Forensic Science International 208 (2011) 124–128

Raman spectroscopy of blood samples for forensic applications

Samantha Boyd a, Massimo F. Bertino b,**, Sarah J. Seashols a,*a Department of Forensic Science, Virginia Commonwealth University, Box 843079, Richmond, VA 23284, USAb Department of Physics, Virginia Commonwealth University, Richmond, Box 842000, VA 23284, USA

A R T I C L E I N F O

Article history:

Received 3 January 2010

Received in revised form 6 October 2010

Accepted 22 November 2010

Available online 3 January 2011

Keywords:

Forensic science

Blood detection

Raman spectroscopy

Hemoglobin

Field analysis

Biospectroscopy

A B S T R A C T

We investigated Raman scattering from human blood as a function of parameters that are relevant for

forensic field analysis, such as substrate, sample dilution, individual from which the sample originates,

and age of the sample. Peaks characteristic of blood components and in particular the hemoglobin peaks

were routinely detected when blood was deposited on substrates that are not strongly luminescent, such

as plastic, metal utensils and dry wall. Raman scattering from blood proved quite sensitive and blood

samples with a dilution up to 1:250 could be measured for an excitation power of �2 mW measured at

the sample plane. The sensitivity of Raman scattering to diluted blood allowed measurement using blood

reconstituted from fabric substrates, thereby alleviating issues related to luminescence and scattering

from the substrate. The dependence of Raman scattering on sample age and individual was also

investigated. We found that the relative intensities of scattering peaks depended on sample age and

history. For example, the relative intensity of oxyhemoglobin peaks decreases after blood has dried.

Fresh blood drawn directly from a donor without intermediate storage exhibits also scattering peaks at

1155 and 1511 cm�1 which disappear after drying. The origin of these peaks is under investigation. We

noticed, however, that they were not found in blood that had been stored for longer than one week in

EDTA containers before analysis, thus requiring the use of fresh blood for future studies and validation

purposes. The relative intensity of scattering peaks was also found to be somewhat dependent on the

donor and, for a same donor, on the day on which blood was drawn.

� 2010 Elsevier Ireland Ltd. All rights reserved.

Contents lists available at ScienceDirect

Forensic Science International

journal homepage: www.e lsev ier .com/ locate / forsc i in t

1. Introduction

Raman spectroscopy is a very sensitive technique that is oftenused in forensic laboratories, mostly to analyze textile fibers andpaints. [1,2] Surprisingly, Raman spectroscopy has seldom beenused to analyze blood and body fluids for forensic applications.Recent research by Lednev et al. [3–6] and by De Wael et al. [7,8]has shown, however, that Raman spectroscopy can be employed todiscriminate blood from other body fluids [3,4], human blood fromfeline blood [5] and blood from paint and stains [8]. Thanks to thesefindings and to the recent availability of hand-held and portableRaman spectrometers [9], Raman field analysis of blood and bodyfluids may soon become a reality. To this end, several parameterssuch as influence of substrates, dilution limits and excitationwavelength must be investigated to establish reliable collectionprotocols. Studies should also be carried out to determine, forexample, if Raman scattering from blood depends on the donor, onthe clinical condition of the donor, and on the history of the sample.

* Corresponding author. Tel.: +1 804 828634.

** Corresponding author.

E-mail addresses: mfbertino@vcu.edu (M.F. Bertino), sseashols@vcu.edu

(S.J. Seashols).

0379-0738/$ – see front matter � 2010 Elsevier Ireland Ltd. All rights reserved.

doi:10.1016/j.forsciint.2010.11.012

To our knowledge, only one study of this kind has been carried out[6]. This study showed that Raman spectra of blood samples do notdepend strongly on the donor. It was also shown that the spectraarise from two main components. The major component ishemoglobin, which yields scattering peaks at 1000, 1368, 1542and 1620 cm�1, respectively, for excitation at 752 nm. The secondcomponent is likely fibrin, which gives rise to peaks at 967, 1248,1342 and 1575 cm�1, respectively, again for excitation at 752 nm.

In this study, we present data that are relevant to establish fieldcollection protocols such as preferred excitation wavelength,dilution limits and influence of substrates. We show that dilutionsof up to 1:250 can be detected using a laser power of a few mW.Thus, Raman scattering has a sensitivity comparable to manypresumptive tests for blood, such as the phenolphthalein (Kastle–Meyer), leucomalachite green, Hemastix1 and the forensic lightsource tests. [10,11] These tests are highly sensitive (up to1:5,000,000 dilutions can be measured in forensic laboratories),but in the field a cotton swab test is generally employed which canonly measure dilutions on the order of 1:150. Because of thesensitivity of Raman scattering to diluted blood, a protocol for thereconstitution of blood from fabrics was established whichalleviates issues related to luminescence and scattering fromsubstrates. Reconstitution is a destructive procedure, but Ramanscattering yields definite answers regarding the identity of a

[()TD$FIG]

Fig. 1. Dependence of Raman scattering on excitation wavelength.

S. Boyd et al. / Forensic Science International 208 (2011) 124–128 125

sample. Therefore Raman analysis of reconstituted blood can bemore advantageous than presumptive techniques. We alsoanalyzed the dependence of Raman scattering on sample ageand donor. We found substantial differences between fresh and dryblood and we also found that the spectra of freshly drawn blooddiffered from the spectra of blood stored in EDTA containers. Inaddition to the peaks of dry blood, fresh blood analyzedimmediately after drawing from a donor exhibited peaks at1155 and 1511 cm�1 which disappeared upon drying. These peakswere evident for up to 1 h after drying, and they could be used todetermine the age of blood found at a crime scene. The peaks at1155 and 1511 cm�1 were not noticed in blood that had been keptin EDTA containers for longer than one week. A reduction in therelative intensity of the oxyhemoglobin peaks was also noticed indry blood. The relative intensities of Raman scattering from dryblood were found to be relatively consistent between donors andbetween different days for a same donor. Some variations,however, were noticed in the oxyhemoglobin peaks.

Overall, our results indicate that Raman scattering can be safelyemployed to identify blood deposited on a substrate, even forcomparatively high dilutions. However, in our opinion, anyanalysis based on the relative intensities of Raman peaks shouldbe accompanied by a thorough study of the dependence onindividuals.

2. Materials and methods

2.1. Sample preparation

Blood was collected from eight volunteers of both genders andvarious ethnicities. In most cases, blood was drawn from donorsand immediately measured at different times after collection. Fordilution and reconstitution studies a few ml of blood were drawnfrom a volunteer, collected in acid (EDTA) Vacutainers1 and storedat 4 8C until use. For analysis, approximately 100 ml of blood wasplaced on a substrate. Generally, the substrate was a microscopeslide coated with commercial aluminum foil. For comparisonpurposes, silicon wafers and uncoated microscope glass slideswere also employed. To simulate blood from the crime scene, oneor more 100 ml aliquots of neat blood were placed on commercialfabrics such denim, flannel, or cotton, on dry wall and eatingutensils such as plastic forks and allowed to dry. Reconstitutedblood was prepared by excising an approximately 3 mm2 portionof the stained fabric. The excised portion was placed in a 2 mlcentrifuge tube with a small amount of water (typically 500 ml).The tube was capped and flick-mixed for approximately 1 min.100 ml of the reconstituted blood was pipetted from the tube andallowed to dry on the aluminum foil.

2.2. Raman spectroscopy

A Horiba Jobin Ivon HR 800 Raman Spectrometer was used forall measurements. This spectrometer is coupled to an Olympusconfocal microscope and is equipped with a liquid nitrogen-cooledCCD detector (Synapse). The excitation sources were a He–Ne laseremitting at 632.8 nm with a power of 30 mW and a diode-pumpedsolid state (DPSS) laser emitting at 532.11 nm with a power of50 mW. The laser power at the sample plane was measured with apower meter (Thor Labs) to be 9 mW at 632.8 nm and 2 mW at532.1 nm. The lower power of the 532.1 nm laser is due to adifferent optical path. The DPSS laser had a higher divergence thanthe 632.8 nm laser and it was mounted about 1.4 m away from thespectrometer. The 532.8 nm laser was mounted about 10 cm fromthe sample. The excitation light was focused on the sample with a10� or a 50� objective. In the initial stages of the experiments,measurements were repeated several times at regular time

intervals to ensure that the incident light was not altering thesamples. These experiments ensured that Raman analysis wasnon-destructive at the powers employed for the measurements. Allspectra presented here were not manipulated except for thesubtraction of a constant background and removal of cosmic rays.This simplified analysis was deliberate, since in field applicationstime and simplicity of analysis are of the essence.

3. Results and discussion

The goal of our experiments was to outline optimal proceduresand to pinpoint potential pitfalls for the use of Raman spectroscopyfor forensic examination of blood.

3.1. Excitation wavelength

Fig. 1 reports spectra of blood samples excited by red light(632.8 nm, lower spectrum) and, respectively, by green light(532.1 nm, upper spectrum). The laser power was measured at thesample position and it was about 9 mW for the 632.8 nm laser andabout 2 mW for the 532.1 nm laser. Excitation with 532.1 nm lightyielded peaks that were in overall agreement with those reportedby previous experiments focusing on Raman scattering from blood.The frequencies of the Raman peaks are reported in Fig. 1 and theirattribution can be found in standard reference works such as[12,13]. For the purposes of this paper, the most relevant peaks arethose at 1365 and 1636 cm�1, which are attributed to oxyhemo-gobin, and, respectively, to the ferrous low spin band. Illuminationwith red light yielded a scattering signal that was barelydistinguishable from the background. The higher scatteringcross-section of the 532.1 nm excitation is in agreement withmeasurements recently reported in [8] and is due to the differentexcitation mechanisms. Both 532.1 and 632.8 nm wavelengthsscatter resonantly by coupling with electronic transitions of bloodporphyrins. However, green light couples to the highly intenseSoret band (located between 410 and 450 nm in human blood). Redlight couples to an absorption at 625 nm. This absorption arisesfrom porphyrin–Fe charge transfer and it is more than 100 timesless intense than the Soret band. It is not surprising that the Ramanscattering cross-section scales with the absorption cross-sections.

3.2. Dilution

For forensic applications, a determination of the maximumsample dilution for which measurements can be carried out is very

[()TD$FIG]

Fig. 2. Raman scattering of diluted blood samples. The excitation wavelength was

532.1 nm.

S. Boyd et al. / Forensic Science International 208 (2011) 124–128126

important. In Fig. 2, we show that a 1:50 dilution yields Ramanscattering peaks which are clearly identifiable. The maximumdilution for which at least the most intense peaks were detectablewas 1:250. Higher dilutions yielded weak, unresolved features, asshown in the lower spectrum of Fig. 2. Dilutions higher than 1:250could probably be measured by employing a higher excitationpower, which, however, might damage the sample. The dilutionlimit observed in our experiments also allows us to quantify thescattering probabilities of 532.1 and 632.8 nm light shown in Fig. 1.Our apparatus can measure blood dilutions of at least 1:250 when532.1 nm light is employed. Therefore, we can estimate theexcitation probability with 632.8 nm light to be at least 250 � 9/2 = 1125 times lower than the excitation probability with532.1 nm light (the factor 9/2 accounts for the different laserpowers measured at the sample plane, as described in theExperimental Section).

3.3. Substrate

In Raman spectroscopy, fluorescence and scattering fromsubstrates can prevent detection of analyte signals. We comparedRaman scattering from blood samples deposited on substratessuch as a microscope slide (borosilicate glass), a silicon wafer, apolyethylene cup and a microscope slide coated with commercialaluminum foil. Scattering peaks characteristic of blood, and in

[()TD$FIG]

Fig. 3. Raman scattering of blood on different substrates. (a) Non-luminescent substrates

flannel). The indicated amounts of blood were deposited on a commercial flannel subs

particular the hemoglobin peaks at 1584 and 1636 cm�1 wereidentifiable on all substrates. In many cases, substrate peaks werealso detected, except for the aluminum foil, which should beconsidered as the preferred substrate for the analysis of fresh andreconstituted blood. Raman spectra of blood deposited onsubstrates commonly encountered at the crime scene, such asplastic and metal utensils, dry wall and fabrics yielded thecharacteristic hemoglobin and fibrin peaks provided that thesubstrate was not strongly luminescent. Fig. 3a shows Ramanspectra of blood deposited on a plastic fork and on a dry wall. Thepeaks of hemoglobin were well-evident on both substrates. Ondrywall, additional peaks at 445 and 610 cm�1 were also detectedwhich could be reconciled with TiO2 [14]. On luminescentsubstrates such as most commercial fabrics, however, Ramanscattering was only marginally useful. Even when luminescencewas manageable, the spectra were dominated by scattering peaksfrom the fabric. Fig. 3b shows the Raman spectrum of blood onflannel cloth. The spectrum obtained by depositing 100 ml of bloodon the fabric shows Raman peaks superimposed on a strongluminescence background. Of the Raman peaks only those in the1600 cm-1 region could be reconciled with blood. When a largeramount (300 ml) of blood was deposited on the fabric, lumines-cence decreased and Raman scattering peaks were more evident.However, hardly a feature could be attributed unequivocally toblood. In our opinion, analysis of blood samples on stronglyluminescent substrates should be regarded as a last-resortanalysis.

3.4. Reconstitution of blood

Given the complications of the analysis of blood on fabrics, weattempted to measure Raman scattering from reconstituted blood.For this, approximately 3 mm2 of the blood-stained fabric wereexcised and placed in approximately 500 ml of water in a 1.5 mlmicrocentrifuge tube. After flick-mixing the tube for 1 min,approximately 100 ml of the liquid was allowed to dry onaluminum foil. Reconstituted blood samples exhibited a strongRaman signal as shown in Fig. 4. More importantly, luminescenceand scattering peaks deriving from the fabric such as the peak at1550 cm�1 in the flannel sample did not dominate the spectrum.

3.5. Sample age and history

Fig. 5a shows Raman spectra of fresh and dry blood. The bloodwas drawn from a donor and analyzed immediately. Most peakscoincide; however, fresh blood exhibits peaks at 1155 and

(a plastic fork, and, respectively, dry wall). (b) Luminescent substrates (commercial

trate and measured. The excitation wavelength was 532.1 nm for all spectra.

[()TD$FIG]

Fig. 4. Raman scattering from reconstituted blood samples. The excitation

wavelength was 532.1 nm.

S. Boyd et al. / Forensic Science International 208 (2011) 124–128 127

1511 cm�1 which are not found in dry blood. The origin of thesepeaks is not clear; however, they were found in all donors and theirintensity decreased rapidly during drying. The intensity of thepeaks at 1155 and 1511 cm�1 was below the detection limit of ourapparatus about 60 min after drawing the blood. The relativeintensity of the oxyhemoglobin peaks at 1369 cm�1 and, respec-tively, 1636 cm�1 was also found to decrease in the dry sample.

[()TD$FIG]

Fig. 5. Raman scattering from fresh and dry blood. (a) Blood analyzed immediately after b

at least one week. The excitation wavelength was 532.1 nm.

[()TD$FIG]

Fig. 6. Variations of the relative scattering intensities (a) between

When blood was drawn from a donor and stored refrigerated in aEDTA Vacutainer for longer than about one week, the peaks at 1155and 1511 cm�1 were very weak or altogether absent, as shown inFig. 5b. These findings show that the relative intensity of theRaman scattering peaks depends on sample history. Mostimportantly, the findings show that forensic protocols should beestablished using freshly drawn blood.

3.6. Dependence on donor and day of collection

We will now focus on differences in the relative scatteringintensities of blood drawn from different donors and from thesame donor at different times. In all but a few cases, the relativeintensities of the spectra coincide within error with the spectrumof donor A reported in Fig. 6a. However, the relative intensities ofthe oxyhemoglobin peaks can vary between individuals (Fig. 6a)and can also depend on the time of collection (Fig. 6b). Thevariation of the relative scattering intensities between individualsand for the same individual is in apparent contrast with the resultsreported by Lednev et al., who found scattering intensities to bequite consistent for a sample of fourteen individuals [6]. However,we used a different excitation wavelength, and this might explainour results. In our case, 532.1 nm light was used as excitationsource. When an excitation wavelength of 532.1 nm is used like inour study, hemoglobin Raman bands in the 1550–1650 cm�1

region are particularly strong, and their intensity can be used tomonitor the level of blood oxygenation [15]. The differences

eing drawn from a donor. (b) Blood stored in an EDTA container and stored at 4 8C for

different donors and (b) for the same donor in different days.

S. Boyd et al. / Forensic Science International 208 (2011) 124–128128

between donors and within the same donor reflect, probably, adifferent level of blood oxygenation, and thus could potentially beuseful in forensic pathology applications.

4. Conclusions

In conclusion, identification of scattering peaks characteristic ofhemoglobin and fibrin in dry blood is possible using Ramanspectroscopy to dilutions up to 1:250. Direct examination of driedbloodstains on evidence is recommended when luminescencefrom the substrate is not strong. Reconstitution in a minimalvolume of water (on the order of 500 ml), followed by placing asmall quantity of the reconstituted sample on a non-fluorescentsubstrate such as aluminum foil will yield optimal spectra foridentification of the sample as containing hemoglobin. Realistical-ly, a portion of the sample is consumed, just as it would be forpresumptive tests for blood. However, given that the quantity ofsample required for a positive identification is very small, Ramanspectroscopy certainly has potential for blood identification, bothin the lab and at the crime scene. We also found that relativeintensities of Raman peaks depend on the individual and thehistory of the samples, but can also vary considerably for the sameindividual. Therefore, any Raman analysis based on relativeintensities should be accompanied by a robust analysis of thedependence within an individual and between individuals.

References

[1] J. Thomas, P. Buzzini, G. Massonnet, B. Reedy, C. Roux, Raman spectroscopy andthe forensic analysis of black/grey and blue cotton fibres part 1: investigation ofthe effects of varying laser wavelength, Forensic Sci. Int. 152 (2005) 189–197.

[2] E.M. Suzuki, M. Carrabba, In situ identification and analysis of automotive paintpigments using line segment excitation Raman spectroscopy: I. Inorganic topcoatpigments, J. Forensic Sci. 46 (2001) 1053–1069.

[3] K. Virkler, I.K. Lednev, Raman spectroscopy offers great potential for the non-destructive confirmatory identification of body fluids, Forensic Sci. Int. (2008).

[4] K. Virkler, I.K. Lednev, Raman spectroscopic signature of semen and its potentialapplication to forensic body fluid identification, Forensic Sci. Int. 193 (2009)56–62.

[5] K. Virkler, I.K. Lednev, Blood species identification for forensic purposes usingRaman spectroscopy combined with advanced statistical analysis, Anal. Chem. 81(2009) 7773–7777.

[6] K. Virkler, I.K. Lednev, Raman spectroscopic signature of blood and its potentialapplication to forensic body fluid identification, Anal. Bioanal. Chem. 396 (2010)525–534.

[7] L. Lepot, K. De Wael, F. Gason, B. Gilbert, Application of Raman spectroscopy toforensic fibre cases, Sci. Justice 48 (2008) 109–117.

[8] K. De Wael, L. Lepot, F. Gason, B. Gilbert, In search of blood—detection of minuteparticles using spectroscopic methods, Forensic Sci. Int. 180 (2008) 37–42.

[9] See, for example, the RMP-300 series of portable Raman instrumentation offeredby Jasco Inc., http://www.jascoinc.com/literature/pdf/spect/rmp_300.pdf (lastaccessed: Dec. 23, 2009), the InPhotote instrument offered by InPhotonics,http://www.inphotonics.com/spectrometers2.htm (last accessed: Dec. 23,2009), or the EZRaman-M offered by Enwave Optropnics, Inc., http://www.enwaveopt.com/EZRaman-M.htm (last accessed: Dec. 23, 2009).

[10] J.L. Webb, J.I. Creamer, T.I. Quickenden, A comparison of the presumptive luminoltest for blood with four non-chemiluminescent forensic techniques, Lumines-cence 21 (2006) 214–220.

[11] M. Cox, A study of the sensitivity and specificity of four presumptive tests forblood, J. Forensic Sci. 36 (5) (1991) 1503–1511.

[12] T.G. Spiro, T.C. Strekas, Resonance Raman spectra of hemoglobin and cytochromeC: inverse polarization and vibronic scattering, Proc. Natl. Acad. Sci. U.S.A. 69(1972) 2622–2626.

[13] X. Hu, T.G. Spiro, Tyrosine and tryptophan structure markers in hemoglobinultraviolet resonance Raman spectra: mode assignments via subunit-specificisotope labeling of recombinant protein, Biochemistry 36 (1997) 15701–15712.

[14] M.F. Bertino, B. Smarsly, A. Stocco, A. Stark, Densification of nanoparticles withvisible light, Adv. Funct. Mater. 19 (1–6) (2009).

[15] I.P.F. Torres, J. Terner, R.N. Pittman, E. Proffitt, K.R. Ward, Measurement ofhemoglobin oxygen saturation using Raman microspectroscopy and 532 nmexcitation, J. Appl. Physiol. 104 (2008) 1809–1817.