International Journal of Biological Macromolecules 37 (2005) 179–188

Wet spinning ofBombyx mori silk fibroin dissolved inN-methylmorpholineN-oxide and properties of regenerated fibres

Enrico Marsanoa,∗, Paola Corsinia, Cristina Arosiob, Alessandra Boschib,Michele Morminoc, Giuliano Freddib

a Dipartimento di Chimica e Chimica Industriale, Universita di Genova,Via Dodecaneso 31-16146 Genova, Italyb Stazione Sperimentale per la Seta, Via G. Colombo 83-20133 Milano, Italy

c ISMAC-CNR, Sezione di Genova, Via De Marini 6-16149 Genova, Italy

Received 7 September 2005; received in revised form 14 October 2005; accepted 14 October 2005Available online 21 November 2005

Abstract

Silk fibroin (SF) was dissolved inN-methyl morpholineN-oxide (NMMO) at a polymer concentration of 13% (w/w); thermal and rheologicalsolution properties were characterized. The melting/crystallization behaviour of NMMO was influenced by SF presence. Melting of NMMOh hinga ction oft extrudedfi hysical, andm erallys inning draww x did notc al behaviourt ur. Typicalv©

K

1

raitm(wicl

,onge,bio-and

ther tot, the

erat-ated

c

herecess.prop-inallvent

0d

ydrate decreased to 71◦C and a cold crystallization peak appeared at 35◦C on heating. None crystallization occurred on cooling. Quenct a temperature of 50◦C or higher did not induce any crystallization on heating. Viscosity of SF–NMMO solutions decreased as a fun

emperature. At 75◦C, viscosity remained constant for 360 min. SF–NMMO dope was spun by using a lab-scale wet spinning line. Thelament was coagulated in an ethanol bath. Regenerated SF fibres were collected at different draw ratios and their morphological, pechanical properties were characterized. Fibre diameters ranged from 133 to 19�m, cross-section was regularly circular, and surface was gen

mooth, with a very fine granular aspect. Birefringence increased with increasing the draw ratio, especially when take up and post-spere coupled. FT-IR spectra and DSC thermograms confirmed that SF fibres crystallized into Silk II structure. The IR crystallinity indehange as a function of drawing. Regenerated SF fibres undrawn or drawn only during the coagulation step showed the mechanicypical of a brittle material. However, when both take up and post-spinning draw were applied, fibres displayed a ductile-stable behavioalues of the mechanical parameters of regenerated SF fibres were:E = 8.7 GPa,σb = 120 MPa andεb = 35%.2005 Elsevier B.V. All rights reserved.

eywords: Silk fibroin; N-methyl morpholineN-oxide; Regenerated fibres

. Introduction

Bombyx mori silk is one of the most valuable starting mate-ials for the clothing industry owing to its outstanding aestheticnd sensorial properties (shiny aspect, handle, etc.). Physiolog-

cal properties are attractive as well. In fact, silk by itself ishermo-insulating, well tolerated by the skin, and capable ofaintaining an optimally moisturized environment. Silk fibroin

SF) is also an attractive renewable fibrous biopolymer endowedith an excellent biocompatibility, which strongly recommends

ts use as a means to develop innovative biomaterials intended forlinical applications, such as substitutes for the anterior cruciate

igament, bioactive dressings, and scaffolds for connective tis-

∗ Corresponding author. Tel.: +39 010 3538727; fax: +39 010 3538727.E-mail address: marsano@unige.it (E. Marsano).

sue engineering[1]. To this aim, not only native SF in fibre formbut also various kinds of regenerated SF materials (gel, sppowder, film) exert a strong attractiveness owing to theirlogical properties and to the ability to support cell adhesionproliferation[2–8]. Currently, efforts are aimed at adjustingstructure of biomaterials to their required functions in ordeaddress a broad range of biomedical needs. In this contexpossibility of engineering silk-based biomaterials by regening the native biopolymer in other forms, including regenerSF fibres, pure or blended with other polymers[9], with theinclusion of new useful functionalities[10] is a crucial scientifiand technological challenge.

Although SF dissolves in several solvent systems, tare still some problems associated with the spinning proRegenerated SF fibres not always show good mechanicalerties, frequently they are brittle compared to their origforms. Improvements could be brought using a new so

141-8130/$ – see front matter © 2005 Elsevier B.V. All rights reserved.oi:10.1016/j.ijbiomac.2005.10.005

180 E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188

and/or developing fibre-processing methods. Numerous papersare present in the literature concerning both the study of new sol-vents, able to give spinnable silk solutions, and the improvementof spinning conditions, in order to prepare innovative regener-ated SF filaments. Yazawa[11] used a concentrated aqueous SFsolution as a spinning dope and a saturated ammonium sulfatesolution as coagulant bath. Ortho phosphoric acid (H3PO4) asa solvent and a mixture of ammonium and sodium sulfate ascoagulant bath were used by Ishizaka et al.[12] and Matsumotoet al. [13] reported the results obtained with a system contain-ing LiBr·H2O, EtOH, and H2O to prepare SF dope. The roleof additives in the dope solution or in the coagulant, tempera-ture and spinning parameters on fibres properties were investi-gated too. Nevertheless, these spinning systems showed seriousproblems with polymer degradation process and the reportedstrength and elongation values did not exceed 240 MPa and 11%,respectively.

The results obtained by dissolving and spinning SF inhexafluoro-2-propanol (HFIP), by using methanol as coagulant,showed the possibility to produce SF filament with excellentmechanical properties. In particular, tenacity and elongation val-ues of about 570 MPa and 8.9%, respectively, similar to those ofnatural silk fibre were obtained[14–16]. Yao et al.[17] were alsoable to produce regenerated SF fibres using hexafluoroacetonehydrate (HFA) as solvent and methanol as coagulant, and theeffects of post-spinning by stretching and of steam-annealing toi rtedH entea pin-n g SFi ta singt n byh batha , i.et f then menh goos

uralp res iN -p hiss sea etterm dC ars oo kindo solub , ans uctsT wedt cesi genb merd MOp ion o

the CE pulp. Additionally, produced fibres are environmentallybenign because NMMO is a non-toxic solvent and is almostcompletely recycled (NMMO recovery is usually greater than99%)[25]. Owing to these advantages, a great deal of technicaldevelopments of the NMMO-based spinning process, aiming atproducing fibres with the desired properties, were reported bymany manufacturers (Lyocell-process)[26–29].

One of the authors[30] has shown that the NMMO–H2Osolvent system can be successfully applied for SF dissolution,which was completed within 2 h at about 100◦C under mechan-ical stirring. An amount of silk as large as 20% (w/w) could bedissolved in these conditions. NMMO–H2O/SF solutions withhigh silk concentration were highly viscous and potentially suit-able for fibre spinning. In this study, regenerated wet-spun SFfilaments were fabricated using NMMO monohydrate as thesolvent and ethanol (EtOH) as the coagulant. The thermal andrheological behaviour of SF–NMMO dope solutions were stud-ied in order to optimize the spinning parameters of the lab-scalespinning line. A series of SF filaments collected at differenttake up rate and subsequent post-spinning draw ratios were thenproduced and extensively characterized by optical microscope(OM), SEM, FT-IR, and DSC in order to elucidate the effect ofspinning conditions on fibre morphology, structure, and mechan-ical performance.

2. Materials and methods

2

bydf m-p ents.T bres( sso-l ousL edt leter ofp tests.

,a omA

2

on ofN ctt cen-t ht bout1 Fis-c )w on inN tem-p , witha s of

mprove the mechanical property of SF filament were repoowever, the extremely high cost of these solvents has prevny further scale-up for a possible industrialization of the sing process. More recently, results obtained from spinnin

n formic acid were reported by Park et al.[18] and Hudson el. [19], who particularly focused on the role of the proces

echnique on fibre properties. “As spun” fibres were drawand in a wet state, after 4 h of residence in a coagulationnd fibres with very interesting properties were obtained

enacity of about 1 GPa and elongation of 30%. In spite oumerous studies on the fabrication of regenerated SF filaowever, none was really successful due to the lack of aolvent system.

A new, renewable, friendly solvent, able to dissolve natolymers and already used in the industry of man-made fib-methyl morpholineN-oxide (NMMO) monohydrate, a comound belonging to the organic cyclic amine oxide family. Tolvent has proved to be commercially successful becaullows to obtain regenerated cellulose (CE) fibres with bechanical properties than viscose rayon[20–24]. RegenerateE materials have been largely based for more than 100 yeld viscose technology or on cuprammonium process. Thisf industry produced regenerated fibres via a metastablele CE derivative (cellulose xanthogenate) and they weretill are accompanied by environmental hazardous by-prodhe newly developed NMMO-based spinning process allo

o overcome most of these problems. Though chemical prong of CE, as well as SF, is generally difficult due to its hydroonded structure, which involves that these natural polyo not melt and are not soluble in usual solvents, the NMrocess seems to achieve a physical and direct dissolut

.d

,.

t,d

s

it

n

-d.

s-

s

f

.1. Materials

Silk fibroin (SF) fibres were purified from silk cocoonsegumming with water, in autoclave, at 120◦C for 15 min,

ollowed by extensive rinsing with warm distilled water to coletely remove sericin, a gum protein surrounding SF filamhe degree of polymerization of starting degummed SF fi

Mv) ranged from 220 to 240 kDa. To prepare SF for diution in NMMO, fibres were dissolved in saturated aqueiBr (9–10 M) at 60◦C for 3 h. The solution was, then, filtrat

o remove insoluble foreign matters, dialysed until compemoval of salt, and freeze-dried. The resulting SF in formorous, sponge-like material was used for the dissolution

A commercial aqueous solution, NMMO–H2O 50/50 (w/w)nd the antioxidantn-propyl gallate (PG) were purchased frldrich Co.

.2. Dope preparation

SF was dispersed into the commercial aqueous solutiMMO containing 0.7% (w/w)n-propyl gallate with respe

o NMMO monohydrate. The dispersion was, then, conrated under vacuum at 55◦C to obtain NMMO hydrates withe desired water content (the maximum value was a4.7%, w/w). The water content was determined by Karlher method. SF concentration (Cp) was fixed to 13% (w/with respect to the solution. In order to prepare a SF solutiMMO monohydrate, the dispersion was heated to diverseeratures for different times, under a nitrogen atmospherefast energetic stirring, controlling time by time the progres

E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188 181

Fig. 1. Schematic representation of the dry-jet wet spinning line.

dissolution with an optical microscope. Air bubbles were trappedin the SF solution, due to the fast energetic stirring, and causedflow instability in the dry-jet wet spinning line. Centrifugation athigh temperature of the dope before spinning was used in orderto disengage air bubbles.

2.3. Spinning

A scheme of the dry-jet wet spinning line used to produceregenerated SF fibres is reported inFig. 1. This dry-jet wet spin-ning line is composed by an extrusion unit, in which extrusionrate and spinneret diameter can be varied, a coagulation bathand two sets of spools, take up and roller. Collecting rate canbe varied so that fibres are collected with different drawn ratios,defined as the ratio between spools’ speed and fibre velocity atthe spinneret hole, or rather when no stretch was applied dur-ing coagulation. Collected fibres were further washed in runningwater for about 48 h in order to extract NMMO and, then, theywere dried under vacuum at 40◦C for 24 h and stored over CaCl2until testing.

2.4. Viscosity

The study of the rheological properties of SF solutions inNMMO monohydrate was performed by using a Brookfieldviscometer (LVDV-II+ with SC4-8R(P) chamber and SC4-16s opev earr

d-i lvedi3 asm ofp ngt areα

2

thea The

samples were cycled at 5◦C/min from 0 to 100◦C, cooled at5◦C/min from 100 to 0◦C, and heated again at 5◦C/min to100◦C. Some samples were annealed from 100 to 20◦C or 50◦Cfor different time (t = 5–180 min) before performing the secondheating cycle.

Differential scanning calorimetry (DSC) measurements onSF fibres were performed with a DSC-30 instrument (MettlerToledo), from room temperature to 500◦C, at a heating rate of10◦C/min, on 2–3 mg samples. The open aluminium cell wasswept with N2 during the analysis.

2.6. Microscopic analyses

Fibre diameters were determined using an optical microscopePolyvar Pol (OM). The average diameter was the result of at least80 measurements.

The birefringence of regenerated SF fibres was measuredusing a polarizing microscope with crossed polars light fit-ted with an Ehringhaus compensator. SF fibres were placedon a glass slide and the measurements were done in air.The angle of retardation was measured and converted to bire-fringence.

Surface morphology was examined by means of scanningelectron microscopy (SEM), using a Jeol microscope modelJSM-6380LV. Samples were observed at 10 kV accelerationv herew

2

exuss cellm

2

singa ngtha rateo to8 te of

pindle). Measurements were carried out with a 4.2 ml dolume, in the 75–100◦C temperature range, at different shates (0.8–2 s−1).

The intrinsic viscosity [η] of SF was determined accorng to the SNV 195595 standard method. SF was disson a saturated LiBr aqueous solution (9–10 M), at 60◦C, forh. After dilution 1:1 (v/v) with water–glycerine, viscosity weasured at 20◦C with a capillary viscometer. The degreeolymerization of SF (Mv) was, then, calculated by applyi

he empirical Mark–Houwink equation, whose constants= 0.95 and logK =−2.53[31].

.5. Thermal analysis

A DSC Modulated 2920 (TA Instruments) was used fornalysis of the thermal behaviour of SF–NMMO dopes.

oltage, after gold coating under reduced Argon atmospith a Med 020 Coating System (BAL-TEC).

.7. FT-IR spectroscopy

FT-IR spectra were measured in the ATR mode with a Npectrometer Thermo Nicolet, equipped with a ZnSe ATRod. Smart Performer.

.8. Mechanical properties

Mechanical properties of single fibres were measured un Instron dynamometer mod. 5500, with a 25 mm gauge let a crossbar rate of 15 mm/min, corresponding to strainf 1.0× 10−2 s−1 if elongation at break of specimen was up%, and at a crossbar rate of 10 mm/min, relative strain ra

182 E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188

6.7× 10−3 s−1 if elongation at break was under 8% according toASTM D3822-01-2004 standard method. The elastic modulus(E), the breaking strength (σb), and the elongation to break (εb)were calculated as the average of at least 15 measurements fromstress–strain curves.

3. Result and discussion

3.1. SF–NMMO solutions

Solutions of SF in NMMO monohydrate, at a polymer con-centrationCp = 13% (w/w), were obtained by direct dissolu-tion of SF fibres according to the method reported by Freddi[30] and recently confirmed by Shao et al.[32]. The solu-tions appeared homogeneous when observed with the opticalmicroscope under the polarising light. However, we found someproblems during the extrusion of these dopes in the spinningline. The thread tended to break randomly at the spinneret prob-ably due to the presence of micro-inclusions of material notcompletely dissolved. These micro-particles were not visibleunder the OM. Higher temperature (T = 130◦C) and longer stir-ring time (60 min) were attempted in order to avoid this problemand to prepare spinnable dopes. The new SF solutions could bewet-spun but the increase of temperature during their prepara-tion brought a lower viscous system and, as a consequence, anextensive degradation of SF, whose degree of polymerizationd

erh time( ed.D medn ee n off uentc r,i void-i andv fromf merc

3

ouro t. AN edm ,a lb atedb eltingt l-lc . Thegv[ ribedt ter-

Fig. 2. [A] Thermal behaviour of SF solutions atCp = 13% (w/w), from−70 to100◦C with scanning rate fixed to 5◦C/min: (a) heating; (b) cooling. [B] Heatingcurves of SF solutions atCp = 13% (w/w) after an isothermal treatment at 20and 50◦C for different times: (a) 20◦C, 5 min; (b) 20◦C, 180 min; (c) 50◦C,180 min.

action between NMMO hydrate and SF. The same behaviourwas reported for cellulose (CE) solutions in NMMO mono-hydrate, where a higher drop of the melting temperature wasreported[33,35,36]. Interestingly, CE–NMMO solutions at apolymer concentrationCp = 13% (w/w) did not show any crys-tallization both in the cooling of the melt from 100 to 20◦Cat 5◦C/min and in the subsequent heating scan. The inhibi-tion of NMMO crystallization in CE–NMMO solutions wascorrelated to the strong interactions between the polymer andthe solvent. The different behaviour of SF solutions in NMMOmonohydrate, with particular reference to the absence of crys-tallization on cooling and to the presence of cold crystallizationon heating, can be attributed to milder SF–NMMO interactions,as compared to the CE–NMMO system. Hence, particular caremust be paid to the choice of the spinning temperature in orderto avoid the occurrence of crystallization phenomena and, atthe same time, to keep the viscosity of the system as high aspossible.

Fig. 2B shows the heating scans of SF–NMMO samplesquenched atTc = 20 or 50◦C and maintained at this temperaturefor different times. The sample annealed at 20◦C for 180 min(curve b) crystallized completely, no cold crystallization wasobserved on heating and only the endothermic melting peak at71◦C appeared. For shorter quenching times (5 min), a partialcrystallization occurred, and a wide cold crystallization at about

ecreased of about 45%.Solubilization of freeze-dried SF in NMMO, on the oth

and, required milder conditions of temperature andT = 110◦C; 20 min). The extent of degradation was limitegree of polymerization of the starting material (degumative silk fibre) was in the rangeMv = 220–240 kDa. At thnd of the processing cycle, which comprised preparatio

reeze-dried SF, dissolution in NMMO hydrate, and subseqoagulation, the value never fell belowMv = 190 kDa. Moreove

t was possible to use these solutions in the spinning line ang the problem of thread breaks. The following thermaliscometric characterizations refer to solutions obtainedreeze-dried SF dissolved in NMMO hydrate at a fixed polyoncentration,Cp = 13% (w/w).

.2. Thermal properties of SF–NMMO solutions

It is well known that the melting and crystallization behavif NMMO hydrates is greatly affected by the water contenMMO hydrate with water content of 14.7% (w/w) showelting and crystallization peaks at 78 and 25◦C, respectivelyt a scanning rate of 5◦C/min [33]. Fig. 2 shows the thermaehaviour of 13% (w/w) SF solutions in this solvent investigy DSC measurements. The presence of SF lowered the m

emperature of NMMO hydrates to 71◦C, while a cold crystaization of NMMO appeared atTc = 35◦C on heating (Fig. 2A,urve a). None crystallization appeared on cooling (curve b)lass transition temperature was detected at−45◦C, a loweralue than that of pure NMMO monohydrate (Tg =−41◦C)34]. The decrease of the melting temperature can be asco the fact that crystallization was partiality inhibited by in

E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188 183

Fig. 3. Behaviour of viscosity of the SF solution atCp = 13% (w/w): (a) as afunction of temperature and (b) at constant temperature (75◦C) as a function oftime.

50◦C was observed (curve a). Samples quenched at higher temperature (Tc = 50◦C, curve c), and annealed for 180 min did notshow any thermal transition on heating. These results confirmthe tendency of NMMO hydrate to crystallize in the presenceof SF. Five minutes at 20◦C are enough to trigger the processthat is surely complete at 180 min. As we could observe fromthe thermal characterization, SF–NMMO hydrate did not crys-tallize after quenching atTc = 50◦C for 180 min and this result isextremely important in view of setting the optimum temperaturefor the spinning of the dope.

3.3. Viscosity of SF–NMMO solutions

Fig. 3a shows the temperature dependence of the viscositof SF–NMMO solutions atCp = 13% (w/w). The measurementswere performed in series in the Newtonian plateaux, by gradually increasing the temperature. In order to equilibrate thesystem, viscosity measurements were performed after 10 miof standing at the fixed temperature. An increase of temperaturfrom 75 to 85◦C entailed a decrease of viscosity from 44 to23 Pa s. Taking into account the results of thermal and viscosity characterization of SF solutions in NMMO, the control ofthe dope temperature in the spinning line was set at 70–75◦C.The changes of viscosity of a SF–NMMO solution atCp = 13%(w/w), kept at 75◦C, were plotted as a function of time (Fig. 3b).A constancy of the viscosity with time was observed, confirmingt et Thist tionw , no

undesired crystallization processes occurred under these spin-ning conditions.

3.4. Dry-jet wet spinning of regenerated SF fibres

Fibre formation by dry-jet wet spinning proceeds by a verycomplex mechanism, which entails the combination of rheo-logical and diffusional phenomena.Fig. 1 shows the spinningline used in this study. The SF–NMMO solution was pumpedthrough a spinneret with a 100, 200, or 300�m hole diameter,and the rate of pumping was modified so that the extrusion ratewas fixed at 4 m/min, corresponding to a shear rate of 5300 s−1

(with a 100�m spinneret). When spinnerets of 200 or 300�mwere used, some flow instability was observed, the fibres werecharacterized by heterogeneous diameters, and dried fibres weretoo brittle and difficult to handle. With a 100�m spinneret, anair gap of 5 mm was used to extrude SF–NMMO solutions.This allowed to produce fibres with homogeneous diameter andwith more ductile handling. A greater air gap could not be usedbecause SF solutions began to drip and did not spin, due to theviscoelastic characteristics of the material. Indeed, SF solutionspresent a different viscoelastic behaviour with respect to cel-lulose ones. At the same polymer concentration, CE–NMMOsolutions could be spun with an air gap of 200 mm[21].

The thread emerging from the hole under conditions of freeextrusion (V ) has a velocity much smaller than the averagev factf ons,t m thec me-t well,md iam-e ratio,χ

v ringc es,S at thes con-d alueo ve tou er too

ar-i hreade ac wasc e upw geb ndt pre-c ange.W -g t wasd tch-i l ofs other.

hat the extent of SF degradation at 75◦C for 360 min, i.e. thime usually needed to spin SF solutions, was negligible.emperature allowed to have a highly viscous spinnable soluith viscosity constant for more than 180 min. Furthermore

-

y

-

ne

-

,

felocity of the spinning solutions at the spinneret hole. Inor elastic fluids, such as SF–NMMO concentrated solutihe release of stored elastic energy, as the fluid emerges froapillary, causes the stream of fluid to swell to a larger diaer than the spinneret’s one. This phenomenon, called die sakes the fluid jet to slow down. Die swell ratioχ, usuallyefined as the ratio of maximum jet radius to the spinneret dter, can be expressed approximately in terms of velocity= (Vo/Vf )1/2, whereVo is the extrusion rate andVf is the fibreelocity at the spinneret hole when no stretch is applied duoagulation[37]. Due to their intrinsic viscoelastic propertiF solutions presented a higher die swell than CE solutionsame polymer concentration and under the same extrusionitions (4 m/min extrusion rate), i.e. 3.3 for SF against a vf 2.6 for CE. This means that as extruded SF threads handergo drawing to a higher extent than CE threads in ordbtain filaments with good mechanical properties.

During the spinning process no flow instability or shenduced crystallization phenomena were observed. The txtruded from the spinneret at free rateVf was dipped intooagulation bath containing pure ethanol (EtOH), then, itontinuously removed from the bath by a godet at the takith peripheral velocityV1. In this step, a diffusional interchanetween the freshly formed still fluid thread (SF–NMMO) a

he coagulating solvent caused polymer solidification. SFipitated or crystallized as a consequence of this exchhen the godet velocity was larger thanVf the fluid jet was elon

ated, the protein chains were stretched while the coagulaniffusing in and the solvent out of the still liquid thread. Stre

ng of the individual chains along the fibre axis and removaolvent were simultaneous processes interacting with each

184 E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188

The maximum take up rate (V1) used was 8.8 m/min, because athigher values the thread frequently tended to break. This phe-nomenon could be ascribed to polymer jet cohesive fracture thatseems to be the most important mechanism in the formation offibres in a dry-jet wet spinning line. Different factors appear inthe theory of cohesive fracture, such asVf , polymer content inthe dope, spinning temperature and coagulation rate[37].

The filaments picked up by the take up godet were still in anelastic swelled state. The swelled SF–EtOH–NMMO threadswere further stretched by using a roller rotating spools withperipheral velocityV2 (post-spinning draw). In this step of thedry-jet wet spinning line, the stretching applied on the fibres wasperformed in air and, at the same time, residual ethanol evap-orated. Hence, the structure formed during fibre stretching wasfrozen and a more compact texture with lower fibre diameterwas obtained.

SF fibres were submitted to one or two stretching. The firstwas between the spinneret and the take up, while the second wasbetween take up and roller. The deformation that fibres weresubjected to can be described mathematically by the draw ratio(DR). DR is the ratio between the collecting velocity at the takeup with respect toVf (DR1 = V1/Vf ), or between the roller andthe take up velocity (DR2 = V2/V1).

It is well known that the drawing process is one of the mostimportant procedures used to fabricate fibre materials havinggood mechanical properties. The molecular orientation of fibrem rtiesp ntat ningd r. Int arer

3

3were

o ubseq rizedb bresd ang-i ters( hichd fine

fibres had an average diameter of about 19�m and were pro-duced by using take up (DR1) and post-spinning (DR2) drawratios of 15 and 2.7, respectively. It is worth noting that theextent of variability of fibre diameter was always limited andthat fibres maintained their circular shape. SF fibres drawn dur-ing the coagulation step were quite difficult to handle owing totheir enhanced brittle character. On the other hand, when bothDR1 and DR2 were applied, SF filaments became softer anddeveloped a ductile handling. Birefringence reached a plateauat about 0.018 irrespective of the value of DR1, while it increasedalmost linearly with increasing DR2, at constant DR1 (Table 1).These data highlight the chief contribution of the post-spinningdraw in the fibre-forming process, with particular emphasis onthe improvement of the overall molecular orientation of regen-erated fibres.

SEM revealed the regularity of both longitudinal and cross-sectional shape of regenerated SF fibres (Fig. 4). Surface wasgenerally smooth, with fine details whose morphology changeddepending on the spinning conditions. Undrawn SF fibres werecharacterized by the presence of surface ridges elongated alongthe fibre axis, attributable to protein aggregates only partiallystretched by passing through the spinning die and/or to proteinmasses rapidly coagulated when exposed to the solvent in thecoagulation bath. The ridges completely disappeared when drawat the take up was applied. The fibres became smoother and avery fine granular pattern appeared on their surface. These sur-f rawnb evi-d resw es.

el ofS terialss hainsw inga tos g net-w theh turecT ainsi lues,w t ofn thiss

TS of SF

S

1234567N

aterials obtained through drawing governs their propearticularly, the mechanical properties. A high degree of orie

ion produces high filament tenacity, moreover, a post-spinraw usually entails decreasing variability in fibre diamete

he following, the results obtained in fibre characterizationeported.

.5. Characterization of regenerated SF fibres

.5.1. MorphologyRegenerated SF fibres with different processing history

btained by spinning at various collecting take up rate and suent post-spinning draw. Their morphology was charactey OM and SEM microscopy. The whole regenerated fiisplayed a fairly regular circular shape, with diameters r

ng from 133 to 19�m depending on the spinning parameTable 1). Undrawn SF fibres showed the largest diameter, wrastically decreased as soon as drawing was applied. The

able 1pinning conditions, morphological, physical, and mechanical properties

ample DR1 DR2 Diameter (�m)

1.0 1.0 133± 1215 1.0 26.4± 1.018 1.0 26.1± 0.923 1.0 22.2± 0.415 1.5 22.7± 1.415 2.3 19.5± 0.815 2.7 18.5± 0.8

ative SF fibresa – – 11.5± 0.2

a From Ref.[38].

,-

-

st

ace features remained essentially unchanged for fibres doth at the take up and at the roller. Interestingly, no clearence of the aligned fibrillar structure typical of native silk fibas observed on the whole examined regenerated SF fibrThe sub-micron granules observed at the surface lev

F fibres, often detected on other regenerated SF mauch as films, can be attributed to aggregation of SF chich could not undergo complete unfolding during spinnnd drawing[39]. The well-known tendency of SF chainself-assemble into globular clusters by establishing a stronork of inter and/or intramolecular interactions betweenydrophobic polypeptide blocks forming its primary strucould be responsible for these morphological features[39,40].he hypothesis of an incomplete axially alignment of SF ch

n regenerated SF fibres is supported by birefringence vahich increased on drawing, but remained far below thaative silk fibres, even at the highest draw ratio used intudy.

fibres regenerated in NMMO monohydrate

Birefringence E (GPa) σb (MPa) εb (%)

– 2.6 43 2.00.018 7.4 73 1.00.017 7.5 77 1.50.018 8.6 67 1.10.023 8.3 70 4.00.027 8.7 104 290.030 7.2 120 350.060 15–17 610–690 15

E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188 185

Fig. 4. SEM photographs of regenerated SF fibres. (a) and (b): undrawn fibres [Sample 1]; (c) and (d): fibres drawn during the coagulation step (DR1) [Sample 3];(e) and (f): fibres drawn at the take up (DR1) and at the roller (DR2) [Sample 6]. SeeTable 1for sample codes.

3.5.2. Spectroscopic and thermal analysisStructural and conformational properties of regenerated SF

fibres were investigated by FT-IR (ATR mode) and DSC. TheIR spectra of fibres undrawn and drawn at different ratios arereported inFig. 5, together with that of the original nativesilk sample that was used as a reference. The whole sam-ples, either native or regenerated, showed the typical IR pat-tern of SF fibres with Silk II (�-sheet) structure, as evi-denced by the position and intensity of the amide I, II, andIII bands at 1698–1622, 1520, and 1261–1234 cm−1, respec-tively [41]. Crystallization likely occurred on drying, afterthe passage of as spun fibres into the coagulation bath con-taining ethanol. This solvent, as well as other alcohols, isa well-known annealing agent able to induce SF crystalliza-tion by random-coil→ �-sheet conformational transition[42].Other weak but significant IR bands present in both nativeand regenerated fibres are those at 1015 and 1003–978 cm−1,respectively attributed to Gly-Gly and Gly-Ala motifs whichcharacterize the highly repetitive polypeptide sequence of SF[43,44].

IR spectra provide other interesting information on the struc-ture of SF fibres. TheI1261/I1234 intensity ratio between thetwo amide III components, which is indicative of the degreeof crystallization of SF[45], was calculated and plotted inFig. 6

Fig. 5. FT-IR spectra (ATR mode) of native (d) and regenerated SF fibres. (a)Undrawn fibres [Sample 1]; (b) fibres drawn during the coagulation step (DR1)[Sample 3]; (c) fibres drawn at the take up (DR1) and at the roller (DR2) [Sample7]. SeeTable 1for sample codes.

186 E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188

Fig. 6. Trend of full-width at half-maximum (FWHM) of the IR amide I band(open triangles) and of the IR crystallinity index (black circles) of regeneratedSF fibres as a function of the draw ratio.

as a function of the draw ratio. As it appears evident fromthe graph, even regenerated fibres produced without any drawattained a value of the IR crystallinity index of 0.59, very closeto that of native silk fibres (0.61). This suggests that crystalliza-tion of SF to Silk II structure occurred almost entirely duringthe coagulation process and that it was independent on draw-ing or, at least, it was not significantly influenced by drawingwithin the range of draw ratio examined. Similar results werereported for regenerated SF fibres obtained by a wet spinningprocess using formic acid as solvent and methanol as coagulat-ing agent[18]. Another interesting IR feature is the behaviourof the full-width at half-maximum (FWHM) of the amide Iband at 1622 cm−1, which decreased exponentially as a func-tion of the draw ratio (Fig. 6). This trend cannot be attributedto increase of crystallinity because theI1261/I1234 intensity ratiodid not change. The behaviour of FWHM can be reasonablyrelated to an increase of the degree of overall molecular ori-entation caused by drawing, which entailed an improved axialalignment of SF chains comprising both the amorphous andcrystalline domains. These results are in good agreement withbirefringence measurements (Table 1). Finally, the significantincrease of longitudinal order induced by drawing was also confirmed by the enhanced dichroic character of the IR spectrumof Sample 7 inTable 1obtained at DR1 = 15 and DR2 = 2.7,as evidenced by the change of relative intensity of amide Iand II bands, and of the weak bands at around 1000 cm−1,w celw own[

ratedS hermp lt-ip e thea ighert itionp tem-

Fig. 7. DSC curves of native (a) and regenerated SF fibres. (b) Undrawn fibres[Sample 1]; (c) fibres drawn at the take up (DR1) and at the roller (DR2) [Sample7]. SeeTable 1for sample codes.

perature, in a range characteristic of non-oriented crystallineSF materials[41]. With increasing the degree of molecular ori-entation, such as in the fibre Sample 7 inTable 1obtained atDR1 = 15 and DR2 = 2.7, the peak temperature shifted to highervalues. These findings are in good agreement with birefringenceand IR data.

3.5.3. Tensile propertiesTensile tests were performed to study the effect of spinning

conditions on the basic mechanical properties of regenerated SFfibres (Table 1). The stress–strain curves of some representa-tive samples are reported inFig. 8. First of all, it is important

Fig. 8. Stress–strain of regenerated SF fibres: (a) draw ratio applied during thecoagulation step (DR1) [Sample 3]; (b) draw ratio applied at the take up (DR1)and at the roller (DR2) [Sample 7]. SeeTable 1for sample codes.

hich was observed when fibre orientation in the measuringas changed from parallel to perpendicular (spectra not sh

46].Fig. 7 shows DSC thermograms of native and regene

F fibres. The curves are characterized by intense endoteaking at around 300◦C, which are attributed to me

ng/decomposition of SF chains with�-sheet structure[41]. Theeak of native silk fibres fell at higher temperature becausxially oriented arrangement of SF chains conferred a h

hermal stability on the sample. The melting/decomposeak of undrawn regenerated fibres fell at markedly lower

-

l)

s

E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188 187

to notice that the specimens showed reproducible and uniformmechanical behaviours. Tensile curves typical of brittle mate-rials were obtained for SF fibres undrawn or subjected to drawduring the coagulation step (Fig. 8a). When both coagulation andpost-spinning draw were applied the shape of the curves remark-ably changed (Fig. 8b). The rupture took place at much higherstress values and elongation at breaking significantly increased.Fibres showed a typical ductile-stable behaviour, with a first lin-early increasing trend, followed by a deviation from linearity,which occurred before the strain reached its maximum value. Aregion of plastic deformation with constancy of stress on elon-gation was then observed, before the failure took place and thespecimen broke.

Fig. 9shows the dependence of Young’s modulus (E), break-ing strength (σb), and elongation at break (εb) on the draw ratiosapplied at the take up (DR1) and at the roller (DR2). The valuesof E increased with increasing DR1. In particular, going from 1to 23 the modulus increased from 2.6 to 8.6 GPa. The applica-tion of a post-spinning draw at the roller (DR2) in addition toDR1 did not entail any further increase ofE. Actually, a con-stancy of the values of Young’s modulus at about 8 GPa wasobserved, irrespective of the change of DR2. The behaviour of

the maximum tensile stress was different. As DR1 increased thebreaking strength (σb) increased from 43 to 67 MPa. The appli-cation of a post-spinning draw (DR2) involved a further increaseof σb until 120 MPa. DR1 was almost ineffective on elongationat break (εb) of regenerated SF fibres, which always showed avery brittle behaviour. On the other hand,εb displayed a strongdependence on DR2. The increase of the post-spinning draw upto 2.7 induced a sharp variation ofεb, which increased steadilyfrom 2% up to a maximum value of 35% (Table 1). It is worthnoting that this values is by far greater than that reported fornative silk (εb = 15%)[38].

The tensile behaviour of regenerated SF fibres can beexplained in terms of spinning parameters and fibre properties,with emphasis on molecular orientation because crystallinitydid not change as a function of the spinning conditions. Theclose dependence of the values of modulus on DR1 can beclearly related to the corresponding increase of longitudinalorder, as evidenced by the first sharp rise of birefringence up to0.018. Further increase of birefringence was almost ineffectiveon modulus, but caused significant rise of breaking strength andelongation at break. As previously noticed, DR2, which broughtabout a second marked rise of birefringence, played a key role

Fe

ig. 9. Tensile properties of regenerated SF fibres as function of DR1 (open circlelongation to break.

s) and DR2 (black squares): (a) Young’s modulus; (b) breaking strength; (c)

188 E. Marsano et al. / International Journal of Biological Macromolecules 37 (2005) 179–188

in the perfection of fibre structure. In fact, regenerated SF fibreschanged from brittle to ductile behaviour and their mechani-cal properties significantly improved, though they still remainedlower than those of native silk fibres.

4. Conclusions

This study shows the possibility to obtain regenerated SFfibres, by using a dry-jet wet spinning line, starting fromSF–NMMO solutions. The thermal and rheological behaviourof the SF–MMNO solutions and the drawing effect on the struc-tural characteristics and mechanical properties of SF filamentswere investigated. The crystalline structure and orientation wereexamined by IR and birefringence measurements and it wasobserved that the crystalline and amorphous orientations of theregenerated SF filaments showed different features dependingon the drawing. The crystallization of SF to Silk II structureoccurred almost entirely during the coagulation process and itwas independent on drawing. An increase of the degree of over-all molecular orientation caused by drawing that improved axialalignment of SF chains comprising both the amorphous and crys-talline domains was observed. Tensile property was stronglydependent on the draw ratio, which affects the molecular orien-tation. Although our results do not compete with the mechanicalproperties of the native silk (E = 15–17 GPa;σb = 610–690 MPa)[38], they suggest that the spinning scheme and the solvent, weu wito ningd me-t fibrep ntrat

A

ject“ and”( ano,I hish

R

als

lan,emi-

434.54

6)

ter.

04)

tien,476.

[9] E. Marsano, M. Canetti, G. Conio, P. Corsini, G. Freddi, J. Appl. Polym.Sci., in press.

[10] S. Sampaio, P. Taddei, P. Monti, J. Buchert, G. Freddi, J. Biotechnol.116 (2005) 21–33.

[11] S. Yazawa, J. Chem. Soc. Jpn. 63 (1960) 1428–1434.[12] H. Ishizaka, Y. Watanabe, K. Ishida, O. Fukumoto, J. Seric. Sci. Jpn.

58 (1989) 87–95.[13] K. Matsumoto, H. Uejima, T. Iwasaki, Y. Sano, J. Appl. Polym. Sci. 60

(1996) 503–511.[14] R.L. Lock, US Patent 5,171,505, 1992.[15] K.A. Trabbic, P. Yager, Macromolecules 31 (1998) 462–471.[16] L.W. Jelinski, O. Liivak, A. Blye, N. Shah, Macromolecules 31 (1998)

2947–2951.[17] J. Yao, H. Masuda, C. Zhao, T. Asakura, Macromolecules 35 (2002)

6–9.[18] I.C. Um, C.S. Ki, H.Y. Kweon, K.G. Lee, D.W. Ihm, Y.H. Park, Int. J.

Biol. Macromol. 34 (2004) 107–119.[19] S.W. Ha, A.E. Tonelli, S.M. Hudson, Biomacromolecules 6 (2005)

1722–1731.[20] M.E. Hall, A.R. Horrocks, H. Seddon, Poly. Degrad. Stabil. 64 (1999)

505–510.[21] S.A. Mortimer, A. Peguy, J. Appl. Polym. Sci. 60 (1996) 305–316.[22] S. Davies, Text. Outlook Int. 44 (1993) 8–12.[23] P.G. Bullio, Am. Text. Int. 21 (1992) FW16.[24] D. Loubinoux, S. Chaunis, Text. Res. J. 57 (1987) 61–65.[25] T. Rosenau, A. Potthast, H. Sixta, P. Kosma, Prog. Polym. Sci. 26 (2001)

1763–1837.[26] G. Raven, Int. Fibre J. 6 (1993) 10–14.[27] W.S. Lee, S.M. Jo, Chem. Fibre Int. 49 (1999) 46–48.[28] W. Berger, Lenzinger Berichte 9 (1994) 11–18.[29] D. Eichinger, M. Eibl, Techtextil, May 13, 1997.[ 999)

[ 40

[ 5)

[ 01)

[ ntaldon,

[ lym.

[ 01)

[ ibre

[ Sci.

[[ tone,

[ ka-

[ ppl.

[ . 59

[ , T.cleic

[ 21–

[ and3, pp.

sed, should be capable of generating materials competitivether natural fibres. This would require higher post-spinrawing ratio allowing for a further decreases of fibre dia

er and to investigate the role of other parameters on theroperties as the SF–NMMO monohydrate solution conce

ion and the solvent used in the coagulating baths.

cknowledgements

The work was carried out in the framework of the proNew fibres from renewable resources: fibres on dem2004.1583/10.8441) funded by Fondazione Cariplo, Miltaly and the authors wish to thank Mr. Mario Traverso forelpful work on dry-jet wet spinning line.

eferences

[1] I. Dal Pra, G. Freddi, J. Minic, A. Chiarini, U. Armato, Biomateri26 (2005) 1987–1999.

[2] T. Bunning, H. Iang, W. Adams, R.L. Crane, B. Farmer, D.L. KapSilk Polymers. Materials Science and Biotechnology, American Chcal Society, Washington, 1994.

[3] N. Minoura, M. Tsukada, M. Nagura, Biomaterials 11 (1990) 430–[4] M. Tsukada, G. Freddi, N. Minoura, G. Allara, J. Appl. Polym. Sci.

(1994) 507–514.[5] Y. Liu, X. Zhang, X. Liu, T. Yu, J. Deng, J. Biotechnol. 46 (199

131–138.[6] S. Sofia, M.B. McCarthy, G. Gronowicz, D.L. Kaplan, J. Biomed. Ma

Res. 46 (2001) 139–148.[7] Y. Gotoh, S. Niimi, T. Hayakawa, T. Miyashita, Biomaterials 25 (20

1131–1140.[8] A. Lazaris, S. Arcidiacono, Y. Huang, J.F. Zhou, F. Duguay, N. Chre

E.A. Welsh, J.W. Soares, C.N. Karatzas, Nature 295 (2002) 472–

h

s-

30] G. Freddi, G. Pessina, M. Tsukada, Int. J. Biol. Macromol. 24 (1251–263.

31] G. Freddi, A. Berlin, M. Tsukada, E. Dubini Paglia, Sericologia(2000) 363–373.

32] Y. Xu, Y. Zhang, H. Shao, X. Hu, Int. J. Biol. Macromol. 35 (200155–161.

33] D.B. Kim, W.S. Lee, S.M. Jo, Y.M. Lee, B.C. Kim, Polym. J. 33 (20139–146.

34] I. Quentin, H. Chanzy, M. Paillet, A. Peguy, Integration of FundamePolymer Science and Technology, Elsevier Applied Science, Lon1985.

35] H. Chanzy, A. Peguy, S. Chaunis, P. Monzie, J. Polym. Sci.: PoPhys. Ed. 18 (1980) 1137–1144.

36] D.B. Kim, W.S. Lee, S.M. Jo, Y.M. Lee, B.C. Kim, Polym. J. 33 (2018–26.

37] A. Ziabicki, Fundamentals of Fibre Formation—The Science of FSpinning and Drawing, John Wiley & Sons, London, 1976.

38] J. Perez-Rigueiro, C. Viney, J. Llorca, M. Elices, J. Appl. Polym.75 (2000) 1270–1277.

39] H.-J. Jin, D.L. Kaplan, Nature 424 (2003) 1057–1061.40] S. Putthanarat, S. Zarkoob, J. Magoshi, J.A. Chen, R.K. Eby, M. S

W.W. Adams, Polymer 43 (2002) 3405–3412.41] J. Magoshi, M. Mizuide, Y. Magoshi, K. Takahashi, M. Kubo, S. Na

mura, J. Polym. Sci: Polym. Phys. Ed. 17 (1979) 515–520.42] J. Magoshi, M. Magoshi, S. Nakamura, J. Appl. Polym. Sci.: A

Polym. Symp. 41 (1985) 187–204.43] M. Asai, M. Tsuboi, T. Shimanouchi, S. Mizushima, J. Phys. Chem

(1955) 322–325.44] C.Z. Zhou, F. Confalomieri, N. Medina, Y. Zivanovic, C. Esuault

Yang, M. Jacquet, J. Janin, M. Duguet, R. Perasso, Z.G. Li, NuAcid Res. 28 (2000) 2413–2419.

45] N.V. Bhat, G.S. Nadiger, J. Appl. Polym. Sci. 25 (1980) 9932.

46] R.D.B. Fraser, T.P. MacRae, Conformation in Fibrous ProteinsRelated Synthetic Polypeptides, Academic Press, New York, 197293–343.