Forensic Chemistry 3 (2017) 1420

Contents lists available at ScienceDirect

Forensic Chemistry
journal homepage: www.elsevier.com/locate/forc

Reviews

Comments on the multiple headspace-solid phase microextraction

(MHS-SPME) technique for dating inks
Antonio A. Cant
Scientific Consultant (Retired U.S. Secret Service), 1520 E. Fremont Street, Laredo, TX 78040, USA

a r t i c l e i n f o a b s t r a c t

Article history: A recent article by San Romn et al. (2015) treats the problem of determining the age of ink writing using
Received 10 September 2016 the MHS-SPME technique. Ballpoint pen inks are studied since they tend to age more slowly than other
Accepted 25 October 2016 writing inks (i.e., they contain solvents that evaporate more slowly than those in other writing inks). The
Available online 2 November 2016
MHS-SPME technique basically determines, via multiple extraction steps, the fraction b of ink solvent
molecules that remain after an extraction from a closed vial. During the extraction process, the molecules
Keywords: in the closed vial must be in a state of equilibrium. Using this fraction and the number of solvent mole-
Questioned documents
cules extracted during the first extraction, the total number of solvent molecules in the ink sample can be
Ballpoint ink dating
Volatile ink components
computed. The authors determine the age of ink via the behavior of b with ink age rather than the behav-
Multiple headspace solid phase ior of total amount of solvent molecules in the ink with ink age. They found that ln b increases linearly
microextraction (MHS-SPME) with ln t where t represents the ink age. The authors also found that b is dependent on the amount of
GC/MS ink examined. The purpose of this presentation is to briefly review the MHS-SPME technique and to
address these latter two findings.
2016 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2. Brief derivation of some relevant equations of the MHS-SPME technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1. Basic definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2. The Equation N f aN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3. Multiple sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4. The parameter b . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3. The constancy and mass independence of a and b . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4. Behavior of b with ink age. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.1. Choice of an analytical form for F i t . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2. Consequence of a positive or negative Di ki  ki1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5. Experimental observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.1. Two experiments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.1.1. Experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
5.1.2. Experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.2. Results of experiment 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.2.1. Obtaining the parameters bt for each of five solvent in a one-year old ink. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.2.2. Obtaining the ln btv s: ln t aging graph for PE only. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.3. Results of experiment 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.3.1. Obtaining the parameters bt for each of five solvent in a 15-day old ink . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.3.2. Obtaining the ln btv s: ln t aging graph for each of five solvents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.4. Implications from the two experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
6. Comparison of the theoretically-derived bt and the experimentally-obtained bt. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

E-mail address: aacantu@msn.com

http://dx.doi.org/10.1016/j.forc.2016.10.008
2468-1709/ 2016 Elsevier B.V. All rights reserved.
 A.A. Cant / Forensic Chemistry 3 (2017) 1420
1. Introduction
The recent article by San Romn et al. [1] introduces a novel
technique that examines the volatile components of ballpoint
pen inks for the purpose of determining their age. The authors
use the multiple headspace-solid phase microextraction (MHS-
SPME) technique which basically sequentially extracts multiple
times the volatile components released from a sample housed in
a closed vial. It uses the chemically-coated fiber in the SPME device
to collect the volatile components and this fiber is subsequently
placed in a gas chromatograph with a mass spectrometer detector
(GC/MS) for separating, identifying, and quantifying the volatile
components (analytes) removed.
The volatile components are distributed among several regions
(phases) within the closed vial, the major ones being the head-
space, the sample, and the coated fiber (later the paper on which
the ink lays and the inside surface of the closed vial are added).
There are four key requirements or conditions that must be
met for this technique to work correctly. One is that, during each
extraction step, these volatile components should be in a state of
equilibrium among the several phases. A second condition is that
the partition (distribution) coefficients between the volatile com-
ponents and each of the phases mentioned remain constant after
each extraction step. A third condition (similar to the second one
given) is that the volumes of the places where the volatile com-
ponents reside after reaching equilibrium should also remain
constant after each extraction step. The forth condition is that
the number of analyte molecules in the SPME coated fiber should
be proportional to the peak area of that analyte when it is ana-
lyzed in the gas chromatograph.
We will see that the reason for imposing the first three condi-
tions is so that the ratio of the analyte molecules in the SPME
coated fiber to the total number of analyte molecules present in
the closed vial remains constant for each extraction step.
If these four requirements are met, then a plot of ln Ai , where
Ai is the measured amount of a given (or chosen) volatile compo-
nent (analyte) during the ith extraction, against the extraction
number (i  1) is a linearly decreasing graph whose slope is
ln b. The parameter b is the fraction of the total of number of ana-
lyte molecules existing in the closed vial that remain during an
extraction step. Knowing b and the amount of analyte extracted
during the first extraction, A1 , then the total amount of the ana-
lyte, AT , present in the sample can be computed using an equa-
tion given below.
The theory of the multiple headspace-solid phase microextrac-
tion technique has been treated by numerous authors. Of note is
the article by Ezcuerro et al. [2] and the one by Tena and Carillo
[3], both of which are exemplary and have relevant references on
the topic.
2. Brief derivation of some relevant equations of the MHS-SPME
technique
In the MHS-SPME technique a sample is introduced into a
vial, capped with a cap having a septum, heated, and a SPME
needle is inserted. Once the volatile components that are emit-
ted into the headspace reach equilibrium among the different
phases (or locations) they are in or go into, the SPME needle is
removed and inserted into a GC/MS instrument to determine
the composition of the components (analytes) in the headspace.
A series of peaks result in the chromatogram and the peak area
of each is determined. This process is repeated several more
times keeping in mind that one has to wait until equilibrium
is established.
15
In what follows the possibility that, after equilibrium is reached,
some of the analyte molecules exist in/on the paper and some exist
adhered to the inner glass wall of the closed vial is considered.
2.1. Basic definitions
The following definitions will be used in deriving the key
equations:
N = The total number of analyte molecules in the closed vial at
equilibrium (i.e., the molecules are distributed in the headspace,
the coated fiber, the ink sample, the paper, and the inner glass sur-
face of the vial. This is the number of analyte molecules originally
in the ink before equilibrium in the closed vial is established.)
Note that the number of analyte molecules can be converted to
the mass of the analyte by dividing it by Avogardos number and
multiplying it by the molecular mass of the analyte.
Nf = The number of analyte molecules adsorbed on/in the SPME
coated fiber at equilibrium.
Similarly defined are the numbers N h , N s , N p , and N g which cor-
respond to the number of analyte molecules in the headspace, in
the ink sample, in/on the paper, and on the glass, respectively. It
then follows that,
N Nh Nf Ns Np Ng 1
Vf = The volume of the SPME coated fiber that adsorb the ana-
lyte molecules at equilibrium (units are in a chosen unit of
volume).
Similarly defined are the volumes V h , V s , V p , and V g which cor-
respond to the volume of the headspace, the ink sample, the paper,
and the glass surface, respectively. The volume of the glass surface
can be taken to be the volume of the layers of analyte molecules
that adhere to the glass. In essence it is not critical to know how
the volumes are defined so long as it is known that such a volume
exist. The important thing to know is that these volumes should
not change in going from one extraction step to the next (see the
third condition given above).


N
C f V f = The concentration of the analyte molecules
f
adsorbed on/in the SPME coated fiber at equilibrium (units are in
number of analyte molecules per a chosen unit of volume).
Similarly defined are the concentrations C h , C s , C p , and C g which
correspond to the concentration of the analyte molecules in the
headspace, the ink sample, the paper, and the glass surface,
respectively.


Nf



Cf V N
K fs C s Nfs Nfs VV s = The distribution (partition) coeffi-
V s f
cient between the analyte molecules in/on the SPME coated fiber
(as concentration in the numerator) and those in the ink sample
(as concentration in the denominator).
Since there are five phases in which the analyte molecules can
exist (headspace, fiber, ink sample, paper, and glass), there exist
a total of (5  4)/2 = 10 possible distribution coefficients (also
called distribution constants). It is a simple exercise to show that
from the following four one can generate the remaining six: K hf ,
K hs , K hp , and K hg . To show this, the following properties (which
are also simple to show) are used: for j; k; l h; f ; s; p; and g
1
K jk 2
K kj
K jk K kl K jl 3
Combining Eqs. (2) and (3) yields another useful relation,
K jk K kl
K jk K kl K jl 4
K lk K kj
16 A.A. Cant / Forensic Chemistry 3 (2017) 1420

By the definition of a distribution coefficient K jk , the following During the first extraction:
useful relation follows, N 1 aN analyte molecules are removed
    N  N 1 N  aN N1  a analyte molecules remain
Vj Nj
K jk for j; k h; f ; s; p; and g 5 During the second extraction:
Vk Nk
N 2 aN1  a analyte molecules are removed
N  N 1  N 2 N1  a  aN1  a N1  a1  a N1  a2
2.2. The Equation N f aN analyte molecules remain
During the third extraction:
Using the above equations one can show that for each of the five N 3 aN1  a2  analyte molecules are removed
phases (denoted as h; f ; s; p; and g), there exist a fraction denoted as
N  N 1  N 2  N 3 N1  a2  aN1  a2  N1  a2 1  a
f k that relates N k and N (here k h; f ; s; p; and g). This relation is
N1  a3 analyte molecules remain
Nk f k N 6
During the ith extraction:
By Eq. (1), it follows that, N i aN1  ai1  analyte molecules are removed
fh ff fs fp fg 1 7 N  N 1  N 2      N i N1  ai analyte molecules remain
Defining the parameter b as b 1  a and noting that
A laborious but instructive exercise is the derivation of the frac- N 1 aN, the above equations becomes
tion f f associated with the SPME coated fiber, N i N 1 bi1 analyte molecules are removed
  N  N 1  N 2      N i Nbi analyte molecules remain
K fs V f
ff a 8 The above equation for the number of analyte molecules
K fs V f K hs V h V s K sf V p K sg V g
removed, also shown below, is the first of the two key equations
As noted, this fraction is given the designation of a. This is done of the MHS-SPME method,
in order to have agreement with what is found in the literature for
this fraction. The term in parenthesis, namely, K sf V p K sg V g , Ni N 1 bi1 10a
involves the paper matrix phase and the glass surface phase. As The natural logarithm of this is,
can be seen, its inclusion lowers the value of a. For the case of
the analyte molecules that are adhered to the coated fiber, Eq. ln Ni ln N1 i  1 ln b 10b
(6) becomes, This says that the plot of [ln N i v s:i  1] is a straight line with
Nf aN 9a slope of ln b.
Since N 1 aN and a 1  b, it follows that,
Or, equivalently,
N1
N 11
N 1b
a f 9b
N
This is the second of the two key equations of the MHS-SPME
The parameter a is thus that fraction the total number of ana- method.
lyte molecules distributed throughout the closed vial (i.e., N) that Using Eq. (10a), the following can be shown:
P1 P1 i1
i1 N i aN
are on the SPME coated fiber (it contains N f analyte molecules).
i1 b N. From this, Eq. (11) can also be derived,
These N f molecules are subsequently removed for analysis. but not as easily as it was done above.
In the literature one finds the same equations but using the
2.3. Multiple sampling peak areas resulting from the chromatographic analysis of the con-
tent on the SMPE coated fiber. Here the fourth condition is imposed
Since the focus of the SPME method as well as the MHS-SPME which is that,
technique is in the amount of analyte molecules that adhere in/
Ai kN i ; 12
on the coated fiber (which is the amount that gets extracted from
the closed vial and analyzed), in what follows only this phase will Here Ai is the peak area corresponding to N i , the number of ana-
be considered. The presence of the analyte molecules residing in lyte molecules removed during the ith extraction and k is a propor-
the other phases is incorporated in the parameter a (seen during tionality constant.
the derivation of Eq. (8)). A key condition that must be met (see
condition 2 and 3 above) is that this a parameter remains the same 2.4. The parameter b
during each extraction. This happens when all the distribution
coefficients and all the volumes (e.g., those of the headspace, the Regarding Eq. (10a), note that if i in N i N 1 bi1 is replaced with
SPME coated fiber, the ink sample, the paper, and the glass surface) i 1, then this i begins with 2, (i 2; 3; . . .) and N i N 1 bi1
remain constant during the extraction steps.
becomes N i1 N 1 bi . If N i1 N 1 bi is divided by N i N 1 bi1 , the
In what follows, the number N f of the analyte molecules that
result is
are adhered to the SPME coated fiber at equilibrium during the
ith extraction can be denoted as N f i , however, the subscript f Ni1
b 13a
for fiber will be dropped as it is understood that from here on, Ni
only the coated fiber is considered. Thus, N f i will simply written This can also be written as,
as N i . N2 N3 N4 Ni1
If N is the total number of analyte molecules initially in the b  i 1; 2; 3; . . . 13b
N1 N2 N3 Ni
closed vial and N 1 is the total number of analyte molecules adhered
to the SPME coated fiber during the first extraction, then after this Eq. (13a) with i 1 is simply b NN21 NN
N
1
1  a. Since a NN1
extraction, N 1 analyte molecules are removed and the number of is the fraction of analyte molecules that were removed (extracted)
analyte molecules that remain is N  N 1 . In summary: during the first extraction, b is then the fraction that remains.
 A.A. Cant / Forensic Chemistry 3 (2017) 1420 17

Denote the number of analyte molecules that remain after the are sufficient conditions for the constancy of a. Clearly, if a is con-
ith extraction as, stant then so is the parameter b 1  a.
Since a is a ratio the number of analyte molecules in/on the
Ri N  N1  N2  N3      Ni coated fiber (N f ) to the total number of analyte in the vial (N) at
when i 0; N0  0; thus R0 N 14 the time of extraction, any change in N will change N f in the same
way since a is constant. Thus a and b are independent of the
Then an equivalent equation to Eq. (9a) is,
amount of sample examined; that is, they are mass independent
Ni1 aRi 15a (invariant).
Mass dependence would happen if a (or b) is not constant from
The parameter b then simply becomes one extraction to the next. One way for this to happen is for at least
one of the first three conditions given above is not met. In the work
Ni1 Ri Ri1  Ni
b 16 of San Romn et al. [1] mass independence was not achieved for b
Ni Ri1 Ri1
(see the last line of their page 108) even though in all their mea-
Note that when i 0, R0 N and b NN21 NN 1
(as given above). surements they tried to use the same amount of ink (i.e., they used
N
Eq. (16) basically shows that the parameter b is the fraction of ana- the same number of 1.2 mm ink-on-paper micro-discs removed
lyte molecules that remain after an extraction of what remained from the document). This was done to reduce the mass depen-
from a previous extraction. As indicated above, this fraction is dence between measurements. That is, in comparing measure-
the same for each extraction. b  100% is the percent of analyte ments that tend to depend on the amounts sampled, it is
molecules that remain when an extraction takes place while important that any differences between the samples removed be
a  100% is the percent of analyte molecules that are removed as small as possible (i.e., one should compare samples that have,
(lost) during an extraction. as close as possible, the same amount of ink). However, in their
Since b is a fraction (meaning that b < 1), then it follows that case, that did not provide mass independence.
N 1 > N 2 > N 3 >    > N i > N i1 (i.e., each extraction extracts less Thus, in the case of San Romn et al. [1], a reason why b is not
than what was previously extracted). mass independent may be that one or more of the first three con-
ditions given above was not met. However, it was noted that the
addition of water (as a displacer) appears to correct the variation
3. The constancy and mass independence of a and b of b with mass (private communication with Dr. Magdalena
Ezcurra). Water as a displacer is defined later under experiment 2.
Consider a closed vial with an ink-on-paper sample in it and a
SPME needle inserted in it through the cap septum. If the volatile
4. Behavior of b with ink age
components in closed vial were not in equilibrium when the SPME
needle is removed, then the number of analyte molecules removed
At this point, the dependence on time, t, will be introduced into
is less than the number that would have been removed had one
the terms that depend on time. The time t used by San Romn et al.
waited for the volatile components to come to a state of equilib-
[1], is in days and begins with t 1 day. From Eq. (13a), the param-
rium. This is the key problem of working in non-equilibrium con-
eter bt takes the following time-dependent form,
ditions. This is why it is imperative that during the extraction
process, the components should be in equilibrium. Tena and Car- Ni1 t
bt t P 1
rillo [3] have a section that treats non-equilibrium situations. N i t
Eq. (8) defines a and Eqs. (9a) and (9b) show its key property.
For an ink of age t, N i t is the number of analyte molecules
Values of a are true and valid only if the extracted analytes were
removed by the SPME coated fiber during the ith extraction and
in a state of equilibrium during the extraction process (condition
it decreases to zero as t increases without bound due to the evap-
1 above). For the case of three phases (headspace, fiber, and ink
oration of the analyte molecules with time. If the function F i t is
sample), a takes the following form, t
defined as F i t NNi1 where N i 1 is the number of analyte mole-
2 3 i
  cules on the SPME coated fiber at time equal to one day, then,
K fs V f 1
a 4

5 N i t can be written as,
K fs V f K hs V h V s Vh
1 K hf V K sf VV s
2 3
f f
Ni t N i 1F i t t P 1 18
   
1 Nf Nf The function F i t begins at 1 at t 1 day (i.e., F i 1 1), it
4

5 17
1 Nh
Ns N h N f N s N decreases with time (i.e., F i t 1 > F i t 2 for t1 > t2 ), and it
Nf Nf
approaches zero as the time increases without bound (i.e.,
For the ith extraction step, an equivalent form of Eqs. (15a) and limt!1 F i t 0).
(9b) is,
4.1. Choice of an analytical form for F i t
Ni1
a 15b
Ri An example of an analytical function that behaves like F i t is
Thus, for any extraction step, a is the fraction of the total num- certainly an exponential decreasing function such as
ber of analyte molecules distributed throughout the closed vial
F i t1 eki t1 t P 1 19a
that are on the SPME coated fiber (and are subsequently removed
for analysis). Here ki is the rate constant. But there are other analytical func-
From Eq. (17) it is clear that if any of the distribution coeffi- tions that behave like F i t which could be candidates for F i t. Two
cients or if any of the volumes are not constant between extrac- examples of these (both of which are derived from the above expo-
tions then a will also not be constant. Thus requiring that all the nential function) are the following,
distribution coefficients and all the volumes remain constant
1
(given as condition 2 and 3 above) guarantees that a will be con- F i t2 t P 1 19b
1 ki t  1
stant. These requirements (as well as the equilibrium requirement)
18 A.A. Cant / Forensic Chemistry 3 (2017) 1420

F i t3 tki t P 1 19c between ki and ki1 significantly lengthens the time when bt
can become 1. The time when it does become one is denoted by
F i t2 is derived from F i t1 above by retaining only the linear
tc (c for crossing). It is the time when the graph of N i t and that
term of the Maclaurin expansion of eki t1 as shown below, of N i1 t cross (they are equal at tc ).
1 1 Ni1 t c
F i t1 eki t1 ! F i t2 bt c 1 22
eki t1 1 ki t  1 R Ni t c
1
Note that since Eq. (13b) states that the value of bt is the same
1 ki t  1
for any value of i 1; 2; 3; . . ., it follows that,
Here R is the remainder of the Maclaurin expansion of eki t1
k1  k2 k2  k3 k3  k4    ki  ki1 23
after the linear term.
F i t3 is derived from F i t1 by retaining only the linear term That is, D1 D2 D3    Di . Consequently, what is true for
of the Maclaurin expansion of et1 as shown below, the consecutive pair N i t and N i1 t is true for the consecutive
pair N 1 t and N 2 t. Thus, the above statements could have been
F i t1 eki t1 et1 
ki
^ ki t R
1 t  1 R ^ ki written using D1 k1  k2 instead of Di ki  ki1 .
! F i t3 t ki Converting a rate constant ki to half-life t0:5 i gives a different
view of how N i t changes with time. This conversion, which is
Here R^ is the remainder of the Maclaurin expansion of et1 valid only if the exponential form is assumed, is,
after the linear term.
ln 2
Thus, if F i t1 is used as a candidate for F i t, then bt becomes, ki 24
t0:5 i
  
Ni1 t Ni1 1F i1 t Ni1 1 F i1 t
bt Thus,
Ni t N i 1F i t N i 1 F i t
b1eki ki1 t1 b1eDi t1 20a ln 2
Di ki  ki1 t 0:5 i1  t 0:5 i  23
t 0:5 i t 0:5 i1
Here
The meaning of having a positive Di is the following: If
Di ki  ki1 20b D1 k1  k2 > 0, then the content of the first extraction N 1 t
If F i t2 is used, bt becomes, decreases faster with time (since it has a shorter half-life) than that
   of the second extraction N 2 t (since it has a longer half-life). These
Ni1 t Ni1 1F i1 t Ni1 1 F i1 t two contents become equal at time tc and then they switch roles,
bt
Ni t N i 1F i t N i 1 F i t N 1 t is now smaller than N 2 t; both the crossing and the role-
 
1 ki t  1 reversal are unrealistic. This phenomena is a result of the models
b1 20c used for F i t. As stated earlier, it will be seen that the closeness
1 ki1 t  1
between ki and ki1 significantly lengthens the crossing time tc
If F i t3 is used, bt becomes, when btc 1.
   Even though the above statements about half-lives being valid
Ni1 t Ni1 1F i1 t Ni1 1 F i1 t
bt only if they are used with exponential functions, they can be used
Ni t N i 1F i t N i 1 F i t
for functions such as those like F i t2 and F i t3 shown above
tki1
b1 b1tki ki1 b1t Di 20d (which were derived from an exponential function), provided one
t ki realizes that these half-lives approximate the ones used in the
Note that Eq. (20d) can be obtained from Eq. (20a) as follows: exponential function.
D
eDi t1 et1  i 1 t  1 R ^ Di ! F i t tDi
^ Di t R
3 5. Experimental observations

Of the three analytical functions shown that qualify for describ-
ing F i t, the one that will be used will be t ki of Eq. (19c) as this
will be seen to fit experimental data.
4.2. Consequence of a positive or negative Di ki  ki1
If Di ki  ki1 < 0, then bt b1t Di decreases with time.
This decrease also occurs if the other two choices for F i t are used
in forming bt. For the second choice this decrease is seen when
ki < ki1 is used in Eq. (20c) since Di does not appear implicitly.
For t1 < t 2 ,
bt1 > bt 2 ) bt decreases with time
On the other hand, if Di ki  ki1 > 0, then bt increases
with time. This happens also when the other two candidates given
for describing F i t are used to form bt.
bt1 < bt 2 ) bt increases with time
This increase can go beyond 1, which should not be the case
since bt is a fraction and is always less than 1. This occurs as a
result of the models used for F i t. As will be noted, the closeness
The information provided in this section is taken from the arti-
cle by San Romn et al. [1]. In their work they examined the ink
from four medium point Bic ballpoint pens. These are listed in their
Table 1. Among the solvents found in the inks from the four ball-
point pens are the following: Hexylene glycol (HG), Benzaldehyde
(BZ), Phenol (Ph), Benzyl alcohol (BA), and 2-phenoxyethanol (PE).
Also found among some of the inks were diethylene glycol (DEG)
and 1-methyl-2-pyrrolidinone. The focus, however, is in the first
five listed above. This present article will focus mainly on PE. Of
the several experiments performed by the investigators, only two
of these will be discussed in this present article and these are des-
ignated here as experiment 1 and 2.
21a
5.1. Two experiments
5.1.1. Experiment 1
Six ink-on-paper discs (1.20 mm in diameter) were punched out
21b
from writing made on white paper (Inoscopia) with a Bic blue ball-
point pen (medium point & obtained in Tel Aviv) and these were
extracted (in 4 successive extractions) at 90 C for 15 min using a
65 mm PDMS/DVB fiber for each period of time. For this experi-
 A.A. Cant / Forensic Chemistry 3 (2017) 1420
ment, the inks of known age sampled were of the following age in
days: 1, 15, 61 (2 months), 213 (7 months), 304 (10 months), 912
(2.5 years), 1278 (3.5 years), 1642 (4.5 years), and 1825 (5 years).
5.1.2. Experiment 2
This is the same as experiment 1 but uses only one ink-on-
paper disc (1.20 mm in diameter) and has a trace amount of water
added to function as a displacer (i.e., a substance that has a higher
affinity for the paper matrix than the analyte; it helps to inhibit the
adsorption of the analyte back into the paper matrix); this was
extracted using 7 successive extractions (rather than only 4). For
this experiment the inks of known age sampled were of the follow-
ing age in days: 1, 15, 61 (2 months), 213 (7 months), 304
(10 months), 912 (2.5 years), 1642 (4.5 years), and 1825 (5 years).
5.2. Results of experiment 1
5.2.1. Obtaining the parameters bt for each of five solvent in a one-
year old ink
Figure 4A in the article by San Romn et al. [1] shows the plot of
[ln Ai tv s:i  1] for each of the five solvents found in that ink (i.e.,
HG, BZ, Ph, BA, and PE). The graphs are obtained using the equation
ln Ai t ln A1 t i  1 ln bt (see Eqs. (10b) and (12)). The lin-
earity (R2 ) of each of these five graphs is computed. All values of
R2 are above 0.920, which means that the graphs are quite linear.
The slope, ln bt, of each linear graph is negative, which means
that ln Ai t decreases linearly as the number of extractions, i,
increases. This is expected since the decrease in ln Ai t as i
increases means that ln Ai t > ln Ai1 t, which is equivalent to
Ai t > Ai1 t (i.e., each extraction extracts less than what was pre-
viously extracted, as stated earlier). The values of bt (for
t = 365 days) are: 0.895 for HG, 0.811 for BZ, 0.596 for pH, 0.583
for BA, and 0.508 for PE.
5.2.2. Obtaining the ln btv s: ln t aging graph for PE only
Figure 3 in the article by San Romn et al. [1] shows the plot of
[ln btv s: ln t] for each of the five solvents. The only graph that has
any linearity is the one for PE (R2 0:936). It increases as the age of
the ink increases. The ages used are the 10 listed above under
Experiment 1. Empirically the linear graph for PE has the form,
ln bt ln b1 b ln t 25
The vertical intercept is ln b1 0:905 and the slope is
b 0:028. One discrepancy which is not explained by San Romn
et al. [1] in their article is that in their Fig. 3, ln b1 0:905
(i.e., b1 0:405) and in their Fig. 4A, the value of b1 is com-
puted to be ln b1 0:677 (i.e., b1 0:508). The discrepancy
may be due error bars not being shown. Although the authors state
that precision (repeatability) studies were undertaken, unfortu-
nately little statistical data is given.
5.3. Results of experiment 2
5.3.1. Obtaining the parameters bt for each of five solvent in a 15-
day old ink
As stated above, this experiment uses only one ink-on-paper
sample, uses 7 successive extractions, and uses some water as dis-
placer. The graphs shown in Figure 6 in the article by San Romn
et al. [1] pertain to this experiment and are similar to those shown
in their Figure 4A discussed above for experiment 1. The linearity
of any of the graphs is higher than 0.987 indicating they are quite
linear. The values of bt (for t = 365 days) obtained in this case are:
0.669 for HG, 0.657 for BZ, 0.636 for pH, 0.498 for BA, and 0.361 for
PE.
19
5.3.2. Obtaining the ln btv s: ln t aging graph for each of five
solvents
Figure 7 in the article by San Romn et al. [1] shows the plot of
[ln btv s: ln t] for each of the five solvents and all five show some
linearity. The values of R2 are 0.462 for HG (which has the poorest),
0.769 for BZ, 0.841 for pH, 0.769 for BA, and 0.870 for PE (which has
the highest). The ages used are the 8 listed above under Experi-
ment 2. Empirically the linear graph for each of the five solvent
has the form as shown in Eq. (24). For PE the vertical intercept is
ln b1 1:105 and the slope is b 0:046. For the other solvents,
the vertical intercept and the slope are shown in their Fig. 7.
5.4. Implications from the two experiments
What is learned from these two experiments is that, for dating
this particular Bic ink, the graph of ln bt increases linearly with
ln t, where t is the age of the ink. Based on the linearity and upward
direction of these experimentally-obtained graphs, it could be pos-
sible that the graph could keep increasing, cross the time axis
(which occurs when ln bt 0), and even go beyond that, which
as mentioned above, is unrealistic. This possibility also exists with
the theoretically-derived graph [ln btv s: ln t] which is based on
Eq. (20d). This is discussed next.
6. Comparison of the theoretically-derived bt and the
experimentally-obtained bt
Eq. (20d) gives the theoretically-derived equation for bt,
which can be written as,
 
bt
t Di t P 1 26
b1 theoretical
From Eq. (25), the experimentally-obtained equation for bt
can be obtained as,
 
bt
tb t P 1 27
b1 experimental
The experimentally-derived parameter b corresponds to Di , the
difference between any consecutive pair of in rate constants. For 2-
phenoxyethanol (PE), the value of b 0:046 and b1 0:331
(obtained from ln b1 1:105) comes from experiment 2
described above.
The crossing time, tc , is computed from Eq. (25) using
btc 1.
ln bt c ln b1 b ln tc 0
 ln b1
ln t c
b
1:105
ln t c 24:02
0:046
tc 2:71x1010 days
The fact that Di 0:046 indicates that the difference between ki
and ki1 is small. This smallness is what kept the crossing from
occurring within the experimental age range considered (1 day to
5 years). In this model, it occurs many years later.
The behavior of the experimentally-obtained equation for bt is
identical to that of the theoretically-derived equation; however,
the theoretical treatment offers an explanation for the behavior.
For the case of the first two extractions (or any consecutive pair
of extractions), it shows that the analyte content of the first extrac-
tion decreases faster with time (the ink age) than does the analyte
content of the extraction that follows it.
20 A.A. Cant / Forensic Chemistry 3 (2017) 1420
7. Conclusion
The MHS-SPME technique was applied to study the aging of
ballpoint pen inks by Romn et al. [1]. A sample of ink-on-paper
is placed in a closed vial, heated, and a SPME device with a
chemically-coated fiber is inserted. When equilibrium is reached,
a fraction of the analyte molecules present get adhered to the
coated fiber. These get desorbed and analyzed using a GC/MS
system.
This extraction procedure is repeated several times making sure
that equilibrium is reached during each step. A plot of the natural
logarithm of the number of analyte molecules extracted vs. the
extraction number gives a decreasing linear graph whose slope is
the natural logarithm of bt, the key aging parameter considered
by Romn et al. [1]. The number of analyte molecules extracted
is proportional to the GC peak area corresponding to the analyte.
The aging parameter bt is the fraction of analyte molecules
that remain after an extraction of what remained from a previous
extraction. This parameter is found to vary with the age of the ink.
Because it is a fraction, then, in principle, this parameter should be
mass invariant (i.e., it does not depend on the amount of ink ana-
lyzed). But experimentally it was found to be mass dependent and
reasons are proposed to explain why.
Since one intuitively expects that the fraction of analyte mole-
cules extracted decreases as the age of the ink increases, an analyt-
ical form for this behavior is proposed. When this is used in bt,
the model explains when bt is expected to increase with time
and when it expected to decrease with time. Experimentally it
was found that bt increases with time. Here there is the possibil-
ity that it increases to the point that bt is no longer a fraction, but
becomes one and then larger than one. This is clearly unrealistic. At
least for the inks studied by Romn et al. [1], the model given for
bt shows that this unrealistic behavior occurs when at times
greater than one million days. This is way past the usual number
of years (e.g., under five years) considered in studying the age of
inks.
If one includes in the theoretical development the possibility
that volatile molecules could adhere to the inner wall of the glass
vial and the possibility that they could enter into the porous paper
matrix, then it is shown that their inclusion reduces the magnitude
of at and, thus, increases the magnitude of the aging parameter
bt.
References
[1] I. San Romn, L. Bartolom, M.L. Alonso, R.M. Alonso, M. Ezcurra, DATINK pilot
study: an effective methodology for ballpoint pen ink dating in questioned
documents, Anal. Chim. Acta 892 (2015) 105114.
[2] O. Ezquerro, B. Pons, M.A. Tena, Multiple headspace solid-phase microextraction
for the quantitative determination of volatile organic compounds in multilayer
packagings, J. Chromatogr. A 999 (2003) 155164.
[3] M.A. Tena, J.D. Carrillo, Multiple solid-phase microextraction: Theory and
applications, Trends Anal. Chem. 26 (3) (2007) 206214.