EVIDENTIAL VALUE FROM INK-JET PRINTERS

Stephen Philip DAY, Lee SHUFFLEBOTTOM

Forensic Science Service, Huntingdon, United Kingdom

ABSTRACT: The widespread use of ink-jet printers has made it necessary for docu-
ment examiners to find a method of linking a questioned document to an individual
ink-jet printer. The identification and characterisation of defects produced by ink-jet
cartridges provides a means of achieving this.
This study was primarily to investigate the physical nature of defects in inkjet
printers and show how printer outputs and printers can be linked. Thus it was hoped
to establish whether or not ink-jet printers can be linked to their print-outs through
their physical appearance on the paper and if so to estimate the evidential value of
any defects or other physical features. By identifying such features it was hoped that
a protocol for the examination of this type of printer could be developed.
Defects can be easily introduced into printouts by blocking the electrical defects
that control a particular print-head. The behaviour of the defects thus produced is
different for different types of ink-jet print cartridges. By studying the behaviour of
such defects it is possible to devise a strategy for the examination to distinguish be-
tween different printers. Not all nozzles get blocked with the same frequency so if
a particularly rare nozzle gets blocked then the evidential value in linking that par-
ticular printer to the printout will increase.

KEY WORDS: Ink-jet printers; Document examination; Printing defects; Office ma-
chinery.

Problems of Forensic Sciences, vol. XLVI, 2001, 356–374

Received 24 April 2001; accepted 15 September 2001

INTRODUCTION

In recent years there has been a gradual move away from manual
multi-element fixed type bar typewriters and single element electric type-
writers containing golf balls and print wheels, towards computer based
word processors for the production of “typewritten” documents. This has
brought about an increase in the occurrence of computer printouts which are
produced using a variety of types of printers. These include laser printers,
inkjet printers and dot-matrix printers. As a result, document examiners
are likely to encounter such documents with more frequency, making it nec-
essary to have a means by which they can identify a particular printer as the
one used for production of a particular document.
Evidential value from ink-jet printers 357

The term “word processor” actually refers to the computer program used
by the keyboard operator to prepare a computer file of the document which is
to be printed. The appearance of the printed characters which finally appear
on the paper may be influenced by the type of printer used but to a large ex-
tent is governed by the word processing software which produced it.
With typewriters it is possible to relate a document to a specific machine
by examining and comparing any individual characteristics produced by the
machine i.e. damaged letter blocks, striking out of alignment etc. This is not
the case when attempting to separate computer printed documents which
are not produced mechanically. Printers are compatible with many different
computers which possess a range of fonts, sizes of text and spacing associ-
ated with word processing packages making the overall appearance of the
document and its typeface of little use when seeking to link a particular
printer to a questioned document. However the physical characteristics of
the print itself is determined by the printer and may be used to link a docu-
ment to a printer, particularly if consistent faults can be identified.
In inkjet printers the only moving part in the print-head is the ink itself.
The printer uses a grid of tiny nozzles to which specially formulated ink
flows from a reservoir. The computer sends an electrical signal to the
print-head to determine which nozzles fire for any particular letter. When
the ink from the inkjet printer hits the paper it spreads into the fibres and
gives the print a slightly ragged appearance. This appearance will depend to
some extent on the paper used. In general it is difficult to identify a particu-
lar machine or differentiate between work of several machines.
This study concentrates on the outputs from two printers – the
Hewlett-Packard Deskjet 500 and the Lexmark 1000. These are both single
colour printers, but the principles outlined here can also be applied to most
colour ink-jet printers.

INK-JET PRINTERS

Basic Operation
Hewlett-Packard invented the first non-impact printer in 1979 and com-
mercially introduced the ThinkJet in 1984 [4]. Since then, the ink-jet printer
has become the most dominant output device used with the personal com-
puter as it has several advantages over other printing approaches, including
low printing cost, reduced printing noise, lighter weight and less expensive
printing apparatus. The difference in cost is especially significant for colour
printing. However, the print quality of inkjet printers is generally not as
good as that of other approaches, which can provide higher resolution. The
358 S. P. Day, L. Shufflebottom

technology of ink-jet printing is advancing considerably and many manufac-

turers are adding new features constantly to meet the demand for high qual-
ity resolution and for improved colour graphics [5, 9].

Ink delivery
There are two systems in which ink is delivered to the paper. Early types
of inkjet printers [8] used a continuous-stream method that generated
a steady, constant stream of charged ink droplets which could be deflected
towards or away from the printing surface by means of an electrical field.
Due to their complicated structure this method is only employed now by
some uni-colour inkjet manufacturers and for special applications such as
printing on packaging. Second is the drop-on-demand method that only sup-
plies ink droplets to the paper when they are instructed by the computer to
form a printed image. This is the most popular method of delivery for the
ink-jet print-head in the commercial market. This method is described in
more detail below.

Drop-on demand print mechanism

Today most ink-jet printers use one of two principal drop-on demand
methods to propel ink. Thermal inkjet printers use heat to generate a bubble
that creates an actuating force, and piezoelectric printers use electrically
driven actuators to pump ink from a chamber. In development at the mo-
ment is an electromagnetic printer which uses a tiny ball to squirt ink from
a nozzle. The work described in this paper was performed using thermal
ink-jet printers, so the mechanism for this is described in more detail below.

Thermal printers
A thermal ink-jet print-head consists of series of tiny holes or nozzles in
a small metal plate. Behind each plate is an electrical resistor which is con-
nected by flexible circuitry to a number of electrodes on the outside of a car-
tridge. These in turn connect with other electrical contacts within the
printer and through which the print-head is controlled. The other major
component of this cartridge is the ink reservoir which contains a special ink
with a high thermal expansion coefficient. A diagram of a thermal ink-jet
nozzle is shown in Figure 1.
Evidential value from ink-jet printers 359

heating element bubble

Fig. 1. Thermal ink-jet nozzle.

In operation a voltage pulse is passed through the resistor which causes

it to heat up rapidly. The large heat flow generated is transmitted to the ink
through heat conduction and consequently produces a vapour bubble. The
temperature at which the bubble forms is called the “superheat limit”, which
depends on the chemical composition of the ink and ambient temperature.
A typical value is about 300°C at atmospheric pressure for the ink used in
many commercial inkjet print-heads [2]. The heating resistors can produce
several thousand heat rises per second. As the bubble grows, momentum is
transferred to the surrounding fluid and ink is ejected at velocities of typi-
cally 10 metres per second. After completion of ink ejection, the cavities are
refilled from the ink reservoir by capillary force. Figure 2 shows a diagram of
the stages of operation a typical thermal print-head.
The ejected ink droplet usually has a comet shape with a long tail [6, 10].
The head and tail can separate during travel, the head becoming a separate
drop and the tail forming two or more satellite drops. The result of these sat-
ellite drops severely effects the printing quality of the print-head causing
the appearance of ink spatter.
360 S. P. Day, L. Shufflebottom

Ink heats up
heating resistor

Bubble forms

Bubble
expands

Bubble forces
out ink

Bubble cools
and contracts

Vacuum pulls more

ink into nozzle

Ready to
start again

Fig. 2. Stages in the operation of an ink-jet nozzle.

Inks
When ink is ejected from the cartridge upon the surface of the paper it
evaporates and penetrates to make a permanent mark. The inks used in the
past were made up of a carrier (water or oil) and pigment combination [8]. To
prevent the ink from spreading and thereby reducing the image quality, spe-
cial clay coated papers were used to absorb the carriers. Manufacturers have
recently developed new inks which eliminate the need for coated papers [4].
This has been done in two ways:
– By using fluid inks that dry rapidly by evaporation. Problems have
arisen with this method as evaporation of some of the ink components
takes place in the cartridge causing a hard plug or clogging of the noz-
zles.
– By using special formulated inks which are “solids” at room tempera-
tures and become fluids when heated beyond their low melting points.
These solid plastic inks are liquefied by a heater inside the printer and
are then stored as a liquid inside the cartridge ready to be sprayed
Evidential value from ink-jet printers 361

from the nozzle. Once the ink is ejected the ink cools quickly and solidi-
fies on the paper surface. The advantage of this method lies in the fact
that no volatile carriers are used to dilute the inks so that evaporation
in the nozzle aperture is not a problem [8].

Types of defect and their causes

As with other types of printer, the principles that can be used to link
a given printout to a particular printer are essentially the identification of
defects and an assessment of their significance. It should be remembered
that the print-heads used for many ink-jet printers are associated with the
ink reservoir and are therefore replaced regularly. Consequently some of the
defects described below will disappear when the print-head is changed.
In ink-jet printers the characters are composed of dots such that the den-
sity is generally above 300 dots per linear inch and therefore the individual
dots are very difficult to detect. These printers may develop faults, but it can
be very difficult to detect them due to the fine dot structure. Since the
print-head moves horizontally, the defect can often be seen under a micro-
scope as a horizontal white line running through the characters. The causes
of some of the more common faults are listed below:
– With thermal ink-jet printers, the resistors are subjected to large pres-
sure changes in the nozzle cavity. As the bubble collapses, it creates
a microjet of fluid which impacts the surface of the resistors at high
pressure. This pressure can be up to 13 atmospheres and induces
cracks and craters. Once damaged it can initiate a chain of events that
can cause the resistor to break open and fail [3, 11]. There are also ther-
mal stresses exerted on the print-head due to the large temperature
changes that occur over micro-seconds while firing. Once a resistor has
failed, the nozzle will no longer print and a permanent defect will be
seen in the printout.
– Sometimes at high temperatures, the dye in the ink wall break down
into insoluble fragments that stick to the resistor surface. This build
up of fragments of degraded dye reduces the heat transfer to the ink
and in severe cases the ink cartridge will no longer fire [1]. This effect
is known as Kogation. Again, this will give rise to a permanent defect.
– At low temperatures, the thickening of the dye-carrying agent can
cause the nozzles to clog [7]. This can also happen when a nozzle has
been idle for some time as viscous plugs can form due to evaporation.
This type of defect is likely to be cleared when the head is cleaned and
is not therefore of much use to the document examiner.
– Occasionally, nozzles will misfire due to the continued heating of
a neighbouring nozzle causing the ink to reach its superheat limit.
This defect is intermittent.
362 S. P. Day, L. Shufflebottom

– A nozzle can appear to be blocked when the connection to the printer is

severed, either by an electrical fault (in which case it is likely to per-
sist) or by dirt on the electrical contact (which may or may not persist).
Hewlett-Packard are currently using a thermal inkjet system similar to
the one described above in their Deskjet range. As the print-head is attached
to their ink cartridges, the print-head is changed every time the cartridge
runs out of ink. HP print-heads therefore, are not subjected to thermal
stresses for a prolonged time. Some manufacturers also use Hewlett-Pack-
ard thermal print cartridges in their machines e.g. Brother while Lexmark
and Canon use their own thermal inkjet delivery system for most of their
more common printers. While the designs differ in the arrangement and
number of the nozzles, position of the electrical contacts etc. the basic compo-
nents of the cartridges remain the same.
Piezoelectric systems do not heat the ink and so there are no thermal
stresses exerted on the print-head. This should eliminate the need for peri-
odic print-head replacement, so the print-head will last the life of the
printer. Epson are currently using a system similar to this in their Stylus
range under the trade name MicroPiezoä. MicroPiezoä uses tiny crystal
pumps to control each ink droplet with absolute precision. The manufactur-
ers claim the technology produces more spherical dots and less satellite
droplets (or spatter) than the thermal systems. It also can achieve higher
resolution (up to 1440 dots per inch compared to 1200 dpi) as the print-head
does not have to wait while the ink cools before it can be re-fired, thus it can
operate at higher frequencies.

Ink-jet print cartridges

There are an increasing number of manufacturers of inkjet printers try-
ing to exploit this rapidly growing segment of computer printers. In this in-
vestigation, We chose to look at two machine/cartridge combinations in de-
tail to study the behaviour of faults within these machines. We also looked at
other machine/cartridge combinations from four of the leading manufactur-
ers namely Hewlett-Packard, Lexmark, Canon and Epson. in less detail. Of
the remaining manufacturers, most use the Hewlett-Packard print car-
tridge and as this is the critical part of the printer the results described be-
low will be equally applicable to these systems.
Each cartridge was viewed under a microscope to get a detailed look at
the components. A graticule was used to measure the dimensions of the car-
tridge and the print-head. The print-head plate can be easily removed using
a mounted needle to inspect the electrical components inside he cartridge.
Evidential value from ink-jet printers 363

The printers studied

The printers used for the experiments were the Hewlett-Packard
Deskjet 500, fitted with a 51626A cartridge and the Lexmark 1000 fitted
with a 13620HC cartridge, both being thermal drop-on demand ink-jet sys-
tems. They were operated using the appropriate printer driver installed in
Microsoft Windows NT. The print-head on the Hewlett Packard cartridge
consisted of 54 nozzles arranged in a rectangle with a maximum print height
of 4.1 mm (see Figure 3).The two columns of nozzles are out of phase i.e. the
left-hand side nozzles are positioned in-between those of the right. This
means that microseconds after the right hand side has fired; the left side will
fire ink to fill in the gaps to leave a completed line of text. (The actual length
of the nozzle array is 4.5 mm, but tests show that on this cartridge the corner
nozzles do not fire when printing text).

Nozzle 54

Nozzle 53

Out of phase

Print-head plate
4.5 mm 7.66 mm
27 nozzles
either side
0.8 mm

Nozzles 1 and 2

5.44 mm

Fig. 3. Hewlett Packard print-head 51626A.

The Lexmark cartridge had 60 nozzles and a print-head size of 4.7 mm

but is similar in other respects.
364 S. P. Day, L. Shufflebottom

To study how the different operating conditions affect the distance be-
tween defects printed on different scans of the print-head it is important to
know that the HP Deskjet 500 software prints individual lines of text i.e. it
will try to print one line of text in one scan. If this is not possible then it will
print the remaining part of the characters on its next scan and will print no
other line of text. It also prints moving in both directions, but a forward print
is not necessarily followed by a print scan in the opposite direction. The for-
ward/backward pattern is a function of the software driver.
The Lexmark printer only prints in one direction and prints using all
available nozzles, irrespective of the text it is printing. In applying the find-
ings of this paper it is important to realise that the direction of print and the
number of nozzles used is determined by the software but the physical size of
the print-head and number of nozzles is a property of the print-head.

DEFECT EXPERIMENTS

A number of experiments were designed for each of the cartridges, to an-

swer the following questions:
1. Do all blocked electrical contact pads create a defect?
2. Is a defect consistent throughout the text?
3. How is the same defect affected by different fonts?
4. How is the same defect affected by different sizes of font?
5. How is the same defect affected by different line spacing?
6. Does a cartridge, which produces a defect in one printer, produce the
same defect if it is changed to another printer of the same, make and
model?
It is known that defects occur in ink-jet printers, but there is a perception
that these defects are rare and too variable to be useful. Defects occur in car-
tridges over time due to mechanical and thermal stresses. For these experi-
ments, a known defect was introduced into a printer and then to study its be-
haviour under different conditions. Both the Hewlett Packard cartridge and
the Lexmark cartridge have a series of electrical pads or contacts on the
outer surface that engage with contacts in the machines themselves. By
blocking the electrical contact pads it was found that a single line defect was
produced on most occasions. To enable us to link the electrical pad with the
nozzle that it controls each pad was numbered and then blocked in turn us-
ing a small piece of insulating tape. The cartridge was then loaded into the
appropriate printer and asked to print out the mission of the Forensic Sci-
ence Service in 12 point font in Microsoft Wordâ, which is “The Mission of the
Forensic Science Service is maximise the benefit of forensic science to the
criminal justice system by being the leading provider.”
Evidential value from ink-jet printers 365

The printouts were then viewed under a microscope to check for the pres-
ence or absence of a defect. After each individual pad had been taped, neigh-
bouring pads were taped together to see if this affected the number of defects
they produce. The electrical circuitry inside the cartridge was also examined
and the circuit traced.
In this way we were able to determine which pad controlled which nozzle.
In the Hewlett Packard cartridge examined the relationship of pad to nozzle
was fairly straightforward. The exceptions are the contact pad numbers 23,
25, 26, 29, 30, and 56 (see Figure 4). Pad numbers 29 and 30 do not create de-
fects and appear to have no function in this particular print-head (the com-
ponents of this print-head are used in other related printers and no doubt
this accounts for their presence here). The other four pads, 23, 25, 26, and 56,
produce more line defects than one. When pad number 23 is blocked it pro-
duces a number of line defects that affect the bottom half of the print. If num-
ber 56 is blocked, again the bottom half is affected. When both 23 and 56 are
blocked, the whole of the bottom half of the print is missing. The same hap-
pens for pads 25 and 26 but they affect the top half of the print (see Figure 4).
From these results, it seems that they “control” the other heating resistors
on their side of the nozzle array. Thus if one of these pads becomes defective
it will create a very obvious defect in the printing produced by the machine.
Despite this, we have come across this particular defect in casework.

Electrical contact pads on the

print cartridge

Pad number 10
blocked One line defect

Pad number 23 Number of line

blocked defects

Bottom half of
Pad numbers 23 and character is
56 blocked missing

Fig. 4. Generation of line defects.

366 S. P. Day, L. Shufflebottom

RESULTS

Measurement of defects
Once a defect had been introduced into the page of ink-jet printed mate-
rial it was examined with a binocular microscope to identify any defects that
had been introduced into printing.
Attempts were made to measure the distance from the top of the letter
“H” as this letter uses most of the nozzles to print the character. A microme-
ter graticule was used to measure the distance from the top of the printed
character to the top of the defect. This distance was recorded and used to
compare to other samples. The defect can also be measured by counting the
droplets of ink above it. The appearance of the letter under the microscope
allows the individual dots that make up the letter to be counted. This is less
accurate than the graticule method, as two dots can easily become fused to-
gether if a nozzle is firing incorrectly i.e. not firing perpendicular to the paper.
The most practical way of determining whether defects were consistent
was found to be measurement of the distances between defects on different
lines. This allowed a pattern of defects to emerge although, as will be seen
later, this pattern does not necessarily repeat from line to line and may re-
quire several lines to be inspected before the pattern emerges (the distance
to be measured can be seen in the illustration below, Figure 5).

Fig. 5. Measurement of inter-line defect spacing.

Consistency tests
A single pad was taped (which pad does not matter, but we chose pad 10)
to see if the defect it produced occurred in the same position in each line of
text and also if it appears in the same position each time the document is
printed. For these experiments, the first page of this introduction in
Microsoft Wordâ was printed. A microscope was used to observe the posi-
Evidential value from ink-jet printers 367

tions of the defect down the page. This experiment was run throughout the
period of study to see if the defects are consistent over time. It was found
that, providing exactly the same conditions were used each time, then the
defect did not vary and occurred in exactly the same place in each line of text.

Different fonts
The above experiment was repeated but this time the font of the docu-
ment was changed to see if this affected the positioning and distance be-
tween defects. The point size for each font used was 12 point. The fonts used
in these experiments were Arial, Book Antiqua, Bookman Old Style, Cen-
tury Gothic, Century Schoolbook, Courier New, Footlight MT Light, and
Times New Roman.
The results of these experiments for a number of different cartridges are
shown in Table I.

TABLE I. DISTANCES BETWEEN THE SAME DEFECT OF DIFFERENT 12 POINT

FONTS FOR FOUR DIFFERENT CARTRIDGES

Distance between defects [mm]

Size of H
Typescript
(mm)
51626A 51629A 51645A 13620HC

Arial 3.1 5.1/5.0 5.1/5.0 14.4/14.3 4.7

Book Antiqua 3 5.1 5.1 14.4 4.7

Bookman Old Style 2.9 5.1/5.0 5.1/5.0 14.4/14.3 4.7

Century Gothic 3.1 5.4/5.3 5.4/5.3 14.7/14.6 4.7

Century Schoolbook 3 5.1 5.1 14.4 4.7

Courier New 2.5 4.5/4.4 4.5/4.4 13.6/13.5 4.7

Footlight MT Light 2.8 4.4 4.4 13.5 4.7

Times New Roman 2.9 5.1/5.0 5.1/5.0 14.4/14.3 4.7

Arial,
Book Antiqua,
Bookman Old Style,
Century Gothic,
Century Schoolbook,
Courier New,
Footlight MT Light,
Times New Roman
Fig. 6. Example of type fonts used.
368 S. P. Day, L. Shufflebottom

For the Hewlett Packard cartridge under study, 51626A, when the font of
the document was changed then the spacing between the same defect in dif-
ferent lines changed. This is because this cartridge prints a line at a time
therefore the defect distance is dependent on the size of the font and the size
of the line spacing between.
In addition to this, in some of the fonts the distance between two consecu-
tive lines of the same defect changed while in others the distance remained
the same. This is because different typescripts of the same point size differ in
their heights. If a typescript is created by an even number of nozzles then the
distance between defects will be constant throughout the document. How-
ever if the document is created from an odd number of nozzles then the dis-
tance will vary in alternate lines. Thus if the distance between a defect in
line 1 and 2 is 5.1 mm, the distance between a defect in line 2 and 3 may be
5 mm, but the defect between lines 3 and 4 will again be 5.1 mm. This pattern
will continue down the page. Table I shows the differences between the dif-
ferent fonts in length between two lines of text produced from the same de-
fect.
Because the Hewlett Packard printer prints one line at a time and prints
travelling in both directions the distance between the same defect on differ-
ent lines will be nearly the same each time. A different size of character and
therefore a different size of line spacing will create a different distance be-
tween the same defect. However the software controlling the print-head is
designed to use slightly different nozzles on the way forward than on the way
back thus giving rise to the differences observed.
The Lexmark printer (cartridge 13620C), on the other hand, prints using
the complete array of nozzles irrespective of the text being produced. There-
fore if spaces between lines coincide with a pass of the print-head then the
corresponding nozzles will not produce an image. Hence the distance be-
tween a defect in different lines will be constant and will be the size of the
print-head (in this case 4.7 mm) no matter which font is used.

Using different point size (character height)

Again the consistency experiment is repeated but this time the point size
was changed to see if this affected the positioning and distance between de-
fects. The sizes used in these experiments were 10 point, 12, 12.5, 13, 13.5,
14, 16, 18, and 20 point using Times New Roman typescript (see Figure 6).
Evidential value from ink-jet printers 369

10 point,
12 point,
12.5 point,
13 point,
13.5 point,
14 point,
16 point,
18 point,
20 point
Fig. 7. Examples of different point sizes.

The Hewlett Packard cartridge is only capable of producing print which

is 4.1 mm high, which corresponds to 12.5 point in Times New Roman.
Therefore when point sizes of less that 12.5 are used the pattern described
above in the study of font is again seen. In a point size requiring an odd num-
ber of nozzles a permanent defect has an alternating pattern while in one re-
quiring an even number the pattern is repeated every line. However, if the
printer is asked to print out text which is larger than the print-head size,
then the printer has to complete the character in more than one scan and so
the defect may or may not occur more than once. With this cartridge, for
point sizes greater than 12.5, the print-head will need to scan across the
page twice to complete the character. If the defect was in a nozzle close in po-
sition to nozzle 1, then the defect is more likely to be repeated in the same
character. If the defect is repeated, it represents the maximum size that the
print cartridge can print. (for this cartridge this distance was 4.1 mm). The
pattern of firing thus becomes complicated and the exact appearance of the
text on the paper will depend on the position of the nozzle which is not firing
and the point size used. In addition, the alternate pattern described above
may be superimposed onto the existing pattern. Figure 8 shows the pattern
of firing for one such situation. For 12 point size characters, the number of
dots in a straight line are the same but the nozzles that ejected those dots
change for alternate lines of text. The first line of text is printed by nozzles
4–47, the second line by nozzles 5–48, the third line is printed like the first
and the fourth like the second and so on. This is the reason why the distance
between defects changes for every alternate line.
For 12.5 point font size characters, all the lines are printed by the same
nozzles. The nozzles that fire are 5–49, thus there is an extra dot of ink for
the change of 12 to 12.5 point size. For this point size all the distances be-
tween defects will be the same.
370 S. P. Day, L. Shufflebottom

For 13 point font size characters, nozzles 4–49 plus nozzles 4 and 5 from
the second scan print the first line of text. Nozzles 5–50 plus nozzles 4 and 5
from the second scan print the second line of text. If here nozzle 5 were
blocked then the line defect would be repeated twice in both lines of text.
For 14 point font size characters, nozzles 4–49 plus nozzles 4–8 from the
second scan print the first line of text. Nozzles 5–50 plus nozzles 5–9 from
the second scan print the second line of text.
These different firing patterns produced by the nozzles creates the differ-
ing distance between defects but by careful study a regular pattern should
emerge if the defect is consistent.
It is interesting to note that if nozzle 4 is showing a defect then some lines
of typing in 12 point will not contain the visible defect while in 13 point the
defect will occur twice in the same line.

l Using Times New Roman

l 12 point uses nozzles 4-47 and 5-48 alternately
l 12.5 point uses 5-49 all the time
l 13 point uses 4-49 and 5-50 alternately and
nozzles 4 & 5 of the return scan

Fig. 8. Firing pattern for Times New Roman in different point size.

Using different line spacing

Again the consistency experiment is repeated but this time the line spac-
ing of the document is changed to see if this affects the positioning and dis-
tance between defects. The spacing used were single, 1.5 lines, and double
again using the Times New Roman typescript in 12 point. The effect of this
in the Hewlett Packard machine was simply to make the distance between
the defects in each line greater while in the Lexmark machine the distance
between the defects remained at 4.7 mm, the size of the print-head. How-
ever, in some circumstances the defect in this type of printer may appear to
vanish for parts of the text down the page because it conicides with a line
spacing. By measuring differences in the spacing of defects down the page
the regular pattern should be apparent and the size of the print-head can be
determined.

Changing cartridges between machines

A print cartridge known to have a defect was used to print a document in
a number of different machines. The experiments were conducted to see if
Evidential value from ink-jet printers 371

the defect remained in the same place from one machine to the next. The ma-
chines used were the Deskjet 500, 510 and 510C which all use the same car-
tridge. The resulting printouts were viewed under a binocular microscope
with a graticule so the defect positions could be compared. It was found that
the defect behaved in the same way whichever machine was used.

Print direction
Some machines can print in both directions while some can only print in
one direction. Further the pattern of printing is not always left to right then
right to left. Sometimes the print-head will travel several times in one direc-
tion then several times in the other, depending on what it has to print. From
viewing the printouts underneath a microscope, the direction of print can be
found by looking at the ink spatter. This predominately occurs on the side of
the printed character towards the end that the print-head is travelling. This
can be seen in Figure 9. Thus if the bulk of the spatter is on the right of the
character then the print-head was travelling from left to right (it occurs on
the leading edge). The reason for this can be easily understood by referring
back to Figures 1 & 2. The comet-like tail of the ejected drop is ejected
slightly after the main drop and therefore reaches the paper slightly further
along the paper than the main drop, resulting in the spatter. The amount of
spatter will vary from machine to machine.
By looking at these spatter patterns one can see when the direction of the
print-head changes direction of print and consequently this provides
a method of determining the size of the print-head nozzle array. This is a fea-
ture which cannot vary and is therefore very useful in distinguishing be-
tween printers using different cartridges or different manufacture.
Occasionally ink droplets from the tail can travel through the main drop-
let and reach the paper in advance of the main dot, creating spatter on both
sides of the character. However, this is uncommon.

Ink spatter

Direction of print-head

Fig. 9. Diagram showing ink spatter.

372 S. P. Day, L. Shufflebottom

Sample printouts
Printout samples were collected from as many machines as possible to try
to identify how common defects occur in printouts and which of the nozzles
were prone to become defective.
Of the 53 printouts, 14 had defects (approximately 1 in 4) of some descrip-
tion. The manufacturer’s reliability goal is for the cartridge to run out of ink
before a nozzle becomes blocked. Of those with defects 71.4% of the defects
were around the centre of the print-head, but this may not be significant due
to the small sample size. Some other defects were detected apart from line
defects in two printouts. These were e print quality defects which resulted in
unevenness of print. This is probably caused by slightly varying drop vol-
umes due to imperfections in the manufacturing process of the nozzles.
One inkjet cartridge containing a defect was changed to another printer
model that also uses the same type of cartridge. The resulting printouts had
the same defect but the defect was in a slightly different place for each ma-
chine. This could be due to the printers being different models. However, the
distance between the defects remained the same from the two printers.

DISCUSSION

While the appearance of a document printed using an ink-jet printer is

largely determined from the way the printer interacts with the controlling
software the above experiments show that there are some features which
are properties of the printer itself. Using these features the document
examiner can develop a protocol for the comparison of print-outs and thus
determine whether two documents share a common origin. The following
steps are one such protocol:
– direction of print-head motion: by looking carefully at the spatter pat-
tern on single colour printed text one can determine how the print-head is
moving. If it is printing one line of text at a time before beginning another
then this can be distinguished from a printer that has software which prints
without regard to lines of text.
– maximum size of print-head: additionally, by examining the transitions
from one print pass to another the size of the print-head can be measured. In
line-at-a-time printers this will be particularly evident in areas of the print
which contain diagrams or print with a point size larger than the print-head,
usually around 13 point. In other printers the print-head can easily be deter-
mined when it changes direction in the middle of the line of text.
– repeating pattern of line defects: defects in a print-head can arise
through a number of reasons. Whatever the cause, most result in a fine
Evidential value from ink-jet printers 373

white line running through the print. Consistent defects occur in about 25%
of inkjet printers. The problem for the document examiner is distinguishing
between defects that are consistent from those which are not. This is done by
looking for regular patterns that repeat in the lines of text. In some printers,
that do not follow the lines of text, this is relatively easy and is simply a mat-
ter of identifying the positions of the defect and measuring the repeating dis-
tance between them. In others the position of the white line may vary de-
pending on the point size, font, line spacing etc. Here the document exam-
iner must look for a pattern that repeats within a particular section of the
text. This pattern can be quite complicated, especially if the point size used
is generally larger than the size of the print-head.
– position of line defect with respect to the top of “H”: by measuring the
height of the block capital letter “H” and determining the position of the line
defect with respect to the top of the “H” in a number of lines it may be possi-
ble to determine which nozzle is actually defective. If the definition of the
dots is clear enough and the print quality is not too high then it may be possi-
ble to count the number of nozzles in operation and the exact origin of any de-
fect.
Limited studies of the type described above have been carried out using
a number of other printers from different manufacturers and with colour
printers and the general examination principles seem to apply to them also.
The machines studied were: Hewlett Packard 51629A, 51645A, 51641A,
51625A; Lexmark 13620HC, 13619HC, 13400HC, 12A1970.
Some Canon and Epson printers were also studied but no detailed mea-
surements were taken from these machines.

CONCLUSION

From the observations made it can be seen that there are a number of
ways in which the products of an inkjet printer and the printer that pro-
duced them can be associated with a greater degree of certainty than has
hitherto been reported. However, these methods are likely to be much more
reliable in showing that different printers are involved and, whatever the
situation, the techniques still need to be used with caution.

Acknowledgements:
Thanks also to Dr. Barbara Daniel Kings College, London, for the use of her printer
and all the people who helped to provide printouts of their printers.
Thanks to Epson UK Limited, Hewlett-Packard UK Limited, Lexmark International
and Canon UK Limited for their literature and help.
374 S. P. Day, L. Shufflebottom

References:
1. A d e n J. J., The third generation HP thermal ink-jet print-head, Hewlett-
-Packard Journal 1994, vol. 45, no.1, pp. 41–45.
2. A l l e n R. R., M e y e r J. D., K n i g h t W. R., Thermodynamics and hydrodynam-
ics of thermal ink-jets, Hewlett-Packard Journal 1985, vol. 36, no. 5, pp. 21–27.
3. A s k e l a n d R. A., C h i l d e r s W. D., S p e r r y W. R., The second generation
thermal ink-jet structure, Hewlett-Packard Journal 1988, vol. 39, no. 4,
pp. 28–31.
4. B u s k i r k W. A. H a c k e l m a n D. E., H a l l S. T. [et. al.], Development of a high
resolution thermal ink-jet print-head, Hewlett-Packard Journal 1988, vol. 39,
no. 5, pp. 55–61.
5. C h e n J., W i s e K. D, A high resolution silicon monoltithic nozzle array for
ink-jet printing, IEEE Transactions of Electron Devices 1999, vol. 44,
pp. 1401–1409.
6. C h e n P. H., Droplet formation of a thermal sideshooter ink-jet print-head, In-
ternational Journal of Heat and Fluid Flow 1998, vol. 19, pp. 382–390.
7. H o l l i s J. L., S c h u l t z P. C., W a l s h W. J., Colour thermal ink-jet printer elec-
tronics, Hewlett-Packard Journal 1988, vol. 39, no. 4, pp. 51–56.
8. K a m p h o e f n e r F. J., Ink-jet printing, IEEE Transactions of Electron Devices
1972, vol. 19, pp. 584–593.
9. L e e C. S., L e e J. D., H a n C. H., New fabrication process of single-crystalline
silicon islands using double diffusion: Application to a heating resistor of a ther-
mal ink-jet print-head, Japanese Journal of Applied Physics 1998, part 1, vol. 37,
no. 12B, pp. 7100–7103.
10. L o r e n z e R. V., K u h m a n D. E., Correlation of misdirected satellite drops and
resultant print quality defects with nozzle face geometries in thermal ink-jet
print-heads, Journal of Imaging Science and Technology 1995, vol. 39,
pp. 489–494.
11. M a z e C. [et al.], Ink and print cartridge development for the HP deskjet 500C /
Deskwriter C print family, Hewlett-Packard Journal 1992, vol. 43, no. 4,
pp. 69–76.