CARBON PERGAMON Carbon38(2000)675-689 The surface chemistry and acid-base properties of a paN-based bon fibre Phil e. vickers John F. Watts,* Christian Perruchot. Mohamed M. Chehimib School of Mechanical and Materials Engineering, University of Surrey, Guildford, Surrey GU2 5XH, UK Institut de Topologie et de dynamique des Systemes(ITODYS) Universite Paris 7, Denis Diderot, associe au CNRS (UPRESA 7086), I rue Guy de la brosse, 75005 Paris, france Received 29 March 1999; accepted 25 June 1999 Abstract Akzo Tenax hta carbon fibres(high strength) with progressive amounts of the standard commercial treatment have been investigated using a number of surface analytical methods. Chemical content and structure have been interrogated by XPs and ToF-SIMS, and compared to thermodynamic and acid-base properties probed by IGC and DCAA. The ys values of the fibres were calculated at ca. 50%C and were found to decrease from 104 mJ m- for untreated fibres to 78 mJ m- for 200% oxidised fibres. The acid-base character at the molecular level and the macroscopic properties of y. are significantly enhanced even at low levels of oxidation, resulting in increased wettability of carbon fibres by polar species. Wettability of the surface by non-polar species appears to be largely unaffected by oxidation. XPS analysis reveals that oxygen introduced progressively at the expense of carbon content, and trace amounts of Na and Cl slowly increase with oxidation oF-SIMS shows that the amount of Ca and Na increases very quickly during the early stages of oxidation, whereas Cl and O increase at a more steady rate. 2000 Elsevier Science Ltd. All rights reserved. Keywords: A. Carbon fibres; C. X-ray photoelectron spectroscopy(XPS): D. Thermodynamic properties, Surface properties 1. Introduction surface area as a result of electrochemical oxidation by BET analysis and other workers have also concluded that The interaction of material(e.g. epoxy resin, surface area of the fibre plays little or no role in the amine curing agent) carbon surfaces is of improvement in the interfacial shear strength(IFSS)[1],or paramount importance to the ultimate mechanical prop. the interlaminar shear strength (ILSS)[4]. Zhdan et al, erties of a carbon fibre reinforced polymer(CFRP)[1, 2]. however, detected an increase in surface roughness of This is because it is the ability of the matrix material to high-strength(HS)carbon fibres by atomic force micro- transfer stress to the reinforcing fibre that results in the scopy(AFM) during in situ electrochemical oxidation [5] high strength weight ratio obtained in high-performanc Previous monitoring of the oxidation of carbon fibre composite materials such as CFRPs. In order to avoid surfaces has involved the use of X-ray photoelectron formation of voids at the fibre-matrix interface, complete spectroscopy(XPS) which has been studied in great detail wetting of the fibre surface by the resin material must be in a series of papers by Sherwood et al. [6-11. These achieved. This is generally attained by oxidising the fibre papers observed the change in C Is and o ls peak shapes surface, thus increasing the surface free energy, and as the oxidation progressed, using a wide variety of therefore wettability, of the fibre surface. The actual reason oxidation methods, such as electrochemical oxidation in for the improvement in performance of the composite is aqueous solutions of sulphuric acid or ammonium bicar unclear, although it is consistently observed. Mahy et al. bonate [6,7]. Scrutiny of the C Is and o Is photoelectron 3 observed that there was no significant increase in peaks indicated that carbonyl groups are introduced to the surface at low oxidation potentials, and the concentration *Corresponding author. Tel. +44-1483-259617, fax: +44 of alcohol-ether groups increases at high oxidation po- 1483-876291 tentials. This observation appears to be consistent with E-mail address: j.watts@surrey. ac uk(F. Watts) cyclic voltammograms recorded during the oxidation pro- 0008-6223/00/S-see front matter 2000 Elsevier Science Ltd. All rights reserved PII:S0008-6223(99)00137-2
PERGAMON Carbon 38 (2000) 675–689 The surface chemistry and acid–base properties of a PAN-based carbon fibre a a, b b Phil E. Vickers , John F. Watts , Christian Perruchot , Mohamed M. Chehimi * a School of Mechanical and Materials Engineering, University of Surrey, Guildford, Surrey GU2 5XH, UK b Institut de Topologie et de Dynamique des Systemes `´ ´ (ITODYS), Universite Paris 7, Denis Diderot, associe au CNRS (UPRESA 7086), 1 rue Guy de la Brosse, 75005 Paris, France Received 29 March 1999; accepted 25 June 1999 Abstract Akzo Tenax HTA carbon fibres (high strength) with progressive amounts of the standard commercial treatment have been investigated using a number of surface analytical methods. Chemical content and structure have been interrogated by XPS d and ToF-SIMS, and compared to thermodynamic and acid–base properties probed by IGC and DCAA. The g values of the s 22 22 fibres were calculated at ca. 508C and were found to decrease from 104 mJ m for untreated fibres to 78 mJ m for 200% p oxidised fibres. The acid–base character at the molecular level and the macroscopic properties of g s are significantly enhanced even at low levels of oxidation, resulting in increased wettability of carbon fibres by polar species. Wettability of the surface by non-polar species appears to be largely unaffected by oxidation. XPS analysis reveals that oxygen is introduced progressively at the expense of carbon content, and trace amounts of Na and Cl slowly increase with oxidation. ToF-SIMS shows that the amount of Ca and Na increases very quickly during the early stages of oxidation, whereas Cl and O increase at a more steady rate. 2000 Elsevier Science Ltd. All rights reserved. Keywords: A. Carbon fibres; C. X-ray photoelectron spectroscopy (XPS); D. Thermodynamic properties, Surface properties 1. Introduction surface area as a result of electrochemical oxidation by BET analysis and other workers have also concluded that The interaction of matrix material (e.g. epoxy resin, surface area of the fibre plays little or no role in the amine curing agent) with carbon fibre surfaces is of improvement in the interfacial shear strength (IFSS) [1], or paramount importance to the ultimate mechanical prop- the interlaminar shear strength (ILSS) [4]. Zhdan et al, erties of a carbon fibre reinforced polymer (CFRP) [1,2]. however, detected an increase in surface roughness of This is because it is the ability of the matrix material to high-strength (HS) carbon fibres by atomic force microtransfer stress to the reinforcing fibre that results in the scopy (AFM) during in situ electrochemical oxidation [5]. high strength:weight ratio obtained in high-performance Previous monitoring of the oxidation of carbon fibre composite materials such as CFRP’s. In order to avoid surfaces has involved the use of X-ray photoelectron formation of voids at the fibre–matrix interface, complete spectroscopy (XPS) which has been studied in great detail wetting of the fibre surface by the resin material must be in a series of papers by Sherwood et al. [6–11]. These achieved. This is generally attained by oxidising the fibre papers observed the change in C 1s and O 1s peak shapes surface, thus increasing the surface free energy, and as the oxidation progressed, using a wide variety of therefore wettability, of the fibre surface. The actual reason oxidation methods, such as electrochemical oxidation in for the improvement in performance of the composite is aqueous solutions of sulphuric acid or ammonium bicarunclear, although it is consistently observed. Mahy et al. bonate [6,7]. Scrutiny of the C 1s and O 1s photoelectron [3] observed that there was no significant increase in peaks indicated that carbonyl groups are introduced to the surface at low oxidation potentials, and the concentration *Corresponding author. Tel.: 144-1483-259617; fax: 144- of alcohol–ether groups increases at high oxidation po- 1483-876291. tentials. This observation appears to be consistent with E-mail address: j.watts@surrey.ac.uk (J.F. Watts). cyclic voltammograms recorded during the oxidation pro- 0008-6223/00/$ – see front matter 2000 Elsevier Science Ltd. All rights reserved. PII: S0008-6223(99)00137-2
P.E. Vickers et al. Carbon 38(2000)675-689 edure. In addition to these observations, it was also noted acid-base properties by in surace as chromatography by Proctor and Sherwood [6, 7 that the plasmon satellite (IGC). It was found that a su eatment of 24 Cg feature observed at a 6-7 ev higher binding energy than increase of the surface by the primary peak decreased in intensity as the oxidation of nearly five-fold. Curve-fitting of the C Is spectrum led to the graphite fibre surface progressed. This was probably a the conclusion that the C-o/C-N moiety increased fro result of the disruption of t-bonding orbitals at the fibre 0% to 4%, C=0(carbonyl)from 0% to 9%, and o-C-Oh surface caused by the introduction of oxygen function- (carboxyl) from 2% to 8% of the total carbon present in alities. Nakao et al. [12 also oxidised ex-PAN fibres the surface. This result was also supported by titrimetric Pyrofil XTR40) in both ammonium bicarbonate and analysis. IGC results showed an increase in acidity and phosphoric acid electrolytes and concluded that the equiva- basicity, attributed to the introduction of alcohol and lent phosphoric acid treatment introduced a greater propor- carboxylic acid groups (amphoteric)and carbonyl groups tion of oxygen to the fibre surface, in agreement with(basic). Somewhat surprising was the observation that y Sherwood et al. It was also found that the ILSS of a CFrP increased significantly as the oxidation treatment pro- increased for fibres that had received both types of gressed. At a treatment level of 24 cg, the fibres electrochemical oxidation, but treatment in the phosphori possessed a y. of 140 mJ m, a value greater than acid electrolyte was detrimental to the transverse flexural previously obtained for highly ordered pyrolytic graphite strength of the composite. This was attributed to the(hoPg) by IGc [14 of 129 mJ m. As the untreated formation of fissures on the fibre surface causing a weak fibres are highly graphitic in nature, y s should be expected to decrease as the oxidation progresses, because of disrup- It is important to relate the chemical composition of the tion to the bonding in the aromatic ring structure. Nardin et fibre surface to the surface free energy so that a measure of al. [15 used IGC to measure the surface free energy of the wettability of the fibre surface may be obtained. The Soficar T300 fibres. These fibres are similar to AKzo surface free energy of a solid, y, may be thought of as the HTA 5000 fibres in both surface chemical and physical sum of the dispersive component to the surface free properties. It was found that y s was virtually unaffected energy, y, and the polar or Lewis acid-base component (y:=50 m m by the oxidation treatment employed and to the surface free energy, y P In order for spontaneous that both the acidity and basicity increased as a result of etting by a liquid to occur, y must exceed the surface oxidation. The change in surface free energy monitored by tension,y, of the liquid in question. It therefore follows, DCAA has also been correlated with data from XPS that. in order for complete wetting of the carbon fibl experiments by Wu et al. [16] for Hercules As(oxidised) surface by a resin to occur, it would be advantageous to and aU (unoxidised) HS fibres. It was found that y produce a fibre with a high surface energy. However, it is decreased slightly as a result of the standard commercial well known that graphite has a much higher surface energy treatment and this was accompanied by a very large (120 mJ m )than water (72. 8 mJ m ) but is not wet increase in yP, as might be expected, leading to an at all well by water. Therefore, instead of the comparison increase in y of slightly less than a factor of two. It was of the surface free energies of a solid and a liquid, it is also found that heating the fibres in a vacuum to tempera more accurate to relate wettability to the surface free tures as high as 1000C led to a significant decrease in y energy of the interface, y. The latter must be minimised presumably because of degradation of the polar surface in order to improve wettability, for example, via strong groups introduced as a result of electrochemical oxidation. acid-base interactions at the solid-liquid (i.e. fibre-resin) In this paper we compare acid-base properties of fibres interface. This is achieved by introducing polar oxygen (AKZO's Tenax hta high-strength fibres) that have containing functionalities which may react chemically to received varying degrees of a manufacturers standard form a covalent bond at the fibre surface. The introduction eectrochemical oxidation, measured by both IGC and of polar groups to the surface will enhance the probability DCAA, with the elemental composition of the fibre surface of a Lewis acid-base interaction between the two. Bradley measured by XPs. IGC brings thermodynamic information [13] compared XPS and dynamic contact angle analysis at the molecular scale whereas dCaa probes the wettabili- (DCAA)results of oxidised and unoxidised fibres(akzo ty of material surfaces at the macroscopic scale. We thus Tenax HTA), and found that the oxygen content of the expect complementary information to be obtained from the fibre surface was increased by a factor of 2. Surprisingly, two methods. Another technique that may also be used to the DCAA results indicate that y i decreases after the elucidate chemical structure of the fibre surface is time-of- oxidation treatment. This effect was attributed to a thin flight secondary ion mass spectrometry (ToF-SIMS).Al- oxygen-rich layer thought to be present on the surface of though this is a more recent development than XPS, it has untreated fibres that is removed during the initial stages of rarely been used for the characterisation of unmodified anodic oxidation. Mahy et al. b3 electrochemically oxi- carbon fibre surfaces. This is because sample preparation is dined AKZO Tenax HTA 5000 fibres in an ammonium not trivial as a result of field emission occurring from stray sulphate electrolyte, and analysed the chemical composi- fibre ends. In addition to this, the highly aromatic nature of tion of the fibre surface by XPS and investigated the the fibre surface leads to relatively simple spectra of
676 P.E. Vickers et al. / Carbon 38 (2000) 675 –689 cedure. In addition to these observations, it was also noted acid–base properties by inverse gas chromatography21 by Proctor and Sherwood [6,7] that the plasmon satellite (IGC). It was found that a surface treatment of 24 C g feature observed at a 6–7 eV higher binding energy than increased the oxygen concentration of the surface by the primary peak decreased in intensity as the oxidation of nearly five-fold. Curve-fitting of the C 1s spectrum led to the graphite fibre surface progressed. This was probably a the conclusion that the C–O/C–N moiety increased from result of the disruption of p-bonding orbitals at the fibre 0% to 4%, C=O (carbonyl) from 0% to 9%, and O=C–OH surface caused by the introduction of oxygen function- (carboxyl) from 2% to 8% of the total carbon present in alities. Nakao et al. [12] also oxidised ex-PAN fibres the surface. This result was also supported by titrimetric (Pyrofil XTR40) in both ammonium bicarbonate and analysis. IGC results showed an increase in acidity and phosphoric acid electrolytes and concluded that the equiva- basicity, attributed to the introduction of alcohol and lent phosphoric acid treatment introduced a greater propor- carboxylic acid groups (amphoteric) and carbonyl groupsd tion of oxygen to the fibre surface, in agreement with (basic). Somewhat surprising was the observation that g s Sherwood et al. It was also found that the ILSS of a CFRP increased significantly as the oxidation treatment pro- 21 increased for fibres that had received both types of gressed. At a treatment level of 24 C g , the fibres d 22 electrochemical oxidation, but treatment in the phosphoric possessed a g s of 140 mJ m , a value greater than acid electrolyte was detrimental to the transverse flexural previously obtained for highly ordered pyrolytic graphite 22 strength of the composite. This was attributed to the (HOPG) by IGC [14] of 129 mJ m . As the untreated d formation of fissures on the fibre surface causing a weak fibres are highly graphitic in nature, g s should be expected boundary layer. to decrease as the oxidation progresses, because of disrupIt is important to relate the chemical composition of the tion to the bonding in the aromatic ring structure. Nardin et fibre surface to the surface free energy so that a measure of al. [15] used IGC to measure the surface free energy of the wettability of the fibre surface may be obtained. The Soficar T300 fibres. These fibres are similar to AKZO surface free energy of a solid, g , may be thought of as the HTA 5000 fibres in both surface chemical and physical s d sum of the dispersive component to the surface free properties. It was found that g s was virtually unaffected d d 22 energy, g s, and the polar or Lewis acid–base component (g s¯50 mJ m ) by the oxidation treatment employed and p to the surface free energy, g . In order for spontaneous that both the acidity and basicity increased as a result of s wetting by a liquid to occur, gs must exceed the surface oxidation. The change in surface free energy monitored by tension, gl , of the liquid in question. It therefore follows, DCAA has also been correlated with data from XPS that, in order for complete wetting of the carbon fibre experiments by Wu et al. [16] for Hercules AS (oxidised)d surface by a resin to occur, it would be advantageous to and AU (unoxidised) HS fibres. It was found that g s produce a fibre with a high surface energy. However, it is decreased slightly as a result of the standard commercial well known that graphite has a much higher surface energy treatment and this was accompanied by a very large 22 22 p (.120 mJ m ) than water (72.8 mJ m ) but is not wet increase in g s, as might be expected, leading to an at all well by water. Therefore, instead of the comparison increase in g of slightly less than a factor of two. It was s of the surface free energies of a solid and a liquid, it is also found that heating the fibres in a vacuum to tempera-p more accurate to relate wettability to the surface free tures as high as 10008C led to a significant decrease in g , s energy of the interface, g . The latter must be minimised presumably because of degradation of the polar surface sl in order to improve wettability, for example, via strong groups introduced as a result of electrochemical oxidation. acid–base interactions at the solid–liquid (i.e. fibre–resin) In this paper we compare acid–base properties of fibres interface. This is achieved by introducing polar oxygen (AKZO’s Tenax HTA high-strength fibres) that have containing functionalities which may react chemically to received varying degrees of a manufacturers standard form a covalent bond at the fibre surface. The introduction electrochemical oxidation, measured by both IGC and of polar groups to the surface will enhance the probability DCAA, with the elemental composition of the fibre surface of a Lewis acid–base interaction between the two. Bradley measured by XPS. IGC brings thermodynamic information [13] compared XPS and dynamic contact angle analysis at the molecular scale whereas DCAA probes the wettabili- (DCAA) results of oxidised and unoxidised fibres (AKZO ty of material surfaces at the macroscopic scale. We thus Tenax HTA), and found that the oxygen content of the expect complementary information to be obtained from the fibre surface was increased by a factor of 2. Surprisingly, two methods. Another technique that may also be used to p the DCAA results indicate that g decreases after the elucidate chemical structure of the fibre surface is time-of- s oxidation treatment. This effect was attributed to a thin flight secondary ion mass spectrometry (ToF-SIMS). Aloxygen-rich layer thought to be present on the surface of though this is a more recent development than XPS, it has untreated fibres that is removed during the initial stages of rarely been used for the characterisation of unmodified anodic oxidation. Mahy et al. [3] electrochemically oxi- carbon fibre surfaces. This is because sample preparation is dised AKZO Tenax HTA 5000 fibres in an ammonium not trivial as a result of field emission occurring from stray sulphate electrolyte, and analysed the chemical composi- fibre ends. In addition to this, the highly aromatic nature of tion of the fibre surface by XPS and investigated the the fibre surface leads to relatively simple spectra of
P.E. Vickers et al. Carbon 38(2000)675-689 aromatic fragment ions, where the fragments observed tend of 30 ns, and a beam current of 1.0 to be very similar to those observed from organic surfaces ion dose of approxi 4×10 I-analysis that have been damaged by ion doses that are beyond the SIMS spectra were over a ange of m/- of generally accepted static limit of 10 ons cm as a 5-800 in both the positive and negative ion modes. The result of this situation, ToF-SIMS analysis is more readily spectrometer was controlled by a vG Scientific data system applied to the study of the interaction of organic molecules based on a dEC PDP11/73 running the uRSX operating such as epoxy resins(used in the size and matrix material system. Sample preparation has been fully described in CFRP's)adsorbed on carbon fibres [17]. However, the elsewhere [17] surface specificity of SIMs(1-2 nm)is greater than that of KPS (2-5 nm) and its high sensitivity to chemical struc- 24. GC ture make it ideal for the identification of contaminants present on the surface, where its excellent detection limit A Girdel 330 gas chromatograph utilising a fame may be exploited [18, 19 ionisation detector was employed to carry out the IGC characterisation of untreated and electrochemically treated carbon fibres. Teflon columns, with an internal diameter of 2. Experiment 6.25 mm and length of 500 mm were filled with tows of carbon fibres. Three tows of 1000 mm of fibre were pulled 2.. Fibre treatment through the column from the middle of the lengths using a wire drawstring, which led to the column cross-section AKZO Tenax HTA 5000 fibres were obtained that had containing approximately 70 000 fibres along the column been subjected to 0, 25, 50, 100 and 200% of the standard corresponding to a sample weight of approximately 3.6 g (proprietary) manufacturers electrochemical oxidation A slight torque was applied to the tows during the packing procedure, described as the degree of fibre treatment procedure, which helped reduce the number of voids in the DFT). Sample preparation of the fibres for each technique column. The ends of the column were plugged with a small is described below amount of clean glass wool to ensure an even gas flow through the column. 2.2. XPS The carrier gas used was high-purity nitrogen supplied by Air Liquide, and methane (Air Liquide), was used as The system used for all XPs analysis was a VG the non-interacting marker, in order to measure the resist Scientific ESCALAB Mkll spectrometer interfaced to a ance to flow of the column. The flow-rate, measured by PDPll/73 minicomputer utilising DEC HRSX software means of a soap-bubble flowmeter, was approximately 25 AlKa(hv=1486.6 ev)was used in all analyses, at a power cm min in most cases. although it was reduced to of 10 kv at 34 mA. two fibre tows of 30 mm length approximately 15 cm min for the acidic probes as their placed in a specially designed stub ensuring that no signal low retention time meant that the probe signal is observed from the sample holder. Survey scans were resolved from that of the methane marker. The acquired with a pass energy of 50 ev, and the high- temperature was set at 50C, and the injector and resolution spectra with a pass energy of 20 eV. Wagner were both held at 110C. Prior to analysis, the columns sensitivity factors [20] were used for the quantification were conditioned at 120C under a flow of 25 cm min procedures, and, for the C Is peak, a Shirley type of nitrogen for 15 h background was subtracted prior to quantification, because The probes were a range of alkanes, acidic and basic of the increase in the background at molecules. The characteristics of the probes are reported in greater that the energy characteristic of the C Is peak at Table 1. The probes were injected manually with a 284.6 ev. The instrument was operated analyser methane marker by means of a Hamilton gas-tight syringe chamber pressure of 10 mbar or less and spectra were The chromatographs were recorded using a Delsi 21 digital acquired over a period of approximately 45 min recorder, and net retention times calculated by measuring the distance from the non-interacting marker to the centre 23. ToF-SIMS of the peak of the probe. This procedure was deemed satisfactory because there was very little asymmetry ToF-SIMS analysis was carried out using a VG Sci- observed in the peaks ntific Type 23 System equipped with a two-stage reflec- tron type analyser and an Mig300PB pulsed liquid metal 2.5. DCAA ion source, and the spectrometer was run at an operating pressure of 10 mbar. Static SIMS conditions were used Contact angles were measured by a Cahn 322 DCAA with a pulsed 5 kHz, 26 keV, Ga primary ion beam system utilising a 16-speed motor, capable of rastered over a frame area of 0. 4X0.4 mm- at 50 between 19.8 and 264 ums, and a microbalance cycles s. The system was operated using a pulse width resolution of 0. I ug. Fibres were analysed using a
P.E. Vickers et al. / Carbon 38 (2000) 675 –689 677 aromatic fragment ions, where the fragments observed tend of 30 ns, and a beam current of 1.0 nA was used giving an 11 22 21 to be very similar to those observed from organic surfaces ion dose of approximately 4310 ions cm analysis . that have been damaged by ion doses that are beyond the SIMS spectra were acquired over a mass range of m/z of 13 22 generally accepted static limit of 10 ions cm . As a 5–800 in both the positive and negative ion modes. The result of this situation, ToF-SIMS analysis is more readily spectrometer was controlled by a VG Scientific data system applied to the study of the interaction of organic molecules based on a DEC PDP11/73 running the mRSX operating such as epoxy resins (used in the size and matrix material system. Sample preparation has been fully described in CFRP’s) adsorbed on carbon fibres [17]. However, the elsewhere [17]. surface specificity of SIMS (1–2 nm) is greater than that of XPS (2–5 nm) and its high sensitivity to chemical struc- 2.4. IGC ture make it ideal for the identification of contaminants present on the surface, where its excellent detection limit A Girdel 330 gas chromatograph utilising a flame may be exploited [18,19]. ionisation detector was employed to carry out the IGC characterisation of untreated and electrochemically treated carbon fibres. Teflon columns, with an internal diameter of 2. Experimental 6.25 mm and length of 500 mm were filled with tows of carbon fibres. Three tows of 1000 mm of fibre were pulled 2.1. Fibre treatment through the column from the middle of the lengths using a wire drawstring, which led to the column cross-section AKZO Tenax HTA 5000 fibres were obtained that had containing approximately 70 000 fibres along the column, been subjected to 0, 25, 50, 100 and 200% of the standard corresponding to a sample weight of approximately 3.6 g. (proprietary) manufacturer’s electrochemical oxidation A slight torque was applied to the tows during the packing procedure, described as the degree of fibre treatment procedure, which helped reduce the number of voids in the (DFT). Sample preparation of the fibres for each technique column. The ends of the column were plugged with a small is described below. amount of clean glass wool to ensure an even gas flow through the column. 2.2. XPS The carrier gas used was high-purity nitrogen supplied by Air Liquide, and methane (Air Liquide), was used as The system used for all XPS analysis was a VG the non-interacting marker, in order to measure the resistScientific ESCALAB MkII spectrometer interfaced to a ance to flow of the column. The flow-rate, measured by PDP11/73 minicomputer utilising DEC mRSX software. means of a soap-bubble flowmeter, was approximately 25 3 21 AlKa (hn51486.6 eV) was used in all analyses, at a power cm min in most cases, although it was reduced to 3 21 of 10 kV at 34 mA. Two fibre tows of 30 mm length are approximately 15 cm min for the acidic probes as their placed in a specially designed stub ensuring that no signal low retention time meant that the probe signal was not is observed from the sample holder. Survey scans were resolved from that of the methane marker. The column acquired with a pass energy of 50 eV, and the high- temperature was set at 508C, and the injector and detector resolution spectra with a pass energy of 20 eV. Wagner were both held at 1108C. Prior to analysis, the columns 3 21 sensitivity factors [20] were used for the quantification were conditioned at 1208C under a flow of 25 cm min procedures, and, for the C 1s peak, a Shirley type of nitrogen for 15 h. background was subtracted prior to quantification, because The probes were a range of alkanes, acidic and basic of the increase in the background at binding energies molecules. The characteristics of the probes are reported in greater that the energy characteristic of the C 1s peak at Table 1. The probes were injected manually with a 284.6 eV. The instrument was operated with an analyser methane marker by means of a Hamilton gas-tight syringe. 29 chamber pressure of 10 mbar or less and spectra were The chromatographs were recorded using a Delsi 21 digital acquired over a period of approximately 45 min. recorder, and net retention times calculated by measuring the distance from the non-interacting marker to the centre 2.3. ToF-SIMS of the peak of the probe. This procedure was deemed satisfactory because there was very little asymmetry ToF-SIMS analysis was carried out using a VG Sci- observed in the peaks. entific Type 23 System equipped with a two-stage reflectron type analyser and an MIG300PB pulsed liquid metal 2.5. DCAA ion source, and the spectrometer was run at an operating 29 pressure of 10 mbar. Static SIMS conditions were used Contact angles were measured by a Cahn 322 DCAA 69 1 with a pulsed 5 kHz, 26 keV, Ga primary ion beam system utilising a 16-speed motor, capable of speeds 2 21 rastered over a frame area of 0.430.4 mm at 50 between 19.8 and 264 mm s , and a microbalance with a 21 cycles s . The system was operated using a pulse width resolution of 0.1 mg. Fibres were analysed using a motor
P.E. Vickers et al./ Carbon 38(2000)675-689 Table I 3. Results Physico-chemical properties of probes used for IGC abbreviation bp"C-1 3.1. Surface chemical composition 8.4 3.1.I. Electron spectroscopy Survey spectra of fibres that had received treatments of n-Nonane 0, 25 and 200% are presented in Fig. 1(a-c). There is very TMP little difference observed between any of the spectra Tert-butyl alcohol t-BuOH recorded, save for the increase in magnitude of theo Is Carbon tetrachloride fiac, peak observed at =530 eV and the O KLL X-ray induced Auger peak at 980 ev. The surface chemical com- positions present within the fibre as measured by xPs are Tetrahydrofuran shown in Table 3. It is assumed that the surface com- Isooctane is 2, 2, 4-trimethyl pentane position does not vary significantly along the tow of fibres, and that fibre heterogeneity is present but at a much smaller scale than the analysis area probed by the X-ray beam. It can be seen that the oxygen content gradually speed of 19.8 um s, giving the greatest number of mass increases at the expense of carbon as the oxidation eadings per unit length of fibre, because mass readings are procedure progresses. The residual nitrogen content fuc- taken at a rate of 1 Hz, irrespective of the speed of th tuates slightly around 3%, and there are traces of Na, Cl motor. Samples were conditioned in an air oven at 160C and Ca. These latter three elements appear to be slightly overnight to remove adsorbed water from the surface of increasing with treatment and are probably introduced to the fibres. A single fibre, approximately 10 mm in length, the surface as a direct result of electrochemical oxidation. was attached to a small piece of adhesive tape, approxi their origin most probably the wash water used in the mately 2X2 mm, that had already been attached to a piece dustrial process, and possibly, in the case of sodium(and of nichrome wire, approximately 60 mm in length, and less likely, chlorine), the electrolyte. The high-resolution a hook at the far end. This was hooked onto the spectra (not shown), show little difference in the C Is sample pan of the microbalance. The hook, tape and fibre envelope for all types of fibre, and the o Is peak reveals ssessed a mass in the range of 4-6 mg, and this was an overall increase in intensity, possibly indicating an balanced to within +2 mg by adding tare weights to the introduction of more carbonyl groups at the fibre surface. other side of the microbalance. Once these were in position The N Is spectra have poor counting statistics, and are the liquid surface was brought to within 5 mm of the fibre centred at approximately 401 ev. Genera ally, the xP end by raising a platform situated underneath the microbal- spectra obtained do not possess sufficient resolution to ance. The fibres were then subjected to one cycle of determine the relative amounts of hydroxyl-ester, car- immersion, i.e. lowered and raised by 10 mm, taking a bonyl-carboxyl, and acid sites on the fibre surface total analysis time of 1020 s. Cyclohexane was used to determine the average fibre perimeter of each type of fibre, 3.1.2. Mass spectrometry and the dispersive and polar contributions to the surface ToF-SIMS has greater surface specificity and better free energies of the fibres were estimated by conducting detection limits than those possible by XPS, and should DCAA experiments using the contact media listed in Table show differences in spectra for different treated fibres 2. 1-Bromonaphthalene(98%), n-hexadecane(99%)and owing to its sensitivity to changes in chemical structure glycerol(99.5%), supplied by Aldrich, were used in the occurring at the surface. Surprisingly though, the spectra DCA analyses obtained were remarkably similar for 0, 25 and 100% treated fibres. However, the spectra revealed a significant increase in inorganic ions with surface treatment. SIMS is very sensitive to inorganic ions which possess low ionia tion energies such as Na, cl and Ca. The increase in these Surface free energy contributions of contact media used for elements is illustrated in Fig. 2, which shows the variation DCAA of the intensity of ions normalised to the total ion signal ontact medium y(kJ mol )y(]J mol)yP(]J mol -) with oxidation treatment. As would be expected from the XPS results, the oxygen content gradually increases and the relative intensity of the CH ions decrease slightly Both Na and Ca show a marked increase at early stages of 26.4 treatment and the Cl content increases steadily. It is interesting to note that there is very little change in the
678 P.E. Vickers et al. / Carbon 38 (2000) 675 –689 Table 1 3. Results Physico-chemical properties of probes used for IGC Probe Abbreviation bp 8C21 3.1. Surface chemical composition n-Hexane C 69 6 3.1.1. Electron spectroscopy n-Heptane C 98.4 7 Survey spectra of fibres that had received treatments of n-Octane C 125.7 8 0, 25 and 200% are presented in Fig. 1(a–c). There is very n-Nonane C 150.8 9 a little difference observed between any of the spectra Isooctane TMP 99.2 Tert-butyl alcohol t-BuOH 83.0 recorded, save for the increase in magnitude of the O 1s Carbon tetrachloride CCl 76.8 4 peak observed at ¯530 eV and the O KLL X-ray induced Chloroform CHCl 61.2 3 Auger peak at ¯980 eV. The surface chemical comEthyl acetate EtAc 77.1 positions present within the fibre as measured by XPS are Tetrahydrofuran THF 66.0 shown in Table 3. It is assumed that the surface com- a Isooctane is 2,2,4-trimethyl pentane. position does not vary significantly along the tow of fibres, and that fibre heterogeneity is present but at a much smaller scale than the analysis area probed by the X-ray beam. It can be seen that the oxygen content gradually 21 speed of 19.8 mm s , giving the greatest number of mass increases at the expense of carbon as the oxidation readings per unit length of fibre, because mass readings are procedure progresses. The residual nitrogen content fluctaken at a rate of 1 Hz, irrespective of the speed of the tuates slightly around 3%, and there are traces of Na, Cl motor. Samples were conditioned in an air oven at 1608C and Ca. These latter three elements appear to be slightly overnight to remove adsorbed water from the surface of increasing with treatment and are probably introduced to the fibres. A single fibre, approximately 10 mm in length, the surface as a direct result of electrochemical oxidation, was attached to a small piece of adhesive tape, approxi- their origin most probably the wash water used in the mately 232 mm, that had already been attached to a piece industrial process, and possibly, in the case of sodium (and of nichrome wire, approximately 60 mm in length, and less likely, chlorine), the electrolyte. The high-resolution possessing a hook at the far end. This was hooked onto the spectra (not shown), show little difference in the C 1s sample pan of the microbalance. The hook, tape and fibre envelope for all types of fibre, and the O 1s peak reveals possessed a mass in the range of 4–6 mg, and this was an overall increase in intensity, possibly indicating an balanced to within 62 mg by adding tare weights to the introduction of more carbonyl groups at the fibre surface. other side of the microbalance. Once these were in position The N 1s spectra have poor counting statistics, and are the liquid surface was brought to within 5 mm of the fibre centred at approximately 401 eV. Generally, the XPS end by raising a platform situated underneath the microbal- spectra obtained do not possess sufficient resolution to ance. The fibres were then subjected to one cycle of determine the relative amounts of hydroxyl–ester, carimmersion, i.e. lowered and raised by 10 mm, taking a bonyl–carboxyl, and acid sites on the fibre surface. total analysis time of 1020 s. Cyclohexane was used to determine the average fibre perimeter of each type of fibre, 3.1.2. Mass spectrometry and the dispersive and polar contributions to the surface ToF-SIMS has greater surface specificity and better free energies of the fibres were estimated by conducting detection limits than those possible by XPS, and should DCAA experiments using the contact media listed in Table show differences in spectra for different treated fibres 2. 1-Bromonaphthalene (98%), n-hexadecane (99%) and owing to its sensitivity to changes in chemical structure glycerol (99.5%), supplied by Aldrich, were used in the occurring at the surface. Surprisingly though, the spectra DCA analyses. obtained were remarkably similar for 0, 25 and 100% treated fibres. However, the spectra revealed a significant increase in inorganic ions with surface treatment. SIMS is very sensitive to inorganic ions which possess low ionisation energies such as Na, Cl and Ca. The increase in these Table 2 Surface free energy contributions of contact media used for elements is illustrated in Fig. 2, which shows the variation DCAA of the intensity of ions normalised to the total ion signal 21 d 21 p 21 with oxidation treatment. As would be expected from the Contact medium g (kJ mol ) g (kJ mol ) g (kJ mol ) sss XPS results, the oxygen content gradually increases and C H 27.6 27.6 0 2 16 34 the relative intensity of the CH ions decrease slightly. C H Br 44.6 44.6 0 10 7 Both Na and Ca show a marked increase at early stages of Glycerol 63.4 37.0 26.4 treatment and the Cl content increases steadily. It is H O 72.8 21.8 51.0 2 interesting to note that there is very little change in the
P.E. Vickers et al. Carbon 38(2000)675-689 URVEY 120 Binding Energy /ev OX FIBRES OUTGASSED 122C 17HR 0.m006010 Binding Energy /eV or TREATED FIERES OUTGASSED 122C 185H 000 n 0 200 600 1000 1200 Binding Energy /ev TREATED FIBRES ot Fig. L. XPS survey scans of:(a)0%,(b)25%; and(c)200% treated fibres
P.E. Vickers et al. / Carbon 38 (2000) 675 –689 679 Fig. 1. XPS survey scans of: (a) 0%; (b) 25%; and (c) 200% treated fibres
P.E. Vickers et al. Carbon 38(2000)675-689 ic(%)composition of the fibre surface measured by XPS after fibres had been dried in an air oven at 122C overnight Atomic(%)concentration Elemental ratios W min nIs Na Is O/C N+O)/C 000 00000 0.l1 13 0.8 elative intensity of the OH ions, which would be appearing in bold. These even-mass ions generally de- expected if a significant number of hydroxyl or carboxylic crease in relative intensity from the untreated fibre surface ing introduced at the after the early stages of oxidation, with the ot be observed if th the ion assigned with the composition of C, H, NO, which present as phenol groups or attached to aromatic systems, appears unaffected by oxidation. The general observation owing to the increased stability to fragmentation. is consistent with the XPs data, which shows a slight A feature that is generally observed in the positive decrease in surface nitrogen concentration, and is also in ToF-SIMS spectra is that the vast majority of ions present agreement with the findings of Alexander and Jones [18] are even-electron ions. This phenomenon arises firstly who observed a change in peak shape in the n Is signal because odd-electron ions are radicals and as a result more They assigned the nitrogen present in the untreated fibre to prone to fragmentation to an even-electron ion and aa protonated amine structure and proposed that the oxi- neutral fragment (radical), and secondly because the dined fibres possessed nitrogen atoms in a configuration method of secondary ion production(sputtering) produces ions that tend to have low internal energies(and therefore decrease in initial nitrogen content and change in chem Bht unlikely to be radicals). This can prove useful in assigning environment is real, then this suggests that nitrogen is organic secondary ions, because the only trivalent element incorporated into the outer layers of the fibre during ommonly found in organic systems is nitrogen, and as a manufacture, where the fibres are subjected to high tem- result, even-electron ions containing an odd number of peratures in inert(N2, Ar)atmospheres. This outer layer is nitrogen atoms will appear at an even mass: charge ratio, then probably removed during the initial stages of oxida- generally most useful for positive ions of low-moderate tion to reveal the residual nitrogen left over from the pan mass precursor. intensity with treatment for a number of even-mass ions in XPS and ToF-SIms both indicate that the surface the positive spectrum, and possible ion assignments are composition of the fibre is modified as a result of the given in Table 4, with the most plausible ion compositions elec ical oxidation process. However, they do not §252出一若 ☆一OH =:°·Ca Fig. 2. Variation of simple ions, Na, Ca, CH, O, OH and Cl, with fibre treatment
680 P.E. Vickers et al. / Carbon 38 (2000) 675 –689 Table 3 Atomic (%) composition of the fibre surface measured by XPS after fibres had been dried in an air oven at 1228C overnight Fibre Atomic (%) concentration Elemental ratios 21 W min g C 1s O 1s N 1s Cl 2p Na 1s Ca 2p O/C (N1O)/C 0 91.7 3.6 3.5 0.3 0.3 0.5 0.04 0.08 25 88.2 7.7 2.7 0.3 0.6 0.4 0.09 0.12 50 86.9 9.7 2.0 0 0.9 0.5 0.11 0.14 100 86.6 9.0 2.5 0.2 1.0 0.6 0.10 0.13 200 83.5 11.6 3.0 0.3 0.9 0.8 0.14 0.17 2 relative intensity of the OH ions, which would be appearing in bold. These even-mass ions generally deexpected if a significant number of hydroxyl or carboxylic crease in relative intensity from the untreated fibre surface acid groups were being introduced at the surface. How- after the early stages of oxidation, with the exception of 1 ever, this may not be observed if the oxygen atoms are the ion assigned with the composition of C H NO , which 2 2 present as phenol groups or attached to aromatic systems, appears unaffected by oxidation. The general observation owing to the increased stability to fragmentation. is consistent with the XPS data, which shows a slight A feature that is generally observed in the positive decrease in surface nitrogen concentration, and is also in ToF-SIMS spectra is that the vast majority of ions present agreement with the findings of Alexander and Jones [18] are even-electron ions. This phenomenon arises firstly who observed a change in peak shape in the N 1s signal. because odd-electron ions are radicals and as a result more They assigned the nitrogen present in the untreated fibre to prone to fragmentation to an even-electron ion and a a protonated amine structure and proposed that the oxineutral fragment (radical), and secondly because the dised fibres possessed nitrogen atoms in a configuration method of secondary ion production (sputtering) produces more similar to a tertiary aromatic nitrogen. If the slight ions that tend to have low internal energies (and therefore decrease in initial nitrogen content and change in chemical unlikely to be radicals). This can prove useful in assigning environment is real, then this suggests that nitrogen is organic secondary ions, because the only trivalent element incorporated into the outer layers of the fibre during commonly found in organic systems is nitrogen, and as a manufacture, where the fibres are subjected to high temresult, even-electron ions containing an odd number of peratures in inert (N , Ar) atmospheres. This outer layer is 2 nitrogen atoms will appear at an even mass:charge ratio, then probably removed during the initial stages of oxidagenerally most useful for positive ions of low-moderate tion to reveal the residual nitrogen left over from the PAN mass. Fig. 3 shows the variation in normalised ion precursor. intensity with treatment for a number of even-mass ions in XPS and ToF-SIMS both indicate that the surface the positive spectrum, and possible ion assignments are composition of the fibre is modified as a result of the given in Table 4, with the most plausible ion compositions electrochemical oxidation process. However, they do not 1 1 22 2 2 Fig. 2. Variation of simple ions, Na , Ca , CH , O , OH and Cl , with fibre treatment
P.E. Vickers et al. Carbon 38(2000)675-689 一CH2NH2 日一C2H4N o00 0 Percentage treatment Fig. 3. Variation of even mass positive ions with surface treatment. offer an adequate explanation for the improvement in required to sweep out an adsorbed species is the net mechanical properties observed in CFRP materials, which retention vol may result from changes in the thermodynamic properties of the surface. An investigation into these aspects of the N=JFCIN fibre surface by IGC and dCaa was therefore undertaker Where is the compression factor, Fc the corrected flow 3. 2. Surface thermodynamics properties by means of rate, and IN the net retention time. The value of IN will therefore be large for a strongly adsorbed species. The free energy of adsorption for a given probe on the surface is related to N by Inverse gas chromatography has long been used to △G=Rrhn(N)+C determine both ys, the dispersive contribution to the surface free energy, and acidity and basicity constants for where R is the gas constant and T the working tempera- polymers [21-23], fillers [24, 25] and fibres [21]. In IGC ture. the term inverse means that the stationary phase of the The constant C is a term that takes the weight and chromatographic column is of interest in contrast to specific surface area of the stationary phase, and the onventional GC. Molecular probes of known physico- standard state of the probes in the gaseous and adsorbed hemical properties are passed through a column packed state into account [26 with the material of interest in an inert carrier gas such as The determination of AG. or RT In( V) values for all N2, at such a low concentration that they behave in- material-probe pairs will permit an estimation of y s and dependently of each other. This is known as IGC at infinite acid-base constants of the materials under test. y. and dilution or zero coverage. This approach permits the acid-base constants will be assessed using n-alkanes and characterisation of the surface or bulk properties of the specific probes, respectively. For all the probes used packing material under test. The volume of carrier gas RT In(N) values are reported in Table 5 Table 4 lon assignments for even mass ions detected on the fibre surface by ToF-SIMS C,HN. C, H, O C,H CrH, NO, C3HN, C3 H,O, C,H&, C,H,N2 CH C,HNO CHN CsH.O C, H.O. CH. C,h,N,o C,HANO, C3H,NO,, CsHIN, C H,O2, CsH1oO, C4H1oN2, CHSN,O
P.E. Vickers et al. / Carbon 38 (2000) 675 –689 681 Fig. 3. Variation of even mass positive ions with surface treatment. offer an adequate explanation for the improvement in required to sweep out an adsorbed species is the net mechanical properties observed in CFRP materials, which retention volume, V . N may result from changes in the thermodynamic properties of the surface. An investigation into these aspects of the V 5 jF t (1) N CN fibre surface by IGC and DCAA was therefore undertaken. Where j is the compression factor, F the corrected flow- C rate, and t the net retention time. The value of V will N N 3.2. Surface thermodynamics properties by means of therefore be large for a strongly adsorbed species. The free IGC energy of adsorption for a given probe on the surface is related to V by: N 3.2.1. Behaviour of the probes Inverse gas chromatography has long been used to 2 DG 5 RT ln(V ) 1 C (2) a N d determine both g , the dispersive contribution to the s surface free energy, and acidity and basicity constants for where R is the gas constant and T the working temperapolymers [21–23], fillers [24,25] and fibres [21]. In IGC ture. the term ‘inverse’ means that the stationary phase of the The constant C is a term that takes the weight and chromatographic column is of interest in contrast to specific surface area of the stationary phase, and the conventional GC. Molecular probes of known physico- standard state of the probes in the gaseous and adsorbed chemical properties are passed through a column packed state into account [26]. with the material of interest in an inert carrier gas such as The determination of DG or RT ln(V ) values for all a N d N , at such a low concentration that they behave in- material-probe pairs will permit an estimation of g s and 2 d dependently of each other. This is known as IGC at infinite acid–base constants of the materials under test. g s and dilution or zero coverage. This approach permits the acid–base constants will be assessed using n-alkanes and characterisation of the surface or bulk properties of the specific probes, respectively. For all the probes used, packing material under test. The volume of carrier gas RT ln(V ) values are reported in Table 5. N Table 4 Ion assignments for even mass ions detected on the fibre surface by ToF-SIMS m/z Possible composition 42 CHN 2 4 , C H O, C H 22 36 56 C H NO, C H N 22 36 , C H O, C H , C H N 34 48 242 70 C H NO, C H N 34 48 , C H O, C H 4 6 5 10 84 C H NO, C H N 4 6 5 10 , C H O, C H O , C H , C H N O 5 8 4 4 2 6 12 3 4 2 86 C H NO, C H NO , C H N 4 8 3 4 2 5 12 , C H O , C H O, C H N , C H N O 4 6 2 5 10 4 10 2 3 6 2
P.E. Vickers et al./ Carbon 38(2000)675-689 interactions, it is assumed that AG has additive dispersive RT In(N) values" for molecular probes adsorbed onto untreated (d)and acid-base(AB)contributions [21, 281 Degree of fibre treatment (% △G=G4+△GAB re a nu 4.42 tion of AG to AG of a material [29-35 8.13 7.24 approach uses the boiling points of the probe 13.8 1.1 [29]. If the AG, values of the n-alkanes are plotte = 5.0 14.1 their boiling points a straight line should be attained TMP 9.22 pecific acid-base interactions between a probe and the CHCI 3.79 04 5.09 3.85 stationary phase cause the retention volume of the probe to 4.19° deviate from the line due to the n-alkanes AG can be t-BuOH 18.7 EtAc 16.2 14.9 15.8 13.7 (△G-△Ga)=RTln(VN/ RT In(N) values calculated with RT in k mol and Nin where IN and VN.ref are the net retention volumes of the Determined using a flow-rate of 15 ml min polar probe and a hypothetical reference n-alkane having the same the same boiling point, respectively 3.2.1.I. Free energy of adsorption of n-alkanes. Fig 4 Fig 5 shows graphs of RT In(n) values plotted against shows a plot of RT In(N) values vs the number of carbon the boiling point of the probe molecules. Again the n- atoms per n-alkane for the homologous series of n-alkanes alkanes yield a straight line. The distance between the adsorbed onto the untreated and the 200% treated fibres. It specific probe markers and the reference line is a measure can be seen that a straight line is obtained when the of AG a the acid-base contribution to the free energy of n-alkane series are used as IGC probe molecules. The adsorption. For the untreated fibre, the markers corre- gradient of each plot is a measure of AG, the free sponding to the specific probes are lying close to the energy of adsorption per methylene dispersive interactions reference line, an indication of proportional to the square root of y he slight weak specific interactions with the solid surface except for decrease in the slope as a result of surface treatment is thus BuoH(see Table I for abbreviations) which undergoes an indication of a decrease of y- as a result of surface substantial acid-base interactions. Chloroform has its treatment marker lying on the reference line and its interaction is thus mainly via dispersive forces. In the case of the 200% 3.2. 1.2. Free energy of adsorption of polar probes and the DFT, the basic probes EtAc and THF and the amphoteric branched alkane. If the material under test and a molecu- l-Buoh exhibit significant deviations from the reference, lar probe interact via both dispersive and acid-base which indicates that the oxidised fibres are showing acidic 25 200% 2E Fig. 4. RT In(VN)vS number of carbons, ne
682 P.E. Vickers et al. / Carbon 38 (2000) 675 –689 Table 5 interactions, it is assumed that DG has additive dispersive a a RT ln(VN ) values for molecular probes adsorbed onto untreated (d) and acid–base (AB) contributions [21,28]: and treated carbon fibres at 508C d AB DG 5 DG 1 DG (3) Degree of fibre treatment (%) aaa Probes 0 25 50 100 200 There are a number of methods in calculating the contribu- AB tion of DG to DG of a material [29–35]. A simple C6 5.21 4.42 a a approach uses the boiling points of the probe molecules C7 9.63 8.13 7.24 8.89 6.65 C8 13.8 12.0 11.1 12.5 10.3 [29]. If the DG values of the n-alkanes are plotted against a C9 16.1 15.0 16.2 14.1 their boiling points a straight line should be attained. TMP 9.22 7.97 7.35 8.51 6.69 Specific acid–base interactions between a probe and the b CHCl 3.79 3.54 4.04 5.09 3.85 3 stationary phase cause the retention volume of the probe to CCl 4.30 2.84 2.60 4.19 2.40 b AB 4 deviate from the line due to the n-alkanes. DG can be a t-BuOH 11.1 16.8 18.7 16.9 computed by: EtAc 7.94 14.0 15.5 16.2 14.5 AB d THF 8.09 13.8 14.9 15.8 13.7 2 DG 5 2 (DG 2 DG ) 5 RT ln(V /V ) (4) a a a N N,ref a 21 RT ln(VN ) values calculated with RT in kJ mol and VN in 3 where V and V are the net retention volumes of the cm . N N,ref b Determined using a flow-rate of 15 ml min . 21 polar probe and a hypothetical reference n-alkane having the same the same boiling point, respectively. 3.2.1.1. Free energy of adsorption of n-alkanes. Fig. 4 Fig. 5 shows graphs of RT ln(V ) values plotted against N shows a plot of RT ln(V ) values vs. the number of carbon the boiling point of the probe molecules. Again the n- N atoms per n-alkane for the homologous series of n-alkanes alkanes yield a straight line. The distance between the adsorbed onto the untreated and the 200% treated fibres. It specific probe markers and the reference line is a measure AB can be seen that a straight line is obtained when the of DG , the acid–base contribution to the free energy of a n-alkane series are used as IGC probe molecules. The adsorption. For the untreated fibre, the markers corre- CH2 gradient of each plot is a measure of DG , the free sponding to the specific probes are lying close to the a energy of adsorption per methylene group, which is dispersive interactions reference line, an indication of d [27] proportional to the square root of g . The slight weak specific interactions with the solid surface except for s decrease in the slope as a result of surface treatment is thus t-BuOH (see Table 1 for abbreviations) which undergoes d an indication of a decrease of g as a result of surface substantial acid–base interactions. Chloroform has its s treatment. marker lying on the reference line and its interaction is thus mainly via dispersive forces. In the case of the 200% 3.2.1.2. Free energy of adsorption of polar probes and the DFT, the basic probes EtAc and THF and the amphoteric branched alkane. If the material under test and a molecu- t-BuOH exhibit significant deviations from the reference, lar probe interact via both dispersive and acid–base which indicates that the oxidised fibres are showing acidic Fig. 4. RT ln(V ) vs. number of carbons, n . N c
P.E. Vickers et al. Carbon 38(2000)675-689 22 86420 86420 8E32 86420 86420 110 Fig. 5. Plot of RT In(N)vs. T, for:(a)0%; and (b)200% carbon fibres +, n-CE-9: D, CHCI,, O, THF; +CCl4: A, EtAc, -1-BuOH; X TME Chloroform has its marker above but close to the I-BuoH(38%)*THF(27%)> EtAc(18%)> CHCI, (2%) us indicating a weakly basic character of the T treated fibres. CCla is a weak acid and usually The trend observed in the probes relevant to this study is interacts mainly via dispersive interactions. For this reason shown above. where the numbers in brackets refer to the its corresponding marker is very close to the reference line degree of self-association. CCl, may undergo some repul As far as TmP is concerned retention data are almost sive interactions with the surfaces as AGa invariably identical to those of n-heptane since both alkanes have takes positive values(negative deviation from the disper- very close boiling points. For this apolar probe we define a △ BA parameter as[36 Table 6 △BA=RTin(N(TMP)/Ne Free energy of adsorption per methylene group(-AG4 ), ABA for isooctane and acid-base contribution to the free energy of where Ba stands for branched alkane. Since TMP and C7 adsorption(aGa )of polar probes adsorbed onto untreated and have comparable boiling points, VN.ref for TMP is almost treated carbon fibres at 50C that of c7 Table 6 reports △ ba and ag values for ethylene groups Kanes, TMP and the specific -△GH4.3053.9763864 3.718 probes, respectively results reported in Table 6 0.565-0.1810.052 0.304 0.005 oxidised fibres than with the untreated fibres. One can also 9 9cT quantitatively indicate that there is a significantly greater interaction between the acidic and basic probes with the CHO 1.14 2.35 1.67 139 145 note that the strength of acid-base interactions roughly 1-BuOH 3.78 11.13 10.1612.49 increase with the degree of self-association reported in the EtAc 9.24 1145 728 10.99 THF literature 37, 38 3.3510.721249 1.73 11.76
P.E. Vickers et al. / Carbon 38 (2000) 675 –689 683 Fig. 5. Plot of RT ln(V ) vs. T for: (a) 0%; and (b) 200% carbon fibres. ♦, n-C ; h, CHCl ; d, THF; 1CCl ; m, EtAc; 2 t-BuOH; 3, N b 6–9 3 4 TMP. character. Chloroform has its marker above but close to the t-BuOH(38%) | THF(27%) . EtAc(18%) . CHCl (2%) 3 n-alkane thus indicating a weakly basic character of the 200% DFT treated fibres. CCl is a weak acid and usually 4 The trend observed in the probes relevant to this study is interacts mainly via dispersive interactions. For this reason shown above, where the numbers in brackets refer to the its corresponding marker is very close to the reference line. degree of self-association. CCl may undergo some repul- 4 AB As far as TMP is concerned, retention data are almost sive interactions with the surfaces as DGa invariably identical to those of n-heptane since both alkanes have takes positive values (negative deviation from the dispervery close boiling points. For this apolar probe we define a DBA parameter as [36]: Table 6 CH2 DBA 5 RT ln(V (TMP)/V ) (5) Free energy of adsorption per methylene group (2DGa ), DBA N N,ref for isooctane and acid–base contribution to the free energy of AB where BA stands for branched alkane. Since TMP and C7 adsorption (2DGa ) of polar probes adsorbed onto untreated and have comparable boiling points, V for TMP is almost treated carbon fibres at 508C N,ref that of C7. 0 25 50 100 200 CH2 AB Table 6 reports DG , DBA and DG values for a a CH2 methylene groups in 2DGa 4.305 3.976 3.864 3.887 3.718 n-alkanes, TMP and the specific DBA 20.565 20.181 0.052 20.304 0.005 probes, respectively. The results reported in Table 6 AB quantitatively indicate that there is a significantly greater — DGa interaction between the acidic and basic probes with the CHCl3 20.22 1.14 2.35 1.67 2.55 oxidised fibres than with the untreated fibres. One can also CCl4 22.07 21.92 21.39 21.45 21.12 t-BuOH 3.78 11.13 – 10.16 12.49 note that the strength of acid–base interactions roughly EtAc 1.52 9.24 11.45 7.28 10.99 increase with the degree of self-association reported in the THF 3.35 10.72 12.49 11.73 11.76 literature [37,38]:
P.E. Vickers et al. Carbon 38(2000)675-689 ve interaction line) greater than I kJ mol. ABA can be 3.3. Wettability studies by dCaa considered within the accuracy of the determination of the deviation from the dispersive interaction line. Therefore, DCAA exploits the Wilhelmy plate technique to de- the interaction of TMP is consistent with its boiling point termine the contact angle made by the appropriate liquid and reveals no unexpected anomalies. probe with the surface of the material under investigation The technique has been described by Chang et al. [40] and 3. 2. 2. Properties of the carbon fibres measures the mass deflection incurred when the probe Table 6 is useful to establish the liquid is raised in order to come into contact with fibre carbon fibre physicochemical properties, that is y -, the urface using the relationship described below dispersive contribution to the surface free energy and cOs 6= acid-base constants derived from the Saint Flour-Papirer approach [30] The value of y. is proportional to the square of AG CH2 where, P is the perimeter of the fibre, g is the gravitational and determined from the slopes shown in Fig. 4 [27] constant, m is the mass(experimental reading from mi- crobalance) y4=(1/4xm)△G/Nacm2) Initially, 0, 25 and 100% dFT fibres were using hexadecane(see Table 2)as the probe where N is the Avogadro number, acH2 is the cross- order to gain an estimate of the fibre perimeter sectional area of an adsorbed CH, group (6 A), y is cane wets the fibre surface efficiently(0<30%)because it the surface free energy of a solid containing only methyl possesses a very low surface free energy, and, being a CI)ps such as polyethylene ()H2=368-0.058T large molecule, is unlikely to fill micropores on the fibre surface, a likely source for contact angle hysteresis. All The relative acid-base properties of the fibres can be three types of fibres gave force curves which showed very investigated by using a combination of the adsorption data little hysteresis, and the fibre diameter of all the fibres was for CHCI, and THF taken as reference Lewis acidic and found to be 6.9 um, in good agreement with Mahy et al basic probes, respectively. Lara and Schreiber[39] defined [31, who reported a fibre diameter of 7. 1 um for the same the following parameters type of fibres 0, 25 and 100% DFT fibres were then analysed using aTHe =-AGa (THF)in kJ mol- ater,glycerol and 1-bromonapthalene as the probe media (7) These were chosen because of they provide a wide range of dispersive and polar contributions to the total surface free energy. as listed in Table 2. The contact angles of BcHCE=-AGA (CHCI, )in kJ mol -I ( 8) fibres with the contact media are reported in Table 8, each of which is an average of five readings. It is apparent that where aThe and BcHcu are the acidity and basicity all receding angles are lower than advancing angles constants, respectively, derived for the solids under test. recorded during the same experiment. This is a widely The constants aTHE and BcHcu together with y, values reported effect [41] and is attributed to a combination of re reported in Table 7. They suggest that carbon fibres chemical heterogeneity and microroughness of the fibre have a high dispersive component to the surface energy, surface. Contact angle hysteresis occurs because the ad y., which however decreases as a result of surface vancing liquid front more easily wets high-energy patches treatment. The value of y. for the untreated fibre is similar present on the surface and is held up by the low-energy to that (129 mJ m at 45C)of HOPG [14] e, resulting in a contact angle more characteristic of treatment,y. value decreases by a value as large as 26.5 the low-energy patches. During emersion, the converse is mJ m- for the 200% DFT fibres. In contrast, the acidity true, with the wetting liquid remaining on the high-energy considerably increases as well as the basicity, although to a phase and holding up the retreat of the liquid from the lesser extent solid surface, thus giving a contact angle characteristic of Dispersive and acid-base constants for carbon fibres with a range of oxidation treatments y(mJ m CTHE 2(aTHE B Untreated 1045 500 l1.8 l1.8 10.86
684 P.E. Vickers et al. / Carbon 38 (2000) 675 –689 21 sive interaction line) greater than 1 kJ mol . DBA can be 3.3. Wettability studies by DCAA considered within the accuracy of the determination of the deviation from the dispersive interaction line. Therefore, DCAA exploits the Wilhelmy plate technique to dethe interaction of TMP is consistent with its boiling point termine the contact angle made by the appropriate liquid and reveals no unexpected anomalies. probe with the surface of the material under investigation. The technique has been described by Chang et al. [40] and measures the mass deflection incurred when the probe 3.2.2. Properties of the carbon fibres liquid is raised in order to come into contact with fibre The data reported in Table 6 is useful to establish the surface using the relationship described below. d carbon fibre physicochemical properties, that is g , the s mg dispersive contribution to the surface free energy and cos u 5 ] (9) acid–base constants derived from the Saint Flour-Papirer Pglv approach [30]. where, P is the perimeter of the fibre, g is the gravitational d CH2 The value of g s is proportional to the square of DG constant, m is the mass (experimental reading from mi- and determined from the slopes shown in Fig. 4 [27]: crobalance). d CH2 2 Initially, 0, 25 and 100% DFT fibres were analysed g 5 (1/4g )(DG /Na ) (6) s CH2 CH2 using hexadecane (see Table 2) as the probe liquid, in order to gain an estimate of the fibre perimeter. Hexade- where N is the Avogadro number, a is the cross- CH2 2 cane wets the fibre surface efficiently (u ,308) because it ˚ sectional area of an adsorbed CH group (6 A ), g is 2 CH2 possesses a very low surface free energy, and, being a the surface free energy of a solid containing only methyl- large molecule, is unlikely to fill micropores on the fibre ene groups such as polyethylene (gCH2536.8–0.058T surface, a likely source for contact angle hysteresis. All (8C)). three types of fibres gave force curves which showed very The relative acid–base properties of the fibres can be little hysteresis, and the fibre diameter of all the fibres was investigated by using a combination of the adsorption data found to be 6.9 mm, in good agreement with Mahy et al. for CHCl and THF taken as reference Lewis acidic and 3 [3], who reported a fibre diameter of 7.1 mm for the same basic probes, respectively. Lara and Schreiber [39] defined type of fibres. the following parameters: 0, 25 and 100% DFT fibres were then analysed using AB 21 water, glycerol and 1-bromonapthalene as the probe media. a 5 2DG (THF) in kJ mol (7) THF a These were chosen because of they provide a wide range of dispersive and polar contributions to the total surface and free energy, as listed in Table 2. The contact angles of AB 21 b 5 2DG (CHCl ) in kJ mol (8) fibres with the contact media are reported in Table 8, each CHCl3 a 3 of which is an average of five readings. It is apparent that where a and b are the acidity and basicity all receding angles are lower than advancing angles THF CHCl3 constants, respectively, derived for the solids under test. recorded during the same experiment. This is a widely d The constants a and b together with g values reported effect [41] and is attributed to a combination of THF CHCl3 s are reported in Table 7. They suggest that carbon fibres chemical heterogeneity and microroughness of the fibre have a high dispersive component to the surface energy, surface. Contact angle hysteresis occurs because the ad- d g , which however decreases as a result of surface vancing liquid front more easily wets high-energy patches s d treatment. The value of g s for the untreated fibre is similar present on the surface and is held up by the low-energy 22 to that (129 mJ m at 458C) of HOPG [14]. Upon phase, resulting in a contact angle more characteristic of d treatment, g value decreases by a value as large as 26.5 the low-energy patches. During emersion, the converse is s 22 mJ m for the 200% DFT fibres. In contrast, the acidity true, with the wetting liquid remaining on the high-energy considerably increases as well as the basicity, although to a phase and holding up the retreat of the liquid from the lesser extent. solid surface, thus giving a contact angle characteristic of Table 7 Dispersive and acid–base constants for carbon fibres with a range of oxidation treatments d 22 1/2 DFT (%) g (mJ m ) a b 2(a b ) s THF CHCl3 THF CHCl3 Untreated 104.5 3.4 20.223 |0 25 89.2 10.7 1.1 6.86 50 84.3 12.5 2.4 10.95 100 85.2 11.8 1.7 8.96 200 78.0 11.8 2.5 10.86