Date of Award

Spring 2003

Document Type


Degree Name

Doctor of Philosophy (PhD)


Mechanical Engineering

First Advisor

Hisham Hegab


A simple thermodynamic, heat transfer, and fluid flow model was developed for microscale Joule-Thomson refrigerators (JT devices). For a given geometry, the model predicted that the cooling capacity of the refrigerator increased with the inlet refrigerant pressure. The effectiveness of the JT device also increased with the inlet pressure, and the heat exchanger channel length. At a constant inlet pressure, the effectiveness, and the refrigeration capacity of a given JT device increased as the aspect ratio of heat exchanger channels was increased. For nitrogen refrigerant, the model predicted that it was possible to obtain approximately 250 mW of refrigeration capacity at 82 K with 10 MPa (100 atm) of inlet pressure and a flow rate of 15.17 ml/s at standard pressure and temperature (STP). This prediction was justified by experimental values of Little (1984) who obtained 250 mW of refrigeration capacity at 83 K with 10 MPa (100 atm) of inlet pressure and a flow rate of 18 ml/s at STP. The simulation model was also used to design a novel JT device based on a layered arrangement of the evaporator, capillary, and the heat exchanger. The proposed JT device would have produced approximately 250 mW of refrigeration capacity at 100 K, for an inlet pressure of 6 MPa (60 atm). This proposed JT device was fabricated on silicon wafers using photolithography. The heat exchanger channels had a cross section of 50 x 20 μm and a length of 6 cm. The capillary channel cross section was 20 x 20 μm and its length was 6 cm. Both the length and the width of the evaporator was 30 mm, and its depth was 20 μm. Pyrex 7740 glass wafers (3 mm thick) were used to separate the evaporator from the capillary and the capillary from the heat exchanger. The heat exchanger was bonded with a top glass cover plate. Most layers were successfully bonded using the anodic bonding procedure. After bonding the evaporator to a glass wafer, subsequent anodic bonding was carried out by applying voltage from sides of each glass and silicon wafer. This bonding attempt demonstrated that the anodic bonding procedure could be used in packaging several silicon and glass wafers. The packaged device held together briefly but later separated due to poor bonding quality of the capillary and the heat exchanger. This poor bonding quality may have resulted from inadequate surface quality of silicon wafers. However, the knowledge and the experience gained in this work will be very useful in future development of JT devices.