Morphological, nanomechanical and cellular structural

Journal of Microscopy, Vol. 220, Pt 2 November 2005, pp. 96 112 Received 29 April 2005; accepted 28 July 2005 Morphological, nanomechanical and cellular structural Blackwell Publishing, Ltd. characterization of human hair and conditioner distribution using torsional resonance mode with an atomic force microscope N. CHEN & B. BHUSHAN Nanotribology Laboratory for Information Storage and MEMS/NEMS (NLIM), 650 Ackerman Road, Suite 255, Ohio State University, Columbus, OH 43210, U.S.A. Key words. Atomic force microscopy (AFM), cellular structure, conditioner, human hair, in-plane heterogeneity, torsional resonance (TR) mode. Received 29 April 2005; accepted 28 July 2005 Summary Characterization of the cellular structure and chemical and physical properties of hair are essential to develop better cosmetic products and advance the biological and cosmetic sciences. Although the morphology of the fine cellular structure of human hair has traditionally been investigated using scanning electron microscopy and transmission electron microscopy, atomic force microscopy can be used for characterization in ambient conditions without requiring specific sample preparations and surface treatment. In this study, the tapping and torsional resonance modes in an atomic force microscope are compared for measurements of stiffness and viscoelastic properties. The materials were mapped using amplitude and phase angle imaging. The torsional resonance mode showed advantages in resolving the in-plane (lateral) heterogeneity of materials. This mode was used for investigating and characterizing the fine cellular structure of human hair. Various cellular structures (such as the cortex and the cuticle) of human hair and fine sublamellar structures of the cuticle, such as the A-layer, the exocuticle, the endocuticle and the cell membrane complex were easily identified. The distribution and thickness of conditioner on the treated hair surface affects the tribological properties of hair. The thickness of the conditioner was estimated using force distance measurements with an atomic force microscope. Introduction Hair care is a huge industry. Various hair products and treatments are topically applied, ranging from hair shampoos and conditioners to permanent waving, bleaching, and dyeing treatments (Zviak, 1986; Robbins, 1994; Feughelman, 1997; Correspondence to: Bharat Bhushan. Tel.: 1-614-292-0651; fax: 1-614-292-0325; e-mail: bhushan.2@osu.edu Jollès et al., 1997). Shampoo is used to clean hair and conditioner is used to protect it and provide a desirable look and feel. These products can alter the tribological (surface roughness, friction and adhesion) properties of the hair surface. Permanent waving, bleaching and dyeing treatments are all involved in altering the chemical and physical properties of the internal cellular structure of the hair. Characterization of the cellular structure and chemical and physical properties of human hair are of great interest to cosmetic scientists seeking to develop better cosmetic products and advance the biological and cosmetic sciences. The schematic diagram in Fig. 1(a) provides an overall view of various cellular structures of human hair (Robbins, 1994; Swift, 1997). Human hair is a complex tissue consisting of several morphological components (Fig. 1a) and each component consists of several different chemical species. Depending on its moisture content, human hair consists of approximately 65 95% proteins, which are condensation polymers of amino acids. The remaining constituents are water, lipids (structural and free), pigments and trace elements. Table 1 summarizes the main chemical species present in human hair. Among numerous amino acids in human hair, cystine is one of the most important and has been involved in most studies of the individual amino acids in human hair. The differences in cystine content between various cellular structures in human hair result in significant differences in their physical properties. Generally, hair fibres (about 50 100 µm in diameter) consist of a central core (the cortex) of closely packed spindleshaped cortical cells (1 6 µm thick and 100 µm long), each filled mainly with macrofibrils (about 0.1 0.4 µm in diameter), which in turn consist of intermediate filaments (also called microfibrils, about 7.5 nm in diameter) and the matrix. The intermediate filaments are low in cystine ( 6%), and the matrix is rich in cystine ( 21%). The long axes of the cells and their fibrous constituents are orientated along the long axis of the hair fibre. The cortex takes up the majority of the hair fibre 2005 The Royal Microscopical Society

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 97 Fig. 1. (a) Schematic diagram of hair fibre structure (Robbins, 1994) and sublamellar structure of the human hair cuticle. (b) SEM of the surface of human hair (Wei et al., 2005) and TEM of a cross-section through the cuticle for a section stained with ammonical silver nitrate (Swift, 1997). composition and is thought to play a large role in the mechanical properties of the hair. The cortex is covered and protected by sheet-like cuticle cells which are internally laminated. These cuticle cells (scales) form a pattern similar to tiles on a roof: overlapping from the root end towards the tip of the fibre. Each cuticle cell is approximately 0.3 0.5 µm thick and the visible length of each cuticle cell is approximately 5 10 µm. Cuticles in human hair are generally 5 10 scales thick. Each cuticle cell consists of various sublamellar layers (the epicuticle, the A-layer, the exocuticle, the endocuticle and the inner layer) and the cell membrane complex. The outer layer is the epicuticle which is covered with a thin layer of covalently attached lipid 18-methyl eicosanoic acid (18-MEA). The A-layer is a component with a high cystine content (> 30%) and is located on the outer-facing aspect of each cell. This layer is highly crosslinked which gives it considerable mechanical toughness and chemical resilience, and the swelling in water is presumed to be minimal. The exocuticle, which is immediately adjacent to the A-layer, also has a high cystine content ( 15%). On the inner-facing aspect of each cuticle cell is a thin layer of material which is known as the inner layer. Between the exocuticle and inner layer is the endocuticle which is low in cystine ( 3%). The cell membrane complex itself is a lamellar structure, which consists of the inner β-layer, the δ-layer and the outer β-layer. To date, most information about the detailed structure of human hair has been obtained using scanning electron microscopy (SEM) and transmission electron microscopy (TEM) (Swift, 1991, 1997; Robbins, 1994; Wei et al., 2005). Figure 1(b) shows typical images of human hair obtained by SEM and TEM. SEM has long been the standard means of investigating the surface topography of human hair. SEM uses an electron beam to give a photographic image of the sample, but cannot provide quantitative data regarding the surface. SEM requires the hair sample to be covered with a very thin layer of a conductive material and the microscope needs to be operated under vacuum during both metallization and measurements. Surface metallization and vacuum exposure could potentially induce modifications to the surface details. TEM examinations provide fine detailed internal structure of human hair. However, thin sections of 50 100 nm thickness and a heavy metal compounds staining treatment are required for TEM. The cutting of these thin sections with the aid of an ultra-microtome is not an easy task. Moreover, because neither SEM nor TEM techniques can measure the physical properties (mechanical, tribological, etc.) of various cellular structures of interest in human hair and do not allow ambient imaging conditions, many outstanding issues remain to be answered. For example: how do the various cellular structures of hair behave physically in various environments (changing temperature, humidity, etc.)? How do they swell in water? For conditioner-treated hair, how thick is the conditioner layer and how is the conditioner distribution on hair surface? Atomic force microscopy (AFM) has been commonly used for characterization of tribological and mechanical properties of surfaces (Bhushan, 1999, 2004). As a noninvasive technique, AFM can evaluate the effect of sample treatment because it can be used without requiring any specific treatment of the surface. Most importantly, it can be operated in ambient conditions in order to study the effect of environment. AFM provides the potential to be able to see the cellular structure and molecular assembly of hair, for determining various properties of hair, such as friction, adhesion, wear, elastic stiffness and viscoelastic properties, and for investigating the physical behaviour of various cellular structures of hair in various environments. So far, most AFM studies on hair fibres have focused on surface topographic

98 N. CHEN & B. BHUSHAN Table 1. Summary of chemical species presented in human hair. imaging (Smith & Swift, 2002; LaTorre & Bhushan, 2005a) and friction, adhesion and wear properties (McMullen & Kelty, 2001; LaTorre & Bhushan, 2005a,b). In addition, a nanoindentation technique has been used to investigate nanomechanical properties, such as the hardness and stiffness of various cellular layers of keratin fibres (Parbhu et al., 1999; Wei et al., 2005). However, these studies had limited capabilities to identify the fine details of various cellular structures of hair, which prevented further in situ studies of the physical behaviours of these cellular structures of hair in various environments. In this study, different AFM operating modes, the tapping mode and the recently introduced torsional resonance (TR) mode (Huang & Su, 2004; Kasai et al., 2004), were compared for measurements of material stiffness and viscoelastic property mapping using amplitude and phase angle imaging. The new TR mode was used to investigate and characterize the fine cellular structures of human hair. The conditioner thickness on conditioner-treated hair surfaces was also estimated using force distance measurement with AFM. Materials and methods Samples Caucasian hair samples were obtained from Procter & Gamble (Cincinnati, OH) (see Appendix). Three categories of hair samples were studied and are referred to here as: virgin, chemically damaged and damaged treated hair. Virgin samples were considered to be baseline specimens and were free from chemical damage. Chemically damaged hair fibres had been exposed to two cycles of permanent wave treatment, washing and drying, which were taken to be representative of common hair management and alteration. From now on, this category is referred to simply as damaged hair in the text and figures. Damaged treated samples had been treated with three cycles of a Procter & Gamble commercial conditioner, Pantene Smooth & Sleek. All hair samples had undergone two rinse/wash cycles of commercial shampoo treatment (Procter and Gamble) (in the case of treated samples this was prior to conditioner treatments). Samples arrived as hair swatches approximately 0.3 m long. Although the exact location from the root was unknown, it was estimated that hair samples used were between 0.1 and 0.2 m from the scalp. In most cases, the tests were conducted on the middle parts of the hair samples. In the case of virgin hair, the samples were cut both from the locations near the scalp and near the tip in order to compare the topographical properties of hair at different locations. Hair specimens ( 10 mm) were mounted onto AFM sample pucks using Liquid Paper correction fluid (Sanford Corporation, Oak Brook, IL). A thin layer of the fluid was brushed onto the puck, and when the fluid hardened into a tacky state, the hair sample was carefully placed onto it. The Liquid Paper dried quickly to keep the hair firmly in place. An optical microscope was used to preliminarily image the specimen to ensure none of the Liquid Paper was deposited on the hair surface. For hair cross-section and longitudinal section studies (no special sample preparations were needed for hair surface studies), the hair sample was put vertically into a mixture of Buehler Epoxicure TM resin and Buehler Epoxicure TM hardener (Buehler, Lake Bluff, IL) of weight ratio 5 : 1. One end of the hair was attached to a thin pole, and the other end was tied with a weight. After 24 h curing, the epoxy containing the hair was taken out and cut (Wei et al., 2005). Then the cross-section or longitudinal section face of the hair sample was polished with a microtone. The samples were mounted on AFM sample pucks using double-sided tape. In order to better understand the different AFM operating modes, measurements on metal particle (MP) tapes (Imation Corp., Oakdale, MN) were also conducted. The cross-sectional

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 99 Fig. 2. Cross-sectional schematic diagram of metal particle (MP) tape. schematic of the MP tape (Bhushan, 1996) is shown in Fig. 2. The magnetic coating of an MP tape consists of a nanocomposite structure containing magnetic and nonmagnetic particles and lubricant in a polymeric binder matrix. The magnetic particles are about 70 nm long with an aspect ratio of 10. The composites of the magnetic coating of MP tape have distinct stiffness and viscoelastic properties, which made MP tape an ideal nanocomposite surface for our study. Apparatus and measurement techniques An atomic force microscope (Multimode Nanoscope IIIa, DI- Veeco, Santa Barbara, CA) with modifications for the TR mode and a nanoscope extender electronic circuit for the measurement of the phase angle were used in this study. All measurements were conducted in ambient conditions (22 ± 1 C, 50 ± 5% relative humidity). The probes used in this study were single beam etched Si probes (MikroMasch) with a fundamental flexural mode frequency of 75 khz and a fundamental TR frequency of 835 khz with a quality factor of around 1000. The dimension of the cantilever was typically 230 40 3 µm with a flexural spring constant of 1 5 N m 1 and a torsional spring constant estimated to be 30 150 N m 1. The radius of curvature of the tip was about 10 nm. Surface height images were processed using the first order planefit command available in AFM software, which eliminates tilt in the image. Amplitude and phase angle images were processed using zero-order flatten command, which only modifies the offset of the image. Figure 3(a) is a schematic diagram of a tip-cantilever assembly in an atomic force microscope. The cantilever can scan a sample with its tip in constant contact, intermittent contact or without contact with the sample surface (Bhushan, 1999, 2004). The scanning is implemented by the motion of a cylindrical piezoelectric tube, which can act as the holder of either the cantilever or the sample. The deflection of the cantilever is generally measured using an optical lever method. A laser beam is projected on and reflected from a location on the upper surface of the cantilever close to the tip and led by a mirror into a four-segment photodiode. The normal and lateral deflections of the cantilever at that location can then be obtained by Fig. 3. (a) Top: schematic diagram of an atomic force microscope. Middle: three different settings of the microscope are compared: the tapping mode (TM), torsional resonance (TR) mode and contact mode. The TR mode is a dynamic approach with a laterally vibrating cantilever tip that can interact with the surface more intensively than the tapping mode; therefore, more detailed near surface information is available. Bottom: the phase angle is defined and two examples of the phase angle response are shown: one is for materials exhibiting viscoelastic properties (left) and the other nearly elastic properties (right). (b) Typical experimental results of TR amplitude and phase angle change as a function of frequency of polydimethylsiloxane (PDMS) before and after engagement.

100 N. CHEN & B. BHUSHAN Fig. 3. Continued calibrating the voltage output of the photo diode. AFM measurements can be performed with one of the several modes: tapping mode, TR mode (Huang & Su, 2004; Kasai et al., 2004) or contact mode, as shown in Fig. 3(a). The phase angle is also defined in Fig. 3(a) as a phase delay between input/output strain and/or stress profiles. Table 2 summarizes the characteristics of the tapping, TR and contact modes. The contact mode is a static mode and uses a nonvibrating tip, therefore a phase analysis is not available. Other modes, such as acoustic atomic force microscopy (AAFM) which involves oscillation of the samples and was primarily developed to allow friction measurements at high velocities, are not discussed here. The TR mode measures surface height and phase angle (and amplitude) images as follows. The tip is vibrated in the TR mode at the resonance frequency of the cantilever beam in air driven by a specially designed cantilever holder. The torsional vibration amplitude of the tip (TR amplitude) is detected by the lateral segments of the photodiode. A feedback loop system coupled to a piezo-electric scanning stage is used to control the vertical z position of the sample, which changes the degree of in-plane (lateral) tip sample interaction of interest. The z displacement of the sample gives a surface height image of the sample. There are two possible operating modes depending on which parameter is controlled (see Table 2): 1 TR mode I: constant TR amplitude; 2 TR mode II: constant normal cantilever deflection (constant load). Both modes are operated at the resonance frequency of the cantilever in air, which is different from the TR friction mode used in a previous study (Bhushan & Kasai, 2004) in which the tip is vibrated at the resonance frequency of the cantilever after engagement. The TR mode II was implemented here for the first time for surface imaging. During the measurement, the cantilever/tip assembly was first vibrated at its resonance at some amplitude before the tip engaged the sample. Then, for TR mode I, the tip engaged the sample at some setpoint. For convenience, the setpoint is reported as a ratio of the vibration amplitude after engagement to the vibration amplitude in free air before the engagement (Bhushan & Qi, 2003; Scott & Bhushan, 2003; Kasai et al., 2004), shown below: Setpoint (%)= vibrationamplitudeof the tip after engagement 100 vibrationamplitudeof the tip inair before engagement A lower setpoint gives a reduced vibration amplitude. This can be done by bringing the tip closer to the sample. The closer mean tip-to-sample distance leads to more interaction between the tip and the sample surface, therefore the vibration amplitude is lowered. A phase angle image can be obtained by measuring the phase lag of the cantilever vibration response during engagement with respect to the cantilever vibration response in free air before engagement. For TR mode II, instead of keeping a constant setpoint, a constant normal load measured using vertical segments of the photodiode is applied. The feedback control and measurements of the tapping mode is similar to that of TR mode I, except that the tip is oscillated vertically. Under in-plane tip sample interaction, the TR frequency, amplitude and phase of the cantilever all change from those when it is far away from the sample surface, and can be used Table 2. Summary of the various atomic force microscope operating modes used for surface imaging. Operating modes Direction of cantilever vibration Parameter controlled Data obtained Tapping Vertical Setpoint (constant amplitude) Surface height, phase angle (normal viscoelasticity) TR mode I Torsional (lateral) Setpoint (constant amplitude) Surface height, phase angle (lateral viscoelasticity) TR mode II Torsional (lateral) Constant load Surface height, amplitude and phase angle (lateral stiffness and viscoelasticity) Contact n/a Constant load Surface height, friction TR, torsional resonance.

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 101 for contrasting and imaging of in-plane lateral surface properties. Figure 3(b) shows the typical amplitude and phase angle response before and after engagement of polydimethylsiloxane (PDMS), a commonly used viscoelastic material in research and industry. Before engagement, the tip vibrates at the resonance frequency of cantilever in air (not interacting with the sample) and the phase angle is reported as zero; after engagement, TR amplitude decreases to about 70% of the amplitude in air due to the in-plane tip sample interaction, the vibration frequency increases and a 10 phase angle shift occurs. For convenience, TR amplitude is recorded in volts, and 1 V corresponds to about 0.5 nm TR amplitude. Although surface topography and phase contrast images are traditionally obtained using a conventional tapping mode technique, it is recognized that the TR mode has some inherent advantages in qualitative/quantitative surface properties imaging over the widely used tapping mode: (1) In tapping mode, the cantilever vibrates vertically to the sample surface and the tip touches the sample surface only intermittently. In TR mode, the cantilever tip vibrates laterally (parallel) to the sample surface. During measurement, the tip surface distance remains close to the sample surface, ensuring more intensive tip sample interaction and more surface material propertiesrelated information; (2) The torsional stiffness of a cantilever is typically two orders of magnitude larger than vertical bending. Therefore, most of the deformation in TR mode occurs in the sample and more in-plane material related information can be revealed with the TR mode; (3) The principal component of tip surface interaction of the tapping mode is perpendicular to the sample surface. Therefore, the tapping mode is insensitive to lateral force gradients. Compared to the tapping mode, which usually only measures normal surface properties, TR mode can readily measure in-plane (lateral) surface properties of samples. Results and discussion Comparison of different AFM modes In order to better understand the different AFM operating modes, measurements on silicon, human hair and MP tape samples using tapping mode, TR mode I (constant amplitude mode) and TR mode II (constant deflection mode) were conducted. Figure 4(a) shows the typical TR amplitude and phase angle changes of silicon, virgin Caucasian hair, and MP tape as a function of tip surface distance (only the approach branch is shown). In these measurements, the feedback loop was turned off and x y position of the tip was fixed. The sample was brought to the tip until an amplitude drop was detected. Then the sample was retracted to let the cantilever vibration fully recover. For an elastic sample, silicon in this case, the TR amplitude decreases monotonically from point A to point B (see Fig. 4a). As the tip surface distance decreases and quickly reaches a plateau indicating that the tip and sample surface are now stuck and oscillate together. The phase angle tends to increase first then decrease sharply as the TR amplitude reaches the plateau. The initial increase on the phase angle is caused by the adhesion between the tip and a superficial layer on the silicon surface which could simply be a contamination layer or an absorbed water layer on silicon. By contrast, for a viscoelastic sample, such as human hair or MP tape, the changes of TR amplitude and phase angle are much more complicated as the tip surface distance decreases. TR amplitude first decreases sharply when the tip starts to touch the sample surface. As the tip moves further down, the TR amplitude can increase and reaches a peak and then starts to decrease gradually. The phase angle tends to increase first because of the adhesion between the tip and the sample surface; then the phase angle decreases sharply as the tip touches the sample surface. Again, the phase angle increases and then decreases as the tip moves further down into the sample. The exact shape and change of TR amplitude and phase angle may vary depending on the sample and the shape of the tip, but the general trend will follow the pattern described above. By using different feedback loop mechanisms, TR AFM can be performed in two different modes: constant amplitude mode (TR mode I) and constant deflection (normal load) mode (TR mode II). As shown in Fig. 4(a), for an elastic sample, as TR amplitude decreases monotonically with the tip surface distance, it is appropriate to operate the atomic force microscope on TR mode I. In this case, a setpoint will be sufficient to obtain the images. By contrast, because the change of TR amplitude of a viscoelastic sample is no longer monotonic, the feedback control will fail if the setpoint is too small. Therefore, TR mode I can be operated only on a narrow range, for example, the possible operating range for our MP tape sample was only between point A and point B (Fig 4a). Any setpoint below point B will have multiple tip-surface distance corresponding to the same TR amplitude; therefore no stable images can be obtained. Moreover, in the range between point A and point B, the adhesion, which may be caused only by the thin water layer absorbed on the sample surface, has significant effects on the measurement results. These results will only reflect the properties of the surface water layer, which aren t the real sample surface properties. In order to detect the properties of the real surface underneath the absorbed water layer, more intensive tip sample interaction has to be applied. One of the ways to detect samples with more intense tip sample interaction is to operate the atomic force microscope in TR mode II. Now, the deflection of the cantilever is kept constant (for example, to the left of point C), a constant normal load is applied on the tip which is sufficient to let the tip penetrate the absorbed water layer on the sample surface and therefore real surface properties are detected. Figure 4(b) shows some typical images of MP tape using different AFM operating modes. At 60% setpoint, TR mode I phase angle image shows a much larger contrast compared to the tapping mode phase angle image. This setpoint is almost

102 N. CHEN & B. BHUSHAN Fig. 4. (a) Typical torsional resonance (TR) amplitude and phase angle change of silicon, virgin hair and MP tape as a function of tip surface distance. There are two possible operating modes for the TR mode: constant amplitude feedback loop control (mode I) which can only be operated at a very narrow range between points A and B and constant deflection feedback loop control (mode II) which can be operated at a very broad range to the left of point C. (b) Images of metal particle (MP) tape captured using different modes: tapping, TR mode I (constant amplitude) and TR mode II (constant deflection). A better contrast, which may result from variations in viscoelastic behaviour, can be seen in the TR mode I phase angle image compared to the tapping mode (TM) phase angle image. TR mode II amplitude and phase angle images have the largest contrast among these techniques. the largest tip sample interaction (smallest setpoint) that could be applied to our MP tape samples in order to get a stable image (see Fig. 4a). Note that the magnetic and nonmagnetic particles are actually buried in (covered by) a polymeric binder matrix on the MP tape surface, the weak tip surface interaction in the tapping mode is not sufficient to reveal the structure underneath the polymer binder matrix, therefore little contrast can be observed in our TM phase image. In TR mode I phase image,

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 103 the distribution of magnetic particles can be clearly seen because of the stronger tip surface interaction. By contrast, the tip interacts with the surface more intensively in TR mode II therefore more detailed in-plane surface information is obtained. Compared to the tapping mode phase angle image and TR mode I phase angle image, by using TR mode II, TR amplitude and TR phase angle images show even larger contrast. As shown in Fig. 4(b), MP tape samples show granular structure with elliptical shape magnetic particle aggregates (50 100 nm in diameter). Previous studies (Bhushan & Qi, 2003; Scott & Bhushan, 2003; Kasai et al., 2004) indicated that the phase shift can be related to the energy dissipation through the viscoelastic deformation process between the tip and the sample. Recent theoretical analysis (Song & Bhushan, 2005) established a quantitative correlation between the lateral surface properties (lateral stiffness and viscoelasticity) of materials and amplitude/phase angle shift in TR measurements. The contrast in the TR amplitude and phase angle images is due to the in-plane (lateral) heterogeneity of the surface. Based on the TR amplitude and phase angle images, the lateral surface properties (lateral stiffness and viscoelasticity) mapping of materials can be obtained. In summary, tapping mode can be used to detect surface normal mechanical properties, but is insensitive to surface in-plane (lateral) properties. TR mode can readily resolve inplane lateral heterogeneity of materials, and is a good complementary technique to the tapping mode for surface imaging. There are two possible TR operational modes: TR mode I (constant TR amplitude) and TR mode II (constant cantilever deflection or constant normal load). Comparing these two modes, TR mode I allows tip interact with the sample more intensively compared to the tapping mode, however, it is still strongly affected by adhesion. Especially, for viscoelastic materials, the possible operating range could be very narrow and sometimes no real surface properties information can be obtained. In this case, TR mode II can overcome this problem and reveal real surface in-plane properties. Compared to the tapping mode and TR mode I, TR mode II amplitude and phase images can give larger contrast which is related to the materials lateral properties (lateral stiffness and viscoelasticity). Unless stated otherwise in the text or figures, the following studies were conducted with the TR mode II technique. Fine details of human hair and various hair cellular structures Traditionally, most cellular structure characterizations of human hair or wool fibre were done using SEM and TEM. Little work has been done to characterize the cellular structure of keratin fibres using AFM, mainly because the tapping mode is insensitive to lateral surface heterogeneity and has limited ability to resolve various cellular components of hair even though these regions have distinct mechanical and viscoelastic properties. Parbhu et al. (1999) studied the ultrastructure of native wool fibres with AFM. In their studies, they managed to resolve various subcellular components of the wool fibres using nanoindentation technique. However, nano-indentation measurement is a destructive technique, which prevents further in situ studies. Smith & Swift (2002) used lateral force microscopy (LFM) and force modulation AFM to examine human hair. Their study indicated that sublamellar components of the cuticle of human hair show friction and hardness contrast. TR mode II has advantages to detect the in-plane lateral heterogeneity of materials. It is a nondestructive method, and as a dynamic technique, it has higher sensitivity than the normal static lateral force (contact mode) microscopy technique. TR mode II was used in this study to characterize human hair and study fine details and various hair cellular structures. Cross-section and longitudinal section of Caucasian hair. Figure 5(a) shows AFM images of a Caucasian hair cross-section. The hair fibre embedded in epoxy resin can be easily seen. From the TR amplitude image and the TR phase image, the cortex, cuticle and epoxy resin regions can be easily identified. In the cuticle region, five layers of cuticle cells are seen, and the total thickness of the cuticle region is about 2 µm for this sample. In the detailed images of the cuticle region, three layers of cuticle cells are shown, and various sublamellar structure of the cuticle can be seen. The thickness of the cuticle cell varies from 200 to 500 nm. The cortex region shows a very fine circular structure of about 50 nm in size, which represents the transverse face of the macrofibril and matrix. At this scale, no intermediate filament structure can be revealed. Figure 5(b) shows AFM images of a longitudinal section of virgin Caucasian hair. Different regions (the cortex and cuticle regions and embedding epoxy resin) are easily seen. As shown in the detailed image of the cuticle region, various sublamellar structures of the cuticle, the A-layer, the exocuticle, the endocuticle and the cell membrane complex, which cannot be easily revealed in a TR surface height image, are easily resolved in a TR amplitude image and a TR phase image because of the different contrast. Most sublamellar structural features of the cuticle shown in the TEM image of Fig. 1(b) can be identified in the TR amplitude and phase angle images. Previously, these sublamellar structures were only able to be distinguished by TEM (Swift, 1997). Various cellular sublamellar structures in the cuticle have very different chemical contents (Robbins, 1994; Swift, 1997): the A-layer is rich in disulphide crosslinks due to a very high cystine content of up to 35%; The exocuticle is also rich in disulphide crosslinks (15% cystine); by contrast, the endocuticle is relatively lightly crosslinked containing only about 3% cystine. Consequently, these layers exhibit distinct stiffness and viscoelastic properties, and the TR mode II imaging technique (TR amplitude and phase angle images) can easily detect these differences. Note that this longitudinal section is not perfectly parallel to the long axis of the hair fibre but at a small angle to it therefore the thickness of various sublamellar layers of the cuticle do not represent the real thickness. In the cortex region, two different morphological regions

104 N. CHEN & B. BHUSHAN Fig. 5. (a) Cross-sectional images of virgin Caucasian hair and fine detailed images of the cuticle region and cortex region. (b) Longitudinal section images of virgin Caucasian hair and fine detailed images of the cuticle region and cortex region. Note that the longitudinal section is not perfectly parallel to the long axis of the hair fibre but is at a small angle to it, therefore the thickness of sublamellar layers of the cuticle is not the real thickness. can be seen: the macrofibril and the matrix. The macrofibris, which are a bundle of the intermediate filaments, align parallel to each other and looks like a tree trunk; the matrix surrounds the macrofibril region. The matrix region has high cystine content compared to the low cystine content of the macrofibril region. This chemical content difference between the macrofibril and the matrix make it possible to reveal the fine internal cellular structure of hair using the AFM TR mode II technique. Surface of human hair. Three categories of Caucasian human hair surfaces were studied: virgin, damaged (two permanent wave treatments) and damaged treated hair (three cycles of conditioner treatment). Virgin hair. Figure 6 shows AFM images of the surface of virgin Caucasian hair. Two typical sample positions are shown: position 1 is near the root end of the hair and position 2 is near the tip end. In position 1, one cuticle edge is shown. (This is also seen in the TR amplitude and phase images of position 1 in Fig. 6 as the black strips because of the topographic effect near the cuticle edge. The topographic effect tends to be significant only when there is a large local geometry change.) The cuticle edge shows little natural weathering damage and is still intact with a step height of about 500 nm, and the general cuticle surface which is covered with a lipid layer (the outer β-layer) is relative uniform at large scale. By contrast, the surface near the tip of the hair (position 2) shows lots of damage which may

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 105 Fig. 5. Continued be simply because of natural weathering and mechanical damage from the effects of normal grooming actions, such as combing, brushing and shampooing. Parts of the cuticle outer sublamellar layers were removed and the underneath layers (the A-layer, endocuticle and inner layer) were exposed. Because the different chemical contents of various sublamellar layers of the cuticle results in different stiffnesses and viscoelastic properties, a large contrast can be seen in the TR amplitude and TR phase angle images. Note that the surface height within each of the individual sublamellar layers (the A-layer, outer β-layer and inner layer) is relatively uniform, therefore the topographic effect on the TR amplitude and phase is minimum. Detailed images of the outer β-layer, the A-layer and the endocuticle are shown at the bottom of Fig. 6. All these layers show distinct morphology which can be readily revealed in the TR amplitude and phase angle images: the outer β-layer shows very fine granular structure; the A-layer shows few discriminatory features; and the endocuticle has a much rougher granular structure (Swift, 1991) than that of the outer β-layer. Previous LFM studies (Smith & Swift, 2002) on keratin fibres indicated that for untreated (virgin) fibres, no image contrast was observed on the outer facing surfaces of the scales. Our results indicate that the outer lipid layer may form fine domains, which results in the fine granular structure shown in the TR amplitude and phase images (Fig. 6). The TR mode II technique has higher sensitivity compared to the LFM technique therefore the fine chemical distribution, which cannot normally be detected by LFM and other techniques, can readily be revealed.

106 N. CHEN & B. BHUSHAN Fig. 6. Images of the surface of virgin Caucasian hair. Two typical samples are shown: position 1 near the root end in which intact cuticle edges are seen and position 2 near the tip end in which damage occurred, part of the cuticle top layers were removed and underneath sublamellar layers exposed. Detailed images of the outer β-layer, A-layer and the endocuticle are shown at the bottom. Damaged hair. Figure 7 shows AFM images of the surface of damaged Caucasian hair. Two typical samples are shown. More damage can be seen compared to the surface of the virgin hair. More cuticle edges were removed and often larger areas of rough granular endocuticle layer were exposed (see sample I). Of the components within each cuticle cell (the A-layer, the exocuticle, the endocuticle, etc.), the endocuticle is the least crosslinked (Robbins, 1994). Under wet conditions it will swell preferentially and is the preferred plane for lamellar fracture under mechanical stress. Indeed, we have seen many examples where the cuticle has come off to leave this granular endocuticle layer of approximately half of the thickness of original scale and located at the scale margins. As shown in sample II, the endocuticle layers were further eliminated, entire pieces of

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 107 Fig. 7. Images of the surface of damaged Caucasian hair. Two typical images are shown: sample I in which large areas of the cuticle tops sublamellar layers were removed and the much rougher endocuticle layer was exposed; and sample II in which entire pieces of the cuticle were removed and only the cuticle ghost edges were left. Detailed images of the outer β-layer with the endocuticle and with the epicuticle are shown at the bottom. cuticles were removed and some fine lines on cuticle surface which delineate the original boundaries of the cuticle edges are clearly seen in TR amplitude images. These lines are referred to cuticle edge ghost in the literature (Swift, 1991; Smith & Swift, 2002). The actual fracture occurs at the interface between the outer β-layer and the δ-layer (see Fig. 1a) to leave a new surface covered with covalently bound lipid (the outer β-layer). The ghost edges are thought to have arisen by the indented impression of each cuticle cell upon the one underlying it during formative moulding process in the hair follicle. At the bottom of Fig. 7, the comparison of fine details of the endocuticle and the outer β-layer are shown. The endocuticle and the outer β-layer show distinct TR phase contrast. Note that the topographic effect is minimal except for the boundary between the endocuticle and the outer β-layer. The distinct phase contrast is the result of the different chemical content between these two sublamellar layers. The endocuticle shows a much rougher granular structure than the outer β-layer. As much larger areas of rougher endocuticle were exposed on the damaged hair surface than on the virgin hair surface, this could be part of the reason why damaged hair loses shine. Detailed views of the cuticle ghost edge are also shown at the bottom of Fig. 7. In TR amplitude and phase angle images, the region near the ghost edge exhibits a different TR phase contrast, indicating the presence of a surface with a different chemical nature from the outer β-layer. We believe that the newly formed surface is the epicuticle layer which was originally covered by the outer β-layer (see Fig. 1a). The outer β-layer was originally present,

108 N. CHEN & B. BHUSHAN Fig. 8. Schematic diagram of the progress of hair damage. Top: top view images of the hair surface at each stage. Bottom: cross-sectional views, taken from the direction of the corresponding arrows in the top images. but because of its location, i.e. under the original overlying cuticle but close to the scale edge, it may have undergone oxidative loss through environmental or chemical exposure. For easy visualization, Fig. 8 illustrates the progress of hair damage (Swift, 1991). Virgin hair has an intact smooth cuticle edge; as damage occurs (natural weathering or mechanical damage), parts of the cuticle outer sublamellar layers wear off and underlying layers (for example, the endocuticle) are exposed. Further damage will cause the entire piece of cuticle to be broken off and the ghost which delineates the original boundary of the cuticle edge is seen. Damaged treated hair (three cycles of conditioner treatment). Various sublamellar layers of cuticle can be exposed on the surfaces of virgin and damaged hair depending on the degree of damage. Because of their distinct chemical nature, these layers may have different interactions with conditioner (or other hair care products) which will affect the adsorption of conditioner on the hair surface. Figure 9 shows AFM images of the surface of damaged treated hair. Two typical samples are shown. In sample I, intact cuticle edges can be seen. From the TR phase angle image, a higher contrast can be seen near cuticle edges. Previous studies (LaTorre & Bhushan, 2005a,b) show that conditioner is unevenly distributed across the hair surface, and that a thicker conditioner layer can be found near cuticle edges. This build-up of conditioner might be caused simply by physical entrapment of the conditioner at the steps of the cuticle edges. Uneven thickness of conditioner caused the contrast on the TR phase angle images. In sample II, a sharp cuticle edge and a cuticle ghost edge can be readily seen in the TR amplitude image. No endocuticle or other sublamellar layers could be found because further treatments removed these layers. As shown in the TR phase angle image of sample II, the region between the cuticle edge and the cuticle ghost edge shows a different contrast from the other parts of hair surface. The epicuticle layer may have been exposed on this newly formed region while the other parts of the hair were still covered with the outer β-layer. The outer β-layer is basically a hydrophobic covalently attached lipid covered layer. The interactions between the conditioner and the outer β-layer or other sublamellar layers of the cuticle (such as the epicuticle) can be very different; therefore, the adsorptions of conditioner on these layers are different. Fine details of the damaged treated general cuticle surface (the outer β-layer) are shown at the bottom of Fig. 9. Compared to the fine granular structure of the outer β-layer shown in Fig. 6, no discriminatory features can be seen because the entire surface is covered with a layer of conditioner. Although the morphology of the fine cellular structures of human hair has traditionally been investigated using SEM and TEM, these techniques have limited capabilities to study environmental effects on the physical behaviour of hair in situ. In this study, AFM TR mode II technique was used for the first time to characterize the fine cellular structure of human hair and many features previously only seen with SEM and TEM were identified. Our results have proved that this technique can readily reveal the fine cellular structure of human hair. The technique provides the possibility to carry out further in situ studies of the effects of environment (temperature, humidity, etc.) and hair-care product treatment on the physical behaviour of human hair.

MORPHOLOGICAL AND CELLULAR STRUCTURAL CHARACTERIZATION OF HUMAN HAIR 109 Fig. 9. Images of the surface of damaged treated hair (three cycles of conditioner treatment). Two hair samples are shown (Samples I and II). Conditioner is unevenly distributed on the hair surface. A thicker conditioner layer can be seen near the cuticle edges. Fine details of the conditioner-covered hair surface are shown at the bottom. No discernible features can be seen. Conditioner thickness Conditioner is used to coat the hair with a thin film for protection and provide a desirable look and feel. Conditioner consists of a gel network chassis (cationic surfactant, fatty alcohols and water) for a superior wet feel and a combination of conditioning active components (silicones, fatty alcohols and cationic surfactant) for a superior dry feel. The wet feel benefits are a creamy texture, ease of spreading, a slippery feel while applying and a soft rinsing feel. The dry feel benefits are moistness, softness and ease of dry combing. Many other ingredients are also added in order to meet the needs of consumers. The conditioner distribution and thickness on the hair surface are essential for its functions. In order to determine the thickness of conditioner on hair, several possible techniques can be applied (Bhushan, 1999): Fourier transform infrared spectroscopy, ellipsometry, angle-resolved X-ray photon spectroscopy and AFM. Ellipsometry and angle-resolved X-ray photon spectroscopy have excellent vertical resolution in the order of 0.1 nm, but their lateral resolutions are in the order of 1 and 0.2 mm, respectively. By contrast, AFM can measure the thickness of a liquid film with a lateral resolution in the order of the tip radius (about 100 nm for our tips), which is not possible to achieve by other techniques. In this study, force-distance measurements were conducted to estimate the local conditioner thickness on damaged-treated hair surfaces. The conditioner thickness was obtained by measuring the forces on the tip as it approaches, contacts and pushes through the conditioner layer. The inset of Fig. 10 is a diagram of an AFM tip interacting with a conditioner-covered hair surface. A typical force vs. distance curve between a silicon tip and the damaged treated hair surface is shown in Fig. 10. The surface is first brought into contact with the tip and then pulled away. The force curve at small separation is zoomed in. Details are shown at the bottom of Fig. 10. Because of the van der Waal s force as well as the meniscus formation between the tip and the conditioner liquid layer, as the surface approaches the tip at a finite distance H s of about 30 nm, a mechanical instability occurred where the conditioner layer suddenly jumped into contact with the AFM tip (snap-in), causing a sharp onset of attractive force. The attractive meniscus force experienced by the tip is 2πRγ(1 + cos θ), where R is the radius of the tip, γ is the surface tension of the liquid and? is the contact angle (Bhushan, 2002). The attractive force measured in Fig. 10 is about 50 nn.

110 N. CHEN & B. BHUSHAN Fig. 10. Forces between the tip and the hair surface as a function of tip sample separation. A schematic diagram of the measurement for localized conditioner thickness is shown in the inset at the top. The expended scale view of the force curve at small separation is shown at the bottom. The hair sample is first brought into contact with the tip and then pulled away. The sample is moved with a velocity of 400 nm s 1, and the zero tip sample separation is defined as the position where the force on the tip is zero when in contact with the sample. A negative force indicates an attractive force. Conditioner-treated hair surfaces show much longer ranges of interaction with the tip compared to the very short range of interaction between virgin hair surfaces and the tip. Typically, the tip will suddenly snap into contact with the conditioner layer at a finite separation H s that is proportional to the actual conditioner thickness h. When the sample is withdrawn, the forces on the tip slowly decrease to zero as a long meniscus of liquid is drawn out from the hair surface. As the sample further approaches the tip, the force on the tip remains almost constant while the tip is in the liquid conditioner film on the hair surface. When the tip contacts the underlying hair surface, the force quickly becomes repulsive as the tip is pushed up against the hard wall of the hair surface. As the sample is pulled away from the tip, the forces on the tip slowly decrease to zero as a long meniscus bridge of liquid is drawn out from the surface and eventually breaks at a tip sample separation of about 160 nm. Note that because the conditioner is unevenly distributed across the hair surface, the measured curve may vary depending on the local thickness of the conditioner, but the general features will be the same as described above. For comparison, the force vs. distance curves between the microscope and the virgin hair surface are also shown in Fig. 10. The range of the interactions between the tip and virgin hair surface is much shorter than those between the tip and conditioner-treated hair surface. As the virgin hair surface approaches the tip, the tip will jump into contact with the hair surface at a smaller separation of around 4 nm because of the van der Waal s attractive interaction as well as a small meniscus formation due to the presence of condensed water from the environment between the hair surface and the tip, then the tip reaches the hard wall contact with the hair surface. The measured attractive force between the tip and the virgin hair surface is only about 20 nn, which is much smaller than that between the tip and treated hair surface. When the virgin hair surface is withdrawn, the tip will simply jump out and no large liquid deformations occur because of the lack of liquid layer. The distance H s between the sharp snap-in at the liquid surface and the hard wall contact with the substrate is not the real conditioner thickness h. It tends to be thicker than the actual film thickness, but can still provide a very good estimate of the actual film thickness. Previous measurements on the localized lubricant film thickness on a particulate type magnetic rigid disk indicated that the measured thickness using an atomic force microscope is about 2 nm larger than the actual thickness based on the ellipsometry measurements (Bhushan & Blackman, 1991). Forcada et al. (1991) theoretically analysed the thickening of a liquid film on a substrate surface due to its interaction with an AFM tip. A number of theoretical publications have addressed the issue of liquid coalescence in terms of the effective stiffness of a liquid surface or interface (Attard & Miklavcic, 2001; Bhatt et al., 2001), and have concluded that a liquid surface behaves like a Hookian spring with an effective stiffness K eff equal to its surface or interfacial tension γ. For one of the main components of conditioner, silicone (PDMS) in air, we therefore expect an effective stiffness of about 20 mn m 1 (Brandup et al., 1999), which is a very low value compared to the spring constant, 5 N m 1, of the cantilever used in our study. It suggests that liquid surfaces become highly distorted by even the smallest force. Recent surface forces apparatus experiments (Chen et al., 2004) indicated that the snap-in (jump in) distance will be closer to the actual film thickness with an increase of the approach rate, the liquid viscosity and the interfacial tension γ, and the decrease of the initial film thickness h. For a thickness of 25 nm liquid PDMS film in air, the snap-in distance is about 100 nm at a very slow (0.3 nm s 1 ) approach rate. At a faster approach rate (400 nm s 1