Evaluation of a micro-climate for short term loan

Introduction/Problem
As conservators are aware, loans can be both a blessing and a curse for museums.  Recently, a museum curator who was faced with a challenging loan request asked me for assistance.  Another well-respected museum wanted to borrow a small group of old master prints for a short exhibit, but unfortunately the prints would need to be displayed in a building with no HVAC system.  The exhibit was scheduled to run for two weeks in the fall, during which time the average temperature in the region normally fluctuates between 40 and 80° F and the relative humidity (RH) between 30 and 80%.  We calculated the likely cumulative increase in temperature ¹ and agreed that the resulting risk to the prints would be minimal. However, the inevitable fluctuations in RH were a major concern for the prints, especially as they would be hinged into window mats.

Objective
We wanted to create an environment where even with radical temperature fluctuations, changes in RH would be minimal.  The RH in the lending museum is normally 50%+/-4 per annum and we wanted to stay within that range.  Using silica gel to control sealed environments is a well known and accepted conservation practice and we decided to test how well silica gel would work using the museum's standard frames. 

Experimental: Testing effectiveness of sealed enclosure with silica gel
I first designed a test to evaluate the effectiveness of the proposed methods.  I selected a variety of materials with a range of hygroscopicities: Kozo, Gampi, rag paper with letter press printing, vellum and Glassine.  The first four samples were hinged to mat board with tissue and paste; the Glassine was hinged with heat-set tissue activated with a tacking iron.  All samples and framing materials were conditioned to 50% RH before testing.

Two sample sets were prepared; the first set was placed with a datalogger in one of the museum's standard frames, glazed with Plexiglas, and backed with corrugated board.  Tyvek tape replaced the wooden inset and screws that normally secure the back of the frame. ²   The second sample set was placed in an enclosure with two compartments as shown in the illustration below.  The first compartment held the sample set and datalogger. ³   A second chamber behind the backing board held four Tyvek sachets among which one half pound of silica gel was divided, and another datalogger. ⁴   (Illustration 1)  Marvelseal was wrapped around the back and sides of the package, heat-sealed to the Plexiglas and reinforced with strips of Tyvek tape.





In all, four Hobo dataloggers accompanied the materials; one enclosed with each sample set, one in the silica gel chamber and another outside the packages.  Small digital hygrometers were mounted with each sample set, visible through the Plexiglas, and another was kept outside the packages to allow visual monitoring of the conditions.

Over the next three weeks I subjected the packages to fluctuating conditions as much as possible.  I took advantage of the Spring weather conditions and moved the packages outdoors in the cool early mornings.  I placed the packages in humidity chambers several times, sometimes heating them with incandescent lamps or heating pads.  This was not designed to be a fair test of expected conditions, rather I tried to create extreme conditions and compare the effectiveness of the sealed and unsealed packages.  Ultimately the packages were cycled between 30 and 88% RH and between 60 and 90°F with some of those changes occurring very quickly.  As expected, the samples in the package with silica gel fared well while the unsealed package showed severe distortion, even to the extent that the heat-set hinges holding the Glassine popped.  The images below illustrate the effects on each set.  No photos were taken before the experiment was begun, but the set on the left is largely unchanged.





Experimental:  Silica gel vs. blotter
The first experiment indicated that the sealed package with silica gel would protect the prints adequately, but budget constraints and health concerns⁵ made silica gel somewhat problematic.  Therefore, I investigated using multiple sheets of blotter as a buffer.  The experiment described above was repeated using similar test samples and frames, but this time both packages had a chamber behind the backing board.  The first chamber held the same four Tyvek sachets filled with silica gel used in the first experiment, and the second held blotter.⁶  All materials were conditioned to 50% RH before being assembled and sealed. (Illustration 2)  Kozo was not included in this test for two reasons; first it was unresponsive to the first test, and second, this allowed a more prominent placement of the rag paper with letterpress, the sample most indicative of how the old master prints would respond.





Testing conditions were limited by logistics to some extent.  The packages were placed in a small building with open windows and no HVAC for one month during which time conditions in the room fluctuated between 58 and 78°F and 37 and 76% RH.  For the second month they were placed in a room where temperature was controlled between 73 and 74°F with minor exceptions, but RH varied between 54 and 79%.  I made the decision to explore more gradual conditions here that might mimic what will occur in the exhibit, rather than the quick shocks of the incandescent lights, heating pads and humidity chambers.

Results and discussion
While the results of the first experiment had been dramatic, the results of the second were much more subtle.  Inside the silica gel package, datalogger readings showed that T fluctuated between 60 and 78°F over the two month period, reflective of the ambient conditions. For the first month, while the package was in the room with no HVAC, the temperature rose during the day and fell at night, reflective of the changing Spring weather.  Daytime highs were recorded between 63 and 78° F. and nighttime lows between 58 and 74° F.  The RH cycled accordingly; got lower as the temperature rose and rose as the temperature fell, but the changes were smaller than what would be expected.  For example, when the temperature jumped from 70°F to 78°F, the RH only fell from 53 to 52%, not to 40 % as one might expect if calculating from a psychrometric chart.  In the second month, when T was stable and ambient RH was generally above 65%, the RH rose steadily from 53.5 to 55%.

The conditions inside the package with blotters were almost identical to the package with silica gel.  The temperature ranged between 60 and 78°F over the two month period.  The RH began at 53%, showed the same cycling for the first month, and in the second month slowly rose from 53.2 to 55%.  At the end of two months, when the packages were unsealed, they were left unsealed, but with Plexi placed over enclosures in a room at 65%RH for a week in order to see how quickly the materials equilibrated to ambient conditions when not protected by the Marvelseal, but only the weight of the Plexiglas on the frames.  There are no photos from when the packages were initially unsealed because it would have meant lifting the Plexiglas and exposing the samples to the ambient RH.  During this week the humidity in both enclosures rose steadily, but the package with the silica gel stayed slightly lower and more stable than the package with the blotters.  

        

The samples had been relatively flat when the packages were un-sealed.  The photographs here were taken after the samples had sat unsealed but covered in a room at 65% RH and show the resulting distortion, especially to the Gampi and Glassine samples. The distortion appeared similar on both sets.  There are no photos from when the packages were initially unsealed.


Conclusion
I was not able to test for the full range of expected conditions, but the results are encouraging.  The slow rise in RH in the second month of the second test probably indicates that the packages were not sealed perfectly, or that the rough treatment they received moving them from place to place weakened the seals.  Although the samples and framing materials were conditioned to 50% RH before the testing, they tended to stabilize at 52%RH when packaged.  This most likely occurred because the RH in the room where I was working tended to be between 60 and 65%.  The wooden frames may not have been conditioned for a long enough period of time.

One interesting finding came from the first experiment.  When I shocked the packages with severe increases or decreases in temperature, the relationship between T and RH was opposite what would be expected.  According to the psychrometric chart, as T rises, RH should fall, and vice versa.   But when I used a heating pad to raise the temperature inside the package with the silica gel, the RH rose as well.  As the temperature inside the enclosure rose from 72 to 87°F the RH rose from 52 to 60%.  The following morning when I put the packages outside in the cool morning air, T went from 74 to 64°F and RH went from 54 to 50%.  My suspicion is that what I was seeing was a temporary "reconditioning" of the silica gel but the equilibrium moisture content of the sample materials was not significantly affected.  This observation is supported by the fact that while quick and severe fluctuations appear on the datalogger charts, the sample materials inside the sealed package remained relatively undistorted.  The samples inside the unsealed frame, where highs and lows were sustained for longer periods of time, were severely distorted.

At the end of the second experiment when the packages were unsealed but left with Plexi covers in a humid room, the silica gel package appeared to control the RH more effectively.  The two frames were not constructed identically and is not clear whether this is the result of the buffering capacity of the silica gel, or the amount of air exchange in each package.

This was not a rigorous study, but a quick experiment designed to test potential enclosure strategies for a single, short-term loan and to satisfy some personal curiosity.  The photography was not as systematic as would have been useful.  The charts from the dataloggers are not reproduced here because they are difficult to read in small formats.  Although there are recommendations available for how much silica gel to use, the choice of the number of sheets of blotter was more intuitive.  All these flaws should be considered if the study is repeated.



1 The temperature in the lending museum is set at 72°F.  The temperature in the borrowing institution is likely to get to 80°F for three hours a day at most for the two week period, but the temperature is likely to be below 72°F for at least 18 hours each day.  
2 The frames are designed to hold matted prints and had to be modified to accommodate the thickness of a datalogger.   So the wooden inset that normally is screwed in behind the backing board was transferred to the front of the backing board to create space behind the Plexiglas.
3 Sides were constructed from mat board to accommodate the thickness of the datalogger.
4 The total dimensions of both chambers of the enclosure were 16 x 20 x 2" and an excess amount of silica gel was used.
Weintraub, S. (2002). Demystifying silica gel. Retrieved July 24, 2013, from
      http://www.apsnyc.com/uploads/Demystifying%20Silica%20Gel.pdf
       (Reprinted from Object Specialty Group Postprints, 9.)
5 Material safety data sheet: Silica gel dessiccant. (1997, September 2).
     Retrieved July 24, 2013, from http://www.atmos.umd.edu/~russ/MSDS/
     silicagel28200.html
6 The total dimensions of the first package were 16 x 20 x 2" and one pound of silica gel was placed inside the chamber.  The total dimensions of the second package were 16 x 20 x 1 1/2" and eight sheets of blotter were cut to fill the chamber.  The number of blotter sheets was chosen according to what would fit in the chamber without crowding.