A protocol for rheological characterization of hydrogels for tissue

engineering strategies

Jonathan M. Zuidema,1 Christopher J. Rivet,1 Ryan J. Gilbert,1 Faith A. Morrison2

1
Department of Biomedical Engineering, Center for Biotechnology and Interdisciplinary Studies,
Rensselaer Polytechnic Institute, Troy, New York 12180-3590
2
Department of Chemical Engineering, Michigan Technological University, Houghton, Michigan, 49931

Received 30 July 2013; revised 6 November 2013; accepted 16 November 2013

Published online 6 December 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/jbm.b.33088

Abstract: Hydrogels are studied extensively for many tissue Frequency sweep to determine the linear equilibrium modu-
engineering applications, and their mechanical properties lus plateau of the hydrogel. (4) Time sweep with values
influence both cellular and tissue compatibility. However, it is obtained from strain and frequency sweeps to accurately
difficult to compare the mechanical properties of hydrogels report the equilibrium moduli and gelation time. Finally, the
between studies due to a lack of continuity between rheologi- rheological characterization protocol was evaluated using a
cal protocols. This study outlines a straightforward protocol composite MatrigelTM-methylcellulose hydrogel blend whose
to accurately determine hydrogel equilibrium modulus and mechanical properties were previously unknown. The proto-
gelation time using a series of rheological tests. These proto- col described herein provides a standardized approach for
cols are applied to several hydrogel systems used within tis- proper analysis of hydrogel rheological properties. V C 2013

sue engineering applications: agarose, collagen, fibrin, Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater,
MatrigelTM, and methylcellulose. The protocol is outlined in 102B: 10631073, 2014.
four steps: (1) Time sweep to determine the gelation time of
the hydrogel. (2) Strain sweep to determine the linear- Key Words: biomaterial, hydrogel, rheology, agarose, colla-
viscoelastic region of the hydrogel with respect to strain. (3) gen, fibrin, MatrigelTM, methylcellulose

How to cite this article: Zuidema JM, Rivet CJ, Gilbert RJ, Morrison FA. 2014. A protocol for rheological characterization of
hydrogels for tissue engineering strategies. J Biomed Mater Res Part B 2014:102B:10631073.

INTRODUCTION ties of hydrogels, such as equilibrium modulus and gelation

There is signicant interest in the use of hydrogels for tis- time, is of critical importance.
sue engineering applications, as displayed in a recent review Rheology is an appropriate method for characterizing
by Van Vlierberghe et al.1 This interest is driven by the ben- hydrogel mechanical properties since it is quick, sensitive,
ecial characteristics that hydrogels present for both in vitro requires small sample sizes, and is revealing of differences
and in vivo applications. In vitro, hydrogels may serve as in architecture such as degree of crosslinking, proximity of
either a two-dimensional (2D) growth-supporting substrate2 the glass transition, structural homogeneity/heterogeneity,
or as a three-dimensional (3D) microenvironment.3 Recent and molecular weight. There are a number of reviews of the
reviews4,5 describe the complex, mechanically driven com- rheological properties of hydrogels composed of proteins,11
munications that occur between cells and their environment polysaccharides,12 or both.13 The main rheological technique
in vitro. Hydrogels are important for studying mechanical to characterize hydrogels is small-amplitude oscillatory
interactions because their mechanical properties are tunable shear (SAOS). In SAOS, a small amount of sample (<1 g) is
through the use of cross-linking agents6 or by altering the placed between parallel disks or a cone and a disk, and a
physical2 and chemical7 composition of the formulation. small-amplitude torsional oscillation generates shear ow in
These ndings may then be translated to in vivo and clinical the sample.14 In the linear-viscoelastic regime (LVE), the
applications, as many hydrogels are injectable and conform frequency-dependence of the dynamic moduli G0 (x) and
to their implantation site geometry. Hydrogels are also capa- G00 (x) may be used to deduce important aspects of gel
ble of sustained release for drug8 or gene therapies9 and structure and mechanical behavior. For example, with fre-
can provide an injection vehicle/temporary scaffold for cell- quency sweeps we can detect gel character (degree of cross
based therapies.10 For all these applications, understanding linking), entanglement, and the glass transition, and some
and properly quantifying the underlying mechanical proper- details of chain architecture.15 Essential to the use of SAOS

Correspondence to: F.A. Morrison (e-mail: fmorriso@mtu.edu)

Contract grant sponsor: NIH; contract grant numbers: HD061096, NS062392
Contract grant sponsor: NSF; contract grant number: NSF CAREER: 1150125

C 2013 WILEY PERIODICALS, INC.

V 1063
is that the moduli be independent of the amplitude of the chosen examples are in widespread use: agarose, collagen,
deformation (LVE limit). brin, MatrigelTM, and methylcellulose. Agarose and methyl-
In particular, SAOS is effective in identifying the equilib- cellulose are polysaccharide hydrogels that exhibit opposing
rium shear modulus of a gel, Ge,12 for gels that are not too gelation proles; agarose gelation occurs upon cooling
stiff. To determine the equilibrium modulus of a gel, G0 (x) whereas methylcellulose gelation occurs upon heating. Colla-
and G00 (x) are measured in the linear-viscoelastic limit, and gen, brin, and MatrigelTM are protein-based hydrogels that
the limiting value of G0 at low frequency is identied as the undergo gelation by pH changes, enzymatic cross-linking,
shear modulus. Other methods of measuring Ge for hydro- and heating, respectively. The protocol presented herein pro-
gels include the creep experiment11 and the unconstrained vides a clear method for accurately determining the rheologi-
tensile test.15 SAOS and the tensile test are both quick tests cal characteristics of hydrogels, and it is recommended that
that are effective on stronger gels; the creep test is used investigators wanting to assess these properties follow the
when a precise measurement of Ge is desired on a very thoroughness of this protocol. We evaluated the ability of
weak gel.11 Note that the conned elongation test16,17 meas- the protocol to measure the equilibrium strength and the
ures bulk modulus, a mechanical property not related to the time to gelation of a MatrigelTM-methylcellulose hydrogel
shear modulus.18 composite whose mechanical properties were unknown. Our
SAOS is ideal for monitoring changing systems (such as suggested protocol is a cycle of the three tests: time sweep,
gelation of a hydrogel) since the measurement is quick and strain sweep, frequency sweep, followed by a second time
can be monitored as a function of time. The SAOS frequency sweep. The protocol reported ensures that the assumptions
sweep used to measure Ge also allows us to identify x1, the of SAOS are respected and permits meaningful comparisons
lower frequency limit at which solid (gel, constant modulus) among studies in hydrogel-based tissue-engineering strat-
behavior is observed. Once x1 is known, the evolution of a egies. Although we limit our discussion to obtaining Ge and
sol to a gel may be monitored with SAOS. To do this, G0 and the time to gelation, determining the limits of linear viscoe-
G00 are measured at a frequency below x1, as a function of lasticity and the frequency dependence of the moduli are
time or extent of reaction. The SAOS measurement thus both important rst steps towards decoding other rheologi-
tracks the growth of the G0 value from zero (sol) to the cal signatures present in SAOS.
value of the equilibrium modulus Ge. The same congura-
tion may be used to monitor structural breakdown in METHODS
hydrogels by imposing a nonlinear strain (large-amplitude Hydrogel fabrication
oscillatory shear, for example, or a proscribed amount of Agarose. SeaPrepTM Agarose was purchased from Lonza
preshear19) on the gel and subsequently monitoring the (Walkersville, MD) and used to prepare a 2% (wt./wt.)
structural reformation with SAOS.7 Detailed characterization hydrogel in saline solution (0.85% sodium chloride in
of the gelation point is also possible with SAOS, following deionized, distilled water and 0.5 mM HEPES buffer) using a
the methods established by Chambon and coworkers.2022 previously published protocol.2 The hydrogel was fabricated
In those tests, both frequency and time are varied on a gel- by rst dispersing agarose in cold saline solution and subse-
ling system, and the gelation point is recognized as the quently heating to 60 C for 15 min under constant stirring.
moment when G0 5 G00 for all frequencies. The hydrogel was kept at this temperature until a 250 lL
Using SAOS on fragile hydrogels presents a challenge, sample was removed and dispensed onto the rheometer
because the measurement requires a mechanical perturba- plate preheated to 37 C. The test geometry was lowered to
tion of the system. To not destroy or modify hydrogel sam- the desired gap height and excess hydrogel was discarded.
ples, strains and stresses need to be kept low during
rheological testing. Rheometers have a nite sensitivity, Collagen. Type I, rat-tail collagen was purchased from Invi-
however, and thus the choice of imposed strain amplitude trogen (Grand Island, NY) and was prepared according to
must be a compromise between generating adequate signal the manufacturers protocol. In brief, the stock collagen
and preserving the integrity of the sample. Attention to the solution (5 mg/mL in acetic acid) was diluted to 4 mg/mL
effect of frequency is important as well, since the choice of with a neutralization buffer [1N sodium hydroxide in
frequency determines the relaxation processes that are phosphate-buffered saline (PBS)] while kept on ice. A 300
observed. Always using the same frequency across many lL sample of the neutralized solution was immediately dis-
types of hydrogels is not correct since each gel has its own pensed on the rheometer plate chilled to 4 C and the test
frequency (i.e., relaxation) behavior. It can be a challenge to geometry was lowered to the desired height. Collagen was
get all the details of a rheological test right, particularly as easier to manipulate at larger volumes, so 300 lL of solu-
one moves from one hydrogel system to another (e.g., pro- tion and a 1000-lm gap were chosen for collagen instead of
tein hydrogels to polysaccharide hydrogels). the 250 lL hydrogel volume and 500-lm gap height used
The goal of this work is to develop and evaluate a proto- for the other hydrogels.
col to provide the equilibrium strength, Ge, as well as the
time to gelation for hydrogels frequently used in tissue Fibrin. A brin hydrogel was prepared by using a modied
engineering applications. A panel of hydrogels was chosen version of a previously published procedure.23 The brin
to display the utility of the protocol across multiple gelation hydrogel was created by crosslinking brinogen using the
mechanisms (e.g., temperature, pH, and enzymatic). The enzyme thrombin. Fibrinogen and thrombin were purchased

1064 ZUIDEMA ET AL A PROTOCOL FOR RHEOLOGICAL CHARACTERIZATION OF HYDROGELS

 ORIGINAL RESEARCH REPORT

TABLE I. Parameter Table.

Hydrogel Parameter Strain Sweep Frequency Sweep Time Sweep

Agarose Temperature ( C) 10 10 10
Gap Size (lm) 500 500 500
Equilibrium Time (min) 30 30
Frequency (Hz) 1 0.01100 1
% Strain 0.1100 10 10
Collagen Temperature ( C) 37 37 37
Gap Size (lm) 1000 1000 1000
Equilibrium Time (min) 45 45
Frequency (Hz) 1 0.01100 1
% Strain 0.1100 10 10
Fibrin Temperature ( C) 37 37 37
Gap Size (lm) 500 500 500
Equilibrium Time (min) 30 30
Frequency (Hz) 1 0.01100 1
% Strain 0.1100 5 5
MatrigelTM Temperature ( C) 37 37 37
Gap Size (lm) 500 500 500
Equilibrium Time (min) 30 30
Frequency (Hz) 1 0.01100 1
% Strain 0.1100 10 10
Methylcellulose Temperature ( C) 37 37 37
Gap Size (lm) 500 500 500
Equilibrium Time (min) 30 30
Frequency (Hz) 1 0.01100 1
% Strain 0.1100 5 5
MatrigelTM-MC Temperature ( C) 37 37 37
Gap Size (lm) 500 500 500
Equilibrium Time (min) 30 30
Frequency (Hz) 1 0.001100 1
% Strain 0.1100 10 10

The temperatures, gap sizes, equilibrium times, frequencies, and percent strains are presented for all tests ran in the study.

from Sigma Aldrich, (St. Louis, MO) and dissolved in Dulbec- dispensed on the rheometer plate cooled to 4 C. The test
cos phosphate-buffered saline (DPBS) with calcium. Fibrino- geometry was lowered to the desired gap height and excess
gen and thrombin were dissolved in DPBS at concentrations solution was discarded.
of 10 mg/mL and 2 U/mL, respectively. The rheometer plate
was chilled to 4 C and the brinogen solution was rst dis- MatrigelTM-methylcellulose hydrogel blend. MatrigelTM
pensed followed by the thrombin. Gelation occurs rapidly so and methylcellulose were combined to investigate the ability
the test geometry was lowered to the desired gap height of methylcellulose to alter the gelation characteristics and
immediately after the thrombin was dispensed, ensuring gel mechanical properties of MatrigelTM. To fabricate the com-
formation while conned between the plate and the test posite hydrogel, MatrigelTM was rst thawed overnight on
geometry. ice at 4 C. 5% methylcellulose (wt./wt.) in saline solution
was created as described above. MatrigelTM was then mixed
MatrigelTM. MatrigelTM was purchased from BD Biosciences with methylcellulose on ice at the proportions of 4:1 (vol./
(Franklin Lakes, NJ) and prepared according to the manufac- vol.) of MatrigelTM to 5% methylcellulose solution. This cre-
turers protocol. In brief, MatrigelTM was thawed overnight ated a hydrogel that had the composition of 80% (vol./vol.)
on ice at 4 C and a 250 lL sample of the solution was dis- MatrigelTM and 1% (wt./wt.) methylcellulose. A 250 lL
pensed on the rheometer plate that was chilled to 4 C. The sample was removed and dispensed on the rheometer plate
test geometry was lowered to the desired gap height and cooled to 4 C. The test geometry was lowered to the desired
excess solution was discarded. gap height and excess solution was discarded.

Methylcellulose. MethocelTM methylcellulose was provided Rheological characterization

by Dow Chemical (Midland, MI). Using a previously pub- Rheological characterization was performed on all hydrogel
lished protocol,2 a 6% (wt./wt.) hydrogel was prepared by samples using a TA Instruments AR-G2 rheometer (New
rst dispersing the methylcellulose in saline solution heated Castle, DE). The test geometry was a 20 mm diameter plate,
to 60 C. The dispersion was then placed on ice to dissolve and Table I outlines the testing parameters. Each sample
the methylcellulose, forming a clear liquid. The solution was was dispensed on the preheated/cooled rheometer plate
kept on ice until use. A 250 lL sample was removed and while in liquid state. The test geometry was lowered to the

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH B: APPLIED BIOMATERIALS | JUL 2014 VOL 102B, ISSUE 5 1065
FIGURE 1. Strain sweeps. The linear-viscoelastic limit was determined with respect to strain. G0 was determined from 0.1 to 100% strain for aga-
rose (A), collagen (B), fibrin (C), MatrigelTM (D), and methylcellulose (E). No values of G00 were obtained due to reasons stated in the methods.

desired gap height (500 or 1000 lm), and excess hydrogel these experiments; for experiments that seek long-time
was discarded. Rapidly heating or cooling the rheometer behavior of gels such precautions are essential.
plate initiated gelation. The testing sequence was then The accuracy of SAOS moduli measurements depends on
applied to the sample. Each hydrogel sample was used for the magnitude of the torque generated by the deformation
only one test. Each test was performed in triplicate and the and by the instruments ability to resolve the phase angle d
data represents the average of the three tests with corre- between the strain wave and the stress wave. Smaller gaps
sponding standard deviation. The testing time to determine produce larger torques; larger plates produce larger tor-
gelation time and modulus was very short and there was ques. The AR-G2 is rated to measure a lower limit of 0.003
therefore no need to use a humidied chamber or trap for mNm of torque; we found that the accuracy of G00

1066 ZUIDEMA ET AL A PROTOCOL FOR RHEOLOGICAL CHARACTERIZATION OF HYDROGELS

 ORIGINAL RESEARCH REPORT

FIGURE 2. Typical linear-viscoelastic behavior of a crosslinked gel. At high frequencies, the glassy behavior of polymers is observed (G0 5 109 Pa). At
low frequencies, G0 approaches a plateau, the gel shear modulus. The transition of G0 between these two plateaus characterizes the glass transition of
the polymer. The loss modulus G00 for a gel is very low and cannot be measured; G00 goes like x through the region of the glass transition as shown;
in the glassy region G00 goes through a maximum.14,15 [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

measurements were compromised when low torque signals in vivo at or near this temperature. In cases such as agarose,
combined with phase angles of close to 0 ; likewise, the which forms a gel by cooling, lower temperatures are
accuracy of G0 measurements were compromised when low required (i.e., 10 C). Furthermore, each sample is loaded
values of torque are combined with phase angles near 90 . onto the rheometer plate as a liquid and the test geometry
The reported values of G0 in this article correspond to a is lowered to the desired height while still a liquid sample.
minimum torque of at least 0.003 mNm; the reported values This ensures that the physical structure of the hydrogel,
of G00 correspond to a minimum value of torque of at least whether it is generated by chemical crosslinking or by phys-
0.010 mNm and G00 /G0 5 tan d of at least 0.01. Plate diame- ical electrostatic associations, occurs conned between the
ter was chosen to be as large as possible within the con- rheometer plate and the test geometry. If a hydrogel is
straints formed by the amount of sample available. formed prior to dispensing on the rheometer plate and sub-
sequently compressed by lowering the test geometry to the
RESULTS desired height, then the resulting data is indicative of a
Gelation time deformed hydrogel network and not the naturally forming
When working with a new hydrogel formulation, it is difcult network.
to know what SAOS settings to use for initial characterization
[e.g., strain amplitude (c0) and the frequency (x0)]. The values Strain sweeps
used for the rst gelation time test should be, if possible, val- The second recommended test is to check for the LVE limit
ues that are referenced in the literature for gels that are simi- by performing a strain sweep on a fully formed gel. A new
lar to the one being tested (we chose 510% strain and 1 Hz hydrogel solution was placed on the rheometer plates and
to start). For a gel about which little is known, it is not possi- the corresponding thermal conditions were applied to initi-
ble to choose the strain amplitude (c0) and the frequency (x0) ate gelation. The gel was allowed to reach equilibrium for
according to appropriate rheological criteria. We recommend the appropriate amount of time as determined by the pre-
that the rst step in the process be to produce a gel while ceding time sweep step in the absence of rheological moni-
monitoring with SAOS at an arbitrary strain and frequency; toring. A strain sweep from 0.1 to 100% strain was then
this time (or extent of reaction) sweep then serves as an ini- conducted at a frequency of x0 5 1.0 Hz (6.3 rad s21) (cho-
tial data point for the studyproviding a general indication of sen arbitrarily) on the fully formed gel. The results of the
gelation time. This is especially critical to ensure that subse- strain sweeps for the gels studied here are shown in Figure
quent tests (e.g., strain sweep and frequency sweep) are per- 1. G00 was too small to measure for all hydrogels, as is to be
formed on a fully gelled sample. For the hydrogels tested in expected for fully formed hydrogels (Figure 2; more discus-
this study, the gelation times are either 30 or 45 min (Table I). sion of this below). Agarose showed linear behavior (con-
Each test should be conducted at 37 C if possible, since stant values as strain is varied) of G0 up to 15 % strain
in tissue engineering applications the gel will be forming [Figure 1(A)], and a strain of 10% was selected for

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH B: APPLIED BIOMATERIALS | JUL 2014 VOL 102B, ISSUE 5 1067
FIGURE 3. Frequency sweeps. The linear equilibrium modulus plateau (Ge) with respect to frequency was determined. G0 and G00 were deter-
mined from 0.01 to 100 Hz for agarose (A), collagen (B), fibrin (C), MatrigelTM (D), and methylcellulose (E). Values of G00 were removed if the
data were not reliable as discussed in the report.

subsequent sweeps. Collagen had a linear strain region also Frequency sweeps
up to 15% strain [Figure 1(B)], and 10% strain was The presence of a gel is associated with the low frequency
selected for subsequent sweeps. Fibrin exhibited a linear plateau of the SAOS frequency sweep (Figure 2). To use
region up to 10% strain [Figure 1(C)], and 5% strain was SAOS to monitor gel formation or reformation, the testing
selected for subsequent sweeps. MatrigelTM had a linear frequency must correspond to this low-frequency plateau.
region up to 13% strain [Figure 1(D)], and 10% strain was In the initial time sweep conducted at the beginning of this
selected for subsequent sweeps. Methylcellulose had a lin- protocol, we arbitrarily chose the value of frequency to use;
ear strain region up to 10 % strain [Figure 1(E)], and 5% the results of the frequency sweep using the appropriate
strain was chosen for subsequent sweeps. strain (as determined in the preceding section) allow us to

1068 ZUIDEMA ET AL A PROTOCOL FOR RHEOLOGICAL CHARACTERIZATION OF HYDROGELS

 ORIGINAL RESEARCH REPORT

FIGURE 4. Final time sweeps. Gel formation was monitored with a time sweep for 30 min for agarose (A), collagen (B), fibrin (C), MatrigelTM (D),
and methylcellulose (E). G0 is presented for all tests, while G00 is only presented for methylcellulose (E) due to limits on instrument sensitivity.

rigorously choose a frequency to ensure that the time ter plate and the appropriate amount of time elapsed for
sweeps we perform reect the formation of the gel network the gel to reach equilibrium (no rheological monitoring).
and not some other structural change. Frequency sweeps from 0.01 to 100 Hz were conducted at
The purpose of the third recommended test is to nd an the LVE strain amplitude determined in the preceding sec-
appropriate value of frequency to employ in a second time tion. The results of frequency sweeps are shown in Figure
sweep that denes the equilibrium strength and gelation 3. For agarose, all frequencies measured reected gel behav-
time. As in the case of the strain sweeps, for frequency ior and a frequency of 1 Hz was selected for future tests
sweeps a new hydrogel solution was placed on the rheome- based on good torque signal at that frequency. For collagen,

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH B: APPLIED BIOMATERIALS | JUL 2014 VOL 102B, ISSUE 5 1069
brin and MatrigelTM, the gel region was observed below 20
Hz; 1 Hz was selected for future testing. For methylcellulose
the inuence of the glass transition extended to approxi-
mately 510 Hz; 1 Hz was therefore also an acceptable fre-
quency choice for this system.

Time sweeps
The time sweep performed as step one of this protocol
employed values of strain and frequency that were unveri-
ed; that is, it was unknown if strain was small enough to
be in the LVE and if the frequency was a value in the low-
frequency gel plateau. Once these limits were established,
we were able to select appropriate values for future tests,
for the intent of the rst time sweep was solely to identify
gelation time. The nal step of the protocol is a second time
sweep at the now veried strain and frequency values. This
second time sweeps was performed in order to obtain the
hydrogel equilibrium modulus.
Agarose time sweeps were conducted at 10% strain and
1 Hz at 10 C. Agarose slowly formed a gel, reaching stability
in 20 min near 1200 Pa [Figure 4(A)]. Collagen time sweeps
were conducted at 10% strain and 1 Hz at 37 C. Collagen
quickly formed a gel, reaching stability at 120 Pa in 3 min
[Figure 4(B)]. Fibrin time sweeps were conducted at 5%
strain and 1 Hz at 37 C. Fibrin formed a stabilized gel in 10
min near 160 Pa [Figure 4(C)]. MatrigelTM time sweeps
were conducted at 10% strain and 1 Hz at 37 C. MatrigelTM
formed a stable gel in 5 min near 90 Pa [Figure 4(D)].
Methylcellulose time sweeps were conducted at 5% strain
and 1 Hz at 37 C. For methylcellulose G0 crossed G00 at 5
min, showing when the material changed from more of a
liquid state to being dominated by the gel state [Figure
4(E)]. Methylcellulose then continued to increase the G0
modulus until stabilizing at 25 min near 340 Pa [Figure
4(E)]. The rst four hydrogels formed rapidly and the cross-
over of G0 and G00 was not observed (G00 was too low to mea-
sure for the extent of the tests). The gelation kinetics of
methylcellulose were slower than the other hydrogels, and
the crossover of G0 and G00 was captured for this material,
indicating a transition from a primarily liquid sample to a
predominately solid material. FIGURE 5. MatrigelTM-Methylcellulose hydrogel characterization. The
linear-viscoelastic limit was determined with respect to strain (A), the
linear equilibrium modulus plateau (Ge) with respect to frequency was
MatrigelTM-methylcellulose hydrogel blend determined using frequency sweeps (B), and gel formation and equilib-
To start the analysis for this previously uncharacterized rium modulus (Ge) was monitored by time sweeps for 30 min (C).
hydrogel blend, a time test was rst used to determine the
time of gelation (data not shown). The parameters for strain determine the frequency domain within the LVE for the
and frequency were taken from the results of the MatrigelTM MatrigelTM-methylcellulose hydrogel blend. The hydrogel
strain [10% strain (Figure 1(D)] and frequency [1 Hz (Fig- blend exhibited LVE behavior up to 9 Hz [Figure 5(B)], and
ure 3(D)] sweeps. After running the time sweep at 37 C, a a frequency of 1 Hz was chosen to perform subsequent
stable gel formed in under 10 min, so we utilized a 30 min time sweeps. Time sweeps were conducted on MatrigelTM-
preincubation time at 37 C before running the strain and methylcellulose blends using 10% strain and 1 Hz at 37 C.
frequency sweeps. A strain sweep from 0.1 to 100% strain The MatrigelTM-methylcellulose blend formed a stable gel in
was then conducted at a frequency of x0 5 1.0 Hz on the 5 min near 30 Pa [Figure 5(C)].
fully formed gel. It was found that the MatrigelTM -methyl-
cellulose blend was in the LVE range up to nearly 20%
strain [Figure 5(A)]. Therefore, we chose a strain of 10% DISCUSSION
for the subsequent frequency sweep. Frequency sweeps Characterization of a new hydrogel with SAOS requires a
from 0.01 to 100 Hz were conducted at 10% strain to trial-and-error approach, because it is not possible to know

1070 ZUIDEMA ET AL A PROTOCOL FOR RHEOLOGICAL CHARACTERIZATION OF HYDROGELS

 ORIGINAL RESEARCH REPORT

a priori the appropriate values of strain, frequency, and reects the equilibrium modulus of the crosslinked gel
gelation time to employ towards obtaining a value for the rather than the entanglement plateau. For the hydrogels
equilibrium gel strength. The choices for these parameters studied here there was no entanglement plateau, and fre-
must comply with the capabilities of the SAOS instrument, quency tests revealed the location of the gel plateau. In all
which presents limitations in terms of strain and phase- cases studied, a frequency of 1 Hz was within that plateau.
angle sensitivity. The chosen parameters must also comply One goal of SAOS characterization of gels in tissue engi-
with the material, which must not be destroyed by the neering is to monitor gelation (i.e., gel formation), and this
applied SAOS test. Furthermore, even when equipment and is provided by the time sweep test. Gelation time is an espe-
material issues are addressed, it is important that the mea- cially important parameter to consider when delivering
surement is interpretable; that is, we must be able to relate drugs, genes, or cells to an injury site, because it impacts
the moduli measured to the material characteristics of inter- how the hydrogel conforms to implantation site geometry
est. In this work, we illustrate such a process that appropri- and may inuence the outcomes of tissue regeneration.
ately characterizes hydrogels through SAOS. Slow gelation times may permit hydrogel leakage from the
Characterization begins with an initial time sweep to delivery site, potentially modifying therapeutic delivery,
determine the time to gelation. This value is then employed while rapid gelation times may inhibit the hydrogel from
in a subsequent strain sweep to determine the maximum conforming to the lesion geometry, creating a tissue-
strain that delivers a material response within the LVE limit. hydrogel mismatch.
The issue of strain is very important in SAOS measure- Time sweeps of a forming gel are straightforward to
ments. The strain needs to produce a sufciently high signal perform once the appropriate values of strain and frequency
without destroying the gel. In this work we performed are identied, as mentioned above. Time sweeps were con-
strain sweeps to determine the LVE limit and employed a ducted for 30 min at 37 C for all gels except for agarose,
value chosen within the LVE from the strain sweep in sub- which were conducted at 10 C (Figure 4). This was a suf-
sequent tests, such as the frequency sweep where strain is cient time for the gels tested to reach their nal moduli val-
held constant. An alternative choice is to use automatic ues. When working with gels that take longer to form, the
instrument features that allow users to choose stress and tests can be extended as necessary.
strain maxima and minima to control the sample deforma- When the gelling of a sample is observed with the SAOS
tion and signal strength; these features can be convenient technique, the sample begins as a pure liquid and is charac-
and may be used when users are accustomed to the per- terized by very low values of both G0 and G00 . As the gel
formance of their instrument and test materials. For our forms, the viscosity of the sample (related to G00 ) grows, as
instrument and for the hydrogels studied herein, between 5 does the elastic behavior (G0 ). While the sample is a sol, G00
and 10% strain was sufcient to generate appropriate signal is greater than G0 . When the gel forms, G0 becomes equal to
while keeping the material response within the LVE limit G00 20 and as the reaction proceeds, G0 exceeds G00 . Ultimately
(Table I). the elastic modulus G0 levels off at the value of the equilib-
Frequency is another parameter of SAOS tests that rium modulus, Ge, and G00 approaches zero. Because of the
affects signal strength, because higher frequencies generate large changes in G0 and G00 during gelation, SAOS is an ideal
higher torques. The probed physics change with frequency, tool for monitoring this change; however, the large changes
however, and thus to interpret the SAOS response the mea- and large relative values of the two dynamic moduli cause
surement frequency must be chosen with care. The fre- their own measurement issues, as we now discuss.
quency behavior of long-chain polymeric gels has been There are two raw signals in an SAOS rheometer: (1)
discussed at length in the literature15 (see Figure 2). A poly- torque, which combined with geometry information gives a
mer gel at high frequency exhibits G0 5 109 Pa and G0 value of complex modulus (G*), and (2) the angle d, which
greater than G00 (glassy solid behavior). As lower frequen- is the measured phase difference between the oscillating
cies are tested, both G0 and G00 drop (the glass transition), strain and stress waves. From these two measurements the
with G0 dropping more rapidly than G00 , causing the dynamic instrument calculates G0 and G00 14:
moduli to cross. At lower frequencies of testing, the elastic
response grows in importance and G0 and G00 cross again, G00 5G sin d
with G0 tending at low frequencies towards a plateau, the
equilibrium strength (Ge) of the gel, and in the same region G0 5G cos d
G00 decreases proportional to the frequency as lower and G00 =G0 5tan d
lower frequencies are tested. Note that for polymeric gels
that have entangled strands or other sources of strong elas- Because the sample starts as a low viscosity liquid, the
tic behavior, G00 does not decrease as the rst power of fre- initial signal of torque is small, and the initial phase angle is
quency at low frequency (see, for example, the polymer near 90 (sin d 5 0). As the gelation reaction occurs, the tor-
featured in Figure 2). For these materials G0 is strongly que increases and G00 becomes measureable. The storage
affected by entanglement, exhibiting an entanglement pla- modulus G0 is not capable of being measured at these early
teau at intermediate frequencies14,15 before progressing stages, however, because of the intrinsic nature of liquids.
towards glassy behavior at high frequency. Because of this, For a liquid, the strain and the stress are 90 out of phase,
care must be taken to verify that an observed plateau in G0 and thus G00 5 G*sin90 5 G* and G0 5 G*cos90 is zero. Until

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH B: APPLIED BIOMATERIALS | JUL 2014 VOL 102B, ISSUE 5 1071
sufcient elasticity develops in the sample, G0 remains too three, the values are 2.73 6 1.1 kPa. Our average of 1.2 kPa
low to measure, although accurate values of G00 may be is slightly less than the average of 2.73 kPa, but hydrogel
obtained if there is sufcient torque [Figure 4(E)]. As the preparation, the type of agarose used, and testing parame-
sample progresses towards the gel point, the phase angle d ters can account for these differences. The similarity
grows, and G0 becomes measurable. Good values of both G0 between the two results validates both methods, but draws
and G00 are obtained through the gel point, but after the gel awareness to the need for a standardized approach to
point, G0 grows strongly and G00 decreases as the loose directly compare hydrogel mechanical properties between
chains and dangling ends that are responsible for the vis- different studies.
cous signal are incorporated into the purely elastic gel. For The protocol outlined here was used on a gel formulation
the ripening gel G0 is large and G00 is small, leading to a very not evaluated previously in the literature in order to show
small phase angle that eventually becomes too small to its utility. The MatrigelTM-methylcellulose hydrogel formed a
measure. When the phase angle is too small to record, the gel in less than 5 min and reached an equilibrium modulus
measurement of G0 is still quite accurate (cos d 5 1), but value of 30 Pa (Figure 5). This value is much lower than the
accurate values of G00 may no longer be obtained [Figure 6% methylcellulose hydrogel (340 Pa) and is also less than
4(AD)]. As a rule of thumb, when the ratio of G0 to G00 (or the 100% MatrigelTM formulation (90 Pa) (Figure 4). The
G00 to G0 ) is larger than 10 the phase angle reliability of reduction in the percentages of the hydrogel materials, and
rheological instruments are approaching their limits. Com- the possibility of reducing methylcellulose chain entangle-
plex moduli ratios of greater than 102 or less than 1022 are ment and MatrigelTM gelation by the competition between
unlikely to be accurate for any instrument due to inherent the two different hydrogels are the two most likely causes
noise in detecting weak phase-angle signals. Because of this for the reduction in the equilibrium modulus. Regardless of
issue, care should be taken in reporting values of G0 and G00 the cause, this gel could have some utility in applications
to avoid assigning meaning to signals that reect noise that require hydrogels with very low moduli.
rather than material behavior. The difference in the mechanical properties of tissues is
The experiments discussed here are all in the LVE an important parameter that needs to be taken into account
regime. Within this regime the SAOS behavior discussed in when developing new hydrogels. Cells in different tissue
Figure 2 is expected of all gels, strong and weak. For gels microenvironments are presented with different elasticities:
with entangled strands between crosslinks, an intermediate brain 0.21 kPa, adipose 2.53.5 kPa, muscle 8.515 kPa,
entanglement plateau is observed between Ge and the glass cartilage 2025 kPa, and bone 2545 kPa.5 Several studies
transition behavior.14,15 The division of gels into strong and have been conducted detailing the effects of substrate stiff-
weak12 is a distinction based on the fracture strength of the ness on cell responses in vitro. Soft gels (200 Pa) pro-
gel, which results from a deformation that is outside the moted enhanced neurite growth of mixed cortical cultures
LVE limit. Tests that disrupt a gel may be of interest in tis- compared with stiffer gels (9000 Pa).25 Substrates with
sue engineering applications and may be pursued after LVE mechanical property gradients affected bone marrow stro-
characterization is complete. Nonlinear tests include frac- mal cell attachment and differentiation, with the stiffer sub-
ture strength,12 steady shear,19 large-amplitude oscillatory strates (80 kPa) resulting in increased gene expression of
shear (LAOS), start-up of steady shearing,12 and superposed osteogenic markers.26 Smooth muscle cells have also been
shear and SAOS.12 Results from such tests would permit dis- shown to respond to differences in substrate stiffness.27
cussion of the relative strengths of different gels. The gels presented in this study are mostly weak gels, and
The time sweep results for the ve systems studied here would be possible tissue engineering substrates for brain or
are shown in Figure 4. The methylcellulose time sweep spinal cord. Agarose, however, is about 200 Pa above the
gives an example of a gelation crossover point [Figure 4(E)]. modulus range of brain tissue, so it may not be the ideal
G0 crosses over and rises above G00 at about 5 min. This material of choice for central nervous system regeneration.
shows that before 5 min the material is mostly a liquid, low If agarose is pursued for central nervous system repair, one
viscosity material, and then after 5 min the material forms may fabricate an agarose hydrogel with a lower weight per-
a gel as G0 rises above G00 . G0 rose quickly above G00 for the centage than the hydrogel tested here (2%) in order to cre-
other gels tested, showing that the gel state dominates at an ate a hydrogel that is similar to the mechanical properties
early time for these hydrogels; G00 was too small to measure of central nervous system tissue.28 Whenever a new hydro-
accurately throughout the time sweeps for all but the meth- gel material is developed for a specic tissue engineering
ylcellulose. The nal values of equilibrium modulus, Ge, for situation, it is important to accurately characterize the
the ve gels were found to be: agarose 1200 Pa; collagen mechanical properties of the hydrogel to ensure that cells of
120 Pa; brin 160 Pa; MatrigelTM 90 Pa; methylcellulose the tissue of interest interact favorably with the implanted
340 Pa. Small amplitude testing in a compression mode material.
gives the same information as small-amplitude testing in a
shear mode (when divided by a factor of three), and we
compared our shear modulus results from a 2% agarose gel SUMMARY
to published results for the shear modulus determined by This study outlines a protocol to assess the linear-
unconned compression tests.24 The shear modulus that viscoelastic mechanical properties of hydrogels using small-
Dekosky et al. reported was 8.2 6 3.3 kPa. When divided by amplitude oscillatory shear. The rst step of the protocol is

1072 ZUIDEMA ET AL A PROTOCOL FOR RHEOLOGICAL CHARACTERIZATION OF HYDROGELS

 ORIGINAL RESEARCH REPORT

a time sweep with arbitrary strain and frequency choices; pigment epithelium-derived factor combined with chemotherapy.
Biomaterials 2009;30:48154823.
this rst test provides an idea of the gelation kinetics for
10. Wang C, Varshney RR, Wang DA. Therapeutic cell delivery and
the system and ensures that subsequent tests will be per- fate control in hydrogels and hydrogel hybrids. Adv Drug Deliv
formed on fully gelled samples. The initial time sweep is fol- Rev 2010;62:699710.
lowed by a strain sweep to determine the linear-viscoelastic 11. Yan C, Pochan DJ. Rheological properties of peptide-based hydro-
gels for biomedical and other applications. Chem Soc Rev 2010;
limit and a frequency sweep to determine a frequency that 39:35283540.
appropriately reects the equilibrium gel modulus Ge. 12. Ross-Murphy SB, Shatwell KP. Polysaccharide strong and weak
Finally, the time sweep is repeated with appropriate values gels. Biorheology 1993;30:217227.
of strain and frequency. The nal time sweep thus appropri- 13. Picout DR, Ross-Murphy SB. Rheology of biopolymer solutions
and gels. Scientific World J 2003;3:105121.
ately reects the growth of the gels strength and structure 14. Morrison FA. Understanding Rheology. New York, NY: Oxford
in the course of gel formation. Subsequent nonlinear tests University Press; 2001.
may be performed to probe samples nonlinear behavior 15. Ferry JD. Viscoelastic Properties of Polymers, 3rd ed. New York,
NY: Wiley; 1980.
and fracture strength.
16. Behravesh E, Timmer MD, Lemoine JJ, Liebschner MAK, Mikos
The outcome of this protocol presents a robust procedure AG. Evaluation of the in vitro degradation of macroporous hydro-
that is not often conducted in the eld of tissue engineering. gels using gravimetry, confined compression testing, and micro-
Following the outlined procedure ensures accurate results computer tomography. Biomacromolecules 2002;3:12631270.
17. Spiller KL, Laurencin SJ, Charlton D, Maher SA, Lowman AM.
when reporting equilibrium gel modulus and gelation time of Superporous hydrogels for cartilage repair: Evaluation of the
a hydrogel. These two features are integral to tissue engi- morphological and mechanical properties. Acta Biomater 2008;4:
neering as the implanted material must not only interface 1725.
18. Bird RB, Armstrong RC, Hassager O. Dynamics of Polymeric
mechanically well with the surrounding tissue (equilibrium
Liquids, Volume 1, 2nd ed. New York, NY: Wiley; 1987.
gel modulus), but it must also do so in a clinically relevant 19. Zhao GH, Kapur N, Carlin B, Selinger E, Guthrie JT. Characterisa-
manner (gelation time). Researchers in tissue engineering tion of the interactive properties of microcrystailline cellulose-
may nd this protocol useful as a rst step in developing carboxymethyl cellulose hydrogels. Int J Pharmaceutics 2011;415:
95101.
hydrogel-based strategies; providing a screening mechanism 20. Winter HH, Chambon F. Analysis of linear viscoelasticity of a
for desirable characteristics prior to implantation that will crosslinking polymer at the gel point. J Rheol 1986;30:367382.
increase efciency and ultimately implant effectiveness. 21. Chambon F, Petrovic ZS, MacKnight WJ, Winter HH. Rheology of
model polyurethanes at the gel point. Macromolecules 1986;19:
21462149.
REFERENCES 22. Chambon F, Winter HH. Linear viscoelasticity at the gel point of a
1. Van Vlierberghe S, Dubruel P, Schacht E. Biopolymer-based crosslinking PDMS with imbalanced stoichiometry. J Rheol 1987;
hydrogels as scaffolds for tissue engineering applications: A 31:683697.
review. Biomacromolecules 2011;12:13871408. 23. Taylor SJ, McDonald JW III, Sakiyama-Elbert SE. Controlled
2. Zuidema JM, Pap MM, Jaroch DB, Morrison FA, Gilbert RJ. Fabri- release of neurotrophin-3 from fibrin gels for spinal cord injury. J
cation and characterization of tunable polysaccharide hydrogel Control Release 2004;98:281294.
blends for neural repair. Acta Biomater 2011;7:16341643. 24. DesKosky BJ, Dormer NH, Ingavle GC, Roatch CH, Lomakin J,
3. Ananthanarayanan B, Kim Y, Kumar S. Elucidating the mecha- Detamore MS, Gehrke SH. Heirarchically designed agarose and
nobiology of malignant brain tumors using a brain matrix- poly(ethylene glycol) interpenetrating network hydrogels for carti-
mimetic hyaluronic acid hydrogel platform. Biomaterials 2011;32: lage tissue engineering. Tissue Eng Part C Methods 2010;16:1533
79137923. 1542.
4. Discher DE, Janmey P, Wang YL. Tissue cells feel and respond to 25. Georges PC, Miller WJ, Meaney DF, Sawyer ES, Janmey PA. Mat-
the stiffness of their substrate. Science 2005;310:11391143. rices with compliance comparable to that of brain tissue select
5. Buxboim A, Ivanovska IL, Discher DE. Matrix elasticity, cytos- neuronal over glial growth in mixed cortical cultures. Biophys J
keletal forces and physics of the nucleus: how deeply do cells 2006;90:30123018.
feel outside and in? J Cell Sci 2010;123:297308. 26. Sharma RI, Snedeker JG. Biochemical and biomechanical gra-
6. Hennink WE, van Nostrum CF. Novel crosslinking methods to dients for directed bone marrow stromal cell differentiation
design hydrogels. Adv Drug Deliv Rev 2002;54:1336. toward tendon and bone. Biomaterials 2010;31:76957704.
7. Liao SW, Yu TB, Guan Z. De novo design of saccharide-peptide 27. Peyton SR, Kim PD, Ghajar CM, Seliktar D, Putnam AJ. The
hydrogels as synthetic scaffolds for tailored cell responses. J Am effects of matrix stiffness and RhoA on the phenotypic plasticity
Chem Soc 2009;131:1763817646. of smooth muscle cells in a 3-D biosynthetic hydrogel system.
8. Hamidi M, Azadi A, Rafiei P. Hydrogel nanoparticles in drug deliv- Biomaterials 2008;29:25972607.
ery. Adv Drug Deliv Rev 2008;60:16381649. 28. Balgude AP, Yu X, Szymanski A, Bellamkonda RV. Agarose gel
9. Ta HT, Dass CR, Larson I, Choong PF, Dunstan DE. A chitosan stiffness determines rate of DRG neurite extension in 3D cultures.
hydrogel delivery system for osteosarcoma gene therapy with Biomaterials 2001;22:10771084.

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH B: APPLIED BIOMATERIALS | JUL 2014 VOL 102B, ISSUE 5 1073