10
11 investigation (Shadfan et al., 2015; Su et al., 2015), heavy metals in water sources (Apilux et al.,
12
2010; Apilux et al., 2012; Nath, Arun, & Chanda, 2014; Shi et al., 2012), heavy metals in industrial
13
14 waste (Nurak, Praphairaksit, & Chailapakul, 2013), bacterial contaminants (Gunda et al., 2014; Park
15
16 et al., 2013), food analysis (Escarpa, 2014; Tee-ngam et al., 2013), airborne contaminants (Cate et
17 al., 2015; Rattanarat et al., 2013), and other biological and pollution targets (Jayawardane,
18
19 McKelvie, & Kolev, 2012; Nie et al., 2010; P. Wang et al., 2012; S. Wang et al., 2012; Zakir
20 Hossain et al., 2009). PADs are ideal for applications in low-resource settings due to their low cost,
21
22 ease of use, and passively-driven flow that allows for independence from auxiliary equipment (Hu et
23 al., 2014; Martinez et al., 2010). Testing for a broad range of analytes across various settings results
24
25 in substantial variation of test conditions for PADs, particularly for environmental conditions such
26 as temperature and humidity. It is necessary to understand the impact of these parameters on the
27
28 fluid flow behaviour in paper-based devices to ensure accurate results over a range of environmental
29 conditions.
30
31 The analytical capacity of PADs has been increasing rapidly due to the development of
32 enhanced features such as multi-dimensional device designs (Fu et al., 2010; Martinez et al, 2008;
33
34 Martinez et al., 2010; P. Wang et al., 2012), enrichment-based techniques (Feng et al., 2013),
35 filtering and multi-step reactions (Li et al., 2008), dissolvable fluidic time delays (Lutz et al., 2013),
36
37 flow control mechanisms such as pumps (Zimmermann, Schmid, & Delamarche, 2007), and analyte
38 concentration and transport after wetting (Gong et al., 2014). These enhanced features have enabled
39
40 more complex chemistries to be incorporated into paper-based platforms, which have expanded the
41
contaminants that can be detected. Many of these contaminants occur in the presence of other
42
43 similar constituents, and have low detection limits, thus increasing the need for sensitivity and
44
specificity. Engineering highly sensitive and specific PADs requires precise fluid flow control that
45
46 enables pre-programming of the chemical reactions to yield predictable and consistent signal
47
readouts. The current demands for precise flow control have necessitated an in-depth understanding
48
49 of which parameters influence the flow behaviour and what their impact is. Models that account for
50
51 these parameters can then be developed with precise flow predictions to aid in the design of accurate
52 PADs.
53
54 Models have been developed to predict fluid flow behaviour in paper based on
55 modifications of the classical Lucas-Washburn equation for capillary-driven flow (Fu et al., 2011;
56
57 Hong & Kim, 2015; Roberts et al., 2003; Schuchardt & Berg, 1990; Songok et al., 2014), or Darcys
58 law for flow in porous media (Fu et al., 2011; Hwang et al, 2011; Masoodi & Pillai, 2010; Mendez
59
60 et al., 2010; Osborn et al., 2010; Yu et al., 2011). In order to generate a model that accurately
61 predicts the flow behaviour in paper, with inputs that can be reliably and consistently determined,
62
63 3
64
65
 Microfluid Nanofluid Page nr. 4 of 20

we must ascertain which variables influence the flow through paper and incorporate their influence
1 into the models. Some models have been used to investigate flow behaviour when the width of the
2
3 strip changes in the flow direction (Fu et al., 2011; Mendez et al., 2010), or to analyze the influence
4 of wax boundaries on the flow (Hong & Kim, 2015). Parameters such as the width of uniform strips,
5
6 length of the strip, and external environmental conditions, such as temperature and humidity have
7 not yet been explored. In order to generate high accuracy predictions of flow behaviour in paper, a
8
9 deeper understanding of the variables that influence the wicking behaviour in a paper medium is
10 necessary.
11
12 In this paper, we characterize fluid flow in paper-based devices by exploring which
13 parameters influence imbibition and need to be considered when developing a model capable of
14
15 predicting flow behaviour in PADs. Specifically, we perform experiments to investigate the
16
influence of temperature, humidity, paper machine direction, strip length, and strip width on the
17
18 wicking behaviour of fluid in a paper strip. We fabricate simple PADs using chromatography paper
19
in strips of varying dimensions and configurations, and experimentally analyze the parameters. We
20
21 analyze the experimental results and develop a model based on Darcys law using empirically
22
determined values for the permeability length scale. This analysis of influential parameters can be
23
24 incorporated in design considerations for paper-based devices, and the suggested model can be used
25
26 for predicting fluid flow in paper-based devices in various conditions.
27
28 2 Materials and Methods
29
30
31 2.1. Fabrication of paper strips
32
33 Paper strips were fabricated using pure cellulose chromatography paper (Whatman Grade 1 CHR
34
35 and 17 CHR). Table 1 gives the properties of these papers, as provided by the supplier. Strip
36 dimensions were printed onto the paper using an inkjet printer (HP Deskjet 2540), and cut using a
37
38 craft paper cutter.
39
40
41 Table 1. Properties of Whatman chromatography paper
42
43 Paper Type Basis Weight Thickness Porosity Flow Rate
44 (g/m2) (mm) (mm/min)
45
46 Whatman 1 CHR 87 0.18 67.8% 4.33
47 Whatman 17 CHR 325 0.70 69.1% 6.33
48
49
50
51
2.2. Experimental apparatus
52
53
54 Experiments were conducted in a temperature and humidity controlled chamber measuring 51.5 cm
55 x 41.5 cm x 41 cm, as shown in Fig. 1a. The temperature in the chamber was adjusted using a 250
56
57 W heat lamp, measured using a digital thermometer, and controlled using a digital temperature
58 controller (TC). The humidity in the chamber was established using a submerged heater in an open
59
60 container of water, monitored using a digital hygrometer, and controlled using a digital humidity
61
controller. There was no active mixing of the air in the chamber in order to avoid convection, and
62
63 4
64
65
 Microfluid Nanofluid Page nr. 5 of 20

after changes in conditions the chamber was allowed to reach equilibrium prior to conducting
1 experiments.
2
3 (a)
4
5
6
7
8
9
10
11
12
13
14 (b)
15
16
17
18
19
20
21
22
23
24 Fig. 1. (a) Schematic of experimental setup to examine wicking behaviour in PADs using paper strips dipped
25 in a reservoir (b) Paper strip wicking from reservoir during experimentation
26
27 The paper strip (Whatman, GE Healthcare, Pittsburgh, PA, USA) was folded and placed in
28 a 10 cm diameter petri dish filled with Allura red food colour dye (Sigma Aldrich, Oakville, ON,
29
30 Canada) and to keep the strip level the opposite end was supported by an inverted petri dish as
31
shown in Fig. 1b. The wicking process was recorded using a Nikon Digital SLR camera with an AF-
32
33 S DX Micro-NIKKOR 40mm f/2.8G lens. A JEOL 6400 scanning electron microscope (SEM) was
34
used to take micrograph images.
35
36
37 3 Results and Discussion
38
39
40 Experiments were conducted to study the influence of temperature, humidity, paper machine
41 direction, strip length, and strip width on the wicking behaviour of fluid in a paper strip.
42
43
44 3.1. Influence of temperature on wicking
45
46 Since PADs are well suited as test platforms for resource-poor settings, they may be applied across
47
48 a wide range of climate zones with different temperatures and humidity values. Therefore, it is
49 important to understand the influence of the temperature and humidity of the surroundings on fluid
50
51 flow behaviour in paper-based devices to determine the design elements to be incorporated in a
52 PAD.
53
54 Fig. 2 shows the experimental results for fluid flow in paper strips with a 10 mm width and
55 a 45 mm length, tested in temperature conditions ranging from 15oC to 45oC at a fixed relative
56
57 humidity of 30%. After running the experiments, it was clear from the data that the temperature of
58 the fluid in the reservoir was what influenced the wicking behaviour and not the air temperature.
59
60 The fluid temperature is shown in brackets in the legend of Fig. 2. It can be observed in Fig. 2 that
61 total wicking time decreased as temperature increased. In conditions of 15oC fluid and air
62
63 5
64
65
 Microfluid Nanofluid Page nr. 6 of 20

temperature, a wicking time of approximately 11 minutes was required for the fluid front to travel a
1 distance of 45 mm. With a fluid temperature of 35oC (air temperature of 45oC), wicking time was
2
3 reduced by three minutes, and a total duration of approximately 8 minutes was required for the fluid
4 front to travel 45 mm in the paper strip. These results demonstrate that the time required for the fluid
5
6 front to travel the length of the paper strip decreases as temperature increases, and correspondingly
7 speed increases. This increase in wicking speed can be attributed to the decrease in the viscosity of
8
9 water as its temperature increases. To analyze this justification, Fig. 3 shows a comparison of the
10 experimental results for wicking distance at 6 minutes versus the inverse root of viscosity at varying
11
12 fluid temperatures. This correlation between wicking distance and viscosity follows Darcys law
13 (listed in Section 4 as Equation 4), where the wicking distance is inversely proportional to the root
14
15 of viscosity. The corresponding dependence of wicking distance and the viscosity term on
16
temperature shown in Fig. 3 confirms that the fluid viscosity accounts for the observed variations
17
18 with temperature.
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43 Fig. 2. Experimental results for wicking in 1 CHR strips 10 mm in width and 45 mm in length at temperature
44 conditions varying from 15oC to 45oC. Fluid temperature is indicated in brackets.
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63 6
64
65
 Microfluid Nanofluid Page nr. 7 of 20

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19 Fig. 3. Comparison of Darcys Law model to experimental data for wicking distance at 6 minutes in 1 CHR, 10
20 mm width strips at varying temperature conditions.
21
22 3.2. Influence of humidity on wicking
23
24 Wicking in paper strips with the same dimensions as the temperature tests were tested in
25
26 humidity conditions ranging from 30% to 85% relative humidity, at a fixed air temperature of 20C,
27 and the experimental results are shown in Fig. 4. Despite the variations in humidity, the wicking
28
29 time for the fluid to fill the full length of the paper strip had a consistent value of approximately 7
30 minutes. There was no observable increase or decrease in wicking velocity beyond the experimental
31
32 error that corresponded to increases in humidity for the conditions of our experiment. This result
33 indicates that PADs developed for settings with varying humidity can yield consistent flow
34
35 behaviour regardless of the humidity fluctuations.
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Fig. 4. Experimental results for wicking in 1 CHR strips 10 mm in width and 45 mm in length, for humidity
58 conditions varying from 30% to 85%.
59
60
61
62
63 7
64
65
 Microfluid Nanofluid Page nr. 8 of 20

3.3. Influence of machine direction on wicking

1
2 During the paper production process the cellulose pulp is laid into sheets, and it has been observed
3
that paper fibres tend to align parallel to the direction of the machine (Cassidy, 1952; Rasi, 2013).
4
5 The machine direction of the chromatography paper was labelled by the manufacturer, so an
6
7 investigation to determine any potential impacts of fibre arrangement could be conducted.
8 Hypotheses for the specific fluid transport mechanism in paper include wicking through capillary-
9
10 like structures formed by an alignment of pores, or wicking along adjacent fibres (Roberts et al.,
11 2003). It is expected that fluid flowing in the direction of the fibres will encounter fewer
12
13 obstructions from the fibres and therefore travel through the paper strip in a shorter time.
14 The machine direction is labelled in the SEM micrograph inset of Fig. 5 for Whatman 17 CHR
15
16 paper, and the tendency for fibre alignment parallel to machine direction can be observed. Fig. 5
17 shows the results of a comparison for fluid flow in the machine direction and cross direction for
18
19 Whatman 17 CHR paper with a width of 1 cm. It can be seen that fluid flow in the machine direction
20 is faster than in the cross direction, this observation becomes more distinct as wicking distance
21
22 increases. In Fig. 5, it can be seen that the fluid in the machine direction strip wicked to a length of
23 45mm 30% faster than in the cross direction strip. For our experimental investigations we chose to
24
25 use the machine direction, since the chromatography paper is designed to operate in the machine
26 direction and the fluid flow is less inhibited.
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52 Fig. 5. Experimental observations for wicking in a 17 CHR paper strip 10mm in width, in machine and cross
53 machine direction
54
55 3.4. Wicking in paper strips of varying lengths
56
57 To examine the impacts of paper strip dimensions on fluid wicking, strips of varying lengths were
58
59 tested to ascertain if fluid flow behaviour in paper is influenced by the length of the strip. Fig. 6
60 shows the experimental results for fluid flow in PADs of varying length and a fixed width of 10
61
62 mm, with the fluid temperature at 20 oC (ambient temperature of 22oC), and 25% humidity.
63 8
64
65
 Microfluid Nanofluid Page nr. 9 of 20

Experimental data demonstrates that there is no variation in flow behaviour for paper strips with
1 lengths varying from 25 mm to 65 mm. Though the lengths of the paper strips increased, there was
2
3 no subsequent increase or decrease in flow velocity beyond the experimental error. As such, length
4 is a flexible design parameter for PADs, and can be adjusted to meet any size requirements without
5
6 impacting the wicking behaviour in a paper-based device.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Fig. 6. Experimental results for wicking in 1 CHR strips of lengths varying from 25 mm to 65 mm, and a width
31 of 10 mm.
32
33
3.5. Wicking in paper strips of varying widths
34
35
36 Impacts of the paper strip width were also investigated. Previous studies have found different strip
37 widths require varying wicking times in paper strips with hydrophobic barriers (Hong & Kim, 2015;
38
39 Songok et al., 2014). In order to examine the influence of the width of the paper strip on fluid flow
40 behaviour, we used paper strips with a length of 5 cm and a range of widths from 0.5 cm to 4 cm in
41
42 increments of 0.5 cm.
43 Fig. 7 shows the experimental results for fluid flow in PADs of varying width using
44
45 Whatman 1 CHR. The total wicking time for paper strips of 45 mm in length was observed to
46
decrease as the width increased. This confirms that the width can influence the fluid flow behaviour
47
48 for thin strips, with diminishing dependency as the strip width increases. The same set of
49
experiments were repeated with thicker chromatography paper (Whatman 17 CHR) to give the
50
51 results in Fig. 8. For the 17 CHR paper, no variations associated with width were observed within
52
the experimental error.
53
54
55
56
57
58
59
60
61
62
63 9
64
65
 Microfluid Nanofluid Page nr. 10 of 20

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26 Fig. 7. Experimental results for wicking in 1 CHR strips of widths varying from 5 mm to 40 mm, and a length
27 of 45 mm.
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54 Fig. 8. Experimental results for wicking in 17 CHR strips of widths varying from 5 mm to 30 mm, and a length
55 of 45 mm.
56
57 3.6. Post-wetting flow
58
59
Our most unexpected observation was fluid flow in the paper strip after wetting was complete. The
60
61 paper strips remained in contact with the reservoir after the liquid front had travelled the full length
62
63 10
64
65
 Microfluid Nanofluid Page nr. 11 of 20

of the strip. Though no additional fluid transport was expected due to capillary forces, since wetting
1 was complete, further flow, or post-wetting flow, was observed in the paper strip. Post-wetting flow
2
3 caused an increase in the amount of dye in the paper strip, and was detected visually as a darkening
4 in the colour of the paper strip. ImageJ was used to analyze images of post-saturation flow to
5
6 quantify our observations.
7 Fig. 9 compares the colour intensity of a paper strip at the moment of saturation to the
8
9 colour intensity 17 minutes after saturation. Colour intensity is measured by representing each pixel
10 in the image with a numerical value between 0 and 255, where 0 is the value assigned to the darkest
11
12 pixels and 255 is assigned to the brightest red pixels. The colour profile shows a shift to the left that
13 reflects the darkening in colour due to the increase in dye within the paper strip. The darkening can
14
15 also be observed in the paper strips in the inset of Fig. 10.
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37 Fig. 9. Comparison of colour profiles of the red tone in 1 CHR paper strips (a) upon wetting for the full length
38 of the strip, and (b) 17 minutes after wetting to detect flow after wetting.
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61 Fig. 10. Colour intensity of a 1 CHR paper strip after wetting.
62
63 11
64
65
 Microfluid Nanofluid Page nr. 12 of 20

The maximum value for red pixels is the numerical value assigned by the software to the
1 brightest red pixel in the image. This value was measured every minute after saturation, where a
2
3 lower value represents a darker red colour corresponding to the increase in dye concentration, these
4 values are plotted in Fig. 10. The data demonstrates a decreasing relationship between time and
5
6 brightness of the red colour in the paper strip that levels off after approximately 15 minutes. This
7 data indicates that there exists a maximum quantity of water that will fill the paper strip during post-
8
9 wetting flow.
10 To determine if the colour change was due to increased flow from the reservoir, rather than
11
12 the diffusion of dye particles, a fully wetted strip was removed from the reservoir immediately after
13 the fluid front reached the end and the colour was compared to a strip that remained in the reservoir.
14
15 A slight redistribution of the dye was detected as the removed strip evened in colour, however, there
16
was no distinct darkening in the colour of this strip. Additionally, the mass of both strips was
17
18 measured 18 minutes after wetting. The mass of the paper strip that remained in the reservoir was
19
36% higher than the paper strip that was removed from the reservoir, thus confirming the presence
20
21 of post-wetting flow.
22
A close look at the fibre structure of paper, as seen in the SEM micrograph in Fig. 11,
23
24 shows that interfibre pores are created by the spaces between fibres, which tend to be highly variable
25
26 in size (Dullien, 1992; Roberts et al., 2003). A study by Roberts et al. (2003) observed that the bulk
27 filling of pores is not the primary flow mechanism in paper, rather it is flow along the channels
28
29 caused by fibre overlap, in other words, capillary driven film flow. The cellulose fibres also contain
30 pores within the fibre, known as intrafibre pores (Rasi, 2013; Roberts et al., 2003). Filling of these
31
32 intrafibre pores leads to fibre swelling, while the capillary forces of a liquid between fibres can
33 cause fibre deformation (Duprat et al., 2012; Rasi, 2013; Rost et al., 2007). While these fibre
34
35 swelling effects are not immediately apparent during imbibition, they could provide an explanation
36 for the presence of post-wetting flow. Movement of the fibres due to stretching or relaxation caused
37
38 by elastic forces could also account for fluid flow within the paper after wetting.
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fig. 11. Scanning electron micrographs of cellulose fibre networks in (a) 1 CHR at 100x magnification, (b) 1
61
CHR at 10000x, (c) 17 CHR at 100x, and (d) 17 CHR at 10000x
62
63 12
64
65
 Microfluid Nanofluid Page nr. 13 of 20

For design of precise PADs when using a large fluid reservoir, it will be important to
1 consider the post-wetting effects in relation to the intensity of colorimetric signals by controlling the
2
3 time of submersion and elapsed time prior to test readout by the user. This phenomenon is of
4 particular interest as the scope of paper-based microfluidic applications expands to include
5
6 environmental testing, where high sample volumes are readily available.
7
8
9 4 Model
10
11 A common approach to modelling flow in porous media is to apply the Washburn equation, which
12
approximates the porous medium as a bundle of cylindrical capillaries (Washburn, 1921), however
13
14 studies for imbibition of paper have found that the Washburn equation does not always generate
15
16 accurate predictions for fluid flow (Bhaduri et al., 2014; Das, Waghmare, & Mitra, 2012; Roberts et
17 al., 2003; Rost et al., 2007; Schuchardt & Berg, 1990). The Washburn equation is as follows
18
19 (Washburn, 1921):
20
21 (1)
22
23
24 where is distance, is the liquid-air surface tension, is the contact angle, is the kinematic
25
26 viscosity of the fluid, is the capillary radius, and is time. Washburn based models address the
27 dominant capillary forces in the micropores of the paper, however paper can also be classified as a
28
29 fibrous material with details that are not addressed by an assembly of capillaries (Alava, Dub, &
30 Rost, 2004). Fig. 11 shows SEM micrographs that illustrate how the micro-structure of paper is not
31
32 accurately represented by cylindrical flow channels, and how challenging it can be in practice to
33 ascertain what value of the capillary radius to use in models. There is a wide range of pore sizes,
34
35 with varying shapes due to the fibrous network, resulting in a complex matrix structure, and no clear
36 way to select an appropriate capillary radius value. The challenge for modelling is to approximate
37
38 this soft porous media using easy to attain variables in order to generate meaningful and consistent
39 predictions.
40
41 Darcys law has also been used to model paper-based microfluidic systems (Hwang et al.,
42 2011; Masoodi & Pillai, 2010; Mendez et al., 2010; Osborn et al., 2010; Yu et al., 2011). It was
43
44 developed empirically in 1856 to describe fluid flow velocity in porous media using the pressure
45 gradient within the medium as well as its permeability, where permeability is defined as the ability
46
47 of a porous material to allow fluids to pass through it (Darcy, 1856). Darcys law is as follows:
48
49
(2)
50
51
52 where is velocity, is permeability, and is the pressure gradient. Furthermore, if we assume
53
54 that the pressure gradient is constant and linear in the x-direction, Darcys law can be simplified by
55 substituting the Young-Laplace equation:
56
57
58 (3)
59
60 for pressure, rearranging for velocity, separating variables and then integrating, which results in:
61
62
63 13
64
65
 Microfluid Nanofluid Page nr. 14 of 20

1 (4)
( )
2
3
We simplify the model based on Darcys Law (Equation 4) with two unknown parameters,
4
5 the permeability ( ) and the ambiguous capillary radius (reff) term, to a single parameter model by
6
7 grouping the terms into a single parameter ( ) with units of length, that we call the
8 permeability length scale. The permeability length scale can be empirically determined using our
9
10 experimental data. A comparison of Equations 1 and 4 shows the similarity in the forms of the
11 Washburn equation and Darcys law after the capillary substitutions have been made. It is possible
12
13 to compare the permeability length scale to the effective capillary radius from the Washburn
14 equation by equating Equations 1 and 4, yielding the following expression:
15
16
17 ( ) (6)
18
19 The effectiveness of the Darcys law model for PADs ultimately depends on the appropriate
20
21 determination of a permeability length scale.
22 Recent studies (Koponen et al., 1998; Masoodi & Pillai, 2010; Rasi, 2013) that used
23
24 Darcys law to model fluid flow in paper demonstrate that the permeability models vary greatly,
25
have the robustness to describe several different types of porous media, and the versatility to
26
27 incorporate different parameters as necessary. We use our experimental data to generate empirical
28
values for the permeability length scale for the chromatography paper used in this study and
29
30 incorporate these into our model.
31
32
33 4.1. Permeability
34
35 Empirical values for permeability length scale were determined from the experimental data using
36
Darcys law (Equation 4). The 17 CHR paper demonstrates a constant permeability length scale
37
38 value of 5.40 x 10-4 mm, however, the permeability group value for 1 CHR varied with width, as
39
observed in Fig. 12. A minimum permeability group value for 1 CHR was observed to be 8.07 x 10-5
40
41 mm, which increased linearly with width up to a maximum value of 1.50 x 10-4 mm. Resulting
42
permeability length scale values are tabulated in Table 2.
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62 Fig. 12. Experimental observations of permeability varying with width in 1 CHR paper.
63 14
64
65
 Microfluid Nanofluid Page nr. 15 of 20

1 Table 2 Empirical values for permeability length scale in Darcys Law

2
3
Paper Type Permeability Length Scale Value (mm)
4
5 Whatman 1 CHR
6
7 Whatman 17 CHR
8
9
10 Darcys law (Equation 4) is plotted alongside some of our experimental data for
11
comparison in Fig. 13. Liquid properties used in these models are tabulated in Table 3. The surface
12
13 tension value describes the tension force at the liquid-air interface for water, while the contact angle
14
describes the angle at which a droplet of water would rest on paper as determined in previous work
15
16 (Hong & Kim, 2015; Mendez et al., 2010). The viscosity value listed is that of water at 22C
17
(Pritchard, 2011). Empirical values for the permeability length scale from Table 2 are applied in
18
19 Darcys Law. In Fig. 13 we observe good agreement between the modelling predictions and the
20
21 experimental data.
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46 Fig. 13. Comparison of the Washburn equation and Darcy's law models with experimental data for 1 CHR
47 paper 10 mm in width and 45 mm in length.
48
49
50 Table 3 Liquid properties applied in model
51
52 Property Units Value
53
54 Surface Tension ( ) (mN/m) 73.2
55 o
Contact Angle ( ) () 83
56
57 Viscosity ( , at 22oC) (Pas) 9.52 x 10-4
58
59
60 To determine the capillary radius values that will give the same modeling results using the
61 Washburn equation, Equation 6 is applied. For a 1 CHR paper strip 10 mm in width, the
62
63 15
64
65
 Microfluid Nanofluid Page nr. 16 of 20

permeability length scale value is 8.87 x 10-5 mm, which corresponds to a capillary radius value of
1 0.7 m for the Washburn equation. This capillary radius is smaller than the pore sizes observed in
2
3 the micrographs in Fig. 11, as well as the reported pore size of 0.9 to 11 m for other cellulose-
4 based and composite papers (Bhm et al., 2014; Hong & Kim, 2015; Masoodi & Pillai, 2010;
5
6 Mendez et al., 2010). In particular, Hong and Kim use a pore size of 11 m in their adapted
7 Washburn equation and empirically determine a proportionality constant k to be 0.264 m0.5 or
8
9 6.97 x 10-5 mm when calculated in the form used in Darcys law for this study. Though empirical
10 values can be determined for appropriate effective capillary radius to fit the Washburn equation to
11
12 experimental data, the resulting values remain ambiguous, as they do not intuitively correlate to
13 physical observations of paper fibre structure. Additionally, the presence of multiple active length
14
15 scales contributes an uncertainty with regards to the physical dependence between permeability on
16
the micrometer scale, and changes in width on the millimeter scale.
17
18 The permeability length scale term in Darcys law is robust and has the flexibility to
19
address the influence of additional parameters in paper-based devices as they are uncovered. The
20
21 permeability length scale also provides a useful metric to identify appropriate paper types and sizes
22
for the desired application and aid in the design of more complex PADs.
23
24
25 5 Conclusions
26
27
28 In this paper, we investigated the impact of length, width, paper machine direction, and surrounding
29 conditions such as temperature and humidity on fluid flow behaviour in paper strips. Using
30
31 experimental data, we determined that length and humidity do not have a significant impact on flow
32 behaviour in a paper-based device for the conditions we tested. We noticed quicker wicking for flow
33
34 parallel to the machine direction of the paper compared to the cross (perpendicular) direction. Our
35 experimental results indicated that the wicking time decreases as temperature increases in proportion
36
37 to decreases in the viscosity of the fluid. Additionally, we observed that wicking time also decreases
38 with increasing width for thin (1 CHR) paper strips, a dependency which diminishes as width
39
40 increases. Width was not found to influence the wicking behaviour for the thicker paper strips (17
41
CHR). We also observed some post-wetting flow when the paper strips remained in contact with a
42
43 fluid reservoir, which should be considered during design of PADs used in conjunction with large
44
sample volumes, for example, in an environmental testing context.
45
46 To develop the prediction capacity for the wicking behaviour in paper, we used Darcys
47
Law with empirically derived values for the permeability length scale. We determined the
48
49 permeability length scale for Whatman 1 CHR and 17 CHR including the dependence on width. Our
50
51 model and method for determining permeability length scale values can be used to provide accurate
52 models for fluid flow in paper that account for the parameters influencing flow without ambiguous
53
54 variables. A reliable model to help predict and design fluid flow in PADs is a valuable tool,
55 particularly with the increasingly expanding complexity of PADs.
56
57
58
59
60
61
62
63 16
64
65
 Microfluid Nanofluid Page nr. 17 of 20

6 Acknowledgements
1
2 The authors gratefully acknowledge funding support from Grand Challenges Canada through Stars
3
4 in Global Health.
5
6
7 7 Notes and references
8
9 Alava, M., Dub, M., & Rost, M. (2004). Imbibition in Disordered Media. Advances in Physics,
10 53(2), 53. doi:10.1080/00018730410001687363
11
12 Apilux, A., Dungchai, W., Siangproh, W., Praphairaksit, N., Henry, C. S., & Chailapakul, O. (2010).
13 Lab-on-paper with dual electrochemical/colorimetric detection for simultaneous determination
14
15 of gold and iron. Analytical Chemistry, 82(5), 172732. doi:10.1021/ac9022555
16 Apilux, A., Siangproh, W., Praphairaksit, N., & Chailapakul, O. (2012). Simple and rapid
17
18 colorimetric detection of Hg(II) by a paper-based device using silver nanoplates. Talanta, 97,
19 38894. doi:10.1016/j.talanta.2012.04.050
20
21 Bhaduri, S., Sahu, P., Das, S., Kumar, A., & Mitra, S. K. (2014). Capillary imbibition dynamics in
22 porous media. In ASME 12th International Converence on Nanochannels, Microchannels and
23 Minichannels.
24
25 Bhm, A., Carstens, F., Trieb, C., Schabel, S., & Biesalski, M. (2014). Engineering microfluidic
26 papers: effect of fiber source and paper sheet properties on capillary-driven fluid flow.
27 Microfluidics and Nanofluidics. doi:10.1007/s10404-013-1324-4
28
29 Cassidy, H. G. (1952). Investigation of Paper Chromatography. Analytical Chemistry, 24(9), 1415
30 1421.
31
32 Cate, D., Noblitt, S. D., Volckens, J., & Henry, C. (2015). Multiplexed paper analytical device for
33
measuring airborne metal particulates with distance-based detection. Lab on a Chip.
34
35 doi:10.1039/C5LC00364D
36
Chin, C. D., Laksanasopin, T., Cheung, Y. K., Steinmiller, D., Linder, V., Parsa, H., Wang, J.,
37
38 Moore, H., Rouse, R., Umviligihozo, G., Karita, E., Mwambarangwe, L., Braunstein S. L., van
39 de Wijgert, J., Sahabo, R., Justman, J. E., El-Sadr, W., & Sia, S. K. (2011). Microfluidics-
40 based diagnostics of infectious diseases in the developing world. Nature Medicine, 17(8),
41 10151019. doi:10.1038/nm.2408
42
43 Darcy, H. (1856). The public fountains of the city of Dijon.
44
45 Das, S., Waghmare, P. R., & Mitra, S. K. (2012). Early regimes of capillary filling. Physical Review
46 E, 86(6), 067301. doi:10.1103/PhysRevE.86.067301
47
48 Dullien, F. (1992). Porous Media: Fluid Transport and Pore Structure. San Diesgo: Academic
49 Press.
50
51 Duprat, C., Protiere, S., Beebe, A., & Stone, H. A. (2012). Wetting of flexible fibre arrays. Nature,
52 7386(482), 510513.
53
54 Escarpa, A. (2014). Lights and shadows on food microfluidics. Lab on a Chip, 32133224.
55
doi:10.1039/c4lc00172a
56
57 Feng, L., Li, X., Li, H., Yang, W., Chen, L., & Guan, Y. (2013). Enhancement of sensitivity of
58
paper-based sensor array for the identification of heavy-metal ions. Analytica Chimica Acta,
59
60 780, 7480. doi:10.1016/j.aca.2013.03.046
61
62 Fu, E., Lutz, B., Kauffman, P., & Yager, P. (2010). Controlled reagent transport in disposable 2D
63 17
64
65
 Microfluid Nanofluid Page nr. 18 of 20

paper networks. Lab on a Chip, 10(December 2009), 918920. doi:10.1039/b919614e

1
Fu, E., Ramsey, S. A., Kauffman, P., Lutz, B., & Yager, P. (2011). Transport in two-dimensional
2
3 paper networks. Microfluid Nanofluidics, 10(1), 2935.
4
5 Gong, M. M., MacDonald, B. D., Nguyen, T. V., Van Nguyen, K., & Sinton, D. (2014). Lab-in-a-
6 pen: a diagnostics format familiar to patients for low-resource settings. Lab on a Chip, 14,
7 95763. doi:10.1039/c3lc51185e
8
9 Gong, M. M., Zhang, P., MacDonald, B. D., & Sinton, D. (2014). Nanoporous membranes enable
10 concentration and transport in fully wet paper-based assays. Analytical Chemistry, 86, 80907.
11 doi:10.1021/ac502597v
12
13 Gunda, N. S. K., Naicker, S., Shinde, S., Kimbahune, S., Shrivastava, S., & Mitra, S. (2014). Mobile
14 Water Kit (MWK): a smartphone compatible low-cost water monitoring system for rapid
15
16 detection of total coliform and E. coli. Analytical Methods, 6(16), 62366246.
17 doi:10.1039/C4AY01245C
18
19 Hong, S., & Kim, W. (2015). Dynamics of water imbibition through paper channels with wax
20 boundaries. Microfluidics and Nanofluidics. doi:10.1007/s10404-015-1611-3
21
22 Hu, J., Wang, S., Wang, L., Li, F., Pingguan-Murphy, B., Lu, T. J., & Xu, F. (2014). Advances in
23 paper-based point-of-care diagnostics. Biosensors and Bioelectronics, 54, 585597.
24 doi:10.1016/j.bios.2013.10.075
25
26 Hwang, H., Kim, S.-H., Kim, T.-H., Park, J.-K., & Cho, Y.-K. (2011). Paper on a disc: balancing the
27 capillary-driven flow with a centrifugal force. Lab on a Chip, 11(20), 3404.
28 doi:10.1039/c1lc20445a
29
30 Jayawardane, B. M., McKelvie, I. D., & Kolev, S. D. (2012). A paper-based device for measurement
31
of reactive phosphate in water. Talanta, 100, 45460. doi:10.1016/j.talanta.2012.08.021
32
33 Koponen, A., Kandhai, D., Helln, E., Alava, M., Hoekstra, A., Kataja, M., Niskanen, K., Sloot, P.,
34
& Timonen, J. (1998). Permeability of three-dimensional random fiber webs. Physical Review
35
36 Letters, 80(4), 716719. doi:10.1103/PhysRevLett.80.716
37
38 Li, X., Tian, J., Nguyen, T., & Shen, W. (2008). Paper-based microfluidic devices by plasma
39 treatment. Lab on a Chip, 80(2008903553), 91319134.
40 doi:10.1039/b811135a.10.1021/ac801729t
41
42 Lutz, B., Liang, T., Fu, E., Ramachandran, S., Kauffman, P., & Yager, P. (2013). Dissolvable fluidic
43 time delays for programming multi-step assays in instrument-free paper diagnostics. Lab on a
44 Chip, 13(14), 28407. doi:10.1039/c3lc50178g
45
46 Martinez, A. W., Phillips, S. T., Nie, Z., Cheng, C.-M., Carrilho, E., Wiley, B. J., & Whitesides, G.
47 M. (2010). Programmable diagnostic devices made from paper and tape. Lab on a Chip,
48
10(19), 2499504. doi:10.1039/c0lc00021c
49
50 Martinez, A. W., Phillips, S. T., & Whitesides, G. M. (2008). Three-dimensional microfluidic
51
52 devices fabricated in layered paper and tape. Proceedings of the National Academy of Sciences
53 of the United States of America, 105(50), 1960611. doi:10.1073/pnas.0810903105
54
55 Martinez, A. W., Phillips, S. T., Whitesides, G. M., & Carrilho, E. (2010). Diagnostics for the
56 developing world: microfluidic paper-based analytical devices. Analytical Chemistry, 82(1),
57 310. doi:10.1021/ac9013989
58
59 Masoodi, R., & Pillai, K. M. (2010). Darcys Law-based model for wicking in paper-like swelling
60 porous media. AIChE Journal, 56(9), 22572267.
61
62
63 18
64
65
 Microfluid Nanofluid Page nr. 19 of 20

Mendez, S., Fenton, E. M., Gallegos, G. R., Petsev, D. N., Sibbett, S. S., Stone, H. A., Zhang, Y., &
1 Lopez, G. P. (2010). Imbibition in porous membranes of complex shape: quasi-stationary flow
2 in thin rectangular segments. Langmuir, 26(2), 13801385.
3
4 Nath, P., Arun, K., & Chanda, N. (2014). A paper based microfluidic device for the detection of
5 arsenic using a gold nanosensor. RSC Advances, 4, 5955859561. doi:10.1039/C4RA12946F
6
7 Nie, D., Li, P., Zhang, D., Zhou, T., Liang, Y., & Shi, G. (2010). Simultaneous determination of
8 nitroaromatic compounds in water using capillary electrophoresis with amperometric detection
9
on an electrode modified with a mesoporous nano-structured carbon material. Electrophoresis,
10
11 31, 29812988. doi:10.1002/elps.201000275
12
Nurak, T., Praphairaksit, N., & Chailapakul, O. (2013). Fabrication of paper-based devices by
13
14 lacquer spraying method for the determination of nickel (II) ion in waste water. Talanta, 114,
15 2916. doi:10.1016/j.talanta.2013.05.037
16
17 Osborn, J. L., Lutz, B., Fu, E., Kauffman, P., Stevens, D. Y., & Yager, P. (2010). Microfluidics
18 without pumps: reinventing the T-sensor and H-filter in paper networks. Lab on a Chip,
19 10(20), 26592665. doi:10.1039/c004821f
20
21 Park, T. S., Li, W., McCracken, K. E., & Yoon, J. Y. (2013). Smartphone quantifies Salmonella
22 from paper microfluidics. Lab on a Chip, 13, 483240. doi:10.1039/c3lc50976a
23
24 Phillips, S. T., & Thom, N. K. (2013). Three-dimensional, paper-based microfluidic devices
25 containing internal timers for running time-based diagnostic assays. Methods in Molecular
26
27 Biology, 949, 323. doi:10.1007/978-1-62703-134-9
28 Pritchard, P. J. (2011). Fox and McDonalds Introduction to Fluid Mechanics (8th ed.). Hoboken,
29
30 NJ: Wiley.
31
Rasi, M. (2013). Permeability Properties of Paper. University of Jyvaskyla.
32
33 Rattanarat, P., Dungchai, W., Cate, D. M., Siangproh, W., Volckens, J., Chailapakul, O., & Henry,
34
C. S. (2013). A microfluidic paper-based analytical device for rapid quantification of
35
36 particulate chromium. Analytica Chimica Acta, 800, 505. doi:10.1016/j.aca.2013.09.008
37
38 Roberts, R. J., Senden, T. J., Knackstedt, M. A., & Lyne, M. B. (2003). Spreading of aqueous
39 liquids in unsized papers is by film flow. Journal of Pulp and Paper Science, 29(4), 123131.
40
41 Rost, M., Laurson, L., Dub, M., & Alava, M. (2007). Fluctuations in fluid invasion into disordered
42 media. Physical Review Letters, 98(5), 14. doi:10.1103/PhysRevLett.98.054502
43
44 Schuchardt, D., & Berg, J. (1990). Liquid transport in composite cellulose-superabsorbent fiber
45 network. Wood Fiber Sci, (23), 342357.
46
47 Shadfan, B. H., Simmons, A. R., Simmons, G. W., Ho, A., Wong, J., Lu, K. H., Bast R.C, &
48 McDevitt, J. T. (2015). A multiplexable, microfluidic platform for the rapid quantitation of a
49 biomarker panel for early ovarian cancer detection at the point of care. Cancer Prevention
50 Research, 8(37).
51
52 Shi, J., Tang, F., Xing, H., Zheng, H., Bi, L., & Wang, W. (2012). Electrochemical detection of Pb
53 and Cd in paper-based microfluidic devices. Journal of the Brazilian Chemical Society, 23(6),
54
55 11241130. doi:10.1590/S0103-50532012000600018
56
Songok, J., Tuominen, M., Teisala, H., Haapanen, J., Makela, J., Kuusipalo, J., & Toivakka, M.
57
58 (2014). Paper-based microfluidics: fabrication technique and dynamics of capillary-driven
59 surface flow. ACS Applied Materials & Interfaces, 6, 2006020066.
60
61 Su, M., Ge, L., Kong, Q., Zheng, X., Ge, S., Li, N., Yu, J., & Yan, M. (2015). Cyto-sensing in
62
63 19
64
65
 Microfluid Nanofluid Page nr. 20 of 20

electrochemical lab-on-paper cyto-device for in-situ evaluation of multi-glycan expressions on

1 cancer cells. Biosensors and Bioelectronics, 63, 232239.
2
3 Tee-ngam, P., Nunant, N., Rattanarat, P., Siangproh, W., & Chailapakul, O. (2013). Simple and
4 rapid determination of ferulic acid levels in food and cosmetic samples using paper-based
5 platforms. Sensors (Basel, Switzerland), 13(10), 1303953. doi:10.3390/s131013039
6
7 Veigas, B., Jacob, J. M., Costa, M. N., Santos, D. S., Viveiros, M., Incio, J., Martins, R.,
8 Barquinha, P., Fortunato, E., & Baptista, P. V. (2012). Gold on paperpaper platform for Au-
9
nanoprobe TB detection. Lab on a Chip, 12(22), 4802. doi:10.1039/c2lc40739f
10
11 Wang, P., Ge, L., Yan, M., Song, X., Ge, S., & Yu, J. (2012). Paper-based three-dimensional
12
electrochemical immunodevice based on multi-walled carbon nanotubes functionalized paper
13
14 for sensitive point-of-care testing. Biosensors and Bioelectronics, 32(1), 238243.
15 doi:10.1016/j.bios.2011.12.021
16
17 Wang, S., Ge, L., Song, X., Yu, J., Ge, S., Huang, J., & Zeng, F. (2012). Paper-based
18 chemiluminescence ELISA: Lab-on-paper based on chitosan modified paper device and wax-
19 screen-printing. Biosensors and Bioelectronics, 31(1), 212218.
20
doi:10.1016/j.bios.2011.10.019
21
22 Washburn, E. W. (1921). The dynamics of capillary flow. Physical Review, 17, 273283.
23
doi:10.1103/PhysRev.17.273
24
25 Watkins, N. N., Hassan, U., Damhorst, G., Ni, H., Vaid, A., Rodriguez, W., & Bashir, R. (2013).
26
27 Microfluidic CD4+ and CD8+ T lymphocyte counters for point-of-care HIV diagnostics using
28 whole blood. Science Translational Medicine, 5(214). doi:10.1126/scitranslmed.3006870
29
30 Whitesides, G. M. (2006). The origins and the future of microfluidics. Nature, 442(7101), 36873.
31 doi:10.1038/nature05058
32
33 Yen, C. W., de Puig, H., Tam, J. O., Gmez-Mrquez, J., Bosch, I., Hamad-Schifferli, K., &
34 Gehrke, L. (2015). Multicolored silver nanoparticles for multiplexed disease diagnostics:
35 distinguishing dengue, yellow fever, and Ebola viruses. Lab on a Chip, 15, 16381641.
36 doi:10.1039/C5LC00055F
37
38 Yetisen, A. K., Akram, M. S., & Lowe, C. R. (2013). Paper-based microfluidic point-of-care
39 diagnostic devices. Lab on a Chip, 13(12), 221051. doi:10.1039/c3lc50169h
40
41 Yu, J., Ge, L., Huang, J., Wang, S., & Ge, S. (2011). Microfluidic paper-based chemiluminescence
42 biosensor for simultaneous determination of glucose and uric acid. Lab on a Chip, 11(7),
43
44 12861291. doi:10.1039/c0lc00524j
45
Zakir Hossain, S. M., Luckham, R. E., Smith, A. M., Lebert, J. M., Davies, L. M., Pelton, R. H.,
46
47 Filipe, C. D. M., & Brennan, J. D. (2009). Development of a bioactive paper sensor for
48 detection of neurotoxins using piezoelectric inkjet printing of sol-gel-derived bioinks.
49 Analytical Chemistry, 81(13), 54745483. doi:10.1021/ac900660p
50
51 Zimmermann, M., Schmid, H., & Delamarche, E. (2007). Capillary pumps for autonomous capillary
52 systems, (October 2006), 119125. doi:10.1039/b609813d
53
54
55
56
57
58
59
60
61
62
63 20
64
65