Nanoscale control of interfacial processes for latent fi ngerprint enhancement

Latent fingerprints on metal surfaces may be visualized by exploiting the insulating characteristics of the fingerprint deposit as a “mask” to direct electrodeposition of an electroactive polymer to the bare metal between the fingerprint ridges. This approach is complementary to most latent fingerprint enhancement methods, which involve physical or chemical interaction with the fingerprint residue. It has the advantages of sensitivity (a nanoscale residue can block electron transfer) and, using a suitable polymer, optimization of visual contrast. This study extends the concept in two significant respects. First, it explores the feasibility of combining observation based on optical absorption with observation based on fluorescence. Second, it extends the methodology to materials (here, polypyrrole) that may undergo postdeposition substitution chemistry, here binding of a fluorophore whose size and geometry preclude direct polymerization of the functionalised monomer. The scenario involves a lateral spatial image (the whole fingerprint, first level detail) at the centimetre scale, with identification features (minutiae, second level detail) at the 100–200 mm scale and finer features (third level detail) at the 10–50 mm scale. However, the strategy used requires vertical spatial control of the (electro)chemistry at the 10–100 nm scale. We show that this can be accomplished by polymerization of pyrrole functionalised with a good leaving group, ester-bound FMOC, which can be hydrolysed and eluted from the deposited polymer to generate solvent “voids”. Overall the “void” volume and the resulting effect on polymer dynamics facilitate entry and amide bonding of Dylight 649 NHS ester, a large fluorophore. FTIR spectra demonstrate the spatially integrated compositional changes. Both the hydrolysis and fluorophore functionalization were followed using neutron reflectivity to determine vertical spatial composition variations, which control image development in the lateral direction.


Introduction
Fingerprints provide an example of pattern formation in nature, carrying information that uniquely identies an individual.Notwithstanding the rise of sophisticated genetic methods, they remain the cornerstone of many criminal investigations and have a number of non-criminal applications based upon identication of an individual. 1 The efficacy of this approach is in large measure associated with the complexity of a ngerprint and the consequent practical difficulty of forgery; powerful soware tools for analysis and recognition facilitate exploitation of this potential.The power of ngerprint evidence for analytical purposes in general is anecdotally recognized by the use of the term in other contexts, such as the "ngerprint" region of IR spectra and DNA "ngerprinting".The challenge in realizing this analytical opportunity lies in visualization of the interfacial chemical transfer that constitutes a ngerprint, primarily in the case of latent ngerprints, for which the nature 2 and extent 3 of the deposit mean that they are not immediately visible to the eye.Here we explore how this can be accomplished by a novel electroanalytical approach based upon spatially selective deposition of electroactive polymers with variable optical properties.The intellectual novelty lies in the need to control the (electro)chemistry on different length scales and in both the lateral and vertical directions.
When a nger contacts a surface, exchange of material with the surface leaves behind a trace of this contact which resembles the pattern present on the nger.Dependent on the substrate and the nature of the contact, the ngerprint may be visible, latent or plastic. 4Since they are not immediately visible to the eye, and thus less readily "wiped", latent ngerprints are the greatest source of forensic evidence.To give an indication of scale, in the UK on the order of 700 000 objects are ngerprinted per annum.In response to this demand, numerous methods and reagents have been developed to effect latent ngerprint visualization but, perhaps surprisingly, the operational success rate is only ca.10%.While there may be local variations in demand and/or success rate for specic types of object, the global need for improvement is clear.
The existence of ngerprint patterns was recognized in ancient times: 5,6 they have been identied on hand-formed building materials in Jericho dating back to 7000 BC and on the reverse of Chinese clay seals from 300 BC.Much later they were used as accompaniments to signatures by citizens claiming damages following the siege of Londonderry in 1691 5 and by the engraver Bewick who used an engraved ngerprint as a signature on his work. 6The rst documented study of ngerprints was in the 17th century by Grew, 5 followed by attempts at pattern classication in 1823 by Purkinje, 5,6 and later by Henry. 5evelopment of these historical observations for forensic application was reliant upon three crucial deductions made during the 19th century.First, Herschel 7 made the critical observationon his own handsthat ngerprints do not change during the life of an individual. 5Indeed, the friction ridge skin pattern that constitutes a ngerprint persists aer death, thereby enabling post mortem identication. 8Second, by removing skin from the ngers and allowing it to re-grow, it was demonstrated that injury does not change ngerprint patterns. 5hird, in 1892 Galton 9 estimated that the odds of two individuals having identical ngerprints were 64 billion to 1; thus, for all practical purposes, they are unique to an individual.Combination of these observations pointed to the value of ngerprints in criminal investigations. 10The outcome of this, in the rst decade of the 20th century, was establishment of the UK's rst ngerprint bureau 5,7 and the use of ngerprint evidence to secure a murder conviction. 6ingerprint patterns fall into three basic categories (so-called rst level detail): loops, whorls and arches. 5,11While differences at this gross level can clearly eliminate certain individuals, positive identication relies on the minutiae (or second level detail) within the pattern: these include such features as ridge endings, crossovers (bridges), short independent ridges, islands, bifurcations, spurs, dots and lakes. 5The standard for matching a crime scene ngermark to one from a database varies with jurisdiction: in some cases there is a set minimum number of points of similarity (although the number is not universal) and in others (including the UK) it is decreed a matter for a recognized expert to decide.As a rule of thumb, identication of around 16 points of similarity can be expected to be considered conclusive. 7,12There are also ner features (third level detail) present in a ngerprint image: these include the detailed shapes of the ridges and individual sweat pores.While not currently used in ngerprint identication, there is considerable research interest in third level detail, since it may in future permit analysis of smaller fragments of marks le by a nger, i.e. partial ngerprints.
Here we focus on latent ngerprints, since these are the primary source of forensic evidence.In general terms, the traditional approach has been to apply a reagent that interacted with the residues le by contact of the nger.We do not rehearse the diverse methods available, since these have been reviewed elsewhere recently, [13][14][15] but rather show how the apparently diverse methods used in fact have some similarities that limit their efficacy and motivate the novel approach developed here.The classical approach is to apply a powder (either dry or as a suspension) that adheres physically to the sweat residues; the powder may be uorescent, [16][17][18] magnetic 19,20 or thermoplastic.2][23][24][25][26] More recent developments involve (by dipping, spraying or gas phase delivery, according to the chemistry) ninhydrin solution, 27 vacuum metal deposition, 21 small particle reagent, 28 physical developer, 28 cyanoacrylate ("superglue") polymerization in conjunction with a suitable dye, 29,30 S 2 N 2 polymerization 31 and cadmium sulphide nanocomposites. 32In some cases the interaction of the reagent may be relatively unspecic (physical adhesion of powders), in others it may involve moderately specic chemistry (CdS binding with fatty acids and amino acids generally found in ngerprint deposits) and in other cases it may be very specic (reaction with secreted drug metabolites of antibody-functionalised nanoparticles. 33However, the common factor is that the reagent interacts with the deposited residue.This makes all these technologies vulnerable to loss or deterioration of ngerprint residue, e.g. as a consequence of ageing, environmental exposure or abuse (attempted washing).
The approach developed here is complementary to those described above, in that the "reagent" is applied to the bare substrate surface that lies between the deposited ngermark ridges.The means of accomplishing this is to use the ngerprint deposit as a "mask" or template, whose broadly insulating characteristics preclude electron transfer from a metal substrate to a solution precursor.Since electron tunnelling can only take place over very short distances, only a very thin layer (ca.1-2 nm) of ngerprint residue is requiredfar less (by an order of magnitude) than required in the conventional strategies listed above.We have recently demonstrated

View Article Online
proof-of-concept for this strategy 34,35 in the context of electropolymerization of aromatic solution precursors 36 to generate conducting polymer lms.Other applications of electrochemically-based methods of latent ngerprint visualization include imaging using the SECM [38][39][40][41] and deposition of Au nanoparticles. 42n the case of electroactive polymer enhancement, aer transfer to a background electrolyte, the electrochromic properties of these materials were exploited to adjust the visual contrast between the polymer and the metal substrate.Evaluation of the ability of this approach to visualize ngerprints subject to ageing in a range of environments suggests that there are practically relevant situations in which the methodology may be superior to currently employed methods, for example involving powders and cyanoacrylate. 37he aim of the present study is to extend the concept of the electrochromic polymer enhancement strategy in two signicant respects.First, we wish to explore the possibility of combining observation based on optical absorption (as above) with observation based on emission.Since the advantages, notably sensitivity, of uorescence detection are appreciated in ngerprint visualization, translation to practical application would be facilitated by existing instrumentation.In the context of "superglue" enhancement, subsequent use of uorescent dyes is a necessity, since the cyanoacrylate polymer is not coloured.This leads naturally to our second generic goal, use of electropolymerized lms permitting excellent spatial control but with sub-optimal optical properties.In particular, we wish to move from thiophene 35 and aniline 34 based materials to pyrrole-based systems, for which there is much greater opportunity for manipulation of properties by substituent chemistry on the pN-H functionality.Here we focus on the underlying fundamental (electro)chemistry of this approach.
The future viability of enhancement and analysis of latent ngerprints using this strategy relies on "writing" (electropolymerization and post-deposition reaction) and "reading" (absorption and emission observations) of spatial information at different length scales.The precise gures will vary from one individual to another, but typically this is at ca. 1 cm for rst level detail (the whole print), at 100-200 mm for second level detail and at 10-50 mm (third level detail).These feature sizes set the chemical challenge: control of enhancement chemistries with commensurate lateral spatial resolution to the feature size.However, the requirement for vertical resolution is somewhat different.Typical ngerprint deposits may be a few microns thick as-deposited, but evaporative and other environmental losses will typically decrease this to 100s of nm before enhancement is undertaken.This denes the vertical resolution required for the (electro) chemistry: there is an optimum to be found between the lower limit of detection of deposited material and over-lling 35 of the trenches between ngerprint ridge deposits.To summarize, this singular analytical challenge of uniquely identifying an individual based on electroanalytical visualization of their ngerprint requires simultaneous control of (electro)chemical processes on length scales from the nanometre to the centimetre regime and in both lateral and vertical directions.
For the neutron reectivity (NR) experiments, the working electrode was prepared by sputter coating gold onto a polished single-crystal quartz block (100 Â 50 mm, Gooch and Housego) coated with a monolayer of 3-mercaptopropyltrimethoxysilane (MPTS) (Sigma Aldrich) to promote adhesion.The nominal Au lm thickness was 20 nm.For other experiments, the electrodes were metal sheet, as indicated in the gure legends.The counter electrode was in each case a Pt gauze, of adequate size to ensure that the counter electrode reaction was not limiting.The reference electrode was a double junction Ag|AgCl|KCl (saturated) electrode.These were assembled into a standard three electrode cell conguration; for the NR measurements, the purpose built cell has been described elsewhere. 43,44strumentation NR measurements were performed on FIGARO 45 and D17 at the Institut Laue-Langevin (Grenoble, France) and on INTER 46 at the ISIS Facility of the Rutherford Appleton Laboratory (Harwell Oxford, UK).Static neutron reectivity measurements were performed ex situ in air ("dry") and in situ immersed in h 3 -and d 3 -acetonitrile before and aer the lm fabrication stage (see Fig. 3, below).Deuterated solvents were used to maximise contrast between the polymer and the electrolyte so that the solvation within the polymer could be probed.Kinetic measurements were recorded during the hydrolysis stage of the reaction (see Fig. 3), using time-of-ight instrumentation with l ranges of 2-30 Å (ILL) and 1.5-16 Å (ISIS).Using different incident angles, this provides an accessible momentum transfer range of 0.004 < Q/ ÅÀ1 < 0.12, where Q is dened as (4p/l) sinq; l is the wavelength of the neutron and q is the incident angle.(Strictly, this is Q z but since we only consider specular reection we use the simpler notation Q.)The collimation slits were set to give a beam footprint on the sample of 60 mm Â 30 mm; they also dene the Dq/q resolution.The Dl/l resolution is dependent on the source (spallation source at ISIS, reactor at ILL) and associated instrumentation (chopper settings at ILL).In both cases, the resultant resolution in momentum transfer was DQ/Q $ 2-3%.Data acquisition times were ca.1.5 h per static run and 10 min for kinetic runs.
Photographs were taken with a Canon A480 digital camera and were digitally enhanced using the GNU Image Manipulation Program 2.6.7.(G.I.M.P.).3D proles were recorded with a Zeta 200 Optical Proler.Reectance FTIR spectra were acquired with p-polarised radiation incident at a reectance angle of 55 using a Spectra-Tech reectance accessory mounted on a Bomem MB120 infra-red instrument.The lms were grown potentiodynamically (n ¼ 20 mV s À1 ) in the potential range 0.3 < E/V < E max , where the anodic limit E max (typically 1.15 V) was set to cap the anodic current to 4 mA.This procedure was designed to avoid uncontrollably rapid growth; the amount of deposited polymer was varied with adequately ne control by varying the number of deposition cycles.All measurements were made at room temperature, 20 AE 2 C. PPyFMOC lms were deprotected using a 30% v/v piperidine solution in CH 3 CN to yield PPyNH 2 .PPyNH 2 lms were reacted with 0.01 M Dylight 649 NHS ester (hereaer referred to as "Dylight", for brevity) (Thermo Scientic) in DMSO-pH7 phosphate buffer in water (1 : 9 ml).

Data analysis
The principles of neutron reectivity data analysis 47 and the issues arising for samples involving "wet" interfaces under electrochemical control 48 have been described elsewhere.The variation with momentum transfer of reectivity from an interface, R(Q), is determined by the depth prole of the scattering length density, Nb, where N represents the concentration of scattering atoms present and b is their scattering length; the value of b is isotopically unique and medium independent.The scattering length density of a composite medium, here a solvated polymer lm, is a weighted sum of the Nb values of its components: lm composition determines scattering length density and thence reectivity.Experimentally, we invert the process and use reectivity, R, to determine composition in practice, the volume fractions of polymer and solventby model tting the reectivity prole, R(Q).This was accomplished using the box-model approach, implemented in the Motot soware. 49

Electrochromic enhancement of latent ngerprintsbasic observations in absorption mode
Before attempting to exploit the substitution chemistry opportunities presented by the pN-H function in pyrrole, it is rst necessary to demonstrate that PPy lms can in fact be deposited with spatial selectivity directed by a ngermark on a metallic substrate.Fig. 1 and Fig. 2 show representative images of two ngermarks on 304 stainless steel substrates, following enhancement by electrodeposition of PEDOT and PPy, respectively.In panel (a) of each gure, the paler (nominally white) regions correspond to the ngermark itself, i.e. the complex mixture of materials secreted from pores along the ridges on the ngertip.The darker regions (blue for PEDOT in Fig. 1 and black for polypyrrole in Fig. 2) correspond to polymer deposited on the bare metal between the ngerprint deposits.The PEDOT enhanced image acts as a control; this material has previously been demonstrated to provide good visualization of latent ngermarks, with high delity and controllable visual contrast. 35e have quite deliberately shown examples that might be typical of real evidence, rather than highly controlled ("groomed") model examples.Thus, one can see evidence of damage to the ngermark and of adjacent ngermarks in Fig. 1 and of smearing (for example, caused by motion of the nger on the surface) and variable amounts of residue in Fig. 2.These and other imperfections represent practical challenges to be addressed.
At the coarsest level of interpretation, rst level detail, the ngerprints in Fig. 1 and Fig. 2 are, respectively, a loop and a whorl.While this is clearly not sufficient for identication purposes, it is clear that there is much second level detail present within these images.Panel (b) in each of Fig. 1 and Fig. 2 shows higher magni-cation optical images of the samples in panels (a).By any reasonable standard (see above), it would be possible to achieve an evidentially acceptable identication from images such as these using various combinations of second level detail.To illustrate the principle, we simply identify two such features on each image.It is also possible to identify a large number of pores, seen as dots along the ridges where the absence of contact with the substrate permits polymer deposition; these are examples of third level detail.

Overview of extension to emission mode visual enhancement
The qualitative conclusion of the preliminary experiments shown above is that polypyrrole can be added to the set of electrochromic materials (to date, polyaniline 34 and PEDOT 35 ) that permit visualization of latent ngermarks on metallic substrates by means of their optical absorption properties; note that the work of Bersellini 36 in this respect demonstrated the facility to electrodeposit polypyrrole, but did not go on to exploit its electrochromic properties.We therefore proceed to the more challenging goal of developing a polymeric system with suitable absorption and uorescence characteristics.Conceptually, the simplest approach is to functionalise the nitrogen of the pyrrole ring with a uorophore.In practice, the (necessarily) large size of most uorophores creates such steric hindrance that the substituted pyrrole monomers cannot polymerize.We therefore arrive at the three step strategy schematically represented in Fig. 3.
The essential idea is to functionalize pyrrole units with the uorophore postdeposition.However, in order to accomplish this, there is still a requirement to create sufficient free volume within the polymer lm to accommodate the uorophore units.The tactic employed is to polymerize not pyrrole itself, but an N-functionalized derivative that can readily be removed post-deposition.The balance to be struck is use of a substituent that is not so large as to preclude polymerization but that is large enough to create appreciable free volume.The functionality chosen was the widely used FMOC protecting group.Thus, we set out to polymerize the N-substituted FMOC derivative of pyrrole (PyFMOC), then hydrolyze and leach out the protecting group to leave an amine-functionalized polypyrrole lm.The semi-uid nature of the polymer lm means that the randomly distributed free volume can aggregate to generate "voids" of sufficient size to accommodate larger uorophore moieties.Choice of a uorophore with an ester functionality provides the means of covalent bonding by amide formation (see Fig. 3).
In the experimental realisation of the strategy indicated in Fig. 3, critical issues are the completion of hydrolysis, the extent to which departing FMOC is replaced by solvent cf.lm contraction, the vertical (perpendicular to the interface) spatial distribution of replaceable solvent and the penetration of uorophore into the lm.The outcomes, which will ultimately determine performance, are addressed using a combination of spectroscopic and neutron reectivity measurements.

Optimization of lm deposition
The limited aim of this study is establishment and assessment of the strategy of Fig. 3 on clean (i.e.non-ngermarked) metal surfaces.The deliberate absence of lateral spatial variation focuses attention on the required control of lm composition in the vertical direction.Fig. 4 and Fig. 5, respectively, show the voltammetric responses of PPyFMOC and PPyNH 2 lms on Au during electropolymerization (panels (a)) and aer transfer to background electrolyte (panels (b)).Both sets of responses provide information relating to polymer deposition.In the former instance, the integrated current provides cycle-by-cycle monitoring to facilitate deposition of the chosen amount of polymer.In practice, use of the cathodic half cycle response is better, since this is not complicated by contributions from the (irreversible) anodic polymerization current contribution.
The coulometric assay described above is reliant upon complete lm redox conversion on the experimental timescale; that this is accomplished is demonstrated by the data in Fig. 4b and Fig. 5b.In these measurements (following completion of lm deposition) there is no monomer present in the solution, so the issue of distinguishing polymerization and lm redox chemistry is irrelevant.For lms of suitable coverage, i.e. appropriate both to the NR experiment (giving multiple well-dened interference fringes within the accessible Q range) and to future forensic exploitation (not so thick as to obscure all image detail when a ngermark is present), we nd that the peak currents are linearly proportional to the potential scan rate (see Fig. 4c and Fig. 5c).This indicates complete redox conversion of the lm on the experimental timescale, validating a coulometric assay of the spatially integrated surface population of polymer.On this basis, the surface coverage, G/mol cm À2 , is determined as q/nFA, where q/C is the charge, n is the number of electrons transferred ("doping level", n ¼ 0.33) 50 , A is the electrode area and F is the Faraday constant.For the data shown, the nal polymer coverages are G PPyFMOC ¼ 20 nmol cm À2 and G PPyNH2 ¼ 19 nmol cm À2 , where in both cases the surface population is expressed in terms of monomer units.These surface coverages can be used to estimate a physical lm thickness, as follows.The molar volume of monomer units, V m /cm 3 mol À1 (i.e. the reciprocal of the volume concentration of monomer units), can be estimated as the quotient of monomer molar mass and density, RMM/r.The approximation here is that the monomer units in the polymer pack essentially the same as in pure monomer.For a compact, solvent-free ("dry") polymer, the lm thickness h* ¼ V m G.The RMM values of PyFMOC and PyNH 2 are 344 and 122 g mol À1 , respectively.Their respective densities are ca.1.2 and 1.0 g cm À3 .Combining these with the values of G PPyFMOC and G PPyNH2 from the previous paragraph, we estimate "dry" (i.e.collapsed, solventfree) lm thicknesses of ca.600 Å and 240 Å for the PPyFMOC and PPyNH 2 lms, respectively.2][53][54] This overall scenario is appropriate both for a NR experiment (to explore structure) and lling of the interridge "trenches" in a typical ngerprint deposit (to accomplish practical visualization).As a nal observation, the different thicknesses of these two PPyFMOC and PPyNH 2 lms containing essentially the same number of monomer units (irrespective of whether one compares two dry lms or two solvated lms) shows the potential for generation of free volume by FMOC elution from a lm.

Observation of changes in surface composition using spectroscopic measurements
Before addressing the more sophisticated issue of spatial distribution of active components within the lm, it is necessary to demonstrate that the strategy of Fig. 3 does indeed result in uorophore immobilization.In principle, one might consider accomplishing this using either a uorescence measurement or some other spectroscopic probe.Supercially, a uorescence measurement has the attraction of also providing a more direct functional appraisal.However, while this could demonstrate the presence of the uorophore, it might not do so if quenching were an issue; in either instance, it would not provide evidence of its immobilization.The concern here is that, unless covalent attachment to the polymer is achieved, facile entry of the uorophore could just as easily be followed by its elution upon further exposure to electrolyte, thereby jeopardizing the entire surface synthetic strategy.Consequently, we sought a spectroscopic probe able to provide direct evidence of uorophore-polymer binding.Qualitatively, this can be accomplished using vibrational spectroscopy, with particular focus on the presence (or absence) of carbonyl bands associated with the amide functionality, which will be present in an PPyFMOC lm (at the start) and a successfully functionalized PPy-Dylight lm (at the end), but should be absent in PPyNH 2 (following hydrolysis of the PPyFMOC lm).
Representative data are shown in Fig. 6.Trace (a), representing a PPyFMOC lm, has a strong absorption band at 1660 cm À1 .Trace (b), for the hydrolysed lm, has no signicant amide band, demonstrating removal of the FMOC functionality.Trace (c), for the hydrolysed lm aer exposure to Dylight solution, shows a strong band at 1630 cm À1 , showing the formation of an amide.Signicantly, the last of these observations demonstrates not only the permeation of uorophore into the lm but also its chemical immobilization.Overall, these data show elution of FMOC and binding of uorophore, but give no insight into the spatial distribution of the uorophore or the factor(s) limiting its nal population.We now address these issues using NR.

Determination of vertical spatial structure using neutron reectivity
The second stage of the strategy in Fig. 3 is the creation of free volume in the PPyFMOC lm by the hydrolysis and elution of the FMOC moieties.Importantly, we require this free volume to be created throughout the lm.Fig. 7 shows neutron reectivity data acquired at different times during the hydrolysis process, conducted in a deuterated solvent medium to optimise the contrast.The data shown in Fig. 7 in fact represent the secondand, in reaction terms, productivephase of the experiment.Since there was no means a priori to predict the timescale of the hydrolysis-elution process, the lm was initially exposed to a low (0.01 mM) concentration of piperidine, with the aim of slowing the reaction to a measurable rate.This approach turned out to be more than successful, in that the rate of hydrolysis was immeasurably slow.However, this had the advantage of providing a lm solvation prole at a true "t ¼ 0"; the outcome of this is cited below in the discussion of Fig. 7c.The hydrolysis solution was then exchanged to higher piperidine concentration (10 mM) and the data shown in Fig. 7a acquired.The time taken for cell mounting and alignment and for initiation of data acquisition was on the order of 30 min.Hence the time axis in Fig. 7c has an offset of this order; since there is no attempt to extract kinetic or diffusional parameters, this is not critical.Broadly speaking, the R(Q) proles of Fig. 7a comprise three regions.At low Q (here, Q < 0.0055 ÅÀ1 ) there is total external reection (R ¼ 1); this is a consequence of the substrate material the neutrons are transmitted through (quartz) having a lower scattering length density than the material they are reecting from (the deuterated solvent).Beyond this, there is a series of fringes whose periodicity is dictated by the thickness of the lms present (Au electrode and polymer) and whose amplitude is dictated by the scattering length density contrast and sharpness of the interfaces between layers.At high Q (here, Q > 0.06 ÅÀ1 ), the fringes seen are attributable to the Au electrode.Although these are not central to the aims of the experiment per se, their quantitation does assist tting of the full R(Q) prole from which we extract the polymer lm data.At intermediate Q, the Au-derived fringes are seen, but superimposed on these are fringes that result from the polymer lm.The latter do not persist to high Q as a consequence of the diffuse polymer/solution interface.
Independent of any model, we can make four deductions from the data.Trivially, the presence of a critical edge shows that the lm scattering length density is below that of the solution.The position of the critical edge, Q* ¼ (16pDNb) 1/2 , where DNb corresponds to the difference in scattering length densities between the two bulk phases in the sample.In this instance, the relevant bulk phases are the quartz block supporting the electrode and the bathing solution to which the lm is exposed; the scattering length densities are known for these materials, giving DNb ¼ 0.62 Â 10 À6 ÅÀ2 .Inserting this into the expression above, we estimate Q* $ 0.0056 ÅÀ1 , consistent with the experimental data of Fig. 7. Secondly, the time invariance of the fringes at high Q allows these to be assigned to the Au electrode, the composition and thickness of which (necessarily) do not vary with time.Thirdly, an Au electrode thickness of 210 Å can be estimated from the periodicity of the fringes, DQ ¼ 0.03 ÅÀ1 , via the equation DQ ¼ 2p/d (where d ¼ lm thickness); this is entirely consistent with the nominal thickness of 200 Å from the sputtering process.Fourthly, the higher frequency fringes present only in the intermediate Q region can be seen progressively to stretch out with time.Since the momentum transfer, Q, is in reciprocal space, this stretching corresponds to some element of contraction of lm thickness with hydrolysis of the FMOC groups.The quantitative question that follows is whether this contraction results in loss of some or all of the "void" volume generated by the FMOC departure.
To address this last question, detailed tting of the data is required.Full details of this standard procedure are given in the Experimental section and elsewhere, [47][48][49] but the essential points are as follows.The scattering lengths of all the components present (quartz, Au, PPyFMOC, the departing FMOC, the remaining PPy-amine and CD 3 CN solvent) are all known.The Au thickness is known (from a combination of fabrication protocol, bare electrode observations and the high Q data of Fig. 7).The unknowns are therefore the internal composition of the lm (predominantly, solvation level), lm thickness and the roughness of the polymer/solution interface.
The outcomes of the tting process are shown in Fig. 7b and 7c.In the rst of these, the scattering length density of the system, from the quartz block supporting the electrode through to the bulk solution, is shown as a function of

View Article Online
distance, z (perpendicular to the interface).The values for quartz, Au and solvent represent bulk values for the pure components.The sharp dip between the quartz support and the Au electrode represents the MPTS bonding layer used to ensure good Au adhesion.For the purposes of this work, the region of interest is from the Au electrode outwards.The interface between the Au and the lm is relatively sharp; a small amount of diffuseness (typically 10 Å) is required to account for the nite roughness of the Au surface.Interestingly, the polymer lm is not compositionally homogeneous at any point in the process: there is a relatively diffuse outer region (a transition from "bulk" lm to bulk solution over 40 Å) and the interior of the lm comprises compositionally distinct inner and outer regions.A single polymer layer with high interfacial roughness was insufficient to model the lm.The key outcome is a substantial shrinkage of the lm as hydrolysis proceeds.For the example shown, the dry lm thickness (measured in air) was 561 Å, and the solvated lm thickness (measured upon exposure to solvent, prior to hydrolysis) was 639 Å, and the latter shrunk to 443 Å following completion of FMOC hydrolysis and elution.The signicance of this in solvation terms is shown in the solvent volume fraction data of Fig. 7c.While the tting unquestionably demands that some lm inhomogeneity be recognized (see Fig. 7b), the difference in scattering length density (and thence solvation) between the inner and outer regions of the lm is relatively small.We therefore look at the average picture.The lm solvent volume fraction at the outset, as solvated PPyFMOC exposed to a low piperidine concentration (as explained above) is 0.41.By the end of the hydrolysis process, this increases to 0.60, but it subsequently falls to 0.52.The latter decrease is the result of polymer relaxation, which occurs on a longer timescale than the FMOC hydrolysis and elution.
It is of course not possible to make the analogous measurements in hydrogenous solvent (CH 3 CN) for the same lm, since the hydrolysis reaction is a onetime process.However, such measurements were made on a nominally identical lm and the outcome, in summary form for the analogue of Fig. 7c, was an initial solvent volume fraction of 0.35, rising to 0.58 immediately aer hydrolysis and subsequently relaxing back to 0.44.The common conclusion from these experiments is that FMOC removal increases the lm solvent volume fraction by Df S $ 0.2 in the short term, but this increase is subsequently diminished to Df S $ 0.1.While this may seem modest, we note that relatively small changes in solvent content can have profound effects on polymer chain mobility, for example as manifested in viscoelastic properties, 54,55 which would facilitate permeation of uorophore reactant.In an absolute sense, a replaceable solvent volume fraction f S > 0.4 is more than adequate to give a high (and thus visible) uorophore population.
This leads to consideration of the nal step in the scheme of Fig. 3, uorophore functionalization of the PPy-NH 2 lm.What reaction did occurand the signicant changes in R(Q) proles do unequivocally demonstrate change, quantied belowtook place within the rst 15 min of exposure to uorophore solution.We attribute this to the more uid-like environment of the lm following hydrolysis (see above).From a mechanistic perspective, this removed the opportunity to follow the kinetics of the process (largely due to instrumental issues such as sample alignment prior to measurement), but from a practical perspective in future application it is obviously benecial.
Consequently, on a separate (but nominally similar) lm to that of Fig. 7, a separate set of measurements were made, as follows.First, R(Q) data were acquired for a PPyFMOC lm in the dry state (solvent-free) and exposed to h 3 -and d 3 -acetonitrile, prior to hydrolysis.Second, the lm was hydrolysed (with no attempt to monitor the time-dependence of this process) and R(Q) data acquired for the resultant PPyNH 2 lm in the three environments (air, h 3 -and d 3 -acetonitrile).Finally, the lm was Dylight functionalised and R(Q) data acquired in the same three environments.This last part of the experiment is complementary to the hydrolysis step, in that the aim is entry of a large reactant to consume free volume, rather than elution of a large leaving group to generate free volume.As compensation for sacrice of any kinetic information, this suite of measurements provided data in different solvent contrasts for the same lm, which (through co-renement) gives greater certainty in tting.The resulting R(Q) proles are shown in Fig. 8a, grouped according to the lm environment; the general form of the proles is analogous to those of Fig. 7.
The key parameters of interest are lm thickness and solvent content at each stage.To extract these, we need to consider the contributions of the polypyrrole spine, uorophore and solvent components to the scattering length density.For the polypyrrole and solvent components, the scattering length and physical density are known.For the uorophore, whose structure is commercially protected, this is not so straightforward, but acceptable approximations are possible, as follows.The molar mass is on the order of 1000 Daltons and there are four sulfonate groups (to provide adequate solubility) on an essentially aromatic hydrocarbon skeleton.We thus have an entity that comprises ca.320 Daltons of "SO 3 À " and ca.680 Daltons of "CH"; we make the plausible approximation that the overall density is unity.With these physically reasonable approximations, combining the data sets for the h 3 -acetonitrile and d 3 -acetonitrile environments, we estimate that there is one uorophore entity for every ca.ve pyrrole monomer units.Physically, this is plausible, given the geometrical constraints of the cartoon representation of Fig. 3 and practically this is expected to be useful.The model scattering length density proles best tting the data are shown in Fig. 8b and the R(Q) ts (lines) are shown alongside the data (points) in Fig. 8a; the agreement is good across the accessible Q range.We particularly highlight four characteristics.First, the outer interfaces are diffuse.In the R(Q) proles, this accounts for the damping of the lm-based fringes.In terms of reactivity, it undoubtedly contributes to the faster permeation of the uorophore molecules.Second, once one progresses to the interior of the lm, its scattering length density, and thus composition, shows at most only modest dependence on depth (actually, none for PPyNH 2 ).This indicates that diffusion of reactant into the lm is not a limiting factor; if it were, then there would be a clear gradient of composition representing uorophore penetration.Third, despite the entry of uorophore into the polymer, the lms shrink slightly during the process (compare the traces for PPyNH 2 and PPy-Dylight).This indicates that the volume of solvent expelled exceeds the volume of uorophore entering, suggesting that transport processes are not so slow that mobile species (here, solvent) are trapped within the lm.Finally, while the outer interface is slightly sharper for the dry lm (prole not shown), the overall thickness is not much less than for the solventexposed lm.Since (see Table 1) there is still appreciable solvent in the immersed lms, this suggests that the lm does not collapse upon emersion.
The characteristics of the model proles are summarized in Table 1.Considering rst the thickness data, the dramatic collapse of the PPyFMOC lm upon hydrolysis (by ca.40%) is accompanied by an increase in solvent content of the resulting PPyNH 2 lm.This apparently counter-intuitive result is a consequence of the size of the FMOC group; recall the earlier estimations of lm thickness accompanying the coulometry, when it was noted that the FMOC group constitutes ca.60% of the lm volume.Turning to the solvent volume fraction, the values for PPyFMOC (f S ¼ 0.36) and PPyNH 2 (f S ¼ 0.47) are satisfyingly consistent Table 1 Summary of film thickness and solvent content values at each stage of assembly of the surface architecture represented schematically in Fig. 3. PPyNH 2 films could be modelled as a single layer (i.e. were internally homogeneous), so "inner" and "outer" regions are merged with those of f S ¼ 0.38 and f S ¼ 0.48 at the same points in the surface chemistry of Fig. 3 determined by averaging the outcomes of the two kinetic experiments of Fig. 7 and its h 3 -acetonitrile counterpart.

Functional viability of the uorophore-modied lm
This report focuses on construction of the interfacial architecture and, via the NR measurements, on establishing spatial control of uorophore immobilization within the polymer matrix; essentially, this represents the compositional and structural aspects of the strategy.In a subsequent phase of the work, the focus will shi to determination of the polymer chromophore and immobilized uorophore properties; these represent the functional aspects of the strategy.The latter will involve a substantive programme of measurements, notably as functions of excitation and observational wavelengths and of polymer charge state (i.e.doping level, manipulated via applied potential).Nonetheless, in advance of such a future report, there is merit in a forward look to establish at a qualitative level uorophore activity in the polymer lm context.The technical issue here is whether (or not) proximity of the uorophore sites to the underlying electrode results in uorescence quenching.
Fig. 9 shows two images of PPy-based lms representing two stagessimplistically, in the absence and presence of uorophore, respectivelyof the assembly process shown in Fig. 3.This preliminary observation does not attempt to address the spatial issue of imaging a full ngerprint, but focuses solely on the viability of uorophore emission when in the lm environment.The le hand image shows a section of a PPyNH 2 lm deposited on an Au substrate (using the procedure of Fig. 5, terminating at the cathodic end of a potential cycle to establish the undoped redox state), removed from solution and viewed ex situ (dry) under illumination by light of wavelength 640 nm.This control observation shows a few brighter areas, but nothing systematic or substantive.The right hand image shows a partially Dylight 649 ester functionalised lm, prepared as follows.A PPyFMOC lm was deposited on Au and hydrolysed (as discussed earlier with reference to Fig. 7), then a droplet of Dylight 649 NHS ester solution placed on one part of the surface, followed by rinsing with pure water (to remove unbound and surface/ exterior uorophore) and air drying (at room temperature for ca.15 min).The resultant lm was viewed ex situ under illumination with light of wavelength 640 nm.The top le part of the viewed region in panel (b) of Fig. 9 includes part of the droplet-exposed area.The sharply dened region of enhanced brightness is consistent with strong uorophore emission; note that the intensity was attenuated (see legend) so the distinction between the images in panels (a) and (b) is signicant.Although this is not a quantitative measure of the uorescence efficiency and does not totally exclude quenchingperhaps from uorophores sites closer to the electrode (see Fig. 8 for evidence of deep penetration of uorophore)it is clear that at least some uorophore sites are sufficiently distant from the electrode that they are not vulnerable to quenching.Pragmatically, the practical viability of the interfacial (electro)chemical strategy is established.

Conclusions
A combination of spectroscopic, electrochemical and neutron-based techniques provides the capability to follow and quantify deposition and subsequent functionalization of electroactive polymer lms relevant to latent ngerprint visualization.This approach has been used to explore the (electro)chemistry of pyrrole-FMOC electropolymerization and deposition, followed by hydrolysis of the FMOC leaving group, then permeation and bonding of the uorophore Dylight 649 NHS ester.By revealing the presence, removal and reintroduction of amide functionalities, FTIR spectroscopy demonstrates qualitative success of this postdeposition functionalization strategy.Electrochemistry provides control over (and coulometric assay of) the surface population of polymer, the electrochromic matrix into which the uorophore is introduced.Neutron reectivity provides insight into the vertical spatial distribution of the permeating uorophore and the changes in lm population of the solvent that must leave to create space for it.Together, these techniques provide insights into lm composition, structure and (in the cases of the electrochemical and neutron data) dynamics; simplistically, they address the tersely expressed questions "what, how much and where?" We have demonstrated that the portfolio of materials suitable for electrochromic enhancement of latent ngerprints can be extended from the previously used aniline and thiophene (PEDOT) families to include pyrrole-based materials.Specically, this was accomplished for the parent polypyrrole, N-propylamine functionalised pyrrole and FMOC functionalised pyrrole.In future, in addition to the substitution chemistry explored here, this will provide a wider colour palette with which to optimise latent ngerprint visual contrast against the substrate.
In the post-deposition functionalization of PPyFMOC, the hydrolysis process (leading to FMOC removal) is relatively slow (ca. 3 h under the conditions employed), which allowed the progress of the reaction to be monitored by neutron reectivity.The subsequent entry of uorophore (Dylight 649 NHS ester) is much more rapidtoo rapid to follow readily using neutron reectivitywhich is attributed to the higher solvent volume fraction and thence greater uidity of the lm.During this sequence of events, the solvent volume fraction rises by ca.0.2 immediately upon hydrolysis and FMOC elution, but then falls by ca.0.1, presumably as a consequence of polymer relaxation in the now highly plasticised lm.In absolute terms, the solvent volume fractions are 0.37 (AE0.03) for PPyFMOC prior to hydrolysis, 0.59 (AE0.01)immediately aer hydrolysis and 0.47 (AE0.04)aer relaxation (where these data originate from both kinetic and non-kinetic experiments).Aer uorophore entry, the lm contracts slightly.
Future prospects for a combined absorption/uorescence strategy in latent ngerprint enhancement appear promising.The next step is implementation of the strategy described here to ngerprinted surfaces.Having established control of reactivity and composition in the vertical direction at distance scales from 10-100 nm, this ne control can now be applied to the lateral direction.Since the ngerprint feature sizes are at the scale of >10 mm, the prospect of high resolution, high delity ngerprint images is excellent.

Fig. 1
Fig. 1 PEDOT enhanced sebaceous fingerprint on 304 stainless steel.Deposition conditions as in main text; deposition time, t dep ¼ 3000 s.Panel (a): whole fingermark image.Dark (blue) regions correspond to PEDOT and lighter regions to fingerprint deposit.Circles highlight examples of second level detail, a bifurcation and a ridge ending.Panel (b): optical microscope image of selected area from panel (a) (as defined by the rectangle).Larger dark circular regions within light areas represent individual sweat pores.Vertical and horizontal distance scales (expressed in mm) are relative.

Fig. 2
Fig. 2 PPy enhanced sebaceous fingerprint on 304 stainless steel.Deposition conditions as in main text; deposition time, t dep ¼ 3000 s.Panel (a): whole fingermark image.Dark regions correspond to PEDOT and lighter regions to fingerprint deposit.Circles highlight examples of second level detail, a crossover and a ridge ending.Panel (b): optical microscope image of selected area from panel (a) (as defined by the rectangle).Larger dark circular regions within light areas represent individual sweat pores.Vertical and horizontal distance scales (expressed in mm) are relative.

Fig. 4
Fig. 4 i-E responses for a PPyFMOC film.Panel (a): during potentiodynamic electropolymerization (v ¼ 20 mV s À1 ); panel (b): after transfer to a monomer-free background electrolyte (subsequent to the final cycle of panel (a)), during cycling at v ¼ 1, 2, 5, 10, 20, 50, 100 and 200 mV s À1 (increasing "outwards"); panel (c): variation of cathodic peak current (from curves in panel (b)) with scan rate.Solution compositions as described in the main text.In panels (a) and (b), chevron arrows indicate scan direction.In panel (a) large arrows indicate the time sequence and "1" indicates the first deposition cycle (note the nucleation loop).

Fig. 5
Fig. 5 i-E responses for a PPyNH 2 film.Panel (a): during potentiodynamic electropolymerization (v ¼ 20 mV s À1 ); panel (b): after transfer to a monomer-free background electrolyte (subsequent to the final cycle of panel (a)), during cycling at v ¼ 1, 2, 5, 10, 20, 50, 100 and 200 mV s À1 (increasing "outwards"); panel (c): variation of cathodic peak current (from curves in panel (b)) with scan rate.Solution compositions as described in main text.In panels (a) and (b), chevron arrows indicate scan direction.In panel (a) large arrows indicate the time sequence and "1" indicates the first deposition cycle (note the nucleation loop).

Fig. 6
Fig. 6 Reflectance infra-red spectra for a PPyFMOC film subject to the reaction sequence of Fig. 3. Trace (a): PPyFMOC film, as deposited; trace (b): PPyFMOC film after hydrolysis, notionally a PPyNH 2 film; trace (c): film of panel (b) after reaction with Dylight solution.Traces are arbitrarily offset vertically for visual clarity.Reaction conditions as in main text.The asterisks on traces (a) and (c) indicate the amide peaks referred to in the main text.

Fig. 7
Fig. 7 Time-resolved NR experiment for PPyFMOC hydrolysis to give PPy-voids.Panel (a): R(Q) profiles as a function of time (indicated by arrows); panel (b) model fitted scattering length density profiles as a function of time (from the data of panel (a)); panel (c): solvent volume fraction in the film at selected time intervals (see main text for comment on absolute values) during the hydrolysis.( : inner polymer layer;: outer polymer layer).

Fig. 8
Fig. 8 Panel (a): R(Q) data for a PPyFMOC film prior to hydrolysis (B), the PPyNH 2 film resulting from hydrolysis (O) and the PPy-Dylight film following exposure to the fluorophore (,).Points represent data; lines represent fits (see panel (b)).For visual comparison purposes, R(Q) profiles are group according to the ambient medium (see annotations).Data are progressively offset downwards for presentational purposes; for the dry and d 3 -acetonitrile exposed films, a critical edge is seen, below which R ¼ 1. Panel (b): model fitted scattering length density profiles for the film exposed to h 3acetonitrile and d 3 -acetonitrile at each of the three stages of the process: nominally PPyFMOC (red traces), PPyNH 2 (green traces) and PPy-Dylight (blue traces).

Fig. 9
Fig. 9 Panel (a): PPy film on Au (deposition procedure as in main text); this represents the control experiment, in the absence of fluorophore.Panel (b): PPyNH 2 film prepared by hydrolysis of a PPyFMOC film (deposition and subsequent treatment as in main text) after partial exposure to Dylight 649 NHS ester by contact with a droplet of fluorophore solution, and subsequent removal of excess fluorophore by rinsing.The top left part of the viewed region includes part of the droplet-exposed area.In both cases, the films were in the reduced (undoped) state and were viewed ex situ under illumination by light of wavelength 640 nm.In panel (b), the intensity was attenuated by a factor of 4.