0% found this document useful (0 votes)
27 views9 pages

AFM Image Artifacts: Applied Surface Science

This paper discusses various artifacts encountered in atomic force microscopy (AFM) that can affect the accuracy of surface imaging. It highlights the importance of recognizing these artifacts, such as edge overshoot, thermal drift, and friction-related distortions, to ensure reliable data interpretation. The authors provide examples and suggest methods to minimize these artifacts for better imaging results in nanotechnology applications.

Uploaded by

ahamed.raihan19
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
27 views9 pages

AFM Image Artifacts: Applied Surface Science

This paper discusses various artifacts encountered in atomic force microscopy (AFM) that can affect the accuracy of surface imaging. It highlights the importance of recognizing these artifacts, such as edge overshoot, thermal drift, and friction-related distortions, to ensure reliable data interpretation. The authors provide examples and suggest methods to minimize these artifacts for better imaging results in nanotechnology applications.

Uploaded by

ahamed.raihan19
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Applied Surface Science 304 (2014) 11–19

Contents lists available at ScienceDirect

Applied Surface Science


journal homepage: [Link]/locate/apsusc

AFM image artifacts


F. Gołek ∗ , P. Mazur, Z. Ryszka, S. Zuber
Institute of Experimental Physics, University of Wrocław, Pl. Maxa Borna 9, 50-204 Wrocław, Poland

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

Article history: Atomic force microscopy (AFM) has become an important tool in surface science and nanotechnology.
Received 14 October 2013 It is obvious that the intrinsic limitations of AFM must be understood in order to get useful information
Received in revised form 24 January 2014 about surface structure of the material under study. The ability to recognize artifacts should assist in
Accepted 24 January 2014
reliable evaluation of instrument operation and in reporting of data. In this paper, we discuss the most
Available online 31 January 2014
frequently encountered image artifacts in atomic force microscopy. A variety of artifacts are illustrated
by the results obtained with the aid of contact AFM (C-AFM), which can help avoid misinterpretations. It
Keywords:
is shown that, despite inaccuracies in AFM image generation, in many cases valuable information can be
AFM
Artifacts
obtained.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction Images produced by AFM are only approximations of the specimen


surface [11–36] because the probing tips are not perfectly sharp.
Atomic force microscopy (AFM) is getting more and more pop- In this paper the most common AFM image artifacts will be
ular in many different scientific and engineering fields and is the described, with a focus on tip geometry related artifacts, exem-
most widely used tool for measuring, imaging and manipulating plifying some of them by the results from our laboratory.
matter at the nanometer scale [1–6]. Most frequently the AFM is
used to image the topography of a surface, but by modifying the
2. Experimental
probing tip it is possible to measure friction, magnetic structures,
surface charge distribution and so on [1,6–12]. The AFM is espe-
The experiments were performed in UHV STM/AFM Omicron
cially useful for studying insulating materials. It can be used to
system (base pressure <10−8 Pa). Silicon and alkali halide single
probe biological material, even in its physiological environment
crystals were chosen as illustrative samples. Some of them were
[13,14].
covered by thermal evaporation of LiF, NaCl or KCl using Knudsen
Despite the advantages, serious problems arise due to the finite
cells. The deposition rates used were between 0.5 and 1 nm/min,
size of the AFM tip, when an AFM is used for imaging surfaces
as monitored by a quartz microbalance. The substrate tempera-
with features comparable or smaller than the mean radius of cur-
ture was measured using a type K thermocouple spot-welded to
vature of the probing tip. Furthermore, carefully collected data
the Omicron sample holder. The contact AFM (C-AFM) was work-
suggest that the measured contrast can be due to a variation in
ing in a constant force mode under soft contact with contact forces
the chemical reactivity of the top surface atoms or to a tip-induced
typically adjusted in the region of 0–6 nN. AFM images were gen-
atomic-relaxation effect reflecting the stiffness of the surface layer
erated with silicon and silicon nitride pyramidal tips. The analysis
[15].
of AFM results was performed with the WSxM software [37]. The
A serious disadvantage of the AFM technique is that many arti-
AFM images are presented as obtained, without filtering.
facts can appear in AFM images which are difficult to recognize
even by experienced users [16–29]. Moreover, the AFM probing tip
may influence and even change the state of the investigated sam- 3. Image artifacts
ple [30–34]. Artifacts can occur for a variety of reasons, including
finite tip dimensions, creep and hysteresis of piezo-scanner, ther- An image artifact is defined as any feature which appears in the
mal gradients, vibrations and the electronics (feedback circuits). image which is not present in the original probed object [27–29].
As a matter of fact, any component of the AFM (sensing probes,
xyz-piezoelectric ceramics, feedback electronic circuit, computer
system for generating and presenting images, etc.) can be a source
of artifacts. Furthermore, the sample preparation techniques and
∗ Corresponding author. Tel.: +48713759308. external factors (electromagnetic and mechanical noise and so on)
E-mail address: golek@[Link] (F. Gołek). can be a source of artifacts as well. It is impossible to address and

0169-4332/$ – see front matter © 2014 Elsevier B.V. All rights reserved.
[Link]
12 F. Gołek et al. / Applied Surface Science 304 (2014) 11–19

illustrate artifacts generated by all sources; therefore, we limit our it has strangely shaped structures. The example of tip contamina-
presentation to the most frequently encountered ones. tion induced noise is shown in Fig. 4. As long as this type of noise
is not too intense, it is possible to find out the surface structure of
3.1. Edge overshoot the probed sample. To avoid such a noise, the tip should be cleaned
before (or replaced by the new and clean one) and maintained clean
This artifact occurs when the AFM tip crosses the step edge. It during the scan.
is generated because of piezoelectric scanner hysteresis and some It has to be stressed that internal noise is the main obstacle
inertia of the cantilever holder. Edge overshoot artifacts can be seen limiting the sensitivity and resolution of the AFM/STM apparatus.
as artificial, narrow hills and valleys at the edges of surface terraces. When noise appears as a large and periodic modulation of the AFM
In Fig. 1 such an artifact is schematically drawn and illustrated by topography image, its origin is rather easy to be recognized and
the image of the surface of LiF crystal previously heated up to about eliminated.
800 K. Electrical and mechanical noise from the street traffic can be
As one can see, this artifact does not destroy information about minimized when the measurements are performed at night or
the surface step arrangement and heights. Moreover, it even makes weekend (when the building vibrations and the electric power fluc-
the step lines better pronounced. The effect of step pinning by sur- tuations are weaker). Some types of noise can be corrected and
face contamination clusters is clearly seen. It may be concluded that almost eliminated by proper filtering [44–46]. For example, high
edge overshoots do not alter the main features of the probed sam- frequency noise can be removed by Matrix Filtering which averages
ple. To minimize the overshoot amplitudes one should consider the several adjacent points together in the image. Common practice is
scan speed reduction. to use 2D fast Fourier transformations (2D FFT) to remove noise
from the images and then analyze the frequencies due to the peri-
3.2. Edge elevation odic features of the sample [47].
It is also known that for better resolution the laser beam and
This artifact looks similar to overshoot, but does not decline the whole instrument should be allowed to warm up for about 1 h
with slowing down the scan. Such an artifact may occur in situa- with the measuring head plugged in. Very powerful and promis-
tions when attractive forces contribute significantly to the effective ing in noise reduction are the recently developed Cryo-AFM/STM
sample–tip force. The example of artificial edge elevation is shown measuring systems [48–54].
in Fig. 2. Fig. 2c illustrates why the attraction part is smaller at the
step edges and how it results in the artificial edge elevation.
3.5. Light interference artifact
3.3. Thermal drift and piezoelectric creep
AFM based on optical beam deflection can generate artifacts
due to interference on the quadrant photodiode between the light
Thermal drift and piezoelectric scanner creep may introduce
reflected from the cantilever and stray light from the sample sur-
serious deformations into the topography images of the investi-
face. Example of this type of artifact is presented in Fig. 5.
gated samples when not effectively reduced. The cantilever has a
This artifact is easy to be filtered out, as shown in Fig. 5c.
very small mass, and thus, very small heat capacity, therefore very
Oscillations in images typically have a period comparable to the
small amount of heat due to shining light or touching to the sample
wavelength of the laser light. If this is the case, the dominant
may change its temperature and the relative position. Thermal drift
interference is between incident and sample-reflected laser beams
can be minimized by allowing thermal equilibrium to be reached
[55]. In other cases, interference between cantilever- and sample-
or by using faster scan rates. Unfortunately, ringing or noise may
reflected laser beams is to be considered at the photo-detector. It
occur when scanning too fast. Creep, on the other hand, can be
is worth to notice that this type of artifact is significantly reduced
minimized by the usage of position sensor and a closed-loop con-
in systems with low-coherence lasers. Another proposed strategy
trol unit [38–41]. Thermal drift and creep, when not eliminated, can
is to modulate the laser current at high frequency [56].
greatly affect the apparent dimensions in the slow scan direction,
and thus, produce images with substantial distortions, especially at
the beginning of the scan. Such distortions can be easily recognized 3.6. Friction-related artifacts
for the samples containing a straight terrace edges or well defined
objects with known dimensions. An example of such a distortion In C-AFM, the friction forces contribute significantly to the total
is shown in Fig. 3. The bottom of the image is the beginning of the tip–sample interaction force. These forces tend to induce some tor-
scan. As one can see, the terrace edges look straight in the upper sion of the cantilever which can be interpreted by the deflection
part while in the lower part, closer to the beginning of the scan, detector as a height variation.
bowing of the edge lines is visible. As an extreme case, we regis- Therefore, when the topography images of the sample are cre-
tered images for which drift was initially faster than the scan in the ated, the attention should be paid to the generation of the friction
slow-scan direction. related artifacts [57]. Such an artifact may occur for the samples
that show domains with different friction properties. It was shown
3.4. External and internal noise that the friction force depends not only on the load, but also on the
type of the sample and the probe used (for example, Si, Si3 N4 or
To the external (environmental) noise mostly contribute such some other), the scan velocity, the presence of contaminations or
factors as acoustic noise, mechanical building vibrations, air cur- fluids, and even on the scan direction (also relative to the cantilever
rents, electromagnetic noise and temperature waves. The number axes) [57–68]. Consequently, considerable efforts are being made
of internal noise sources is even larger. In fact, each component to avoid friction and adhesion induced height artifacts in topog-
of the system (force sensing tip, cantilever, piezoelectric scanning raphy images, as well as to avoid topography related artifacts in
stage, laser, optical system, electronic preamplifiers and amplifiers, the friction map measurements [69–72]. It is interesting to notice
etc.) can contribute to the total noise, influencing the topographic that what is considered a friction related artifact in the topogra-
signal and limiting both the resolution and sensitivity of the micro- phy measurements, usually provides important data in the friction
scope [42]. In some cases, the extra noise can be induced by tip and adhesion related experiments, and inversely, an artifact in
contamination [43]. Dirty tips can also make the image appear as if micro and nano-tribology experiments (for example, step-related
F. Gołek et al. / Applied Surface Science 304 (2014) 11–19 13

Fig. 1. (a) Scheme showing the edge overshoots. (b) Topography image of the LiF crystal surface. The crystal was cleaned by partial sublimation before the examination. (c)
Height profile along the line drawn in (b).

Fig. 2. Image of the KBr crystal surface with artificial edge elevations. (a) Topography image. (b) Height profile along the line drawn in the topography image. (c) Drawing
explaining the origin of the elevation effect.

Fig. 3. AFM topography image of the LiF crystal surface registered with thermal drift present (left) and after its reduction (right).

features), is considered as fundamental data in the topography is an indication of strong influence of friction on AFM topography
measurements. images. As it is well known, the friction related artifacts can be min-
In Fig. 6, an example of the friction related artifact in the topog- imized by reducing the tip–sample interaction force or, better, by
raphy image is shown. As one can see, there are four domains that using the tapping or non-contact AFM operation mode.
are shown as valleys when scanned in the forward direction, and As one can see, it is really useful to compare images obtained
as hills when scanned in the backward direction. Such an effect for both scan directions in order to avoid the cross-talk of the

Fig. 4. Left panel: two AFM images exemplifying a tip contamination induced noise. Right panel: AFM image with tip contamination induced noise significantly reduced.
Images were taken for the LiF crystal surfaces under UHV conditions.
14 F. Gołek et al. / Applied Surface Science 304 (2014) 11–19

Fig. 5. Light interference artifact: (a) topography image with interference oscillations. (b) Height profile along the line drawn in (a) showing the artificial oscillations. (c)
Topography image without the interference effect that was removed by flattening function of the WSxM program [37].

normal force into the lateral signal and vice-versa. If the cross-talk The scale of deformation is a dramatic problem for soft materials,
of the lateral forces is negligible, the topography images should be such as, for example, polymers or biological objects. Owing to the
identical for the forward and backward scan directions. high elasticity of biological cells and membranes, the tip can deeply
In the context of friction, it is worth mentioning that many indent the probed species, so the contact area is rather large and
authors investigating friction phenomena with the aid of friction increases with the increasing applied force. Therefore, the nanome-
force microscopy (FFM, which is AFM working in the lateral force ter resolution cannot be easily achieved in this case. First of all, the
mode) have paid attention to the enhancement of friction at step height profiles are substantially altered and the measured heights
edges [73–78] and have attributed it to the Ehrlich–Schwoebel bar- are significantly lower than the expected (and true) ones. Some bio-
rier (E–S barrier) [79,80]. Conventional E–S barrier is defined for a logical and sensitive objects can be even destroyed in the course of
single atom or molecule as depicted in Fig. 7a. scanning by the cantilever tip.
To illustrate the complexity and possible source of artifacts In the case of soft materials, the limitation of AFM may come
in nano-scale friction experiments, we present schematically in from the sample alteration due to the effect of mechanical pushing
Fig. 7b a typical scenario of the tip-induced relaxation. From this of the top surface material. It was shown in the literature that not
simplified illustration, it is easily seen that images of lateral forces only atoms [97] and large molecules [98–100], but also biomaterials
may differ from experiment to experiment. Recently published [101–105] can be pushed and replaced using AFM.
results [73–78,81–84] clearly show that the lateral force image at Another source of sample alteration is the material transported
a step edge strongly depends on the shape of the probing tip and from the AFM tip into the sample surface. Such events are very
experimental conditions. Two contributions of the opposite sign, likely to happen for contaminated probing tips. Sample surface
namely, geometric (that may reduce the lateral force) and frictional alteration or scan damage, which may occur especially in the con-
(that increases the lateral force) are responsible, at least partially, tact mode, is easily indicated by the changes of topography images
for the poor reproducibility of lateral force images at step edges. when the scan is repeated in the same place of the sample. This
From recently published results [78], we know that atomically type of disadvantage is usually examined by repeated scanning of,
sharp tips easily slip into the bottom of the edge, thus experienc- say, 3 ␮m × 3 ␮m area and then scanning, say, 5 ␮m × 5 ␮m area
ing a friction force which is temporarily oriented in the direction of that centrally contains the initially scanned 3 ␮m × 3 ␮m area. If
motion. This behavior may even contradict the presence of E-S bar- damage occurs, the final image will contain clearly visible changes
rier. Therefore, in our opinion, it is not appropriate to interpret the of topography in the central area. An example of such a struc-
FFT images at step edges directly in terms of E–S barriers. However, ture is presented in Fig. 8a and b. In Fig. 8a, a topography image
in many cases such images contain valuable information about the of 3 ␮m × 3 ␮m KBr crystal surface area is shown that centrally
probing tip parameters and the interactions involved [85–89]. contains 300 nm × 300 nm area scanned two times before. The
alteration of the sample can be deduced from the altered topog-
raphy of the central part of the image. It is obvious that pushing
3.7. Deformation, damage and sample alteration the residual contaminations already present on the sample as well
as the contaminations dropped by the probing tip may contribute
In general, in the AFM both the probing tip and sample are to this result. In the case presented in Fig. 8b, the sample was the
deformed when they are in contact [31,90–96] (see also Fig. 7b). KBr crystal with potassium enriched surface. The potassium enrich-
Therefore, the tip–surface separation is not exactly known, since ment was created by an electron stimulated desorption process (in
when the tip contacts the sample, there will be a small, but which the halogen is known to be more effectively desorbed). As
unknown, elastic deformation of the tip and/or the sample. it is visible, the topography image of this sample is also altered by

Fig. 6. The topography images with friction related artifacts. The scanned sample is a contaminated (1 0 0) surface of NaCl crystal. The left-hand-side part shows the AFM
image and the height profile obtained for the forward (right) scan direction, while the right side part shows the AFM image and height profile obtained for the backward
(left) scan direction. Valleys in the forward scan image occur as hills in the backward scan image.
F. Gołek et al. / Applied Surface Science 304 (2014) 11–19 15

Fig. 7. (a) Illustration of energy barrier for diffusion Ediff and Ehrlich–Schwoebel barrier EE–S . (b) Simplified illustration of the relaxation of atomic structures at the interface
between the sample and the probing tip during FFM scanning in the region of step edge. It is easy to notice that the change of effective area of interaction results in significant
changes of distribution of both pressure and local relaxation.

the tip movement. In this case, pushing the surplus potassium by the most typical and AFM images in this case are extremely difficult
the moving tip, is, most likely, the dominating effect. for correct interpretation. Broken, multiple, blunt or contaminated
In Fig. 8c–e, besides the blunt tip self-images (tip scanned by tips can produce strange results. It is well known that the ability to
very sharp sample protrusions), the effect of breaking of such pro- image deep trenches, steps, or small particles in AFM is limited by
trusions is captured two times. The effect of breaking is deduced the sidewall angles of the tip. It is obvious that the tip is unable to
from the vanishing of the features (tip self-images) indicated by image features that have greater sidewall angles than those of the
arrows. tip. Extreme case resulting in self-imaging of the tip is depicted in
Fig. 9c.
3.8. Tip–sample convolution The AFM images occur as a convolution of the sample and the
tip geometry plus some switching effects (like switching lines that
Since the tip radius is usually in the range of 10 nm, convolution are defined and explained below, see Figs 11–12). Such a convo-
effects have to be considered whenever the finest details of the lution is, in our opinion, one of the main sources of AFM image
AFM images are interpreted. Among variety of situations, the three artifacts. This type of artifacts can, to some extent, be corrected by
depicted schematically in Fig. 9 are the most typical ones. In the case computer manipulation of the topographic data [106–108]. Unfor-
of a flat sample surface with well separated steps of only single or tunately, when the aspect ratio of the sample features is larger than
double atomic height, sketched in Fig. 9a, the probing tip sharpness that of the probing tip, the complete correction is impossible. This
is not crucial for the visualization of the atomic step lines. In such is because there will be sample areas not accessible for the probing
situations only one (the highest) protrusion of the AFM tip will be tip, as illustrated in Fig. 11a. The scanning tip has no chance to touch
allowed for the contact with the sample and will properly measure every place of the lower (internal) terraces. In another words, if the
the surface topography. Two examples illustrating these situations inclination angles of the surface feature are greater than those on
are shown in Fig. 10. In this figure both images show the terrace the tip, the image will show more of the tip sidewalls than of the
edges in a satisfactory manner. actual sample surface.
In the case sketched in Fig. 9b, the dimensions of the sample In the case of rectangular shaped fences (indicated in Fig. 11a
features are comparable to those of the probing tip. This situation is by a dotted line), the AFM image will show them as broadened

Fig. 8. AFM topography images showing damaging activity of the contact mode AFM. Image (a) shows clean KBr surface as a topography image of 3 ␮m × 3 ␮m area that
centrally contains the 300 nm × 300 nm area scanned two times before. Image (b) shows the topography of the KBr crystal surface previously decomposed by an electron
beam. Such a surface is potassium enriched. The area of 5 ␮m × 5 ␮m centrally contains the 3 ␮m × 3 ␮m area scanned fourth times before. In (c)–(e), three consecutively
recorded images of the same place of dirty LiF crystal are shown. These images show the effect of breaking of narrow protrusions one by one. Image of the tip apex (self-image)
indicated by an arrow vanishes in the subsequent image.
16 F. Gołek et al. / Applied Surface Science 304 (2014) 11–19

Fig. 9. Three general types of the tip-sample convolution. (a) Sharp tip is not required. (b) Sharp tip is required and the knowledge about the tip shape is needed for the
topography interpretation. (c) Self-imaging of the tip by the sharp islands.

Fig. 10. AFM topography images of (a) KBr(0 0 1) crystal surface and (b) Si(0 0 1) crystal surface.

and trapezoidal ones, as indicated by dashed line in Fig. 11a, and for one pair of tip and sample contact points, but is continued for a
illustrated by the topography image and the profile line in Fig. 11b quite different pair of tip and sample contact points.
and c, respectively. Furthermore, the narrow trenches appear to be Observation of the above-defined switching lines is a good indi-
narrower than expected and true ones. cation for the investigator to suspect image artifacts and to replace
For the recognition of the image artifacts it is important, in our the probing tip by the new and sharper one.
opinion, to pay attention to the so-called switching lines defined Another convolution-related artifact, which is frequently
as a set of S1 type points (see Fig. 11a) or a set of S2 type points. encountered, but much easier to recognize, is the multiple-tip
At S1 , the scanning of the sample is switched into the scanning of effect. The example AFM images generated by multiple tips are
the tip (tip self-imaging), while at S2 , the tip self-imaging switches shown in Fig. 13. From this figure, it is clear that the effect of fea-
back into the scanning of the sample. The existence of such switch- ture repetition (multiple-tip effect) is better manifested in the AFM
ing lines is illustrated in Fig. 12, where the AFM image of LiF thin error signal images than in the topography ones. The AFM image
layer composed of islands with various sizes is shown. The pres- of rather high (up to about 30 nm) rectangular shaped NaCl islands
ence of these lines gives the impression that some objects start to probed by a triple tip is shown in Fig. 14.
be imaged before imaging of some others is completed. Another As it is seen in Fig. 14, some valuable information can be obtained
impression is that some objects are partially covered by others. from the data despite the triple tipsample convolution. The near-
Such a switching effect occurs every time when the tip-surface con- zero glitch in the height diagram (Fig. 14b) informs us that the flat
tact point jumps from one place on the sample and the tip to another substrate is approached by the probing tip many times around the
(often relatively far away). At that moment, an island loses its con- islands of the interest. Thanks to such a situation the height of the
tact with the probing tip in favor of another one. Obviously, in such island can be measured. The top flat part of the convoluted image
moments the contact and the convolution (“imaging”) is disrupted can be considered as the top surface of the island (see Fig. 14a and c).

Fig. 11. Sample feature broadening due to tip-sample convolution. (a) Drawing explaining the image artifact generation. S letters indicate the characteristic switching points.
Sample surface parts indicated by dotted lines, e.g., between S1 and S2 , are skipped by the tip and do not contribute to the topography data. During the scan of this region,
the tip is probed instead. (b) AFM image of the rectangular shaped LiF fences grown on the NaCl(0 0 1) surface. (c) Height profile along the line drawn in the AFM image. The
shaded parts show the sample contribution, while the clear lines show the tip contribution to the profile. Note that there are inclination angles much smaller than 45◦ that
could, in principle, be interpreted as the proper imaging of the side walls of the fences. However, the fences height is only about 10 nm and the probing tip usually has also
its inclination angles much smaller than 45◦ close to the apex.
F. Gołek et al. / Applied Surface Science 304 (2014) 11–19 17

Fig. 12. AFM image of LiF thin layer formed on Si(0 0 1) showing a number of so-called switching lines (lines of switching points defined in Fig. 11), a few of which are
indicated by arrows. (a) Topography image generated by the tip-sample convolution. (b) AFM error signal image. (c) Drawing explaining the origin of the switching lines and
the artifacts. It shows how far the measured profile is from the true one.

Fig. 13. Images generated by multiple blunt AFM tips. (a) LiF thin layer, deposited at room temperature on Si(0 0 1), imaged by a double tip (topography image). Figures (b–d)
show convolutions of multiple tips and sample protrusions formed in thin-layer KCl/Si(0 0 1), NaCl/Si(0 0 1) and LiF/Si(0 0 1) systems, respectively. The (a) and (c) figures are
topography images, while (b) and (d) are AFM error signal images.

Fig. 14. The NaCl islands formed on Si(0 0 1) surface. (a) Topography image generated by triple tip. (b) Diagram showing the topography height distribution. (c) Height profile
along the line drawn in the topography image. Dashed lines show the expected profile of the island (nano-crystallite).

In general, we can conclude that the lateral dimensions of small Last but not least, it is worth to answer the question often asked
particles deposited on flat substrates appear to be larger than by the beginners in the field, namely, is it possible to exclude mul-
expected while the height of the particles appears to be correct. tiple or blunt tip artifacts in the case of single protrusion on the
This is because the width of the AFM probe is added to the width of sample relying only on the images generated for different scan
the measured structure, whereas the height is not changed (ignor- directions? The answer is no. Drawings shown in Fig. 16 clear up
ing negligible effect due to the fact that tip-substrate interaction the situation. It should also be mentioned that rotation of the sam-
may differ from the tip-island interaction). ple or the probe allows one to distinguish between the blunt tip
When the substrate is not flat, both the lateral and vertical artifact and the proper image generated by a sharp tip.
dimensions become altered. Such a situation is depicted schemat-
ically in Fig. 15. The particle in the valley appears as much lower
3.9. Remarks on deconvolution
and flatter than the identical particle on the hill (h1 > h2).

In experimental practice, characterization of AFM tips is often


neglected even when quantitative measurements, such as rough-
ness or friction and adhesion experiments, are performed. In such
experiments the shape of the probing tip plays an important role.
Moreover, the mesoscopic tip determines the attractive part of the
total tip-sample interaction in atomic resolution images [106]. As
we know, the finite dimensions of the tip apex are responsible for
producing topography images of the samples that appear wider
than they really are, and this is the most frequently encountered
Fig. 15. Dependence of the particle image on the tip radius and local topography of
the substrate. The particle in the valley appears as much lower and flatter than the topography artifact. To overcome this problem, at least partially,
identical particle on the hill (h1 > h2). one can employ deconvolution methods already described in the
18 F. Gołek et al. / Applied Surface Science 304 (2014) 11–19

Fig. 16. Drawing illustrating that change of the scan direction does not allow one to distinguish between the tip image and the real sample topography.

Fig. 17. Self-imaging of the AFM tip by the NaCl/Si(0 0 1) system, previously annealed and oxidized. (a) AFM topography image. (b) AFM error signal image. (c) Height profile
along the line drawn in the topography image.

literature [107–115]. Unfortunately, for all deconvolution meth- artifacts is extremely difficult, if possible at all. We propose to
ods the probe geometry must be analytically known. In practice, look for the switching lines, defined in Section 3.8, especially in
it is difficult to characterize the probing tip accurately and non- cases when the registered image shows features higher than sev-
destructively. One solution relies on the use of test samples or eral nanometers. In our opinion such an inspection is very useful
calibration arrays (gratings or nanoholes with well-known dimen- for the disclosure of the image artifacts.
sions, etc.). From the images obtained for standard test samples We have illustrated and discussed a number of AFM image arti-
one can deduce the shape of the AFM tip. Another solution is self- facts, some of which can be compensated or minimised. We hope
imaging of the probing tip, in which the AFM probe is scanned over. that presented illustrations, hints and discussion will help avoid
Appropriate for this purpose are the materials that exhibit very misinterpretation of registered images. As has been shown, despite
sharp (needle like) and stable protrusions. Fortunately, this type of the fact that practically always the AFM images contain tip-related
protrusions can be prepared by thermal treatment of a thin layer artifacts, quite a useful information can often be drawn from the
of one material deposited on another with a different lattice con- collected results.
stant. As an example, we present in Fig. 17 the topography image
of the probing tip self-imaging by sharp NaCl islands formed on References
Si(0 0 1) substrate after annealing at elevated temperatures. The
same fine features repeated many times in Fig. 17 inform us that [1] R.J. Colton, J. Vac. Sci. Technol. B 22 (2004) 1609.
[2] C.M. Mate, G.M. McClelland, R. Erlandsson, S. Chiang, Phys. Rev. Lett. 59 (1987)
we have to do with quite sharp protrusions on the probed sam-
1942.
ple that are able to reproduce the actual shape of the tip. When [3] N.A. Burnham, R.J. Colton, J. Vac. Sci. Technol. A 7 (1989) 2906.
the actual geometry of the AFM probe is known, the topography [4] H.K. Wickramasinghe, Sci. Am. 261 (1989) 98.
[5] B. Drake, C.B. Prater, A.L. Weisenhorn, S.A.C. Gould, T.R. Albrecht, C.F. Quate,
image of the material under study can be deconvoluted, giving rise
D.S. Cannell, H.G. Hansma, P.K. Hansma, Science 243 (1989) 1586.
to a much better approximation of the specimen than the origi- [6] F.J. Giessibl, Rev. Mod. Phys. 75 (2003) 949.
nal unprocessed result. However, it is always worth remembering [7] I. Palaci, Atomic force microscopy studies of nanotribology and nanomechan-
that deconvolution works only in favorable relations of the speci- ics, in: Ph.D. Thesis, EPFL de Lasuanne, Lasuanne, Switzerland, 2007.
[8] K.B. Jinesh, Atomic-scale friction: Thermal effects and capillary condensation,
men and probe geometries. If probing tip is not sharp enough for in: Ph.D. Thesis, University of Leyden, Leyden, MA, 2006.
the sharp sample features, as illustrated in Figs 11, 12c, and 15, [9] M. Brink, Imaging single-electron charging in nanostructures by low-
there are areas that cannot be touched and probed. In such situa- temperature scanning force microscopy, in: Ph.D. Thesis, Cornell University,
Cornell, NY, 2007.
tions, only pieces of the approximated image of the tip are obtained [10] P. Avouris, T. Hertel, R. Martel, Appl. Phys. Lett. 71 (1997) 285.
instead of the sample topography. [11] R. Wiesendanger, Scanning Probe Microscopy and Spectroscopy, Methods and
Applications, Cambridge University Press, Cambridge, 1994.
[12] M. Kuwahara, H. Abe, H. Tokumoto, T. Shima, J. Tominaga, H. Fukuda, Mater.
Charact. 52 (2004) 43.
4. Conclusion [13] S. Kasas, V. Gotzos, M.R. Celio, Biophys. J. 64 (1993) 539.
[14] P.C. Braga, D. Ricci (Eds.), Methods in Molecular Biology, Volume 242, Atomic
force microscopy biomedical methods and applications, Humana Press, New
The AFM is susceptible to a variety of factors, among which the
Jersey, 2004.
temperature variations and mechanical vibrations are very com- [15] R. Erlandsson, L. Olsson, P. Mårtensson, Phys. Rev. B: Condens. Matter 54
mon. Tip related artifacts are the most frequently encountered and (1996) R8309.
the most difficult to avoid in AFM images. In cases in which the [16] P. Griitter, W. Zimmermann-Edling, D. Brodbeck, Appl. Phys. Lett. 60 (1992)
2741.
dimensions of the sample features are comparable to that of the [17] U.D. Schwarz, H. Haefke, P. Reimann, H.-J. Guntherodt, J. Microsc. 173 (1994)
blunt, broken or dirty probing tip, the recognition of the image 183.
F. Gołek et al. / Applied Surface Science 304 (2014) 11–19 19

[18] P.M. Williams, K.M. Shakesheff, M.C. Davies, D.E. Jackson, C.J. Roberts, S.J.B. [66] A. Beyder, F. Sachs, Ultramicroscopy 106 (2006) 838.
Tendler, Langmuir 12 (14) (1996) 3468. [67] R.W. Carpick, M. Salmeron, Chem. Rev. 97 (1997) 1163.
[19] K.I. Westra, A.W. Mitchell, D.J. Thomson, J. Appl. Phys. 74 (1993) 3608. [68] R.W. Carpick, Q. Dai, D.F. Ogletree, M. Salmeron, Tribol. Lett. 5 (1998) 91.
[20] J.E. Griffith, D.A. Grigg, J. Appl. Phys. 74 (1993) R83. [69] C.M. Mate, Phys. Rev. Lett. 68 (1992) 3323.
[21] H. Edwards, R. McGlothlin, U. Elisa, J. Appl. Phys. 83 (1998) 3952. [70] B. Bhushan, J.N. Israelachvili, U. Landman, Nature 375 (1995) 607.
[22] Y.J. Chen, I.H. Wilson, C.S. Lee, J.B. Xu, J. Appl. Phys. 82 (1997) 5859. [71] B. Bhushan, T. Kasai, Nanotechnology 15 (2004) 923.
[23] R.V. Gainutdinov, P.A. Arutyunov, Russ. Microlectron. 30 (2001) 219. [72] R.J. Cannara, M.J. Brukman, R.W. Carpick, Rev. Sci. Instrum. 76 (2005) 053706.
[24] S.B. Velegol, S. Pardi, Xu Li, D. Velegol, B.E. Logan, Langmuir 19 (2003) 851. [73] G. Meyer, N.M. Amer, Appl. Phys. Lett. 57 (1990) 2089.
[25] P.J. Godowski, V. Maurice, P. Marcus, Chem. Anal. (Warsaw) 40 (1995) 231. [74] J.A. Ruan, B. Bhushan, J. Appl. Phys. 76 (1994) 8117.
[26] A.P. Gunning, A.R. Mackie, P.J. Wilde, V.J. Morris, Langmuir 20 (2004) 116. [75] E. Meyer, R. Lüthi, L. Howald, M. Bammerlin, M. Guggisberg, H.-J. Güntherodt,
[27] P.H. Osanna, Meas. Sci. Rev. 1 (2001) 15. J. Vac. Sci. Technol., B 14 (1996) 1285.
[28] P. West, N. Starostina, Microsc. Today 11 (May/June) (2003) 20. [76] T. Müller, M. Lohrmann, T. Kässer, O. Marti, J. Mlynek, G. Krausch, Phys. Rev.
[29] Y. Chen, J.Y. Cai, M.L. Liu, G.C. Zeng, Q. Feng, Z.W. Chen, Scanning 26 (4) (2004) Lett. 79 (1997) 5066.
155. [77] H. Hölscher, D. Ebeling, U.D. Schwarz, Phys. Rev. Lett. 101 (2008) 246105.
[30] I. Schmitz, M. Schreiner, G. Friedbacher, M. Grasserbauer, Anal. Chem. 69 [78] P. Steiner, E. Gnecco, F. Krok, J. Budzioch, L. Walczak, J. Konior, M. Szymonski,
(1997) 1012. E. Meyer, Phys. Rev. Lett. 106 (2011) 186104.
[31] F.M. Ohnesorge, Phys. Rev. B: Condens. Matter 61 (2000) R5121. [79] G. Ehrlich, F.G. Hudda, J. Chem. Phys. 44 (1966) 1039.
[32] J. Vesenka, S. Manne, R. Giberson, T. Marsh, E. Henderson, Biophys. J. 65 (1993) [80] R.L. Schwoebel, E.J. Shipsey, J. Appl. Phys. 37 (1966) 3682.
992. [81] Y. Dong, X.Z. Liu, P. Egberts, Z. Ye, R.W. Carpick, A. Martini, Tribol. Lett. 50
[33] A.V. Bolshakova, O.I. Kiselyova, I.V. Yaminsky, Biotechnol. Prog. 20 (2004) (2013) 49.
1615. [82] M.A. Lantz, S.J. O’Shea, M.E. Welland, Phys. Rev. B: Condens. Matter 55 (1997),
[34] P. Eaton, P. West, Atomic Force Microscopy, Oxford University Press, Oxford, 10 776.
2010, pp. 121–138 (Chapter 6). [83] Y. Dong, Q. Li, A. Martini, J. Vac. Sci. Technol. A 31 (2013) 030801.
[35] C. Barth, A.S. Foster, C.R. Henry, A.L. Shluger, Adv. Mater. 23 (2011) 477. [84] P.E. Sheehan, Chem. Phys. Lett. 410 (2005) 151.
[36] Y. Gan, Surf. Sci. Rep. 64 (2009) 99. [85] M. Urbakh, J. Klafter, D. Gourdon, J. Israelachvili, Nature (London) 430 (2004)
[37] I. Horcas, R. Fernandez, J.M. Gomez-Rodriguez, J. Colchero, J. Gomez-Herero, 525.
A.M. Baro, Rev. Sci. Instrum. 78 (2007) 013705. [86] C.M. Mate, Tribology on the Small Scale: A Bottom Up Approach to Friction,
[38] R. Changhai, S. Lining, Sens. Actuators, A 122 (2005) 124. Lubrication, and Wear, Oxford University Press, Oxford, 2007.
[39] K.R. Koops, P.M.L.O. Scholte, W.L. de Koning, Appl. Phys. A 68 (1999) 691. [87] H. Hölscher, A. Schirmeisen, U.D. Schwarz, Philos. Trans. R. Soc., A 366 (2008)
[40] G. Schitter, A. Stemmer, Nanotechnology 13 (2002) 663. 1383.
[41] H.G. Xu, T. Ono, M. Esashi, J. Micromech. Microeng. 16 (2006) 2747. [88] I. Szlufarska, M. Chandross, R.W. Carpick, J. Phys. D: Appl. Phys. 41 (2008)
[42] Y. Liu, Magnetic dissipation force microscopy, in: Ph.D. Thesis, McGill Univer- 123001.
sity, Montreal, 1997. [89] E. Meyer, R.M. Overney, K. Dransfeld, T. Gialog, Nanoscience Friction and
[43] H.-Y. Nie, N.S. Mclntyre, Rev. Sci. Instrum. 78 (2007) 023701. Rheology on the Nanometer Scale, World Scientific Publishing Co. Pte. Ltd,
[44] J.E. Castle, P.A. Zhdan, P. Singiai, J. Phys. D: Appl. Phys. 31 (1998) 3437. Singapore, 1998.
[45] F. Kiendl, Advanced data acquisition and exploitation in scanning near-field [90] U. Landman, W.D. Luedtke, M.W.J. Ribarsky, J. Vac. Sci. Technol., A 7 (1989)
(magneto-)optical microscopy, in: Ph.D. Thesis, Aachen University, Aachen, 28299.
2005. [91] J. Vesenka, M. Guthold, C.L. Tang, D. Keller, E. Delaine, C. Bustamante, Ultra-
[46] F. Kiendl, G. Güntherodt, Ultramicroscopy 102 (2005) 299. microscopy 42–44 (1992) 1243.
[47] T.R. Keel, The application of in situ AFM to the study of molecular and [92] R.E. Mahaffy, C.K. Shih, F.C. MacKintosh, j. Käs, Phys. Rev. Lett. 85 (2000) 880.
macromolecular crystallization, in: Ph.D. Thesis, University of Nottingham, [93] R. Garcia, R. Perez, Surf. Sci. Rep. 47 (2002) 197.
Nottingham, 2004. [94] A.S. Paulo, R. Garcia, Biophys. J. 78 (2000) 1599.
[48] L.H. Pope, M.C. Davies, C.J. Roberts, S.J.B. Tendler, P.M. Williams, Anal. Com- [95] F. Braet, R. De Zanger, C. Seynaeve, M. Baekeland, E. Wisse, J. Electron. Microsc.
mun. 35 (1998) H5. 50 (2001) 283.
[49] A. Radenovic, E. Bystrenova, L. Libioulle, M. Taborelli, J.A. DeRose, G. Dietler, [96] R.P. Richter, A.R. Brisson, Biophys. J. 88 (2005) 3422.
Rev. Sci. Instrum. 74 (2003) 1022. [97] Y. Sugawara, Y. Sano, N. Suehira, S. Morita, Appl. Surf. Sci. 188 (2002) 285.
[50] S. Tobenas, E. Bystrenova, A. Radenovic, G. Di Santo, G. Dietler, Jpn. J. Appl. [98] M. Kageshima, H. Ogiso, H. Tokumoto, Surf. Sci. 517 (2002) L557.
Phys. 45 (2006) 2345. [99] Y. Guo, J. Wu, Y. Zhang, Chem. Phys. Lett. 362 (2002) 314.
[51] D. Yamamoto, K. Tani, T. Gotoh, T. Kouyama, Micron 34 (2003) 9. [100] L. Roschier, R. Tarkiainen, M. Ahlskog, M. Paalanen, P. Hakonen, Microelectron.
[52] K. Sagisaka, M. Kitahara, D. Fujita, Jpn. J. Appl. Phys. 42 (2003) L126. Eng. 61–62 (2002) 687.
[53] K. Saitoh, K. Hayashi, Y. Shibayama, K. Shirahama, J. Low Temp. Phys. 150 [101] K.D. Jandt, Surf. Sci. 491 (2001) 303.
(2008) 561. [102] D. Fotiadis, S. Scheuring, S.A. Muller, A. Engel, D.J. Muller, Micron 33 (2002)
[54] I. Shiraki, Y. Miyatake, T. Nagamura, K. Miki, Rev. Sci. Instrum. 77 (2006) 385.
023705. [103] A. Ikai, A. Idiris, H. Sekiguchi, H. Arakawa, S. Nishida, Appl. Surf. Sci. 188 (2002)
[55] G. Haugstad, Atomic Force Microscopy: Understanding Basic Modes and 506.
Advanced Applications, Wiley & Sons, Hoboken, New Jersey, 2012. [104] J. Hu, Y. Zhang, H. Gao, M. Li, U. Hartmann, Nano Lett. 2 (2002) 55.
[56] R. Kassies, K.O. van der Werf, M.L. Bennink, C. Otto, Rev. Sci. Instrum. 75 (2004) [105] Y-R. Ma, Y. Liou, Y-D. Yao, J. Magnet. Magnet. Mat. 282 (204) 342.
689. [106] F.O. Goodman, N. Garcia, Phys. Rev. B: Condens. Matter 43 (2004) 4728.
[57] J.I. Paredes, A. Martinez-Alonso, J.M.D. Tascon, J. Microsc.—Oxford 200 (2000) [107] G.M. Fuchs, T. Prohaska, G. Friedbacher, H. Hutter, M. Grasserbauer, J. Anal.
109. Chem. 351 (1995) 143.
[58] Z. Tao, B. Bhushan, J. Tribol. 128 (2006) 865. [108] P. Markiewicz, M.C. Goh, Rev. Sci. Instrum. 66 (1995) 3186.
[59] G. Shang, X. Qiu, C. Wang, C. Bai, Appl. Phys. A 66 (1998) S333. [109] [Link]
[60] D. Tománek, Theory of atomic-scale friction, in: R. Weisendanger, H.J. Gün- [110] J.S. Villarrubia, J. Res. Natl. Inst. Stand. Technol. 102 (1997) 425.
terdot (Eds.), Scanning Tunneling Microscopy III Springer Series in Surface [111] A.A. Bukharaev, N.V. Berdunov, D.V. Ovchinnikov, K.M. Salikhov, Scanning
Sciences, 29, Springer, Berlin, 1993, pp. 269–292 (Chapter 11). Microsc. 12 (1998) 225.
[61] E. Gnecco, E. Meyer, Fundamentals of Friction and Wear on the Nanoscale [112] R. Bernardes-Filho, O.B.G. de Assis, Braz. Arch. [Link]. 48 (2005) 667.
Nanoscience and Technology, Springer, Berlin Haidelberg, 2007. [113] L. Udpa, V.M. Ayres, Yuan Fan, Qian Chen, S.A. Kumar, IEEE Signal Process.
[62] M.W. Such, D.E. Kramer, M.C. Hersam, Ultramicroscopy 99 (2004) 189. Mag. 23 (2006) 73.
[63] B. Bhushan, A.V. Kulkarni, V.N. Koinkar, M. Boehm, L. Odoni, C. Martelet, M. [114] P.M. Williams, K.M. Shakesheff, M.C. Davies, D.E. Jackson, C.J. Roberts, S.J.B.
Belin, Langmuir 11 (1995) 3189. Tendler, J. Vac. Sci. Technol. B14 (1996) 1557.
[64] R. Szoszkiewicz, B. Bhushan, B.D. Huey, J. Appl. Phys. 99 (2006) 014310. [115] M. Xu, D. Fujita, K. Onishi, Rev. Sci. Instrum. 80 (2009) 043703.
[65] R.W. Stark, G. Schitter, A. Stemmer, Ultramicroscopy 100 (2004) 309.

You might also like