Polyurethane urea membranes for membrane blood ......Polyurethane urea membranes for membrane blood...

Polyurethane urea membranes for membrane blood

oxygenators: synthesis and gas permeation properties

Tiago Mendonça Eusébio

Thesis to obtain the Master of Science Degree in

Chemical Engineering

Supervisors:

Dr. Eduardo Jorge Morilla Filipe

Dr. Mónica Cristina Faria Besteiro

Examination Committee

Chairperson: Dr. Carlos Manuel Faria de Barros Henriques

Supervisor: Dr. Eduardo Jorge Morilla Filipe

Member of the committee: Dr. Carla Maria Carvalho Gil Brazinha de Barros

Ferreira

November 2017

Polyurethane urea membranes for membrane blood

oxygenators: synthesis and gas permeation properties

Tiago Mendonça Eusébio

Thesis to obtain the Master of Science Degree in

Chemical Engineering

Supervisors:

Dr. Eduardo Jorge Morilla Filipe

Dr. Mónica Cristina Faria Besteiro

Dr. Maria Norberta Neves Correia de Pinho

Examination Committee

Chairperson: Dr. Carlos Manuel Faria de Barros Henriques

Supervisor: Dr. Eduardo Jorge Morilla Filipe

Member of the committee: Dr. Carla Maria Carvalho Gil Brazinha de Barros

Ferreira

November 2017

i

Abstract

Nonporous symmetric (PU) and integral asymmetric (PEU) poly(ester urethane urea)

membranes were synthetized and characterized in terms of: i) structure by Scanning Electron

Microscopy (SEM), and ii) gas permeation properties in a custom-made set-up constructed and

optimized for the measurement of the N2, CO2 and O2 permeation fluxes at constant temperature.

The membranes were synthesized by a modified version of the phase inversion technique

where polyurethane (PUR) and polycaprolactone-diol (PCL-diol) prepolymers react in a solvent

mixture of dimethyl formamide (DMF) and diethyl ether (DEE), during the casting solutions

preparation step. Total polymer to solvent weight ratio, solvent evaporation time and PCL quantity

were varied.

SEM micrographs showed that the integral asymmetric poly(ester urethane urea)

membranes have a characteristic cross section structure with no visible dense layer but instead

three distinct porous regions.

The CO2 permeabilities obtained for the nonporous symmetric membranes with 0, 5 and

15% wt. ratio of PCL were 163 Barrer, 94 Barrer and 218 Barrer, respectively.

The average permeances obtained for the integral asymmetric poly(ester urethane urea)

membranes prepared with, total polymer/total solvent ratio 1/1, 5 minutes solvent evaporation

time and PCL content between 0-15 wt.% were 0.13 ± 0.01 × 10−5 cm3cm−2s−1cmHg−1 for CO2,

0.011 ± 0.003 × 10−5 cm3cm−2s−1cmHg−1 for O2 and 0.004 ± 0.001 × 10−5 cm3cm−2s−1cmHg−1

for N2.

iii

Resumo

Membranas densas simétricas (PU) e membranas assimétricas (PEU) de poliéster-

uretano-ureia foram preparadas e caracterizadas em termos de: i) estrutura, por microscopia

eletrónica de varrimento (MEV), e ii) propriedades de permeação gasosa, numa unidade

construída para medir fluxos de permeação de N2, CO2 e O2 a temperatura constante.

As membranas foram sintetizadas pelo método de inversão de fase modificado, em que

se fez reagir os pré-polímeros de poliuretano (PUR) e policaprolactona-diol (PCL-diol) num

sistema de solventes de dimetilformamida (DMF) e dietiléter (DEE), durante a etapa de

preparação da solução de casting. Variou-se o rácio mássico entre polímero e solvente, o tempo

de evaporação de solvente e a quantidade de PCL.

Imagens MEV mostraram que as membranas assimétricas de poliuretano-ureia são

caracterizadas por uma secção de corte constituída por três regiões porosas distintas, em vez

de uma camada densa bem definida.

Os valores de permeabilidade de CO2 obtidos para as membranas densas com 0, 5 e 15

%(m/m) de PCL foram 163 Barrer, 94 Barrer e 218 Barrer, respetivamente.

As permeancias médias obtidas para as membranas assimétricas de poliuretano-ureia,

preparadas com um rácio mássico entre polímero e solvente de 1/1, um tempo de evaporação

de solvente de 5 minutos e quantidades de PCL entre 0-15 %(m/m) foram 0.13 ± 0.01 × 10−5

cm3cm−2s−1cmHg−1 para CO2, 0.011 ± 0.003 × 10−5 cm3cm−2s−1cmHg−1 para O2 e 0.004 ±

0.001 × 10−5 cm3cm−2s−1cmHg−1 para N2.

v

Keywords

gas permeability

bi-soft segment polyurethanes

integral asymmetric membranes

Membrane Blood Oxygenators

vii

Agradecimentos

Ao Professor Eduardo Filipe pela proposta e orientação desta tese. Grato por todo o tempo

dispensado, pelas suas explicações e por todo o apoio.

À Doutora Mónica Faria pela orientação desta tese. Agradeço os seus esclarecimentos, o seu

tempo e toda a motivação e energia que me transmitiu.

À Professora Maria Norberta de Pinho pela orientação desta tese, por todos os conselhos e

conhecimentos transmitidos.

Ao Doutor Pedro Morgado pelas incontáveis horas dedicadas a este trabalho e pela sua

resiliência em ultrapassar obstáculos.

À Cíntia e ao Gonçalo por toda a ajuda no laboratório.

Aos meus pais, aos meus avós e ao meu irmão pelo seu apoio incondicional ao longo destes

anos.

À Cinara pela paciência, companhia e incentivo, sem os quais teria sido muito mais difícil.

ix

Contents

1. Introduction ............................................................................................................................ 1

1.1 Extracorporeal Membrane Oxygenation Systems ......................................................... 2

1.2 Blood Oxygenators ........................................................................................................ 3

1.3 Membrane Blood Oxygenators ...................................................................................... 3

1.4 Membrane Blood Oxygenator Specifications ................................................................ 4

2. Literature Review on Polyurethane Membranes for MBOs ................................................... 7

2.1 Polyurethane Membranes ............................................................................................. 7

2.2 Gas permeability of segmented polyurethane symmetric membranes ......................... 8

2.3 Gas permeability of bi-soft segment polyurethane membranes .................................... 9

2.4 Hemocompatibility of bi-soft polyurethane membranes .............................................. 10

2.5 Integral asymmetric poly(ester urethane urea) membranes ....................................... 11

3. Framework and Thesis Objectives ...................................................................................... 13

4. Mass Transport Phenomena in Homogeneous Membranes .............................................. 15

5. Experimental ........................................................................................................................ 21

5.1 Materials ...................................................................................................................... 21

5.1.1 Commercial Membrane ....................................................................................... 21

5.1.2 Materials for the synthesis of poly(urethane urea) membranes .......................... 22

5.1.3 Gases .................................................................................................................. 23 5.2 Synthesis of integral asymmetric poly(ester urethane urea) membranes ................... 24

5.2.1 Variation of total polymer to total solvent weight ratio ......................................... 24

5.2.2 Variation of solvent evaporation time .................................................................. 25

5.2.3 Variation of PUR/PCL weight ratio maintaining solvent evaporation time .......... 25 5.3 Synthesis of nonporous symmetric poly(urethane urea) membranes ......................... 26

5.4 Characterization of the membranes structure by Scanning Electron Microscopy ...... 27

5.5 Gas Permeation experiments - Volumetric Method .................................................... 27

5.6 Gas Permeation experiments - Pressure Method ....................................................... 28

6. Results and Discussion ....................................................................................................... 31

6.1 Surface morphology and cross section structure analysis of integral asymmetric PEU

membranes by Scanning Electron Microscopy ....................................................................... 31

6.1.1 PEU membranes prepared with different total polymer to total solvent weight

ratio ...................................................................................................................................... 31

6.1.2 PEU membranes prepared with different solvent evaporation time .................... 33

6.1.3 PEU membranes prepared with different PUR to PCL weight ratios .................. 35 6.2 Surface morphology and cross-section structure analysis of nonporous symmetric PU

membranes by SEM ................................................................................................................ 38

6.3 Gas Permeation by the Volumetric Method ................................................................ 39

6.4 Gas Permeation by the Pressure Method ................................................................... 41

6.4.1 Volumetric method vs Pressure method ............................................................. 44 6.5 CO2 permeation properties of the PU nonporous symmetric membranes .................. 45

6.6 CO2 permeation properties of the PEU integral asymmetric membranes prepared with

different solvent evaporation time ........................................................................................... 47

6.7 CO2, N2, and O2 permeation properties of the PEU membranes with different PU/PCL

weight ratios. ........................................................................................................................... 48

6.7.1 CO2 permeation of integral asymmetric PEU Membranes .................................. 48

6.7.2 N2 permeation of integral asymmetric PEU Membranes ..................................... 49

6.7.3 O2 permeation of integral asymmetric PEU membranes .................................... 51

6.7.4 Comparing the different gases permeation of asymmetric PEU membranes ..... 52 6.8 Determination of CO2 diffusion and solubility coefficients ........................................... 56

6.8.1 Time-lag method .................................................................................................. 56

6.8.2 Early approximation method ................................................................................ 57 7. Conclusions ......................................................................................................................... 59

x

8. Perspectives of the future work ........................................................................................... 61

9. Bibliography ......................................................................................................................... 63

Annex – Custom-made set-up calibrations and tests.................................................................. 65

Paroscientific pressure transmitter calibration .................................................................... 65

Set up Volume Calibration ................................................................................................... 66

Setra pressure transmitter calibration ................................................................................. 68

Air bath temperature ............................................................................................................ 69

xi

List of Figures

Figure 1 – Schematic representation of an ECMO system. [3] .................................................... 2

Figure 2 — (a) A true silicone membrane (sheet) oxygenator (Avecor) [8]; (b) A Hollow fibers

oxygenator (Xenios Medos hilite oxygenator). ............................................................ 4

Figure 3 – Principle of the membrane blood oxygenator [2]. ........................................................ 5

Figure 4 – Polyurethane structural unit. [11] ................................................................................. 7

Figure 5 – Polyurethanes’ synthesis reaction [11]. ....................................................................... 7

Figure 6 — Hard segment (HS) and soft segment (SS) components of polyurethane elastomers.

..................................................................................................................................... 8

Figure 7 – Reaction of the prepolymer with a chain extender (diol) to obtain a bi-soft segmented

polyurethane [11]. ........................................................................................................ 9

Figure 8 - Representative pressure vs time graph with the intercept of the linear section with the

time axis, tc, and the pressure axis, pc...................................................................... 18

Figure 9 – Representative lntdpdt vs 1t plot for the early approximation. .................................. 19

Figure 10 – The 0600 MBO (left) and ECMOtherm II (right) commercialized by

Avecor/Medtronics [3] ................................................................................................ 21

Figure 11 – SEM images of upper surface, cross-section and the bottom support surface of

membrane Avecor/Medtronics oxygenator model 0600. [31] ................................... 21

Figure 12 – Prepolymers’ molecular structure. [24] .................................................................... 22

Figure 13 — Molecular structure of poly(urethane urea) prepared by PUR prepolymer and

composed of only one SS (PPO) and HS type I. [22] ............................................... 23

Figure 14 – Molecular structure of poly(urethane urea) prepared with PUR and PCL-diol

prepolymers containing two types of SSs (PPO and PCL) and type I and type II HSs.

[22] ............................................................................................................................. 23

Figure 15 – Schematic representation of the permeation cell. .................................................... 27

Figure 16 – Scheme of the experimental set-up used in permeation flux measurements by the

volumetric method. .................................................................................................... 28

Figure 17 – Scheme of the set-up used in permeation flux measurements by constant volume

method. ...................................................................................................................... 29

Figure 18 – Photography of apparatus used for permeation measurements by barometric

method inside air bath ............................................................................................... 30

Figure 19 – SEM images of samples of PEU3: (a) top, (b) cross-section, (c) bottom; PEU4: (d)

top, (e) cross-section, (f) bottom; PEU5: (g) top, (h) cross-section, (i) bottom. ........ 33

Figure 20 – SEM images of samples of PEU-1-100: (a) top, (b) cross-section, (c) bottom; PEU-

5-100: (d) top, (e) cross-section, (f) bottom; PEU-10-100: (g) top, (h) cross-section,

(i) bottom. .................................................................................................................. 34

Figure 21 – SEM images of samples of PEU-1-95: (a) top, (b) cross-section, (c) bottom; PEU-1-

90: (d) top, (e) cross-section, (f) bottom; PEU-1-85: (g) top, (h) cross-section, (i)

bottom. ....................................................................................................................... 36

Figure 22 – SEM images of samples of PEU-5-95: (a) top, (b) cross-section, (c) bottom; PEU-5-

90: (d) top, (e) cross-section, (f) bottom; PEU-5-85: (g) top, (h) cross-section, (i)

bottom. ....................................................................................................................... 37

Figure 23 – SEM images of samples of PU1: (a) top surface, (b) cross-section. ....................... 38

Figure 24 – Volume of N2 measured as a function of time for CM membrane sample by the

constant pressure method. Pf = 3.6 bar. ................................................................... 39

Figure 25 – Volume of N2 measured as a function of time for three different experiments of the

CM membrane by the constant pressure method at a feed pressure of = 5.0 bar. .. 40

Figure 26 – Volume of N2 measured as a function of time for the same sample of the CM

membrane by the constant pressure method at feed pressures of 3.6 bar (orange

dots) and 5.0 bar (blue dots). .................................................................................... 40

Figure 27 – Volumetric fluxes as a function of TMP of CM membrane. ...................................... 41

Figure 28 – CO2 permeate pressure (mbar) measurement as a function of time (s) for the PEU-

5-100 membrane (CO2 feed pressure of 2.3 bar). ..................................................... 42

xii

Figure 29 – Permeate pressure (mbar) measurements as function of time (s) for samples i)

(green), ii) (blue); iii (black) of the PEU-5-100 membrane at CO2 feed pressures of

1.6±0.1 bar; 2.0±0.1 bar; 2.3±0.1 bar; 2.8±0.1 bar. .................................................. 42

Figure 30 – CO2 volumetric fluxes (10 − 5 cm3cm − 2s − 1) as function of TMP (cmHg),

obtained for samples i) (green); ii) (blue); iii) (black) and iv) (grey) of PEU-5-100

membrane. ................................................................................................................ 44

Figure 31 – N2 volumetric fluxes flux J(10 − 5 cm3cm − 2s − 1) as function of TMP (cmHg)

obtained by the pressure method (dots) and volumetric method (triangles). ............ 45

Figure 32 – Average CO2 volumetric fluxes J(10 − 5 cm3cm − 2s − 1) as function of TMP

(cmHg) for the PU100 (blue), PU95(orange) and PU85(grey) membranes. ............. 46

Figure 33 – (a) Average CO2 volumetric fluxes J10 − 5 cm3cm − 2s − 1 as function of TMP

(cmHg) for the PEU-1-100 (green), PEU-5-100 (blue), PEU-10-100 (orange) and

PEU-15-100 (grey) membranes; (b) Average CO2 volumetric fluxes J10 − 5 cm3cm −

2s − 1 as function of the transmembrane pressure (cmHg) for the PEU-5-100 (blue),

PEU-10-100 (orange) and PEU-15-100 (grey) membranes ..................................... 48

Figure 34 – Average CO2 volumetric fluxes 𝐽(10 − 5 𝑐𝑚3𝑐𝑚 − 2𝑠 − 1) as function of TMP

(cmHg) of PEU-5-100 (blue), PEU-5-95 (orange), PEU-5-90 (grey) and PEU-5-85

(yellow) membranes. ................................................................................................. 49

Figure 35 – Average N2 volumetric fluxes 𝐽(10 − 5 𝑐𝑚3𝑐𝑚 − 2𝑠 − 1) as function of TMP (cmHg)

of PEU-1-100 (blue), PEU-1-95 (orange), PEU-1-90 (grey), PEU-1-85 (yellow). ..... 50

Figure 36 – Average N2 volumetric fluxes 𝐽10 − 5 𝑐𝑚3𝑐𝑚 − 2𝑠 − 1 as function of TMP (cmHg) of

the PEU-5-100 (blue), PEU-5-95 (orange), PEU-5-90 (grey), PEU-5-85 (yellow). ... 51

Figure 37 – Average O2 volumetric fluxes 𝐽10 − 5 𝑐𝑚3𝑐𝑚 − 2𝑠 − 1 as function of TMP (cmHg) of

PEU-5-100 (blue), PEU-5-95 (orange), PEU-5-90 (grey), PEU-5-85 (yellow). ......... 52

Figure 38 – Average N2 (blue), CO2 (orange) and O2 (grey) volumetric fluxes 𝐽(10 − 5 𝑐𝑚3𝑐𝑚 −

2𝑠 − 1) as function of TMP (cmHg) of PEU-5-100 (a), PEU-5-95 (b), PEU-5-90 (c),

PEU-5-85 (d). ............................................................................................................ 54

Figure 39 – Measurements of N2 (blue), CO2 (orange) and O2 (grey) flux J(10 − 5 cm3cm − 2s −

1) as function of TMP (cmHg) for the CM membrane. .............................................. 55

Figure 40 – CO2 permeate pressure (mbar) measurement as a function of time (s) for the PU95

membrane (CO2 feed pressure of 1.9 bar). ............................................................... 57

Figure 41 – lntdpdt vs 1t plot obtained from the CO2 permeate pressure (mbar) measurement

as a function of time (s) for the PU95 membrane (CO2 feed pressure of 1.9 bar). ... 58

Figure 42 – Values of pressure (mbar) in an installation during a vacuum test, read by each

transmitter: model 1100A-CE (blue dots); model 6100A-CE (orange dots). ............ 65

Figure 43 — Plot of model 6100A-CE pressure measurements in function of model 1100A-CE

pressure measurements. ........................................................................................... 66

Figure 44 – Pressure measurements of Paroscientific transmitter during the test. .................... 68

Figure 45 – Pressure measurements of the Setra transmitter (blue dots) and of the reducing

valve manometer (orange dots) as function of pressure measurements of the

Paroscientific transmitter at room temperature. ........................................................ 68

Figure 46 — Temperature peak measured in air bath (Keithley multimeter) .............................. 69

Figure 47 – Internal temperature(ºC) (orange dots) and set-up pressure (mbar) (blue dots)

(Paroscientific transmitter) ......................................................................................... 70

Figure 48 – Constant phase temperature measurements in air bath (Keithley multimeter). ...... 70

xiii

List of Tables

Table 1 – Chemical composition of the PEU membranes prepared with varying contents of PCL

diol, total polymer to total solvent weight ratio and the types of HSs and SSs that they

contain. ......................................................................................................................... 24

Table 2 – Chemical composition of PEU membranes prepared varying solvent evaporation times.

..................................................................................................................................... 25

Table 3 – Chemical composition of PEU membranes prepared varying contents of PCL diol,

solvent evaporation times and the types of hard and soft segments that they contain.

..................................................................................................................................... 26

Table 4 – Chemical composition of PU membranes prepared varying contents of PCL diol and

the types of hard and soft segments that they contain. ............................................... 27

Table 5 – total membrane thickness obtained for the PEU membranes and the respective

standard deviation. ....................................................................................................... 38

Table 6 – Membranes upper denser layer mean thickness (μm) and the membrane mean

thickness (μm) and measurements’ standard deviations. ........................................... 39

Table 7 – Slopes of the linear part of the permeate pressure vs time curves of samples i), ii) and

iii) of the PEU-5-100 membranes at CO2 feed pressures of 1.6±0.1 bar; 2.0±0.1 bar;

2.3±0.1 bar; 2.8±0.1 bar. .............................................................................................. 43

Table 8 – Method used and permeance values measured for each CM sample. ...................... 45

Table 9 – Values of total membrane thickness, mean Permeation and permeability coefficient

obtained for the PU membranes. ................................................................................. 47

Table 10 – Values of average permeance obtained for PEU membranes ................................ 48

Table 11 – CO2 average permeance obtained for PEU membranes. ......................................... 49

Table 12 – N2 average permeance obtained for PEU membranes. ............................................ 50

Table 13 – N2 average permeance obtained for PEU membranes. ........................................... 51

Table 14 – O2 average permeance obtained for PEU membranes. ........................................... 52

Table 15 – Average permeance of N2, CO2 and O2, obtained for PEU membranes. ................. 54

Table 16 – Values of membrane thickness, and the N2, CO2 and O2 mean permeance and

permeability obtained for the commercial membrane. ................................................. 56

Table 17 – TMP, membrane thickness and CO2 permeation properties of the PU95 membrane

obtained by the time-lag method. ................................................................................ 57


obtained by the early approximation method. .............................................................. 58

xv

Abbreviations and Symbols

MBO – Membrane blood oxygenator

PUR – Polyurethane prepolymer

PPO – polypropylene oxide

PBDO – Polybutadine-diol

PDMS – Polydimethylsiloxane

PCL – Polycaprolactone

PU – Nonporous symmetric poly(ester urethane urea) membranes

PEU – Integral asymmetric poly(ester urethane urea) membranes

SEM – Scanning electron microscopy

P, k – Permeability coefficient

D – Diffusion coefficient

S, K – Solubility coefficient

Perm – Permeance

TMP – Transmembrane pressure

J – Volumetric flux

𝑐 – Volumetric concentration

δ – Membrane thickness

𝑛 – Number of moles

pf – Feed pressure

pp – Permeate pressure

a – Membrane active surface area

t – Time

V – Volume

T – Temperature

R – Ideal gas constant

PpT – Pressure transmitter on the permeate side

PfT – Pressure transmitter on the feed side

tc – characteristic time

pc – characteristic pressure

1

1. Introduction

The human cardiovascular system is composed by blood vessels, lymphatic vessels and the

heart. Its main function is to circulate blood through the body to guarantee the supply of oxygen and

other nutrients to cells and to remove the products resulting from the cellular metabolism. Blood flows

through a closed and pressurized system of vessels called the circulatory system, which contains a

constant volume of five liters of blood in an average adult. If stretched, the network of blood vessels

would have a length of 150,000 kilometers. The activity of the cardiovascular system defines the life or

death of an organism, as it is one of the first systems to become functional, during the third week of

gestation, as the primordial heart begins to beat. [1]

In order to survive and be fully functional, the human body requires a constant supply of oxygen

to the cells, tissues and organs. The respiratory system is responsible for simultaneously providing

oxygen and removing carbon dioxide to the circulatory system, and it can be divided in three major parts:

the airway, the lungs, and the muscles of respiration.

The function of the lungs is to capture air from the environment to the body and expel the

products of cellular metabolism.

The airway, which consists of the nose, mouth, pharynx, larynx, trachea, bronchi, and

bronchioles, carries air between the lungs and the body’s exterior. The trachea supports the bronchial

structure that carries the air to the lungs. The trachea extends from the bottom of the throat to the middle

of the thorax where it branches into the left bronchus that goes to the left lung and the right bronchus

that goes to the right lung. Each bronchus rapidly divides into branches in the pulmonary lobes, further

subdividing into bronchioles, in turn ending in clusters of alveoli.

The alveoli act as gas exchangers and each lung contains about 150 million of these structures.

The alveoli are microscopic bags which are organized in alveolar clusters. In the alveoli, a cellular

membrane of a single cell thickness (1 μm) constitutes an interface between the pulmonary system and

the cardiovascular system, and allows oxygen and carbon dioxide to be received from the compressed

air to the blood stream in the capillaries and vice versa. The cluster arrangement greatly exceeds the

surface area available for exchange of oxygen in the confined space in the chest. A total surface area

of alveoli in an adult individual is about 80 m2, which corresponds to a tennis court. [2]

The lungs of an adult, on average, expand and squeeze between fifteen and twenty times per

minute. The expansion or inhalation draws air into lungs, while compression or exhalation expels carbon

dioxide and other gases resulting from the metabolic activity of the cells. The heart pumps the entire

volume of blood from the lungs, around the body and back to the heart in about one minute, during

which time gas exchange takes place in a dense network of capillaries that surround the pulmonary

alveoli.

Pulmonary and cardiovascular mechanisms maintain an intrinsic balance between blood flow

and airflow, constantly adjusting the blood flow in the capillaries surrounding the alveoli, according to

the internal receptor. This mechanism promotes better efficiency in the transport of oxygen into the

bloodstream. [1]

2

1.1 Extracorporeal Membrane Oxygenation Systems

Extracorporeal membrane oxygenation (ECMO) is a medical technique of providing prolonged

artificial breathing and heart support to patients whose cardiovascular and or pulmonary systems are

not functioning normally. The technology of ECMO stemmed from cardiopulmonary bypass (CPB),

which provides shorter-term support. The use of CPB technology allows surgeons to perform cardiac

procedures, such as coronary artery bypass grafts, repair of valves and repair of aortic aneurysms, in a

motionless, bloodless environment. CPB technology incorporates an extracorporeal circuit that provides

physiological support by temporarily replacing the functions of the heart and lungs, maintaining blood

circulation and ensuring gas exchange. Unlike standard CPB, which is used under general anesthesia

for short-term support (hours), ECMO is used for longer-term support (days) and its purpose is to allow

the heart and lungs time to recover and heal [3]. The ECMO circuit is similar to the one used in CPB.

Figure 1 shows the representation of a typical circuit in which, the blood is drained from the heart and

lungs by gravity through the venous cannulation and tubing into a reservoir and then pumped to an

artificial lung. After being oxygenated, the blood is pumped back into the arterial system [4]. The artificial

lung also known as the blood oxygenator is the main component of the extracorporeal circuit as it is

responsible for supplying adequate amounts of oxygen to the blood and efficiently removing from it

carbon dioxide.

Figure 1 – Schematic representation of an ECMO system. [3]

3

1.2 Blood Oxygenators

Throughout history three main types of blood oxygenators have been developed: oxygenators

where the gas exchange was provided by direct contact with a blood film (e.g. disc oxygenators);

oxygenators where oxygen was bubbled through blood (e.g. bubble oxygenators); and indirect contact

oxygenators, where gas exchange occurs across a membrane that separates the blood from the gas

phase (e.g. sheet and hollow-fiber membrane blood oxygenators). [4]

The first surgery involving blood oxygenation dates back to 1953 and consisted in the direct

oxygenation of the patient's blood in a small countercurrent filled tower. This equipment was adopted

until the mid-1980s when disc oxygenators were developed, where gas exchange was performed by

keeping a blood film in contact with an oxygen-rich atmosphere. Later, bubble oxygenators were

developed, in which oxygen was bubbled into the blood. Direct blood contact procedures required a

considerable amount of blood at start-up, and partially damaged it during the operation. Due to these

facts, membrane blood oxygenators (MBOs) were developed. This type of oxygenator required a much

smaller blood volume and caused much less damage, being the only type of oxygenators used

nowadays. [2]

1.3 Membrane Blood Oxygenators

Early MBOs were composed of silicon rubber and Teflon membranes. Nowadays, many types

of material are used with polyolefin polymers like TPX poly(4methyl pentene-1) being one of the most

widespread [5]. Companies such as MembranaGmbH developed asymmetric hydrophobic hollow fiber

membranes of TPX and or mixed with other types of polyolefins such as polyurethanes. This type of

membranes are prepared by the TIPS (thermally induced phase separation) method [2].

There are three main groups of MBOs commercially available:

i) Plaque oxygenators, equipped with Z-folded expanded polypropylene microporous

membranes. Examples of this type of oxygenator are Cobe Excel, Cobe VPCML,

Shirley M-2000;

ii) Spiral oxygenators, where the silicon membrane is wound around a central axis. A

widely used oxygenator of this type is produced by Avecor (Figure 2 (a));

iii) Hollow fiber oxygenators, manufactured with microporous polypropylene membranes

arranged in bundles of hollow fibers where blood flows on the outer side of the fibers

while oxygen flows in the lumen (Figure 2 (b)). This is the most common type of

oxygenator used nowadays. [6]

Each blood oxygenator has an approximate cost between 500$ and 600$, being the annual

market worth about 500 million dollars. [7]

4

(a) (b)

Figure 2 — (a) A true silicone membrane (sheet) oxygenator (Avecor) [8]; (b) A Hollow fibers oxygenator (Xenios Medos hilite oxygenator).

1.4 Membrane Blood Oxygenator Specifications

The ideal MBO has to fulfil a two-fold goal: i) promote efficient gas exchange and ii) be blood

compatible or hemocompatible. It is important that the MBO efficiently oxygenate up to 5 L / min of

venous blood with 95-100% hemoglobin saturation, in periods between several minutes and several

hours; Simultaneously, it should remove the adequate amount of CO2 sufficient to prevent

hemoacidosis, but not in excess which can cause alkalosis. Another vital property of the ideal MBO is

to be blood compatible, i.e. to be blood friendly and cause no damage to the blood cells and blood

components avoiding hemolysis, protein denaturation, clotting, activation of the complement system,

etc. The ideal MBO should also have a reasonable priming volume (1-4L), be simple and safe to use,

easily cleaned and sterilized. Figure 3 shows the principle of the MBO: in order to be efficient, the MBO

should provide approximately 250 cm3 (STP) / min of oxygen to the blood and must remove from it

approximately 200 cm3 (STP) / min of carbon dioxide. Blood flows of 2 to 4 L/min, which is approximately

10 times the blood flow through a kidney dialyzer, are required because the solubility of these gases in

the blood is limited. In a MBO, the driving force for O2 is 15 times higher than for CO2. In the lung, the

driving force for O2 is about 13 times higher than for CO2, although this organ is 20 times more

permeable to CO2 than to O2. Thus the key aspect in the design of a membrane oxygenator is the

transport of CO2. [2]

The average membrane surface area of commercial oxygenators is approximately 2 m2.

Considering the referred volumetric fluxes and a feed pressure of 76 cmHg, the membrane should

exhibit CO2 and O2 permeances of approximately 0.22 × 10−5 cmSTP3 cm−2s−1cmHg−1 and 0.27 ×

10−5 cmSTP3 cm−2s−1cmHg−1, respectively. [9]

5

Figure 3 – Principle of the membrane blood oxygenator [2].

7

2. Literature Review on Polyurethane Membranes for MBOs

2.1 Polyurethane Membranes

In 1937, Otto Bayer and his co-workers of I.G.Farbenindustrie in Leverkusen, Germany, made

the historical discovery of a class of polymers known today as urethanes or polyurethanes.

Polyurethanes are a special group of heterochain polymers with the structural unit represented in Figure

4. [10]

Figure 4 – Polyurethane structural unit. [11]

Polyurethanes are formed by reacting a polyol (an alcohol with more than two reactive hydroxyl

groups per molecule) with a diisocyanate or a polymeric isocyanate in the presence of additives and

catalysts, as described in Figure 5. There is an enormous variety of diisocyanates and a wide range of

polyols that can be used to produce polyurethanes. Due to this fact, a broad spectrum of materials can

be produced to meet the needs for specific applications. Polyurethanes exist in several forms such as

flexible foams, rigid foams, chemical-resistant coatings, specialty adhesives and sealants, and

elastomers.[12]

Figure 5 – Polyurethanes’ synthesis reaction [11].

There are two main categories of polyurethanes: elastic polyurethanes, e.g., flexible foams,

elastomers, coatings, adhesives, fibers etc., and rigid polyurethanes, e.g., rigid polyurethane foams,

structural foams, wood substitutes, solid polyurethanes, etc. This common classification of elastic and

rigid polyurethanes is mainly based on the oligo-polyol structure that can vary between 300 and 10000

daltons and the number of hydroxyl groups/molecule of oligo-polyol (the oligo-polyol functionality) varies

generally in the range of 2-8 OH groups/mol. A polyol of low functionality, having around 2-3 hydroxyl

groups/mol, with a high molecular weight of 2000-10000 daltons, leads to an elastic polyurethane. A low

molecular weight oligo-polyol of 300-1000 daltons, with a high functionality of around 3-8 hydroxyl

groups/mol leads to a rigid crosslinked polyurethane. The urethane linkages (and urea linkages)

http://dii.americanchemistry.com/

8

generate the “hard segment” (HS) of a polyurethane elastomer because of the possibility of association

by hydrogen bonds. The high elasticity of the polyurethane elastomer is assured by the high mobility of

high molecular weight polyol chains, known as “soft segment” (SS). [11]

Figure 6 — Hard segment (HS) and soft segment (SS) components of polyurethane

elastomers.

Polyurethanes, having extensive structure and property diversity are one of the most bio- and

blood compatible material and have played a major role in the development of many medical devices.

Polyurethanes are characterized by durability, elasticity, elastomer-like character, fatigue resistance,

compliance, and acceptance or tolerance by the body [13]. Works by Boretos et. al. [14] and Marzec

et. al. [15] describe biomedical applications of polyurethanes.

2.2 Gas permeability of segmented polyurethane symmetric membranes

Schneider proved that gas permeation characteristics of polyurethanes are directly related to

the chemical composition of their backbones [16]. On the other hand, the chemical composition of

polyurethanes’ backbones is strongly dependent on the composition, type and molecular weights of the

SS, the degree of hydrogen bonding between groups, the HS/SS ratio, etc. Moreover, studies have

shown that there is no obvious correlation between these parameters and gas permeability.

Knight and Lyman [17] and Cao et al. [18] found that gas permeability of symmetric polyurethane

membranes increases with the decrease of the HS content and the increase of the SS molecular weight.

Knight and Lyman studied the effect of chemical structure and fabrication variables on the gas

diffusion properties of various poly(ether urethane) and poly(ether urethane urea) membranes it was

studied. It was found that the type of chain extender used affected gas permeability, as copolymers

containing polyethylene glycol (PEG) as a SS were less permeable than those containing polypropylene

glycol (PPG) segments. Also, the gas permeability of the poly(ether urethane urea) membranes

increased linearly with an increase in the PPG segment molecular weight, possibly explained by the

degree of phase separation and by the nature of chain packing with the different extenders (the better

the packing the lower the permeability).

9

Hsieh et al., studied the relationship between structure and gas permeation of polyurethane

membranes containing different types and molecular weights of polyols and different HS/SS distributions

[19]. It was found that the degree of crystallinity was the most reasonable variable to describe the

behavior of permeation of gas or vapor in symmetric polyurethane membranes. The permeability of gas

and vapor consistently increased by decreasing the degree of crystallinity in polyurethane membranes

of all the various compositions.

In conclusion, the type, length and content of the SS as well as the crystallinity and hydrogen

bonding between HS‘s and SS‘s, influence the gas permeability properties of the polyurethane

membranes. Until the present moment a direct relationship between these properties and the gas

permeability has not been established, especially when comparing membranes of different components

and compositions.

2.3 Gas permeability of bi-soft segment polyurethane membranes

Bi-soft segment polyurethanes contain not one but two different types of SSs and are formed

when the second SS is used to extend an isocyanate terminated prepolymer containing the first SS.

Figure 7 shows the synthesis of a bi-soft segmented polyurethane from the reaction between the

polyurethane prepolymer and a chain extender, which contains the second SS. In bi-soft segment

polyurethanes, the two SSs can also show different extents of phase separation, apart from the

possibility of occurrence of different extents of phase separation between SSs and HSs. The introduction

of a second SS further increases the options for tuning membrane properties.

Figure 7 – Reaction of the prepolymer with a chain extender (diol) to obtain a bi-soft segmented polyurethane [11].

Bi-soft segment poly urethane/urea membranes have been thoroughly studied by Professor de

Pinho’s research group [20], [21], [22]. In particular, poly(urethane urea) membranes containing

polypropylene and polybutadiene as SSs, prepared from a polypropylene oxide-based (PPO)

prepolymer (PUR) and polybutadiene diol (PBDO) were studied [20]. Results showed that the ratio of

PBDO and PUR in the membranes changes the degree of cross-liking and the degree of phase

separation, and thus influences the gas permeability. It was found that the CO2 permeability of the

membranes is dependent of the ratio of PU/PBDO. A 67 wt.% PBDO membrane has a lower degree of

cross-liking and displays a linear behavior of CO2 permeability vs pressure, ranging from 90 to 550

10

Barrer. A 20 wt.% PBDO membrane has a higher degree of cross-linking and displays phase separation,

leading to higher CO2 permeability, ranging from 150 to 950 Barrer.

The same group also synthesized poly(urethane urea) membranes containing PPO and

polydimethylsiloxane (PDMS), by extending a PUR prepolymer with PDMS. For the PUR/PDMS

membranes, it was concluded that the two SS’s were phase separated and the HS’s formed small

aggregates somewhere in these two phases. Gas permeation experiments led to the conclusions that

the CO2, O2 and N2 permeabilities increased with the increase of PDMS membrane content, while the

permeability ratios, P(CO2)/P(N2) and P(O2)/P(N2) did not significantly change. For all the membranes,

the permeabilities decrease in the order P(CO2)>P(O2)>P(N2). Also, the lowest degree of cross-linking

and the lowest contribution of hydrogen bonding between HSs, was seen in the membrane with highest

content of PDMS. [21]

A third type of poly(urethane urea) membranes containing PPO and polycaprolactone (PCL), by

extending a PUR prepolymer with PCL diol, have also been synthesized and characterized by the de

Pinho’s group. Studies of the chemical structure and phase segregation properties of homogeneous

dense bi-soft segment PUR/PCL membranes, were performed [22] and it was found that the urethane

groups form HS aggregates dispersed in the soft segment phase and that this aggregation increases

with the increase of the PCL content. Gas permeation experiments led to the conclusion that

permeabilities of CO2 increased from 188 Barrer to 337 Barrer when the PCL content increased from 0

wt.% to 10 wt.% and was lowest, 113 Barrer, for the membrane containing 15 wt.% of PCL. O2

permeability through the membranes was independent of PCL content and was between 10 and 11

Barrer. The membrane with the highest CO2 permeability contains 10 wt.% of poly(caprolactone) and

was characterized by the highest contribution of hydrogen bonding between urethane and urea hard

segments.

Studies by the de Pinho’s research group show that the gas permeation properties of bi-soft

segment poly(urethane urea) membranes were influenced by the type and content of the second soft

segment. The CO2 permeability increases with the type of second SS in the order PCL, PBDO, and

PDMS; For PUR/PBDO membranes CO2 permeability increased with the decrease of PBDO; For

PUR/PDMS membranes CO2 permeability increased with the increase of PDMS; For PUR/PCL

membranes CO2 permeability was very low for membranes containing 0-15% of PCL.

2.4 Hemocompatibility of bi-soft polyurethane membranes

The blood compatibility properties of the PUR/PBDO and PUR/PDMS bi-soft segment

membranes were studied by de Queiroz et. al. [21] and Zhou [23]. Results showed that the

hemocompatibility of these types of membranes was limited particularly in terms of thrombogenicity for

the PUR/PBDO membranes, and of platelet adhesion for the PUR/PDMS membranes.

The hemocompatibility of bi-soft segment PUR/PCL membranes with PCL-diol content ranging

from 0% to 15% (w/w), was also studied by the de Pinho’s research group ([24], [25]) and it was

concluded that these presented enhanced hemocompatibility properties when compared to the

11

PUR/PBDO and PUR/PDMS membranes. Characterization studies showed that a variation of the

PUR/PCL weight ratio effects the surface energy, morphology, topography and hemocompatibility. The

PUR/PCL membranes were found to be nonhemolytic, with hemolysis index values of 1 and 2.1. The

in-vitro thrombosis degree was between 27% and 42% for a blood contact time of 15 minutes [26].

Phase segregation and chemical composition studies showed that a higher aggregation between

HS’s and mixture between the two types of SS’s led to less platelet adhesion with the membranes

containing 5 and 15% weight of PCL showing a very low number of adhered platelets in early stages of

activation. It was also concluded that enhancement of membrane hemocompatibility is achieved through

the control of the surface morphology. It was shown that the top dense surfaces of asymmetric poly(ester

urethane urea) membranes can be tailored with different morphologies when the ratio of the two SSs

PPO/PCL varies. The asymmetric membrane with the greatest amount of PCL displayed minimal

platelet deposition and inhibition of extreme stages of platelet activation. [22] [25]

2.5 Integral asymmetric poly(ester urethane urea) membranes

Despite the very promising results in terms of hemocompatibility, the gas permeability of the

dense homogeneous PUR/PCL membranes was very low. One method to increase the gas

permeabilities and simultaneously preserve the enhanced hemocompatibility of the PUR/PCL

membranes is to synthesize them as integral asymmetric PUR/PCL membranes [9]. These membranes

are composed of a very thin top dense layer, where all of the resistance to the gas transport phenomena

is present, and by a bottom thicker porous support layer which offers little or no resistance to gas

transport.

De Pinho e Faria et al. synthetized bi-soft segment poly(urethane urea) by a modified version

of the phase inversion technique where a polymerization reaction was introduced in the casting solution

preparation step. The reaction between the PUR prepolymer and PCL-diol, took place in a solvent

system of dimethyl formamide (DMF) and diethyl ether (DEE). The weight ratio of DMF/DEE was varied

over 11, 5 and 3. The evaporation times during the membranes casting were 30, 60 and 90 seconds.

The feed pressure for the gas permeation test with CO2 and O2 varied between 76 cmHg and 380 cmHg.

It was concluded that increasing the amount of DEE in the casting solution results in a decreasing of the

dense layer thickness. The membranes with the higher amount of DEE and lower evaporation times

were the ones that showed higher permeabilities. The CO2 permeation fluxes of these membranes were

of the same order of magnitude of commercial oxygenators demand, although, when tested with O2,

these membranes showed permeation fluxes below the demand for commercial oxygenators. The

synthesis of asymmetric poly(ester urethane urea) membranes with thickness of dense layer minimized,

enhanced the permeation to O2 and CO2. This fact associated with the introduction of PCL-diol,

responsible for a higher hemocompatibility, are evidences of possible application of these membranes

in blood oxygenators. [9] [27]

13

3. Framework and Thesis Objectives

The literature review reveals that the versatility of polyurethane synthesis is a strong asset on the

design of membrane material properties. In the present work, this versatility is further extended to

membrane gas permeation properties by the variation of the casting parameters involved in the

synthesis of integral asymmetric membranes of bi-soft segmented polyurethanes.

The main objectives of this work are:

1- Build and validate a novel experimental set-up capable of measuring the evolution of pressure

online, in the milibar range, at constant temperature, for extended periods of time;

2- Synthesis of integral asymmetric and nonporous symmetric poly(ester urethane urea)

membranes and characterization of the membranes surface and cross-section structures by

Scanning Electron Microscopy;

3- Determination of the permeation properties of the poly(ester urethane urea) membranes

towards CO2, O2 and N2.

15

4. Mass Transport Phenomena in Homogeneous Membranes

The mass transport is non-porous homogeneous polymer membranes is assumed to occur by

a solution-diffusion-desorption mechanism. The steady state diffusive flux in the y-direction, JAy, is

described by the Fick's First Law (Eq. (1)):

JAy = −DAm

dcAm

dy

(1)

where JAy is the flux of species A in terms of moles per unit of time and unit of membrane surface area.

The gradient of concentration is dcA

dy, with cAm being expressed as the molar concentration of solute A in

the polymer. The quantity DAm is the diffusion coefficient and can be regarded as a proportionality

between the flux and the concentration gradient . [28]

The integration of the First Fick’s Law over the total membrane thickness, δ, with the boundary

conditions: i) On the feed side, y = 0, the penetrant concentration in the polymer is cAm0; ii) On the

permeate side, y = δ, the penetrant concentration is cAmδ, results in Eq.(2).

JA =

DAm

δ(𝑐Am0 − 𝑐Amδ)

(2)

For ideal systems, where the solubility is independent of concentration, the concentration inside

the polymer is proportional to the applied pressure. This behavior is normally observed with gases in

elastomers. Since the solubility of the gases in elastomeric polymers is very low, Henry's law can be

applied. The equilibrium at the membrane/gas interfaces is therefore described by the relationship of

the concentration inside the polymer with the external pressure, given by the sorption coefficient, SA,

(Eq.(3)).

SA =cAm0

pf

=cAmδ

pp

(3)

In the considered system, cAm0 and cAmδ are not known, unlike pf and pp that are known.

Applying Henry’s Law (Eq. (3)), to Eq.(2), and considering the pressure pf on the feed side and pp on

the permeate side, it is obtained Eq. (4):

JA =

SADA

δ(pf − pp)

(4)

Considering the Solution-diffusion model, Permeability (P𝐴) is a function of Diffusivity (D𝐴) and

Solubility (S𝐴):

16

P𝐴 = D𝐴. S𝐴 (5)

Considering Eq. (5), Eq. (4) turns in Eq. (6):

Eq. (6) shows that the flux of a component through a membrane is proportional to the pressure

difference across the membrane and inversely proportional to the membrane thickness. [29]

Usually, Permeability is represented in Barrer units:

Barrer = 10−10 (cm3cm

cm2 s cmHg)

In transient state, the mass balance of penetrant gas A through a nonporous symmetric

membrane is given by Eq. (7).

∂JAm

∂y=

∂cAm

∂t (7)

Substituting the flux by the Fick’s first law, it is obtained the Fick’s second Law, expressed by

Eq. (8).

∂cAm

∂t= DA

∂2cAm

∂y2 (8)

The initial and boundary conditions of a system where the concentration of gas in one side of

an initially gas-free membrane is raised to cAm0, at time t = 0, while the other side is maintained at zero,

can be expressed by the Eq. (9b), Eq. (9b) and Eq. (9c):

cAm(y, 0) = 0 (9a)

𝑐Am(0, t) = cAm0 (9b)

cAm(δ, t) = cAmδ ≅ 0 (9c)

Rogers et al. [30] suggests a solution for Fick’s second Law admitting that the relation between

the concentration of gas at the interface of the polymer and the external gas pressure is given by Eq.

(10):

KA =nAm0

pf

(10)

where nAm0 is the quantity of gas, as the pressure-volume product, dividing by the volume of polymer.

Despite the sorption coefficient, KA, be defined without units, can be understood as equivalent units of

(pressure-volume of dissolved gas/volume of polymer)/external pressure.

JA =

PA

δ(pf − pp) (6)

17

Taking into consideration the units of the sorption coefficient, it is possible to obtain the

permeability coefficient, kA, by Eq. (11).

kA = DA𝐾𝐴 (11)

Thus, considering the relation of Eq. (10), and the Solution-diffusion model (Eq. (11)), the Fick’s

second Law can be solved by a commonly used method of assuming a solution in the form of Fourier

series assigning coefficients to meet the boundary and initial requirements (Eq. (12)). [30]

dpp

dt=

AD𝐴𝐾𝐴

Vδpf [1 + ∑ 2cos (πm) exp (−

D𝐴m2π2t

δ2)

∞

m=1

] (12)

Where V is the receiving chamber volume in which the diffused gas accumulates, 𝐴 is the

membrane surface area and dpp

dt is the variation of the permeate pressure with time.

For sufficient long time, series of Eq.(12) can be neglected and dpp

dt attains a constant value of:

dpp

dt=

AD𝐴𝐾𝐴

Vδpf

(13)

Considering the permeability coefficient, introduced in Eq. (11), and rearranging Eq. (13), one

can obtain the kA permeability value, given by Eq.(14):

kA =

Vδ

A

1

pf

dpp

dt

(14)

Integrating the Eq. (13) from time 0 to time t, one obtains for the pressure p at time t (Eq.(15)):

pp =

AD𝐴𝐾𝐴

Vδpf [t −

δ2

6D𝐴

+2

π2D𝐴

∑(−1)m

m2exp (−

D𝐴m2π2t

δ2)

∞

m=1

] (15)

When t → ∞ , exponential terms are elapsed, becoming negligibly small and pressure becomes

linear with time (Eq. (16)).

pp = (

AD𝐴𝐾𝐴

Vδ) pf (t −

δ2

6D𝐴

) (16)

The plot of permeate pressure versus time, in conjunction with the plot of the asymptotic linear

permeation flow in steady state, results in an intercept on the time axis, denoted by tc (characteristic

time), and an intercept in the pressure axis, denoted by pc (characteristic pressure), represented in

Figure 8.

18

Figure 8 - Representative pressure vs time graph with the intercept of the linear section with the time axis, tc, and the pressure axis, pc.

The intercept on the time axis, tc, can be obtained by solving the Eq. (16) for 𝑡 when pp = 0,

resulting in Eq. (17), allowing the calculation of the diffusion coefficient, 𝐷𝐴.

tc =δ2

6𝐷𝐴

(17)

Analogously, the intercept on the pressure axis, pc affords a simple calculation of the solubility,

𝐾𝐴, from Eq. (18). [30]

𝐾𝐴 = −6V

Aδ

pc

pf

(18)

Determined DA and 𝐾𝐴, the permeability coefficient, 𝑘𝐴, can be obtained, according to the

Solution-diffusion model, by Eq.(11).

Eq. (17) and Eq. (18), allow to determine DA and 𝐾𝐴, by recording the pressure of gas A into V

as a function of time until a straight line is determined, and extrapolating this line graphically to locate

the characteristic time tc and the characteristic pressure pc. These are the equations used for the late

approximation or time-lag method.

Rogers et al. [30] suggested an alternative method, known as the early approximation, specially

used for permeation experiments where the time required to establish the linear relation between the

pressure and the time is very long. It is based on a treatment of the exact solution in Fourier series (Eq.

(12)) in which a transformation formula is applied to the right hand leading to another solution to Fick’s

second law (Eq. (19)).

dpp

dt=

2A

V𝐾𝐴pf√

DA

πt∑ e

[−(δ2

4DAt)(2m+1)2]

∞

m=0

(19)

-10

0

10

20

30

40

50

0 100 200 300 400 500 600 700

pp(m

bar

)

t(s)

𝑝𝑐

t𝑐

19

Because of the inverted placement of t in the exponentials, this series converges most rapidly

for very small values of t rather than for large values. For times sufficiently short, Eq. (19) may be

approximated by neglecting all but the leading term in the series of exponentials. Then it is convenient

to multiply by √t, and take the logarithms on both sides. Thus, Eq. (20) is obtained.

Plotting ln (√tdpp

dt) against

1

t, represented in Figure 9, a straight line is determined.

Figure 9 – Representative ln (√tdp

dt) vs

1

t plot for the early approximation.

From the slope of this line, DA can be obtained according to the relation expressed by Eq. (21).

slope = −

δ2

4DA

(21)

After determined DA, the solubility can be obtained by solving Eq. (22), for KA.

𝐾𝐴 = √π

DA

V

2A

1

pf

(√tdp

dt) exp (

δ2

4DAt) (22)

Then, the permeability coefficient, 𝑘𝐴, can be obtained according to the Solution-diffusion

model, by Eq.(11).

-6,15

-6,10

-6,05

-6,00

-5,95

-5,90

-5,85

-5,80

-5,75

-5,70

-5,65

0,05 0,10 0,15 0,20

Ln(d

pp

/dt.

t^0

.5)

1/t (s^-1)

ln (√t

dpp

dt) = ln (

2A

V𝐾𝐴pf

2√DA

π) −

δ2

4DAt (20)

21

5. Experimental

5.1 Materials

5.1.1 Commercial Membrane

Figure 10 shows a photograph of a MBO marketed by Avecor/Medtronics, model 0600.

Membrane samples taken from this MBO were tested in the experimental gas permeation set-up built

in this work. Figure 11 shows Scanning electron microscopy (SEM) images of top surface, cross-section

and the bottom support surface micrographs of the silicon rubber symmetric nonporous membrane

supported by a woven structure which acts as a spacer between successive folds of the spiral wound

membrane module. The mean thickness of the silicone membrane measured from SEM images of the

cross section using the ImageJ software, is equal to 66.1±1.2 μm (excluding the supporting web).

Figure 10 – The 0600 MBO (left) and ECMOtherm II (right) commercialized by Avecor/Medtronics [3]

Top surface Cross-section Bottom support

Figure 11 – SEM images of upper surface, cross-section and the bottom support surface of membrane Avecor/Medtronics oxygenator model 0600. [31]

22

5.1.2 Materials for the synthesis of poly(ester urethane urea) membranes

Integral asymmetric and nonporous symmetric poly(ester urethane urea) membranes containing

one and two types of soft segments were synthesized. For the preparation of the poly(ester urethane

urea) membranes two types of prepolymers were used: polypropylene oxide (PPO)-based polyurethane

prepolymer with three isocyanate end groups (PUR), and the chain extending prepolymer

polycaprolactone-diol (PCL-diol). The chemical structures of the two prepolymers are shown in Figure

12.

Figure 12 – Prepolymers’ molecular structure. [24]

The PUR prepolymer, supplied by Fabrires-Produtos Químicos, SA, has a molecular weight of

3500 Da, and the prepolymer PCL-diol, provided by Sigma-Aldrich, has a molecular weight of about 530

Da. The solvents used in membranes’ synthesis were dimethylformamide (DMF) (w / w% grade, 99.8%)

and diethyl ether (DEE) (w / w% grade, 99.7%) provided by Panreac. Stannous octoate (C16H30O4Sn)

(wt.%, 95%) also of the Aldrich brand was used as the catalyst.

In the absence of the PCL-diol prepolymer membranes were synthesized with a single type of

soft segment (SS) and a type of hard segment (HS). Poly(ester urethane urea) membranes with two

SSs were synthesized from the two prepolymers, PUR and PCL-diol. The PUR prepolymer contains the

first SS, PPO, while PCL confers the second SS. In terms of HSs, two types can be formed: - Type I:

where the HSs were originated by the reaction of two PUR segments, consisting of two urethane groups

linked by two toluene groups and one urea group. - Type II: where the HS was originated by the reaction

of a PUR segment with a PCL segment, consisting of two urethane groups linked by a toluene group.

This type II HS is shorter than type I.

Figure 13 shows the molecular structure of poly(urethane urea) synthesized only with PUR and

containing only one SS (PPO) and HS type I and Figure 14 shows the molecular structure of poly(ester

urethane urea) containing two types of SSs (PPO and PCL) and type I and type II HSs, synthesized

from the PUR and PCL prepolymers.

23

Figure 13 — Molecular structure of poly(urethane urea) prepared by PUR prepolymer and composed of only one SS (PPO) and HS type I. [22]

Figure 14 – Molecular structure of poly(ester urethane urea) prepared with PUR and PCL-diol

prepolymers containing two types of SSs (PPO and PCL) and type I and type II HSs. [22]

5.1.3 Gases

Permeation tests were performed using nitrogen (purity ≥ 99.999%), carbon dioxide (purity ≥

99.98%) and industrial oxygen (purity ≥ 99.5%) provided by Air Liquide. All gases were used as received.

24

5.2 Synthesis of integral asymmetric poly(ester urethane urea)

membranes

5.2.1 Variation of total polymer to total solvent weight ratio

Integral asymmetric poly(ester urethane urea) membranes, designated PEU membranes, were

synthesized by a modified version of the phase inversion technique using different total polymer to total

solvent weight ratios. In the first step, a casting solution was prepared, where the prepolymers were

reacted in a solvent system of DMF and DEE (DMF / DEE = 3/1 by weight), and a stannous octoate

catalyst (Tin(II) 2-ethylheaxanoate) in the quantity of 6 drops. Upon this preparation, the casting solution

was left in agitation for about 2 hours. In a second step, the casting solution was spread on a glass plate

using a 250 μm casting knife. After a solvent evaporation time of 30 seconds, the glass plate was

subsequently introduced into the coagulation bath (distilled water) where it was left for about 12 hours.

When removed from the bath, the membranes were dried in an oven at 35 °C for at least 36 hours.

Polyurethane-urea membranes were synthesized starting from PUR and PCL prepolymers, and

the mass ratio between total polymer and total solvent of 2/3, designated PEU2 and PEU3, wherein the

mass ratios between PUR and PCL were 95/5 and 90/10, respectively. Poly(ester urethane urea)

membrane designated by PEU1, PEU4 and PEU5, was synthesized starting from only a PUR

prepolymer, and the mass ratio of total polymer to total solvent was 2/3, 1/1 and 3/2, respectively. Table

1 shows the chemical composition, solvent time evaporation and total polymer to total solvent weight

ratio of the PEU membranes.

Table 1 – Chemical composition of the PEU membranes prepared with varying contents of PCL

diol, total polymer to total solvent weight ratio and the types of HSs and SSs that they contain.

PEU

membrane

PUR/PCL

diol (wt.%)

PUR/PCL-

diol molar

ratio

HS Type SS Type

Solvent

evaporation

time (sec)

total

polymer/total

solvent (wt.

%)

PEU1 100/0 1/0.00 I PPO 30 2/3

PEU2 95/5 1/0.35 I and II PPO & PCL 30 2/3

PEU3 90/10 1/0.73 I and II PPO & PCL 30 2/3

PEU4 100/0 1/0.00 I PPO 30 1/1

PEU5 100/0 1/0.00 I PPO 30 3/2

25

5.2.2 Variation of solvent evaporation time

Integral asymmetric poly(ester urethane urea) membranes, were synthesized by a modified

version of the phase inversion technique varying the solvent evaporation time. The membranes were

named by PEU-X-Y, where X refers to the solvent evaporation time, and Y refers to the PUR wt.%.

In a first step, a casting solution was prepared, where the prepolymer PUR were reacted in a

solvent system of DMF and DEE and the Tin-Octoate catalyst. Upon this preparation, the casting

solution was left in agitation for about 2 hours. In a second step, the casting solution was spread on a

glass plate using a 250 μm casting knife.

The evaporation time of PEU membranes were 1, 5, 10 and 15 minutes, for the membranes

PEU-1-100, PEU-5-100, PEU-10-100, PEU-15-100, respectively. After this, the glass plate was

introduced into the coagulation bath (distilled water) where it was left for about 12 hours. When removed

from the bath, the membranes were dried in an oven (35 °C) for at least 36 hours. Table 2 shows the

chemical composition, solvent time evaporation and total polymer to total solvent weight ratio of the PEU

membranes.

Table 2 – Chemical composition of PEU membranes prepared varying solvent evaporation

times.

PEU

membrane

PUR/PCL-diol

(wt.%)

PUR/PCL-

diol molar

ratio

HS

Type

SS

Type

Solvent

evaporation

time (min)

total

polymer/total

solvent ratio (wt.

%)

PEU-1-100 100/0 1/0.00 I PPO 1 1/1

PEU-5-100 100/0 1/0.00 I PPO 5 1/1

PEU-10-100 100/0 1/0.00 I PPO 10 1/1

PEU-15-100 100/0 1/0.00 I PPO 15 1/1

5.2.3 Variation of PUR/PCL weight ratio maintaining solvent evaporation time

Integral asymmetric poly(ester urethane urea) membranes, were synthesized by a modified

version of the phase inversion technique with solvent evaporation time of 1 or 5 minutes, varying the

PCL content. The membranes were named by PEU-X-Y, where X refers to the solvent evaporation time,

and Y refers to the PUR wt.%.

In a first step, a casting solution was prepared, where the prepolymers were reacted in a solvent

system of DMF and DEE and a Tin-Octoate catalyst. Upon this preparation, the casting solution was left

in agitation for about 2 hours. In a second step, the casting solution was spread on a glass plate using

a 250 μm casting knife. Thereafter, the evaporation time of 30 seconds, the glass plate was

subsequently introduced into the coagulation bath (distilled water) where it was left for about 12 hours.

When removed from the bath, the membranes were dried in an oven (35 °C) for at least 36 hours. The

PEU membranes were synthesized starting from two PUR and PCL prepolymers, where the time of

26

evaporation was 1 minute, designated by PEU-1-100, PEU-1-95, PEU-1-90, PEU-1-85, wherein the

mass ratios between PUR and PCL were 100/0, 95/5, 90/10 and 85/15, respectively. The PEU

membranes were synthesized starting from two PUR and PCL prepolymers, where the time of

evaporation was 5 minutes, designated by PEU-5-100, PEU-5-95, PEU-5-90, PEU-5-85, wherein the

mass ratios between PUR and PCL were 100/0, 95/5, 90/10 and 85/15, respectively. Table 3 shows the

chemical composition, solvent time evaporation and total polymer to total solvent weight ratio of the PEU

membranes.

Table 3 – Chemical composition of PEU membranes prepared varying contents of PCL diol, solvent

evaporation times and the types of hard and soft segments that they contain.

PEU

membrane

PUR/PCL

diol (wt.%)

PUR/PCL-

diol molar

ratio

HS

Type SS Type

Solvent

evaporation

time (min)

total

polymer/total

solvent ratio

(wt. %)

PEU-1-100 100/0 1/0.00 I PPO 1 1/1

PEU-1-95 95/5 1/0.35 I and II PPO & PCL 1 1/1



PEU-5-100 100/0 1/0.00 I PPO 5 1/1




5.3 Synthesis of nonporous symmetric poly(ester urethane urea)

membranes

The nonporous symmetric poly(ester urethane urea) membranes, designated PU membranes,

were synthesized by the solvent evaporation method, varying the PCL content. In a first step, a casting

solution was prepared, where the PUR and PCL prepolymers were reacted in a solvent system of DMF

and DEE and the Tin-Octoate catalyst. Upon this preparation, the casting solution was left in agitation

for about 2 hours, after which the casting solution was spread on a glass plate using a 250 μm casting

knife and the films were left to dry for about 24 hours. After completely dried and with all the solvent

evaporated, the membranes were detached from the glass plate and thoroughly washed in distilled

water and dried an oven, at 35 °C, for 3 hours. The PU membranes, designated PU100, PU90 and

PU85, wherein the mass ratios between PUR and PCL were 100/0, 95/5 and 85/15, respectively. Table

4 shows the chemical composition and total polymer to total solvent weight ratio of the PU membranes.

27

Table 4 – Chemical composition of PU membranes prepared varying contents of PCL diol and the types of hard and soft segments that they contain.

PU

membrane

PUR/PCL

diol

(wt.%)

PUR/PCL-

diol molar

ratio

HS Type SS Type

total

polymer/total

solvent ratio

(wt. %)

PU100 100/0 1/0.00 I PPO 1/1

PU95 95/5 1/0.35 I and II PPO & PCL 1/1

PU85 85/5 1/1.20 I and II PPO & PCL 1/1

5.4 Characterization of the membranes structure by Scanning Electron

Microscopy

Samples of the PEU and PU membranes were observed by Scanning Electron Microscopy

(SEM) (JM-7001F FEG-SEM, JOEAL, Tokyo, Japan). The samples were obtained by fracturing pieces

of membranes in liquid nitrogen, which were then mounted on a stub and sputter-coated with gold. For

each sample pictures of the top surface, cross-sections and bottom surface were taken. The average

total thickness of membranes was determined from five measurements on different sections of SEM

images of the cross section using the software ImageJ version 1.4.3.67 (NIH ImageJ, USA) [32].

5.5 Gas Permeation experiments - Volumetric Method

Initial gas permeation flux measurements were realized by the volumetric method for samples

of the CM membrane. This method consisted in the direct measurement of the volume of gas that passed

through the membrane as a function of time. The experiments occurred in a permeation cell, shown in

Figure 15, with an effective surface area of 9.62 cm2. The permeation cell consists of two stainless steel

plates between which a membrane sample was placed to be tested.

Figure 15 – Schematic representation of the permeation cell.

28

Figure 16 shows a scheme of the experimental set-up used in permeation flux measurements

by the volumetric method. Permeation tests were performed in one unit where a gas feed line and a

purge were attached to the bottom permeation cell plate, while the porous steel top plate was connected

to a serological pipette filled with water, in which the bubbles of the gas that reached the permeate were

identifiable. A pressure sensor (Setra, Model 205-2, range 0-3000 PSIA, Massachusetts, USA [33]),

connected to a Setra Datum 2000TM dual channel meter display unit [34], was mounted on the feed side

before the cell, to guarantee a precise measurement of pressure values of the feed gas, regulated by a

pressure reducing valve (PRV) assembled to the gas cylinder. The tubbing system of the set-up

consisted mainly of stainless steel 316 tube with 1/8 inch internal diameter, provided by Hook®. Several

types of tube fitting, of different materials (stainless steel, titanium and brass) GYROLOK®, and needle

valves 3700 Series, also provided by Hook®, were used. [35]

To perform a measurement by the volumetric method, the gas is fed, by regulating the PRV in

the gas cylinder, to a membrane sample placed in the permeation cell. Simultaneously, the operator

starts a chronometer and record the evolution of the gas bubble in the serological pipette in function of

time. This method requires the constant presence of an operator. For measurements of very little

permeable membranes, it can take long time to the operator to be able to distinguish a variation of

volume in the pipette, while for very permeable membranes the measurement is highly dependent of

the operator criteria

Figure 16 – Scheme of the experimental set-up used in permeation flux measurements by the volumetric method.

5.6 Gas Permeation experiments - Pressure Method

An in house built set-up was used for gas permeation measurements by the pressure method

at constant temperature. A temperature controlled unit was built and calibrated to perform gas

29

permeation tests, where pressure variation is recorded online, at very short intervals of time with a very

high precision. Figure 17 represents a diagram of the set-up where the permeation flux measurements

of the CM, PEU and PU membranes were carried out. The unit was composed of a feed pressure sensor

(PfT) (Setra, Model 205, Massachusetts, USA [36]), a permeation cell like the one described in Figure

15, a cylindrical buffer of 12.6 ± 0.1 cm3 and a pressure transmitter (PpT), (Intelligent Transmitter

Paroscientific, Series 6000, model 6100A-CE Inc. Washington, USA), attached to a Paroscientific model

710 display unit, connected to a computer. The pressure values were recorded in the software Digiquartz

Assistant® version 1.0 (Paroscientific Inc, Washington, USA). The receiving chamber volume was 27.7

± 0.1 cm3.

The tubbing system of the set-up consisted mainly of stainless steel 316 tube with 1/8 inch

internal diameter, provided by Hook®. Several types of tube fitting, of different materials (stainless steel,

titanium and brass) GYROLOK®, and needle valves 3700 Series (V1, V2, V3 and V4) also provided by

Hook®, were used. [35]

Before the beginning of each measurement, a single gas (N2, CO2 or O2) was fed, by regulating

the pressure reducing valve (PRV) in the gas cylinder, to a membrane sample placed in the permeation

cell, while valves V1 and V3 were kept opened and valve V2 closed. The feed pressure was kept

constant and was measured in the pressure transmitter PfT. The permeation measurement begins when

the operator closes the valve V3 and the pressure in the receiving chamber, of known volume, starts

being measured as a function of time by the PpT sensor. After one measurement, valve V3 is opened

and the permeate pressure returns to atmospheric pressure.

To reduce the feed pressure, valve V2 and valve V3 must be opened. Valve V4 allows to set

the volume of the receiving chamber (27.7 cm3 or 15.1 cm3).

Figure 17 – Scheme of the set-up used in permeation flux measurements by constant volume method.

To ensure that pressure measurements were not affected by temperature variations, the set-up

was installed inside an air bath where the temperature was controlled by regulating a heat source,

30

keeping constant a cold source. Figure 18 shows the interior of the air bath with set-up and apparatus.

Calibrations and tests performed in the set-up are described in the Annex.

The experimental set-up is capable of measuring online the evolution of pressure as function of

time, in intervals of 1.4 seconds, with milibar precision. It is possible to perform measurements with a

duration up to 90 minutes, at constant temperature, being the maximum variation of the air bath

temperature of 0.02 ºC.

Figure 18 – Photography of apparatus used for permeation measurements by barometric method inside air bath

31

6. Results and Discussion

6.1 Surface morphology and cross section structure analysis of integral

asymmetric PEU membranes by Scanning Electron Microscopy

6.1.1 PEU membranes prepared with different total polymer to total solvent weight ratio

The PEU1, PEU2 and PEU3 integral asymmetric membranes, containing 0, 5 and 10% wt. of

PCL, respectively, were synthesized with total polymer to total solvent weight ratio of 2/3, DMF to DEE

weight ratio of 3, and solvent evaporation time of 30 seconds.

Upon visual inspection, it was verified that the membranes did not present a cohesive and

continuous structure but instead net or web-like morphology. The PEU1 membrane showed the least

homogeneous structure of varying thickness and with visible holes; the PEU2 membrane also presented

a fragile net-like structure; and the PEU3 membrane presented a more homogeneous structure,

continuous, cohesive and with no visible holes. These results indicate that the increase of the amount

of PCL in the PEU membranes improves the structural characteristics. Furthermore, the structure of the

PEU membranes produced with the contents and quantities described in Table 1 of section 5.2.1,

suggests that the batch of the prepolymer PUR used in this study was manufactured with higher

amounts of solvent, than the one used in previous studies ([9], [22]), supplied by another producer.

Taking into account this assumption, other membrane casting solutions and conditions were tested in

order to obtain the optimal film forming conditions, such as: variation of total polymer/total solvent ratio

and variation of the solvent evaporation time.

The PEU1, PEU4 and PEU5 membranes, containing only PUR and no PCL, were synthesized

with polymer/solvent weight ratio of 2/3, 1/1 and 3/2, respectively, while maintaining the mass

proportions among DMF / DEE solvents equal to 3, and the solvent time evaporation of 30 seconds.

When the membranes were visually compared in terms of physical appearance, it was found that the

PEU1 membrane presented a web like incomplete film structure, with several holes and zones of limited

thickness; that the PEU5 membrane was slightly more rigid, malleable and sticky, which difficulted its

manipulation, than the PEU1 membrane. On the other hand, the PEU4 membrane was more

homogeneous and consistent, malleable, elastic and with a uniform, non-web like structure. It was

observed that the greater the total polymer to total solvent ratio, the easier it is to obtain a membrane

with a complete film. However, an excess of polymer may result in less elastic and very sticky

membranes. This indicates that to obtain membranes with optimal structural characteristics, there must

be a balance between the amount of polymer and the solvent system. It was determined that, for the

materials used, the most suitable total polymer to total solvent ratio is 1/1.

Figure 19 shows S.E.M. images of the top surface, bottom surface and cross-section of the

PEU3, PEU4 and PEU5 membranes. Comparing the SEM images of the top surface of the PEU3, PEU4

and PEU5 membranes ((a), (b) and (c)), it is observed that the PEU3 membrane presents an active

surface fully constituted of concavities and pores. Unlike the PEU3 membrane, it is observed in the

32

PEU4 membrane active surface a complete film structure, with pores of two different sizes spread along

the surface. Like the PEU3 membrane, it is observed that the PEU5 membrane top surface is fully

constituted of pores, apparently larger and deeper like the PEU3 membrane pores.

Relatively to the bottom surface SEM images ((d), (e), (f)), it is observed that PEU3 membrane

has a bottom surface full of pores and concavities, like its top surface. The PEU4 membrane shows a

bottom surface characterized by two pore regimes of two different sizes, spread along the film surface.

The PEU5 membrane bottom surface has pores, apparently in a lower number than its top surface,

spread along the surface.

Observing the SEM images of the cross sections ((g), (h) and (i)), the PEU3, PEU4 and PEU5

membranes appears to have a distinct cross section structure with no visible dense layer but instead

two porous regions: for the PEU3 and PEU5 membranes, close to the upper and bottom surface small

pores in large number are observed, while on the most inner part of the membrane a thicker phase with

larger pores and less numerous can be noted; The same is observed for the PEU4 membrane, except

the fact that near bottom surface are observed large concavities, correspondent to the large pores

observed in the bottom surface image (e). The cross-section of the PEU5 membrane has a torn aspect.

When the samples were prepared and cut for the SEM, regard its elastomeric and sticky nature, it was

not achieved a clean cut of the membrane pieces. Note that the cross-section structures look drained

by the knife, contrary to the PEU4 membrane cross-section. Due this, the two pores region became

harder to identify and to limit.

33

PEU3 PEU4 PEU5

Top surface

(a) (b) (c)

Bottom surface

(d) (e) (f)

Cross-section

(g) (h) (i)

Figure 19 – SEM images of samples of PEU3: (a) top, (b) cross-section, (c) bottom; PEU4: (d) top, (e) cross-section, (f) bottom; PEU5: (g) top, (h) cross-section, (i) bottom.

6.1.2 PEU membranes prepared with different solvent evaporation time

In order to study the effect of the solvent evaporation time on the membrane morphology, PEU

membranes, containing only PUR, were prepared with a polymer/solvent ratio of 1/1, a DMF/DEE weight

ratio equal to 3 and solvent evaporation times of 1, 5, 10 and 15 minutes to render PEU-1-100, PEU-5-

100, PEU-10-100 and PEU-15-100 membranes, respectively. It was observed that all of these

membranes presented structural sustainability, were homogeneous and elastic.

Figure 20 shows S.E.M. images of the top surface, bottom surface and cross-section of the

PEU-1-100, PEU-5-100 and PEU-10-100 membranes.

The top surface SEM images ((a), (b) and (c)) show that the porous definition decreases in the

order of the PEU1, PEU2 and PEU3 membranes.

Comparing the SEM images of the bottom surface ((d), (e) and (f)) for the PEU-1-100 membrane

is observed two types of pores, large concavities and smaller ones between them; for PEU-5-100 and

PEU-10-100 membrane it is observed that pores are fully spread along the surface, being smaller and

34

in large number in the PEU-5-100 membrane and larger and in small number for the PEU-10-100

membrane. Actually, the top and the bottom surfaces of the PEU-10-100 membrane are very similar.

The cross-section of the PEU-1-100, PEU-5-100 and PEU-10-100 membranes represented in

((g), (h) and (i)), appears to have a distinct cross section structure with no visible dense layer but instead

two porous regions: close to the upper and bottom surface small pores in large number are observed,

while in the most inner part of the membrane is observed a thicker phase with larger pores and less

numerous. Note that the upper layer seems to be the less porous region (upper denser layer),

resembling the dense layer of the integrally skinned membranes. It is observed that the upper denser

layer becomes more distinct in the order of the PEU-1-100, PEU-5-100 and PEU-10-100 membranes.

In the same order, it is verified an apparent increase of the membrane thickness. A possible explanation

is the increase of the solvent evaporation time promotes the formation of a dense layer in the upper

surface but has no effect on the pore regime of the inner part.

PEU-1-100 PEU-5-100 PEU-10-100

Top surface

(a) (b) (c)

Bottom surface

(d) (e) (f)

Cross-section

(g) (h) (i)

Figure 20 – SEM images of samples of PEU-1-100: (a) top, (b) cross-section, (c) bottom; PEU-5-100: (d) top, (e) cross-section, (f) bottom; PEU-10-100: (g) top, (h) cross-section, (i) bottom.

35

6.1.3 PEU membranes prepared with different PUR to PCL weight ratios

With the objective of studying the effect of the PCL content on the membrane morphology, the

PEU membranes were prepared with varying percentages of PCL and with constant polymer/solvent

weight ratio 1/1, constant DMF/DEE weight ratio equal to 3. PEU-1-100, PEU-1-95, PEU-1-90 and PEU-

1-85 membranes were synthesized with 0, 5, 10 and 15% of PCL, respectively, and solvent evaporation

time of 1 minute. The PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85 membranes were prepared with

0, 5, 10 and 15%, of PCL respectively, and solvent evaporation time of 5 minutes. It was observed that

every PEU membrane presented structural sustainability, were homogeneous and elastic. It was also

noted that membranes were whiter, less yellow and also stickier with the increase of PCL amount used

in its preparation.

Figure 21 shows SEM images of top surface, bottom surface and cross-section of the PEU-1-

95, PEU-1-90 and PEU-1-85 membranes prepared with solvent evaporation time of 1 minute and PCL

wt.% of 5, 10 and 15%, respectively. It can be observed that the number of pores and its diameter

decreases in the order of the PEU-1-95, PEU-1-90 and PEU-1-85 membrane, being almost completely

invisible in the PEU-1-85 membrane top surface.

Observing the SEM images of the bottom surface of the PEU-1-95, PEU-1-90 and PEU-1-85

membranes ((d), (e) and (f)) it is observed that all of the three have similar aspect, with pores in the

same number and of the same size.

In the SEM images of the cross sections ((g), (h) and (i)), one can see that the PEU-1-95, PEU-

1-90 and PEU-1-85 membranes appears to have a distinct cross section structure with no visible dense

layer but instead two porous regions: close to the upper and bottom surface small pores in large number

are observed, while in the most inner part of the membrane is observed a thicker phase with larger pores

and less numerous. Note that the upper layer seems to be the less porous region (upper denser layer),


layer becomes more distinct and the number and the size of the pores in the inner region decreases, in

the order of the PEU-1-95, PEU-1-90 and PEU-1-85 membrane.

The cross-section of the PEU-1-90 and PEU-1-85 membranes have a torn aspect. When the

samples were prepared and cut for the SEM, regard its elastomeric and sticky nature, it was not

achieved a clean cut of the pieces of the PEU-1-90 and PEU-1-85 membranes. Note that the cross-

section structures look drained by the knife, contrary to the PEU-1-95 cross-section. Due this, the two

pores region became harder to identify and to limit.

36


Figure 22 shows SEM images of top surface, bottom surface and cross-section of the PEU-5-

95, PEU-5-90 and PEU-5-85 membranes prepared with a solvent evaporation time of 5 minutes and

PCL wt.% of 5, 10 and 15%, respectively.

It can be observed that the number of pores and its diameter decreases in the order of the PEU-

5-95, PEU-5-90 and PEU-5-85 membrane, being almost completely invisible in the PEU-5-90 and PEU-

5-85 membranes top surface.

Observing the SEM images of the bottom surface of the PEU-5-95, PEU-5-90 and PEU-5-85

membranes ((d), (e) and (f)) it is observed that the PEU-5-95, PEU-5-90 membranes have similar

aspect, with pores in the same number and of the same size. The PEU-5-85 membrane shows a bottom

surface with a greater number of pores of the same size of the other membranes.

In the SEM images of the cross sections ((g), (h) and (i)), one can see that the PEU-5-95, PEU-

5-90 and PEU-5-85 membranes appears to have a distinct cross section structure with no visible dense

layer but instead two porous regions: close to the upper and bottom surface small pores in large number

are observed, while in the most inner part of the membrane is observed a thicker phase with larger pores

and less numerous. Note that the upper layer seems to be the less porous region (upper denser layer),


PEU-1-95 PEU-1-90 PEU-1-85

Top surface

(a) (b) (c)

Bottom surface

(d) (e) (f)

Cross-section

(g) (h) (i)

37

layer becomes more distinct, and the number of pores in the inner region decreases, in the order of the

PEU-5-95, PEU-5-90 and PEU-5-85 membrane.

PEU-5-95 PEU-5-90 PEU-5-85

Top surface

(a) (b) (c)

Bottom surface

(d) (e) (f)

Cross-section

(g) (h) (i)


Table 5 shows the average total membrane thickness obtained for the PEU membranes and

the respective standard deviation. The thickness of the nonporous symmetric PEU membranes was

obtained from 5 measurements using the ImageJ software, which allows to estimate lengths from SEM

images of the cross-section, comparing the image scale with a chosen length by the image pixels.

Comparing the measured values of thickness for the PEU membranes it is verified an increase

in order of the PEU3, PEU4 and PEU5 membrane, being an evidence that the increasing of the polymer

to solvent ratio results in thicker membranes. It is also verified an increase in the order of the PEU-1-

100, PEU-5-100 and PEU-10-100 membrane. One can admit that the increasing of the solvent

evaporation time results in the increasing of the total membrane thickness. Relatively to the other PEU

membranes, the total membrane thickness seems not to respect any specific tendency.

Because of the non existent of a well-defined dense layer in the PEU membranes, the

measurement of a possible upper denser layer may be very dependent of the criteria of thickness

38

measurements and have large uncertainty. Nevertheless, the upper denser layer thickness of the PEU

membranes seems to vary between 8 and 37 µm.

Table 5 – total membrane thickness obtained for the PEU membranes and the respective standard deviation.

Membrane δ(μm) σ(μm)

PEU3 131 4

PEU4 162 2

PEU5 174 2

PEU-1-100 88 3

PEU-5-100 114 2

PEU-10-100 203 7

PEU-1-95 163 3

PEU-1-90 130 2

PEU-1-85 123 1

PEU-5-95 142 6

PEU-5-90 112 2

PEU-5-85 121 2

6.2 Surface morphology and cross-section structure analysis of

nonporous symmetric PU membranes by SEM

The nonporous symmetric PU membranes prepared were translucent, glassy and very sticky.

Figure 23 shows the SEM images of the top surface and cross-section of the PU100 membrane. The

PU100 membrane is completely dense with no visible pores. The same was observed for the PU95 and

PU85 membranes.

Top surface Cross Section

(a) (b)

Figure 23 – SEM images of samples of PU100: (a) top surface, (b) cross-section.

Table 6 shows the average total membrane thickness obtained for the PU membranes and the

respective standard deviation of these measurements. The thickness of the nonporous symmetric

39

PU100 membranes was obtained from 5 measurements using the ImageJ software. The total thickness

of the PU95 and PU85 membranes was obtained from 5 measurements using a digital caliper.

The symmetric dense PU100, PU95 and PU85 membranes have total membrane thickness of

107 Barrer, 126 Barrer and 114 Barrer, respectively.

Table 6 – Membranes upper denser layer mean thickness (μm) and the membrane mean thickness (μm) and measurements’ standard deviations.

Membrane δ(μm) σ(μm)

PU100 107 0

PU95 126* 3

PU85 114* 16

*measured using a digital caliper.

6.3 Gas Permeation by the Volumetric Method

In the gas permeation experiments measured by the volumetric method (constant pressure

method), volume variations as function of time were taken in a serological pipette as was described in

section 5.5 Samples of the CM membrane were tested with N2.

Figure 24 shows an experiment performed with a sample of the CM membrane, where the N2

was fed at a pressure, pf, of 3.6 bar.

Figure 24 – Volume of N2 measured as a function of time for CM membrane sample by the constant pressure method. Pf = 3.6 bar.

Figure 25 shows the results obtained for different tests using the same membrane several times

with the same feed pressure (5.0 bar).

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0 20 40 60 80 100 120

V(c

m3 )

time (s)

40

Figure 25 – Volume of N2 measured as a function of time for three different experiments of the CM membrane by the constant pressure method at a feed pressure of = 5.0 bar.

Figure 26 shows the volume measurements as a function of time of the same sample of the CM

membrane at feed pressures of 3.6 bar and 5.0 bar.

Figure 26 – Volume of N2 measured as a function of time for the same sample of the CM membrane by the constant pressure method at feed pressures of 3.6 bar (orange dots) and 5.0 bar (blue dots).

The volumetric flux is obtained by dividing the slopes of the lines of volume vs time, dV

dt, (Figure

26) by the effective surface area, A. The factor of temperature correction, TSTP

T, must be considered in

order to obtain the volumetric flux for STP conditions which is defined by:

J𝑁2

=dV

A. dt TSTP

T [

10−5cm3

cm2s]

(23)

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0 10 20 30 40 50 60 70

V(c

m3)

time (s)

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0 20 40 60 80 100 120

V(mL)

t(s)

41

The volumetric flux was obtained for each of the volume vs time plots obtained for different feed

pressures. For example, the value of volumetric flux obtained for the plots represented in Figure 26 are

shown in Figure 27 as a function of the transmembrane pressure (TMP). The TMP is defined as the

difference between the feed pressure, pf, and the atmospheric pressure and is represented by TMP =

pf − patm. The N2 permeation is defined by:

Perm𝑁2=

dJ𝑁2

d(TMP) [

10−5cm3

cm2 s cmHg] (24)

where dJ𝑁2

d(TMP) is the slope of the line flux vs TMP, plotted in Figure 27. The N2 permeance obtained was

0.35 × 10−5 cm3cm−2s−1cmHg−1.

Figure 27 – Volumetric fluxes as a function of TMP of CM membrane.

6.4 Gas Permeation by the Pressure Method

In the pressure method, the permeate pressure, 𝑝𝑝, (mbar) is measured directly as function of

time (s) in experimental setup, resulting in plots similar to the one showed in Figure 28. At the beginning

of each experiment (t = 0s) the relative permeate pressure was 0 mbar and all of the experiments were

performed at feed pressures ranging between 0.5 e 4.0 bar absolute.

The N2, CO2 and O2 permeation parameters of the PEU and PU membranes were determined by

the pressure method, described in section 5.6 .

The permeation flux of the membranes was determined by the slope of the steady state zone of

the permeate pressure vs time graph (Figure 28).

0

20

40

60

80

100

120

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

42

Figure 28 – CO2 permeate pressure (mbar) measurement as a function of time (s) for the PEU-5-100

membrane (CO2 feed pressure of 2.3 bar).

Figure 29 shows the results of CO2 permeate pressure as a function of time obtained for the

samples i), ii) and iii) of the PEU-5-100 membrane, with different feed pressures. Note that, the curve in

Figure 28 is also represented by one of the blue curves in Figure 29.

Table 7 shows the obtained slopes of the linear part of the permeate pressure vs time curves of

samples i), ii), iii) (represented in Figure 29) and iv) of the PEU-5-100 membranes at different CO2 feed

pressures.

Figure 29 – Permeate pressure (mbar) measurements as function of time (s) for samples i) (green), ii)

(blue); iii (black) of the PEU-5-100 membrane at CO2 feed pressures of 1.6±0.1 bar; 2.0±0.1 bar;

2.3±0.1 bar; 2.8±0.1 bar.

0

5

10

15

20

25

0 50 100 150 200 250

Pp

(mb

ar)

t(s)

0

2

4

6

8

10

12

14

16

0 50 100 150

Pp

(mb

ar)

t(s)

43

Table 7 – Slopes of the linear part of the permeate pressure vs time curves of samples i), ii) and iii) of the PEU-5-100 membranes at CO2 feed pressures of 1.6±0.1 bar; 2.0±0.1 bar; 2.3±0.1 bar; 2.8±0.1

bar. dp/dt (mbar/s)

pf(bar) i) ii) iii) iv <dp/dt> (mbar/s)

σ(mbar/s)

1.6 0.029 0.054 0.054 0.045 0.046 0.010

2.0 0.050 0.069 0.086 0.101 0.076 0.019

2.4 0.068 0.094 0.119 0.131 0.103 0.024

2.8 0.107 0.118 0.155 0.164 0.136 0.024

To obtain the flux of gas passing through the membrane from the feed compartment to the

permeate side, some calculations were required to convert the effected pressure variation measurement

into volume variation. As the measurements were carried out at 25 °C and permeate pressure was

always small, by considering the ideal gas law, one can transform the slopes of the linear part of Figure

28, (dP

dt), to a volumetric flow (volume per unit time) (

dV

dt). Eq. (25) was used to calculate the STP volume

that crossed the membrane at each instant of time.

dV

dt=

dn

dt

RTSTP

PSTP

(25)

Where dn

dt is the molar flow of gas passing to the permeate side, obtained from Eq. (26),

considering the bath temperature (T) at the time of the experiments, and the volume of the permeate

compartment in the experimental setup (Vinst).

dn

dt=

dpdt

⁄ Vinst

RT

(26)

Substituting Eq. (26) in Eq. (25), one obtains Eq. (27).

dV

dt=

dpdt

⁄ Vinst

RT

RTSTP

PSTP

(27)

The volumetric flux (J), defined by Eq. (28), is obtained by dividing the volumetric flow (Eq. (27))

by the effective surface area of the permeation cell, A.

The slopes of the straight lines of the permeate pressure vs time plots, presented in Table 7,

were used to obtain the volumetric flux (J) at each feed pressure. Considering that at the beginning of

the measurement, the receiving chamber has 1 atm of air, one can define TMP as the difference

between the feed pressure and atmospheric pressure (𝑇𝑀𝑃 = 𝑝𝑓 − 𝑝𝑎𝑡𝑚).

J =

dVdt⁄

A

(28)

44

J vs TMP curves for the four PEU-5-100 membrane samples are presented in Figure 30. The

permeance (Perm) of each sample of the PEU-5-100 membrane was obtained by the Eq. (29). Results

show that the permeance for each of the four PEU-5-100 membranes samples was 0.10, 0.12, 0.12,

and 0.17 (10−5 cmSTP3 cm−2s−1cmHg−1).

Figure 30 – CO2 volumetric fluxes (10−5 cm3cm−2s−1) as function of TMP (cmHg), obtained for

samples i) (green); ii) (blue); iii) (black) and iv) (grey) of PEU-5-100 membrane.

The CO2 permeance (Perm) was obtained from each of the J vs TMP graphs and an average

value was obtained for the PEU-5-100 membrane after using Dixon’s Q test to identify and reject outliers.

For the PEU-5-100 membrane the mean Perm value was 0.14 ± 0.03 × 10−5 cm3(STP)cm−2s−1cmHg−1.

In the next sections the CO2, N2 and O2 average permeance values obtained for the PU and

PEU membranes are presented and discussed.

6.4.1 Volumetric method vs Pressure method

The N2 permeation fluxes of CM membrane were tested using volumetric and pressure

methods. In order to compare the two methods, the J vs TMP graphs obtained for different samples of

the CM membrane using both methods are represented in Figure 31.

Table 8 shows the N2 permeance, Perm, value and the method used to determine it, for different

samples of CM membrane.

0

5

10

15

20

25

30

35

0 50 100 150 200

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

Perm =

dJ

d(TMP) [

cmSTP3

cm2. s. cmHg] (29)

45

Figure 31 – N2 volumetric fluxes flux J(10−5 cm3cm−2s−1) as function of TMP (cmHg) obtained by the

pressure method (dots) and volumetric method (triangles).

Table 8 – Method used and permeance values measured for each CM sample.

Sample Method 𝐏𝐞𝐫𝐦 𝐍𝟐(𝟏𝟎−𝟓𝐜𝐦𝟑

𝐜𝐦𝟐𝐬𝐜𝐦𝐇𝐠 )

i pressure 0.34

ii pressure 0.27

iii pressure 0.37

iv volumetric 0.35

v volumetric 0.31

Results show that the average values of N2 permeance of the CM membrane were 0.33 × 10−5 ±

0.04 𝑐𝑚3𝑐𝑚−2𝑠−1𝑐𝑚𝐻𝑔−1 for the pressure method and 0.33 × 10−5 ± 0.02 𝑐𝑚3𝑐𝑚−2𝑠−1𝑐𝑚𝐻𝑔−1 for the

volumetric method. The average values permeance obtained show a very low variation, and therefore

an evidence of the consistency of both methods. Furthermore, in the built set-up of pressure method,

the measurements are made online, which allows a more efficient and a quickest obtainment of the

results and it is capable to detect and record very low permeation fluxes, which is a major advantage of

the pressure method over the volumetric method.

6.5 CO2 permeation properties of the PU nonporous symmetric

membranes

The PU nonporous symmetric membranes were tested with CO2 by the pressure method. Figure

32 shows the average volumetric fluxes J(10−5 cm3cm−2s−1) of membranes PU100, PU95 and PU85 as

0

20

40

60

80

100

120

140

0 100 200 300 400

J(1

0^-

5 c

m^3

cm

^-2

s^-

1)

TMP(cmHg)

46

a function of the transmembrane pressure (cmHg) and Table 9 shows the average total membrane

thickness (δ), the average permeance (Perm) and permeability coefficient (PCO2).

The permeability coefficient (P), in units of Barrer, was obtained by multiplying the mean

permeance (Perm) by the dense layer thickness (Eq. (30)).

P = Perm × δ [cmSTP

3 . cm

cm2. s. cmHg10−10]

(30)

The only difference between the three PU membranes was the PCL content: 0, 5 and 15 %wt.

for the PU100, PU95 and PU85 membranes, respectively. The average CO2 permeance of the PU100,

PU95 and PU85 membranes is 0.15 × 10−5cmSTP3 cm−2s−1cmHg−1, 0.07 × 10−5 cmSTP

3 cm−2s−1cmHg−1,

0.19 × 10−5 cmSTP3 cm−2s−1cmHg−1, respectively. The values of CO2 Perm have the same order of

magnitude and increase in the order of the PU95, PU100, and PU85 membranes. The CO2 permeability

obtained for the PU membranes was 163, 94, and 218 Barrer for the PU100, PU95 and PU85

membranes, respectively. Thus, it can be concluded that there is no direct correlation between the

increasing amounts of PCL polymer in these membranes and the CO2 permeability properties.

Figure 32 – Average CO2 volumetric fluxes J(10−5 cm3cm−2s−1) as function of TMP (cmHg) for the PU100 (blue), PU95(orange) and PU85(grey) membranes.

0

10

20

30

40

50

60

70

80

90

0 100 200 300 400

J(1

0^-

5 c

m^3

cm

^-2

s^-

1)

TMP(cmHg)

47

Table 9 – Values of total membrane thickness, mean Permeation and permeability coefficient obtained for the PU membranes.

Membrane 𝛅(𝛍𝐦) 𝐏𝐞𝐫𝐦 𝐂𝐎𝟐(𝟏𝟎−𝟓 𝐜𝐦𝟑

𝐜𝐦−𝟐𝐬−𝟏𝐜𝐦𝐇𝐠−𝟏)

𝐏𝐂𝐎𝟐

[𝐁𝐚𝐫𝐫𝐞𝐫]

PU1 107 0.15 163

PU2 126 0.07 94

PU3 114 0.19 218

6.6 CO2 permeation properties of the PEU integral asymmetric

membranes prepared with different solvent evaporation time

Figure 33 (a) shows the average volumetric fluxes J(10−5 cm3cm−2s−1) of the PEU-1-100, PEU-

5-100, PEU-10-100 and PEU-15-100 membranes as a function of TMP (cmHg) and Table 10 shows the

values average permeance obtained for the PEU membranes. The only difference between the four

PEU membranes was the solvent evaporation time used for the synthesis: 1, 5, 10 and 15 minutes for

the PEU-1-100, PEU-5-100, PEU-10-100 and PEU-15-100 membranes, respectively. Results show that

the average permeance of the PEU-1-100 membrane (10 ± 8 × 10−5 cmSTP3 cm−2s−1cmHg−1) is several

orders of magnitude greater than the other PEU membranes.

Figure 33 (b) shows the same data depicted in Figure 33 (a), but only for the PEU-5-100, PEU-

10-100 and PEU-15-100 membranes. The average CO2 permeance of the PEU-5-100, PEU-10-100 and

PEU-15-100 membranes is 0.14 ± 0.03 × 10−5cmSTP3 cm−2s−1cmHg−1, 0.11±0.01 ×

10−5 cmSTP3 cm−2s−1cmHg−1, 0.14±0.01 × 10−5 cmSTP

3 cm−2s−1cmHg−1, respectively, and a ANOVA-test

performed for these results found that there is no significant difference for a confidence interval of 99.5%.

Results show that there is no correlation between the solvent evaporation time and the CO2 permeance.

In fact, the CO2 permeance value obtained for the nonporous symmetric PU membranes is the same

order of magnitude as the PEU-5-100, PEU-10-100 and PEU-15-100 asymmetric membranes.

For a solvent evaporation time equal to and greater than 5 mins the solvent evaporation time

seems to have no effect on the CO2 permeance, but for a solvent evaporation of 1 minutes a permeance

approximately 100 times higher is observed. A possible explanation for this phenomenon can be that a

solvent evaporation time of 1 min is insufficient for the formation of an upper denser layer, near the

surface of the membranes, which in turn offers a higher resistance to gas permeation when compared

to microporous membranes. In order to obtain a thin denser layer at the surface of the membrane and

increase the resistance to gas transport, there needs to be enough time to allow the evaporation of the

more volatile solvent, DEE, before the polymer membrane be quenched in the water bath [29]. The

thickness of this denser layer is expected to be greater for increasing solvent evaporation times. This is

confirmed by the fact that the membranes synthesized with solvent evaporation times larger than 1 min

showed much higher resistance to CO2 permeation. Taking into account these results one can conclude

48

that the PEU-1-100 membrane has a structure and behavior that resembles more a microporous

membrane rather than an integral asymmetric membrane.

(a) (b)

Figure 33 – (a) Average CO2 volumetric fluxes J(10−5 cm3cm−2s−1) as function of TMP (cmHg) for the PEU-1-100 (green), PEU-5-100 (blue), PEU-10-100 (orange) and PEU-15-100 (grey) membranes; (b)

Average CO2 volumetric fluxes J(10−5 cm3cm−2s−1) as function of the transmembrane pressure (cmHg) for the PEU-5-100 (blue), PEU-10-100 (orange) and PEU-15-100 (grey) membranes

Table 10 – Values of average permeance obtained for PEU membranes

Membrane 𝐏𝐞𝐫𝐦 𝐂𝐎𝟐(𝟏𝟎−𝟓 𝐜𝐦𝟑


PEU-1-100 10

PEU-5-100 0.14

PEU-10-100 0.11

PEU-15-100 0.14

6.7 CO2, N2, and O2 permeation properties of the PEU membranes with

different PU/PCL weight ratios.

6.7.1 CO2 permeation of integral asymmetric PEU Membranes

Figure 34 shows the average volumetric fluxes J(10−5 cm3cm−2s−1) vs TMP (cmHg) and

Table 11 shows the average permeation (Perm).

These four PEU membranes were prepared under the same conditions, with a solvent

evaporation time of 5 mins being the only difference between them the PCL content which is 0, 5, 10

and 15 wt.% for the PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85 membranes, respectively.

Results show that the CO2 permeance measured for the PEU membranes were

0.14±0.03 × 10−5 cmSTP3 cm−2s−1cmHg−1, 0.11 ± 0.01 × 10−5 cmSTP

3 cm−2s−1cmHg−1, 0.12 ± 0.02 ×

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 100 200 300 400

J(1

0^5

cm

^3/c

m^2

/s)

TMP(cmHg)

0

10

20

30

40

50

60

0 100 200 300 400

J(1

0^5

cm

^3/c

m^2

/s)

TMP(cmHg)

49

10−5 cmSTP3 cm−2s−1cmHg−1 and 0.13 ± 0.03 × 10−5 cmSTP

3 cm−2s−1cmHg−1 for the PEU-5-100, PEU-5-

95, PEU-5-90 and PEU-5-85 membranes, respectively, and a ANOVA-test performed for these results

found that there is no significant difference for a confidence interval of 99.5%. The values of CO2 Perm

have the same order of magnitude and increase in the order of the PEU-5-95, PEU-5-90 and PEU-5-85

and PEU-5-100 membranes. The CO2 permeability obtained for the PEU membranes was 18, 26, 15

and 22 Barrer for the PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85 membranes, respectively. Thus,

it can be concluded that there is no direct correlation between the amount of PCL polymer in these

membranes and the permeability properties.

Figure 34 – Average CO2 volumetric fluxes 𝐽(10−5 𝑐𝑚3𝑐𝑚−2𝑠−1) as function of TMP (cmHg) of PEU-5-

100 (blue), PEU-5-95 (orange), PEU-5-90 (grey) and PEU-5-85 (yellow) membranes.

Table 11 – CO2 average permeance obtained for PEU membranes.

Membrane 𝐏𝐞𝐫𝐦 𝐂𝐎𝟐(𝟏𝟎−𝟓 𝐜𝐦𝟑


PEU-5-100 0.14

PEU-5-95 0.11

PEU-5-90 0.12

PEU-5-85 0.13

6.7.2 N2 permeation of integral asymmetric PEU Membranes

Figure 35 shows the average volumetric fluxes of N2 for the PEU-1-100, PEU-1-95, PEU-1-90

and PEU-1-85 membranes as function of transmembrane pressure. Table 12 shows the average

permeation (Perm). These four PEU membranes were prepared under the same conditions, with a

solvent evaporation time of 1 min being the only difference between them the PCL content which is 0,

5, 10 and 15 wt.% for the PEU-1-100, PEU-1-95, PEU-1-90 and PEU-1-85 membranes, respectively.

Results show that the N2 permeance measured for the PEU membranes was 7 ×

10−5 cmSTP3 cm−2s−1cmHg−1, 19 × 10−5 cmSTP

3 cm−2s−1cmHg−1, 179 × 10−5 cmSTP3 cm−2s−1cmHg−1 and

0

10

20

30

40

50

60

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

50

20 × 10−5 cmSTP3 cm−2s−1cmHg−1 for the PEU-1-100, PEU-1-95, PEU-1-90 and PEU-1-85 membranes,

respectively.

Note that the N2 permeance obtained for the PEU-1-95, PEU-1-90 and PEU-1-85 are orders of

magnitude greater than the CO2 permeance obtained for the PEU-1-100 membrane, also prepared with

a solvent evaporation time of 1 minute. Due to the lack of resistance to transport of these membranes,

the measurements were very fast regard to the small volume of the receiving chamber. These results

were not reproducible and are associated to a large uncertainty. Thus, these measurements must be

questioned.

Figure 35 – Average N2 volumetric fluxes 𝐽(10−5 𝑐𝑚3𝑐𝑚−2𝑠−1) as function of TMP (cmHg) of PEU-1-100 (blue), PEU-1-95 (orange), PEU-1-90 (grey), PEU-1-85 (yellow).

Table 12 – N2 average permeance obtained for PEU membranes.

Membrane 𝐏𝐞𝐫𝐦 𝐍𝟐(𝟏𝟎−𝟓 𝐜𝐦𝟑


PEU-1-100 7

PEU-1-95 19

PEU-1-90 179

PEU-1-85 20

Figure 36 shows the average volumetric fluxes of N2 as function of the TMP for the PEU-5-100,

PEU-5-95, PEU-5-90 and PEU-5-85 membranes and Table 13 shows the average permeation (Perm).

These four PEU membranes were prepared under the same conditions, with a solvent evaporation time

of 5 mins being the only difference between them the PCL content which is 0, 5, 10 and 15 wt.% for the

PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85 membranes, respectively.

Results show that the N2 permeance measured for the PEU membranes was 0.003±0.001 ×

10−5 cmSTP3 cm−2s−1cmHg−1, 0.006 ± 0.003 × 10−5 cmSTP

3 cm−2s−1cmHg−1, 0.004 ± 0.001 ×

10−5 cmSTP3 cm−2s−1cmHg−1 and 0.004 ± 0.001 × 10−5 cmSTP

3 cm−2s−1cmHg−1 for the PEU-5-100, PEU-

5-95, PEU-5-90 and PEU-5-85 membranes, respectively. The values of N2 Perm have the same order

0

2000

4000

6000

8000

10000

12000

14000

0 100 200 300 400

J (1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

51

of magnitude and increase in the order of the PEU-5-100, PEU-5-90, PEU-5-85 and PEU-5-95

membranes.

Results show that the PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85 average N2 permeances

are several orders of magnitude lower than N2 average permeances obtained for the PEU-1-100, PEU-

1-95, PEU-1-90 and PEU-1-85 membranes, prepared with 1 min of solvent evaporation time. This

indicates that the resistance to transport is highly dependent of the solvent evaporation time and there

are no evidences of a direct correlation between the PCL content of the PEU membranes and the N2

permeances.

Figure 36 – Average N2 volumetric fluxes 𝐽(10−5 𝑐𝑚3𝑐𝑚−2𝑠−1) as function of TMP (cmHg) of the PEU-5-100 (blue), PEU-5-95 (orange), PEU-5-90 (grey), PEU-5-85 (yellow).

Table 13 – N2 average permeance obtained for PEU membranes.



PEU-5-100 0.003

PEU-5-95 0.006

PEU-5-90 0.004

PEU-5-85 0.004

6.7.3 O2 permeation of integral asymmetric PEU membranes

Figure 37 shows the average volumetric fluxes of O2 for the PEU-5-100, PEU-5-95, PEU-5-90

and PEU-5-85 membranes as function of TMP. Table 14 shows the O2 permeance (Perm) for the PEU-

5-100, PEU-5-95, PEU-5-90 and PEU-5-85 membranes.

Results show that the O2 permeance measured for the PEU membranes was 0.011 ×

10−5 cmSTP3 cm−2s−1cmHg−1, 0.008 × 10−5 cmSTP

3 cm−2s−1cmHg−1, 0.010 × 10−5 cmSTP3 cm−2s−1cmHg−1

and 0.015 × 10−5 cmSTP3 cm−2s−1cmHg−1 for the PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85

membranes, respectively. The values of O2 Perm have the same order of magnitude and increase in

0

0,5

1

1,5

2

2,5

0 50 100 150 200 250 300 350 400

J(1

0^-

5 c

m^3

/cm

^/s)

TMP(cmHg)

52

the order of the PEU-5-95, PEU-5-90, PEU-5-100 and PEU-5-85 membranes. Thus, it can be concluded

that there is no direct correlation between the amount of PCL polymer in these membranes and the

permeability properties.

Figure 37 – Average O2 volumetric fluxes 𝐽(10−5 𝑐𝑚3𝑐𝑚−2𝑠−1) as function of TMP (cmHg) of PEU-5-100 (blue), PEU-5-95 (orange), PEU-5-90 (grey), PEU-5-85 (yellow).

Table 14 – O2 average permeance obtained for PEU membranes.

Membrane 𝐏𝐞𝐫𝐦 𝐎𝟐

(𝟏𝟎−𝟓 𝐜𝐦𝟑𝐜𝐦−𝟐𝐬−𝟏𝐜𝐦𝐇𝐠−𝟏)

PEU-5-100 0.011

PEU-5-95 0.008

PEU-5-90 0.010

PEU-5-85 0.015

6.7.4 Comparing the different gases permeation of asymmetric PEU membranes

Figure 38 shows the average volumetric fluxes of N2, CO2 and O2 obtained for the PEU-5-100,

PEU-5-95, PEU-5-90 and PEU-5-85 membranes, as function of transmembrane pressure. Membranes

PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85 were prepared with 1/1 total polymer to total solvent

weight ratio, solvent evaporation time of 5 minutes, and 100/0, 95/5, 90/10 and 85/15 of PUR/PCL wt.

%, respectively.

Table 15 shows the upper denser layer thickness, average N2, CO2 and O2 permeation (Perm)

and N2, CO2 and O2 permeability coefficient (P) for the PEU-5-100, PEU-5-95, PEU-5-90 and PEU-5-85

membranes.

Results show that the O2 and N2 permeation fluxes are much lower than the CO2 permeation

fluxes. Permeance values of CO2 for the PEU membranes were one order of magnitude higher than O2

permeance values and two orders of magnitude higher than N2 permeance. The molecular diameter of

0

1

2

3

4

5

6

7

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

53

O2 is 3.5 Å, of N2 is 3.7 Å and of CO2 is 4.3 Å [37]. It is known that the diffusion coefficient decreases as

the size of the gas molecules increases, which is verified by the obtained results of O2 permeance which

is higher than the N2 permeance. This is also verified in several studies where the tested polyurethanes

have higher permeability, diffusivity and solubility for O2 than for N2. [38] [39]

Considering only the diameter aspect, one would expect that the permeability would be bigger

for the smallest gas, O2, which is not the case. A possible explanation for this is the fact that the transport

of a gas through dense membranes can be described by a 3 step mechanism, sorption-diffusion-

desorption. It is known that the solubility coefficient increases with increasing molecular dimensions and

N2 and O2 may be considered non-interacting gases, since the interaction with the polymer is in general

very small, contrary to CO2 which can show some interaction. [29]

Results show that the membranes with the higher permeance are the PEU-5-90 and the PEU-

5-85 with CO2 permeances of 0.14 × 10−5 cm3(STP)cm−2s−1cmHg−1, 0.13 ×

10−5 cm3(STP)cm−2s−1cmHg−1, and O2 permeances of 0.01 × 10−5 cm3(STP)cm−2s−1cmHg−1, 0.02 ×

10−5 cm3(STP)cm−2s−1cmHg−1, respectively. These two PEU membranes were prepared under the

same conditions, with a solvent evaporation time of 5 mins being the only difference between them the

PCL content which is 0 and 15 wt.% for the PEU-5-100 and PEU-5-85 membranes, respectively. Taking

into account these results together with the enhanced hemocompatibility of the membranes containing

15 wt.% of PCL found by Faria et al. [25], one can consider that the PEU-5-85 membrane, prepared

with 1/1 total polymer to total solvent weight ratio, 5 minutes of solvent evaporation time and 85/15 of

PUR/PCL wt. %, as being the most promising membrane to be applied in future MBOs.

54

(a) (b)

(c) (d)

Figure 38 – Average N2 (blue), CO2 (orange) and O2 (grey) volumetric fluxes 𝐽(10−5 𝑐𝑚3𝑐𝑚−2𝑠−1) as function of TMP (cmHg) of PEU-5-100 (a), PEU-5-95 (b), PEU-5-90 (c), PEU-5-85 (d).

Table 15 – Average permeance of N2, CO2 and O2, obtained for PEU membranes.



𝐏𝐞𝐫𝐦 𝐂𝐎𝟐(𝟏𝟎−𝟓 𝐜𝐦𝟑


𝐏𝐞𝐫𝐦 𝐎𝟐(𝟏𝟎−𝟓 𝐜𝐦𝟑


PEU-5-100 0.003 0.14 0.011

PEU-5-95 0.006 0.11 0.008

PEU-5-90 0.004 0.12 0.010

PEU-5-85 0.004 0.13 0.015

According to [2], the MBOs must remove 200 cm3/min of CO2, which is the same as 3.3 cm3/s.

Thus, to obtain gas flows suitable for future MBOs, the PEU-5-85 membrane, with a CO2 permeance of

0.13 × 10−5 cm3(STP)cm−2s−1cmHg−1, and considering a TMP of 76 cmHg, must have a surface area

0

5

10

15

20

25

30

35

40

45

50

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

0

5

10

15

20

25

30

35

40

45

50

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

0

5

10

15

20

25

30

35

40

45

50

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

0

5

10

15

20

25

30

35

40

45

50

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

55

of 3.4 m2 to ensure adequate CO2 fluxes. The improvement of the permeation fluxes, can further be

achieved with the tailoring of membranes with thinner upper denser layers.

In order to compare the gas permeation properties of the integral asymmetric PEU membranes

to the ones found for a commercial membrane, the permeation properties of the CM membrane were

studied.

Figure 39 shows the average N2, CO2 and O2 volumetric fluxes vs TMP curves obtained by the

pressure method for the CM membrane, and Table 16 shows the values of the membrane thickness (δ),

and the N2, CO2 and O2 mean permeance and permeability obtained for the CM membrane. Results

show that, for the CM membrane, the average N2, CO2 and O2 permeance were 0.3 ×

10−5 cm3cm−2s−1cmHg−1, 3.6 × 10−5 cm3cm−2s−1cmHg−1 and 1.1 × 105 cm3cm−2s−1cmHg−1,

respectively. These values are an order of magnitude greater than the permeances required for efficient

MBOs, (0.22 × 10−5 cm3cm−2s−1cmHg−1 for CO2 and 0.27 × 10−5 cm3cm−2s−1cmHg−1 for O2) [2]. Even

though the N2, CO2 and O2 permeances obtained for the CM membrane are much higher than the ones

found for the PEU membranes it is important to remember that the hemocompatibility of the

poly(urethane urea) membranes is significantly better, particularly for the membranes containing 10 and

15 wt.% of PCL. [31]

It is also important to note that the permeation experiments performed in this work take place in

a gas/membrane/gas set-up as opposed to operational MBOs that operate in a gas/membrane/liquid

environment, where the liquid is blood that has to be oxygenated at the same time that the carbon

dioxide is being removed. In the latter conditions, the resistance to gas transport is higher and therefore

it is plausible to admit that the permeance values obtained for the CM membrane in this work are

equivalent to those stated by the membrane fabricator without considering the additional resistance to

gas transfer offered by the blood.

Figure 39 – Measurements of N2 (blue), CO2 (orange) and O2 (grey) flux J(10−5 cm3cm−2s−1) as

function of TMP (cmHg) for the CM membrane.

0

200

400

600

800

1000

1200

1400

1600

0 100 200 300 400

J(1

0^-

5 c

m^3

/cm

^2/s

)

TMP(cmHg)

56

Table 16 – Values of membrane thickness, and the N2, CO2 and O2 mean permeance and permeability obtained for the commercial membrane.

𝛅(𝛍𝐦) 𝐏𝐞𝐫𝐦 (𝟏𝟎−𝟓 𝐜𝐦𝟑

𝐜𝐦−𝟐𝐬−𝟏𝐜𝐦𝐇𝐠−𝟏) 𝐏[𝐁𝐚𝐫𝐫𝐞𝐫]

N2 66 0.3 275

CO2 66 3.6 2371

O2 66 1.1 715

6.8 Determination of CO2 diffusion and solubility coefficients

As was described in Figure 8 in Section 4, the permeate pressure measurements versus time

curves exhibit, at short times, a zone of non-linearity, corresponding to the transient state of permeation.

Analysis of this section of the graphs together with Equations pertaining to the time-lag/late

approximation method and the early approximation method were carried out, in order to obtain CO2

diffusion and solubility coefficients for the PU95 non-porous symmetric membrane.

A point of divergence between theory and application in this work, was the operational

conditions and the initial and boundary conditions. Both, time-lag and early approximations are solutions

of the Fick’s second Law under the conditions that initially the membrane is gas free and that the

pressure in the receiving chamber is maintained near zero, which means that measurements must be

initialized with vacuum in both chambers. However, during this work, pressure on the receiving chamber

was maintained near atmospheric pressure during measurements, and at initial instants the membrane

samples were under gas feed pressure. Thereafter, in the calculations where equation related to the

time-lag and the early approximation were necessary, instead of the feed pressure, pf, it was used the

pressure difference between permeate and feed side of the membrane, TMP.

Because of this, values obtained and discussed in this section, must be considered preliminary

results, and a motivation for future work.

6.8.1 Time-lag method

In order to determine the DCO2and KCO2

coefficients by the time-lag method, the graphical

extrapolation of the straight line in the pressure vs time graphs (Figure 8, Section 4), which corresponds

to the steady state, was performed to the pressure vs time curve for the PU95 membrane (Figure 40) to

obtain the values of the characteristic time, tc, and the characteristic pressure, pc. Eq. (17) and Eq.

(18) of Section 4, were used to determine DCO2 and KCO2

, respectively, and the permeability coefficient,

𝑘CO2, was obtained from Eq. (11).

Table 17 shows the transmembrane pressure, TMP, the membrane thickness, δ, the slope of

the linear section of permeate pressure vs time plot (dp/dt), the interceptions tc and pc, and the DCO2,

KCO2, kCO2

and PCO2 obtained by the time-lag method for the dense PU95 membrane.

57

Results show that the diffusion coefficient were 5.6 × 10−7cm2s−1 and the solubility coefficients were

124.7 × 10−2cmHg 𝑐𝑚𝐶𝑂23 cmHg−1𝑐𝑚𝑚𝑒𝑚𝑏

3 . The permeability coefficients obtained were 6.9 ×

10−7𝑐𝑚2𝑠−1, that corresponds to 106 Barrer.

Figure 40 – CO2 permeate pressure (mbar) measurement as a function of time (s) for the PU95 membrane (CO2 feed pressure of 1.9 bar).


obtained by the time-lag method.

𝐓𝐌𝐏 (𝐜𝐦𝐇𝐠)

𝛅(µ𝐦) 𝐝𝐩

𝐝𝐭⁄

(𝟏𝟎−𝟑 𝐜𝐦𝐇𝐠𝐬−𝟏) 𝐭𝒄(𝐬) 𝐩𝒄(𝐜𝐦𝐇𝐠)

𝐃𝑪𝑶𝟐

(𝟏𝟎−𝟕 𝐜𝐦𝟐𝐬−𝟏)

𝐊𝑪𝑶𝟐

(𝟏𝟎−𝟐)

𝐤𝑪𝑶𝟐

(𝟏𝟎−𝟕𝐜𝐦𝟐𝐬−𝟏)

𝐏𝑪𝑶𝟐

(𝐁𝐚𝐫𝐫𝐞𝐫)

65.8 126 1.26 47 -0.060 5.6 124.7 6.9 105

6.8.2 Early approximation method

According to the early approximation method, Eq.(21), Eq. (22) and Eq. (11), of Section 4, were

used to determine DCO2, KCO2

and kCO2, respectively, using the slope and y-intercept.

Figure 41 shows the lower times of the ln (√tdp

dt) vs

1

t plot obtained from the CO2 permeate

pressure measurement as a function of time for the PU95 membrane. The plot points correspond to the

early times of Figure 40, for a maximum time of 12 seconds, which is assuredly in the range of the

transient section.

Table 18 shows the transmembrane pressure, TMP, membrane thickness, δ, the slope, 𝑚, and

y-axis interception, 𝑏, of ln (√tdp

dt) vs

1

t plot, and the DCO2

, KCO2, kCO2

and PCO2 obtained for the PU95

membrane by the early approximation method. Results show that the diffusion coefficient were

-0,08

-0,06

-0,04

-0,02

0

0,02

0,04

0,06

0,08

0,1

0,12

0,14

0 25 50 75 100 125 150

pp(c

mH

g)

t(s)

𝑝𝑐

𝑡𝑐

58

96.9 × 10−7cm2s−1 and the solubility coefficients were 5.7 × 10−2cmHg 𝑐𝑚𝐶𝑂23 cmHg−1𝑐𝑚𝑚𝑒𝑚𝑏

3 . The

permeability coefficients obtained were 5.6 × 10−7𝑐𝑚2𝑠−1, that corresponds to 85 Barrer.

Figure 41 – ln (√tdp

dt) vs

1

t plot obtained from the CO2 permeate pressure (mbar) measurement as a

function of time (s) for the PU95 membrane (CO2 feed pressure of 1.9 bar).

Table 18 – TMP, membrane thickness and CO2 permeation properties of the PU95 membrane obtained by the early approximation method.

𝐓𝐌𝐏 (𝐜𝐦𝐇𝐠) 𝛅(µ𝐦) 𝐒𝐥𝐨𝐩𝐞 𝐛 𝐃𝑪𝑶𝟐

(𝟏𝟎−𝟕 𝐜𝐦𝟐𝐬−𝟏)

𝐊𝑪𝑶𝟐

(𝟏𝟎−𝟐)

𝐤𝑪𝑶𝟐

(𝟏𝟎−𝟕𝐜𝐦𝟐𝐬−𝟏)

𝐏𝑪𝑶𝟐

(𝐁𝐚𝐫𝐫𝐞𝐫)

65.8 126 -4.1 -5.4 96.9 5.7 5.6 85

The CO2 permeability obtained for the PU95 membrane by the time-lag method and by the early

approximation method was 105 Barrer and 85 Barrer, respectively. The experimental CO2 permeability

obtained from the permeation experiments, discussed in Section 6.5 for the PU95 membrane was 94

Barrer. Thus, it is verified that this permeability value, obtained by multiplying the permeance and the

average total membrane thickness, is in accordance with the permeability coefficients calculated from

the permeation parameters, DCO2 and KCO2

, obtained from the transient state of the permeation

measurements.

-6,15

-6,10

-6,05

-6,00

-5,95

-5,90

-5,85

-5,80

-5,75

-5,70

-5,65

0,05 0,10 0,15 0,20

Ln(d

pp

/dt.

t^0

.5)

1/t (s^-1)

59

7. Conclusions

A novel experimental set-up capable of recording the evolution of pressure online, in intervals

of 1.4 seconds, with milibar precision, at constant temperature was built and validated. The set-up allows

to perform measurements, at chosen constant temperature, with a duration up to 90 minutes.

For membranes with a range of permeances between 0.003 × 10−5𝑐𝑚3𝑐𝑚−2𝑠−1𝑐𝑚𝐻𝑔−1 and

3.6 × 10−5𝑐𝑚3𝑐𝑚−2𝑠−1𝑐𝑚𝐻𝑔−1, it was possible to obtain reproducible results, being the measurements

uncertainty lower than the variability associated to the membranes synthesis.

It was verified that for measurements of high permeation fluxes, like the ones obtained for the

integral asymmetric PEU-1-* membranes, the receiving chamber volume is too small, resulting in

measurements of low reproducibility and with high uncertainty associated.

Poly(ester urethane urea) membranes prepared by a modified version of the phase inversion

technique, by reacting a polyurethane (PUR) prepolymer with a polycaprolactone (PCL) prepolymer,

displayed asymmetric cross-sectional structures that were tailored upon the variations of the casting

solutions and conditions. The increase of the total polymer to total solvent weight ratio resulted in more

uniform, completely formed membranes, however, an excess of polymer resulted in less elastic, very

sticky membranes which are difficult to manage. It is concluded that, to be able to obtain membranes

with optimal structural characteristics there must be a certain balance between the total amount of

polymer and solvent and, for the particular case of the prepolymers used in this work, the most suitable

composition was total polymer/total solvent weight ratio 1/1.

The integral asymmetric poly(urethane urea) membranes appear to have a characteristic cross

section structure where the expected very thin active layer and much thicker porous bottom layer are

not easily identified. Instead three distinct regions can be seen: two regions that boarder the upper

(upper denser layer) and bottom surface composed of very small pores in large number and a thicker

region located at the center part of the membrane composed of larger pores and less numerous.

The increase of the amount of PCL in the PEU membranes improves the structural

characteristics and membranes become whiter, less yellow and stickier. It was also verified that as the

PCL content increases, the distinction between the smaller-pore regions and the internal large-pore area

become more noticeable and, apparently, the number of pores in the inner region decreases even more.

The increase of the polymer to solvent ratio and the increase of the solvent evaporation time

results in the increasing of the total membrane thickness. The measurements of the total membrane

thickness for the integral asymmetric membranes were between 88 and 203 μm.

The nonporous symmetric membranes, PU100, PU95 and PU85, prepared with a PCL content

of 0, 5 and 15%, respectively, were dense, translucent, glassy and very sticky. The total membrane

thickness of these membranes was between 107 and 126 μm.

The experimental measurements of the permeation fluxes obtained by the pressure method

were coherent with the ones obtained with the volumetric method. Furthermore, in the built set-up of

pressure method, the fact of the measurements are made online allows a more efficient and a quickest

obtainment of the results and it is capable to detect and record very low permeation fluxes, which is a

major advantage of the pressure method over the volumetric method.

60

The CO2 permeance values obtained for the nonporous symmetric membranes with 0, 5 and

15% weight ratio of PCL were 0.15 × 10−5, 0.07 × 10−5 and 0.19 × 10−5 cm3cm−2s−1cmHg−1,

respectively.

The highest CO2 permeance value (10.4 × 10−5 cm3cm−2s−1cmHg−1) was measured for the

asymmetric membranes prepared with the lowest evaporation time (1 minute). The permeance values

obtained for the asymmetric membranes prepared with higher solvent evaporation time were two orders

of magnitude lower: 0.14 × 10−5, 0.11 × 10−5 and 0.14 × 10−5cm3cm−2s−1cmHg−1, for 5, 10 and 15

minutes of solvent evaporation time, respectively, and these are of the same order of magnitude as for

the nonporous symmetric membranes. Thus, the solvent evaporation time, above 5 minutes, does not

affect the permeability properties of the membranes.

The CO2, N2 and O2. permeances of the asymmetric membranes prepared with a solvent

evaporation time of 5 minutes and varying content of PCL, were determined and results showed that: i)

the CO2 permeances were 0.14 × 10−5, 0.11 × 10−5, 0.12 × 10−5, 0.13 × 10−5 cm3cm−2s−1cmHg−1

for the membranes containing 0, 5, 10, 15% wt. of PCL, respectively; ii) the N2 permeances were

0.003 × 10−5, 0.006 × 10−5, 0.004 × 10−5, 0.004 × 10−5 cm3cm−2s−1cmHg−1 for the membranes with

0, 5, 10, 15% wt. of PCL, respectively; iii) the measured O2 permeances were 0.011 × 10−5,

0.008 × 10−5, 0.010 × 10−5, 0.015 × 10−5 cm3cm−2s−1cmHg−1 for the membranes with 0, 5, 10, 15%

wt. of PCL, respectively. It is therefore concluded that the amount of PCL in these membranes is not

directly correlated to the gas permeability properties.

Results show that the membranes with the higher permeance are the PEU-5-90 and the PEU-

5-85, which were prepared with polymer to solvent weight ratio of 1/1, a solvent evaporation time of 5

minutes and PCL content of 10 and 15 wt.%, respectively. Considering the enhanced hemocompatibility

of the membranes containing 15 wt.% of PCL, found in previous studies ([24], [25]), it is concluded the

PEU-5-85 membrane, prepared with 1/1 of total polymer to total solvent weight ratio, 5 minutes of

solvent time evaporation and 85/15 of PUR/PCL wt. %, is the most promising membrane to be applied

in future MBOs.

In order to be considered for future MBOs, and functioning at a TMP of 76 cmHg, the PEU-5-85

membrane, must have a surface area of 3.4 m2 to ensure adequate CO2 fluxes. One method of achieving

higher gas fluxes is by tailoring of the PEU-5-85 membrane with a thinner upper denser layer.

The CO2 permeability obtained for the PU95 membrane by the time-lag method and by the early

approximation method was 105 Barrer and 85 Barrer, respectively, and are coherent with the

experimental CO2 permeability obtained from the permeation experiments (94 Barrer). Thus, it is verified

that this permeability value, obtained by multiplying the permeance and the average total membrane

thickness, is in accordance with the permeability coefficients calculated from the permeation

parameters, DCO2 and KCO2

, obtained from the transient state of the permeation measurements.

61

8. Perspectives of the future work

It is suggested that the future work concentrate the synthesis of integral asymmetric poly(ester

urethane urea) membranes, containing PUR and PCL prepolymers, with solvent evaporation times

between 1 and 5 minutes. In order to verify the permeance enhancement of these membranes,

permeation measurements may be performed, using the built set-up.

To allow a more precise measurement of highly permeable membranes, it is suggested the

increase of the volume of the receiving chamber of the experimental set-up. This can be easily

performed by substituting the buffer cylinder by a vessel of known volume.

In order to obtain validated values of the solubility and diffusion coefficients, the use of a vacuum

pump (prior to the gas permeation measurements) to create vacuum on both feed and receiving gas

sides, is recommended. This way, the initial and boundary conditions of the presented second Fick’s

law solution are respected. Furthermore, to ensure the stabilization of the feed pressure is

recommended the assembling of a vessel in the feed side of the set-up. Applied these technical

considerations, it is suggested to perform permeation measurements with CO2 and O2 in order to obtain

the solubility and diffusion coefficients of nonporous symmetric poly(ester urethane urea) membranes

with different quantities of PUR/PCL wt.%, using the same mathematical treatment (time-lag and early

approximation methods).

For studies of the membrane selectivity, it is suggested the assembling of a vessel to the set-

up, before the permeation cell, where all of the gases can be stored before being fed to the membrane

sample. Finally, in order to accurately detect the concentration of gas passing through the membrane,

a gas concentration sensor assembled to the receiving chamber tubbing is recommended.

63

9. Bibliography

[1] I. Facts on File, The Facts on File Encyclopedia of Health and Medicine, no. vol. 1. Facts On File, 2007.

[2] D. F. Stamatialis et al., “Medical applications of membranes: Drug delivery, artificial organs and tissue engineering,” J. Memb. Sci., vol. 308, no. 1–2, pp. 1–34, 2008.

[3] N. C. Nanda, N. Trehan, B. Airan, S. A. Conrad, and Y. Mehta, “Manual of Extracorporeal Membrane Oxygenation (ECMO) in the ICU,” Jaypee Brother. Med. Publ., 2014.

[4] D. Machin and C. Allsager, “Principles of cardiopulmonary bypass,” Contin. Educ. Anaesthesia, Crit. Care Pain, vol. 6, no. 5, pp. 176–181, 2006.

[5] F. C. Lin, D. M. Wang, and J. Y. Lai, “Asymmetric TPX membranes with high gas flux,” J. Memb. Sci., vol. 110, no. 1, pp. 25–36, 1996.

[6] M. Drummond, D. M. Braile, A. P. M. Lima-oliveira, A. S. Camim, R. S. K. Oyama, and G. H. Sandoval, “Technological evolution of membrane oxygenators,” vol. 20, no. 4, pp. 432–437, 2005.

[7] R. W. Baker, Membrane technology and applications. 2004.

[8] G. P. Gravlee, Cardiopulmonary Bypass: Principles and Practice. Wolters Kluwer Health/Lippincott Williams & Wilkins, 2008.

[9] M. Faria, M. Rajagopalan, and M. N. De Pinho, “Tailoring bi-soft segment poly (ester urethane urea) integral asymmetric membranes for CO2 and O2 permeation,” J. Memb. Sci., vol. 387–388, no. 1, pp. 66–75, 2012.

[10] R. W. Fuest, Polyurethane Elastomers. 2009.

[11] M. Ionescu, Chemistry and Technology of Polyols for Polyurethane. 2005.

[12] I. American Chemistry Council, “How Polyurethane Is Made,” 2017. [Online]. Available: https://polyurethane.americanchemistry.com/How-Polyurethane-is-Made/. [Accessed: 13-Sep-2017].

[13] R. J. Zdrahala and I. J. Zdrahala, “Biomedical Applications of Polyurethanes: A Review of Past Promises, Present Realities, and a Vibrant Future,” J. Biomater. Appl., vol. 14, no. 1, pp. 67–90, 1999.

[14] J. W. Boretos and W. S. Pierce, “Segmented Polyurethane: A New Elastomer for Biomedical Applications,” Science (80-. )., vol. 158, no. 3807, pp. 1481–1482, 1967.

[15] M. Marzec, J. Kucińska-Lipka, I. Kalaszczyńska, and H. Janik, “Development of polyurethanes for bone repair,” Mater. Sci. Eng. C, vol. 80, pp. 736–747, 2017.

[16] N. S. Schneider, “Water vapor transport in structurally varied polyurethanes,” J. Macromol. Sci. , Part B Phys., vol. 3, no. February 2014, pp. 37–41, 1969.

[17] P. M. Knight and D. J. Lyman, “Gas permeability of various block copolyether-urethanes,” J. Memb. Sci., vol. 17, no. 3, pp. 245–254, 1984.

[18] N. Cao, M. Pegoraro, F. Bianchi, L. Di Landro, and L. Zanderighi, “Gas transport properties of polycarbonate–polyurethane membrance,” J. Appl. Polym. Sci., vol. 48, no. 10, pp. 1831–1842, 1993.

[19] K. H. Hsieh, C. C. Tsai, and S. M. Tseng, “Vapor and gas permeability of polyurethane membranes. Part I. Structure-property relationship,” J. Memb. Sci., vol. 49, no. 3, pp. 341–350, 1990.

[20] D. P. Queiroz and M. N. Pinho, “Gas permeability of polypropylene oxide / polybutadiene bi-soft segment urethane / urea membranes,” Science (80-. )., vol. 145, pp. 379–383, 2002.

64

[21] D. P. Queiroz and M. N. De Pinho, “Structural characteristics and gas permeation properties of polydimethylsiloxane/poly(propylene oxide) urethane/urea bi-soft segment membranes,” Polymer (Guildf)., vol. 46, no. 7, pp. 2346–2353, 2005.

[22] M. Faria and M. N. de Pinho, “Phase segregation and gas permeation properties of poly(urethane urea) bi-soft segment membranes,” Eur. Polym. J., vol. 82, pp. 260–276, 2016.

[23] M. Zhou et al., “Blood Platelet’s Behavior on Nanostructured Superhydrophobic Surface,” J. Nano Res., vol. 2, pp. 129–136, 2008.

[24] M. Faria, V. Geraldes, and M. N. De Pinho, “Surface characterization of asymmetric Bi-soft segment poly(ester urethane urea) membranes for blood-oxygenation medical devices,” Int. J. Biomater., vol. 2012, 2012.

[25] M. Faria, P. Brogueira, and M. N. de Pinho, “Sub-micron tailoring of bi-soft segment asymmetric polyurethane membrane surfaces with enhanced hemocompatibility properties,” Colloids Surfaces B Biointerfaces, vol. 86, no. 1, pp. 21–27, 2011.

[26] M. C. Besteiro, A. J. Guiomar, C. A. Gonçalves, V. A. Bairos, M. N. De Pinho, and M. H. Gil, “Characterization and in vitro hemocompatibility of bi-soft segment, polycaprolactone-based poly(ester urethane urea) membranes,” J. Biomed. Mater. Res. - Part A, vol. 93, no. 3, pp. 954–964, 2010.

[27] M. N. de Pinho, “Process of synthesis asymmetric polyurethane based membranes with hemocompatibility characteristics and membranes obtained by said process,” US 2010/0111761 A1, 2010.

[28] G. S. Park, “Transport Principles—Solution, Diffusion and Permeation in Polymer Membranes,” in Synthetic Membranes: Science, Engineering and Applications, P. M. Bungay, H. K. Lonsdale, and de P. M. N., Eds. 1986, pp. 57–107.

[29] M. Mulder, Basic Principles of Membrane Technology. Springer, 1996.

[30] W. A. Rogers, R. S. Buritz, and D. Alpert, “Diffusion coefficient, solubility, and permeability for helium in glass,” J. Appl. Phys., vol. 25, no. 7, pp. 868–875, 1954.

[31] M. Cristina, F. Besteiro, M. Cristina, and F. Besteiro, “Development of poly ( ester urethane urea ) membranes for Extracorporeal Membrane Blood Oxygenators INSTITUTO SUPERIOR TÉCNICO.”

[32] W. N. Rasband, “ImageJ.” .

[33] P. Transducer, “Model 205.”

[34] “Datum 2000 TM,” 2000.

[35] C. Improving and F. C. Worldwide, “Installation Manual for Tube & Precision Instrument Pipe Fittings.”

[36] H. Stability and F. Calibrated, “Intelligent Transmitters Digiquartz ® Pressure Instrumentation Digiquartz ® Pressure Instrumentation.”

[37] A. S. Michaels and H. J. Bixler, “Flow of gases through polyethylene,” J. Polym. Sci., vol. 50, no. 154, pp. 413–439, 1961.

[38] K. Haraya and S. T. Hwang, “Permeation of oxygen, argon and nitrogen through polymer membranes,” J. Memb. Sci., vol. 71, no. 1–2, pp. 13–27, 1992.

[39] K. D. Ziegel, “Gas transport in segmented block copolymers,” J. Macromol. Sci. Part B, vol. 5, no. 1, pp. 11–21, 1971.

[40] Keithley Instruments, “Model 2000 Multimeter User’s Manual,” no. August, p. 256, 2010.

[41] I. Cibulka, “Saturated liquid densities of 1-alkanols from C1 to c10 and n-alkanes from C5 to C16: A critical evaluation of experimental data,” Fluid Phase Equilib., vol. 89, no. 1, pp. 1–18, 1993.

[42] C. Headquarters, “User manual 2100 Temperature Controller,” 2013.

65

Annex – Custom-made set-up calibrations and tests Paroscientific pressure transmitter calibration

The pressure transmitter Paroscientific model 6100A-CE, has a compensation functionality that

fully thermally compensates the output pressure using a quartz crystal temperature signal. However, it

was observed that when the device was exposed atmospheric pressure the measured relative pressure

was different of zero. This was an evidence of the presence of gas in the device’s reference, which

made that the measurements were affected of an off-set.

The pressure transmitter Paros 6100A-CE have been calibrated against an Intelligent

Transmitter Paroscientific, Series 1000 (model1100A-CE, Inc. Washington, USA), connected to a

Paroscientific model 715 display unit, present in the lab and calibrated with a maximum error 0.01%FS.

To calibrate and verify the scale linearity of the Paros 6100A-CE, discrete pressure measurements were

registered by both devices at the same time, under vacuum and room temperature. Figure 42 shows

the values of pressure measured by the two transmitters. Despite the difference between the pressure

values measured by the two transmitters, both devices had a similar behavior to the pressure variation.

show the plot of the model 6100A-CE pressure measurements as function of the model 1100A-

CE pressure measurements. This line proof the linearity of the model 6100-CE scale, even at pressure

below zero, with a maximum deviation to linearity of 0.15 mbar.

The transmitter model 1100-CE reads absolute pressure, unlike model 6100-CE that reads

relative pressure and is affected of an off-set. Adding the atmospheric pressure (1013.25 mbar) and

subtracting the off-set relative to that measurement (51.57 mbar), to every measurement, it was found

that the two pressure transmitters had a maximum difference between measurements of 4.39 mbar.

Figure 42 – Values of pressure (mbar) in an installation during a vacuum test, read by each transmitter: model 1100A-CE (blue dots); model 6100A-CE (orange dots).

-1500

-1000

-500

0

500

1000

1500

0 5 10 15 20 25 30 35P(m

bar

)

measurement

66

Figure 43 — Plot of model 6100A-CE pressure measurements in function of model 1100A-CE pressure measurements.

Set up Volume Calibration

In the pressure method, the quantity of gas passing through the membrane are obtained by

measuring the pressure variation in the set-up receiving chamber of known volume. To identify the set-

up receiving chamber volume, a cylinder has been calibrated (named buffer in Figure 17). Thus, the

cylinder was properly washed and dried in an oven. After 12h, it was filled with Hexadecane

(CH3(CH2)14CH3) 99%, provided by Aldrich, in a water bath at constant temperature of 25ºC. The

temperature was measured by a platinum resistance sensor Keithley 2000 Multimeter, (Keithley

Instruments, Inc. Ohio, U.S.A. [40]). When it was achieved a constant temperature, the cylinder was

closed and the outside was dried using acetone and a drier to certify that all the hexadecane has

evaporated. The full cylinder was then weighted. The cylinder’s inside was re-washed and dried in the

oven, repeating the whole process.

Different density values of hexadecane were compared at bath temperature and were found to

be concordant up to the third decimal place. The temperatures were obtained by converting the

resistance measured by the multimeter, according to the equation provided by device’s constructors.

The density of hexadecane, at operations temperature, was obtained by the equation in Cibulka’s work

[41]. Having the mass of liquid and the hexadecane’s density, the cylinder’s volume was obtained, being

12.6 ± 0.1 cm3.

After the cylinder volume calibration, this was used to calibrate the set-up receiving chamber

volume. The first step to accomplish the calibration was to pressurize the facility with the cylinder

opened. Thereafter, the cylinder was closed and pressure was measured. Then, the installation was

opened for atmospheric pressure. After this, an impermeable membrane was placed in the cell to not

y = 0,999x - 957,089R² = 1,000

-1000

-800

-600

-400

-200

0

-200 0 200 400 600 800 1000 1200

P 6

100A

-CE

(m

bar)

P 1100A-CE (mbar)

67

let the gas flow through the feed section of the set-up. After close de permeation cell and all exit valves,

the cylinder was opened and pressure was measured.

The total volume of the unit corresponds to the sum of the cylinder’s volumes (𝑉1) and the

volume of the remaining unit (𝑉2):

𝑉𝑇𝑜𝑡𝑎𝑙 = 𝑉1 + 𝑉2

(31)

Considering the Ideal Gases Law:

𝑛 =

𝑝𝑉

𝑅𝑇

(32)

Thus, the sum of moles stored in the cylinder (𝑛1) with the mols in the remaining unit (𝑛2), is

equal to the total moles in the set-up (𝑛𝑇𝑜𝑡𝑎𝑙).

𝑛𝑇𝑜𝑡𝑎𝑙 = 𝑛1 + 𝑛2

(33)

Substituting Eq. (32) in Eq. (33), it is obtained the Eq. (34) where 𝑃1 is the pressure read in the

moment of cylinder closure, 𝑃2 is the unit’s pressure at the moment right before the cylinder opening,

and 𝑃 is the unit pressure after cylinder opening.

𝑝𝑉𝑇𝑜𝑡𝑎𝑙

𝑅𝑇=

𝑝1𝑉1

𝑅𝑇+

𝑝2𝑉2

𝑅𝑇

(34)

Considering that temperature is constant during the process.

𝑝𝑉𝑇𝑜𝑡𝑎𝑙 = 𝑝1𝑉1 + 𝑝2𝑉2

(35)

It is considered that 𝑃2 = 𝑃𝑎𝑡𝑚, because the set-up was opened to atmospheric pressure after

the cylinder closuring:

𝑝𝑉𝑇𝑜𝑡𝑎𝑙 = 𝑝1𝑉1 + 𝑝𝑎𝑡𝑚(𝑉𝑇𝑜𝑡𝑎𝑙 − 𝑉1) (36)

Rearranging the Eq. (37) in function of 𝑉𝑇𝑜𝑡𝑎𝑙 it is obtained the total volume of the receiving

chamber by the expression:

𝑉𝑇𝑜𝑡𝑎𝑙 = 𝑉1

𝑝1−𝑝𝑎𝑡𝑚

𝑝−𝑝𝑎𝑡𝑚

(37)

The total volume of the receiving chamber of the set-up was 27.7± 0.1 cm3.

68

Setra pressure transmitter calibration

After testing the pressure transmitter Paroscientific model 6100-CE, and assembling it in the

set-up, the feed side pressure transmitter, Setra Model 205, was tested pressurizing the set-up with N2.

This test consisted on recording thirty discrete pressure measurements of the set-up pressure (without

a membrane sample in the permeation cell), at room temperature, with both sensors and the pressure

reducing valve manometer on the N2 cylinder. The pressure was increased by feeding N2 to the set-up,

and reduced by opening a vent valve. Figure 44 shows the pressure measurements of the Paroscientific

transmitter.

In Figure 44 the pressure measurements of the Setra transmitter and of the reducing valve

manometer are represented as function of Paroscientific transmitter. The linearity of every pressure

scale was verified. Note that the Setra transmitter measures relative pressure and is affected by an off-

set, like the Paroscientific transmitter. Therefore, when an atmosphere is added and the off-set is

subtracted to every pressure measurement of the Setra transmitter, it was verified that the maximum

difference between transmitter was 27.0 mbar.

Figure 44 – Pressure measurements of Paroscientific transmitter during the test.

Figure 45 – Pressure measurements of the Setra transmitter (blue dots) and of the reducing valve

manometer (orange dots) as function of pressure measurements of the Paroscientific transmitter at

room temperature.

0

1000

2000

3000

4000

5000

6000

7000

0 5 10 15 20 25 30

p(m

bar

)

Measurment number

0

1000

2000

3000

4000

5000

6000

7000

8000

0 1000 2000 3000 4000 5000 6000

P(m

bar

)

Pparos(mbar)

69

Air bath temperature

To ensure that the permeation flux measurements were made at constant temperature, the set

up was assembled in a thermostatic air bath (Figure 18). The apparatus of the air bath consisted on a

wine cooler (Candy brand, model CCV 1420GI) as cold source, a thermo-ventilator (Electronia, model

NSB-200A7 of 2000 watts) as heat source, attached to a temperature controller (Hart Scientific, model

2100, Fluke Corp. WA, U.S.A.[42]), located on the bottom of the wine cooler. The controller probe, a

platinum sensor, was allocated next to the heat source, the apparatus components were distributed

along the air bath using the wine cooler shelfs and the permeation flux set-up was allocated to medium

height of the wine cooler. The bath temperature was measured by a platinum resistance, allocated to

the permeation cell, sensor attached to the calibrated Keithley 2000 multimeter. To improve the air

circulation and guarantee a homogeneous temperature inside the air bath, a small fan was assembled

on the upper shelf of air bath.

Figure 46 show the air bath temperature measured on Keithley multimeter for 90 minutes of

continuous operation. It was observed that, periodically, the temperature of the air bath increased 1ºC

in some minutes, returning to the initial temperature after some time. A possible explanation for these

temperature fluctuations is the fact that the wine cooler compressor stops periodically, pausing the cool

source for some minutes, before starting it again. In general, the temperature increases approximately

1 ºC over 10 minutes and then takes approximately 1h to reach the initial temperature.

Figure 46 — Temperature peak measured in air bath (Keithley multimeter)

Figure 47 shows the Paroscientific transmitter measurements of internal temperature and the

pressure in the closed circuit set-up. It was verified that every 90 minutes, a temperature peak occurs,

taking one hour to reach the initial and constant temperature. It should be noted that the internal

291,80

291,90

292,00

292,10

292,20

292,30

292,40

292,50

292,60

0 20 40 60 80 100

T(K

)

t(min)

70

temperature is always late in relation to the pressure measurements. A possible explanation for this is

the low heat conductivity of the thick transmitter walls.

Figure 47 – Internal temperature(ºC) (orange dots) and set-up pressure (mbar) (blue dots) (Paroscientific transmitter)

It is known that temperature variations in gas permeation systems affect the pressure

measurements, particularly in low flux conditions. For this reason, the pressure was only measured

during intervals of time at which the temperature of the system was constant. Figure 48 shows the

temperature measured throughout a 20 minutes time frame, for a set-point temperature of 28.00 ºC,

where the maximum temperature variation inside the air bath was 0.02 ºC.

Figure 48 – Constant phase temperature measurements in air bath (Keithley multimeter).

1158

1159

1160

1161

1162

1163

1164

1165

0 50 100 150 200 250

17,95

18,05

18,15

18,25

18,35

18,45

18,55

18,65

18,75

P(m

bar

)

t(min)

T(ºC

)

292,265

292,270

292,275

292,280

292,285

292,290

292,295

292,300

0 5 10 15 20

T(K

)

t(min)

Polyurethane urea membranes for membrane blood ......Polyurethane urea membranes for membrane blood...

Documents

Transcript of Polyurethane urea membranes for membrane blood ......Polyurethane urea membranes for membrane blood...