Supporting information
XPS
3500
Br 3d area (CPS*eV)
3000
2500
2000
1500
1000
500
0
0
500
1000
1500
2000
XPS analysis time (s)
Figure S1: Bromine signal area from high-resolution scans of the Br 3d region as a function of
XPS analysis time. Bromine appears to be labile under XPS analysis conditions, so we limit our
measurements to a total of 240 s of X-ray exposure; a 60 s high-resolution scan of the Br 3d region,
followed by a 60 s high-resolution scan of the C 1s region, followed by a 120 s survey scan.
Table S1: Full XPS composition for UVO treatment time series according to the survey scan data.
Error bars represent the standard deviation of the integrated area as calculated by CasaXPS’s
“calculate error bars” method.
UVO treatment Br 3d
time (s)
(Atomic %)
C 1s
(Atomic %)
Si 2p
(Atomic %)
O 1s
(Atomic %)
0
1.3 ± 0.1
34.5 ± 0.4
34.9 ± 0.3
29.3 ± 0.3
10
1.1 ± 0.1
34.7 ± 0.4
32.3 ± 0.3
31.3 ± 0.3
20
0.67 ± 0.06
33.7 ± 0.4
33.6 ± 0.3
32.0 ± 0.3
30
0.61 ± 0.06
32.8 ± 0.4
31.5 ± 0.3
35.0 ± 0.3
60
0.20 ± 0.06
24.9 ± 0.5
36.0 ± 0.3
38.9 ± 0.3
100
-0.01 ± 0.06
15.4 ± 0.6
42.8 ± 0.4
41.8 ± 0.4
180
0.01 ± 0.09
11.4 ± 0.7
42.6 ± 0.4
46.0 ± 0.4
300
0.03 ± 0.07
5.3 ± 0.7
49.4 ± 0.4
45.3 ± 0.4
Figure S2: XPS survey of a Br-SI film on a Si wafer. Peaks representing key atoms carbon (C 1s),
oxygen (O 1s), silicon (Si 2p) and bromine (Br 3d) are labeled and integrated using a Shirley
baseline.
Figure S3: XPS survey of a Br-SI film on a Si wafer after exposure to 10 seconds of UVO
treatment. Peaks representing key atoms carbon (C 1s), oxygen (O 1s), silicon (Si 2p) and bromine
(Br 3d) are labeled and integrated using a Shirley baseline.
Figure S4: XPS survey of a Br-SI film on a Si wafer after 300 seconds of UVO treatment. Peaks
representing key atoms carbon (C 1s), oxygen (O 1s), silicon (Si 2p) and bromine (Br 3d) are
labeled and integrated using a Shirley baseline.
1000
0s
800
10 s
Br 3d signal (CPS)
20 s
600
30 s
60 s
400
100 s
180 s
200
300 s
0
75
74
73
72
71
70
69
Photoelectron energy (eV)
-200
Figure S5: Overlaid high-resolution XPS scans of the Br 3d region at 0, 10, 20, 30, 60, 100, 180,
and 300 seconds of UVO treatment (light blue, orange, gray, yellow, dark blue, green, black, and
brown, respectively). The signal has in each case had the Shirley baseline subtracted. The peak
decreases in order of oxidation time.
Normalized Br 3d area (CPS*eV)
1.2
1
0.8
0.6
0.4
0.2
0
0
50
100
150
200
250
300
350
UVO treatment time (s)
Figure S6: Replicate UVO treatments on four different Br-SI films. Blue dots represent the series
with half-life 19.1 ± 0.9 s. Red dots represent the series with half-life 15.8 ± 0.4 s. Grey dots
represent the series with half-life 18.8 ± 1.0 s. Yellow dots represent the series pictured in Figure
2 corresponding to the Br-SI batch discussed in the rest of the paper with half- life 20.9 ± 1.1 s.
The XPS peak Br 3d areas have been normalized to the pre-exponential factor of the exponential
fit to each series.
ToF-SIMS
Secondary Ion Mass Spectral Data
Both positive ion and negative ion mass spectra were acquired from the samples. For these
samples, it turned out that the negative ion mass spectral data were more representative of the
chemistry of the polymer surface, due to the fact that the positive spectra were dominated by
siloxane peaks (a common surface contaminant), which resulted in the significant and nonlinear
suppression of polymer signals. Figure S7 shows a representative negative ion mass spectra
acquired from two samples, a virgin polymer film and one that has been exposed to 300 s of UVO.
In this particular example, it can be seen that the two samples have drastically different signals
within each mass region. For example, it can be seen clearly within the m/z 5 to m/z 100 window
that the Br signals are significantly diminished when the film is exposed to UVO for a prolonged
amount of time. Likewise, a strong substrate signal becomes prominent as the film is being etched
away (m/z 100 to m/z 200 window).
(a) 0 s Ablation
m/z 79, 81
79
81
Br-, Br-
m/z 123, 125
79
81
CO2 Br-, CO2 Br-
m/z 158, 160, 162
79
79
81
81
Br2-, Br Br-, Br2-
(b) 300 s Ablation
m/z 76, 77
CO4-, CHO4-
m/z 136, 137, 138
Si2O5-, Si2HO5-, Si2H2O5-
Figure S7: Negative ion mass spectra obtained from (a) a virgin polymer film and one that was
exposed to 300 s of UVO. Both spectra compare the three mass regions (m/z 5 to 100, m/z 101 to
200, and m/z 201 to 300) using the same vertical intensity scales.
Principal Components Analysis
Peak areas were selected from each mass spectrum by using the Ion-ToF SurfaceLab software’s
Auto Peak Search algorithm, selecting all peaks below m/z 800 with an intensity of 100 counts or
more. A final peak list was generated and used for integration of all peaks from the entire dataset.
The integrated results were tabulated and imported directly into a script written in-house for
MATLAB (MathWorks, Inc., Natick, MA), where it was subjected to principal components
analysis (PCA).24,28
In PCA, the singular value decomposition of the variance-covariance matrix (X) is calculated
from a pretreated data matrix, where all datasets were: 1) normalized to the total ion intensity to
focus on relative variations in peak intensities, rather than total ion intensities, 2) log decay scaling
was applied to emphasize smaller variances in the dataset, which are typically observed at higher
masses, and 3) the dataset was mean-centered so that any of the variances observed could be
attributed to differences from the mean.
The data will have several principal components (PC’s) which describe mass spectral differences
observed in the dataset. These PC’s are usually ordered by the amount of variance in the dataset
they describe and are typically defined as PC1, PC2, PC3, … PCn, where n is the number of
principal components obtained for the dataset. The PCA transform is then used to rotate the dataset
such that the data are orthogonal, allowing for the areas of maximum variance to be easily
visualized. This process can be described mathematically as:
X·M = R = T·PT
Where X is the mean centered data matrix of covariances, M is the PCA transform matrix used
to rotate the data such that it is orthogonal, and R is the resultant orthogonal matrix. R is equal to
T·PT where T is the orthogonal matrix of sample “scores”, and P T is the transposed orthogonal
matrix of mass spectrometry peak “loadings”. These values (scores and loadings) are frequently
used to describe the differences in the PCA data, where scores indicate the length of the projection
of a particular spectrum onto a principal component, and loadings are the mass spectral peaks
(variables) that describe that particular component. Hence if the values of the scores obtained from
two different mass spectra are similar for a particular principal component, say PC1, the
corresponding mass spectral peak intensities described by PC1 are also similar. Similarly, if the
scores for the two mass spectra are different, then the corresponding mass spectra data will be
different. Several examples of this method have been given in the literature, and it is now used as
a tool for SIMS data analysis on a regular basis.24,25,29,30
Figure S8 shows the scores and loadings plots generated from the negative spectra in this
study, where the first principal component (PC) captured roughly 91.1 % of the variance. The
linear scores relationship shows that in this PC, increasing UVO exposure time corresponded
positively with peaks having positive loadings, while it corresponded negatively with peaks having
negative loadings. More specifically, increasing UVO exposure time resulted in oxidation of the
polymer (indicated by presence of sputtered species such as CHO2-, CO4-, CHO4-, and CH2O4-) as
well as chain-scission, where shortening of the polymer chain led to the increased intensity of
substrate fragments such as SiO2, SiHO2-, Si2O5-, and Si2HO5-, as well as substrate contaminants
such as buffer salts (indicated by presence of SO4H- and NaP2O7-). Although the mechanism of
chain-scission or polymer oxidation is difficult to determine from the peaks found in the positive
loadings, the peaks in the negative loadings (species that decrease in intensity with UVO exposure
time) shed some light. For example, highest intensity peaks are 79Br- and 81Br-, which indicate that
the C-Br bond is most sensitive to UVO. The next highest intensity peaks are Br dimers, whose
presence most likely suggests that free Br species are recombining in the sputtered flux. This is
followed by CO279Br- and CO281Br-, which may also be recombining in the sputtered flux with free
CO2 molecules. The most intriguing finding was that C4H5O2- was also present, suggesting that
the C-O bond holding the head group to the alkyl chain is the second mostly likely cleaved bond
in the UVO process.
Figure S8: Negative mass spectral PCA results, showing the scores and loadings plots from the
first principal component. In this principal component, roughly 91 % of the variance was captured.
Each data point in the scores plot represents a single mass spectrum, and three spectra were
obtained from each sample.
Contact angle
The water contact angle provides a qualitative measure of the availability of hydrophilic groups
at the first 1 nm to 2 nm of the surface of a film. Since the UVO treatment of our Br-SI wafers
introduces additional hydrophilic oxygen-containing moieties into the film, we expected the
decrease in water contact angle shown in Figure S9.
Contact Angle - 10 (Degrees)
100
10
1
0
50
100
150
200
250
300
UVO Treatment Time
Figure S9: Advancing water contact angle measurements (circles) on Br-SI wafers as a function
of oxidation time. The points represent single measurements on each chip, and therefore have no
associated error bars. An exponential fit is plotted (dotted line). The data are shifted by 10° in order
to properly display the exponential trend between 98° and 10° on a semilog plot.
The trend is not so clearly exponential in this case; the argument could be made for a piecewise
linear model between 0 s and 100 s. Nevertheless, to be consistent with our other experiments we
fit an exponential model, decaying from 98° to 10° with a half-life of 56 ± 8 s. This larger half-life
is an interesting discrepancy with the bromine signals from XPS and ToF-SIMS. The addition of
hydrophilic groups is not a binary change like the removal of bromine, however. Earlier work18
has shown that hydroxyls, ketones, aldehydes and acids are forming at various rates, each
presenting a different surface energy and affinity towards water. Therefore, the contact angle
measurement is not so much tracking the removal of bromine and reduction in final grafting
density, but rather the change from bromine to carboxylic acid and silanol or other more polar
functionalities, a multi-step, multi-rate process which appears exponential. We do not expect
roughness to play much of a role, since the films are only ~ 5 nm thick.