Confocal Raman Spectroscopy of Whole Hairs Paul D. A. Pudney,a,* Eleanor Y. M. Bonnist,a Kevin J. Mutch,a Rachel Nicholls,a Hugh Rieley,b Samuel Stanfielda a b
Unilever Discover, Colworth Laboratory, Sharnbrook, Bedfordshire, MK44 1LQ UK Unilever R&D Port Sunlight Laboratory Quarry Road East, Bebington, Wirral, Merseyside, CH63 3JW UK
This paper describes the application of Raman spectroscopy to whole hair fibers. Previously this has proved difficult because the hairs are relatively opaque, and spatial resolution diminishes with depth because of the change in refractive index. A solution is to couple confocal Raman with multivariate curve resolution (MCR) data analysis, which separates spectral differences with depth despite this reduction in resolution. Initially, it is shown that the cuticle can be separated from the cortex, showing the differences in the proteins, which can then be plotted as a function of depth, with the cuticle factor being seen only at the surface as expected. Hairs that had been treated in different ways, e.g., by bleaching, treatment with the active molecule resorcinol followed by rinsing and treatment with a full hair care product, were also examined. In all cases, changes to the hair are identified and are associated with specific parts of the fiber. Since the hair fiber is kept intact, it can be repeatedly treated and measured, hence multistep treatment processes can be followed. This method expands the potential use of Raman spectroscopy in hair research. Index Headings: Raman; Hair; Keratin; Penetration; Bleaching.
INTRODUCTION Hair is one of the defining characteristics of mammals. It is made mainly of protein in a filamentous structure. It has a number of functions such as providing warmth and camouflage, as well as for sending signals for danger warnings and mating in some species. In humans many of these functions have diminished, but it does still provide some function in thermoregulation. The head is the main area where hair is in abundance and where people can relatively easily vary their appearance. Since people spend a great deal of time and effort cleaning, coloring, and styling their hair, it is of significant importance to the cosmetic industry. The structure of the hair fiber has been described elsewhere.1 It consists mainly of the protein keratin in a filamentous structure and has three main parts: The cortex, which constitutes the bulk of the hair, and is formed of microfibrils and macrofibrils arranged parallel to each other and separated by a matrix; the cuticle, flattened cells arranged perpendicular to the cortical cells that form the outer protective layer; and the medulla is at the center of the fiber; however, this is often not continuous or absent from some human hair fibers. Other components in hair fibers include lipids, water, and melanin, the pigment responsible for hair color. The constituent protein keratin varies in the order and types of amino acids present in the different parts of the hair. These areas thus have different interactions between groups, i.e., hydrogen bonds, polar interactions, Coulombic attractions, and covalent bonds, which lead to different conformations and Received 27 March 2013; accepted 3 August 2013. * Author to whom correspondence should be sent. E-mail: paul.pudney@ unilever.com. DOI: 10.1366/13-07086
1408
Volume 67, Number 12, 2013
secondary structures. For example the cortex material is mainly a-helical, and the cuticle has high b-sheet character.2 Particularly important are disulfide bonds that form between the –SH groups on cysteine residues, which can be intra- or inter-chain and are more prevalent in the cuticle than the cortex.3 These are the principal forces in maintaining the structure and properties of the hair fiber and are resistant to many chemical treatments.1 Vibrational spectroscopy is sensitive to molecular structure and conformation of proteins4 and so is well placed to investigate the properties of hair. Indeed there are a number of studies using both infrared (IR) and Raman spectroscopy5,6,7,8,9,10,11 and also coherent anti-Stokes Raman scattering spectroscopy (CARS).12,13 When cross-sectioned, the different structures of the cuticle and cortex have been demonstrated by these methods.2,6,14 Changes in hairs have also been followed when treated in various ways, the most widely used treatment being bleaching.3,13,15 A treatment of hair with cosmetic products often involves diffusion of active compounds into hair.16 The penetration of actives has been previously shown by Raman and IR.9,17 Despite much progress having been made, some challenges still remain, which in combination prevent valuable insights into hair properties and hair treatment from being uncovered. Most studies have concerned microtomed hair rather than whole hair fibers, allowing access to the different areas of the fiber by scanning along the cross-section. The drawback to this is that often one may want to follow external molecule penetration over time or examine the hair throughout a successive treatment regime. Microtoming destroys the sample so that this type of repeat measurement is not possible and also negates one of the advantages of Raman spectroscopy of being noninvasive. This approach is taken because, when sampling whole hair, especially when doing depth profiles through whole hair fibers, it is difficult to distinguish clearly the different layers of the hair structure.18 This is because the hair structure is composed predominantly of keratin, and it is only relatively minor inhomogeneities in its chemical and structural composition that distinguish different layers, as highlighted above. The difficulty in unambiguously detecting and mapping these layers is further compounded by the fact that the Raman spectroscopic signals of different proteins and their different structural conformations are all very similar and heavily overlap. Furthermore, the natural optical properties of hair leads to a large change in refractive index, ongoing from air into the hair fiber, which causes the focus of the Raman laser beam to be enlarged, thus reducing the confocal resolution with increasing depth and condensing of the depth scale (the apparent depth problem).19 Recently, Zhang et al.20 made use of the normal method of reducing the loss of depth resolution19 by using an oil-immersion objective in order to refractive-index match the sample. They did this by placing the hair under a
0003-7028/13/6712-1408/0 Ó 2013 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
FIG. 1. Schematic diagram showing how the whole hair fibers were sampled. Spectra are collected along the length of the hair with a depth profile collected at each point.
coverslip to prevent oil from interfering with the hair. This provided a good depth profile and showed protein features changing with depth, although spectral overlap was still apparent when compared with the separated cuticle and cortex spectra2 and when microtomed.6,13,14,21 They did not compare data collected using immersion and dry objectives. Hence, there is a need for an improved methodology that can accommodate the measurement of whole hair fibers and sufficiently discriminate between the different sections of the hair within the data that is generated. Essentially whole hairs are a multicomponent microstructure, and thus a Raman map produces a large spectral dataset. In a conventional analysis of the resultant spectrum, an attempt is made to attribute specific individual spectral intensities to specific chemical components and then derive, as a function of position, the spatial component maps. For the spectra generated by a multicomponent microstructure, this is generally not possible since it is rarely possible to unambiguously attribute specific spectral intensities in this way. This is clearly the case in spectra of hair as described in the above section, as it comprises mainly keratin but differently folded. Consequently, a different approach is needed, in which the aim of the analysis is to produce pure separated spectra and their intensities from the spectral dataset. In a lot of chemometric techniques, the spectra are separated on the basis of variations, in this case in the spatial and spectral dimensions. However, in most methods, this does not produce pure and thus interpretable factors. However, there are methods for doing this, which generally
come under the title of self-modeling curve resolution (SMCR) or multivariate curve resolution (MCR).22 This method has now been successfully used for a number of years on different types of microstructures that have been Raman-mapped. These include emulsions,23 phase separated biopolymers,24 food products,25,26 and plant structures.27 It has also been used on depth profiles into skin,28 which is most similar to the system being examined in this study. This paper reports the spectral separation of the different parts of hair fibers from depth-profile maps of whole hairs with multivariate curve resolution (MCR) analysis. This is demonstrated with yak hair (a good model of human hair without melanin) and blonde or gray human hair. The method is further applied to a number of hair problems, specifically to investigate bleaching and external molecule treatment without microtoming the hair sample. It is found that bleaching mainly affects the cuticle layer, and its extent diminishes with depth. We show the uptake of resorcinol into the hair fiber and observe that the amount inside the hair is reduced on rinsing. Finally, we identify the penetration of a select ingredient into the hair fiber after treatment with a full hair-care product formulation. The use of air objectives allows the most flexibility with treatment conditions, but the approach would be equally applicable to data collected with immersion objectives in those cases where the sample allows it. This approach expands the applicability of Raman in applied hair research.
METHODS Hair Samples. Yak hair was obtained from Hugo Royer Ltd. as premade 0.75 g, 5 cm length switches of yak belly hair, which is a close approximation to human hair. It is very low in color and is usually obtained with little damage and is of consistent quality. Natural white hair was obtained from International Hair Importers, USA; selected by optical microscopy were naturally blonde virgin human hair. To bleach the hair strands, L’Oreal Platine Precision bleach powder and 9% Excel cre`me peroxide were mixed in a 1 : 2 ratio. The bleaching mix was applied to each hair strand using a tinting brush and left to develop for 30 min. Each hair strand was rinsed for 2 min and left to air dry for 2 h. Following this, another bleach powder/peroxide cre`me mixture was prepared and applied once more to the bleached hairs. After leaving for 30 min, the hair strands were washed using a base shampoo (17.10 wt % sodium laureth sulfate, 5.33 wt % cocoamidopropyl betaine,0.40 wt % DMDM hydantoin, 1.40 wt % sodium chloride, in water) and air-dried.
FIG. 2. The two main factors/spectra obtained from mapping a virgin yak’s hair using the Kaiser holoprobe 5000R spectrometer. (a)The blue spectrum is typical of keratin with high a-helix content, typical of the cortex. (b) The red spectrum, shows an S-S rich/b-sheet protein of the cuticle.
APPLIED SPECTROSCOPY
1409
FIG. 3. A plot of the two main factors obtained from MCR analysis of a xz scan through a single yak hair fiber. The y-axis is a line of points along the top of the hair at 2 lm intervals (x-scan), and the x-axis is the z direction, i.e., penetrating through the hair, with points taken at 5 mm intervals. A collection time of 60 s was used at each point.
The procedure for treating yak hair with resorcinol (1,3dihydroxy benzene) was as follows: 0.4 g of 15 cm long yak hair was packed into a glass pipette. This was then rinsed with 1 mL of a 2% by weight solution of resorcinol. For the rinse sample, the treated hair was left for 24 h then rinsed with 2 mL of distilled water. Blonde human hairs were treated with a full formulation NeXXus product, Youth Renewal, whereby the 1% glycerol in the formulation was replaced with deuterated glycerol. This is a leave-on formulation, so hairs were treated with the product and left for approximately 24 h before measurement. Data Collection. Raman spectra were collected using identical procedures on two different spectrometers, both using an excitation wavelength of 785 nm. The first was a WITec Alpha 300 R system. The 785 nm laser was used at an operating power of 50 mW before the objective; this was selected such as to avoid sample damage while still resulting in sufficient sample excitation. A Zeiss 1003 (N.A. 0.9) air/metallurgical objective and a Zeiss Achromoplan 1003 (N.A. 1.25) immersion objective were used. The second spectrometer used was a Kaiser holoprobe 5000R confocal Raman spectrometer equipped with an Olympus air/1003 metallurgical objective (N.A. 0.8), and also employing less than 50 mW of laser power at the sample. Hair strands were fixed to a glass microscope slide using tape, ensuring that the hair fiber was pulled straight without stretching. When using the oil-immersion objective, a cover slip was placed on top of the hair so the immersion oil did not interfere with the hair. A line depth profile along the length of the hair was collected, i.e., an xz-map of the hair fiber, as
shown in Fig. 1. The collection of data along the length of the fiber, rather than across the fiber, prevents some more serious problems concerning the mixing of spatial information.29 A collection time of 60 s for each point was used with both spectrometers unless otherwise stated. Data Analysis. Multivariate curve resolution (MCR) analysis separates the data into two modes: one which describes the spectra of the different components present (the factors) and another describing the intensity of each spectral factor, at each location sampled (the scores)22. To do this, a principal factor analysis (PFA), or equivalent method, is performed on the unscaled data. The factors represent the changes which occur in the spectra and are calculated by separating them out according to their variance from each other, i.e., as different from each other as possible. A relative score is given to each factor to show the distribution of these factors through the sample. After PFA, the resulting factors are not easily interpreted as they may still be mixtures of the components and often do not represent the real components.30 Further constraints must be applied to transform the results into real spectra and pure scores; these constraints are applied using an alternating least squares regression (ALS) algorithm.29 A non-negativity constraint is applied to make the data practically meaningful, i.e., it is not possible to have a negative concentration (score) or negative Raman peaks. An equality constraint is also applied; this aims to normalize the factors ensuring that all intensity variance is found in the scores and not in the spectra.30 The constraints are applied alternatively to the factors and scores until a solution converges, thus
FIG. 4. Spectral factors from MCR analysis of the data from mapping an untreated whole human hair fiber using the WITec Alpha 300 R spectrometer. Blue is the cortex. Green is the cuticle. For label meanings see text.
1410
Volume 67, Number 12, 2013
TABLE I. Assignments of cuticle and cortex spectrum derived from the spectral factors produced from MCR analysis. Assignments using references 2, 3, 5, 14, and 36. Cuticle/cm1 v (SS) (gauche–gauche– gauche) v (CS) (gauche) v (CS) p (CH2) in-phase d (CCH) aliphatic (tyrosine side chains) d (CCH) aromatic (tyrosine side chains) p (CH2) tryptophan v (CC) helix-a/p (CH3) terminal p (CH3) d (CCH) olefinic v (CC) aromatic ring/ phenylalanine v (CC) skeletal/cis conformation v (CC) skeletal/trans conformation v (CC) skeletal v (CC) skeletal Tyrosine Amide III disordered Amide III Amide III a helix CH2 bending mode Amide I
Cortex/cm1
509 644, 663, 671 686, 719 –
509 – – 742/743
–
828
– 880–891
853 –
– – 982, 989
933 959 –
–
1002-1003
1033
–
1044 1052
1047 – 10821103 and 1125 – – – 1342 1448 1657 (1649–1669)
– 1176, 1187, 1199 1241, 1252 1281 – 1451, 1465 1675 (1668–1679)
producing factors representing the real spectra of the components present and scores that represent the relative concentration of each factor at every position sampled. This is the basic ALS approach; the modified alternating least squares
(MALS) algorithm is an improved implementation of this method and has been described elsewhere.31 When conducting a depth profile, one of the main problems is the loss of ongoing signal deeper into the sample, mostly caused by physical scattering and spherical aberration. The same problems are encountered when doing depth profiles into skin.32,33 This problem is overcome by using the protein signal as an internal standard as the concentration of protein is approximately constant with depth. Here, the same approach can be taken using the keratin signal. Another problem with measuring depth is that the depth axis is not real depth into the sample but rather apparent depth; the depth axis here refers to mechanical depth. This is the distance moved by the objective from the surface position of the hair fiber to collect the spectrum at that point.
RESULTS AND DISCUSSION Whole Untreated Hairs. Spectra were collected in a line along the length of the hair taking a depth profile at each point (an xz-map, see Fig. 1). Firstly, yak hair was examined. This xz-map data was then analyzed using the MCR method; the two main spectral factors resulting from this are shown in Fig. 2. The associated scores (intensities) are plotted in Fig. 3. This experiment has been repeated on human whole blonde hairs, and the same analysis carried out; the spectral factors are comparable and shown in Fig. 4. The associated score maps are very similar to those of yak hair that are shown in Fig. 3, so are not shown here. From the spatial distribution and the features of the spectral factors, these can be associated with the cuticle (Figs. 2b and 4 green) and cortex (Figs. 2a and 4 blue). The spectral factors highlight the differences between the two proteins in both molecular structure and conformation; the key differences and peak assignments5,14,34 are listed in Table 1. Here, the features from the human hair are discussed, but this applies equally for the yak hair results as well.
FIG. 5. Plots of the scores of the cortex and cuticle from MCR analysis of xz depth profiles from the same untreated yak’s hair. (a) A plot of the cortex scores using the air objective. (b) The plot of cuticle scores using the air objective. (c) A plot of the cortex scores using the oil objective. (d) The plot of cuticle scores using the oil objective. Data collected using WITec Alpha 300 R spectrometer.
APPLIED SPECTROSCOPY
1411
FIG. 6. Spectral factors from untreated (blue) and bleached blonde human hair (red). (a) Cuticle factors. (b) Cortex factors. Data collected using WITec Alpha 300 R spectrometer.
The spectra reflect differences in amino acid content, specifically that the cuticle is more sulfur-rich than the cortex due to a higher number of cysteine residues14 and that the cortex has more tyrosine and phenylalanine.2 The band at 509 cm1 in Fig. 4a, which arises from the disulfide (-C-S-S-C-) bonds,35 is more intense in the cuticle spectrum. The narrower shape of this band in the cuticle, suggests that the SS bonds here have a high population of the gauche-gauche-gauche conformation. An all-gauche conformer is centered at ;510 cm1, but this increases in frequency when a trans conformation is introduced (these conformers are centered at 525, 540 cm1). The higher intensity of the 645 cm1 band for the cuticle, Fig. 4b, from the C-S bonds, is also consistent with higher levels of cysteine. The bands at 828 and 853 cm1, Fig. 4c, are only evident in the cortex spectrum; these originate from the amino acid tyrosine. The phenylalanine peak at ;1002 cm1 in Fig. 4e is more intense in the cortex spectrum, also fitting with the known amino acid content. Apart from reflecting the different amino acids, the resolved cuticle and cortex spectra further show the differences due to their conformations, the cuticle being in the b-sheet conformation, while the cortex is a-helical. This can be seen from the position of the amide I band, Fig. 4g, which is centered at 1675 cm1 in the cuticle and at 1657 cm1 for the cortex, and of the amide III band, Fig. 4f, which is between 1240 and 1252 cm1 for the cuticle, and at 1342 cm1 for the cortex. In addition to
FIG. 7. Ratio of the sulfur-oxygen peak (1020–1070 cm1) area to phenylalanine peak (990–1016 cm1) area on the same hair before (red) and after bleaching (blue). The five points along each hair have then been averaged to give a representative ratio for the hairs at each treatment stage. Data collected using WITec Alpha 300 R spectrometer.
1412
Volume 67, Number 12, 2013
these are the presence of a band at 933 cm1 in the cortex and the absence of a 959 cm1 band in the cuticle. These spectral factor results for cuticle and cortex are in good agreement with Raman spectra from the same parts isolated from sheep’s hair (i.e., merino wool).2 Spectra from the cuticle and cortex of hair have been identified from a crosssection of hair;14,36 however, the differences observed here are less well defined and often contain some features of the other proteins present. The spectra obtained on a whole hair using an oil objective showed differences with depth consistent with the changes outlined above, including a shift of the amide I band towards the a-helical position and the lower intensity of the S-S band;20 however, the features are not as well separated as seen using the MCR analysis. The plot of the scores of these two spectral factors in Fig. 3 show that the cuticle factor is only observed at the surface of the hair and then appears again when the focus reaches the other side of the hair; although, at this stage, the laser focus is very diffuse, and hence here, it appears much more spatially spread. The cortex factor appears between the location of the cuticle factor, as would be expected. Thus, application of the MCR analysis method has provided a clear resolution of cuticle and the cortex layers in the data from a whole hair fiber, without the need for microtoming. The intact sample is thus available for further measurement, giving scope for continued investigation. Comparing Collection Methods. The use of dry objectives allows the most flexibility with treatment conditions; however, as pointed out in many publications,19,37,38 this leads to degraded resolution in the z-axis. The normal way to negate this is to use immersion objectives to refractive index match. This approach has been used with hair20 where a coverslip is placed on top of the hair. Here, this method is compared with the dry objective for hair. Data was collected from the same yak’s hair using the same collection conditions for the air objective but with an oil-immersion objective and coverslip. This data was analyzed by the MCR methods and results in very similar factors, to those shown in Figs. 2 and 4. The score maps showing the spatial distribution are shown in Fig. 5. Both methods clearly differentiate the cuticle and cortex spatially;
FIG. 8. Spectral factor for resorcinol (blue) compared to spectrum of pure resorcinol (red). Data collected using Kaiser holoprobe 5000R spectrometer.
the main difference is that the signal from the cortex is maintained to greater depth with the oil-immersion objective as the laser focus is maintained much better, whereas the focus through the dry objective is much more attenuated because of the larger refraction effect.38 However, the dry-objective data gives a good signal from the cortex well beyond the cuticle region, suggesting that the cortex can be measured with good reliability. One practical issue that should be mentioned is that good profiles are not always obtained when collecting data using the oil objective because, as the hair fiber is not flat, being both round and rough, the coverslip is not always in good contact with the hair over a large area. Effect of Bleaching on Hair. Whole blonde human hairs were examined before and after bleaching. The data was then analyzed using the MCR method. The cuticle and cortex spectral factors from this analysis are compared before and after the bleaching process, as shown in Fig. 6. The bleached cuticle shows a reduction in the S-S band at 508 cm1 and an increase in the S-O band seen at ;1042 cm1. These changes are also replicated in the cortex factors, see Fig. 6b, but the increase in the S-O band is larger in the cuticle than in the cortex. Thus, the impact of bleaching spans across different layers of the hair fiber. To show this in further detail, a plot of the integrated intensity ratio of the sulfur-oxygen band (1020– 1070 cm1) normalized to the phenyalanine band (990–1016 cm1) with depth is shown for the same hair before and after bleaching, see Fig. 7. This further illustrates that the S-O band increases upon bleaching across the measured depth range and that the extent of this change decreases with depth.
The spectral differences observed here are consistent with the molecular changes known to be caused by bleaching, namely cleavage and oxidation of the disulfide bonds to sulfonic acid (conversion of cystein to cysteic acid39). A number of other studies have investigated bleaching,3,14,15,39 and most are in agreement. The impact is greatest in the cuticle14,39 and diminishes in the deeper layers of the fiber, as is expected from a surface insult. Multivariate curve resolution analysis has enabled specific examination of the cuticle and cortex layers separately from a completely intact hair. Furthermore, variability in the data has been reduced by measuring the same hair fiber before and after bleaching. Commercially, hair is bleached for cosmetic reasons to lighten the color of hair, either as the sole aim or as a base for which the hair can then be colored. Bleaching decolorizes the hair strand by breaking down the melanin pigment but also damages the protein structure, breaking apart the disulfide bonds, which leads to swelling of the fiber, brittleness, and dryness. The ideal bleaching process would break down the melanin without affecting the keratin structure. Consequently, being able to follow the bleaching process on whole hairs, its extent, and where any changes occur spatially is a valuable capability for the investigation of existing hair treatments and for the search for new treatments that are more selective. Penetration into Hairs. Resorcinol. Yak hair was treated with a solution of resorcinol (1,3-dihydroxy benzene), and Raman line depth profiles were collected on both treated and untreated hairs. The MCR analysis of the dataset resolves a spectral factor that closely resembles the Raman spectrum of this molecule, shown in Fig. 8. Figure 9 shows the spatial distribution plots from the treated hair map. Figure 9a is a map of the keratin intensity, which shows the location of the hair substrate; Fig. 9b shows the corresponding resorcinol intensity; and Fig. 9c shows the resorcinol normalized to the keratin component. Together, these illustrate that the resorcinol has penetrated extensively, is widely distributed through the hair fiber, and has an approximately uniform concentration throughout. Figure 10 shows the data obtained from hair that has been rinsed after the treatment with resorcinol. Resorcinol was still present within the hair and remained evenly distributed. The data from both maps were analyzed together with the MCR as a single combined dataset, and so the score values calculated for
FIG. 9. Spatial distribution maps of yak’s hair treated with resorcinol. (a) Keratin factor. Showing the position of the hair, (b) map of resorcinol. (c) Normalized resorcinol factor. Data collected using Kaiser holoprobe 5000R spectrometer.
APPLIED SPECTROSCOPY
1413
FIG. 10. Comparison of resorcinol factors intensity maps before and after rinsing of the hair. Data collected using Kaiser holoprobe 5000R spectrometer.
the resorcinol in each are directly comparable. It shows 60% of the resorcinol initially present on application is retained following rinsing. In a separate chemical analysis the percent retention of resorcinol was measured by using dry weight analysis to quantify the uptake of bulk hair fibers from an initial 2% w/w solution and the amount of resorcinol collected in the water eluent used to rinse the same fibers. This method also measured 60% retention for resorcinol to compare favorably with the results from the Raman study. Penetration of actives has been previously measured with Raman and IR9,17 on microtomed hair samples. But here, we have shown that it is possible to do this with whole hairs and this gives the benefit of being able to follow the same sample through a treatment regime, without affecting the underlying substrate by invasive analysis. Glycerol from a Full Product. Understanding delivery from a full product, given the many components that are normally present, is a bigger challenge than a single compound in solution. Clearly, it is also the crucial test of whether a product is doing what it is expected to do, as other ingredients may interfere. In principle, this is possible using an MCR analysis
approach, as has been shown for a full food product.26 However, a molecule may only make up a small percentage of the product, and then multiple band overlap may become too much of a problem and require acquisition of a very large dataset. One way to overcome this or reduce the size of the needed dataset is to use a deuterated version of the molecule of interest, as has been used in skin penetration, for example.40 This is the approach taken here with glycerol. The product (NeXXus Youth Renewal) contained 1% glycerol, which was replaced with d-glycerol. The treated hairs (see Methods section) were then mapped as previously described, but with longer acquisition times, 4 min per spectrum. A plot of the location of glycerol can be done using the intensity of the C–D stretch alone. However, we have again used the MCR method, as this allows better separation of other components, especially the keratin, which is then used to normalize the glycerol intensity inside the hair (see Methods section) as deuterated glycerol still has bands overlapping many of the protein bands, see Fig. 11. Using MCR overcomes these background interference problems. A plot of the glycerol location is shown in Fig. 12. It can be seen that the glycerol clearly penetrates the hair cortex, which is an important observation to help understand and exploit the potential benefits of this agent when applied as part of a hair product.
CONCLUSION
FIG. 11. (a) Spectral factors for the hair fiber (keratin) and (b) the d-glycerol from the MCR of a hair treated with shampoo product containing 1% dglycerol. (c) The Raman spectrum of d-glycerol. Data collected using WITec Alpha 300 R spectrometer.
1414
Volume 67, Number 12, 2013
The use of confocal Raman spectroscopy has been demonstrated on whole hairs, and it has been shown that, with the right analysis methods, the different sections of the hair (cuticle and cortex) can be differentiated both spectroscopically and spatially, despite the diminished resolution from refractive index changes when scanning into the hair.19 This method was applied to investigations on hair treatment, using whole hair fibers for measurement. Cuticle and cortex were affected to different extents in hairs that had been bleached. Penetration of
FIG. 12. (a) Score maps for the hair fiber (keratin) and (b) the d-glycerol normalized to hair from the MCR analysis of a hair treated with shampoo product containing 1% d-glycerol. Data collected using WITec Alpha 300 R spectrometer.
a target molecule was identified, from both a solution and a full product formulation, which reveals whether a molecule has reached its intended target. It was also shown that a treatment cycle could be followed, in this case rinsing. This demonstrates that Raman spectroscopy is in a position to be more widely applied to hair research with a key role to play in the selection of new benefit agents based on their physical, chemical, and spatial efficacy. ACKNOWLEDGMENTS Eleanor D’Agostino, Ranjit Bhogal, Glynn Roberts, Richard Read, Gez Adams, Janhavi Raut, Cheryl Taylor, Michael Brown. 1. C.R. Roberts. Chemical and Physical Behavior of Human Hair. New York: Springer-Verlag, 2002. 4th ed. 2. J.S. Church, G.L. Corino, A.L. Woodhead. ‘‘The Analysis of Merino Wool Cuticle and Cortical Cells by Fourier Transform Raman Spectroscopy’’. Biopolymers. 1997. 42(1): 7-17. 3. W. Akhtar, H.G.M. Edwards, D.W. Farwell, M. Nutbrown. ‘‘FourierTransform Raman Spectroscopic Study of Human Hair’’. Spectrochim. Acta, Part A. 1997. 53(7): 1021-1031. 4. A. Barth, C. Zscherp. ‘‘What Vibrations Tell Us About Proteins’’. Q. Rev. Biophys. 2002. 35(4): 369-430. 5. A.C. Williams, H.G.M. Edwards, B.W. Barry. ‘‘Raman Spectra of Human Keratotic Biopolymers: Skin, Callus, Hair and Nail’’. J. Raman Spectrosc. 1994. 25(1): 95-98. 6. A. Kuzuhara, N. Fujiwara, T. Hori. ‘‘Analysis of Internal Structure Changes in Black Human Hair Keratin Fibers with Aging Using Raman Spectroscopy’’. Biopolymers. 2007. 87(2–3): 134-140. 7. A. Kuzuhara. ‘‘Protein Structural Changes in Keratin Fibers Induced by Chemical Modification Using 2-Iminothiolane Hydrochloride: A Raman Spectroscopic Investigation’’. Biopolymers. 2005. 79(4): 173-184. 8. K.L.A. Chan, S.G. Kazarian, A. Mavraki, D.R. Williams. ‘‘Fourier Transform Infrared Imaging of Human Hair with a High Spatial Resolution Without the Use of a Synchrotron’’. Applied Spec. 2005. 59(2): 149-155. 9. K.L.A. Chan, F.H. Tay, C. Taylor, S.G. Kazarian. ‘‘A Novel Approach for Study of In Situ Diffusion in Human Hair Using Fourier Transform Infrared Spectroscopic Imaging’’. Applied Spec. 2008. 62(9): 1041-44.
10. K.R. Ackermann, J. Koster, S. Schluecker. ‘‘Polarized Raman Microspectroscopy on Intact Human Hair’’. J. Biophotonics. 2008. 1(5): 419-424. 11. F.I. Bell, R. Skinner. I.M. Tucker, L. Leray, T.E. Lyons, K. Devine P. Pudney, T. Oikawa. ‘‘Biophysical and Mechanical Response of Keratinous Fibres to Changes in Temperature and Humidity’’. J. Cosmet. Sci. 2004. 55(Suppl.): S19-S24. 12. M. Zimmerley, C.-Y. Lin, D.C. Oertel, J.M. Marsh, J.L. Ward, E.O. Potma. ‘‘Quantitative Detection Of Chemical Compounds In Human Hair With Coherent Anti-Stokes Raman Scattering Microscopy’’. J. Biomed. Opt. 2009. 14(4): 044019. 13. K. Bito, M. Okuno, H. Kano, S. Tokuhara, S. Naito, Y. Masukawa, P. Leproux, V. Couderc. and H.-O. Hamaguchi. ‘‘Protein Secondary Structure Imaging with Ultrabroadband Multiplex Coherent Anti-Stokes Raman Scattering (CARS) Microspectroscopy’’. J. Phys. Chem. B. 2012. 116(4): 1452-1457. 14. A. Kuzuhara. ‘‘Analysis of Structural Changes in Bleached Keratin Fibers (Black and White Human Hair) Using Raman Spectroscopy’’. Biopolymers. 2006. 81(6): 506-514. 15. L.J. Hogg, H.G.M. Edwards, D.W. Farwell, A.T. Peters. ‘‘FT RamanSpectroscopic Studies Of Wool’’. J. Soc. Dyers Colour. 1994. 110(5–6): 196-199. 16. L. Wang, L. Chen, L. Han, G. Lian. ‘‘Kinetics and Equilibrium of Solute Diffusion into Human Hair’’. Ann. Biomed. Eng. 2012. 40: 2719-2726. 17. A. Kuzuhara. ‘‘Raman Spectroscopic Analysis of L-Phenylalanine and Hydrolyzed Eggwhite Protein Penetration into Keratin Fibers’’. J. Appl. Polym. Sci. 2011. 122(4): 2680-2689. 18. L.E. Jurdana, K.P. Ghiggino, K.W. Nugent, I.H. Lever. ‘‘Confocal Laser Raman Microprobe Studies of Keratin Fibers’’. Text. Res. J. 1995. 65(10): 593-600. 19. N.J. Everall. ‘‘Confocal Raman Microscopy: Performance, Pitfalls, and Best Practice’’. Appl. Spectrosc. 2009. 63(9): 245A-262A. 20. G. Zhang, L. Senak, D.J. Moore. ‘‘Measuring Changes in Chemistry, Composition, and Molecular Structure Within Hair Fibers by Infrared and Raman Spectroscopic Imaging’’. J. Biomed. Opt. 2011. 16(5): 056009. 21. A. Kuzuhara. ‘‘Analysis of Structural Changes in Permanent Waved Human Hair Using Raman Spectroscopy’’. Biopolymers. 2007. 85(3): 274283. 22. P.D.A. Pudney, T.M. Hancewicz. ‘‘The Role of Confocal Raman Spectroscopy in Food Science’’. Handb. Vib. Spectrosc. 2010:133-167.
APPLIED SPECTROSCOPY
1415
23. J.J. Andrew, M.A. Browne, I.E. Clark, T.M. Hancewicz, A.J. Millichope. ‘‘Raman Imaging of Emulsion Systems’’. Appl. Spectrosc. 1998. 52: 790796. 24. P.D.A. Pudney, T.M. Hancewicz, D.G. Cunningham, C Gray. ‘‘A Novel Method for Measuring Concentrations of Phase Separated Biopolymers: The Use of Confocal Raman Spectroscopy with Self-Modelling Curve Resolution’’. Food Hydrocolloids. 2003. 17(3): 345-353. 25. P.D.A. Pudney, T.M. Hancewicz, D.G. Cunningham. ‘‘The Use of Confocal Raman Spectroscopy to Characterise the Microstructure of Complex Biomaterials: Foods’’. Spectrosc. 2002. 16(2-3): 217-225. 26. P.D.A. Pudney, T.M. Hancewicz, D.G. Cunningham, M.C. Brown. ‘‘Quantifying the Microstructures of Soft Solid Materials by Confocal Raman Spectroscopy’’. Vib. Spectrosc. 2004. 34(1): 123-135. 27. P.D.A. Pudney, L. Gambelli, M.J. Gidley. ‘‘Confocal Raman Microspectroscopic Study of the Molecular Status of Carotenoids in Tomato Fruits and Foods’’. Appl. Spectrosc. 2011. 65(2): 127-134. 28. E.Y.M. Bonnist, J.-P. Gorce, C. Mackay, R.U. Pendlington, P.D.A. Pudney. ‘‘Measuring the Penetration of a Skin Sensitizer and Its Delivery Vehicles Simultaneously with Confocal Raman Spectroscopy’’. Skin Pharmacol. Physiol. 2011. 24(5): 274-283. 29. N.J. Everall. ‘‘Otimising Image Fidelity in 3D Confocal Raman Imaging’’. Paper presented at: SciX Conference. Kansas City, MO; Sept 30-Oct. 5 2012. 30. J.J. Andrew, T.M. Hancewicz. ‘‘Rapid Analysis of Raman Image Data Using Two-Way Multivariate Curve Resolution’’. Appl. Spectrosc. 1998. 52(6): 797-807. 31. J. Wang, P.K. Hopke, T.M. Hancewicz, S.L. Zhang. ‘‘Application of Modified Alternating Least Squares Regression to Spectroscopic Image Analysis’’. Anal. Chim. Acta. 2003. 476(1): 93-109.
1416
Volume 67, Number 12, 2013
32. P.D.A. Pudney, M.E. Melot, P.J. Caspers, A. Van Der Pol, G.J. Puppels. ‘‘An In Vivo Confocal Raman Study of the Delivery of Trans Retinol to the Skin’’. Appl. Spectrosc. 2007. 61(8): 804-811. 33. M. Me´lot, P.D.A. Pudney, A.-M. Williamson, P.J. Caspers, A. Van Der Pol, G.J. Puppels. ‘‘Studying the Effectiveness of Penetration Enhancers to Deliver Retinol through the Stratum Cornum by In-Vivo Confocal Raman Spectroscopy’’. J. Controlled Release. 2009. 138(1): 32-39. 34. R. Paquin, P. Colomban. ‘‘Nanomechanics of Single Keratin Fibres: A Raman Study of the a-Helix ! b-Sheet Transition and the Effect of Water’’. J. Raman Spectrosc. 2007. 38(5): 504-514. 35. M. Gniadecka, O.F. Nielsen, D.H. Christensen, H.C. Wulf. ‘‘Structure of Water, Proteins, and Lipids in Intact Human Skin, Hair, and Nail’’. J. Invest. Dermatol. 1998. 110(4): 393-398. 36. A. Kuzuhara, T. Hori. ‘‘Reduction Mechanism of L-cysteine on Keratin Fibers Using Microspectrophotometry and Raman Spectroscopy’’. Biopolymers. 2005. 79(6): 324-334. 37. N.J. Everall. ‘‘Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think’’. Appl. Spectrosc. 2000. 54(10): 1515-1520. 38. N.J. Everall. ‘‘Modeling and Measuring the Effect of Refraction on the Depth Resolution of Confocal Raman Microscopy’’. Appl. Spectrosc. 2000. 54(6): 773-782. 39. L.J. Wolfram, K. Hall, I. Hui. ‘‘The Mechanism of Hair Bleaching’’. J. Soc. Cosmet. Chem. 1970. 21: 875-900. 40. G. Mao, C.R. Flach, R. Mendelsohn, R.M. Walters. ‘‘Imaging the Distribution of Sodium Dodecyl Sulfate in Skin by Confocal Raman and Infrared Microspectroscopy’’. Pharm. Res. 2012. 29(8): 2189-2201.
Unilever Discover, Colworth Laboratory, Sharnbrook, Bedfordshire, MK44 1LQ UK Unilever R&D Port Sunlight Laboratory Quarry Road East, Bebington, Wirral, Merseyside, CH63 3JW UK
This paper describes the application of Raman spectroscopy to whole hair fibers. Previously this has proved difficult because the hairs are relatively opaque, and spatial resolution diminishes with depth because of the change in refractive index. A solution is to couple confocal Raman with multivariate curve resolution (MCR) data analysis, which separates spectral differences with depth despite this reduction in resolution. Initially, it is shown that the cuticle can be separated from the cortex, showing the differences in the proteins, which can then be plotted as a function of depth, with the cuticle factor being seen only at the surface as expected. Hairs that had been treated in different ways, e.g., by bleaching, treatment with the active molecule resorcinol followed by rinsing and treatment with a full hair care product, were also examined. In all cases, changes to the hair are identified and are associated with specific parts of the fiber. Since the hair fiber is kept intact, it can be repeatedly treated and measured, hence multistep treatment processes can be followed. This method expands the potential use of Raman spectroscopy in hair research. Index Headings: Raman; Hair; Keratin; Penetration; Bleaching.
INTRODUCTION Hair is one of the defining characteristics of mammals. It is made mainly of protein in a filamentous structure. It has a number of functions such as providing warmth and camouflage, as well as for sending signals for danger warnings and mating in some species. In humans many of these functions have diminished, but it does still provide some function in thermoregulation. The head is the main area where hair is in abundance and where people can relatively easily vary their appearance. Since people spend a great deal of time and effort cleaning, coloring, and styling their hair, it is of significant importance to the cosmetic industry. The structure of the hair fiber has been described elsewhere.1 It consists mainly of the protein keratin in a filamentous structure and has three main parts: The cortex, which constitutes the bulk of the hair, and is formed of microfibrils and macrofibrils arranged parallel to each other and separated by a matrix; the cuticle, flattened cells arranged perpendicular to the cortical cells that form the outer protective layer; and the medulla is at the center of the fiber; however, this is often not continuous or absent from some human hair fibers. Other components in hair fibers include lipids, water, and melanin, the pigment responsible for hair color. The constituent protein keratin varies in the order and types of amino acids present in the different parts of the hair. These areas thus have different interactions between groups, i.e., hydrogen bonds, polar interactions, Coulombic attractions, and covalent bonds, which lead to different conformations and Received 27 March 2013; accepted 3 August 2013. * Author to whom correspondence should be sent. E-mail: paul.pudney@ unilever.com. DOI: 10.1366/13-07086
1408
Volume 67, Number 12, 2013
secondary structures. For example the cortex material is mainly a-helical, and the cuticle has high b-sheet character.2 Particularly important are disulfide bonds that form between the –SH groups on cysteine residues, which can be intra- or inter-chain and are more prevalent in the cuticle than the cortex.3 These are the principal forces in maintaining the structure and properties of the hair fiber and are resistant to many chemical treatments.1 Vibrational spectroscopy is sensitive to molecular structure and conformation of proteins4 and so is well placed to investigate the properties of hair. Indeed there are a number of studies using both infrared (IR) and Raman spectroscopy5,6,7,8,9,10,11 and also coherent anti-Stokes Raman scattering spectroscopy (CARS).12,13 When cross-sectioned, the different structures of the cuticle and cortex have been demonstrated by these methods.2,6,14 Changes in hairs have also been followed when treated in various ways, the most widely used treatment being bleaching.3,13,15 A treatment of hair with cosmetic products often involves diffusion of active compounds into hair.16 The penetration of actives has been previously shown by Raman and IR.9,17 Despite much progress having been made, some challenges still remain, which in combination prevent valuable insights into hair properties and hair treatment from being uncovered. Most studies have concerned microtomed hair rather than whole hair fibers, allowing access to the different areas of the fiber by scanning along the cross-section. The drawback to this is that often one may want to follow external molecule penetration over time or examine the hair throughout a successive treatment regime. Microtoming destroys the sample so that this type of repeat measurement is not possible and also negates one of the advantages of Raman spectroscopy of being noninvasive. This approach is taken because, when sampling whole hair, especially when doing depth profiles through whole hair fibers, it is difficult to distinguish clearly the different layers of the hair structure.18 This is because the hair structure is composed predominantly of keratin, and it is only relatively minor inhomogeneities in its chemical and structural composition that distinguish different layers, as highlighted above. The difficulty in unambiguously detecting and mapping these layers is further compounded by the fact that the Raman spectroscopic signals of different proteins and their different structural conformations are all very similar and heavily overlap. Furthermore, the natural optical properties of hair leads to a large change in refractive index, ongoing from air into the hair fiber, which causes the focus of the Raman laser beam to be enlarged, thus reducing the confocal resolution with increasing depth and condensing of the depth scale (the apparent depth problem).19 Recently, Zhang et al.20 made use of the normal method of reducing the loss of depth resolution19 by using an oil-immersion objective in order to refractive-index match the sample. They did this by placing the hair under a
0003-7028/13/6712-1408/0 Ó 2013 Society for Applied Spectroscopy
APPLIED SPECTROSCOPY
FIG. 1. Schematic diagram showing how the whole hair fibers were sampled. Spectra are collected along the length of the hair with a depth profile collected at each point.
coverslip to prevent oil from interfering with the hair. This provided a good depth profile and showed protein features changing with depth, although spectral overlap was still apparent when compared with the separated cuticle and cortex spectra2 and when microtomed.6,13,14,21 They did not compare data collected using immersion and dry objectives. Hence, there is a need for an improved methodology that can accommodate the measurement of whole hair fibers and sufficiently discriminate between the different sections of the hair within the data that is generated. Essentially whole hairs are a multicomponent microstructure, and thus a Raman map produces a large spectral dataset. In a conventional analysis of the resultant spectrum, an attempt is made to attribute specific individual spectral intensities to specific chemical components and then derive, as a function of position, the spatial component maps. For the spectra generated by a multicomponent microstructure, this is generally not possible since it is rarely possible to unambiguously attribute specific spectral intensities in this way. This is clearly the case in spectra of hair as described in the above section, as it comprises mainly keratin but differently folded. Consequently, a different approach is needed, in which the aim of the analysis is to produce pure separated spectra and their intensities from the spectral dataset. In a lot of chemometric techniques, the spectra are separated on the basis of variations, in this case in the spatial and spectral dimensions. However, in most methods, this does not produce pure and thus interpretable factors. However, there are methods for doing this, which generally
come under the title of self-modeling curve resolution (SMCR) or multivariate curve resolution (MCR).22 This method has now been successfully used for a number of years on different types of microstructures that have been Raman-mapped. These include emulsions,23 phase separated biopolymers,24 food products,25,26 and plant structures.27 It has also been used on depth profiles into skin,28 which is most similar to the system being examined in this study. This paper reports the spectral separation of the different parts of hair fibers from depth-profile maps of whole hairs with multivariate curve resolution (MCR) analysis. This is demonstrated with yak hair (a good model of human hair without melanin) and blonde or gray human hair. The method is further applied to a number of hair problems, specifically to investigate bleaching and external molecule treatment without microtoming the hair sample. It is found that bleaching mainly affects the cuticle layer, and its extent diminishes with depth. We show the uptake of resorcinol into the hair fiber and observe that the amount inside the hair is reduced on rinsing. Finally, we identify the penetration of a select ingredient into the hair fiber after treatment with a full hair-care product formulation. The use of air objectives allows the most flexibility with treatment conditions, but the approach would be equally applicable to data collected with immersion objectives in those cases where the sample allows it. This approach expands the applicability of Raman in applied hair research.
METHODS Hair Samples. Yak hair was obtained from Hugo Royer Ltd. as premade 0.75 g, 5 cm length switches of yak belly hair, which is a close approximation to human hair. It is very low in color and is usually obtained with little damage and is of consistent quality. Natural white hair was obtained from International Hair Importers, USA; selected by optical microscopy were naturally blonde virgin human hair. To bleach the hair strands, L’Oreal Platine Precision bleach powder and 9% Excel cre`me peroxide were mixed in a 1 : 2 ratio. The bleaching mix was applied to each hair strand using a tinting brush and left to develop for 30 min. Each hair strand was rinsed for 2 min and left to air dry for 2 h. Following this, another bleach powder/peroxide cre`me mixture was prepared and applied once more to the bleached hairs. After leaving for 30 min, the hair strands were washed using a base shampoo (17.10 wt % sodium laureth sulfate, 5.33 wt % cocoamidopropyl betaine,0.40 wt % DMDM hydantoin, 1.40 wt % sodium chloride, in water) and air-dried.
FIG. 2. The two main factors/spectra obtained from mapping a virgin yak’s hair using the Kaiser holoprobe 5000R spectrometer. (a)The blue spectrum is typical of keratin with high a-helix content, typical of the cortex. (b) The red spectrum, shows an S-S rich/b-sheet protein of the cuticle.
APPLIED SPECTROSCOPY
1409
FIG. 3. A plot of the two main factors obtained from MCR analysis of a xz scan through a single yak hair fiber. The y-axis is a line of points along the top of the hair at 2 lm intervals (x-scan), and the x-axis is the z direction, i.e., penetrating through the hair, with points taken at 5 mm intervals. A collection time of 60 s was used at each point.
The procedure for treating yak hair with resorcinol (1,3dihydroxy benzene) was as follows: 0.4 g of 15 cm long yak hair was packed into a glass pipette. This was then rinsed with 1 mL of a 2% by weight solution of resorcinol. For the rinse sample, the treated hair was left for 24 h then rinsed with 2 mL of distilled water. Blonde human hairs were treated with a full formulation NeXXus product, Youth Renewal, whereby the 1% glycerol in the formulation was replaced with deuterated glycerol. This is a leave-on formulation, so hairs were treated with the product and left for approximately 24 h before measurement. Data Collection. Raman spectra were collected using identical procedures on two different spectrometers, both using an excitation wavelength of 785 nm. The first was a WITec Alpha 300 R system. The 785 nm laser was used at an operating power of 50 mW before the objective; this was selected such as to avoid sample damage while still resulting in sufficient sample excitation. A Zeiss 1003 (N.A. 0.9) air/metallurgical objective and a Zeiss Achromoplan 1003 (N.A. 1.25) immersion objective were used. The second spectrometer used was a Kaiser holoprobe 5000R confocal Raman spectrometer equipped with an Olympus air/1003 metallurgical objective (N.A. 0.8), and also employing less than 50 mW of laser power at the sample. Hair strands were fixed to a glass microscope slide using tape, ensuring that the hair fiber was pulled straight without stretching. When using the oil-immersion objective, a cover slip was placed on top of the hair so the immersion oil did not interfere with the hair. A line depth profile along the length of the hair was collected, i.e., an xz-map of the hair fiber, as
shown in Fig. 1. The collection of data along the length of the fiber, rather than across the fiber, prevents some more serious problems concerning the mixing of spatial information.29 A collection time of 60 s for each point was used with both spectrometers unless otherwise stated. Data Analysis. Multivariate curve resolution (MCR) analysis separates the data into two modes: one which describes the spectra of the different components present (the factors) and another describing the intensity of each spectral factor, at each location sampled (the scores)22. To do this, a principal factor analysis (PFA), or equivalent method, is performed on the unscaled data. The factors represent the changes which occur in the spectra and are calculated by separating them out according to their variance from each other, i.e., as different from each other as possible. A relative score is given to each factor to show the distribution of these factors through the sample. After PFA, the resulting factors are not easily interpreted as they may still be mixtures of the components and often do not represent the real components.30 Further constraints must be applied to transform the results into real spectra and pure scores; these constraints are applied using an alternating least squares regression (ALS) algorithm.29 A non-negativity constraint is applied to make the data practically meaningful, i.e., it is not possible to have a negative concentration (score) or negative Raman peaks. An equality constraint is also applied; this aims to normalize the factors ensuring that all intensity variance is found in the scores and not in the spectra.30 The constraints are applied alternatively to the factors and scores until a solution converges, thus
FIG. 4. Spectral factors from MCR analysis of the data from mapping an untreated whole human hair fiber using the WITec Alpha 300 R spectrometer. Blue is the cortex. Green is the cuticle. For label meanings see text.
1410
Volume 67, Number 12, 2013
TABLE I. Assignments of cuticle and cortex spectrum derived from the spectral factors produced from MCR analysis. Assignments using references 2, 3, 5, 14, and 36. Cuticle/cm1 v (SS) (gauche–gauche– gauche) v (CS) (gauche) v (CS) p (CH2) in-phase d (CCH) aliphatic (tyrosine side chains) d (CCH) aromatic (tyrosine side chains) p (CH2) tryptophan v (CC) helix-a/p (CH3) terminal p (CH3) d (CCH) olefinic v (CC) aromatic ring/ phenylalanine v (CC) skeletal/cis conformation v (CC) skeletal/trans conformation v (CC) skeletal v (CC) skeletal Tyrosine Amide III disordered Amide III Amide III a helix CH2 bending mode Amide I
Cortex/cm1
509 644, 663, 671 686, 719 –
509 – – 742/743
–
828
– 880–891
853 –
– – 982, 989
933 959 –
–
1002-1003
1033
–
1044 1052
1047 – 10821103 and 1125 – – – 1342 1448 1657 (1649–1669)
– 1176, 1187, 1199 1241, 1252 1281 – 1451, 1465 1675 (1668–1679)
producing factors representing the real spectra of the components present and scores that represent the relative concentration of each factor at every position sampled. This is the basic ALS approach; the modified alternating least squares
(MALS) algorithm is an improved implementation of this method and has been described elsewhere.31 When conducting a depth profile, one of the main problems is the loss of ongoing signal deeper into the sample, mostly caused by physical scattering and spherical aberration. The same problems are encountered when doing depth profiles into skin.32,33 This problem is overcome by using the protein signal as an internal standard as the concentration of protein is approximately constant with depth. Here, the same approach can be taken using the keratin signal. Another problem with measuring depth is that the depth axis is not real depth into the sample but rather apparent depth; the depth axis here refers to mechanical depth. This is the distance moved by the objective from the surface position of the hair fiber to collect the spectrum at that point.
RESULTS AND DISCUSSION Whole Untreated Hairs. Spectra were collected in a line along the length of the hair taking a depth profile at each point (an xz-map, see Fig. 1). Firstly, yak hair was examined. This xz-map data was then analyzed using the MCR method; the two main spectral factors resulting from this are shown in Fig. 2. The associated scores (intensities) are plotted in Fig. 3. This experiment has been repeated on human whole blonde hairs, and the same analysis carried out; the spectral factors are comparable and shown in Fig. 4. The associated score maps are very similar to those of yak hair that are shown in Fig. 3, so are not shown here. From the spatial distribution and the features of the spectral factors, these can be associated with the cuticle (Figs. 2b and 4 green) and cortex (Figs. 2a and 4 blue). The spectral factors highlight the differences between the two proteins in both molecular structure and conformation; the key differences and peak assignments5,14,34 are listed in Table 1. Here, the features from the human hair are discussed, but this applies equally for the yak hair results as well.
FIG. 5. Plots of the scores of the cortex and cuticle from MCR analysis of xz depth profiles from the same untreated yak’s hair. (a) A plot of the cortex scores using the air objective. (b) The plot of cuticle scores using the air objective. (c) A plot of the cortex scores using the oil objective. (d) The plot of cuticle scores using the oil objective. Data collected using WITec Alpha 300 R spectrometer.
APPLIED SPECTROSCOPY
1411
FIG. 6. Spectral factors from untreated (blue) and bleached blonde human hair (red). (a) Cuticle factors. (b) Cortex factors. Data collected using WITec Alpha 300 R spectrometer.
The spectra reflect differences in amino acid content, specifically that the cuticle is more sulfur-rich than the cortex due to a higher number of cysteine residues14 and that the cortex has more tyrosine and phenylalanine.2 The band at 509 cm1 in Fig. 4a, which arises from the disulfide (-C-S-S-C-) bonds,35 is more intense in the cuticle spectrum. The narrower shape of this band in the cuticle, suggests that the SS bonds here have a high population of the gauche-gauche-gauche conformation. An all-gauche conformer is centered at ;510 cm1, but this increases in frequency when a trans conformation is introduced (these conformers are centered at 525, 540 cm1). The higher intensity of the 645 cm1 band for the cuticle, Fig. 4b, from the C-S bonds, is also consistent with higher levels of cysteine. The bands at 828 and 853 cm1, Fig. 4c, are only evident in the cortex spectrum; these originate from the amino acid tyrosine. The phenylalanine peak at ;1002 cm1 in Fig. 4e is more intense in the cortex spectrum, also fitting with the known amino acid content. Apart from reflecting the different amino acids, the resolved cuticle and cortex spectra further show the differences due to their conformations, the cuticle being in the b-sheet conformation, while the cortex is a-helical. This can be seen from the position of the amide I band, Fig. 4g, which is centered at 1675 cm1 in the cuticle and at 1657 cm1 for the cortex, and of the amide III band, Fig. 4f, which is between 1240 and 1252 cm1 for the cuticle, and at 1342 cm1 for the cortex. In addition to
FIG. 7. Ratio of the sulfur-oxygen peak (1020–1070 cm1) area to phenylalanine peak (990–1016 cm1) area on the same hair before (red) and after bleaching (blue). The five points along each hair have then been averaged to give a representative ratio for the hairs at each treatment stage. Data collected using WITec Alpha 300 R spectrometer.
1412
Volume 67, Number 12, 2013
these are the presence of a band at 933 cm1 in the cortex and the absence of a 959 cm1 band in the cuticle. These spectral factor results for cuticle and cortex are in good agreement with Raman spectra from the same parts isolated from sheep’s hair (i.e., merino wool).2 Spectra from the cuticle and cortex of hair have been identified from a crosssection of hair;14,36 however, the differences observed here are less well defined and often contain some features of the other proteins present. The spectra obtained on a whole hair using an oil objective showed differences with depth consistent with the changes outlined above, including a shift of the amide I band towards the a-helical position and the lower intensity of the S-S band;20 however, the features are not as well separated as seen using the MCR analysis. The plot of the scores of these two spectral factors in Fig. 3 show that the cuticle factor is only observed at the surface of the hair and then appears again when the focus reaches the other side of the hair; although, at this stage, the laser focus is very diffuse, and hence here, it appears much more spatially spread. The cortex factor appears between the location of the cuticle factor, as would be expected. Thus, application of the MCR analysis method has provided a clear resolution of cuticle and the cortex layers in the data from a whole hair fiber, without the need for microtoming. The intact sample is thus available for further measurement, giving scope for continued investigation. Comparing Collection Methods. The use of dry objectives allows the most flexibility with treatment conditions; however, as pointed out in many publications,19,37,38 this leads to degraded resolution in the z-axis. The normal way to negate this is to use immersion objectives to refractive index match. This approach has been used with hair20 where a coverslip is placed on top of the hair. Here, this method is compared with the dry objective for hair. Data was collected from the same yak’s hair using the same collection conditions for the air objective but with an oil-immersion objective and coverslip. This data was analyzed by the MCR methods and results in very similar factors, to those shown in Figs. 2 and 4. The score maps showing the spatial distribution are shown in Fig. 5. Both methods clearly differentiate the cuticle and cortex spatially;
FIG. 8. Spectral factor for resorcinol (blue) compared to spectrum of pure resorcinol (red). Data collected using Kaiser holoprobe 5000R spectrometer.
the main difference is that the signal from the cortex is maintained to greater depth with the oil-immersion objective as the laser focus is maintained much better, whereas the focus through the dry objective is much more attenuated because of the larger refraction effect.38 However, the dry-objective data gives a good signal from the cortex well beyond the cuticle region, suggesting that the cortex can be measured with good reliability. One practical issue that should be mentioned is that good profiles are not always obtained when collecting data using the oil objective because, as the hair fiber is not flat, being both round and rough, the coverslip is not always in good contact with the hair over a large area. Effect of Bleaching on Hair. Whole blonde human hairs were examined before and after bleaching. The data was then analyzed using the MCR method. The cuticle and cortex spectral factors from this analysis are compared before and after the bleaching process, as shown in Fig. 6. The bleached cuticle shows a reduction in the S-S band at 508 cm1 and an increase in the S-O band seen at ;1042 cm1. These changes are also replicated in the cortex factors, see Fig. 6b, but the increase in the S-O band is larger in the cuticle than in the cortex. Thus, the impact of bleaching spans across different layers of the hair fiber. To show this in further detail, a plot of the integrated intensity ratio of the sulfur-oxygen band (1020– 1070 cm1) normalized to the phenyalanine band (990–1016 cm1) with depth is shown for the same hair before and after bleaching, see Fig. 7. This further illustrates that the S-O band increases upon bleaching across the measured depth range and that the extent of this change decreases with depth.
The spectral differences observed here are consistent with the molecular changes known to be caused by bleaching, namely cleavage and oxidation of the disulfide bonds to sulfonic acid (conversion of cystein to cysteic acid39). A number of other studies have investigated bleaching,3,14,15,39 and most are in agreement. The impact is greatest in the cuticle14,39 and diminishes in the deeper layers of the fiber, as is expected from a surface insult. Multivariate curve resolution analysis has enabled specific examination of the cuticle and cortex layers separately from a completely intact hair. Furthermore, variability in the data has been reduced by measuring the same hair fiber before and after bleaching. Commercially, hair is bleached for cosmetic reasons to lighten the color of hair, either as the sole aim or as a base for which the hair can then be colored. Bleaching decolorizes the hair strand by breaking down the melanin pigment but also damages the protein structure, breaking apart the disulfide bonds, which leads to swelling of the fiber, brittleness, and dryness. The ideal bleaching process would break down the melanin without affecting the keratin structure. Consequently, being able to follow the bleaching process on whole hairs, its extent, and where any changes occur spatially is a valuable capability for the investigation of existing hair treatments and for the search for new treatments that are more selective. Penetration into Hairs. Resorcinol. Yak hair was treated with a solution of resorcinol (1,3-dihydroxy benzene), and Raman line depth profiles were collected on both treated and untreated hairs. The MCR analysis of the dataset resolves a spectral factor that closely resembles the Raman spectrum of this molecule, shown in Fig. 8. Figure 9 shows the spatial distribution plots from the treated hair map. Figure 9a is a map of the keratin intensity, which shows the location of the hair substrate; Fig. 9b shows the corresponding resorcinol intensity; and Fig. 9c shows the resorcinol normalized to the keratin component. Together, these illustrate that the resorcinol has penetrated extensively, is widely distributed through the hair fiber, and has an approximately uniform concentration throughout. Figure 10 shows the data obtained from hair that has been rinsed after the treatment with resorcinol. Resorcinol was still present within the hair and remained evenly distributed. The data from both maps were analyzed together with the MCR as a single combined dataset, and so the score values calculated for
FIG. 9. Spatial distribution maps of yak’s hair treated with resorcinol. (a) Keratin factor. Showing the position of the hair, (b) map of resorcinol. (c) Normalized resorcinol factor. Data collected using Kaiser holoprobe 5000R spectrometer.
APPLIED SPECTROSCOPY
1413
FIG. 10. Comparison of resorcinol factors intensity maps before and after rinsing of the hair. Data collected using Kaiser holoprobe 5000R spectrometer.
the resorcinol in each are directly comparable. It shows 60% of the resorcinol initially present on application is retained following rinsing. In a separate chemical analysis the percent retention of resorcinol was measured by using dry weight analysis to quantify the uptake of bulk hair fibers from an initial 2% w/w solution and the amount of resorcinol collected in the water eluent used to rinse the same fibers. This method also measured 60% retention for resorcinol to compare favorably with the results from the Raman study. Penetration of actives has been previously measured with Raman and IR9,17 on microtomed hair samples. But here, we have shown that it is possible to do this with whole hairs and this gives the benefit of being able to follow the same sample through a treatment regime, without affecting the underlying substrate by invasive analysis. Glycerol from a Full Product. Understanding delivery from a full product, given the many components that are normally present, is a bigger challenge than a single compound in solution. Clearly, it is also the crucial test of whether a product is doing what it is expected to do, as other ingredients may interfere. In principle, this is possible using an MCR analysis
approach, as has been shown for a full food product.26 However, a molecule may only make up a small percentage of the product, and then multiple band overlap may become too much of a problem and require acquisition of a very large dataset. One way to overcome this or reduce the size of the needed dataset is to use a deuterated version of the molecule of interest, as has been used in skin penetration, for example.40 This is the approach taken here with glycerol. The product (NeXXus Youth Renewal) contained 1% glycerol, which was replaced with d-glycerol. The treated hairs (see Methods section) were then mapped as previously described, but with longer acquisition times, 4 min per spectrum. A plot of the location of glycerol can be done using the intensity of the C–D stretch alone. However, we have again used the MCR method, as this allows better separation of other components, especially the keratin, which is then used to normalize the glycerol intensity inside the hair (see Methods section) as deuterated glycerol still has bands overlapping many of the protein bands, see Fig. 11. Using MCR overcomes these background interference problems. A plot of the glycerol location is shown in Fig. 12. It can be seen that the glycerol clearly penetrates the hair cortex, which is an important observation to help understand and exploit the potential benefits of this agent when applied as part of a hair product.
CONCLUSION
FIG. 11. (a) Spectral factors for the hair fiber (keratin) and (b) the d-glycerol from the MCR of a hair treated with shampoo product containing 1% dglycerol. (c) The Raman spectrum of d-glycerol. Data collected using WITec Alpha 300 R spectrometer.
1414
Volume 67, Number 12, 2013
The use of confocal Raman spectroscopy has been demonstrated on whole hairs, and it has been shown that, with the right analysis methods, the different sections of the hair (cuticle and cortex) can be differentiated both spectroscopically and spatially, despite the diminished resolution from refractive index changes when scanning into the hair.19 This method was applied to investigations on hair treatment, using whole hair fibers for measurement. Cuticle and cortex were affected to different extents in hairs that had been bleached. Penetration of
FIG. 12. (a) Score maps for the hair fiber (keratin) and (b) the d-glycerol normalized to hair from the MCR analysis of a hair treated with shampoo product containing 1% d-glycerol. Data collected using WITec Alpha 300 R spectrometer.
a target molecule was identified, from both a solution and a full product formulation, which reveals whether a molecule has reached its intended target. It was also shown that a treatment cycle could be followed, in this case rinsing. This demonstrates that Raman spectroscopy is in a position to be more widely applied to hair research with a key role to play in the selection of new benefit agents based on their physical, chemical, and spatial efficacy. ACKNOWLEDGMENTS Eleanor D’Agostino, Ranjit Bhogal, Glynn Roberts, Richard Read, Gez Adams, Janhavi Raut, Cheryl Taylor, Michael Brown. 1. C.R. Roberts. Chemical and Physical Behavior of Human Hair. New York: Springer-Verlag, 2002. 4th ed. 2. J.S. Church, G.L. Corino, A.L. Woodhead. ‘‘The Analysis of Merino Wool Cuticle and Cortical Cells by Fourier Transform Raman Spectroscopy’’. Biopolymers. 1997. 42(1): 7-17. 3. W. Akhtar, H.G.M. Edwards, D.W. Farwell, M. Nutbrown. ‘‘FourierTransform Raman Spectroscopic Study of Human Hair’’. Spectrochim. Acta, Part A. 1997. 53(7): 1021-1031. 4. A. Barth, C. Zscherp. ‘‘What Vibrations Tell Us About Proteins’’. Q. Rev. Biophys. 2002. 35(4): 369-430. 5. A.C. Williams, H.G.M. Edwards, B.W. Barry. ‘‘Raman Spectra of Human Keratotic Biopolymers: Skin, Callus, Hair and Nail’’. J. Raman Spectrosc. 1994. 25(1): 95-98. 6. A. Kuzuhara, N. Fujiwara, T. Hori. ‘‘Analysis of Internal Structure Changes in Black Human Hair Keratin Fibers with Aging Using Raman Spectroscopy’’. Biopolymers. 2007. 87(2–3): 134-140. 7. A. Kuzuhara. ‘‘Protein Structural Changes in Keratin Fibers Induced by Chemical Modification Using 2-Iminothiolane Hydrochloride: A Raman Spectroscopic Investigation’’. Biopolymers. 2005. 79(4): 173-184. 8. K.L.A. Chan, S.G. Kazarian, A. Mavraki, D.R. Williams. ‘‘Fourier Transform Infrared Imaging of Human Hair with a High Spatial Resolution Without the Use of a Synchrotron’’. Applied Spec. 2005. 59(2): 149-155. 9. K.L.A. Chan, F.H. Tay, C. Taylor, S.G. Kazarian. ‘‘A Novel Approach for Study of In Situ Diffusion in Human Hair Using Fourier Transform Infrared Spectroscopic Imaging’’. Applied Spec. 2008. 62(9): 1041-44.
10. K.R. Ackermann, J. Koster, S. Schluecker. ‘‘Polarized Raman Microspectroscopy on Intact Human Hair’’. J. Biophotonics. 2008. 1(5): 419-424. 11. F.I. Bell, R. Skinner. I.M. Tucker, L. Leray, T.E. Lyons, K. Devine P. Pudney, T. Oikawa. ‘‘Biophysical and Mechanical Response of Keratinous Fibres to Changes in Temperature and Humidity’’. J. Cosmet. Sci. 2004. 55(Suppl.): S19-S24. 12. M. Zimmerley, C.-Y. Lin, D.C. Oertel, J.M. Marsh, J.L. Ward, E.O. Potma. ‘‘Quantitative Detection Of Chemical Compounds In Human Hair With Coherent Anti-Stokes Raman Scattering Microscopy’’. J. Biomed. Opt. 2009. 14(4): 044019. 13. K. Bito, M. Okuno, H. Kano, S. Tokuhara, S. Naito, Y. Masukawa, P. Leproux, V. Couderc. and H.-O. Hamaguchi. ‘‘Protein Secondary Structure Imaging with Ultrabroadband Multiplex Coherent Anti-Stokes Raman Scattering (CARS) Microspectroscopy’’. J. Phys. Chem. B. 2012. 116(4): 1452-1457. 14. A. Kuzuhara. ‘‘Analysis of Structural Changes in Bleached Keratin Fibers (Black and White Human Hair) Using Raman Spectroscopy’’. Biopolymers. 2006. 81(6): 506-514. 15. L.J. Hogg, H.G.M. Edwards, D.W. Farwell, A.T. Peters. ‘‘FT RamanSpectroscopic Studies Of Wool’’. J. Soc. Dyers Colour. 1994. 110(5–6): 196-199. 16. L. Wang, L. Chen, L. Han, G. Lian. ‘‘Kinetics and Equilibrium of Solute Diffusion into Human Hair’’. Ann. Biomed. Eng. 2012. 40: 2719-2726. 17. A. Kuzuhara. ‘‘Raman Spectroscopic Analysis of L-Phenylalanine and Hydrolyzed Eggwhite Protein Penetration into Keratin Fibers’’. J. Appl. Polym. Sci. 2011. 122(4): 2680-2689. 18. L.E. Jurdana, K.P. Ghiggino, K.W. Nugent, I.H. Lever. ‘‘Confocal Laser Raman Microprobe Studies of Keratin Fibers’’. Text. Res. J. 1995. 65(10): 593-600. 19. N.J. Everall. ‘‘Confocal Raman Microscopy: Performance, Pitfalls, and Best Practice’’. Appl. Spectrosc. 2009. 63(9): 245A-262A. 20. G. Zhang, L. Senak, D.J. Moore. ‘‘Measuring Changes in Chemistry, Composition, and Molecular Structure Within Hair Fibers by Infrared and Raman Spectroscopic Imaging’’. J. Biomed. Opt. 2011. 16(5): 056009. 21. A. Kuzuhara. ‘‘Analysis of Structural Changes in Permanent Waved Human Hair Using Raman Spectroscopy’’. Biopolymers. 2007. 85(3): 274283. 22. P.D.A. Pudney, T.M. Hancewicz. ‘‘The Role of Confocal Raman Spectroscopy in Food Science’’. Handb. Vib. Spectrosc. 2010:133-167.
APPLIED SPECTROSCOPY
1415
23. J.J. Andrew, M.A. Browne, I.E. Clark, T.M. Hancewicz, A.J. Millichope. ‘‘Raman Imaging of Emulsion Systems’’. Appl. Spectrosc. 1998. 52: 790796. 24. P.D.A. Pudney, T.M. Hancewicz, D.G. Cunningham, C Gray. ‘‘A Novel Method for Measuring Concentrations of Phase Separated Biopolymers: The Use of Confocal Raman Spectroscopy with Self-Modelling Curve Resolution’’. Food Hydrocolloids. 2003. 17(3): 345-353. 25. P.D.A. Pudney, T.M. Hancewicz, D.G. Cunningham. ‘‘The Use of Confocal Raman Spectroscopy to Characterise the Microstructure of Complex Biomaterials: Foods’’. Spectrosc. 2002. 16(2-3): 217-225. 26. P.D.A. Pudney, T.M. Hancewicz, D.G. Cunningham, M.C. Brown. ‘‘Quantifying the Microstructures of Soft Solid Materials by Confocal Raman Spectroscopy’’. Vib. Spectrosc. 2004. 34(1): 123-135. 27. P.D.A. Pudney, L. Gambelli, M.J. Gidley. ‘‘Confocal Raman Microspectroscopic Study of the Molecular Status of Carotenoids in Tomato Fruits and Foods’’. Appl. Spectrosc. 2011. 65(2): 127-134. 28. E.Y.M. Bonnist, J.-P. Gorce, C. Mackay, R.U. Pendlington, P.D.A. Pudney. ‘‘Measuring the Penetration of a Skin Sensitizer and Its Delivery Vehicles Simultaneously with Confocal Raman Spectroscopy’’. Skin Pharmacol. Physiol. 2011. 24(5): 274-283. 29. N.J. Everall. ‘‘Otimising Image Fidelity in 3D Confocal Raman Imaging’’. Paper presented at: SciX Conference. Kansas City, MO; Sept 30-Oct. 5 2012. 30. J.J. Andrew, T.M. Hancewicz. ‘‘Rapid Analysis of Raman Image Data Using Two-Way Multivariate Curve Resolution’’. Appl. Spectrosc. 1998. 52(6): 797-807. 31. J. Wang, P.K. Hopke, T.M. Hancewicz, S.L. Zhang. ‘‘Application of Modified Alternating Least Squares Regression to Spectroscopic Image Analysis’’. Anal. Chim. Acta. 2003. 476(1): 93-109.
1416
Volume 67, Number 12, 2013
32. P.D.A. Pudney, M.E. Melot, P.J. Caspers, A. Van Der Pol, G.J. Puppels. ‘‘An In Vivo Confocal Raman Study of the Delivery of Trans Retinol to the Skin’’. Appl. Spectrosc. 2007. 61(8): 804-811. 33. M. Me´lot, P.D.A. Pudney, A.-M. Williamson, P.J. Caspers, A. Van Der Pol, G.J. Puppels. ‘‘Studying the Effectiveness of Penetration Enhancers to Deliver Retinol through the Stratum Cornum by In-Vivo Confocal Raman Spectroscopy’’. J. Controlled Release. 2009. 138(1): 32-39. 34. R. Paquin, P. Colomban. ‘‘Nanomechanics of Single Keratin Fibres: A Raman Study of the a-Helix ! b-Sheet Transition and the Effect of Water’’. J. Raman Spectrosc. 2007. 38(5): 504-514. 35. M. Gniadecka, O.F. Nielsen, D.H. Christensen, H.C. Wulf. ‘‘Structure of Water, Proteins, and Lipids in Intact Human Skin, Hair, and Nail’’. J. Invest. Dermatol. 1998. 110(4): 393-398. 36. A. Kuzuhara, T. Hori. ‘‘Reduction Mechanism of L-cysteine on Keratin Fibers Using Microspectrophotometry and Raman Spectroscopy’’. Biopolymers. 2005. 79(6): 324-334. 37. N.J. Everall. ‘‘Confocal Raman Microscopy: Why the Depth Resolution and Spatial Accuracy Can Be Much Worse Than You Think’’. Appl. Spectrosc. 2000. 54(10): 1515-1520. 38. N.J. Everall. ‘‘Modeling and Measuring the Effect of Refraction on the Depth Resolution of Confocal Raman Microscopy’’. Appl. Spectrosc. 2000. 54(6): 773-782. 39. L.J. Wolfram, K. Hall, I. Hui. ‘‘The Mechanism of Hair Bleaching’’. J. Soc. Cosmet. Chem. 1970. 21: 875-900. 40. G. Mao, C.R. Flach, R. Mendelsohn, R.M. Walters. ‘‘Imaging the Distribution of Sodium Dodecyl Sulfate in Skin by Confocal Raman and Infrared Microspectroscopy’’. Pharm. Res. 2012. 29(8): 2189-2201.
